Neural Stem Cell Differentiation

ARTICLE IN PRESS

Journal of Molecular Structure xxx (2010) xxx–xxx

Contents lists available at ScienceDirect

Journal of Molecular Structure
journal homepage: www.elsevier.com/locate/molstruc

Neural differentiation of mouse embryonic stem cells studied by FTIR spectroscopy
Waraporn Tanthanuch a, Kanjana Thumanu a, Chanchao Lorthongpanich b, Rangsun Parnpai b,*,
Philip Heraud c,*
a
Synchrotron Light Research Institute (Public Organization), Nakhon Ratchasima 30000, Thailand
b
Embryo Technology and Stem Cell Research Center, School of Biotechnology, Institute of Agricultural Technology, Suranaree University of Technology, Nakhon Ratchasima, Thailand
c
Center for Biospectroscopy, and the Monash Immunology and Stem Cell Laboratories, Monash University, Clayton, Victoria 3800, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Embryonic Stem-derived Neural Cells (ESNCs) hold potential as a source of neurons for a cell-based ther-
Received 15 September 2009 apy for the treatment of brain tumors, and other neurological diseases and disorders in the future. The
Received in revised form 6 January 2010 sorting of neural cell types is envisaged to be one of the most important processed for clinical application
Accepted 7 January 2010
of these cells in cell-based therapies of the central nervous system (CNS). In this study, laboratory-based
Available online xxxx
FTIR and Synchrotron-FTIR (SR-FTIR) microspectroscopy were used to identify FTIR marker for distin-
guishing different neural cell types derived from the differentiation of mouse embryonic stem cells
Keywords:
(mESCs). Principal Component Analysis (PCA) and Unsupervised Hierarchical Cluster Analysis (UHCA)
FPA-FTIR
Synchrotron-FTIR
were shown to be able to distinguish the developmental stage of mESCs into three cell types: embryoid
Embryonic stem cells bodies (EBs), neural progenitor cells (NPCs), and ESNCs. Moreover, PCA provided the mean for identifying
Neural differentiation potential FTIR ‘‘marker bands” that underwent dramatic changes during stem cell differentiation along
Principal Component Analysis (PCA) neural lineages. These appeared to be associated with changes in lipids (bands from CH2 and CH3 stretch-
Unsupervised Hierarchical Cluster Analysis ing vibrations at $2959, 2923 and 2852 cmÀ1) and proteins (changes in the amide I band at $1659 and
(UHCA) 1637 cmÀ1). The results suggested that lipid content of cells increased significantly over the time of dif-
ferentiation, suggesting increased expression of glycerophospholipids. Changes in the amide I profile,
suggested concomitant increases in a-helix rich proteins as mESCs differentiated towards ESNCs, with
a corresponding decrease in b-sheet rich proteins, corresponding with changes in cytoskeleton protein
which may have been taking place involved with the establishment of neural structure and function.
Crown Copyright Ó 2010 Published by Elsevier B.V. All rights reserved.

1. Introduction tenin, musashi) [16], and in fact no single positive marker specific
for neural stem cells is currently available. Therefore, in order to
Neural stem cells (NSCs) hold greater promise for treatment of increase the certainty of correct selection of the desired cell type,
brain tumors, and other pathologies of the CNS [1,2]. NSCs can be several antibodies are required for the identification process for
derived from various pre-cursor cell types, for instance: embryonic one particular cell type [17]; leading to high detection costs, as
stem (ES) cells [3]; stem-like cells obtained from the embryonic well as the exercise being very time consuming. In addition, large
CNS [4–6]; the Peripheral Nervous System (PNS) [7–9]; stem-like or fragile cells have been shown to be damaged by FACS [18].
cells from adult brain [10,11]; and spinal cord [12,13]. Immunoflu- FTIR microspectroscopy is a powerful technique, which has been
orescent staining of specific cell-surface markers in combination widely used in biophysical and biochemical research, proven to pro-
with fluorescence-activated cell sorting (FACS) [14,15] have typi- vide sensitive and precise measurement of biochemical changes in
cally been used to discriminate neural cell types derived from stem biological cells and tissues [19]. This approach provides structural
cells. Even though FACS is a potential approach for the discrimina- information of biological molecules such as proteins, nucleic acids,
tion of neural cells destined for clinical application, many types of carbohydrates and lipids, allowing detection, identification, and
neural stem cells have their unique expression and there exists quantification of changes in these macromolecular cellular compo-
many hitherto unidentified markers. Moreover, many of the mark- nents [20]. Therefore, characterization of neural stem cells could
ers previously thought to be specific for stem cells have been be performed at a level of biological molecule differences, which
shown to be expressed by differentiated cells (e.g., nestin, nucleos- may provide the intrinsic ‘‘biochemical signatures” of particular cell
types. Moreover, detection of many different biomolecules changes
at the same time during one measurement can be accomplished. This
* Corresponding authors. Tel./fax: +66 44223164 (R. Parnpai), tel.: +61
technique is very rapid, non-destructive and non-invasive for the
399055771; fax: +61 399055613 (P. Heraud).
E-mail addresses: rangsun@g.sut.ac.th (R. Parnpai), Phil.Heraud@med.monash. sample, and does not require special time consuming sample prepa-
edu.au (P. Heraud). ration. Recently a few publications have started to appear in the lit-

