This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy

Vibrational Spectroscopy 60 (2012) 3442

Contents lists available at SciVerse ScienceDirect

Vibrational Spectroscopy
journal homepage: www.elsevier.com/locate/vibspec

Label-free non-destructive in situ biochemical analysis of nematode Steinernema

kraussei using FPA-FTIR and Raman spectroscopic imaging
Katherine Lau a,,1 , Alison Hobro b,1 , Tim Smith a , Thomas Thurston a , Bernard Lendl b
a
b

Spectroscopy Product Division, Renishaw plc, Old Town, Wotton-Under-Edge, Gloucestershire GL12 7DW, England, United Kingdom
Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9/164-AC, A-1060 Vienna, Austria

a r t i c l e

i n f o

Article history:
Received 9 September 2011
Accepted 16 January 2012
Available online 25 January 2012
Keywords:
FTIR imaging
FPA
Raman imaging
Nematode
Model organism
Steinernema kraussei

a b s t r a c t
Here we present Fourier transform infrared (FTIR) and Raman images of Steinernema kraussei nematode
worms and, in conjunction with chemometric analysis. We also distinguished the biochemical differences
associated with different regions of these images. The nematodes are complex, multicellular organisms
that have a simple, but dened body plan with distinct digestive, muscular and reproductive systems. As
such, nematodes have often been used as model organisms for the study of biological processes in higher
organisms. We show that FTIR spectra, collected in transmission mode, contain information from the
entire thickness of the nematode, but still exhibit biochemical differences reecting different tissues or
anatomical structures, such as the digestive tract. Raman spectroscopy, in contrast, can be used to investigate the distribution of biological molecules on one plane within a constrained depth of the nematode,
allowing imaging of ne details within the body of the worm, including the distribution of lipid, protein
and collagen. Together these vibrational spectroscopic techniques provide complementary, non-invasive
and label-free information on the spatial distribution of biomolecules within multicellular organisms and,
therefore, have potential future applications in developmental studies of such model organisms.
2012 Elsevier B.V. All rights reserved.

1. Introduction
Genetically amenable multicellular organisms have been instrumental in elucidating gene function and regulation, cellular
organisation and the developmental stages from embryo to adult.
Multicellular organisms are evolutionarily highly conserved. Simple multi-cellular organisms contain direct or similar orthologues
to developmental processes in humans. They also have faster life
cycles and reproduce more rapidly compared to higher organisms, allowing studies of development and metabolism on realistic
timescales. According to Nobel Laureate Sydney Brenner, the
nematode Caenorhabditis elegans is, the simplest differentiated
organism. Since Brenner proposed its use as a model organism
in the 1960s, C. elegans has been widely adopted as a model for
understanding gene functions [1] and geneprotein interaction, as
well as analogues of human diseases [2]. In this study on the nematode species Steinernema kraussei, we explore the use of FTIR and
Raman spectroscopies as non-invasive and label-free techniques to
provide in situ biochemical information within the nematode.

Corresponding author. Tel.: +44 1453 532935; fax: +44 1453 523901.
E-mail address: katherine.lau@renishaw.com (K. Lau).
1
Equal contribution to this work.
0924-2031/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.vibspec.2012.01.009

Genetic manipulation can result in phenotypic changes, which

are currently studied by techniques that are either invasive or
involve labelling [3]. Although invasive techniques such as quantitative polymerase chain reaction [1], Western blotting [1] and
mass spectrometry, can provide very specic information, spatial
information of the bio-molecules in the organism is lost. Labelling
can retain spatial information, but requires a priori knowledge.
Through the use of confocal laser scanning microscopy, 3D noninvasive examination can be achieved [3]. Fluorescent labels, in this
case, can be applied to visualise the presence and locality of target molecules [3]. However, the number of labels applicable to an
organism is limited, since uorescent bands are broad and overlap
in the electromagnetic spectrum.
FTIR and Raman spectroscopies have been applied to a wide
range of cellular and tissue studies [48], as well as microbes [9].
Vibrational spectroscopy gathers information on the biomolecules
present in biological samples based on their vibrational characteristics. FTIR spectroscopy measures the infrared (IR) radiation
of the mid-IR range absorbed by the sample. Raman spectroscopy
measures the Raman scattering of the volume illuminated by the
laser. Through measuring the vibrational modes associated with
the molecules present in dened regions of a sample, and reorganising the data into a hyper spectral data cube, biochemical contents
can be presented in a spatial manner. Vibrational spectroscopic
techniques are label-free, and prior knowledge of the sample is

