![Academic Press is an imprint of Elsevier
525 B Street, Suite 1800, San Diego, CA 92101-4495, USA
225 Wyman Street, Waltham, MA 02451, USA
The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK
32 Jamestown Road, London, NW1 7BY, UK
Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands
Copyright © 2014, 2006 Elsevier, Inc. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form
or by any means electronic, mechanical, photocopying, recording or otherwise without the prior
written permission of the publisher.
Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in
Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: permissions@elsevier.com.
Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.
com/locate/permissions, and selecting Obtaining permission to use Elsevier material.
Notice
No responsibility is assumed by the publisher for any injury and/or damage to persons or property as
a matter of products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions or ideas contained in the material herein. Because of rapid advances in the
medical sciences, in particular, independent verification of diagnoses and drug dosages should be
made.
Library of Congress Cataloging-in-Publication Data
Computational systems biology (Kriete)
Computational systems biology / edited by Andres Kriete, Roland Eils. -- Second edition.
p. ; cm.
Includes bibliographical references and indexes.
ISBN 978-0-12-405926-9 (alk. paper)
I. Kriete, Andres, editor of compilation. II. Eils, Roland, editor of compilation. III. Title.
[DNLM: 1. Computational Biology. 2. Systems Biology. QU 26.5]
QH324.2
570.1’13--dc23
2013045039
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library.
ISBN: 978-0-12-405926-9
For information on all Academic Press publications
visit our website at store.elsevier.com
Printed in the United States of America
14 15 10 9 8 7 6 5 4 3 2 1](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-8-320.jpg)
![8 Robustness and fragility and application to the cell cycle 37
Figure 3.3 shows that the three couples of Clb cyclins (Clb5,6, Clb3,4, and Clb1,2) undergo
waves with amplitudes at different times. We shall first focus on the Clb5/6 couple and on the
onset and the decay of the peak of their level. Figure 3.2 shows that Clb5/6 has a maximum
at t = 23 and that at t = 8 it is hallway reaching that maximum and at t = 31.5 it is again half-
way down. We computed the time-control coefficient Tc at those times, and this amounted to
0.87 and −4.57, respectively. We also computed the robustness R (defined as the inverse of the
time-control coefficients) of the amplitude of Clb5/6 at the two halfway time points. We first
computed the robustness of the Clb5/6 amplitude exercised by all processes in the network
when perturbed simultaneously and equally in the same direction. The corresponding robust-
nesses were 1.14849 and 0.21881 (see Table 3.1).
Quinton-Tulloch et al. (2013) identified the inverse of the robustness with fragility, which
in turn is equal to the control coefficient. As a consequence the above (Equation 3.10):
can be reformulated as:
For this case of equal perturbations of all processes amounts to:
And the same for the other time point analyzed. The last two columns in Table 3.1 confirms
this computationally.
For the onset of the peak the time-control coefficient is positive. Hence the positive robust-
nesses must be smaller (typically those with respect to kinases perturbations) than the robust-
nesses with respect to the phosphatases (which are not considered explicitly in the system
yet). For the decay of the peak, the inverse should be true.
(3.10)
n
j=1
CXk
ej
(t) = CXk
t (t)
(3.22)
n
j=1
1
ReXk
ej
(t)
= CXk
t (t)
(3.23)
n
j=1
1
ℜXk
total for equal(t 1
2on
)
= CXk
t (t 1
2on
)
0 10 20 30 40 50 60
0
0.3
0.6
0.9
1.2
Time
Clb3,4
Clb5,6
Clb1,2
[Clb]
FIGURE 3.3 Computational time course of Clb cyclins couples over time.](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-48-320.jpg)
![38 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS
9 PERFECT ADAPTATION AND INTEGRAL CONTROL IN
METABOLISM
Biochemical reaction networks can exhibit properties similar to those of control system
structures in control engineering, but are they identical? The robustness of cellular adapta-
tion to environmental conditions is often related to negative feedback control structures. For
example, robust adaptations in a bacterial chemotaxis signaling network, in mammalian iron
and calcium homeostasis, and in yeast osmoregulation have been interpreted as integral feed-
back control systems (Yi et al. 2000; El-Samad et al. 2002; Ni et al. 2009; Muzzey et al. 2009).
A recent study identified the three different types of control structures used in control engi-
neering, i.e. proportional, integral, and derivative feedback control, in regulations of energy
metabolism (Cloutier and Wellstead 2010). In addition, specific nonlinear dynamics in signal-
ing networks, such as oscillation or bi-stability, can be induced by positive feedback loops.
Feed-forward control structures are also observed in gene regulatory networks (Mangan and
Alon 2003), as well as in the regulation of glycolytic intermediates (Bali and Thomas 2001).
Regulation in living cells tends to occur at multiple levels simultaneously with a hierarchi-
cal structure (Westerhoff 2008). In a metabolic network the regulation of a reaction rate can be
TABLE 3.1 Calculation of time-control coefficient and robustness for Clb cyclins couples.
Clb5,6 α(5,6) β(5,6)
Max [5,6] Time Max [5,6]/2 t1/2 α(5,6) Max [5,6] Time Max [5,6]/2 t1/2 β(5,6)
1,41473 23 0,707365 8 0,076988 1,41473 23 0,789371 31,5 0,11453
Clb3,4 α(3,4) β(3,4)
Max [3,4] Time Max [3,4]/2 t1/2 α(3,4) Max [3,4] Time Max [3,4]/2 t1/2 β(3,4)
0,959239 30 0,4796195 19,5 0,023046 0,959239 30 0,60479 37 0,05489
Clb1,2 α(1,2) β(1,2)
Max [1,2] Time Max [1,2]/2 t1/2 α(1,2) Max [1,2] Time Max [1,2]/2 t1/2 β(1,2)
1,1148 34 0,5574 27 0,094539 1,1148 34 0,780522 41,5 0,049
Tcα Tcβ Rα Rβ 1/Rα 1/Rβ
Tcα(5,6) Tcβ(5,6) Rα(5,6) Rβ(5,6) Rα(5,6) Rβ(5,6)
0,87071 4,57023 1,14849 0,21881 0,87071 4,57023
Tcα(3,4) Tcβ(3,4) Rα(3,4) Rβ(3,4) Rα(3,4) Rβ(3,4)
0,937 3,358332 1,06723 0,29777 0,937 3,35833
Tcα(1,2) Tcβ(1,2) Rα(1,2) Rβ(1,2) Rα(1,2) Rβ(1,2)
4,57937 2,605393 0,21837 0,38382 4,57937 2,60539](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-49-320.jpg)
![Sustained by this tenacious but fantastic tradition, Ministers have not
infrequently engaged in policies which wiser men would have avoided.
They have uttered protests, warnings, threats which have gone unheeded.
They have presumed to say what would and would not be tolerated in
certain spheres; but having nothing better behind their despatches than a
mere assumption which did not correspond with the facts, they have been
compelled to endure rebuffs and humiliations. As they had not the prudence
to cut their coat according to their cloth, it was only natural that
occasionally they should have had to appear before the world in a
somewhat ridiculous guise.
British statesmen for nearly half a century had persisted in acting upon
two most dangerous assumptions. They had assumed that one branch of the
national armaments conformed to their policy, when in fact it did not. And
they had assumed also, which is equally fatal, that policy, if only it be
virtuous and unaggressive, is in some mysterious way self-supporting, and
does not need to depend on armaments at all.
The military preparations of Britain were inadequate to maintain the
policy of Security, which British Governments had nevertheless been
engaged in pursuing for many years prior to the outbreak of the present war.
[7] On the other hand, the abandonment of this policy was incompatible
with the continuance of the Empire. We could not hope to hold our scattered
Dependencies and to keep our Dominions safe against encroachments
unless we were prepared to incur the necessary sacrifices.
[1] American writers have urged criticism of this sort against the
armaments of the U.S.A., which they allege are inadequate to uphold the
policy of the 'Monroe Doctrine.' The German view of the matter has
been stated by the Chancellor (April 7, 1913) when introducing the
Army Bill:—History knows of no people which came to disaster
because it had exhausted itself in the making of its defences; but history
knows of many peoples which have perished, because, living in
prosperity and luxury, they neglected their defences. A people which
thinks that it is not rich enough to maintain its armaments shows merely
that it has played its part.](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-53-320.jpg)
![[2] So the argument runs, and the course of our naval policy since Mr.
Stead's famous press campaign in 1884 will be cited as an
encouragement.
[3] E.g. in the winter of 1908 and spring of 1909, when an influential
section of the supporters of the present Cabinet chose to believe the false
assurances of the German Admiralty, and freely accused their own
Government of mendacity.
[4] Innovations of this particular sort have possibly a better chance of
preserving their existence than some others. 'Boards are screens,' wrote
John Stuart Mill, or some other profound thinker; and in politics screens
are always useful.
[5] This is obvious from the White Paper without seeking further
evidence in the ministerial press or elsewhere.
[6] Of the six infantry divisions included in the Expeditionary Force only
four were sent in the first instance; a fifth arrived about August 24; a
sixth about mid-September.
[7] Our Army, as a belligerent factor in European politics, is almost a
negligible quantity. This Empire is at all times practically defenceless
beyond its first line. Such an Empire invites war. Its assumed security
amid the armaments of Europe, and now of Asia, is insolent and
provocative (Lord Roberts, October 22, 1912). Nothing indeed is more
insolent and provocative, or more likely to lead to a breach of the peace,
than undefended riches among armed men.
CHAPTER IV
THE BALANCE OF POWER
During the whole period of rather more than thirteen years—which has
been referred to in previous pages as the post-Victorian epoch, and which
extended roughly from January 1901, when Queen Victoria died, to July
1914, when war was declared—the British Army remained inadequate for
the purpose of upholding that policy which British statesmen of both](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-54-320.jpg)
![DERELICT
MAXIMS
anticipated aggressions of the Triple Alliance, wherein Germany was the
predominant partner.
The tendency of phrases, as they grow old, is to turn into
totems, for and against which political parties, and even
great nations, fight unreasoningly. But before we either yield
our allegiance to any of these venerable formulas, or decide to throw it out
on the scrap-heap, there are advantages in looking to see whether or not
there is some underlying meaning which may be worth attending to. It
occasionally happens that circumstances have changed so much since the
original idea was first crystallised in words, that the old saying contains no
value or reality whatsoever for the present generation. More often, however,
there is something of permanent importance behind, if only we can succeed
in tearing off the husk of prejudice in which it has become encased. So,
according to Disraeli, the divine right of Kings may have been a plea for
feeble tyrants, but the divine right of government is the keystone of human
progress. For many years the phrase British interests, which used to figure
so largely in speeches and leading articles, has dropped out of use, because
it had come to be associated unfavourably with bond-holders' dividends.
The fact that it also implied national honour and prestige, the performance
of duties and the burden of responsibilities was forgotten. Even the doctrine
of laissez faire, which politicians of all parties have lately agreed to abjure
and contemn, has, as regards industrial affairs, a large kernel of practical
wisdom and sound policy hidden away in it. But of all these derelict
maxims, that which until quite recently, appeared to be suffering from the
greatest neglect, was the need for maintaining the Balance of Power in
Europe. For close on two generations it had played no overt part in public
controversy, except when some Tory matador produced it defiantly as a red
rag to infuriate the Radical bull.
If this policy of the maintenance of the Balance of Power has been little
heard of since Waterloo, the reason is that since then, until quite recently,
the Balance of Power has never appeared to be seriously threatened.[1] And
because the policy of maintaining this balance was in abeyance, many
people have come to believe that it was discredited. Because it was not
visibly and actively in use it was supposed to have become entirely useless.](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-57-320.jpg)
![DEFENCE
AND
INVASION
combination of nations. But we cannot build ships against half Europe. If
Western Europe, with all its ports, its harbours, its arsenals, and its
resources, was to fall under the domination of a single will, no effort of ours
would be sufficient to retain the command of the sea. It is a balance of
power on the continent, which alone makes it possible for us to retain it.
Thus the maintenance of the balance of power is vital to our superiority at
sea, which again is vital to the security of the British Empire.[2]
Security in the widest sense was the ultimate end of our policy—security
of mind, security from periodic panic, as well as actual military security.
Looked at more closely, the immediate end was defence—the defence of the
British Empire and of the United Kingdom.
In the existing condition of the world a policy of
'splendid isolation' was no longer possible. Conditions with
which we are familiar in commercial affairs, had presented
themselves in the political sphere, and co-operation on a
large scale had become necessary in order to avoid bankruptcy. England had
entered into the Triple Entente because her statesmen realised, clearly or
vaguely, that by doing so we should be better able to defend our existence,
and for no other reason.
After 1911 it must have been obvious to most people who considered the
matter carefully that in certain events the Triple Entente would become an
alliance. It is the interest as well as the duty of allies to stand by one another
from first to last, and act together in the manner most likely to result in
victory for the alliance. What then was the manner of co-operation most
likely to result in victory for that alliance which lay dormant under the
Triple Entente?
But first of all, to clear away one obscurity—Invasion was not our
problem; Defence was our problem; for the greater included the less.
The word 'defence' is apt to carry different meanings to different minds.
The best defence of England and British interests, at any given time, may or
may not consist in keeping our main army in the United Kingdom and
waiting to be attacked here. It all depends upon the special circumstances of
each case. The final decision must be governed by one consideration, and](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-60-320.jpg)
![disapprovingly 'foreign adventures,' and that we should be wise to await
attack behind our own shores. Recent events have wrought such a complete
and rapid conversion from this heresy, that it is no longer worth while
wasting words in exposing it. It is necessary, however, to recall how
influential this view of the matter was, not only up to the declaration of war,
but even for some time afterwards.
As to the precise form of co-operation between the members of the
Triple Entente in case of war, there could be no great mystery. It was
obvious to any one who paid attention to what happened during the summer
and autumn of 1911, that in the event of Germany attacking France over the
Agadir dispute, we had let it be understood and expected, that we should
send our Expeditionary Force across the Channel to co-operate with the
French army on the north-eastern frontier.
[1] It can hardly be overlooked, however, that this principle, rightly or
wrongly interpreted, had something to do with the Crimean War (1854-
56) and with the British attitude at the Congress of Berlin (1878).
[2] Viscount Milner in the United Service Magazine, January 1912.
CHAPTER V
THE MILITARY SITUATION
(August 1911)
The full gravity of the Agadir incident, though apparent to other nations,
was never realised by the people of this country. The crisis arose suddenly
in July 1911. Six weeks later it had subsided; but it was not until well on in
the autumn that its meanings were grasped, even by that comparatively
small section of the public who interest themselves in problems of defence
and foreign affairs. From October onwards, however, an increasing number](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-62-320.jpg)
![THE
EXPEDITION
ARY FORCE
began to awake to the fact, that war had only been avoided by inches, and to
consider seriously—many of them for the first time in their lives—what
would have happened if England had become involved in a European
conflict.
From various official statements, and from discussions
which from time to time had taken place in Parliament, it
was understood that our 'Expeditionary Force' consisted of
six infantry divisions, a cavalry division, and army troops;
[1] also that the national resources permitted of this force being kept up to
full strength for a period of at least six months, after making all reasonable
deductions for the wastage of war. Was this enough? Enough for what? ...
To uphold British policy; to preserve Imperial security; to enable the Triple
Entente to maintain the balance of power in Europe. These were vague
phrases; what did they actually amount to? ... The adequacy or inadequacy
of such an army as this for doing what was required of it—for securing
speedy victory in event of war—or still better for preserving peace by the
menace which it opposed to German schemes of aggression—can only be
tested by considering the broad facts with regard to numbers, efficiency,
and readiness of all the armies which would be engaged directly, or
indirectly, in a European struggle.
War, however, had been avoided in 1911, and not a few people were
therefore convinced that the menace of the available British army, together
with the other consequences to be apprehended from the participation of
this country, had been sufficient to deter Germany from pursuing her
schemes of aggression, if indeed she had actually harboured any notions of
the kind. But others, not altogether satisfied with this explanation and
conclusion, were inclined to press their enquiries somewhat further.
Supposing war had actually been declared, would the British force have
been sufficient—acting in conjunction with the French army—to repel a
German invasion of France and Belgium, to hurl back the aggressors and
overwhelm them in defeat? Would it have been sufficient to accomplish the
more modest aim of holding the enemy at his own frontiers, or even—
supposing that by a swift surprise he had been able to overrun Belgium—at
any rate to keep him out of France?](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-63-320.jpg)
![NEUTRALIT
Y OF ITALY
When people proceeded to seek for answers to these questions, as many
did during the year 1912, they speedily discovered that, in considerations of
this sort, the governing factor is numbers—the numbers of the opposing
forces available at the outbreak of war and in the period immediately
following. The tremendous power of national spirit must needs be left out
of such calculations as a thing immeasurable, imponderable, and uncertain.
It was also unsafe to assume that the courage, intelligence, efficiency,
armament, transport, equipment, supplies, and leadership of the German
and Austrian armies would be in any degree inferior to those of the Triple
Entente. Certain things had to be allowed for in a rough and ready way;[2]
but the main enquiry was forced to concern itself with numerical strength.
There was not room for much disagreement upon the broad facts of the
military situation, among soldiers and civilians who, from 1911 onwards,
gave themselves to the study of this subject at the available sources of
information; and their estimates have been confirmed, in the main, by what
has happened since war began. The Intelligence departments of London,
Paris, and Petrograd—with much ampler means of knowledge at their
disposal—can have arrived at no other conclusions. What the English War
Office knew, the Committee of Imperial Defence likewise knew; and the
leading members of the Cabinet, if not the whole Government, must be
presumed to have been equally well informed.
It was assumed in these calculations, that in case of tension between the
Triple Entente and the Triple Alliance, the latter would not be able—in the
first instance at all events—to bring its full strength into the struggle. For
unless Germany and Austria managed their diplomacy before the outbreak
of hostilities with incomparable skill, it seemed improbable that the Italian
people would consent to engage in a costly, and perhaps ruinous, war—a
war against France, with whom they had no quarrel; against England,
towards whom they had long cherished feelings of friendship; on behalf of
the Habsburg Empire, which they still regarded—and not altogether
unreasonably—with suspicion and enmity.
But although the neutrality of Italy might be regarded as
a likelihood at the opening of the war, it could not be
reckoned on with any certainty as a permanent condition.](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-64-320.jpg)
![SUPERIORIT
Y OF
GERMAN
NUMBERS
For as no one can forecast the course of a campaign, so no one can feel
secure that the unexpected may not happen at any moment. The
consequences of a defeat in this quarter or in that, may offer too great
temptations to the cupidity of onlookers; while diplomacy, though it may
have bungled in the beginning, is sure to have many opportunities of
recovering its influence as the situation develops. Consequently, unless and
until Italy actually joined in the struggle on the side of the Triple Entente, a
considerable section of the French army would, in common prudence, have
to be left on guard upon the Savoy frontier.
In a war brought on by the aggressive designs of Germany, the only
nations whose participation could be reckoned on with certainty—and this
only supposing that Britain stood firmly by the policy upon which her
Government had embarked—were Russia, France, and ourselves on the one
side, Germany and Austria-Hungary on the other.
It would certainly be necessary for Germany, as well as Austria, to
provide troops for coast defences, and also for the frontiers of neutral
countries, which might have the temptation, in certain circumstances, to
deneutralise themselves at an inconvenient moment, if they were left
unwatched. On the north and west were Denmark, Holland, and Belgium,
each of which had a small field army, besides garrison and fortress troops
which might be turned to more active account upon an emergency. On the
south and east were Montenegro, Servia, and Roumania, whose military
resources were on a considerable scale, and whose neutrality was not a
thing altogether to be counted on, even before the Balkan war[3] had
lowered the prestige of Turkey. In addition there was Italy, who although a
pledged ally in a defensive war was not likely, for that reason, to consider
herself bound to neutrality, benevolent or otherwise, if in her judgment, the
particular contingencies which called for her support had not arisen at the
outset.
After taking such precautions as seemed prudent under
these heads, Germany would then be obliged to detach for
service, in co-operation with the Austrians in Poland, and
along the whole eastern border, a sufficient number of army
corps to secure substantial superiority over the maximum](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-65-320.jpg)
![forces which Russia, hampered by an inadequate railway system and
various military considerations,[4] could be expected to bring into the field
and maintain there during the first few months of the war.
It was reckoned[5] after taking all these things into account, that
Germany would have available, for the invasion of France, an army
consisting of some ninety divisions—roughly, rather more than a million
and three-quarters of men—and that she could maintain this force at its full
strength—repairing the wastage of war out of her ample reserves—for a
period of at least six months. It was assumed that the Kaiser, relying upon
the much slower mobilisation of Russia, would undoubtedly decide to use
the whole of this huge force in the west, in the hope that before pressure
could begin to make itself felt in the east, France would either have been
crushed, as she was in 1870, or so much mangled that it would be possible
to send reinforcements of an overwhelming character to make victory
secure in Poland.
Against this German force of 1,800,000, France, according to the best
information available, could put into the field and maintain at full strength
for a similar period of six months about 1,300,000 men. But this was the
utmost that could be expected of the French, and the initial discrepancy of
500,000 men was very serious. It precluded all reasonable hope on their
part of being able to take the offensive, to which form of warfare the genius
of the people was most adapted. It would compel them to remain on the
defensive, for which it was believed at that time—though wrongly, as
events have proved—that they were ill suited by temperament as well as
tradition.
If England joined in the war by land as well as sea the numerical
deficiency would be reduced to 340,000 on the arrival of our Expeditionary
Force. In this connection, as well as for other reasons, the attitude of
Holland and Belgium, and that of Germany with respect to these two
countries, were clearly matters of high importance.
Holland had a field army of four divisions, and her interests could be
summed up in the words, 'preservation of independence.' She would](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-66-320.jpg)
![likely to occur. But even if our Expeditionary Force did go, it was
altogether inadequate to redress the adverse balance; still more inadequate
to bring an immediate victory within the range of practical possibility. It
was inadequate to hold back the premeditated invasion, either at the
German frontier, or even at the French frontier. It was inadequate to make
Belgian resistance effective, even if that nation should determine to throw
in its lot with the Triple Entente.
As a matter of the very simplest arithmetic our land forces were
inadequate for any of these purposes. They were unequal to the task of
maintaining the balance of power by giving a numerical superiority to the
armies of the Triple Entente. Our armaments therefore did not correspond
with our policy. It was clear that they would not be able to uphold that
policy if it were put to the supreme test of war. It was impossible to
abandon our policy. It was not impossible, and it was not even in 1912 too
late, to have set about strengthening our armaments. Nothing of the kind,
however, was undertaken by the Government, whose spokesmen, official
and unofficial, employed themselves more congenially in deriding and
rebuking Lord Roberts for calling attention to the danger.