0022-2860/$ - see front matter Crown Copyright Ó 2010 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.molstruc.2010.01.007

Please cite this article in press as: W. Tanthanuch et al., J. Mol. Struct. (2010), doi:10.1016/j.molstruc.2010.01.007

ARTICLE IN PRESS

2 W. Tanthanuch et al. / Journal of Molecular Structure xxx (2010) xxx–xxx

erature reporting the use of FTIR spectroscopy to study a number of science) for 30 min, followed by incubation for 1 h with blocking
different stem cell types e.g., murine embryonic stem cells [21], hu- buffer consisting of 0.2% Triton X-100 (Sigma), 3 mM sodium azide
man corneal stem cells [22], human mesenchymal stem cells [23] (Sigma), 0.1% saponin (Sigma), 2% BSA (Sigma), and 5% house ser-
and cardiac stem cell [24]. This is the first study to our knowledge um (HyClone) in PBS, without added calcium and magnesium.
which has reported FTIR spectroscopic changes in mESCs undergo- The samples were then thoroughly washed with PBS and incubated
ing differentiation towards neural lines. with primary antibodies at 1 in 100 dilutions in PBS. Primaries
Focal Plane Array (FPA) detectors based FTIR (FPA-FTIR) has the antibodies for specific cell types were as follows: NPCs were
ability to sample a large sample area with high spatial resolution stained with nestin; successful differentiation of ESNCs was con-
(5–20 lm), and significant time savings, compared to mapping firmed by the expression of neuron specific bIII-tubulin (TuJ1), tyro-
using single element IR detectors [25]. Despite these advantages, sine hydroxylase (TH) and Musashi (MUS). The samples were then
however, FPA-FTIR still has limited spatial resolution to be effec- incubated with secondary antibodies conjugated with appropriate
tive in many experiments, especially those where the sample con- fluorescent tag. Cells were subsequently incubated with the nu-
tains low concentrations of the analyte in question or small sample cleic acid dye, Hoechst 33342 (H3570; Molecular Probes, Invitro-
size below the spatial resolution of the technique. Synchrotron- gen), for 30 min and examined by fluorescence microscopy.
based FTIR microspectroscopy (SR-FTIR) has proven to be a more
powerful technique to overcome such problems [26–28]. The high 2.3. Sample preparation for FTIR microspectroscopy
brightness (estimated to be 100–1000 times greater than conven-
tional laboratory-based thermal globar sources) has been proven The spectroscopic analysis of cell clumps (>100 cells/clump)
to enable superior spatial resolution (2.5–10 lm) compared to lab- was performed using focal plane array microspectroscopic imag-
oratory-based instruments enabling the capacity to study single ing. The cellular clumps were prepared by washing EBs and neur-
cell [29] and large intracellular organelles within cells [30] with ospheres for three times with 0.9% NaCl. The cellular clumps were
good signal to noise ratios. then deposited onto indium tin oxide-coated, silver-doped glass
The aim of this study was to use FTIR microspectroscopy to slide (MirrIR, Yienta Sciences, OH, USA) air-dried, and stored in a
examine the differentiation of mESCs into ESNCs. mESCs are plu- desiccator until spectra were acquired.
ripotent cells which are universally recognized as the most prom- For single cell analysis using SR-FTIR: single cell suspensions
ising source for neural progenitor proliferation as they have the were prepared from EBs and neurospheres by trypsinization with
potential to differentiate into any cell type [31–33]. We employed 0.05% Trypsin in 0.5 mM EDTA solution for 3 min at room temper-
two types of FTIR microspectroscopy: FPA-FTIR, which was used to ature. The trypsinized cells were then harvested and washed three