Author's personal copy

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

not necessary as the full range of chemical information is recorded

simultaneously.
Hitherto, FTIR has been used to differentiate nematode species
based on single point spectra [10,11] and, recently, FTIR imaging
has been used to study the distribution of biomolecules within
nematodes [12]. Previous spontaneous Raman scattering studies,
however, have generally relied on the presence of dye molecules,
such as nile red, incorporated into the nematode, localising within
particular tissues [13], or have concentrated on specic molecules
within the nematode, which are strong Raman scatterers, such as
S8 sulfur [14].
FTIR and Raman imaging are often used together as they provide
complementary and insightful information. This study shows that
FTIR and Raman spectroscopies can be applied to nematode species
S. kraussei to reveal chemical structural contents over the entire
organism length, in a non-invasive and label-free manner.
2. Materials and methods
2.1. Material and sample preparation for spectroscopic imaging
Living S. kraussei was purchased from a garden centre and came
in a water soluble gelatinous substance. The nematodes are typically approximately 20 m in diameter and 80100 m in length.
For FTIR imaging, a small spatula of nematodes was rehydrated
in 1 ml of distilled water in an Eppendorf tube. The suspension
was mixed thoroughly, centrifuged and the supernatant removed.
The resultant pellet was re-suspended in water and the process
repeated twice to ensure any media from the dried nematode
preparation was removed from the rehydrated nematodes prior to
analysis. After the nal centrifugation the pellet was re-suspended
in water and allowed to stand for approximately 20 min before
opening the tube carefully (to prevent disturbing any debris that
may have settled to the bottom). Finally, a small volume of the
supernatant containing the nematodes was spotted onto a midinfrared (IR) transparent ZnSe plate. As the water evaporated the
nematodes adhered to the ZnSe slide and were therefore immobilised prior to the IR measurements.
For Raman imaging, a small spatula of S. kraussei nematodes
was rehydrated using distilled water in a 10 ml beaker and washed
by increasing the dilution. A small volume of single nematodes in
water was pipetted onto a quartz slide and separated by a small
brush. The nematodes were dehydrated and immobilised on the
quartz surface. The sample was immersed in 0.9% NaCl solution for
Raman analysis.
2.2. Methods
2.2.1. FPA-FTIR imaging
IR images of the nematode were collected using a Bruker Hyperion 3000 IR microscope (Bruker, USA), attached to a Tensor 37
FT-IR spectrometer, operating in focal plane array (FPA) imaging
mode. A 15 objective was used. The image spectra were collected
over the spectral range 9003400 cm1 , at 4 cm1 resolution. Each
spectrum was a result of 128 accumulated scans. The spectra were
recorded using an FPA detector with 64 64 pixels covering an area
of 170 m 170 m (approx 2.7 m 2.37 m per pixel). For an
area covering a whole nematode, the image presented here is a
composite of 2 3 FPA images. The spatial resolution in FTIR spectroscopy is wavelength dependent. For the mid-IR range it ranges
from approximately 3 m to 10 m. All image spectra were collected as absorbance spectra in transmission mode. A background
spectrum was collected from the ZnSe plate outside of the areas
with the nematode suspension, prior to recording the nematode
images.