Of course if it had been possible to place reliance upon the statement of
the English War Minister, made little more than a year before war broke
out,[6] that every soldier under the voluntary system is worth ten conscripts,
we and our Allies would have been in a position of complete security. In
that case our force of 160,000 would have been the equivalent of 1,600,000
Germans, and we should from the first have been in a superiority of more
than a million over our enemies.
Even if we could have credited the more modest assumption of the
Attorney-General—made nearly four months after war broke out—that one
volunteer was worth three 'pressed' men, the opposing forces would have
been somewhere about an equality.[7]
Unfortunately both these methods of ready-reckoning were at fault,
except for their immediate purpose of soothing, or deluding the particular
audiences to which they were addressed. The words were meaningless and
absurd in a military sense; though conceivably they possessed some occult](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-69-320.jpg)
![THE THREE
PERIODS OF
WAR
political virtue, and might help, for a time at least, to avert the retribution
which is due to unfaithful stewards.
Both these distinguished statesmen, as well as many of their colleagues
and followers, were beset by the error of false opposites. A soldier who has
enlisted voluntarily, and another who is a conscript or 'pressed' man, have
equally to fight their country's enemies when they are ordered to do so. In
both cases the particular war may be against their consciences and
judgments; and their participation in it may therefore be involuntary.
Of two men—equal in age, strength, training, and courage—one of
whom believes his cause to be just, while the other does not, there can be no
doubt that the former will fight better than the latter—even though the latter
was enlisted under the voluntary system while the former was a conscript or
'pressed' man. In this sense the superiority of the 'voluntary' principle is
incontestable. But is there any evidence to show, that either the original
soldiers, or the new levies, of the German army are risking their lives in this
war any less willingly than our own countrymen, who went out with the
Expeditionary Force, or those others who have since responded to Lord
Kitchener's appeal? Is there any reason to suppose that they are fighting any
less bravely and intelligently?[8]
Another matter of importance in these calculations with regard to the
military strength of the Triple Entente and the Triple Alliance was the time
limit.
There are three periods in war. There is the onset of war,
where swiftness of action is what tells most; there is the grip
of war, where numbers of trained men are what tell most;
and there is the drag of war, when what tells most is the
purse.
Speaking by the book, it is of course numbers which tell all the way
through. At the beginning—in the onset—the aim is to hurl superior
numbers at a vital point—taking the enemy by surprise, and thereby](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-70-320.jpg)
![LIMITATION
S OF SEA
or German soil—whether there was to be a German invasion of France or a
Franco-British invasion of Germany. Consequently, if our Expeditionary
Force was to render assistance at the critical time, it must reach its position
on the frontier within a fortnight of the outbreak of war.
As to the drag of war, the Triple Entente had the advantage, if that stage
were ever reached. For the purses of England, France, and Russia were
much longer than those of Germany and Austria. It was important, however,
to remember that there would be no hope for us in the drag of war, if
Germany could deliver a heavy enough blow at the beginning, as she did in
1870.
These were the considerations as to time, which presented themselves to
students of the military situation during the breathing space which followed
upon the Agadir crisis. The substantial accuracy of this forecast was
confirmed by what happened during August and September of last year. In
1914 war was declared by Germany on August 1st. For several days before
she had been engaged actively in mobilisation. Three weeks later three
important battles—on the road to Metz, at Charleroi, and at Mons[9]—were
won by the Germans. If it had not been for the unexpected obstacle of Liège
the last two engagements would in all probability have been fought at an
even earlier date, and in circumstances much more unfavourable to the
Franco-British forces. But in the early days of September, instead of the
crushing defeat of Sedan, there was the victory of the Marne, and the
Germans were forced to retreat to entrenched positions north of the Aisne.
[10]
The onset period was ended; but the issue had not been settled as in
1870. France and England had not been knocked out in the first round. To
this extent the supreme German endeavour had miscarried. Nevertheless a
great advantage had been secured by our enemies, inasmuch as it was now
apparent that the ensuing campaign—the grip of war—would be contested,
not on German soil, but in France and Belgium.
The value of the assistance which the British Navy would
be able to render to the cause of the Triple Entente was a](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-72-320.jpg)
![Power which is only a Sea Power will be overborne with numbers, and
finally worsted by the victorious Land Power. For how is it possible to fight
with one hand against an enemy with two hands? The fleets of Europe
which at last must be combined against us, if we allow any rival to obtain a
European predominance, are too heavy odds. German preparations alone
were already causing us grave anxiety nearly three years before the Agadir
crisis occurred. How then could we hope to build against the whole of
Europe? Or even against half of Europe, if the other half remained coldly
neutral?
[1] In all about 160,000 men, of whom some 25,000 were non-
combatants.
[2] Such, for instance, as the fact that the time-table of German
mobilisation appeared to be somewhat more rapid than that of the
French, and much more so than that of the Russians.
[3] The first Balkan war broke out in the autumn of 1912.
[4] Russia had anxieties of her own with regard to the intentions of
Roumania, of Turkey in Persia and the Caucasus, and of China and
Japan in the Far East.
[5] These calculations were worked out in various ways, but the net
results arrived at were always substantially the same. In view of the fact
that the main conclusions have been amply proved by the results of the
present war, it does not seem worth while to weary the reader with more
sums in arithmetic than are absolutely necessary.
[6] Colonel Seely at Heanor, April 26, 1913.
[7] Sir John Simon (Attorney-General and a Cabinet Minister), at
Ashton-under-Lyne, November 21, 1914.... This speech is instructive
reading. It is also comforting for the assurance it contains, that if the
speaker approved of our taking part in this war (as he vowed he did) his
audience might rest satisfied that it was indeed a righteous war; seeing
that war was a thing which, on principle, he (Sir John Simon) very much
reprehended. And yet we are not wholly convinced and reassured. There
is a touch of over-emphasis—as if perhaps, after all, the orator needed
the support of his own vehemence to keep him reminded of the](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-74-320.jpg)
![righteousness. The pacifist in war-paint is apt to overact the unfamiliar
part. One wonders from what sort of British officer at the front the
Attorney-General had derived the impression that 'one' of our own
voluntary soldiers—gallant fellows though they are—is the equal of
'three' of the Germans who face him, or of the Frenchmen who fight by
his side.... This speech puts us not a little in mind of Evangelist's
warning to Christian, with regard to Mr. Legality's fluent promises to
relieve him of his burden—There is nothing in all this noise save a
design to beguile thee of thy salvation.
[8] Sir John Simon clinched his arithmetical calculation of 'three' to 'one,'
by stating that 'the Kaiser already knew it'; and this reassuring statement
was received with 'laughter and cheers.' The laughter we can understand.
[9] The battle in Northern Alsace was fought on August 21 and 22. A
French army was driven back at Charleroi on the 22nd, and the British at
Mons on the 23rd.
[10] September 6-12.
CHAPTER VI
THE MILITARY SITUATION
(August 1914)
Such was the position of affairs at July 1911, as it appeared to the eyes
of people who—during the ensuing period—endeavoured to arrive at an
understanding of the problem without regard to the exigencies of party
politics. Between that date and July 1914, when war broke out, various
changes took place in the situation. The general effect of these changes was
adverse to Britain and her allies.
In 1911 the German estimates provided for considerable increases,
especially in artillery and machine-guns. The peace strength of the Army
was raised.](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-75-320.jpg)
![entertained conscientious and insuperable objections to bearing its due
share of the burden.
Already, prior to the sensational expansion of Germany in 1913, France
had endeavoured to counteract the current yearly increases in the military
estimates of her neighbour, by various reorganisations and regroupings of
active units, and by improvements calculated to improve the efficiency of
the reserves. But when information was forthcoming[1] as to the nature and
extent of the developments proposed under the German Army Bill of 1913,
it was at once realised that more drastic measures were essential to national
safety.
Before the German projects were officially announced, the French
Government took the bold step of asking the legislature to sanction a
lengthening of the period of active military service from two years to three,
and an extension of the age limit of the reserves from forty-seven to forty-
nine. Power was also taken to summon, in case of emergency, the annual
contingent of recruits a year before their due time. Increases in artillery,
engineers, railways, barrack accommodation, and subsidiary services were
asked for and obtained. The cost of these, when the whole sum came to be
calculated, was found to amount to £32,000,000.
Apart, therefore, from material preparations of one kind and another,
Germany was taking steps to add 200,000 men to her striking force, and the
intentions of France were approximately the same. In the case of Germany,
however, the increases of strength would be operative by Midsummer 1914,
while with France they would not take effect until two years later.[2]
Germany, moreover, was arranging to take 63,000 more recruits
annually. France was unable to obtain any more recruits, as she already took
all that were fit to bear arms. The increase in her striking force was made
mainly at the expense of her reserves. Year by year, therefore, the numerical
inferiority of France must become more marked.
Russia meanwhile was proceeding with her programme of military
extension and reorganisation which had been decided on after the Japanese
war. A great part of her expenditure was being devoted to the improvement](https://image.slidesharecdn.com/129074-250520083015-07bf2466/85/Full-Computational-Systems-Biology-second-edition-Roland-Eils-77-320.jpg)
Full Computational Systems Biology second edition Roland Eils
- 1. Full Computational Systems Biology second edition Roland Eils install download https://ebookmeta.com/product/full-computational-systems-biology- second-edition-roland-eils/ Download more ebook from https://ebookmeta.com
- 2. We believe these products will be a great fit for you. Click the link to download now, or visit ebookmeta.com to discover even more! An Introduction to Computational Systems Biology Systems Level Modelling of Cellular Networks 1st Edition Karthik Raman https://ebookmeta.com/product/an-introduction-to-computational- systems-biology-systems-level-modelling-of-cellular-networks-1st- edition-karthik-raman/ Computational Systems Biology in Medicine and Biotechnology Methods and Protocols Methods in Molecular Biology 2399 Sonia Cortassa (Editor) https://ebookmeta.com/product/computational-systems-biology-in- medicine-and-biotechnology-methods-and-protocols-methods-in- molecular-biology-2399-sonia-cortassa-editor/ Bioinformatics and Computational Biology 1st Edition Hamid R. Arabnia https://ebookmeta.com/product/bioinformatics-and-computational- biology-1st-edition-hamid-r-arabnia/ Ancient Arsenal 1st Edition Jeffery H Haskell https://ebookmeta.com/product/ancient-arsenal-1st-edition- jeffery-h-haskell/
- 3. Introduction to Simple Shock Waves in Air With Numerical Solutions Using Artificial Viscosity 2nd Edition Seán Prunty https://ebookmeta.com/product/introduction-to-simple-shock-waves- in-air-with-numerical-solutions-using-artificial-viscosity-2nd- edition-sean-prunty/ A Guide to UX Design and Development 1st Edition Tom Green https://ebookmeta.com/product/a-guide-to-ux-design-and- development-1st-edition-tom-green/ Universe Robert Geller https://ebookmeta.com/product/universe-robert-geller/ Dissent and the Bible in Britain c 1650 1950 1st Edition Mandelbrote Scott Ledger Lomas Michael https://ebookmeta.com/product/dissent-and-the-bible-in- britain-c-1650-1950-1st-edition-mandelbrote-scott-ledger-lomas- michael/ The New Civilization Upon Data: How Big Data Reshapes Human Civilization, Business And The Personal World 1st Edition Zipei Tu https://ebookmeta.com/product/the-new-civilization-upon-data-how- big-data-reshapes-human-civilization-business-and-the-personal- world-1st-edition-zipei-tu/
- 4. Insight Guides Belize Travel Guide eBook 6th Edition Insight Guides https://ebookmeta.com/product/insight-guides-belize-travel-guide- ebook-6th-edition-insight-guides/
- 7. COMPUTATIONAL SYSTEMS BIOLOGY SECOND EDITION Edited by Roland Eils Andres Kriete AMSTERDAM • BOSTON • HEIDELBERG • LONDON NEW YORK • OXFORD • PARIS • SAN DIEGO SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO Academic Press is an imprint of Elsevier
- 8. Academic Press is an imprint of Elsevier 525 B Street, Suite 1800, San Diego, CA 92101-4495, USA 225 Wyman Street, Waltham, MA 02451, USA The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK 32 Jamestown Road, London, NW1 7BY, UK Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands Copyright © 2014, 2006 Elsevier, Inc. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher. Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: permissions@elsevier.com. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier. com/locate/permissions, and selecting Obtaining permission to use Elsevier material. Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. Library of Congress Cataloging-in-Publication Data Computational systems biology (Kriete) Computational systems biology / edited by Andres Kriete, Roland Eils. -- Second edition. p. ; cm. Includes bibliographical references and indexes. ISBN 978-0-12-405926-9 (alk. paper) I. Kriete, Andres, editor of compilation. II. Eils, Roland, editor of compilation. III. Title. [DNLM: 1. Computational Biology. 2. Systems Biology. QU 26.5] QH324.2 570.1’13--dc23 2013045039 British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. ISBN: 978-0-12-405926-9 For information on all Academic Press publications visit our website at store.elsevier.com Printed in the United States of America 14 15 10 9 8 7 6 5 4 3 2 1
- 9. ix Frédéric Crémazy Department of Synthetic Systems Biology and Nuclear Organization, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, The Netherlands Matteo Barberis Department of Synthetic Systems Biology and Nuclear Organization, Swam merdam Institute for Life Sciences, University of Amsterdam, Amsterdam, The Netherlands Chapter 4 Ursula Klingmüller, Marcel Schilling, Sonja Depner, Lorenza A. D’Alessandro Division Systems Biology of Signal Transduction, German Cancer Research Center (DKFZ), Heidelberg, Germany Chapter 5 Christina Kiel EMBL/CRG Systems Biology Research Unit, Centre for Genomic Regulation (CRG), Barcelona, Spain Universitat Pompeu Fabra (UPF), Barcelona, Spain Luis Serrano EMBL/CRG Systems Biology Re search Unit, Centre for Genomic Regulation (CRG), Barcelona, Spain Universitat Pompeu Fabra (UPF), Barcelona, Spain ICREA, Barcelona, Spain Chapter 6 Seiya Imoto Human Genome Center, Institute of Medical Science, The University of Tokyo, Minatoku, Tokyo, Japan Hiroshi Matsuno Faculty of Science, Yamaguchi University, Yoshida, Yamaguchi, Japan Satoru Miyano Human Genome Center, Insti tute of Medical Science, The University of Tokyo, Minatoku, Tokyo, Japan Chapter 1 Roland Eils Division of Theoretical Bio informatics(B080),GermanCancerResearch Center (DKFZ), Heidelberg, Germany Department for Bioinformatics and Func tional Genomics, Institute for Pharmacy and Molecular Biotechnology (IPMB) and BioQuant, Heidelberg University, Heidelberg, Germany Andres Kriete School of Biomedical Engineering, Science and Health Systems, Drexel University, Philadelphia, PA, USA Chapter 2 Robert B. Russell, Gordana Apic, Olga Kalinina, Leonardo Trabuco, Matthew J. Betts, Qianhao Lu CellNetworks, University of Heidelberg, Heidelberg, Germany Chapter 3 Hans V. Westerhoff Department of Synthetic Systems Biology and Nuclear Organization, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, The Netherlands Department of Molecular Cell Physiology, Faculty of Earth and Life Sciences, VU University Amsterdam, The Netherlands Manchester Centre for Integrative Systems Bio logy (MCISB), Manchester, UK Fei He Manchester Centre for Integrative Systems Biology (MCISB), Manchester, UK Department of Automatic Control and systems Engineering, The University of Sheffield, Sheffield, UK EttoreMurabito ManchesterCentreforIntegrative Systems Biology (MCISB), Manchester, UK Contributors
- 10. x CONTRIBUTORS Chapter 11 Reinhard Laubenbacher Virginia Bioinformatics Institute, Virginia Tech, Blacksburg VA, USA Pedro Mendes Virginia Bioinformatics Institute, Virginia Tech, Blacksburg VA, USA School of Computer Science, The University of Manchester, Manchester, UK Chapter 12 Joseph Xu Zhou, Xiaojie Qiu, Aymeric Fouquier d’Herouel, Sui Huang Institute for Systems Biology, Seattle, WA, USA Chapter 13 John Cole, Mike J. Hallock, Piyush Labhsetwar, Joseph R. Peterson, John E. Stone, Zaida Luthey-Schulten University of Illinois at Urbana-Champaign, USA Chapter 14 Jean-Luc Bouchot Department of Mathematics, Drexel University, PA, Philadelphia, USA William L. Trimble Institute for Genomics and SystemsBiology,ArgonneNationalLaboratory, University of Chicago, Chicago, IL, USA Gregory Ditzler Department of Electrical and Computer Engineering, Drexel University, PA, Philadelphia, USA Yemin Lan School of Biomedical Engineering, Science and Health, Drexel University, PA, Philadelphia, USA Steve Essinger Department of Electrical and Com puter Engineering, Drexel University, PA, Philadelphia, USA Gail Rosen Department of Electrical and Com puter Engineering, Drexel University, PA, Philadelphia, USA Chapter 15 Helder I Nakaya Department of Pathology, Emory University, Atlanta, GA, USA Vaccine Research Center, Emory University, Atlanta, GA, USA Chapter 7 Hong-Wu Ma Tianjin Institute of Industrial Bio technology, Chinese Academy of Sciences, Tianjin, P.R. China School of Informatics, University of Edinburgh, Edinburgh, UK An-Ping Zeng Institute of Bioprocess and Bio systems Engineering, Hamburg University of Technology, Denickestrasse, Germany Chapter 8 Stanley Gu Department of Bioengineering, Uni versity of Washington, Seattle, WA, USA Herbert Sauro Department of Bioengineering, University of Washington, Seattle, WA, USA Chapter 9 Juergen Eils Division of Theoretical Bioinformat ics, German Cancer Research Center (DKFZ), Heidelberg, Germany Elena Herzog Division of Theoretical Bioinformat ics, German Cancer Research Center (DKFZ), Heidelberg, Germany Baerbel Felder Division of Theoretical Bioinforma tics, German Cancer Research Center (DKFZ), Heidelberg, Germany Department for Bioin formatics and Functional Genomics, Institute for Pharmacy and MolecularBiotechnology(IPMB)andBioQuant, Heidelberg University, Heidelberg, Germany Christian Lawerenz Division of Theoretical Bio informatics, German Cancer Research Center (DKFZ), Heidelberg, Germany Roland Eils Division of Theoretical Bioinformat ics, German Cancer Research Center (DKFZ), Heidelberg, Germany Department for Bioinformatics and Functional Genomics, Institute for Pharmacy and Molec ular Biotechnology (IPMB) and BioQuant, Heidelberg University, Heidelberg, Germany Chapter 10 Jean-Christophe Leloup, Didier Gonze, Albert Goldbeter UnitédeChronobiologiethéorique, Faculté des Sciences, Université Libre de Bru xelles, Campus Plaine, Brussels, Belgium x
- 11. xi CONTRIBUTORS Chapter 18 Hang Chang, Gerald V. Fontenay, Cemal Bilgin, Bahram Parvin Life Sciences Division, Law rence Berkeley National Laboratory, Berkeley, CA, USA Alexander Borowsky Center for Comparative Medicine, University of California, Davis, CA, USA. Paul Spellman Department of Biomedical Engi neering, Oregon Health Sciences Univer sity, Portland, Oregon, USA Chapter 19 Stefan M. Kallenberger Department for Bio informatics and Functional Genomics, Divi sion of Theoretical Bioinformatics, German Cancer Research Center (DKFZ), Institute for Pharmacy and Molecular Biotechnology (IPMB) and BioQuant, Heidelberg University, Heidelberg, Germany Stefan Legewie Institute of Molecular Biology, Mainz, Germany Roland Eils Department for Bioinformatics and Functional Genomics, Division of Theoretical Bioinformatics, German Cancer Research Center (DKFZ), Institute for Pharmacy and Molecular Biotechnology (IPMB) and Bio Quant, Heidelberg University, Heidelberg, Germany Department of Clinical Analyses and Toxicology, University of Sao Paulo, Sao Paulo, SP, Brazil Chapter 16 Julien Delile Institut des Systèmes Complexes Paris Ile-de-France (ISC-PIF), CNRS, Paris, France Neurobiology and Development Lab, Terrasse, Gif-sur-Yvette Cedex, France René Doursat Institut des Systèmes Com plexes Paris Ile-de-France (ISC-PIF), CNRS, Paris, France School of Biomedical Engineering, Drexel Uni versity, Philadelphia, PA, USA Nadine Peyriéras Neurobiology and Develop ment Lab, Terrasse, Gif-sur-Yvette Cedex, France Chapter 17 Andres Kriete School of Biomedical Engineering, Science and Health Systems, Drexel Univer sity, Bossone Research Center, Philadelphia, PA, USA Mathieu Cloutier GERAD and Department of Chemical Engineering, Ecole Polytechnique de Montreal, Montreal, QC, Canada xi
- 12. xiii in this area. If compared to the first edition published in 2005, the second edition has been specifically extended to reflect new frontiers of systems biology, including modeling of whole cells, studies of embryonic development, the immune systems, as well as aging and cancer. As in the previous edition, basics of informa- tion and data integration technologies, standards, modeling of gene, signaling and metabolic networks remain comprehensively covered. Contributions have been selected and compiled to introduce the different meth- ods, including methods dissecting biological complexity, modeling of dynamical proper- ties, and biocomputational perspectives. Beside the primary authors and their respective teams who have dedicated their time to contribute to this book, the editors would like to thank numerous reviewers of individual chapters, but in particular Jan Eufinger for support of the editorial work. It is often mentioned that biological sys- tems in its entirety present more than a sum of its parts. To this extent, we hope that the chapters selected for this book not only give a contemporary and comprehensive over- look about the recent developments, but that this volume advances the field and encour- ages new strategies, interdisciplinary coop- eration, and research activities. Roland Eils and Andres Kriete Heidelberg and Philadelphia, September 2013 Computational systems biology, a term coined by Kitano in 2002, is a field that aims at a system-level understanding by modeling and analyzing biological data using computation. It is increasingly recognized that living system cannot be understood by studying individual parts, while the list of molecular components in biology is ever growing, accelerated by genome sequencing and high-throughput omics techniques. Under the guiding vision of systems biology, sophisticated computational methods help to study the interconnection of parts in order to unravel complex and net- worked biological phenomena, from protein interactions, pathways, networks, to whole cells and multicellular complexes. Rather than performing experimental observations alone, systems biology generates knowledge and understanding by entering a cycle of model construction, quantitative simulations, and experimental validation of model predic- tions, whereby a formal reasoning becomes key. This requires a collaborative input of experimental and theoretical biologists work- ing together with system analysts, computer scientists, mathematicians, bioengineers, physicists, as well as physicians to contend creatively with the hierarchical and nonlinear nature of cellular systems. This book has a distinct focus on computa- tional and engineering methods related to sys- tems biology. As such, it presents a timely, multi-authored compendium representing state-of-the-art computational technologies, standards, concepts, and methods developed Preface
- 13. 1 © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/B978-0-12-405926-9.00001-0 Computational Systems Biology, Second Edition 1 Introducing Computational Systems Biology Roland Eilsa,b , Andres Krietec a Division of Theoretical Bioinformatics (B080), German Cancer Research Center (DKFZ), Heidelberg, Germany b Department for Bioinformatics and Functional Genomics, Institute for Pharmacy and Molecular Biotechnology (IPMB) and BioQuant, Heidelberg University, Heidelberg, Germany c School of Biomedical Engineering, Science and Health Systems, Drexel University, Philadelphia, PA, USA C H A P T E R C O N T E N T S 1 Prologue 1 2 Overview of the content 4 3 Outlook 6 References 7 We need to turn data into knowledge and we need a framework to do so. S. Brenner, 2002. 1 PROLOGUE The multitude of the computational tools needed for systems biology research can roughly be classified into two categories: system identification and behavior analysis (Kitano 2001). In molecular biology, system identification amounts to identifying the regulatory relation- ships between genes, proteins, and small molecules, as well as their inherent dynamics hid- den in the specific kinetic and binding parameters. System identification is arguably one of the most complicated problems in science. While behavior analysis is solely performed on a model, model construction is a process tightly connected to reality but part of an iterative process between data analysis, simulation, and experimental validation (Figure 1.1). A typical
- 14. 2 1. Introducing Computational Systems Biology modeling cycle begins with a reductionist approach, creating the simplest possible model. The modeling process generates an understanding of the underlying structures, and components are represented graphically with increasing level of formalization, until they can be converted into a mathematical representation. The minimal model then grows in complexity, driven by new hypotheses that may not have been apparent from the phenomenological descriptions. Then, an experiment is designed using the biological system to test whether the model predic- tions agree with the experimental observations of the system behavior. The constitutive model parameters may be measured directly or may be inferred during this validation process, how- ever, the propagation of errors through these parameters present significant challenges for the modeler. If data and predictions agree, a new experiment is designed and performed. This pro- cess continues until sufficient experimental evidence in favor of the model is collected. Once the system has been identified and a model constructed, the system behavior can be studied, for instance, by numerical integration or sensitivity analysis against external perturbations. Although the iterative process is well defined, the amount of data to be merged into this process can be immense. The human genome project is one of the hallmarks indicating a turn from a reductionistic approach in studying biological systems at increasing level, into a dis- covery process using high-throughput techniques (Figure 1.2). Ongoing research increases the wealth of contemporary biological information residing in some thousand public databases providing descriptive genomics, proteomics and enzyme information, gene expression, gene variants and gene ontologies. Refined explorative tools, such as new deep sequencing, along with the emergence of new specialized -omics (metabolomics, lipidomics, pharmacogenom- ics) and phenotyping techniques, constantly feed into this data pool and accelerate its growth. Given the enormous and heterogeneous amount of data, computational tools have become indispensable to mine, analyze, and connect such information. The aggregate of statistical FIGURE 1.1 Key to systems biology is an iterative cycle of experimentation, model building, simulation and validation.