acquire spectra from clumps of cells, while the superior signal to times with 0.9% NaCl and deposited onto IR-transparent 2 mm
noise ratio of SR-FTIR microspectroscopy enabled the targeting of thick barium fluoride windows, air-dried, and stored in a desicca-
single cells. We proposed that the FTIR spectroscopy has the poten- tor until spectra were acquired.
tial to provide unambiguous discrimination of ESNCs from their
stem cell pre-cursors. Moreover, the FTIR technique may provide
new insights into the biochemical changes occurring during the 2.4. FPA-FTIR microspectroscopy
differentiation process.
FTIR spectral images were acquired at the Center for Biospec-
troscopy at Monash University in Melbourne, Australia, using a
2. Experimental Varian FTIR spectrometer (Model FTS 7000; Varian Inc., Palo Alto,
CA, USA) coupled with an infrared microscope (model 600 UMA;
2.1. Mouse ES cell culturing and differentiation Varian) using a 15 Â Varian objective equipped with a 64 Â 64 pix-
el MCT (HgCdTe) liquid nitrogen cooled FPA detector (Varian). FTIR
mESCs (mESCs-SUT-1, strain C57BL/6) were cultured in suspen- spectra were acquired with 256 co-added scans, at 6 cmÀ1 spectral
sion of mouse fetal fibroblasts for 7 days for the formation of resolution, with binning of the signal from each four adjacent pixel
embryoid bodies (EBs). EBs were plated onto gelatin-coated tissue in the FPA (resulting in a 32 Â 32 pixel array image). The area of the
culture dishes and cultured in N1 medium that consisted of MEM/ sample from which single spectra were acquired was approxi-
F12 (Invitrogen) supplemented with minimum essential amino mately 11 lm Â 11 lm. Apodization was performed using a
acid (Invitrogen), 200 mM of L-glutamine (Invitrogen) and N2 sup- Happ–Genzel function.
plement (Invitrogen). Culturing was continued for 7 days to allow
the differentiation of cells into neural progenitor cells (NPCs). NPCs
2.5. Synchrotron infrared microspectroscopy (SR-FTIR)
were subsequently sub-cultured for 14 days with N2 medium that
was composed of N1 medium supplemented with 20 lg/mL basic
Measurements were carried out on the IR Microspectroscopy
fibroblast growth factor (bFGF). Mature ESNCs were established
Beamline at the Australian Synchrotron, Melbourne, Australia.
from NPCs by plating onto N3 medium composed of DMEM/F12
SR-FTIR absorption spectra were acquired using a Bruker Vertex
supplemented with 1% fetal bovine serum (FBS; Hyclone) and
80v-IR spectrometer coupled with an IR microscope (Hyperion
B27 supplement (Invitrogen) for 7 days. For each step of culturing,
2000) and connected with MCT (HgCdTe) detector cooled with li-
the media was changed media every second day.
quid nitrogen over a wavenumber range from 4000 cmÀ1 to
Samples were collected from different culture stages as follows:
600 cmÀ1. Spectral acquisition was made in transmission mode
EBs culturing for 2 days (EBs day2); EBs culturing for 7 days (EBs
with a focal plane aperture size of 5 lm Â 5 lm at 6 cmÀ1 spectral
day7); EBs cultured in N1 medium for 7 days (N1day7); NPCs cul-
resolution. Thirty two interferograms were co-added and con-
tured in N2 medium for 14 days (N2 day14); and, NPCs cultured in
verted to absorbance spectra using OPUS 6.5 software (Bruker Op-
N3 medium for 7 days (N3 day7).
tics Ltd., Ettlingen, Germany). Baseline correction for all the spectra
was done using the rubber band baseline correction method and
2.2. Immunocytochemistry spectra was normalized to the amide I (1654 cmÀ1) absorbance
band using the OPUS 6.5 (Bruker) software. For each sample, mean
Samples were washed twice in phosphate buffer saline (PBS), spectra were calculated from at least twenty different
fixed onto glass slides with 4% Q4 paraformaldehyde (PFA; Poly- measurements.