35

2.2.2. FPA-FTIR data processing and analysis

Spectra were corrected for Mie scattering effects using a correction algorithm [15] operating in Matlab R2010b (MathWorks,
USA), prior to spectral analysis. The non-resonant Mie scattering
correction option, based on Ref. [16] was applied using a matrigel
reference spectrum, and 8 principal components (PCs). The size of
the scattering particle was assumed to be 218 m and the refractive index was assumed to be 1.11.5. These settings have been
shown to provide effective scattering correction for FTIR spectra of
nematodes [12]. Once corrected, the image was cropped in order
to remove pixels recorded from outside the worm, using the MIA
toolbox (Eigenvector Research Inc., WA, USA) operating in Matlab.
All IR image reconstruction, display and cluster analyses were
performed in CytoSpec software (http://www.cytospec.com/).
Two data processing steps were performed on the non-resonant
Mie scattering corrected data prior to cluster analysis. Initially, the
nematode spectra were selected as the region of interest and spectra originating from the ZnSe slide or cellular debris were therefore
omitted from further analysis. Subsequently, the IR spectra in
the region of interest were transformed into second derivative
spectra (using an 11 point smoothing function). Hierarchical cluster analysis (HCA) was limited to the spectral ranges between
10001770 cm1 and 28003000 cm1 . HCA was performed using
the D-value distance method and the Wards algorithm cluster
method. This combination has been shown to provide a good correlation between spectroscopic and histological data [7]. In this paper
the appropriate number of clusters was dened according to the
dendrogram, using the minimum number of clusters required to
explain the biological condition, where the standard deviation of
each cluster was low.
2.2.3. Raman microspectroscopic mapping
Raman map data of the entire worm, the middle section, and
the tail section of the same organism, were collected using an inVia
Reex Raman Microscope (Renishaw plc, Wotton-Under-Edge,
Gloucestershire, England). A 532 nm laser source and 2400 l/mm
grating were used, giving better than 1 cm1 spectral resolution. A
50 water immersion objective NA 0.75 (Nikon) was used resulting in sub-micrometre lateral resolution. The spectral range was
approximately 7501700 cm1 . The map collected over the entire
length of the nematode was obtained using StreamLine PlusTM
Raman imaging, which uses a line focus. Using a line focus geometry
enabled multiple spectra to be collected simultaneously and power
density was thus minimised, preventing photothermal degradation. For the full organism map the step size was 1.2 m in both
x and y directions. The mapping completed in just over 60 min.
The maps collected from the middle and tail sections were
obtained using StreamLineHRTM imaging, which is a high spatial
resolution mode using a spot focus with spatial over-sampling. The
step size was 0.5 m in both x and y directions. The tail section
map took approximately 45 min to complete. The middle section
map took under 6 h to complete.
2.2.4. Raman data analysis
The Raman data were analysed using principal component analysis (PCA). PCA looks for the most signicant trends in the data set.
The data volume is rotated in such a way that the rst PC explains
the largest possible proportion of the variance in the data matrix.
The second PC explains the largest possible proportion of the residual variance in the data matrix that is not explained by the rst PC.
Subsequent PCs continue to explain the largest possible proportion
of the residual variance not accounted for by earlier components.
The full length nematode Raman data set was imported to Matlab
R2010b (MathWorks, USA) and analysed using an in-house script.
The mid and tail section maps were analysed using WiRE 3 software (Renishaw plc, Wotton-Under-Edge, Gloucestershire, UK). In

Author's personal copy

36

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

either method, standard normal variate pre-processing [17] was

applied to the data sets before they were analysed with PCA. For
the full nematode map, 8 PCs explained 99% of the data in the data
set. PCs 1, 2, 5 and 7 represent the background and quartz. PCs 3,
4 and 8, which represent nematode biochemical contents, are presented here. For visual presentation of the PCs and their distribution
in the maps, colours are assigned to different PCs and the most
intensely coloured (or the brightest regions) represent high scores
for the respective PC. While different components are resolved
in different images, the colour scheme was kept as consistent as
possible.

3. Results
3.1. Complete nematode FTIR imaging distinguishes anatomical
regions based on IR spectral variation
The FTIR spectra recorded in this study arose from the
absorbance of IR light through the entire thickness of the nematode. Therefore, an overview of the overall biochemical contents of
the nematode was obtained. The blunt ended head and the sharp
ended tail regions of a nematode can be easily distinguished in the
white light image (Fig. 1A). By applying HCA to the IR spectra collected from the S. kraussei nematode and selecting an appropriate
number of clusters (in this case, 5) different regions were identied (Fig. 1B), which correspond to anatomical features such as the
digestive tract and body cavity. Comparisons of the average spectra
of each of the clusters (shown in the same colours as the clusters
they to which they correspond) show that these regions vary significantly from one another biochemically. In general, all the clusters
exhibit a similar spectral prole containing a mix of lipid, protein,
carbohydrate and nucleic acid bands, but the proportions of each
of these components differ between the different clusters.
The blue cluster is found in the head and the tail regions, as well
as some parts of the mid-length of the body. According to the IR
spectrum, the blue cluster is low in total lipids, as evidenced by
the absence of a C O band at 1740 cm1 and a lack of prominent
CH2 and CH3 symmetric and asymmetric stretching vibrations in
the high wavenumber region. Its high protein content is conrmed
by the prominent amide I and amide II bands at 1652 cm1 and
1540 cm1 , respectively (Fig. 1C). The grey cluster runs along the
length of the nematode. Similar to the blue cluster, the grey cluster
is also dominated by high protein presence and a low lipid level, as
shown by the high amide I and II peaks at 1652 cm1 and 1540 cm1
and little contribution from C O stretching at 1740 cm1 . The main
difference between the spectra originating from the blue and grey
clusters is found in the relative intensity of the two bands at
2962 cm1 and 2935 cm1 , suggesting that, although the lipid content is low in both cases, the grey cluster exhibits slightly lower
asym (CH3 ) [18,19]. The higher asym (CH3 ) in the blue spectrum
suggests higher acyl branching [8]. Both the grey and blue clusters are protein rich and are associated with the head and body
regions, suggesting these clusters represent the body cavity, which
are relatively rich in proteins associated with muscle or hypodermal proteins [20]. The small difference in lipid content between
these two clusters is likely to reect differences in the types of tissue
present at these positions.
The remaining three clusters, depicted in cyan, red, and green,
are all associated with the centre of the nematode, close to the
digestive tract. The cluster in cyan contains signicant contributions from protein, as shown by the amide I and amide II bands
at 1652 cm1 and 1547 cm1 , and a moderate lipid component,
highlighted by the C O stretching band at 1740 cm1 . Additional
contributions from phosphate and carbohydrate associated bands
are observed between 1230 cm1 and 1000 cm1 [18,19].