- 15. 1 Prologue 3 bioinformatics tools to collect, store, retrieve, visualize, and analyze complex biological data has repeatedly proven useful in biological decision support and discovery. Deciphering the basic building blocks of life is a necessary step in biological research, but provides only lim- ited knowledge in terms of understanding and predictability. In the early stages the human genome project stirred the public expectation for a rapid increase in the deciphering of dis- ease mechanisms, more effective drug development and cure. However, it is well recognized that the battery of mechanisms involved in the proliferation of complex diseases like cancer, chronic diseases, or the development of dementias cannot be understood solely on the basis of knowing all its molecular components. As a consequence, a lack of system level understanding of cellular dynamics has prevented a substantial increase in the number of new drugs available for treatment, drug efficacy, or eradication of any specific diseases. In contrast, pharmaceutical companies are currently lack- ing criteria to select the most valuable targets, R&D expenses skyrocket, and new drugs rarely hit the market and often fail in clinical trials, while physicians face an increasing wealth of information that needs to be interpreted intelligently and holistically. Analysis of this dilemma reveals primary difficulties due to the enormous biomolecular complexity, structural and functional unknowns in a large portion of gene products and a lack of understanding of how the concert of molecular activities transfers into physiological alterations and disease. It has been long recognized that the understanding of cells as open systems, interacting with the environment, performing tasks and sustain homeostasis, or bet- ter homeodynamics (Yates 1992), requires the development of foundations for a general sys- tems theory that started with the seminal work of Bertalanffy (Von Bertalanffy 1969). FIGURE 1.2 By the evolution of scientific disciplines in biology over time, ever-smaller structures have come into focus and more detailed questions have been asked. With the availability of high-throughput sequencing techniques in genetics a turning point was reached at the molecular basis of life. The frontiers of research extended to hypothesis- free data acquisition of biological entities, with genomics becoming the first in a growing series of “-omics” disci- plines. Although functional genomics and proteomics are far from being completed, “omics” -type approaches addressing the phenotypical cellular, tissue and physiological levels constitute themselves as new scientific disci- plines, filling up an otherwise sparse data space. Computational systems biology provides methodologies to com- bine, model, and simulate entities on diverse (horizontal) levels of biological organization, such as gene regulatory and protein networks, and between these levels by using multiscale (vertical) approaches.
- 16. 4 1. Introducing Computational Systems Biology It appears that with the ever increasing quality and quantity of molecular data, mathematical models of biological processes are even more in demand. For instance, an envisioned blue- print of complex diseases will not solely consist of descriptive flowcharts as widely found in scientific literature or in genomic databases. They should rather be based on predictive, rigor- ously quantitative data-based mathematical models of metabolic pathways, signal transduc- tion cascades, cell-cell communication, etc. The general focus of biomedical research on complex diseases needs to change from a primarily steady-state analysis at the molecular level to a systems biology level capturing the characteristic dynamic behavior. Such biosimu- lation concepts will continue to transform current diagnostic and therapeutic approaches to medicine. 2 OVERVIEW OF THE CONTENT This completely revised, second edition of this book presents examples selected from an increasingly diverse field of activities, covering basic key methods, development of tools, and recent applications in many complex areas of computational systems biology. In the follow- ing, we will broadly review the content of the chapters as they appear in this book, along with specific introductions and outlooks. The first section of this book introduces essential foundations of systems biology, princi- ples of network reconstruction based on high-throughput data with the help of engineering principles such as control theory. Robert B. Russell, Gordana Apic, Olga Kalinina, Leonardo Trabuco, Matthew J. Betts, and Qianhao Lu provide an introduction (Chapter 2) on “Structural Systems Biology: modeling interactions and networks for systems studies.” Molecular mechanisms provide the most detailed level for a mechanistic understanding of biological complexity. The current challenges of a structural systems biology are to integrate, utilize, and extend such knowledge in conjunction with high-throughput studies. Understanding the mechanistic consequences of multiple alterations in DNA variants, protein structures, and folding are key tasks of structural bioinformatics. Principles of protein interactions in pathways and networks are introduced by Hans V. Westerhoff, Fei He, Ettore Murabito, Frédéric Crémazy, and Matteo Barberis in Chapter 3. Their contribution is entitled “Understanding principles of the dynamic biochemical networks of life through systems biology” and discusses a number of basic, more recent and upcoming discover- ies of network principles. The contributors review analytical procedures from flux balance in metabolic networks to measures of robustness. In Chapter 4, Ursula Klingmüller, Marcel Schilling, Sonja Depner, and Lorenza A. D‘Alessandro review the “Biological foundations of signal transduction and aberrations in disease.” Signaling pathways process the external signals through complex cellular networks that reg- ulate biological functions in a context-dependent manner. The authors identify the underly- ing biological mechanisms influential for signal transduction and introduce the mathematical tools essential to model signaling pathways and their disease aberrations in a quantitative fashion. Further acceleration of progress in pathway reconstruction and analysis is contingent on the solution of many complexities and new requirements, revolving around the question of how high-throughput experimental techniques can help to accelerate reconstruction and
- 17. 2 Overview of the content 5 simulation of signaling pathways. This is the theme of the review in Chapter 5 by Christina Kiel and Luis Serrano on the “Complexities underlying a quantitative systems analysis of signaling networks.” Chapter 6 by Seiya Imoto, Hiroshi Matsuno, Satoru Miyano presents “Gene net- works: estimation, modeling and simulation.” The authors describe how gene networks can be reconstructed from microarray gene expression data, which is a contemporary problem. They also introduce software tools for modeling and simulating gene networks, which is based on the concept of Petri nets. The authors demonstrate the utility for the modeling and simulation of the gene network for controlling circadian rhythms. Section 2 provides an overview of methods, mathematical tools, and examples for model- ing approaches of dynamic systems. “Standards, platforms, and applications,” as presented by Herbert Sauro and Stanley Gu in Chapter 8, reviews the trends in developing standards indic- ative of increasing cooperation within the systems biology community, which emerged in recent years permitting collaborative projects and exchange of models between different soft- ware tools. “Databases for systems biology,” as reviewed in Chapter 9 by Juergen Eils, Elena Herzog, Baerbel Felder, Christian Lawerenz and Roland Eils provide approaches to integrate information about the responses of biological system to genetic or environmental perturba- tions. As researchers try to solve biological problems at the level of entire systems, the very nature of this approach requires the integration of highly divergent data types, and a tight coupling of three general areas of data generated in systems biology: experimental data, ele- ments of biological systems, and mathematical models with the derived simulations. Chapter 10 builds on a classical mathematical modeling approach to study patterns of dynamic behav- iors in biological systems. “Computational models for circadian rhythms - deterministic versus sto- chastic approaches,” Jean-Christophe Leloup, Didier Gonze and Albert Goldbeter demonstrates how feedback loops give rise to oscillatory behavior and how several results can be obtained in models which possess a minimum degree of complexity. Circadian rhythms provide a par- ticular interesting case-study for showing how computational models can be used to address a wide range of issues extending from molecular mechanism to physiological disorders. Reinhard Laubenbacher and Pedro Mendes review “Top-down dynamical modeling of molecu- lar regulatory networks,” Chapter 11. The modeling framework discussed in this chapter con- siders mathematical methods addressing time-discrete dynamical systems over a finite state set applied to decipher gene regulatory networks from experimental data sets. The assump- tions of final systems states are not only a useful modeling concept, but also serve an explana- tion of fundamental organization of cellular complexities. Chapter 12, entitled “Multistability and multicellularity: cell fates as high-dimensional attractors of gene regulatory networks,” by Joseph X. Zhou and Sui Huang, investigates how the high number of combinatorially possible expression configurations collapses into a few configurations characteristic of observable cell fates. These fates are proposed to be high-dimensional attractors in gene activity state space, and may help to achieve one of the most desirable goal of computational systems biology, which is the development of whole cell models. In Chapter 13 John Cole, Mike J. Hallock, Piyush Labhsetwar, Joseph R. Peterson, John E. Stone, and Zaida Luthey-Schulten review “Whole cell modeling strategies for single cells and microbial colonies,” taking into account spatial and time-related heterogeneities such as short-term and long-term stochastic fluctuations. Section 3 of this book is dedicated to emerging systems biology application including mod- eling of complex systems and phenotypes in development, aging, health, and disease. In Chapter 14, Jean-Luc Bouchot, William Trimble, Gregory Ditzler, Yemin Lan, Steve Essinger,
- 18. 6 1. Introducing Computational Systems Biology and Gail Rosen introduce “Advances in machine learning for processing and comparison of metage- nomic data.” The study of nucleic acid samples from different parts of the environment, reflect- ing the microbiome, has strongly developed in the last years and has become one of the sustained biocomputational endeavors. Identification, classification, and visualization via sophisticated computational methods are indispensable in this area. Similarly, the decipher- ing immune system has to deal with a large amount of data generated from high-throughput techniques reflecting the inherent complexity of the immune system. Helder I. Nakaya, in Chapter 15, reports on “Applying systems biology to understand the immune response to infection and vaccination.” This chapter highlights recent advances and shows how systems biology can be applied to unravel novel key molecular mechanisms of immunity. Rene Doursat, Julien Delile, and Nadine Peyrieras present “Cell behavior to tissue deforma- tion: computational modeling and simulation of early animal embryogenesis,” Chapter 16. They pro- pose a theoretical, yet realistic agent-based model and simulation platform of animal embryogenesis, to study the dynamics on multiple levels of biological organization. This con- tribution is an example demonstrating the value of systems biology in integrating the differ- ent phenomena involved to study complex biological process. In Chapter 17, Andres Kriete and Mathieu Cloutier present “Developing a systems biology of aging.” The contribution reviews modeling of proximal mechanisms of aging occurring in pathways, networks, and multicel- lular systems, as demonstrated for Parkinson’s disease. In addition, the authors reflect on evolutionary aspect of aging as a robustness tradeoff in complex biological designs. In Chapter 18, Hang Chang, Gerald V Fontenay, Ju Han, Nandita Nayak, Alexander Borowsky, Paul Spellman, and Bahram Parvin present image-based phenotyping strategies to classify cancer phenotypes on the tissue level, entitled “Morphometric analysis of tissue het- erogeneity in Glioblastoma Multiforme.” Such work allows to associate morphological heteroge- neities of cancer subtypes with molecular information to improve prognosis. In terms of a multiscale modeling approach the assessment of phenotypical changes, in cancer as well as in other diseases, will help to build bridges toward new spatiotemporal modeling approaches. Stefan M. Kallenberger, Stefan Legewie, and Roland Eils demonstrate “Applications in cancer research: mathematical models of apoptosis” in Chapter 19. Their contribution is focused on the mathematical modeling of cell fate decisions and its dysregulation of cell death, contributing to one of the ramifications of the complexities in cancer biology. 3 OUTLOOK It is commonly recognized that biological multiplicity is due to progressive evolution that brought along an increasing complexity of cells and organisms over time (Adami et al. 2000). This judgement coincides with the notion that greater complexity is “better” in terms of com- plex adaptive systems and ability for self-organization, hence robustness (Csete and Doyle 2002 Kitano 2004). Analyzing or “reverse” engineering of this complexity and integrating results of today’s scientific technologies responsible for the ubiquitous data overload are an essential part of systems biology. The goals are to conceptualize, abstract basic principles, and model biological structures from molecular to higher level of organization like cells, tissues, and organs, in order to provide insight and knowledge. The initial transition requires data
- 19. REFERENCES 7 cleansing and data coherency, but turning information into knowledge requires interpret- ing what the data actually means. Systems biology addresses this need by the development and analysis of high-resolution quantitative models that recapitulate, but more importantly predict cellular behavior in time and space and to determine physiology from the underlying molecular and cellular capacities on a multiscale (Dada and Mendes 2011). Once established, such models are indicators to the detailed understanding of biological function, the diagnosis of diseases, the identification and validation of therapeutic targets, and the design of drugs and drug therapies. Experimental techniques yielding quantitative genomic, proteomic, and metabolomic data needed for the development of such models are becoming increasingly common. Computer representations describing the underlying mechanisms may not always be able to provide complete accuracy due to limited computational, experimental, and methodical resources. Increase in data quality and coherence, availability within integrated databases or approaches that can manage experimental variability, are less considered but may be as essen- tial for robust growth of biological knowledge. Still, the enormous complexity of biological systems has given rise to additional cautionary remarks. First, it may well be that our models and future super-models correctly predict experimental observations, but may still prevent a deeper understanding due to complexities, non-linearities, or stochastic phenomena. This notion may initially sound quite disappointing, but is a daily experience of all those who employ modeling and simulations of large-scale phenomena. Yet, it shows the relevance of computational approaches in this area, and suggestions to link biological with computational problem solving has been suggested (Navlakha and Bar-Joseph 2011). Systems biology should follow strict standards and conventions, and progress in theory and computational approaches will always demand new models that can provide new insights if applied to an existing body of information. Many areas, including cancer model- ing, have demonstrated how models evolve over many cycles of investigation and refinement (Byrne 2010). Once established, new models can be reimplemented into existing platforms to be more broadly available. In the long run, the aim is to develop user-friendly, scalable and open-ended platforms that also handle methods for behavior analysis and model-based dis- ease diagnosis, and support scientists in their every-day practice of decision-making and bio- logical inquiry, as well as physicians in clinical decision support. Systems biology has risen out of consensus in the scientific community, initially driven by visionary scientific entrepreneurs. Now, as its strength becomes obvious, it is recognized as a rapidly evolving mainstream endeavor, which requires specific educational curricula and col- laboration among computational scientists, experimental and theoretical biologists, control and systems engineers, as well as practitioners in drug development and clinical research. These collaborative ties will move this field forwards toward a formal, quantitative, and pre- dictive framework of biology. References Adami, C., Ofria, C., and Collier, T. C. (2000). Evolution of biological complexity. Proc Natl Acad Sci USA 97:4463–4468. Byrne, H. M. (2010). Dissecting cancer through mathematics: From the cell to the animal model. Nat Rev Cancer 10:221–230.
- 20. 8 1. Introducing Computational Systems Biology Csete, M. E., and Doyle, J. C. (2002). Reverse engineering of biological complexity. Science 295:1664–1669. Dada, J. O., and Mendes, P. (2011). Multi-scale modelling and simulation in systems biology. Integr Biol (Camb) 3:86–96. Kitano, H. (2001). Foundations of Systems Biology. MIT-Press. Kitano, H. (2004). Biological Robustness. Nat Rev Genet 5:826–837. Navlakha, S., and Bar-Joseph, Z. (2011). Algorithms in nature: The convergence of systems biology and computa- tional thinking. Mol Syst Biol 7:546. Von Bertalanffy, L. (1969). General Systems Theory. George Brazillar Inc. Yates, F. E. (1992). Order and complexity in dynamical systems: Homeodynamics as a generalized mechanics for biol- ogy. Math and Comput Model 19:49–74.
- 21. 9 © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/B978-0-12-405926-9.00002-2 Computational Systems Biology, Second Edition 2 Structural Systems Biology: Modeling Interactions and Networks for Systems Studies Robert B. Russell, Gordana Apic, Olga Kalinina, Leonardo Trabuco, Matthew J. Betts, Qianhao Lu CellNetworks, University of Heidelberg, Heidelberg, Germany C H A P T E R C o n t e n t s 1 Introduction 10 2 A brief history of structural bioinformatics 10 3 Structural analysis of interaction data 11 4 Other interaction types 13 5 Systems biology applications 13 6 New datasets-specific protein sites 14 7 Current and future needs 14 8 Concluding remarks 16 References 16 Abstract The best understanding of complex biological systems ultimately comes from details of the underlying atomic structures within it. In the absence of known structures of all protein complexes and interactions in a system, structural bioinformatics or modeling fill an important niche in providing predicted mechanistic information which can guide experiments, aid the interpretation of high-throughput datasets and help provide key details to model biological systems. This introductory review discusses the current state of this field and suggests how current datasets in systems studies can profit from a better integration of predicted or known structural information.
- 22. 10 2. STRUCTURAL SYSTEMS BIOLOGY: MODELING INTERACTIONS AND NETWORKS 1 INTRODUCTION We are clearly today in the era of high-throughput biology. In every area of biology— from plant sciences to human health—one increasingly sees systematic screens that identify hundreds or thousands of molecules regulated or changed in response to some stimulus or perturbation. More than ever there is a need to understand what such large sets of molecules mean when identified together in terms of system functions, and to use these data to suggest therapies, vaccines, diagnostics, herbicides, etc. Invariably scientists wish to use the results from high-throughput experiments to unlock the underlying biological mechanism. The molecular mechanism—in the broadest sense— ultimately provides the details that give a deeper understanding of a biological process, or suggest means to perturb a system with small molecules or other agents. To address this, many efforts have been undertaken to capture systematically all of the mechanistic detail that has been captured by low-throughput experiments in the past decades. Pathway resources such as KEGG (Kotera et al. 2012) or Reactome (Croft et al. 2011) and ontological tools such as GO (Gene Ontology Consortium 2006) provide a means to state for a large set of genes, pro- teins, or metabolites which processes are likely being affected. These tools remain central to most high-throughput studies. However, the ultimate understanding of a biological process comes only from a view of the actual molecular details underlying it. Specifically, the availability of multiple three-dimen- sional (3D) structures provides information down to the specific atoms involved in a process. Today, thanks to more than a decade of Structural Genomics driven advances in structure determination by X-ray, NMR, and electron microscopy, there are structural representatives for almost every globular domain, and the number of multi-protein complexes of known structures is also growing at an impressive rate. Concurrent advances in techniques to model protein structures by homology also means that increasingly accurate modeled structures are readily available for at least globular parts of most proteins of interest. There are also many tools for interrogating proteins structurally and increasingly these are addressing the needs of the high-throughput biologist. This chapter discusses recent advances in this broad area of Structural Systems Biology and Bioinformatics, and suggests future directions to meet new challenges of high-throughput biology. 2 A BRIEF HISTORY OF STRUCTURAL BIOINFORMATICS Structural Bioinformatics began with the first attempts to study and predict protein struc- tures (Blundell et al. 1987). While structure and sequences databases were small, the pri- mary focus was the grand challenge to predict protein 3D structures from primary sequences. Methods to predict protein secondary structure or 3D structure were approached by a variety of informatics-or physics-based methods, and had mixed success until the arrival of system- atic community wide assessment exercises (Critical Assessment of Structure Prediction, CASP (Moult et al. 2011)) where double-blind assessments of predictions (i.e. where the structures were unknown to both predictors and experimentalists during the predictions). These experi- ments identified the strengths and weaknesses of all approaches and ultimately have led to mature methods to predict secondary structure and tertiary structure either de novo or via homology modeling techniques. Today models for virtually all proteins that are modelable
- 23. 3 Structural analysis of interaction data 11 are now systematically available via online databases such as ModBase (Pieper et al. 2011) and Swissmodel (Kiefer et al. 2009). Structural bioinformatics now often focuses on methods that predict function of individual proteins of known structure, rather than methods that pre- dict structure per se. For instance, numerous methods have been developed to study protein surfaces to predict functional sites using a variety of geometrical or evolutionary criteria (e.g. Aloy et al. 2001; Capra et al. 2009; Casari et al. 1995; Landgraf et al. 2001; Wilkins et al. 2012; Yang et al. 2012). The initial genome sequencing projects produced the first large sets of genes and encoded proteins for which little information was available. Structural bioinformatics played a crucial role in identifying overall features of the genome in terms of domain distributions and com- binations (e.g. Apic et al. 2001; Gerstein and Levitt 1997), a process that was greatly aided by the availability of structure classification databases (Andreeva et al. 2008; Cuff et al. 2009; Holm and Rosenström 2010). These analyses ultimately matured and were incorporated into the protein databases used today, such as Pfam (Punta et al. 2012) and CDD (Marchler-Bauer et al. 2013) and are readily visible in primary databases such as Uniprot (Wu et al. 2006) or Refseq (Pruitt et al. 2005). 3 STRUCTURAL ANALYSIS OF INTERACTION DATA The arrival of various interaction datasets produced a new challenge for computational structural biologists. Suddenly thousands of new interactions and complexes became known with little or no structural information available. Modeling interactions is, of course, possible if one has a suitable template of known structure containing two or more interacting proteins in contact. However, early analyses of interaction data from a structural perspective high- lighted the relative paucity of these interaction templates (Aloy and Russell 2002). Indeed, while solving structures for single, small, globular proteins are now a relatively straightfor- ward process, solving experimental structures involving multiple proteins continues to be a challenge. Nevertheless, improved experimental techniques, and the increased focus on studying protein complexes in structural biology, means that there is now an exponential growth in the number of distinct interactions of known structure (Aloy and Russell 2004; Kim et al. 2006b; Tuncbag et al. 2008). There are currently several tools that allow biologists to study interactions in three-dimen- sions. Early tools such as InterPReTS (Aloy and Russell 2003) and MULTIPROSPECTOR (Lu et al. 2002) were designed to rapidly assess how well homologous sequences fit onto interact- ing proteins with 3D structures. Systematic analysis of thousands of interactions of known 3D structure showed that sequence similar proteins retain similar interactions, and a drop in sequence similarity increases a tendency to interact differently (Aloy and Russell 2004; Kim et al. 2006b; Tuncbag et al. 2008). Analyses also showed that structural interfaces (Figure 2.1) could be used to infer details about whether or not interactions could occur simul- taneously (Aloy and Russell 2006; Kim et al. 2006a) which helped the classification of protein interaction centers in terms of “party” or “date” hubs (Han et al. 2004). However, interroga- tion of interaction sources showed that the picture for many promiscuous proteins (in terms of interactions) is more complicated, with many having the ability to interact with multiple partners and multiple interfaces (e.g. Figure 2.2). This early work has since led to a number of databases that allow users to query interactions of known 3D structure, including 3DID (Stein
- 24. 12 2. STRUCTURAL SYSTEMS BIOLOGY: MODELING INTERACTIONS AND NETWORKS RAS RAS SOS1 RASSF5 RAFRBD RAS CDK6 CDK6 v-Cyclin P18-INK4C FIGURE 2.1 Structural interfaces can be used to assess whether interactions between proteins can occur simulta- neously. The top of the figure shows a schematic of a protein (hexagon) that can either bind multiple proteins at one interface or simultaneously via different interfaces and that this is not obvious when looking at interaction networks alone. The bottom left of the figure shows three structures of Ras or Ras-like proteins in complex with three structur- ally different proteins that all bind on the same interface; the bottom right shows how the CDK6 structure can accom- modate interactions with three proteins simultaneously. Interface 1: 17 interactors Interface 0: 14 interactors Interface 4: 3 interactors Interface 2: 7 interactors Interface 5: 6 interactors Interface 3: 7 interactors p115 FIGURE 2.2 An example of a highly promiscuous protein (p117) uncovered during a screen of interactions within Mycoplasma pneumoniae (Kühner et al. 2009) and how it can apparently interact with multiple partners on multiple interfaces as predicted by interface modeling techniques. Protein p117 is colored in gray with the interaction partners in other colors. The number of interactors given for each interface is taken from the TAP dataset generated in the same screen.