ARTICLE IN PRESS

W. Tanthanuch et al. / Journal of Molecular Structure xxx (2010) xxx–xxx 3

2.6. Principal Component Analysis (PCA) peaks that were overlapping in the raw spectra (Fig. 2b and c).
The second derivative spectrum gives a negative peak for every
The spectra were processed by taking the second derivative band and shoulder in the absorption spectrum, thus it allows easier
using the Savitzky–Golay algorithm with nine points of smoothing identification of individual peaks in complex spectra.
and normalized with Extended Multiplicative Signal Correction The intensity of the msCH2, masCH2 stretching modes
(EMSC). PCA was performed using The Unscrambler software (ver- ($2850 cmÀ1 and 2920 cmÀ1, respectively) were observed to in-
sion 9.2, CAMO Software AS, Oslo, Norway) employing the com- crease dramatically during the differentiation of neural progenitors
bined spectral ranges of 3000–2800 cmÀ1 and 1750–900 cmÀ1. and mature neural cells. Additionally, the carbonyl band associated
Six PCs were selected for data analysis. Score plots (2D) were with the lipid head-group (msC@O stretching at 1740 cmÀ1) showed
examined for any clustering of spectra and loading plots identified significant spectral changes in both peak position and intensity
the spectral features explaining any observed clustering in the through ES cell differentiation (Fig. 2c). If these spectral changes
scores plots in each data set. were associated with alterations in lipids membranes then it is log-
ical that the changes observed might be associated with glycero-
2.7. Unsupervised Hierarchical Cluster Analysis (UHCA) phospholipids (GPLs), e.g., phosphatidylcholine (PC),
phosphatidylethanolamine (PE), phosphatidylserine (PS), and
UHCA of FTIR spectral data sets was done using Ward’s algo- phosphatidylinositol (PI), which are membrane lipids responsible
rithm, which utilises a matrix defining inter-spectral distances to for neural cell proliferation, differentiation, and signal transduction
identify the two most similar IR spectra. These spectra are com- [34,35].
bined into a new object (called a ‘‘cluster” or ‘‘hierarchical group”). The amide I band ($1700–1600 cmÀ1) was observed to shift sig-
The spectral distances between all remaining spectra and the new nificantly (Fig. 2c) during the differentiation process from EBs
cluster are then recalculated. UHCA was performed using the Cyto- through NPCs to ESNCs, which could be associated with secondary
spec 1.3 IR imaging software (Cytospec Inc., Boston, MA) carried structure changes in proteins. It should be stated that the attribu-
out using the spectral regions from 3000–2800 cmÀ1 to 1750– tion of loadings in the amide region of the spectrum to protein sec-
900 cmÀ1. ondary structure should be approached with caution in FTIR
spectroscopic studies of cells and tissues as recent studies have
shown this region of the spectrum to be sensitive to light scatter-
3. Results and discussion ing, particularly in synchrotron IR measurements when particles
sizes range from cell nuclei to whole cells [36]. Nevertheless, we
3.1. ES cells differentiation and morphology observed none of the characteristic spectral baseline features asso-
ciated with Mie scattering in either the FPA or synchrotron FTIR
mESCs derived from C57BL/6 embryo were demonstrated to be spectra (data not shown) and hence it was concluded that the
fully pluripotent (data not shown) as they were able to differenti- changes in the amide I band observed during the differentiation
ate in vitro into EBs comprising the three embryonic germ layers. of neural cells from ES progenitors were associated with secondary
The 7-days EBs appeared in light microscopy images as oval, shiny structure changes related to changes in protein composition that
colonies with clear boundaries (data not shown). Immunostaining occurred during the differentiation process. Indeed, changes ob-
of the 7-day EBs was positive for a-fetoprotein (AFP) an endoderm served in the amide I band shape during ES cells differentiations
maker and vimentin an ectoderm marker (data not shown). could be related to five components of the amide I band seen in
For neural differentiation, EBs were further cultured in N1 med- spectroscopic studies of isolated proteins [37]. The first component
ium for 7 days leading to differentiation into NPCs. Immunostain- at 1637 cmÀ1 was assigned to intramolecular b-sheet. Absorbance
ing of NPCs was positive for nestin, a specific neural cell-surface in the range from 1650 cmÀ1 to 1659 cmÀ1 was associated with the
marker (Fig. 1b). These results indicated that the neural selection presence of a-helix secondary structure. The third component at
protocol effectively directed differentiation towards NPCs. After 1660 cmÀ1 was assigned to turn and bends in the peptide second-
in vitro differentiation of NPCs into N2 and N3 medium, a neural ary structure. The fourth component at 1680 cmÀ1 was assigned to
phenotype was observed exhibiting neural tube-like structures b-turn structure, while the last component at 1690 cmÀ1 was as-
developed in the embryoid body outgrowths (Fig. 1d–o). The living signed to anti-parallel b-sheet.
population of neural cell showing that every cell (cell nucleus Day 2 EBs were observed to have low intensity in relation to the
stained blue, Fig. 1f, i, l, o) expresses the neural cell markers, a-helix band component ($1650–1652 cmÀ1), which increased
bIII-tubulin (Fig. 1e), tyrosine hydroxylase (Fig. 1h), Musashi gradually during the 7 days of cell culturing. Significant changes
(Fig. 1k), and CHAT (Fig. 1n), confirming that these cells were fully in a-helix structure in terms of band intensity and peak position
differentiated into mature neurons. were observed in the differentiated neural stem cells after subject-
ing EBs to culture in N1, N2 and N3 medium. The a-helix band was
3.2. FTIR microspectroscopy observed to shift from 1650 cmÀ1 to 1657 cmÀ1 when culturing
EBs in N1 medium for 7 days. Similarly, shifts of a-helix compo-
In this experiment we aimed to identify FTIR ‘‘marker bands” nent band to 1659 cmÀ1 was observed in NPCs and ESNCs follow-
related to biochemical changes that occurred during the differenti- ing 14 days of differentiation in N2 medium and 7 days of
ation of mESCs leading to neural cells. FPA-FTIR microspectroscopy differentiation in N3 medium, respectively. The most dramatic in-
was performed to analyse clump of cells. Quality spectra were se- crease in intensity of a-helix component band was observed dur-
lected by taking 2nd derivative spectra, normalizing with EMSC, ing the differentiation from EBs through to the fully
and conducting PCA in the spectral range of 3000–2800 cmÀ1 and differentiated neural cells. The intramolecular b-sheet component
1750–900 cmÀ1 containing bands of biological origin. Spectra were band ($1637 cmÀ1) of day 2 EBs was observed to have a lower
rejected (< 5% of the total number acquired) on the basis of high intensity compared with the a-helix component band, whilst in
residual variance across the first six PCs. The average FTIR absor- day 7 EBs, the intramolecular b-sheet component band intensity
bance spectra of mESCs cells and differentiated neural cells are increased significantly to a higher intensity than a-helix band.
shown in Fig. 2a. As it was problematic to compare precisely peak Notably, the intensity of intramolecular b-sheet decreased steadily
positions of bands by visual inspection of raw spectra, spectra were during cell differentiation from EBs to NPCs, and then dramatically
processed by taking the second derivative, which better resolved decreased during differentiation from NPCs to ESNCs. By contrast,