Those in red and green are both lipid rich with intense peaks
at 2927 cm1 and 2855 cm1 originating from asym (CH2 ) and
sym (CH3 ) of methylene groups in fatty acids, respectively [18,19].
The main differences between these two clusters are associated
with the protein content, which is relatively high in the green
cluster, as shown by higher intensities for the amide I, amide II,
amide III and COO stretching in amino acid side chains [18,19,21]
positioned at 1652 cm1 , 1540 cm1 , 1310 cm1 and 1401 cm1 ,
respectively. The green cluster also exhibits a more intense spectral
envelope centred around 1075 cm1 originating from PO2 stretching in phosphodiesters. The red cluster, however, shows a relative
increase in intensity for the IR band at 1162 cm1 , which is associated with C O in proteins and carbohydrates [8]. In this particular
case, the relative increase in intensity is likely to be associated with
carbohydrates, given the relatively low contributions from other
protein based bands in the red spectrum. Given the localisation of
the red and green clusters, both are anticipated to be associated
with the digestive tract. When considering their spectral differences, the green cluster, as it reects higher protein content, may
depict the digestive tract itself. The red cluster, on the other hand
displays relatively high lipid and carbohydrate, but low protein
contributions, and may reect the contents of the digestive tract.
3.2. Complete nematode Raman imaging identies lipid rich,
protein rich and collagen rich domains
Raman imaging over the entire length of a different nematode
(Fig. 2A) at approximately 1 m resolution, successfully captured
biochemical information on a dened plane across the centre of the
nematode, in a spatially resolved manner. The application of PCA to
the Raman data resulted in three main PCs that contain biochemical information of the nematodePC3 (magenta), PC4 (cyan) and
PC8 (yellow). These colours are combined in the image such that
overlapping cyan and magenta results in green colour (Fig. 2B).
Comparing the PC loadings (not shown) to the band assignments in Table 1, PC3 (magenta) is characterised positively by peaks
at 1265 cm1 , 1301 cm1 , 1440 cm1 , 1462 cm1 and 1659 cm1 ,
assigned to lipid, and the amide I and amide III protein bands
[2224]. It is strongly negatively characterised by a peak at
1209 cm1 associated with the amino acids phenylalanine and
tyrosine [24], as well as peaks at 1347 cm1 and 1393 cm1 . PC3
therefore represents regions consisting mainly of lipid with some
protein. This cluster is mainly distributed in the centre of the nematode and, as such, is similar to the red and green clusters identied
in the FTIR analysis which are also lipid rich with some contributions from protein components.
PC4 (cyan) is predominantly found at the ventral edge and
the tail region of the nematode. According to its PC loadings,
it is positively characterised by collagen and collagen components (proline and hydroxyproline) peaks at 950 cm1 , 1028 cm1
and 1072 cm1 [25], as well as carotenoid peaks at 1192 cm1 ,
1512 cm1 , 1515 cm1 , 1523 cm1 and 1536 cm1 [22,26]. It is
negatively characterised by a range of protein, lipid and carbohydrate peaks at 1003 (phenylalanine), 1129 cm1 (lipid and protein),
1255 cm1 and 1265 cm1 (amide III), 1347 cm1 (C H deformations), 1394 cm1 , 1449 cm1 (C H deformations) and 1659 cm1
(amide I) [24,27], which are the main protein and lipid constituents
of a multicellular organism. The PC loadings suggest that PC4 represents domains biochemically different from the main nematode
core body, and contain mainly collagen and carotenoid. As collagen
is the main component of the outer cuticle of nematodes, and as
PC4 occupies the ventral edge position, the presence of collagen is
to be expected.
PC8 (yellow) is distributed throughout the worm, especially
in the head region and the edge of the body, predominantly the
dorsal side, showing a similar distribution to the grey cluster