- 25. 5 Systems biology applications 13 et al. 2011), SCOPPI (Winter et al. 2006) and Interactome3D (Mosca et al. 2013). There have also been a number of applications of these tools to whole genomes to understand globally the structural repertoire of interactions and complexes present in an organism (Aloy et al. 2004; Kühner et al. 2009; Zhang et al. 2012) which has led to numerous insights into individ- ual complexes and the nature of protein-protein interactions in general. 4 OTHER INTERACTION TYPES Protein interactions come in many different flavors. Most of the above approaches work best when pairs of globular (i.e. folded) proteins or domains interact with one another. It has long been known that many interactions in biology do not occur in this way, but instead involve one globular protein or domain interacting with short peptide segments from other proteins. These peptide segments often show a particular pattern or motif that captures the features most responsible for binding to the globular partner. There are now several resources that capture these motifs systematically and allow users to search for motifs in query proteins (e.g. Dinkel et al. 2012). The fact that these motifs are more difficult to detect than globular segments using conventional sequence analysis tools (owing mostly to their short length) has led to various methods to identify new motif candidates (Davey et al. 2010; Neduva and Russell 2006) and most recently these approaches have been extended to methods to predict protein-peptide interactions using known 3D structures if available (Petsalaki et al. 2009). All of this work is complementary to earlier developments on protein-protein or protein- small-molecule docking. Whereas previous docking efforts were focused on individual pairs of proteins of interest, there are now a growing number of studies whereby hundreds or thou- sands of pairs of proteins are docked together in an attempt either to find a handful of likely biologically meaningful docked structures (Mosca et al. 2009) or to use docking as a means to predict protein-protein interactions (Wass et al. 2011). Other efforts have attempted to use docking to combine pairwise docking methods (i.e. that attempt to dock two proteins or domains together) model higher order complexes (Inbar et al. 2005; Lasker et al. 2009) that are known from protein complex discovery experiments (Gavin et al. 2006; Guruharsha et al. 2011). Protein-small-molecule docking is now applied in a systems-wide fashion. Specifically, virtual screening—whereby thousands of molecules can be docked simultaneously to one or often multiple proteins—is now commonplace and indeed a standard complementary approach to virtual screening (Lavecchia and Di Giovanni 2013). 5 SYSTEMS BIOLOGY APPLICATIONS Exciting applications of structural bioinformatics techniques to systems modelling are already emerging. Structures, for example, provide a means to provide critical missing parameters for metabolic modeling processes (Gabdoulline et al. 2007; Stein et al. 2007). On a large scale, struc- tures (experimental or modeled) can be used to identify missing substrates and products for metabolic reconstruction, which enables more accurate simulation and interpretation (Chang et al. 2013; Yus et al. 2009; Zhang et al. 2009). It is likely that these approaches will be applicable to more complex processes such as signaling or DNA repair in the future, but currently too little
- 26. 14 2. STRUCTURAL SYSTEMS BIOLOGY: MODELING INTERACTIONS AND NETWORKS structural information is available and there are additional challenges to be overcome, such as the ability to reliable estimate thermodynamic or kinetic parameters for protein-protein interac- tions. There are various hints that this will be possible, coming from several studies that attempt to predict interaction specificity across diverse sets of proteins such as (Kiel et al. 2008). 6 NEW DATASETS-SPECIFIC PROTEIN SITES With the advent of next generation sequencing thousands of new individual genomes of a species become available and these data are increasing at an explosive rate (Hanahan and Weinberg 2011; Xuan et al. 2012). While the previous goal was to understand the function of specific genomes or sets of proteins (i.e. a set of dysregulated genes or proteins), now one typically is presented with both a set of genes/proteins and multiple modifications within them. Therefore, these data can profit from computational predictions about the mechanistic consequences of alterations. Most tools for assessing DNA variations consider both protein sequence and structural information to some degree. For instance, tools like PolyPhen and MutationAssessor (Reva et al. 2011; Sunyaev et al. 2000) consider known or modeled struc- tures to assess whether a mutation or variant lies in the interior or at the surface of a protein which helps to suggest how deleterious the change is likely to be, and the latter considers additional contacts to small molecules or other proteins. General principles are also emerging, for example analysis of SNPs that lie within known or predicted 3D structures shows that they tend to be on protein surfaces and to lie at protein interaction interfaces (David et al. 2012). Other new datasets are also in need of the kind of mechanistic interpretation that struc tures can provide. Perhaps most significant among these are proteomics datasets related to the identification of post-translational modifications (PTMs) (Choudhary and Mann 2010; Pflieger et al. 2008). Here too the datasets consist of individual positions within hundreds or thousands of proteins that are often related to phenotypic differences or disease. Structural analyses of proteomic PTM datasets have found that these modifications too are enriched and protein-protein interfaces (van Noort et al. 2012) and that they show certain preferences according to type and that they tend to co-occur within interacting proteins (Minguez et al. 2012). 3D structures have also been suggested as a means to filter meaningful modification sites from possibly artifacts: it has been argued that sites known or predicted to be highly buried in a protein structure are less likely accessible to kinases and phosphatases and such sites likely need to be considered carefully in terms of their accuracy (Vandermarliere and Martens 2013). 7 CURRENT AND FUTURE NEEDS The unifying theme to both of these types of datasets is the need to first understand as much as possible about the mechanistic consequences of mutating or modifying a particular residue in a particular protein, and then, if possible, to identify from hundreds or thousands of data-points those that are most likely to have biological consequences. Thus, beyond the analysis of individual sites within large datasets, there is an increasing need to understand an entire set of genes, proteins, or their modifications in a kind of mechanistic context.
- 27. 7 Current and future needs 15 Q61R Q61R I36M I36M KRAS RASSF2 KRAS SOS1 FIGURE 2.3 View of KRAS mutations in terms of known or predicted interactions between functional elements. The top of the figure shows an interac- tion network of KRAS and a selection of interaction partners. The inset zooms in on interactions with KRAS and various Ras-binding domain proteins. Each protein is shown a series of domains (squares) and linear motifs (diamonds) connected N- to C-terminally. Boxes around regions of the protein denote regions of protein 3D structures that are either in contact with part of another protein in the network (darker or red lines) or with themselves (circular lines connecting individual proteins). Known interactions between linear motifs and domains are also shown as yellow/lighter lines. The location of two KRAS mutations at interaction interfaces are shown to the right of the figure (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this book.).
- 28. 16 2. STRUCTURAL SYSTEMS BIOLOGY: MODELING INTERACTIONS AND NETWORKS The combination of existing individual tools into a more systems-ready view of structural or mechanistic information appears to be a desirable development goal. For instance, consid- ering data on mutations in colorectal cancer (Kilpivaara and Aaltonen 2013) readily identifies sets of proteins of interest, such as KRAS (Lièvre et al. 2006), though mutations are also seen in many other cancers or developmental disorders. The STRING database (Franceschini et al. 2013) provides nine proteins that interact with KRAS (Figure 2.3). Like most eukaryotic pro- teins they are modular, consisting of several distinct modules, or domains with discrete func- tions and often with a discrete 3D structure. Considering predicted interactions via InterPReTS (Aloy and Russell 2002) and potential interactions between linear motifs (Dinkel et al. 2012) and protein domains (Punta et al. 2012) provides a set of potential interactions between these domains that provides various key insights about how KRAS interacts can interaction with its partners. For instance, structural analysis shows that it is unlikely that KRAS can interact with SOS1, RAF1, RALGDS, or RASSF2 at the same time as these interactions are predicted to occur that the same interface (Figure 2.1). The structures also suggest which of the known mutations within KRAS are likely to affect interactions (Figure 2.3 labeled) and how some of these interactions seem to contain mutations for cancer (Q61R) or NS2 (I36M). Analysis of individual KRAS modeled structures also helped to reveal how several key mutations affect nucleotide binding within the RAS domain (not shown). 8 CONCLUDING REMARKS Structural biology and structural bioinformatics have much to offer for systems-level studies. There is still a considerable gap between systems, assemblies or complexes that are understood in terms of their component molecules, but that lack most or all information about how the molecules come together at the atomic level or about the kinetic or thermo- dynamic parameters that are so important to model systems accurately. The ability to exploit and interpret known or predicted structural information quickly for these systems is of grow- ing importance as the datasets related to how these systems are modified either genetically or via PTMs grows. Tools and know how in structural bioinformatics thus provides a great boost to anybody wishing to understand molecular mechanism and how it can be perturbed by variation, modification, or the addition of other molecules. Acknowledgments This work was supported by the European Community’s Seventh Framework Programme FP7/2009 under the grant agreement no: 241955, SYSCILIA. References Aloy, P., and Russell, R. B. (2002). Interrogating protein interaction networks through structural biology. Proc. Nat. Acad. Sci. U.S.A. 99:5896–5901. Aloy, P., and Russell, R. B. (2003). InterPreTS: protein interaction prediction through tertiary structure. Bioinformatics 19:161–162. Aloy, P., and Russell, R. B. (2004). Ten thousand interactions for the molecular biologist. Nat. Biotechnol. 22:1317–1321.
- 29. REFERENCES 17 Aloy, P., and Russell, R. B. (2006). Structural systems biology: modelling protein interactions. Nat. Rev. Mol. Cell Biol. 7:188–197. Aloy, P., Querol, E., Aviles, F. X., and Sternberg, M. J. (2001). Automated structure-based prediction of functional sites in proteins: applications to assessing the validity of inheriting protein function from homology in genome anno- tation and to protein docking. J. Mol. Biol. 311:395–408. Aloy, P., Bottcher, B., Ceulemans, H., Leutwein, C., Mellwig, C., Fischer, S., Gavin, A. C., Bork, P., Superti-Furga, G., Serrano, L., et al. (2004). Structure-based assembly of protein complexes in yeast. Science 303:2026–2029. Andreeva, A., Howorth, D., Chandonia, J.-M., Brenner, S. E., Hubbard, T. J. P., Chothia, C., and Murzin, A. G. (2008). Data growth and its impact on the SCOP database: new developments. Nucleic Acids Res. 36:D419–D425. Apic, G., Gough, J., and Teichmann, S. A. (2001). Domain combinations in archaeal, eubacterial and eukaryotic pro- teomes. J. Mol. Biol. 310:311–325. Blundell, T. L., Sibanda, B. L., Sternberg, M. J., and Thornton, J. M. (1987). Knowledge-based prediction of protein structures and the design of novel molecules. Nature 326:347–352. Capra, J. A., Laskowski, R. A., Thornton, J. M., Singh, M., and Funkhouser, T. A. (2009). Predicting protein ligand binding sites by combining evolutionary sequence conservation and 3D structure. PLoS Comput. Biol. 5:e1000585. Casari, G., Sander, C., and Valencia,A. (1995).Amethod to predict functional residues in proteins. Nat. Struct. Biol. 2:171–178. Chang, R. L., Andrews, K., Kim, D., Li, Z., Godzik, A., and Palsson, B. O. (2013). Structural systems biology evalua- tion of metabolic thermotolerance in Escherichia coli. Science 340:1220–1223. Choudhary, C., and Mann, M. (2010). Decoding signalling networks by mass spectrometry-based proteomics. Nat. Rev. Mol. Cell Biol. 11:427–439. Croft, D., O’Kelly, G., Wu, G., Haw, R., Gillespie, M., Matthews, L., Caudy, M., Garapati, P., Gopinath, G., Jassal, B., et al. (2011). Reactome: a database of reactions, pathways and biological processes. Nucleic Acids Res. 39:D691–D697. Cuff, A., Redfern, O. C., Greene, L., Sillitoe, I., Lewis, T., Dibley, M., Reid, A., Pearl, F., Dallman, T., Todd, A., et al. (2009). The CATH hierarchy revisited-structural divergence in domain superfamilies and the continuity of fold space. Structure 17:1051–1062 (London, England: 1993). Davey, N. E., Haslam, N. J., Shields, D. C., and Edwards, R. J. (2010). SLiMFinder: a web server to find novel, signifi- cantly over-represented, short protein motifs. Nucleic Acids Res. 38:W534–W539. David, A., Razali, R., Wass, M. N., and Sternberg, M. J. E. (2012). Protein-protein interaction sites are hot spots for disease-associated nonsynonymous SNPs. Hum. Mutat. 33:359–363. Dinkel, H., Michael, S., Weatheritt, R. J., Davey, N. E., Van Roey, K., Altenberg, B., Toedt, G., Uyar, B., Seiler, M., Budd, A., et al. (2012). ELM–the database of eukaryotic linear motifs. Nucleic Acids Res. 40:D242–D251. Franceschini, A., Szklarczyk, D., Frankild, S., Kuhn, M., Simonovic, M., Roth, A., Lin, J., Minguez, P., Bork, P., von Mering, C., et al. (2013). STRING v9.1: protein-protein interaction networks, with increased coverage and inte- gration. Nucleic Acids Res. 41:D808–D815. Gabdoulline, R. R., Stein, M., and Wade, R. C. (2007). qPIPSA: relating enzymatic kinetic parameters and interaction fields. BMC Bioinformatics 8:373. Gavin, A.-C., Aloy, P., Grandi, P., Krause, R., Boesche, M., Marzioch, M., Rau, C., Jensen, L. J., Bastuck, S., Dümpelfeld, B., et al. (2006). Proteome survey reveals modularity of the yeast cell machinery. Nature 440:631–636. Gene Ontology Consortium (2006). The Gene Ontology (GO) project in 2006. Nucleic Acids Res. 34:D322–D326. Gerstein, M., and Levitt, M. (1997). A structural census of the current population of protein sequences. Proc. Nat. Acad. Sci. U.S.A. 94:11911–11916. Guruharsha, K. G., Rual, J.-F., Zhai, B., Mintseris, J., Vaidya, P., Vaidya, N., Beekman, C., Wong, C., Rhee, D. Y., Cenaj, O., et al. (2011). A protein complex network of Drosophila melanogaster. Cell 147:690–703. Han, J.-D.J., Bertin, N., Hao, T., Goldberg, D. S., Berriz, G. F., Zhang, L. V., Dupuy, D., Walhout, A. J. M., Cusick, M. E., Roth, F. P., et al. (2004). Evidence for dynamically organized modularity in the yeast protein-protein interac- tion network. Nature 430:88–93. Hanahan, D., and Weinberg, R. A. (2011). Hallmarks of cancer: the next generation. Cell 144:646–674. Holm, L., and Rosenström, P. (2010). Dali server: conservation mapping in 3D. Nucleic Acids Res. 38:W545–W549. Inbar, Y., Benyamini, H., Nussinov, R., and Wolfson, H. J. (2005). Prediction of multimolecular assemblies by multiple docking. J. Mol. Biol. 349:435–447. Kiefer, F., Arnold, K., Künzli, M., Bordoli, L., and Schwede, T. (2009). The SWISS-MODEL repository and associated resources. Nucleic Acids Res. 37:D387–D392. Kiel, C., Beltrao, P., and Serrano, L. (2008). Analyzing protein interaction networks using structural information. Annu. Rev. Biochem. 77:415–441.
- 30. 18 2. STRUCTURAL SYSTEMS BIOLOGY: MODELING INTERACTIONS AND NETWORKS Kilpivaara, O., and Aaltonen, L. A. (2013). Diagnostic cancer genome sequencing and the contribution of germline variants. Science 339:1559–1562. (New York, N.Y.) Kim, P. M., Lu, L. J., Xia, Y., and Gerstein, M. B. (2006a). Relating three-dimensional structures to protein networks provides evolutionary insights. Science 314:1938–1941. (New York, N.Y.) Kim, W. K., Henschel, A., Winter, C., and Schroeder, M. (2006b). The many faces of protein-protein interactions: A compendium of interface geometry. PLoS Comput. Biol. 2:e124. Kotera, M., Hirakawa, M., Tokimatsu, T., Goto, S., and Kanehisa, M. (2012). The KEGG databases and tools facilitat- ing omics analysis: latest developments involving human diseases and pharmaceuticals. Methods Mol. Biol. 802:19–39. (Clifton, N.J.) Kühner, S., van Noort, V., Betts, M. J., Leo-Macias, A., Batisse, C., Rode, M., Yamada, T., Maier, T., Bader, S., Beltran- Alvarez, P., et al. (2009). Proteome organization in a genome-reduced bacterium. Science 326:1235–1240. (New York, N.Y.) Landgraf, R., Xenarios, I., and Eisenberg, D. (2001). Three-dimensional cluster analysis identifies interfaces and func- tional residue clusters in proteins. J. Mol. Biol. 307:1487–1502. Lasker, K., Topf, M., Sali, A., and Wolfson, H. J. (2009). Inferential optimization for simultaneous fitting of multiple components into a CryoEM map of their assembly. J. Mol. Biol. 388:180–194. Lavecchia, A., and Di Giovanni, C. (2013). Virtual screening strategies in drug discovery: a critical review. Curr. Med. Chem. 20:2839–2860. Lièvre, A., Bachet, J.-B., Le Corre, D., Boige, V., Landi, B., Emile, J.-F., Côté, J.-F., Tomasic, G., Penna, C., Ducreux, M., et al. (2006). KRAS mutation status is predictive of response to cetuximab therapy in colorectal cancer. Cancer Res. 66:3992–3995. Lu, L., Lu, H., and Skolnick, J. (2002). MULTIPROSPECTOR: an algorithm for the prediction of protein-protein inter- actions by multimeric threading. Proteins 49:350–364. Marchler-Bauer, A., Zheng, C., Chitsaz, F., Derbyshire, M. K., Geer, L. Y., Geer, R. C., Gonzales, N. R., Gwadz, M., Hurwitz, D. I., Lanczycki, C. J., et al. (2013). CDD: conserved domains and protein three-dimensional structure. Nucleic Acids Res. 41:D348–D352. Minguez, P., Parca, L., Diella, F., Mende, D. R., Kumar, R., Helmer-Citterich, M., Gavin, A.-C., van Noort, V., and Bork, P. (2012). Deciphering a global network of functionally associated post-translational modifications. Mol. Syst. Biol. 8:599. Mosca, R., Pons, C., Fernández-Recio, J., and Aloy, P. (2009). Pushing structural information into the yeast interac- tome by high-throughput protein docking experiments. PLoS Comput. Biol. 5:e1000490. Mosca, R., Céol, A., and Aloy, P. (2013). Interactome3D: adding structural details to protein networks. Nat. Methods 10:47–53. Moult, J., Fidelis, K., Kryshtafovych, A., and Tramontano, A. (2011). Critical assessment of methods of protein struc- ture prediction (CASP)–round IX. Proteins 79(Suppl 1):1–5. Neduva, V., and Russell, R. B. (2006). DILIMOT: discovery of linear motifs in proteins. Nucleic Acids Res. 34:W350 –W355. Van Noort, V., Seebacher, J., Bader, S., Mohammed, S., Vonkova, I., Betts, M. J., Kühner, S., Kumar, R., Maier, T., O’Flaherty, M., et al. (2012). Cross-talk between phosphorylation and lysine acetylation in a genome-reduced bacterium. Mol. Syst. Biol. 8:571. Petsalaki, E., Stark, A., García-Urdiales, E., and Russell, R. B. (2009). Accurate prediction of peptide binding sites on protein surfaces. PLoS Comput. Biol. 5:e1000335. Pflieger, D., Jünger, M. A., Müller, M., Rinner, O., Lee, H., Gehrig, P. M., Gstaiger, M., and Aebersold, R. (2008). Quantitative proteomic analysis of protein complexes: concurrent identification of interactors and their state of phosphorylation. Mol. Cell. Proteomics 7:326–346. Pieper, U., Webb, B. M., Barkan, D. T., Schneidman-Duhovny, D., Schlessinger, A., Braberg, H., Yang, Z., Meng, E. C., Pettersen, E. F., Huang, C. C., et al. (2011). ModBase, a database of annotated comparative protein structure mod- els, and associated resources. Nucleic Acids Res. 39:D465–D474. Pruitt, K. D., Tatusova, T., and Maglott, D. R. (2005). NCBI Reference Sequence (RefSeq): a curated non-redundant sequence database of genomes, transcripts and proteins. Nucleic Acids Res. 33:D501–D504. Punta, M., Coggill, P. C., Eberhardt, R. Y., Mistry, J., Tate, J., Boursnell, C., Pang, N., Forslund, K., Ceric, G., Clements, J., et al. (2012). The Pfam protein families database. Nucleic Acids Res. 40:D290–D301. Reva, B., Antipin, Y., and Sander, C. (2011). Predicting the functional impact of protein mutations: application to cancer genomics. Nucleic Acids Res. 39:e118.