ARTICLE IN PRESS


Fig. 1. Morphological changes occurring during the differentiation of mESCs into neural cells types. The first column from the left is the transmission light image, the second
column contains images showing immunostaining with antibodies specific for each cell type, and the third column shows the overlay of immunostaining and the DNA
fluorescence image (stained with Hoechst 333342 causing nuclei to appear blue). The first row (a–c) represents NPCs cells 7 days after plating in N1 medium showing that the
cells were nestin positive (b). The later images show images representing mature ESNCs neural cells at N3 culture stage stained with bIII-tubulin (e), tyrosine hydroxylase (h),
Musashi (k), and CHAT (n). Bars = 100 lm (a–c); 50 lm (d–l); 20 lm (m–o).

the b-turn component band at 1680 cmÀ1 and anti-parallel b-sheet pressed in differentiated neurons cells [38]. Indeed, the expression
component band at 1690 cmÀ1 showed only minor changes during of b-III tubulin and tyrosine hydroxylase was confirmed in mature
the differentiation process. These results suggested that during the ESNCs by immunostaining in our study (Fig. 1e).
differentiation process of neural cells from embryonic stem cells, The results described above reported the biochemical changes
there is an increased expression of specific proteins which have during neural cell differentiation by interrogating clumps of cells
predominantly a-helix structure at the expense of those with a using FPA-FTIR microspectroscopy. It was viewed as important to
predominantly b-sheet structure. observe whether it was possible to discriminate differences during
We hypothesise that the increased expression of a-helix rich neural differentiation at the single cell level. To achieve this aim we
protein in ESNCs compared with ES progenitor cells might be con- used SR-FTIR spectroscopy with a focused spot size of 5 lm Â 5 lm
nected to the expression of cytoskeleton protein that have been in order to target individual cells. Comparison of spectra acquired
shown in previous studies to be important for the establishment between SR-FTIR and FPA-FTIR were remarkably similar in terms of
of neural structure and function. For instance, neurofilament peak positions, corroborating the FPA-FTIR results. Similar to what
(NF), tubulin, and actins proteins are a major constituent of the was observed with clumps of cells, the single cell measurements
contractile apparatus, which have been reported to be highly ex- showed that the main changes seen between ES cells and differen-


ARTICLE IN PRESS


Fig. 3. Average 2nd derivative SR-FTIR spectra of mESCs suspensions from EBs and
those differentiated along neural lineages to the N1 and N2 culture stages, showing
differences in the spectral region attributable mainly to lipid absorbance from
2800 cmÀ1 to 3000 cmÀ1.

in the intensity of the bands in spectral region arising from the
masCH3, masCH2 and msCH2 vibrational modes at 2958 cmÀ1,
2922 cmÀ1 and, 2851 cmÀ1, respectively. The dramatic increase in
the absorbance of these bands in NPCs at the 14-day N2 culture
stage compared to NPCs from 7-day N1 culture stage and EB stage,
may have indicated the formation of higher concentrations of pre-
dominantly longer chain fatty acids in the later differentiation
stages. This ﬁnding is consistent with the results obtained from
clumps of cells measured using FPA-FTIR microspectroscopy.