- 31. REFERENCES 19 Stein, A., Céol, A., and Aloy, P. (2011). 3did: identification and classification of domain-based interactions of known three-dimensional structure. Nucleic Acids Res. 39:D718–D723. Stein, M., Gabdoulline, R. R., and Wade, R. C. (2007). Bridging from molecular simulation to biochemical networks. Curr. Opin. Struct. Biol. 17:166–172. Sunyaev, S., Ramensky, V., and Bork, P. (2000). Towards a structural basis of human non-synonymous single nucleo- tide polymorphisms. Trends Genet. 16:198–200. Tuncbag, N., Gursoy, A., Guney, E., Nussinov, R., and Keskin, O. (2008). Architectures and functional coverage of protein-protein interfaces. J. Mol. Biol. 381:785–802. Vandermarliere, E., and Martens, L. (2013). Protein structure as a means to triage proposed PTM sites. Proteomics 13:1028–1035. Wass, M. N., Fuentes, G., Pons, C., Pazos, F., and Valencia, A. (2011). Towards the prediction of protein interaction partners using physical docking. Mol. Syst. Biol. 7:469. Wilkins, A. D., Bachman, B. J., Erdin, S., and Lichtarge, O. (2012). The use of evolutionary patterns in protein annota- tion. Curr. Opin. Struct. Biol. 22:316–325. Winter, C., Henschel, A., Kim, W. K., and Schroeder, M. (2006). SCOPPI: a structural classification of protein-protein interfaces. Nucleic Acids Res. 34:D310–D314. Wu, C. H., Apweiler, R., Bairoch, A., Natale, D. A., Barker, W. C., Boeckmann, B., Ferro, S., Gasteiger, E., Huang, H., Lopez, R., et al. (2006). The Universal Protein Resource (UniProt): an expanding universe of protein information. Nucleic Acids Res. 34:D187–D191. Xuan, J., Yu, Y., Qing, T., Guo, L., and Shi, L. (2012). Next-generation sequencing in the clinic: Promises and chal- lenges. Cancer Lett. Yang, H., Qureshi, R., and Sacan, A. (2012). Protein surface representation and analysis by dimension reduction. Proteome Sci. 10(Suppl 1):S1. Yus, E., Maier, T., Michalodimitrakis, K., van Noort, V., Yamada, T., Chen, W.-H., Wodke, J. A. H., Güell, M., Martínez, S., Bourgeois, R., et al. (2009). Impact of genome reduction on bacterial metabolism and its regulation. Science 326:1263–1268. (New York, N.Y.) Zhang, Q. C., Petrey, D., Deng, L., Qiang, L., Shi, Y., Thu, C. A., Bisikirska, B., Lefebvre, C., Accili, D., Hunter, T., et al. (2012). Structure-based prediction of protein-protein interactions on a genome-wide scale. Nature 490:556–560. Zhang, Y., Thiele, I., Weekes, D., Li, Z., Jaroszewski, L., Ginalski, K., Deacon, A. M., Wooley, J., Lesley, S. A., Wilson, I. A., et al. (2009). Three-dimensional structural view of the central metabolic network of Thermotoga maritima. Science 325:1544–1549.
- 32. 21 © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/B978-0-12-405926-9.00003-4 Computational Systems Biology, Second Edition 3 Understanding Principles of the Dynamic Biochemical Networks of Life Through Systems Biology Hans V. Westerhoffa,b,c , Fei Hec,d , Ettore Murabitoc , Frédéric Crémazya , Matteo Barberisa a Department of Synthetic Systems Biology and Nuclear Organization, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, The Netherlands, b Department of Molecular Cell Physiology, Faculty of Earth and Life Sciences, VU University Amsterdam, Amsterdam, The Netherlands, c Manchester Centre for Integrative Systems Biology (MCISB), Manchester, UK d Department of Automatic Control and systems Engineering, The University of Sheffield, Sheffield, UK C H A P T E R C O N T E N T S 1 Principles based on topology of the genome-wide metabolic network: limited numbers of possible flux patterns 22 2 Principles based on topology of the genome-wide metabolic network: toward personalized medicine 25 3 Industrially relevant applications of topology and objective-based modeling 26 4 Applications of topology and objective- based modeling to cancer research and drug discovery 27 5 Principles of control 30
- 33. 22 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS Abstract Systems Biology brings the potential to discover fundamental principles of Life that cannot be discovered by considering individual molecules. This chapter discusses a number of early, more recent, and upcoming dis- coveries of such network principles. These range from the balancing of fluxes through metabolic networks, the potential of those networks for truly individualized medicine, the time dependent control of fluxes and con- centrations in metabolism and signal transduction, the ways in which organisms appear to regulate metabolic processes vis-à-vis limitations therein, tradeoffs in robustness and fragility, and a relation between robustness and time dependences in the cell cycle. The robustness considerations will lead to the issue whether and how evolution has been able to put in place design principles of control engineering such as infinite robustness and perfect adaptation in the hierarchical biochemical networks of cell biology. 1 PRINCIPLES BASED ON TOPOLOGY OF THE GENOME-WIDE METABOLIC NETWORK: LIMITED NUMBERS OF POSSIBLE FLUX PATTERNS The genome-wide reconstructions of enzyme-mediated metabolic activities in various organisms have led to long lists of correspondences between genes, proteins, enzyme activi- ties, and to implied changes in the concentrations of metabolites (Herrgård et al. 2008; Thiele et al. 2013). If any reaction activity is represented by a reaction rate v, then the list of activities may be written as a long column (or vector) of v’s. For a reaction i, one may then write the change it effects in the number of Moles of any metabolite Xj by a stoichiometric number Nji that is defined by the reaction chemistry: Here the final term corresponds to the dilution due to growth at the specific growth rate μ. Doing this for all reactions and generalizing to vectors and matrix this leads to: Here v is a column of all the rates of all the reactions in the organism (i.e. one rate for every gene product at the level of enzyme or transporter) per unit intracellular volume. X is a col- umn of all the molecule numbers (in Moles) of the metabolites in and around the organism, and N the matrix of stoichiometric coefficients of that organism. N represents all single-step (3.1) dXj dt = Nji·vi − µ·Xj (3.2) dX dt = N·v − µ·X 6 Principles of regulation 32 7 Regulation versus control 33 8 Robustness and fragility and application to the cell cycle 35 9 Perfect adaptation and integral control in metabolism 38 References 42
- 34. 1 PRINCIPLES BASED ON TOPOLOGY OF THE GENOME-WIDE METABOLIC NETWORK 23 catalytic capabilities of the organism. In the consensus reconstruction of the human (Thiele et al. 2013), v is a list of 7440 reactions, and X of a list 5063 metabolites, 642 of which are extra- cellular. N is a 5063 by 7440 matrix of numbers like 1, 2, −1, with many zero’s. N is a genome-wide culmination of molecular biochemistry. For any molecule in an organ- ism, say molecule B, it shows from which molecules it can be made in a single step that is cata- lyzed by a protein encoded by the genome of the organism. It also shows into which other molecules the molecule can be converted in a single step. Although of great biochemical inter- est, this does not correspond to the solution of the biological question how an organism builds itself from components it takes from the environment, i.e. of how an organism recreates life from dead materials. For many components an organism cannot be built in a single step from the extracellular components. To address this issue, systems biology is needed, i.e. some way of reflecting how the indi- vidual reactions encoded by the genome integrate their actions. Because it is genome wide, i.e. contains (in principle) all reactions encoded by the genome, matrix N has the potential to do this. N may tell us that molecule A cannot be converted in a single step to molecule B, but may be converted into a molecule C, say by reaction number 5 (i.e. NA5 = −1 and NC5 = +1, while NB5 = 0), and that molecule C can be converted to molecule B by a reaction 9 (i.e. NA9 = 0, NB9 = 1, and NC9 = −1), so that indirectly by collaboration of enzymes 5 and 9, molecule A can be converted to molecule B: networking of enzyme molecules is needed, with metabolite C as communicator (Figure 3.1). If the enzymes work intracellularly and one would start with zero B and C but with a certain amount of A, one would see that the concentration of C would build up first and that only then the concentration of B should begin to increase. If A is kept constant by external supplies, C will increase with time until it becomes constant and the rates of reactions 5 and 9 have become the same. This is called the intracellular steady state. Because the extracellular compartment is much larger, an intracellular steady state will be A A C C B B 5 9 1 2 7 10 D D E E 3 4 F F 6 8 FIGURE 3.1 Example of a network described by matrix N, with molecule A converted to molecule B via equilib- rium reactions.
- 35. 24 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS achieved while the extracellular concentrations are still increasing or decreasing, very slowly with time. If one divides the metabolites into mi intracellular ones (Xi) and me extracellular ones (Xe), reorders the rows of matrix N such that its bottom rows make up the submatrix Nex containing extracellular metabolites, one finds for the upper submatrix of N, Nin: while: Most metabolites are not endpoints of a metabolic pathway, but intermediates with life times much shorter than the cell cycle time. We shall further focus on these cases and thereby be able to neglect the term containing the dilution due to growth. In cases where this does not apply, one may add the growth rate to the vector v and extend N accordingly. At this intracellular steady state, matrix N now puts a strong constraint on all the rates because the latter have to satisfy Equation. 3.3. Only the rate vectors that are in the Kernel of Nin, the subspace of all the possible rate vectors, are admissible. This limitation is enormous, i.e. from the 7444 dimensional space suggested by the length of the rate vector v, the reduc- tion is to a 7444-4421 = 3013 dimensional subspace. Clearly, the intracellular location of most enzymes and the consequent occurrence of steady state, it forces the enzymes to collaborate, to balance their fluxes, and to come to a concerted behavior that produces a steady state. Should a chemical reaction network be created at random, then it would often not relax to a steady state. Here we use a principle of Biology, i.e. that the living organisms we study are viable and hence not subject to metabolic explosions (Teusink et al. 1998), i.e. they exhibit stationary states, and the common stationary metabolic states is the steady state (Westerhoff Van Dam 1987). If life harbored a single linear pathway of 7442 enzymes and two transporters, then the number of intracellular metabolites would be 7443 and the space of possible reaction rates would have been reduced from dimension 7444 to dimension 1: it is the branching of path- ways that is at the basis of the remaining dimensionality of the possible rate distributions at steady state. In actual practice only a single (or a few) steady state is obtained with a single set of rates, although the steady-state conditions still permit an incredibly high number of steady states together filling the 3013 dimensional space. The genes are however expressed to a certain level, as defined by the environment plus the parameters of the intracellular networks, which begin to define the actual vector v, whereto the intracellular metabolite concentrations adjust so that the rates v change until the steady-state condition of Equation 3.3 is met, after which the system becomes constant in time, corresponding to steady state, which corresponds to a zero dimensional space. Evolution has selected the values of all the internal parameters such that a steady state can be obtained (see above) and possibly such that the actual rates are what is optimal for the organism if the extracellular conditions correspond to conditions that reigned during evolution. (3.3) 0 = dXi dt ss = Nin·v − µ·Xi (3.4) dXe dt ss = Nex·v
- 36. 2 PRINCIPLES BASED ON TOPOLOGY OF THE GENOME-WIDE METABOLIC NETWORK 25 If we measure the changes in time of the extracellular metabolites, and insert these into Equation 3.4, then this gives us an additional reduction of dimensionality by 642–2271. This is now not a boundary condition imposed by a fundamental principle, but an experimental observation that could help us to estimate the intracellular behavior of the network. However, we still cannot establish what the intracellular state is: the world of possible states is still 2271 dimensional for the human metabolic map. One approach is to determine intracellular fluxes experimentally by a procedure known as flux analysis, which often employs isotopically labeled growth. Here one may deduce from the growth rate and biomass composition a great many anabolic fluxes and use these to con- fine the possible fluxes (Sauer 2006). All these methodologies are empirical ways to establish what the actual flux distributions are. There are additional ways of limiting the space of possible fluxes. One is that of requiring that no single reaction runs in the direction that is uphill in terms of thermodynamics (Westerhoff Van Dam 1987). In principle the concentrations of intracellular metabolites are then needed, but assuming that these are within reasonable bounds (e.g. between 0.001 and 100 mM) certain directions of reaction can be excluded. Another way is to impose that no reaction rate can become higher than the Vmax of the enzyme that catalyzes the reaction, where the Vmax is determined in cell extracts (Mensonides et al. 2013; and see below). These two principles merely give bounds to values of reaction rates however; they do not reduce the dimensionality of the space of rates (we define reaction rates as net fluxes through processes not as unidirectional fluxes). A more fundamental principle is often used by what is called Flux Balance Analysis (FBA), which assumes that efficiency is maximal in terms of ATP yield, and yet another one assumes maximal biomass synthesis. We shall here discuss the former. For two parallel pathways that hydrolyze different amounts of ATP, this removes the pathways that hydrolyze most ATP. This principle has the advantages that it does not require experimental measurement if it is plainly assumed to apply and that it does reduce the dimensionality of the space of reaction rate distribution appreciably. However, this principle of optimal efficiency has been shown not to apply completely in a number of cases. Organisms such as baker’s yeast for instance do not grow at maximal efficiency when glucose is present in excess (Simeonidis et al. 2010). More in general organisms do not seem to be optimized for thermodynamic efficiency or yield (Westerhoff et al. 1983). On the other hand, the effect of reducing the world of solutions to Equation 3.3, may still be largely appropriate, and the approach may be useful as a first and limited approach in some cases (Reed and Palsson, 2004) and see below. 2 PRINCIPLES BASED ON TOPOLOGY OF THE GENOME-WIDE METABOLIC NETWORK: TOWARD PERSONALIZED MEDICINE If one is interested in whether the organism (through its matrix N) is actually capable of synthesizing a particular intracellular metabolite, say metabolite number 2031, one substi- tutes 1 for the zero at row 2031 of the zero vector at the left-hand side of Equation 3.3 and attempts to find solutions for the rate vector v. Often multiple solutions will be found. One may then ask whether metabolite number 2031 can be synthesized from a certain type of nutrition. To address this issue one should analyze the molecular composition of the nutri- tion, then require the rates of the transport (across the plasma membrane) reactions in v that
- 37. 26 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS correspond to substances that are not present in the nutrition to be nonnegative (positive being defined as outward transport), and again try to find a solution for the rate vector v that is consistent with these conditions. If such a rate vector is found then the map is consistent with producing the intracellular metabolite. Knowledge about the DNA sequence of an individual, enables one to understand where that individual may have inactive gene products on its metabolic map. Requiring in the above computations the corresponding reaction rates (elements of rate vector v) to be zero, one may again try to find a solution for the rate vector that delivers metabolite 2031. If this is impos- sible for the individual then it suffers from an inborn error of metabolism. Repeating this procedure for other nutrients, then enables one to examine whether the disease can be averted by using a special diet. By equating dXin/dt to the biomass composition in terms of all molecules in the living organism, and then solving the resulting equation for the rate vector v, one may ask whether the map is able to make its complete self, and thereby scout for all possible inborn errors of metabolism at the same time. Because of the large size of the matrix N, finding all possible solutions is computationally challenging, but finding one solution often suffices and is possible with modern algorithms. 3 INDUSTRIALLY RELEVANT APPLICATIONS OF TOPOLOGY AND OBJECTIVE-BASED MODELING If the goal of an FBA is to identify a pattern of fluxes fulfilling the steady-state condition imposed by Equation. 3.3 under the assumption that an objective function Z representing the biological process is optimal, the task is: where vL and vU are the lower and upper bound of the fluxes (defining the range of values that the different rates can have), and f is a set of coefficients defining the objective function Z in terms of a linear combination of the rates v. Depending on the specific information we want to retrieve through FBA, Z can also represent a non-biological criterion of optimality, as we shall see below. A promising applications of FBA is in industrial protein production. Proteins require com- plex systems for their synthesis that only living cells are equipped with. The complexity of these “cell factories” is far beyond that of man-made production systems and we are far from understanding their functioning in a comprehensive way. As a consequence protein produc- tion tends to be quite unpredictable. The complexity of these factories which derives from the intricate interconnectivity of its different components has to be taken into account at some level if one wants to make protein production predictable and hence be able to play with the “control knobs” of these factories to adjust the production process to our needs. The applica- tion of FBA, and more in general the adoption of the Systems Biology perspective, may help to make this process more predictable and design strategies to improve protein harvest. (3.5) maximize Z = f T · v subject to N · v = 0; vL ≤ v ≤ vU
- 38. 4 APPLICATIONS OF TOPOLOGY AND OBJECTIVE-BASED 27 Through FBA, for example, it is possible to predict the optimal pattern of internal fluxes rep- resenting the metabolic functioning of the cells cultured under specific conditions. This enables us to attempt to predict and compare the flux patterns of a control culture and a cul- ture expressing the recombinant protein. The superposition of these patterns would provide us with a set of reactions that are either (significantly) active in both scenarios or that turn on/ off when switching from one situation to the other. This set of reactions would host on the one hand the main metabolic processes common to both situations and on the other hand the main metabolic changes that cells undergo when expressing the protein. This would give us some insights on how cells redirect their metabolic trafficking in order to fulfill the new task of producing the recombinant protein. In the scenario illustrated above one would want the flux patterns predicted through FBA for the control and recombinant cultures to be as close as possible to the real functioning of the cells. To this end a good strategy consists of including in the computational representation of the system some experimental data, such as exchange fluxes and growth rate, retrieved in the two conditions. The objective function will be then defined as the negative mismatch between the predicted and the experimental value of the quantities that have been measured: where vi and mi are respectively the predicted and the experimental value of the measured quantity i. From a genetic engineering perspective a relevant question would be whether it is possible to increase the yield of the recombinant protein for a specific growth medium by diverting the internal flux toward more favorable metabolic routes. In this case one would compare the flux pattern obtained for the recombinant culture when Z is defined as in Equation 3.6 and the flux pattern obtained by setting Z = vr where vr is the rate of the pseudo-reaction introduced in the model to represent the recombinant protein production. Another relevant question concerns the growing medium composition. FBA could also be used to identifying the limiting nutrients and suggest alternative optimal feed design to fur- ther increase the protein production. 4 APPLICATIONS OF TOPOLOGY AND OBJECTIVE-BASED MODELING TO CANCER RESEARCH AND DRUG DISCOVERY Drugs are designed to affect one or more specific properties of the cells needing treatment. These properties usually represent what differentiates diseased cells from their normal coun- terparts, or a pathogenic organism from its host. The property one chooses to affect can vary depending on the specific clinical strategy pursued. Because neoplastic cells grow and repli- cate at a considerably faster rate than their normal counterparts, the rationale behind many of the possible choices in cancer treatment consists of halting the proliferative potential of the malignant tissue. Indeed, traditional clinical approaches such as chemotherapy and radio- therapy aim to kill cancer cells by disrupting their replication machinery. Similarly, in drug intervention at the metabolic level, the preliminary step consists of identifying a property (3.6) Z = − i |vi − mi|
- 39. 28 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS which characterizes the altered phenotype and which is therefore sensible to target. In this respect, constraint-based modeling approaches, and particularly FBA, can provide us with a way to identify these properties. If the system under study is known to optimize a certain biological requirement, then that requirement might be considered as the property one may want to target in order to disrupt the metabolic phenotype of the cell. However, the identification of the biological task that the objective function should represent is not always easy. For studies involving E. coli metabo- lism, the objective function Z is usually defined to represent the yield of biomass (Reed and Palsson 2004), assuming that bacteria aim to grow as fast as possible (although this assump- tion does not reflect a generally valid principle in microbiology (Schuster et al. 2008). By contrast, for human cells, things are not so straightforward. Since cancer cells grow at a much higher rate than their normal counterparts, it would seem reasonable to adopt the same approach as for E. coli by choosing the maximization of biomass production rate as the optimization criterion. Although this intuitive choice may seem sensible, the resulting FBA solution highlights a flux pattern which does not match with the observed characteristic of cancer metabolism (Warburg and Dickens 1931). Because of the high demand of ATP in the production of biomass, the flux pattern corresponding to the maximal yield shows the glucose uptake flux entirely entering the TCA cycle, with no lactate production. To retrieve a flux pattern highlighting the cancer metabolic features (a constitutive activation of the branch leading to lactate production and, possibly, the reduc- tion of the flux entering the TCA cycle), the FBA problem has then to be formulated differ- ently. A possible way to do so consists of replacing the maximal yield of biomass with a different criterion of optimality. In a recent work, Simeonidis et al. showed how an appropri- ate reformulation of FBA can be used to reproduce the Crabtree effect, an experimentally observed behavior whereby Saccharomyces. cerevisiae produces ethanol aerobically in the presence of high external glucose concentrations rather than producing biomass through the TCA cycle (Simeonidis et al. 2010) The authors hypothesized that (one of) the “driving forces” behind yeast metabolism is resource preservation (see also León et al. 2008). By mini- mizing the number of active reactions (and hence the number of enzymes) needed to pro- duce a required amount of biomass, the flux patterns obtained as solutions of the FBA problem showed the characteristic switch from respiration to fermentation that occurs when the concentration of glucose in the growing medium is increased above a certain threshold. Because of the commonalities in the metabolic features of fermenting yeast and cancerous cells (Diaz-Ruiz et al. 2009), a similar argument might be applied to reproduce the constitu- tive metabolic changes occurring in carcinogenesis. From an FBA perspective, higher con- centration of glucose in the growth medium and higher rate of glucose uptake due to over-expression of glycolytic enzymes are both implemented by increasing the upper limit of the glucose uptake flux. In both cases, the requirement of resource preservation would force the system to switch from respiration to fermentation/lactate production as soon as the glycolytic flux becomes high enough to provide the cell with the amount of ATP needed for the required production of biomass. A related issue that FBA could address is whether cancer cells are committed to optimize different biological functions concurrently. Indeed, the enhanced replication rate of neo- plastic cells, combined with a predilection for fermentation (which is not the most efficient way to produce ATP) would seem to support a multifunctional optimization hypothesis,