3.3. Statistical analysis

Principal Component Analysis (PCA) was employed to assess the
consistency across the entire spectral data set of the spectral differ-
ences observed between average spectra, described above. PCA
was performed on second derivative spectra from all 236 spectra
acquired from the differentiated mESCs. Analysis demonstrated
that PC1 vs. PC2 consistently provide the best clustering of the dif-
ferent classes of ES cells and neural pre-cursors that were studied.
A two-dimensional PC1 vs. PC2 scores plot is shown in Fig. 4. Four
clusters were observed in the PC1vs. PC2 scores plot: 2 days-EB
spectra, 7 days-EB spectra, 7 days NPC in N1 medium spectra,
and spectra from neural cells following 14 and 7 days of differenti-
ation in N2 and N3 medium, respectively. The ﬁrst two PCs ex-
plained 76% of the total variance in the data set (68% and 8% of
the total variance explained by PC1, and PC2, respectively).
Analysis of the PCA loadings plot (Fig. 4b) was used to deter-
mine regions of the FTIR spectrum which most contribute to the
clustering observed in the scores plot (Fig. 4a). The amide band re-
gion from proteins (1700–1500 cmÀ1), and the C-H stretching re-
Fig. 2. Average raw FTIR spectra of mESCs in different stages of differentiation gion attributable to lipids (3000–2800 cmÀ1) were heavily loaded
towards the formation of ESNCs using FPA microspectroscopy (a). Spectra were
for PC1 which separated EB spectra from neurosphere spectra
normalized over the range of 4000–1000 cmÀ1. (b) Shows an enlargement of the
lipid spectral region from 2800 cmÀ1 to 3000 cmÀ1. (c) Shows amides I and II (Fig. 4b). Spectra from EB cells (after 2-days and 7-days of culture)
spectral region from 1500 cmÀ1 to 1800 cmÀ1 for average second derivative spectra can be distinguished from progenitor neural cell types by having
of mESCs in different stages of the ESNCs differentiation process after nine points of positive PC1 scores (Fig. 4a). This can be explained by these spectra
smoothing and normalized with EMSC.
having highest negative value for PC1 loading (Fig. 4b) with vari-
ables between 1637 cmÀ1 and 1648 cmÀ1, possibly indicating that
the intramolecular b-sheet and a-helix band are most strongly
tiated neural cells lie in the region of the spectrum predominately responsible for discrimination. Moreover, spectra from differenti-
associated with lipid absorbance in the wavenumber range from ated neural cells propagated in N1, N2 and N3 medium are clearly
2800 cmÀ1 to 3000 cmÀ1, as shown in the enlargement of the of separated by the negative correlation of PC1 score (Fig 4a), which
second derivative spectra in Fig. 3. There were clear differences have positive PC1 loadings (Fig. 4b) at 1660 cmÀ1, suggesting a pre-


ARTICLE IN PRESS


Fig. 5. PCA for SR-FTIR spectra of single cells. PC1 vs. PC2 scores plot of the
population data for spectra of mESCs from dissociated EBs and from cells
differentiated along neural lineages to the N1 and N2 culture stages using the
spectral region from 3600 cmÀ1 to 1000 cm–1 for the PCA.

(Fig. 6). Two classes representing EBs (2-days and 7-days) and dif-
ferentiated cells (7-days N1, 14-days N2 and 7-days N3) can be dis-
criminated by UHCA below a linkage distance of 3. The 2-days EBs
and 7-days EBs have a closer spectral similarity with a linkage dis-
tance just below 1, whereas, NPCs (7-days N1) separated into a
cluster that was distinctly separated from NPCs (14-days N2) and
ESNCs (7-days N3).

4. Conclusions

Fig. 4. PCA score plot (a) and loadings plot (b) for individual spectra of mESCs and ESNCs hold great potential for clinical applications involving
cells differentiated to the neural progenitor stage acquired using FPA-FTIR. Spectra therapeutic regeneration and repair in pathology and injuries of
were processed to the 2nd derivative over the spectral range between 3000–
the CNS. Evaluation of the purity and state of differentiation of cell
2800 cmÀ1 and 1750–900 cmÀ1.
populations destined for transplantation will be paramount.
Importantly, differentiated neural cells destined for therapeutic
dominance of a-helical protein structure in neurosphere cells, cor- treatment must be distinguished from ES cells to prevent trans-
roborating the conclusions drawn from examination of the average plantation of potentially teratocarcinoma-forming cells [39,40].
spectra (Fig. 2). This work has demonstrated that FTIR spectroscopy has the po-
PCA of SR-FTIR spectra showed similar findings to those ob- tential to discriminate consistently between EBs and ESNCs. The
tained from PCA of the FPA-FTIR data. Using the first two PCs, clus- combination of FTIR spectra with chemometric analyses such as
tering of data acquired from each cell type was achieved with some PCA and UHCA, provided a reliable technique for ESNCs discrimina-
minor overlapping between cluster groupings (Fig. 5). The greater tion. Indeed, UHCA methods could reliably discriminate and clas-
overlap between clusters in scores plots from the SR-FTIR data may sify each stage of ESNCs differentiation. Moreover, the
be due to the fact that the spectra from single cells have a higher examination of PCA scores and loadings plots provided information
level of variation which is reduced by averaging in the spectra ac- about possible biochemical changes occurring during the differen-
quired from clumps of cells. It is apparent from the scores plots tiation processes, which could direct future proteomic and lipido-
from the SR-FTIR data that the NPC spectra (from the N2 culture mic studies.
stage) can be distinguished from spectra from other cell types by Major band differences were observed during neural cell differ-
having positive PC1 and PC2 scores, which corresponded to high entiation in spectral regions assigned to protein and lipid absor-
negative PC1 and PC2 loadings for bands at 2926, 2921 and bance. Specifically, it was hypothesised that lipid absorbance
2849 cmÀ1 (data not shown), which were assigned to the msCH2 changes could be related to conformational changes in acyl chains
and masCH2 characteristic of long chain saturated fatty acids. These during differentiation of neural cells from ES pre-cursor cells, dem-
results demonstrated that PCA could distinguish between spectra onstrated in both clumps of cells and in single cells using FPA and
from EBs and those from cells differentiated towards neural lin- SR-FTIR microspectroscopy, respectively. Changes in the amide I
eages, with the main changes apparently involving changes in FTIR protein band were interpreted as increases in proteins rich in a-he-
spectral bands assigned to lipids. lix structure and a decrease in those with predominantly b-sheet
UHCA was applied to classify and determine the similarity of secondary structure that occurred during differentiation of NPCs
FPA-FTIR spectra of mESCs and differentiated neural cells. Second and ESNCs. This was corroborated by a number of publications
derivative FTIR spectra were vector normalized in the spectral using non-spectroscopic evidence that also indicated an increase
ranges 3000–2800 cmÀ1 and 1750–900 cmÀ1, and UHCA applied in a-helix rich proteins during neural cell differentiation.
using Ward’s algorithm. UHCA provided clear discrimination of In conclusion, FTIR spectroscopy has been shown to be sensitive
spectra into clusters corresponding to differentiation type for the discrimination of different stages of stem cell differentia-