- 40. 4 APPLICATIONS OF TOPOLOGY AND OBJECTIVE-BASED 29 whereby different criteria of optimization have to be satisfied simultaneously. As initially hypothesized by Gatenby and Gawlinski (1996), the production and excretion of lactic acid constitutes a way for cancer cells to compete with their normal counterpart by creating a hostile environment for normal cells. However, the fact that sometimes the TCA cycle is nevertheless active (although to a smaller extent than its normal capacity) makes evident that competing through excretion of lactate is not the only task that cancer cells try to opti- mize. Using a specular argument, one could say that, despite the enhanced replication rate of cancer cells, the fact that the TCA cycle is somehow hampered shows that replicating most efficiently or at the highest possible rate is not the (only) objective that drives cancer cells, or, in other words, that there are multiple goals pushing the system toward a different metabolic flux pattern. The relevance of different possible optimization criteria in the func- tioning of the system and their relative weights could also help to elucidate why the pheno- typic traits of cancer metabolism are present to different extents in different types of cancer and in different cells in the same tumor. There are other points that an FBA approach might help to elucidate. Knowledge of the metabolic shift occurring in tumorigenesis predominantly involves central carbon metabo- lism. However, the shift may extend beyond central metabolism, and remarkable metabolic differences between normal and cancer cells may lie in pathways not yet studied within the context of cancer research. A further application of FBA could highlight particularly active metabolic pathways in cancer on a genome-scale level, and identify the regions where the flux pattern differs most between cancer and normal cells. Shlomi et al. (2008) have recently used an FBA approach to describe the tissue specificity of human metabolism, where tissue- specific gene and protein expression data were integrated with a genome-scale reconstruction of the human metabolic network. Different integer values were assigned to different gene- expression states, so to distinguish among highly (1), lowly (−1), and moderately (0) expressed genes. The objective function of the FBA problem was then set to account for the differences between the activity of each reaction in the predicted pattern of fluxes and the integer repre- sentation of the corresponding experimental gene-expression level. By minimizing such an objective function, the authors were able to retrieve stoichiometrically and topologically con- sistent flux patterns on a genome-scale level with the maximum number of reactions whose activity was in accordance with their expression state. This study may establish a FBA-based computational approach for the genome-wide study of normal and cancer human metabo- lism in a tissue-specific manner. Another interesting point FBA might address is the following. Given the selective pressure that biological systems undergo when functioning under mutual competition, it seems rea- sonable to assume that cancer cells fulfill their specific biological tasks in the most economical way. In other words, given the available external substrates and given a set of functionally important targets to accomplish, the cell would employ its resources most “effortlessly.” From a metabolic perspective, this would translate into the employment of a minimal number of active reactions, or, more generally, a minimal employment of resources. In E. coli, for exam- ple, experimental results have shown that fitness increases while unused catabolic functions decrease, this reduction being beneficial and therefore favored by selection (Cooper and Lenski 2000). In the context of FBA, this “principle of minimal effort” has been used in differ- ent forms to identify the pattern of fluxes that best portraits the system functioning with respect to specific criteria of optimality (León et al. 2008; Holzhütter 2004). It should be noted
- 41. 30 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS however that many cancer cells appear to secrete more metabolites into their surrounding medium than what may be consistent with minimal use of their resources (Jain et al. 2012). Combinations of optimality criteria with observed flux patterns as constraints for the FBA solutions might be a strategy. On the other hand, there exist different flux patterns that are equally optimal with respect to a certain criterion or set of criteria. Extension of FBA to find alternate optimal solutions (Lee et al. 2000) or alternate optimal patterns of fluxes (Murabito et al. 2009) have been devel- oped. In particular, an algorithm able to find all the minimal and equally optimal flux pat- terns of a metabolic network with respect to a given functional task has been proposed (Murabito et al. 2009). The superposition of all minimal optimal flux patterns allows us to identify those pathways or sets of reactions that must be active in order to optimally fulfill a given function, and other sets of reactions that can be alternatively active. The application of such an approach in the context of cancer research might help to identify and predict the nar- rowest region of human metabolism necessary to observe the carcinogenic metabolic shift. From the perspective of developing a kinetic model of cancer metabolism, these results might also provide modelers with a concise set of reactions that can be used as a backbone for a mechanistic representation of the system under study, as well as an idea about which path- ways and reactions can be reasonably neglected. 5 PRINCIPLES OF CONTROL In biochemical networks, rates of chemical conversions or transport reactions are not just determined by the properties of the enzyme or transporter that catalyzes them, but also by properties of other components of the network. Something similar applies to the concentra- tions of metabolites in the network. It therefore makes sense to define the control of a rate vi of any process in the network by the activity ei of that same process or of any other process ej in the network. The definition of the corresponding flux control coefficient reads as: This definition differs somewhat from the standard definition of the flux control coefficient (Burns et al. 1985), which is limited to the control of steady-state fluxes. Here we are more explicit about the fact that one may also define the flux control coefficient outside of steady state. This does require one to keep track of time, i.e. to be careful about defining the initial (t = 0) condition. The definition compares two effects that a given amount of agent pj, that modulates the rate of process vj specifically, may have on processes i and j (i may or may not equal j). The first is the effect agent pj has on the rate vi of process i when that process func- tions in the system. The second is the effect the same amount of the agent pj would have on process j when the process j would be outside the system but in the same conditions, with (3.7) Cej vi (t) def = ∂lnvi ∂lnej = ∂lnvi(t) ∂pi in the system ∂lnvj (t=0) ∂pj in a constant molecular environment
- 42. 5 Principles of control 31 those conditions frozen. For a network with n processes, Westerhoff (2008) has proven the general property or “law”: The right-hand side is the flux control coefficient of time defined by: It quantifies the extent to which the rate of process i varies with time. We first discuss the example of a signal-transduction cascade with all proteins in the inac- tive un-phosphorylated state, which is then confronted with a sudden activation of a receptor, the activity of which then decays slowly. The rate of phosphorylation of the target of this receptor (which we here assume to be a protein kinase) will jump from zero up to a rate that is almost constant initially, i.e. after that initial jump, the time-control coefficient of that rate will be virtually zero. As a consequence, the above law (Equation 3.8) predicts that all pro- cesses in the network together control the rate of phosphorylation of the target at a control coefficient of 1. However, since there is hardly any phosphorylated target in the beginning, none of the other processes can be active and only the first kinase (the active receptor) can control the rate of phosphorylation of its target. Consequently, that kinase will initially be in full control; a 10% activation of the kinase will produce a 10% higher degree of phosphoryla- tion of the target at any given (short) time t after receptor activation. Because it phosphorylates its target, the kinase will relatively quickly decrease in rate and this decrease will be quicker the more active the kinase is. Consequently, control by the kinase in the rate of phosphorylation of its target at a given moment in time will decrease fairly quickly to below 1 and the time control of the kinase reaction will become negative. As even more of the target gets phosphorylated, its phosphatase becomes active and gains in control. Paradoxically perhaps, this control on the rate of phosphorylation of this first tar- get is positive, as the phosphatase creates more substrate for the kinase reaction. As time proceeds, the control by the kinase will decrease further and that of the phosphatase will increase until the two add up to 1, as the time dependence control coefficient returns to zero, it steady-state value. In general both the kinase and the phosphatase control the rate of phos- phorylation of their substrate. For the concentration of any substance in the system, the time dependent control coeffi- cients sum to zero plus the time-control coefficient: This includes the classical summation law that the sum of all control coefficients with respect to any steady-state concentration equals zero. This law is general in the absence of metabolite channeling (Kholodenko and Westerhoff 1993). The sum must also equal zero when the variation of the concentrations with time exhibits a maximum or minimum. (3.8) n j=1 Cvi ej (t) = 1 + Cvi t (t) (3.9) Cvi t (t) def = ∂lnvi ∂lnt in the system (3.10) n j=1 CXk ej (t) = CXk t (t)
- 43. 32 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS Hornberg et al. (2005) have used this property to prove that in the MAP kinase cascade all the phosphatases (or strictly speaking all the negatively controlling processes) together are as important for the amplitude of the ERK phosphorylation as are all the kinases together. If instead one focuses on the time point where the ERK phosphorylation has decreased again to half its amplitude, then the time-control coefficient is negative, implying that the sum of the control by all the kinases and the control by all the phosphatases must be (equally) negative. Since the phosphatases exercise negative control and the kinases positive control, this implies that the phosphatases are more important for the concentration of Erk-PP at this time point than the kinases are. To the extent that the MAP kinase is important for transcrip- tion regulation, appreciating that transcription integrates the time dependence of Erk-PP, and accepting that the duration of Erk-PP signaling relative to its amplitude is important, Hornberg et al. (2005) concluded that the phosphatases are even more important for signal transduction than the kinases are. This conclusion was perhaps useful because much more attention had been paid to kinases that to phosphatases. Importantly also, the summation law states that all phosphatases together should exercise more control than all kinases together. The principle is not that the first phosphatase must exert more control than the first kinase and that these are the only two controlling enzymes. Indeed, in numerical simulations, con- trol was distributed over all kinases and phosphatases. The biologically important conclusion is that oncogenes and tumor suppressor genes should be sought among all genes encoding kinases and all genes encoding phosphatases or regulating their expression levels, explaining why there are so many of these genes and inducing us to infer that cancer is a systems biology disease (Hornberg et al. 2006). At the time point in which the ultimate signal (Erk-PP in the example of the MAP kinase cascade) has first increased from zero to half its amplitude, the control by time is positive, and the above summation law implies that the total control exercised by the kinases on the signal strength exceeds the total control by the phosphatases. Indeed, early on in signal transduction the kinases should be more important than the phosphatases for the concentration of the sig- nal molecule. Although the example is one of signal transduction, similar considerations apply to meta- bolic and gene-expression networks, and the above may serve to convey that, contrarily to what is often stated, control analysis and the fundamental principles that it brings, are not limited to steady state. 6 PRINCIPLES OF REGULATION The magnitude of the flux control coefficient of a step or of the enzyme catalyzing that step, corresponds to a potential, i.e. to the effect on the flux that an activation of that step or enzyme might have. That magnitude does not indicate whether that step is ever activated either by the network itself, in self-regulation, or by an external influence, e.g. by an engineer (Westerhoff et al. 2009). Regulation coefficients have been introduced to indicate how, when a process is actually regulated, the organism regulates it. The alternatives are regulation through metabolic inter- actions, through single transduction interactions leading to covalent modification of the enzyme, and through gene expression. The gene-expression regulation coefficient has been
- 44. 7 Regulation versus control 33 defined as the change in enzyme concentration divided by the change in flux through the enzyme, i.e. more precisely by: Here ei is the concentration of the enzyme catalyzing the process vi. The rate of an enzyme catalyzed reaction can often be written as the product of three factors, i.e. the enzyme concen- tration, the fraction φa of the enzyme that is in the covalent modification state that is active catalytically, and a factor υm comprising the rate’s dependence on the concentrations of the substrates, the products, and the metabolic modifiers that are not binding covalently or stably. The metabolic and signal-transduction regulation coefficients are defined, respectively by: and Regulation is also subject to a general principle or law: The sum of gene-expression, meta- bolic, and signal-transduction regulation of a metabolic rate is always the same and equal to 1 (Ter Kuile and Westerhoff 2001): 7 REGULATION VERSUS CONTROL As discussed above, regulation differs from control. Yet, it would seem that there might be connections between the two concepts. We shall limit the discussion to linear metabolic pathways. Such a pathway has a single steady-state flux, which is equal to the steady-state rates of all the reactions in the pathway. If the third step of the pathway were completely rate limiting and its expression level would be activated by 30% then the flux would also go up by 30% making its hierarchical regulation coefficient equal to 1. However, if its control on the flux were 0.2 only, then its activation by 30%, in the absence of hierarchical regulation of any of the other enzymes, would increase the flux by 6% only, so that its hierarchical regulation coefficients would equal 5. This suggests that there is some sort of reciprocity between regula- tion and control. When hierarchical regulation involves more enzymes of the pathway, this reciprocity becomes pathway wide, hence again a systems property. For a linear pathway of n enzymes the reciprocity is given by the law: (3.11) ρi g = dlnei dlnvi (3.12) ρi m = dlnϑm,i dlnvi (3.13) ρi s = dlnϕa,i dlnvi (3.14) ρi g + ρi m + ρi s ≡ 1 (3.15) n i=1 CJ i ·ρi h ≡ 1
- 45. 34 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS Here the hierarchical regulation coefficient ρi h comprises both gene-expression and signal- transduction regulation: The proof is as follows: Consider a regulation that results in a change in flux through the enzyme, dlnvi. The increase in that flux may be because gene expression is increased resulting in more enzyme ei, because the altered levels of metabolites (x) have altered the activity of the enzyme, or because signal transduction has led to activation of the enzyme (dlnϕa,i) by cova- lent modification: With the above definition (Equation 3.11) the change in enzyme concentration relates to the change in flux through the enzyme by: The effect of that change in activity of enzyme i on the steady-state flux through the path- way is given by: Here the C refers to the metabolic flux control coefficient, not the hierarchical one; the metabolic pathway is allowed to relax, the gene-expression and the signal-transduction regu- lation are supposed to be fixed by the external world (Westerhoff 2008; Westerhoff et al. 2009). Taking into account all changes in all enzymes, the change in steady-state flux is: And relating the change in enzyme to the change in rate of the enzyme, and then using that flux equals rate, one obtains: Division by dlnJ yields the law we wanted to prove. This law implies that if there is only a single rate-limiting step in the pathway and the pathway is being regulated, the hierarchical regulation coefficient of that enzyme is always 1. It turns out that the classical paradigm of metabolic control and regulation where there was a single rate-limiting step, and where it was not even considered to make a distinction between flux-limiting step and regulated step, corresponds to one and the same special and probably (3.16) ρi h def = ρi g + ρi s (3.17) dlnvi = dlnei + ∂lnϑm,i ∂lnx · dlnx + dlnϕa,i (3.18) dlnei = ρi g · dlnvi (3.19) (dlnJ)as a consequence of the change in activity of ei = CJ ei ·(dlnei + dlnϕa,i) (3.20) dlnJ = n i=1 CJ i ·(dlnei + dlnϕa,i) (3.21) dlnJ = n i=1 CJ i · (ρi g + ρi s) · dlnvi = n i=1 CJ i · (ρi g + ρi s) · dlnJ
- 46. 8 Robustness and fragility and application to the cell cycle 35 rare case. When flux control is distributed and only one pathway step is regulated hierarchi- cally, this needs not be the most rate-limiting step and the regulation coefficient equals the inverse of the control coefficient, i.e. there is much hierarchical regulation if the regulated step has little flux control. Let us consider the example of a three step linear metabolic pathway where the first and the third step have flux control coefficients of 1/3 and 2/3, respectively, and the second step therefore a flux control coefficient of zero. The cell may decide to regulate only the first step in the pathway. This makes the hierarchical regulation coefficient of that enzyme equal 3 (Equation (15)), i.e. the cell will have to increase the concentration of enzyme three times as much as the percentage increase in flux it wishes to obtain. The fluxes through enzymes 2 and 3 would increase due to metabolic regulation only, i.e. the increase in concentration of enzyme 1 would lead to an increase in the concentration of its product, which as substrate of enzyme 2 then would push more flux through enzyme 2. In this example, the metabolic regulation coefficients of enzymes 2 and 3 are 1, while their hierarchical regulation coefficients both equal zero. In the same example, the metabolic regulation of enzyme 1 must be negative, its metabolic regulation coefficient equaling −2 (Equations 3.14 and 3.16). This reflects a strong inhibition by its product or by the substrate of the third enzyme through allosteric feedback regulation. Rossell et al. (2006) have observed such, perhaps nonintuitive aspects of regula- tion experimentally. It could be more efficient for the cell to increase the concentration of the third enzyme and not to regulate the first enzyme hierarchically; then for a 10% increase in flux it would only have to increase the concentration of enzyme 3 by 15, rather than 30%. In the latter case, enzymes 1 and 2 would be regulated metabolically, again with metabolic regulation coeffi- cients of 1. The principles of metabolic regulation can be generalized to branched pathways, but then the meaning of some of the hierarchical regulation coefficients is less obvious. 8 ROBUSTNESS AND FRAGILITY AND APPLICATION TO THE CELL CYCLE To survive evolution, living systems may not only require optimal performance in terms of growth rate, yield, or efficiency, they may also need to be robust against perturbations. Since living systems depend fundamentally on nonequilibrium processes (Westerhoff Van Dam 1987; Westerhoff et al. 2009) an important issue is the robustness of an organism to the sustained perturbation of any one such process. Quinton-Tulloch et al. (2013) defined the robustness of a steady-state biological function vis-à-vis the sustained perturbation of any of its processes, as the percentage change in the activity of that process that would compromise the function by 1% only. Such a robustness is 1 for a process in isolation. Quinton-Tulloch and colleagues then calculated the robustness coefficients for fluxes in some 25 realistic models of biochemical networks. They found that virtually all robustness coefficients were much higher than the in vitro number of 1. Csete and Doyle (2002) had considered robustness with respect to periodic perturbations at various frequencies and found total robustness, in the sense of robustness integrated over all frequencies, to be conserved; making a network more robust at one frequency should
- 47. 36 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS always reduce its robustness at different frequencies. Quinton-Tulloch et al. (2013) examined whether steady-state robustness is conserved over all processes, i.e. whether the sum of the robustness over all perturbed molecular processes in the system should always be the same. They showed that such a conservation of total robustness is not found. The implication is that by increasing the activity of a process and thereby increasing the robustness of a network function with respect to perturbations in that process, one may increase the total robustness of the system. Defining fragility as the inverse of robustness, i.e. the fragility coefficient as the percentage reduction in function for a 1% reduction in the activity of a process in the system, Quinton- Tulloch et al. (2013) found that total fragility is conserved and should equal 1 if the fragility of a flux is considered. They proved this by identifying this fragility coefficient with the flux control coefficient. We here illustrate this principle for a model of an important regulatory aspect of the yeast cell, i.e. the cell cycle. The implementation of Metabolic Control Analysis (MCA) to metabolic pathways at steady state has been frequent, successful and is well-known. MCA has also been applied to mostly metabolic oscillations, either forced or autonomous, with the yeast glycoly- sis oscillations synchronized by acetaldehyde as significant examples (e.g. Richard et al. 1993; Kholodenko et al. 1997; Danø et al. 2001; Reijenga et al. 2001, 2002, 2005; du Preez et al. 2012a, 2012b). The cell cycle may however be a more important oscillation, which is rarely seen as a limit cycle however. Some initial control analysis has been done, revealing again distributed control, but there has been little induction toward general principles of cell cycle. We will here briefly discuss possible developments around unsuspected relationships between robust- ness, fragility and time dependence. Figure 3.2 shows a diagram underlying our dynamic model of a part of the cell cycle of S. cerevisiae, where activation of various mitotic kinase/cyclin (Cdk1/Clb) complexes occurs between DNA duplication (S phase) and cell division (M phase). A kinetic model describing Cdk1/Clb dynamics over time was implemented, where each kinase complex activates the next one in a linear cascade (Barberis et al. 2012). Their activation (and inactivation) occurs in a temporal fashion, and a design principle underlying the oscillatory behavior of Clb waves has been proposed (Barberis 2012). Cdk1-Clb5,6 + Sic1 Cdk1-Clb5,6-Sic1 Cdk1-Clb3,4-Sic1 Cdk1-Clb1,2-Sic1 Cdk1-Clb1,2 + Sic1 Cdk1-Clb3,4 + Sic1 Clb5,6 Clb1,2 Clb3,4 FIGURE 3.2 Signaling network describing Cdk1/Clb regulation from S to M phase of the cell cycle.
- 48. 8 Robustness and fragility and application to the cell cycle 37 Figure 3.3 shows that the three couples of Clb cyclins (Clb5,6, Clb3,4, and Clb1,2) undergo waves with amplitudes at different times. We shall first focus on the Clb5/6 couple and on the onset and the decay of the peak of their level. Figure 3.2 shows that Clb5/6 has a maximum at t = 23 and that at t = 8 it is hallway reaching that maximum and at t = 31.5 it is again half- way down. We computed the time-control coefficient Tc at those times, and this amounted to 0.87 and −4.57, respectively. We also computed the robustness R (defined as the inverse of the time-control coefficients) of the amplitude of Clb5/6 at the two halfway time points. We first computed the robustness of the Clb5/6 amplitude exercised by all processes in the network when perturbed simultaneously and equally in the same direction. The corresponding robust- nesses were 1.14849 and 0.21881 (see Table 3.1). Quinton-Tulloch et al. (2013) identified the inverse of the robustness with fragility, which in turn is equal to the control coefficient. As a consequence the above (Equation 3.10): can be reformulated as: For this case of equal perturbations of all processes amounts to: And the same for the other time point analyzed. The last two columns in Table 3.1 confirms this computationally. For the onset of the peak the time-control coefficient is positive. Hence the positive robust- nesses must be smaller (typically those with respect to kinases perturbations) than the robust- nesses with respect to the phosphatases (which are not considered explicitly in the system yet). For the decay of the peak, the inverse should be true. (3.10) n j=1 CXk ej (t) = CXk t (t) (3.22) n j=1 1 ReXk ej (t) = CXk t (t) (3.23) n j=1 1 ℜXk total for equal(t 1 2on ) = CXk t (t 1 2on ) 0 10 20 30 40 50 60 0 0.3 0.6 0.9 1.2 Time Clb3,4 Clb5,6 Clb1,2 [Clb] FIGURE 3.3 Computational time course of Clb cyclins couples over time.