ARTICLE IN PRESS


on multivariate statistical methods. We are grateful to Dr. Mark To-
bin, Principal Beamline Scientist from IR microspectroscopy beam-
line at the Australian Synchrotron for his expert assistance during
the synchrotron beamtime. Synchrotron beamtime was awarded
by the Australian Synchrotron under the merit-based proposal
scheme. We also thank Dr. Nichada Jearanaikoon, an SLRI research-
er, for her assistance with the measurements during the synchro-
tron beamtime in Australia. Special thank to Dr. Peter Lasch,
Cytospec Inc., Boston Ma, who kindly provided the Cytospec IR
imaging software pro bono, as a way to encourage biospectroscopic
research in Thailand.

References

[1] G. Martino, S. Pluchino, Nat. Rev. Neurosci. 7 (2006) 395–406.
[2] I, Prajerova, P. Honsa, A. Chvatal, M, Anderova, Cell. Mol. Neurobiol., 2009, Aug
26, doi:10.1007/s10571-009-9443-x.
[3] R.S. Goldstein, O. Pomp, I. Brokhman, L. Ziegler, Methods Mol. Biol. 584 (2010)
283–300.
[4] S. Temple, Nature 340 (1989) 471–473.
[5] B.A. Reynolds, S. Weiss, Science 255 (1992) 1707–1710.
[6] T.J. Kilpatrick, P.F. Bartlett, Neuron 10 (1993) 255–265.
[7] D.L. Stemple, D.J. Anderson, Cell 71 (1992) 973–985.
[8] A. Trentin, C. Glavieux-Pardanaud, N.M. Le Douarin, E. Dupin, Proc. Natl. Acad.
Sci. USA 101 (2004) 4495–4500.
[9] E. Dupin, G. Calloni, C. Real, A. Gonçalves-Trentin, N.M. Le Douarin, C.R.
Biologies 330 (2007) 521–529.
[10] V. Coronas, Cent. Nerv. Syst. Agents Med. Chem. 2 (2009) 110–118.
[11] C. Lois, A. Alvarez-Buylla, Proc. Natl. Acad. Sci. USA 90 (1993) 2074–2077.
[12] R. McKay, Science 276 (1997) 66–71.
[13] F.H. Gage, Science 287 (2000) 1433–1438.
[14] W.A. Bonner, H.R. Hulett, R.G. Sweet, L.A. Herzenberg, Rev. Sci. Instrum. 43
(1972) 404–409.
[15] M.H. Julius, T. Masuda, L.A. Herzenberg, Proc. Natl. Acad. Sci. USA 69 (1972)
1934–1938.
[16] T. Strojnik, G. Vatne Røsland, P.O. Sakariassen, R. Kavalar, T. Lah, R. Ayer, J.H.
Zhang, W.C. Low, Surg. Neurol. 68 (2) (2007) 133–144.
[17] S. Erceg, S. Laínez, M. Ronaghi, P. Stojkovic, M.A. Pérez-Aragó, V. Moreno-
Manzano, R. Moreno-Palanques, R. Planells-Cases, M. Stojkovic, PLoS One 3 (5)
(2008) e2122.
[18] W. Weichel, S. Irlenbusch, K. Kato, A. Radbruch, in: A. Radbruch (Ed.), Flow
Cytometry and Cell Sorting, Springer-Verlag, Heildelberg, 1999, p. 264.
[19] M. Diem, P.R. Grifﬁths, J.M. Chalmers (Eds.), Vibrational Spectroscopy for
Medical Diagnosis, John Wiley and Sons, Chichester, UK, 2008.
[20] L.M. Miller, G.D. Smith, G.L. Carr, J. Biol. Phys. 29 (2003) 219–230.
[21] D. Ami, T. Neri, A. Natalello, P. Mereghetti, S.M. Doglia, M. Zanoni, M. Zuccotti,
S. Garagna, C.A. Redi, BBA-Mol. Cell Res. 1783 (1) (2008) 98–106.
Fig. 6. Cluster analysis of mESCs and progenitor cells at different stages of neural [22] M.J. German, H.M. Pollock, B. Zhao, M.J. Tobin, A. Hammiche, A. Bentley, L.J.
differentiation. Cluster analysis employed Ward’s algorithm using second deriva- Cooper, F.L. Martin, N.J. Fullwood, Invest. Ophthalmol. Visual Sci. 47 (2006)
tive, vector normalized spectra, over the spectral ranges from 3000–2800 cmÀ1 to 2417–2421.
1750–900 cmÀ1. [23] C. Krafft, R. Salzer, S. Seitz, C. Ern, M. Schieker, Analyst 132 (7) (2007) 647–653.