- 49. 38 3. UNDERSTANDING PRINCIPLES OF THE DYNAMIC BIOCHEMICAL NETWORKS 9 PERFECT ADAPTATION AND INTEGRAL CONTROL IN METABOLISM Biochemical reaction networks can exhibit properties similar to those of control system structures in control engineering, but are they identical? The robustness of cellular adapta- tion to environmental conditions is often related to negative feedback control structures. For example, robust adaptations in a bacterial chemotaxis signaling network, in mammalian iron and calcium homeostasis, and in yeast osmoregulation have been interpreted as integral feed- back control systems (Yi et al. 2000; El-Samad et al. 2002; Ni et al. 2009; Muzzey et al. 2009). A recent study identified the three different types of control structures used in control engi- neering, i.e. proportional, integral, and derivative feedback control, in regulations of energy metabolism (Cloutier and Wellstead 2010). In addition, specific nonlinear dynamics in signal- ing networks, such as oscillation or bi-stability, can be induced by positive feedback loops. Feed-forward control structures are also observed in gene regulatory networks (Mangan and Alon 2003), as well as in the regulation of glycolytic intermediates (Bali and Thomas 2001). Regulation in living cells tends to occur at multiple levels simultaneously with a hierarchi- cal structure (Westerhoff 2008). In a metabolic network the regulation of a reaction rate can be TABLE 3.1 Calculation of time-control coefficient and robustness for Clb cyclins couples. Clb5,6 α(5,6) β(5,6) Max [5,6] Time Max [5,6]/2 t1/2 α(5,6) Max [5,6] Time Max [5,6]/2 t1/2 β(5,6) 1,41473 23 0,707365 8 0,076988 1,41473 23 0,789371 31,5 0,11453 Clb3,4 α(3,4) β(3,4) Max [3,4] Time Max [3,4]/2 t1/2 α(3,4) Max [3,4] Time Max [3,4]/2 t1/2 β(3,4) 0,959239 30 0,4796195 19,5 0,023046 0,959239 30 0,60479 37 0,05489 Clb1,2 α(1,2) β(1,2) Max [1,2] Time Max [1,2]/2 t1/2 α(1,2) Max [1,2] Time Max [1,2]/2 t1/2 β(1,2) 1,1148 34 0,5574 27 0,094539 1,1148 34 0,780522 41,5 0,049 Tcα Tcβ Rα Rβ 1/Rα 1/Rβ Tcα(5,6) Tcβ(5,6) Rα(5,6) Rβ(5,6) Rα(5,6) Rβ(5,6) 0,87071 4,57023 1,14849 0,21881 0,87071 4,57023 Tcα(3,4) Tcβ(3,4) Rα(3,4) Rβ(3,4) Rα(3,4) Rβ(3,4) 0,937 3,358332 1,06723 0,29777 0,937 3,35833 Tcα(1,2) Tcβ(1,2) Rα(1,2) Rβ(1,2) Rα(1,2) Rβ(1,2) 4,57937 2,605393 0,21837 0,38382 4,57937 2,60539
- 50. Random documents with unrelated content Scribd suggests to you:
- 51. THE NAVAL beforehand to envisage clearly the conditions and consequences involved in their policy of 'Security.' As regards naval preparations, things were better indeed than might have been expected, considering the vagueness of ideas in the matter of policy. We were safeguarded here by tradition, and the general idea of direction had been nearly sufficient. There was always trouble, but not as a rule serious trouble, in establishing the case for increases necessary to keep ahead of German efforts. There had been pinchings and parings—especially in the matter of fast cruisers, for lack of which, when war broke out, we suffered heavy losses—but except in one instance—the abandonment of the Cawdor programme—these had not touched our security at any vital point. Thanks largely to Mr. Stead, but also to statesmen of both parties, and to a succession of Naval Lords who did not hesitate, when occasion required it, to risk their careers (as faithful servants ever will) rather than certify safety where they saw danger—thanks, perhaps, most of all to a popular instinct, deeply implanted in the British mind, which had grasped the need for supremacy at sea—our naval preparations, upon the whole, had kept abreast of our policy for nearly thirty years. As regards the Army, however, it was entirely different. There had been no intelligent effort to keep our military strength abreast of our policy; and as, in many instances, it would have been too bitter a humiliation to keep our policy within the limits of our military strength, the course actually pursued can only be described fitly as a game of bluff. There had never been anything approaching agreement with regard to the functions which the Army was expected to perform. Not only did political parties differ one from another upon this primary and fundamental question, but hardly two succeeding War Ministers had viewed it in the same light. There had been schemes of a bewildering variety; but as the final purpose for which soldiers existed had never yet been frankly laid down and accepted, each of these plans in turn had been discredited by attacks, which called in question the very basis of the proposed reformation. While naval policy had been framed and carried out in accordance with certain acknowledged necessities of
- 52. POSITION TWO INCORRECT ASSUMPTIO NS national existence, military policy had been alternately expanded and deflated in order to assuage the anxieties, while conforming to the prejudices—real or supposed—of the British public. In the case of the fleet, we had very fortunately arrived, more than a generation ago, at the point where it was a question of what the country needed; as regards the Army, it was still a question of what the country would stand. But how could even a politician know what the country would stand until the full case had been laid before the country? How was it that while Ministers of both parties had the courage to put the issue more or less nakedly in the matter of ships, they grew timid as soon as the discussion turned on army corps? If the needs of the Commonwealth were to be the touchstone in the one case, why not also in the other? The country will stand a great deal more than the politicians think; and it will stand almost anything better than vacillation, evasion, and untruth. In army matters, unfortunately, it has had experience of little else since the battle of Waterloo. Mathematicians, metaphysicians, and economists have a fondness for what is termed 'an assumption.' They take for granted something which it would be inconvenient or impossible to prove, and thereupon proceed to build upon it a fabric which compels admiration in a less or greater degree, by reason of its logical consistency. There is no great harm in this method so long as the conclusions, which are drawn from the airy calculations of the study, are confined to the peaceful region of their birth; but so soon as they begin to sally forth into the harsh world of men and affairs, they are apt to break at once into shivers. When the statesman makes an assumption he does so at his peril; or, perhaps, to speak more correctly, at the peril of his country. For if it be a false assumption the facts will speedily find it out, and disasters will inevitably ensue. Our Governments, Tory and Radical alike, have acted in recent times as if the British Army were what their policy required it to be—something, that is, entirely different from what it really was. Judging by its procedure, the Foreign Office would appear to have made the singularly bold assumption that, in a military comparison with other nations, Britain was still in much the same relative position as in the days of Napoleon.
- 53. Sustained by this tenacious but fantastic tradition, Ministers have not infrequently engaged in policies which wiser men would have avoided. They have uttered protests, warnings, threats which have gone unheeded. They have presumed to say what would and would not be tolerated in certain spheres; but having nothing better behind their despatches than a mere assumption which did not correspond with the facts, they have been compelled to endure rebuffs and humiliations. As they had not the prudence to cut their coat according to their cloth, it was only natural that occasionally they should have had to appear before the world in a somewhat ridiculous guise. British statesmen for nearly half a century had persisted in acting upon two most dangerous assumptions. They had assumed that one branch of the national armaments conformed to their policy, when in fact it did not. And they had assumed also, which is equally fatal, that policy, if only it be virtuous and unaggressive, is in some mysterious way self-supporting, and does not need to depend on armaments at all. The military preparations of Britain were inadequate to maintain the policy of Security, which British Governments had nevertheless been engaged in pursuing for many years prior to the outbreak of the present war. [7] On the other hand, the abandonment of this policy was incompatible with the continuance of the Empire. We could not hope to hold our scattered Dependencies and to keep our Dominions safe against encroachments unless we were prepared to incur the necessary sacrifices. [1] American writers have urged criticism of this sort against the armaments of the U.S.A., which they allege are inadequate to uphold the policy of the 'Monroe Doctrine.' The German view of the matter has been stated by the Chancellor (April 7, 1913) when introducing the Army Bill:—History knows of no people which came to disaster because it had exhausted itself in the making of its defences; but history knows of many peoples which have perished, because, living in prosperity and luxury, they neglected their defences. A people which thinks that it is not rich enough to maintain its armaments shows merely that it has played its part.
- 54. [2] So the argument runs, and the course of our naval policy since Mr. Stead's famous press campaign in 1884 will be cited as an encouragement. [3] E.g. in the winter of 1908 and spring of 1909, when an influential section of the supporters of the present Cabinet chose to believe the false assurances of the German Admiralty, and freely accused their own Government of mendacity. [4] Innovations of this particular sort have possibly a better chance of preserving their existence than some others. 'Boards are screens,' wrote John Stuart Mill, or some other profound thinker; and in politics screens are always useful. [5] This is obvious from the White Paper without seeking further evidence in the ministerial press or elsewhere. [6] Of the six infantry divisions included in the Expeditionary Force only four were sent in the first instance; a fifth arrived about August 24; a sixth about mid-September. [7] Our Army, as a belligerent factor in European politics, is almost a negligible quantity. This Empire is at all times practically defenceless beyond its first line. Such an Empire invites war. Its assumed security amid the armaments of Europe, and now of Asia, is insolent and provocative (Lord Roberts, October 22, 1912). Nothing indeed is more insolent and provocative, or more likely to lead to a breach of the peace, than undefended riches among armed men. CHAPTER IV THE BALANCE OF POWER During the whole period of rather more than thirteen years—which has been referred to in previous pages as the post-Victorian epoch, and which extended roughly from January 1901, when Queen Victoria died, to July 1914, when war was declared—the British Army remained inadequate for the purpose of upholding that policy which British statesmen of both
- 55. parties, and the British people, both at home and in the Dominions, were engaged in pursuing—whether they knew it or not—and were bound to pursue, unless they were prepared to sacrifice their independence. The aim of that policy was the security of the whole empire. This much at any rate was readily conceded on all hands. It was not enough, however, that we approved the general aim of British policy. A broad but clear conception of the means by which our Government hoped to maintain this policy, and the sacrifices which the country would have to make in order to support this policy, was no less necessary. So soon, however, as we began to ask for further particulars, we found ourselves in the region of acute controversy. 'Security' was a convenient political formula, which could be accepted as readily by the man who placed his trust in international law, as by his neighbour who believed in battle fleets and army corps. In considering this question of security we could not disregard Europe, for Europe was still the storm-centre of the world. We could not afford to turn a blind eye towards the ambitions and anxieties of the great continental Powers. We were bound to take into account not only their visions but their nightmares. We could not remain indifferent to their groupings and alliances, or to the strength and dispositions of their armaments. That the United Kingdom was a pair of islands lying on the western edge of Europe, and that the rest of the British Empire was remote, and unwilling to be interested in the rivalries of the Teuton, Slav, and Latin races, did not affect the matter in the least. Nowadays no habitable corner of the earth is really remote; and as for willingness or unwillingness to be interested, that had nothing at all to do with the question. For it was clear that any Power, which succeeded in possessing itself of the suzerainty of Europe, could redraw the map of the world at its pleasure, and blow the Monroe Doctrine, no less than the British Empire, sky-high. Looking across thousands of leagues of ocean, it was difficult for the Dominions and the United States to understand how their fortunes, and the ultimate fate of their cherished institutions, could possibly be affected by the turmoil and jealousies of—what appeared in their eyes to be—a number of reactionary despotisms and chauvinistic democracies. Even the hundred and twenty leagues which separate Hull from Emden, or the seven which
- 56. GERMAN AIMS divide Dover from Calais, were enough to convince many people in the United Kingdom that we could safely allow Europe to 'stew in her own juice.' But unfortunately for this theory, unless a great continental struggle ended like the battle of the Kilkenny cats, the outside world was likely to find itself in an awkward predicament, when the conqueror chose to speak with it in the gates, at a time of his own choosing. British policy since 1901 had tended, with ever increasing self- consciousness, towards the definite aim of preventing Germany from acquiring the suzerainty of Western Europe. It was obvious that German predominance, if secured, must ultimately force the other continental nations, either into a German alliance, or into a neutrality favourable to German interests. German policy would then inevitably be directed towards encroachments upon British possessions. Germany had already boldly proclaimed her ambitions overseas. Moreover, she would find it pleasanter to compensate, and soothe the susceptibilities of those nations whom she had overcome in diplomacy or war, and to reward their subsequent services as allies and friendly neutrals, by paying them out of our property rather than out of her own. For this reason, if for no other, we were deeply concerned that Germany should not dominate Europe if we could help it. During this period, on the other hand, Germany appeared to be setting herself more and more seriously to acquire this domination. Each succeeding year her writers expressed themselves in terms of greater candour and confidence. Her armaments were following her policy. The rapid creation of a fleet—the counterpart of the greatest army in Europe—and the recent additions to the striking power of her already enormous army could have no other object. Certainly from 1909 onwards, it was impossible to regard German preparations as anything else than a challenge, direct or indirect, to the security of the British Empire. Consequently the direction of British policy returned, gradually, unavowedly, but with certainty, to its old lines, and became once more concerned with the maintenance of the Balance of Power as the prime necessity. The means adopted were the Triple Entente between Britain, France, and Russia. The object of this understanding was to resist the
- 57. DERELICT MAXIMS anticipated aggressions of the Triple Alliance, wherein Germany was the predominant partner. The tendency of phrases, as they grow old, is to turn into totems, for and against which political parties, and even great nations, fight unreasoningly. But before we either yield our allegiance to any of these venerable formulas, or decide to throw it out on the scrap-heap, there are advantages in looking to see whether or not there is some underlying meaning which may be worth attending to. It occasionally happens that circumstances have changed so much since the original idea was first crystallised in words, that the old saying contains no value or reality whatsoever for the present generation. More often, however, there is something of permanent importance behind, if only we can succeed in tearing off the husk of prejudice in which it has become encased. So, according to Disraeli, the divine right of Kings may have been a plea for feeble tyrants, but the divine right of government is the keystone of human progress. For many years the phrase British interests, which used to figure so largely in speeches and leading articles, has dropped out of use, because it had come to be associated unfavourably with bond-holders' dividends. The fact that it also implied national honour and prestige, the performance of duties and the burden of responsibilities was forgotten. Even the doctrine of laissez faire, which politicians of all parties have lately agreed to abjure and contemn, has, as regards industrial affairs, a large kernel of practical wisdom and sound policy hidden away in it. But of all these derelict maxims, that which until quite recently, appeared to be suffering from the greatest neglect, was the need for maintaining the Balance of Power in Europe. For close on two generations it had played no overt part in public controversy, except when some Tory matador produced it defiantly as a red rag to infuriate the Radical bull. If this policy of the maintenance of the Balance of Power has been little heard of since Waterloo, the reason is that since then, until quite recently, the Balance of Power has never appeared to be seriously threatened.[1] And because the policy of maintaining this balance was in abeyance, many people have come to believe that it was discredited. Because it was not visibly and actively in use it was supposed to have become entirely useless.
- 58. CONDITIONS OF BRITISH FREEDOM This policy can never become useless. It must inevitably come into play, so soon as any Power appears to be aiming at the mastery of the continent. It will ever remain a matter of life or death, to the United Kingdom and to the British Empire, that no continental state shall be allowed to obtain command, directly or indirectly, of the resources, diplomacy, and armaments of Europe. In the sixteenth century we fought Philip of of Spain to prevent him from acquiring European predominance. In the seventeenth, eighteenth, and nineteenth centuries we fought Louis XIV., Louis XV., and Napoleon for the same reason. In order to preserve the balance of power, and with it our own security, it was our interest under Elizabeth to prevent the Netherlands from being crushed by Spain. Under later monarchs it was our interest to prevent the Netherlands, the lesser German States, Prussia, Austria, and finally the whole of Europe from being crushed by France. And we can as ill afford to-day to allow France to be crushed by Germany, or Holland and Belgium to fall into her power. The wheel has come round full circle, but the essential British interest remains constant. The wheel is always turning, sometimes slowly, sometimes with startling swiftness. Years hence the present alliances will probably be discarded. It may be that some day the danger of a European predominance will appear from a different quarter—from one of our present allies, or from some upstart state which may rise to power with an even greater rapidity than the Electorate of Brandenburg. Or it may be that before long the New World, in fact as well as phrase, may have come in to redress the balance of the Old. We cannot say, because we cannot foresee what the future holds in store. But from the opening of the present century, the immediate danger came from Germany, who hardly troubled to conceal the fact that she was aiming at predominance by mastery of the Low Countries and by crushing France. That this danger was from time to time regarded seriously by a section of the British Cabinet, we know from their own statements both before war broke out and subsequently. It was no chimera confined to the imaginations of irresponsible and panic-stricken writers. In sober truth the balance of power in Europe was in as much danger, and the maintenance of
- 59. it had become as supreme a British interest, under a Liberal government at the beginning of the twentieth century, as it ever was under a Whig government at the close of the seventeenth and opening of the eighteenth. The stealthy return of this doctrine into the region of practical politics was not due to the prejudices of the party which happened to be in power. Quite the contrary. Most Liberals distrusted the phrase. The whole mass of the Radicals abhorred it. The idea which lay under and behind the phrase was nevertheless irresistible, because it arose out of the facts. Had a Socialist Government held office, this policy must equally have imposed itself and been accepted with a good or ill grace, for the simple reason that, unless the balance of power is maintained in Europe, there can be no security for British freedom, under which we mean, with God's help, to work out our own problems in our own way. English statesmen had adopted this policy in fact, if unavowedly— perhaps even to some extent unconsciously—when they first entered into, and afterwards confirmed, the Triple Entente. And having once entered into the Triple Entente it was obvious that, without risking still graver consequences, we could never resume the detached position which we occupied before we took that step. It is difficult to believe—seeing how the danger of German predominance threatened France and Russia as well as ourselves—that we should not have excited the ill-will of those two countries had we refused to make common cause by joining the Triple Entente. It was obvious, however, to every one that we could not afterwards retire from this association without incurring their hostility. If we had withdrawn we should have been left, not merely without a friend in Europe, but with all the chief Powers in Europe our enemies—ready upon the first favourable occasion to combine against us. There is only one precedent in our history for so perilous a situation— when Napoleon forced Europe into a combination against us in 1806. And this precedent, though it then threatened our Empire with grave dangers, did not threaten it with dangers comparable in gravity with those which menaced us a century later. The consequences of breaking away from the Triple Entente were sufficiently plain. We may build ships against one nation, or even against a
- 60. DEFENCE AND INVASION combination of nations. But we cannot build ships against half Europe. If Western Europe, with all its ports, its harbours, its arsenals, and its resources, was to fall under the domination of a single will, no effort of ours would be sufficient to retain the command of the sea. It is a balance of power on the continent, which alone makes it possible for us to retain it. Thus the maintenance of the balance of power is vital to our superiority at sea, which again is vital to the security of the British Empire.[2] Security in the widest sense was the ultimate end of our policy—security of mind, security from periodic panic, as well as actual military security. Looked at more closely, the immediate end was defence—the defence of the British Empire and of the United Kingdom. In the existing condition of the world a policy of 'splendid isolation' was no longer possible. Conditions with which we are familiar in commercial affairs, had presented themselves in the political sphere, and co-operation on a large scale had become necessary in order to avoid bankruptcy. England had entered into the Triple Entente because her statesmen realised, clearly or vaguely, that by doing so we should be better able to defend our existence, and for no other reason. After 1911 it must have been obvious to most people who considered the matter carefully that in certain events the Triple Entente would become an alliance. It is the interest as well as the duty of allies to stand by one another from first to last, and act together in the manner most likely to result in victory for the alliance. What then was the manner of co-operation most likely to result in victory for that alliance which lay dormant under the Triple Entente? But first of all, to clear away one obscurity—Invasion was not our problem; Defence was our problem; for the greater included the less. The word 'defence' is apt to carry different meanings to different minds. The best defence of England and British interests, at any given time, may or may not consist in keeping our main army in the United Kingdom and waiting to be attacked here. It all depends upon the special circumstances of each case. The final decision must be governed by one consideration, and
- 61. CO- OPERATION WITH FRANCE one only—how to strike the speediest, heaviest, and most disabling blow at the aggressor. If by keeping our army in England and endeavouring to lure the enemy into our toils, that end is most likely to be accomplished, then it is obviously best to keep our army here. If by sending it into the north of France to combine with the French the supreme military object has a superior chance of being achieved, then it is best to send it into the north of France. A defensive war cannot be defined and circumscribed as a war to drive out invaders, or even to prevent the landing of invaders. The best way to defend your castle may be to man the walls, to fall upon the enemy at the ford, to harry his lands, or even to attack him in his castle. There is no fixed rule. The circumstances in each case make the rule. A war is not less a defensive war if you strike at your enemy in his own territory, or if you come to the aid of your ally, whose territory has been invaded or is threatened. In the circumstances which prevailed for a considerable number of years prior to the outbreak of the present war, it gradually became more and more obvious, that our soundest defence would be joint action with France upon her north-eastern frontier. For there, beyond any doubt, would Germany's supreme effort be made against the Triple Entente. If the attack failed at that point, it would be the heaviest and most disabling blow which our enemy could suffer. If, on the other hand, it succeeded, France and England would have to continue the struggle on terms immensely less favourable. This opinion was not by any means unanimously or clearly held; but during the summer of 1911 and subsequently, it was undoubtedly the hypothesis upon which those members of our Government relied, who were chiefly responsible for the conduct of foreign affairs. Unfortunately Parliament and the country had never accepted either the policy or its consequences; they had never been asked to accept either the one or the other; nor had they been educated with a view to their acceptance. At that time the error was exceedingly prevalent, that it is a more comfortable business fighting in your own country than in somebody else's. From this it followed that it would be folly to engage in what were termed
- 62. disapprovingly 'foreign adventures,' and that we should be wise to await attack behind our own shores. Recent events have wrought such a complete and rapid conversion from this heresy, that it is no longer worth while wasting words in exposing it. It is necessary, however, to recall how influential this view of the matter was, not only up to the declaration of war, but even for some time afterwards. As to the precise form of co-operation between the members of the Triple Entente in case of war, there could be no great mystery. It was obvious to any one who paid attention to what happened during the summer and autumn of 1911, that in the event of Germany attacking France over the Agadir dispute, we had let it be understood and expected, that we should send our Expeditionary Force across the Channel to co-operate with the French army on the north-eastern frontier. [1] It can hardly be overlooked, however, that this principle, rightly or wrongly interpreted, had something to do with the Crimean War (1854- 56) and with the British attitude at the Congress of Berlin (1878). [2] Viscount Milner in the United Service Magazine, January 1912. CHAPTER V THE MILITARY SITUATION (August 1911) The full gravity of the Agadir incident, though apparent to other nations, was never realised by the people of this country. The crisis arose suddenly in July 1911. Six weeks later it had subsided; but it was not until well on in the autumn that its meanings were grasped, even by that comparatively small section of the public who interest themselves in problems of defence and foreign affairs. From October onwards, however, an increasing number
- 63. THE EXPEDITION ARY FORCE began to awake to the fact, that war had only been avoided by inches, and to consider seriously—many of them for the first time in their lives—what would have happened if England had become involved in a European conflict. From various official statements, and from discussions which from time to time had taken place in Parliament, it was understood that our 'Expeditionary Force' consisted of six infantry divisions, a cavalry division, and army troops; [1] also that the national resources permitted of this force being kept up to full strength for a period of at least six months, after making all reasonable deductions for the wastage of war. Was this enough? Enough for what? ... To uphold British policy; to preserve Imperial security; to enable the Triple Entente to maintain the balance of power in Europe. These were vague phrases; what did they actually amount to? ... The adequacy or inadequacy of such an army as this for doing what was required of it—for securing speedy victory in event of war—or still better for preserving peace by the menace which it opposed to German schemes of aggression—can only be tested by considering the broad facts with regard to numbers, efficiency, and readiness of all the armies which would be engaged directly, or indirectly, in a European struggle. War, however, had been avoided in 1911, and not a few people were therefore convinced that the menace of the available British army, together with the other consequences to be apprehended from the participation of this country, had been sufficient to deter Germany from pursuing her schemes of aggression, if indeed she had actually harboured any notions of the kind. But others, not altogether satisfied with this explanation and conclusion, were inclined to press their enquiries somewhat further. Supposing war had actually been declared, would the British force have been sufficient—acting in conjunction with the French army—to repel a German invasion of France and Belgium, to hurl back the aggressors and overwhelm them in defeat? Would it have been sufficient to accomplish the more modest aim of holding the enemy at his own frontiers, or even— supposing that by a swift surprise he had been able to overrun Belgium—at any rate to keep him out of France?
- 64. NEUTRALIT Y OF ITALY When people proceeded to seek for answers to these questions, as many did during the year 1912, they speedily discovered that, in considerations of this sort, the governing factor is numbers—the numbers of the opposing forces available at the outbreak of war and in the period immediately following. The tremendous power of national spirit must needs be left out of such calculations as a thing immeasurable, imponderable, and uncertain. It was also unsafe to assume that the courage, intelligence, efficiency, armament, transport, equipment, supplies, and leadership of the German and Austrian armies would be in any degree inferior to those of the Triple Entente. Certain things had to be allowed for in a rough and ready way;[2] but the main enquiry was forced to concern itself with numerical strength. There was not room for much disagreement upon the broad facts of the military situation, among soldiers and civilians who, from 1911 onwards, gave themselves to the study of this subject at the available sources of information; and their estimates have been confirmed, in the main, by what has happened since war began. The Intelligence departments of London, Paris, and Petrograd—with much ampler means of knowledge at their disposal—can have arrived at no other conclusions. What the English War Office knew, the Committee of Imperial Defence likewise knew; and the leading members of the Cabinet, if not the whole Government, must be presumed to have been equally well informed. It was assumed in these calculations, that in case of tension between the Triple Entente and the Triple Alliance, the latter would not be able—in the first instance at all events—to bring its full strength into the struggle. For unless Germany and Austria managed their diplomacy before the outbreak of hostilities with incomparable skill, it seemed improbable that the Italian people would consent to engage in a costly, and perhaps ruinous, war—a war against France, with whom they had no quarrel; against England, towards whom they had long cherished feelings of friendship; on behalf of the Habsburg Empire, which they still regarded—and not altogether unreasonably—with suspicion and enmity. But although the neutrality of Italy might be regarded as a likelihood at the opening of the war, it could not be reckoned on with any certainty as a permanent condition.
- 65. SUPERIORIT Y OF GERMAN NUMBERS For as no one can forecast the course of a campaign, so no one can feel secure that the unexpected may not happen at any moment. The consequences of a defeat in this quarter or in that, may offer too great temptations to the cupidity of onlookers; while diplomacy, though it may have bungled in the beginning, is sure to have many opportunities of recovering its influence as the situation develops. Consequently, unless and until Italy actually joined in the struggle on the side of the Triple Entente, a considerable section of the French army would, in common prudence, have to be left on guard upon the Savoy frontier. In a war brought on by the aggressive designs of Germany, the only nations whose participation could be reckoned on with certainty—and this only supposing that Britain stood firmly by the policy upon which her Government had embarked—were Russia, France, and ourselves on the one side, Germany and Austria-Hungary on the other. It would certainly be necessary for Germany, as well as Austria, to provide troops for coast defences, and also for the frontiers of neutral countries, which might have the temptation, in certain circumstances, to deneutralise themselves at an inconvenient moment, if they were left unwatched. On the north and west were Denmark, Holland, and Belgium, each of which had a small field army, besides garrison and fortress troops which might be turned to more active account upon an emergency. On the south and east were Montenegro, Servia, and Roumania, whose military resources were on a considerable scale, and whose neutrality was not a thing altogether to be counted on, even before the Balkan war[3] had lowered the prestige of Turkey. In addition there was Italy, who although a pledged ally in a defensive war was not likely, for that reason, to consider herself bound to neutrality, benevolent or otherwise, if in her judgment, the particular contingencies which called for her support had not arisen at the outset. After taking such precautions as seemed prudent under these heads, Germany would then be obliged to detach for service, in co-operation with the Austrians in Poland, and along the whole eastern border, a sufficient number of army corps to secure substantial superiority over the maximum
- 66. forces which Russia, hampered by an inadequate railway system and various military considerations,[4] could be expected to bring into the field and maintain there during the first few months of the war. It was reckoned[5] after taking all these things into account, that Germany would have available, for the invasion of France, an army consisting of some ninety divisions—roughly, rather more than a million and three-quarters of men—and that she could maintain this force at its full strength—repairing the wastage of war out of her ample reserves—for a period of at least six months. It was assumed that the Kaiser, relying upon the much slower mobilisation of Russia, would undoubtedly decide to use the whole of this huge force in the west, in the hope that before pressure could begin to make itself felt in the east, France would either have been crushed, as she was in 1870, or so much mangled that it would be possible to send reinforcements of an overwhelming character to make victory secure in Poland. Against this German force of 1,800,000, France, according to the best information available, could put into the field and maintain at full strength for a similar period of six months about 1,300,000 men. But this was the utmost that could be expected of the French, and the initial discrepancy of 500,000 men was very serious. It precluded all reasonable hope on their part of being able to take the offensive, to which form of warfare the genius of the people was most adapted. It would compel them to remain on the defensive, for which it was believed at that time—though wrongly, as events have proved—that they were ill suited by temperament as well as tradition. If England joined in the war by land as well as sea the numerical deficiency would be reduced to 340,000 on the arrival of our Expeditionary Force. In this connection, as well as for other reasons, the attitude of Holland and Belgium, and that of Germany with respect to these two countries, were clearly matters of high importance. Holland had a field army of four divisions, and her interests could be summed up in the words, 'preservation of independence.' She would
- 67. POSITION OF BELGIUM naturally wish to avoid being actively embroiled in the war on one side or the other; and, fortunately for her, she had every reason to believe that her neutrality would not be disturbed or questioned. Her territories lay to one side of the probable campaign area, and moreover, whatever might be the ulterior designs of Germany with regard to western expansion, it was obvious that her immediate interests must necessarily lie in Dutch neutrality, which would be infinitely more useful to her than a Dutch alliance. For Holland holds the mouths of the Scheldt and Rhine, and so long as she remained neutral, it was anticipated that imports and exports would readily find their way into and out of Germany. This advantage would cease were Britain to establish a blockade of these inlets, as she would certainly do if they belonged to a hostile Power. In certain respects Belgium was in the same case as Holland. She likewise had a field army of four divisions, and her interests could be summed up in the words, 'preservation of independence.' But here all resemblance between the two countries ended. Belgium was not merely the southern portion (Holland being the northern) of that Naboth's vineyard, the possession of which German visionaries had proclaimed to be essential to Teutonic world-power. Belgium was more even than this. If the permanent possession of Belgian territory was a political object in the future, temporary occupation was no less a military necessity of the present. For in order that Germany might benefit in full measure by her numerical superiority, Belgian roads and railways were required, along which to transport her troops, and Belgian hills and plains on which to deploy them. If Germany were confined to the use of her own frontiers she would not only lose in swiftness of attack, but her legions would be piled up, one behind another, like a crowd coming out of a theatre. She needed space on which to spread out her superior numbers in order that her superior numbers might make certain of victory. There was an idea at this time (1911-12) that Germany would be satisfied to keep to the south-east of the fortified line of the Meuse— moving through Luxemburg and the mountains of the Ardennes—and that if Belgium saw fit to yield, under protest, to force majeure, the northern
- 68. INADEQUAC Y OF BRITISH ARMY region, containing the great plain of Flanders and all cities of importance, would be left inviolate. This theory was probably erroneous, for the reason that—as the event has shown—Germany required a greater space and more favourable ground, than would have been provided under this arrangement, in order to bring her great superiority to bear. With the French on the other hand there was no similar advantage to be gained by the violation of Belgian neutrality. From their point of view the shorter the battle front could be kept the better. If Belgium chose to range herself by the side of France as a willing ally it would undoubtedly be a great gain; but if she chose to remain neutral the French could have no object in invading or occupying her territories. It was assumed, and no doubt rightly, that, like Holland, Belgium would prefer to remain neutral—leaving the question of future absorption to take care of itself—provided she could do this without enduring the humiliation of allowing foreign armies to violate her soil. For she knew that, in the event of a French victory, her independence would remain assured; whereas, if the Germans were successful, she would have avoided awakening their hostility and giving them an excuse for annexation. But even if Belgium, under gross provocation, were forced to take sides against Germany, the deficit in numbers on the side of the Triple Entente would only be reduced by some eighty or a hundred thousand men. The deficit would still stand, roughly, at a quarter of a million men. In view of the foregoing considerations it was clearly absurd to think that our own small force was at all adequate, in a military sense, to deter Germany from engaging in a war of aggression. Had we been able, during the years 1912 to 1914, to see into the minds of the German General Staff we should probably have realised that this inadequacy was even greater than it appeared. We should then have known that the numbers of the Kaiser's striking force had been carefully understated; and that the amount of preparations in the way of material had been hidden away with an equal industry. We should also have learned, that the sending of our army abroad was viewed with scepticism in German military circles, as an event hardly
- 69. likely to occur. But even if our Expeditionary Force did go, it was altogether inadequate to redress the adverse balance; still more inadequate to bring an immediate victory within the range of practical possibility. It was inadequate to hold back the premeditated invasion, either at the German frontier, or even at the French frontier. It was inadequate to make Belgian resistance effective, even if that nation should determine to throw in its lot with the Triple Entente. As a matter of the very simplest arithmetic our land forces were inadequate for any of these purposes. They were unequal to the task of maintaining the balance of power by giving a numerical superiority to the armies of the Triple Entente. Our armaments therefore did not correspond with our policy. It was clear that they would not be able to uphold that policy if it were put to the supreme test of war. It was impossible to abandon our policy. It was not impossible, and it was not even in 1912 too late, to have set about strengthening our armaments. Nothing of the kind, however, was undertaken by the Government, whose spokesmen, official and unofficial, employed themselves more congenially in deriding and rebuking Lord Roberts for calling attention to the danger. Of course if it had been possible to place reliance upon the statement of the English War Minister, made little more than a year before war broke out,[6] that every soldier under the voluntary system is worth ten conscripts, we and our Allies would have been in a position of complete security. In that case our force of 160,000 would have been the equivalent of 1,600,000 Germans, and we should from the first have been in a superiority of more than a million over our enemies. Even if we could have credited the more modest assumption of the Attorney-General—made nearly four months after war broke out—that one volunteer was worth three 'pressed' men, the opposing forces would have been somewhere about an equality.[7] Unfortunately both these methods of ready-reckoning were at fault, except for their immediate purpose of soothing, or deluding the particular audiences to which they were addressed. The words were meaningless and absurd in a military sense; though conceivably they possessed some occult
- 70. THE THREE PERIODS OF WAR political virtue, and might help, for a time at least, to avert the retribution which is due to unfaithful stewards. Both these distinguished statesmen, as well as many of their colleagues and followers, were beset by the error of false opposites. A soldier who has enlisted voluntarily, and another who is a conscript or 'pressed' man, have equally to fight their country's enemies when they are ordered to do so. In both cases the particular war may be against their consciences and judgments; and their participation in it may therefore be involuntary. Of two men—equal in age, strength, training, and courage—one of whom believes his cause to be just, while the other does not, there can be no doubt that the former will fight better than the latter—even though the latter was enlisted under the voluntary system while the former was a conscript or 'pressed' man. In this sense the superiority of the 'voluntary' principle is incontestable. But is there any evidence to show, that either the original soldiers, or the new levies, of the German army are risking their lives in this war any less willingly than our own countrymen, who went out with the Expeditionary Force, or those others who have since responded to Lord Kitchener's appeal? Is there any reason to suppose that they are fighting any less bravely and intelligently?[8] Another matter of importance in these calculations with regard to the military strength of the Triple Entente and the Triple Alliance was the time limit. There are three periods in war. There is the onset of war, where swiftness of action is what tells most; there is the grip of war, where numbers of trained men are what tell most; and there is the drag of war, when what tells most is the purse. Speaking by the book, it is of course numbers which tell all the way through. At the beginning—in the onset—the aim is to hurl superior numbers at a vital point—taking the enemy by surprise, and thereby
- 71. RESULTS OF SUCCESS IN ONSET disordering his whole plan of campaign—very much as you knock a limpet off a rock, with a sharp unexpected blow. If this effort fails to settle matters, then we are in the grip. Here it is a case of sheer heavy slogging of all the available trained troops. The weaker side is driven to the defensive. It is found making use of every artificial and natural advantage to counteract the superiority which threatens it, and which must speedily prevail, if only it be superior enough. Finally, after a longer or shorter period of indecisive deadlock, the time comes when trained troops and material of war accumulated in advance begin to run short—when new levies, raised since the war broke out, begin to take the field, well or ill equipped, well or ill armed, as the case may be. When this stage is reached we are in the drag of war; and the side which can best afford to feed, clothe, and arm its fresh reinforcements stands at an enormous advantage. In 1870 war was announced on July 15th, and formally declared on the 19th. Three weeks later, on August 6th, the important battles of Woerth and Spicheren were won by the Germans. On September 2nd, the issue of the war was decided, when the Emperor of the French, with his main army, surrendered at Sedan. Metz fell in the last days of October, and Paris on the first day of March in the following year. In that war the onset settled everything. There was no real grip of the opposing forces. The German attack had been so swift, vigorous, and successful that France was knocked out in the first round. The speed with which great armies can be mobilised and hurled against one another has not diminished in the forty odd years which have elapsed since the débâcle. On the contrary, the art of war has been largely concerned in the interval with the vital question, how to get in the first deadly blow. The military view was, that probably not earlier than the fifteenth day— certainly not later than the twenty-first—a battle would take place which must be of the highest importance, and which might quite well be decisive. It might make ultimate German victory only a matter of time; or it might only determine whether the ensuing campaign was to be waged on French
- 72. LIMITATION S OF SEA or German soil—whether there was to be a German invasion of France or a Franco-British invasion of Germany. Consequently, if our Expeditionary Force was to render assistance at the critical time, it must reach its position on the frontier within a fortnight of the outbreak of war. As to the drag of war, the Triple Entente had the advantage, if that stage were ever reached. For the purses of England, France, and Russia were much longer than those of Germany and Austria. It was important, however, to remember that there would be no hope for us in the drag of war, if Germany could deliver a heavy enough blow at the beginning, as she did in 1870. These were the considerations as to time, which presented themselves to students of the military situation during the breathing space which followed upon the Agadir crisis. The substantial accuracy of this forecast was confirmed by what happened during August and September of last year. In 1914 war was declared by Germany on August 1st. For several days before she had been engaged actively in mobilisation. Three weeks later three important battles—on the road to Metz, at Charleroi, and at Mons[9]—were won by the Germans. If it had not been for the unexpected obstacle of Liège the last two engagements would in all probability have been fought at an even earlier date, and in circumstances much more unfavourable to the Franco-British forces. But in the early days of September, instead of the crushing defeat of Sedan, there was the victory of the Marne, and the Germans were forced to retreat to entrenched positions north of the Aisne. [10] The onset period was ended; but the issue had not been settled as in 1870. France and England had not been knocked out in the first round. To this extent the supreme German endeavour had miscarried. Nevertheless a great advantage had been secured by our enemies, inasmuch as it was now apparent that the ensuing campaign—the grip of war—would be contested, not on German soil, but in France and Belgium. The value of the assistance which the British Navy would be able to render to the cause of the Triple Entente was a
- 73. POWER consideration of the highest importance. But while the fleet, if the national confidence in it were justified, would render invaluable assistance to military operations, it was necessary to bear in mind—what Englishmen in recent times have been very apt to forget—that no success at sea, whether it consisted in the wholesale destruction of hostile ships, or in an absolute blockade of the enemy's coast, could by itself determine the main issue of a European contest of this character. Disaster in a land battle could not be compensated for, nor could the balance of power be maintained, by any naval victory. War would not be brought to an end favourable to the Triple Entente, even by a victory as complete as that of Trafalgar. It is also well to remember that peace came, not after Trafalgar, but after Waterloo, nearly ten years later. The strange idea that the security of the British Empire can be maintained by the Navy alone, seems to be derived by a false process of reasoning, from the undeniable truth, that the supremacy of our Navy is essential to our security. But though it is essential—and the first essential— it is not the only essential of security. An insular Power, largely dependent on sea-borne food supplies and raw materials for its industries—a Power which governs an empire in the East, which has dependencies scattered in every sea, which is politically united with immense but sparsely peopled dominions in the four quarters of the globe—must keep command of the sea. If that supremacy were once lost the British Empire, as an empire, would come to an end. Its early dissolution would be inevitable. Therefore it is true enough to say that if the German Alliance—or any other alliance—were to win a decisive naval victory against Britain, it would end the war completely and effectively so far as we were concerned. But the converse is not the case, and for obvious reasons. In a contest with a continental enemy who conquers on land, while we win victory after victory at sea, the result will not be a settlement in our favour, but a drawn issue. And the draw will be to his advantage, not our own. For having overthrown the balance of power by reason of his successful campaign and invasions, he will then be free to concentrate his whole energies upon wresting away naval supremacy from the British Empire. In time the Sea
- 74. Power which is only a Sea Power will be overborne with numbers, and finally worsted by the victorious Land Power. For how is it possible to fight with one hand against an enemy with two hands? The fleets of Europe which at last must be combined against us, if we allow any rival to obtain a European predominance, are too heavy odds. German preparations alone were already causing us grave anxiety nearly three years before the Agadir crisis occurred. How then could we hope to build against the whole of Europe? Or even against half of Europe, if the other half remained coldly neutral? [1] In all about 160,000 men, of whom some 25,000 were non- combatants. [2] Such, for instance, as the fact that the time-table of German mobilisation appeared to be somewhat more rapid than that of the French, and much more so than that of the Russians. [3] The first Balkan war broke out in the autumn of 1912. [4] Russia had anxieties of her own with regard to the intentions of Roumania, of Turkey in Persia and the Caucasus, and of China and Japan in the Far East. [5] These calculations were worked out in various ways, but the net results arrived at were always substantially the same. In view of the fact that the main conclusions have been amply proved by the results of the present war, it does not seem worth while to weary the reader with more sums in arithmetic than are absolutely necessary. [6] Colonel Seely at Heanor, April 26, 1913. [7] Sir John Simon (Attorney-General and a Cabinet Minister), at Ashton-under-Lyne, November 21, 1914.... This speech is instructive reading. It is also comforting for the assurance it contains, that if the speaker approved of our taking part in this war (as he vowed he did) his audience might rest satisfied that it was indeed a righteous war; seeing that war was a thing which, on principle, he (Sir John Simon) very much reprehended. And yet we are not wholly convinced and reassured. There is a touch of over-emphasis—as if perhaps, after all, the orator needed the support of his own vehemence to keep him reminded of the
- 75. righteousness. The pacifist in war-paint is apt to overact the unfamiliar part. One wonders from what sort of British officer at the front the Attorney-General had derived the impression that 'one' of our own voluntary soldiers—gallant fellows though they are—is the equal of 'three' of the Germans who face him, or of the Frenchmen who fight by his side.... This speech puts us not a little in mind of Evangelist's warning to Christian, with regard to Mr. Legality's fluent promises to relieve him of his burden—There is nothing in all this noise save a design to beguile thee of thy salvation. [8] Sir John Simon clinched his arithmetical calculation of 'three' to 'one,' by stating that 'the Kaiser already knew it'; and this reassuring statement was received with 'laughter and cheers.' The laughter we can understand. [9] The battle in Northern Alsace was fought on August 21 and 22. A French army was driven back at Charleroi on the 22nd, and the British at Mons on the 23rd. [10] September 6-12. CHAPTER VI THE MILITARY SITUATION (August 1914) Such was the position of affairs at July 1911, as it appeared to the eyes of people who—during the ensuing period—endeavoured to arrive at an understanding of the problem without regard to the exigencies of party politics. Between that date and July 1914, when war broke out, various changes took place in the situation. The general effect of these changes was adverse to Britain and her allies. In 1911 the German estimates provided for considerable increases, especially in artillery and machine-guns. The peace strength of the Army was raised.
- 76. MILITARY INCREASES In the following year, 1912, further additions were made to the peace strength, and two new army corps were formed out of existing units—one for the Polish, the other for the French frontier. Artillery and machine-guns were very greatly increased in the ordinary estimates of that year, and again in those of 1913. In addition, Germany at the same time added a squadron to her fleet in the North Sea, by arranging to keep more ships permanently in commission. But early in 1913 it became known, that the German Government was about to introduce an Army Bill, providing for immense and sensational additions. The sum of £50,000,000 was to be raised by loan for initial expenditure. The increased cost of upkeep on the proposed new establishment would amount to £9,500,000 per annum. Sixty-three thousand more recruits were to be taken each year. The total peace strength of the Army was to be raised by approximately 200,000 men. Nearly four millions sterling was to be spent on aircraft, and ten and a half on fortifications; while the war-chest was to be raised from six to eighteen millions. Twenty-seven thousand additional horses were to be purchased. These proposals were timed to take effect the same autumn; so that by the following Midsummer (1914), the military strength of Germany would have reaped the main benefit which was anticipated from the enormous additions. It was not in the power of France to increase the actual total of her numbers, because for many years past she had already taken every man who was physically fit for military service. About eighty per cent of the young Frenchmen who came each year before the revision boards had been enlisted; whereas in Germany—up to the passing of the new Army Law— considerably less than fifty per cent had been required to serve. The German Army as a consequence was composed of picked men, while the French Army contained a considerable proportion who were inferior both in character and physique. But in the face of the new German menace France had to do the best she could. She had to do it alone, for the reason that the British Government
- 77. entertained conscientious and insuperable objections to bearing its due share of the burden. Already, prior to the sensational expansion of Germany in 1913, France had endeavoured to counteract the current yearly increases in the military estimates of her neighbour, by various reorganisations and regroupings of active units, and by improvements calculated to improve the efficiency of the reserves. But when information was forthcoming[1] as to the nature and extent of the developments proposed under the German Army Bill of 1913, it was at once realised that more drastic measures were essential to national safety. Before the German projects were officially announced, the French Government took the bold step of asking the legislature to sanction a lengthening of the period of active military service from two years to three, and an extension of the age limit of the reserves from forty-seven to forty- nine. Power was also taken to summon, in case of emergency, the annual contingent of recruits a year before their due time. Increases in artillery, engineers, railways, barrack accommodation, and subsidiary services were asked for and obtained. The cost of these, when the whole sum came to be calculated, was found to amount to £32,000,000. Apart, therefore, from material preparations of one kind and another, Germany was taking steps to add 200,000 men to her striking force, and the intentions of France were approximately the same. In the case of Germany, however, the increases of strength would be operative by Midsummer 1914, while with France they would not take effect until two years later.[2] Germany, moreover, was arranging to take 63,000 more recruits annually. France was unable to obtain any more recruits, as she already took all that were fit to bear arms. The increase in her striking force was made mainly at the expense of her reserves. Year by year, therefore, the numerical inferiority of France must become more marked. Russia meanwhile was proceeding with her programme of military extension and reorganisation which had been decided on after the Japanese war. A great part of her expenditure was being devoted to the improvement