[24] K. Hoshino, H.Q. Ly, J.V. Frangioni, R.J. Hajjar, Prog. Cardiovasc. Dis. 49 (6)
(2007) 414–420.
[25] B. Rohit, G.W. Bentley, L.K. Jack, Appl. Spectrosc. 54 (4) (2000) 470–479.
tion towards neural lineages, with SR-FTIR microspectroscopy able [26] P. Dumas, L. Miller, J. Biol. Phys. 29 (2003) 201–218.
to achieve discrimination of the changes at the single cell level. [27] A. Kretlow, Q. Wang, J. Kneipp, P. Lasch, M. Beekes, L. Miller, D. Naumann, BBA
Biomembr. 1758 (2006) 948–959.
These results suggest that FTIR spectroscopy shows potential to
[28] A.J. Bentley, T. Nakamura, A. Hammiche, H.M. Pollock, F.L. Martin, S. Kinoshita,
be a complimentary method to conventional methods such as FACS N.J. Fullwood, Mol. Vision. 13 (2007) 237–242.
analysis for discriminating different stages of neural cell differenti- [29] K. Jilkine, K.M. Gough, R. Julian, S.G.W. Kaminskyj, J. Inorg. Biochem. 102
(2008) 540–546.
ation. The spectroscopic method has the advantage over conven-
[30] E. Gazi, J. Dwyer, N.P. Lockyer, J. Miyan, P. Gardner, C.A. Hart, M.D. Brown, N.W.
tional approaches being low cost, rapid, non-destructive and Clarke, Vib. Spectrosc. 38 (1–2) (2005) 193–201.
potentially automatable. It is envisaged that FTIR analysis data [31] S.C. Zhang, M. Wernig, I.D. Duncan, O. Brustle, J.A. Thomson, Nat. Biotechnol.
combined with other complimentary approaches such as proteo- 19 (2001) 1129–1133.
[32] J.W. McDonald, D. Becker, T.F. Holekamp, M. Howard, S. Liu, A.W. Lu, J. Lu, M.M.
mic, lipidomic, or gene micro array analysis, should provide a dee- Platik, Y. Qu, T. Stewart, S. Vadivelu, J. Neurotrauma 21 (2004) 383–393.
per understanding of the biochemical changes occurring during [33] T. Ben-Hur, M. Idelson, H. Khaner, M. Pera, E. Reinhartz, A. Itzik, B.E. Reubinoff,
neural cell differentiation. Stem Cells 22 (2004) 1246–1255.
[34] S.I. Svetlov, T.N. Ignatova, K.K. Wang, R.L. Hayes, V.G. Kukekov, Stem Cells 13
(2004) 685–693.
[35] J. Li, R.J. Wurtman, Brain Res. 803 (1–2) (1998) 44–53.
Acknowledgements [36] P. Bassan, H.J. Byrne, J. Lee, F. Bonnier, C. Clarke, P. Dumas, E. Gazi, M.D. Brown,
N.W. Clarke, P. Gardner, Analyst 134 (2009) 1171–1175.
[37] S. Krimm, J. Bandekar, Adv. Protein Chem. 38 (1986) 181–364.
This study was supported by Synchrotron Light Research Insti- [38] R. Ikeda, M.S. Kurokawa, S. Chiba, H. Yoshikawa, M. Ide, M. Tadokoro, S. Nito, N.
tute (SLRI, Public Organization), and Suranaree University of Tech- Nakatsuji, Y. Kondoh, K. Nagata, T. Hashimoto, N. Suzuki, Neurobiol. Dis. 20 (1)
nology, Thailand. The authors would like to acknowledge Mr Finlay (2005) 38–48.
[39] D.H. Geschwind, J. Ou, M.C. Easterday, J.D. Dougherty, R.J. Jackson, Z. Chen, H.
Shanks from the Monash University Centre for Biospectroscopy for Antoine, A. Terskikh, I.L. Weissman, S.F. Nelson, H.I. Kornblum, Neuron 29
his assistance with the infrared focal plane array microspectro- (2001) 325–339.
scopic imaging, and Dr Keith Bambery for his assistance and advice [40] T. Reya, S.J. Morrison, M.F. Clarke, I.L. Weissman, Nature 414 (2001) 105–111.


Neural Stem Cell Differentiation

More Related Content

What's hot (20)

Similar to Neural Stem Cell Differentiation (20)

Neural Stem Cell Differentiation