Author's personal copy

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

37

Fig. 1. (A) Light microscopy image of the nematode imaged by FPA-FTIR. (B) False colour image resulting from the hierarchical clustering analysis (HCA) performed on
the preprocessed IR spectra. Five clusters are resolved and are coloured differently. (C) The average spectra of the clusters resolved in the HCA. The colours of the spectra
correspond to the colours of the clusters in (B). (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of the article.)

in the FTIR analysis. It overlaps slightly with PC4 (cyan) but

occupies a more proximal layer of the body (Fig. 2B). According to its PC loadings, PC8 is positively characterised by peaks
at 1003 cm1 (phenylalanine), 1032 cm1 (phenylalanine and proline), 1207 cm1 (phenylalanine and tyrosine), 1242 cm1 (amide
III: collagen), and 1674 cm1 (amide I), attributed to protein and
amino acids [23,27]. The peaks at 1332 cm1 (C H deformations in
nucleic acids), 1462 cm1 and 1589 cm1 are tentatively assigned
to deoxyribose and DNA [24]. PC8 therefore represents domains
mainly consisting of protein, of which some is collagen, with some
contributions from nucleic acids, which is in agreement with the
FTIR data.
The dorsal and ventral edges, and the head and the tail of the
nematode, are represented by different PCs. This suggests that the
dorsal and the ventral cuticles are biochemically different. The data
also show that the tail consists mainly of collagen, while the head
contains different protein types and nucleic acids.
3.3. Sub-micrometre resolution Raman imaging unveiled further
chemico-structural details
For the same nematode imaged above, high confocal StreamLineHR Raman imaging was applied to the tail and mid-sections
(Fig. 2C and D). The increase in spatial resolution increased the level
of detail. The spatial distribution and loadings plots for each of the
PCs found in the tail region are shown in Fig. 3 and band assignments are detailed in Table 2. For the mid-section Raman image,

which contains a high level of detail but similar contents to the tail
map, the distribution of the PCs and their loadings plots are shown
in Fig. 4, while for details on band assignments for each of the PC
loadings, readers are referred to Table 3.
3.3.1. Tail section Raman image
In the tail section, PCA resolved 5 biologically relevant domains
(Fig. 3, PC2PC6). PC1 represents the background (data not shown).
PC2 (yellow) is characterised by peaks attributed to protein and
lipid, and is found throughout the tail region. The most intense
regions coincide with those shown in PC3 (magenta) which is positively characterised by lipid bands but negatively characterised
by protein bands, and therefore denotes pure lipid domains. The
overlap between PC2 and PC3 originates from the presence of lipid
associated bands in both loading plots. These domains correlate
to the known lipid droplets in the nematode intestine, which are
important for lipid storage and metabolism [28]. PC5 loadings (Fig. 3
and Table 2) display characteristics for collagen and collagen components hydroxyproline and proline. PC5 (green) is thus likely to
represent the outer cuticle of the nematode, especially as this is
predominantly found at the edge of the nematode.
PC4 (red) is characterised by peaks ascribed to phenylalanine,
glycogen, glucose, carotenoids and globins and therefore is likely
to denote muscular domains. The presence of -carotene in muscle bre has been reported previously [26]. Globins are iron- or
oxygen-binding proteins, which are required in muscles. More than
30 globin-like genes are known to be present in the genome of

Author's personal copy

38

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

Fig. 2. (A) A white light image of the S. kraussei nematode Raman imaged. (B) The Raman image of the full nematode is as presented, which is a resultant principal component
analysis (PCA) image of the full nematode. Scale bar: 100 m. Three principal components (PCs) that are correlated to the biochemical contents of the nematode are overlaid:
PC3 (magenta), PC4 (cyan) and PC8 (yellow). (C) The resultant PCA image of the tail region map. Scale bar: 5 m. The area of the nematode mapped is as indicated by the
red box on the left in (A). Five biochemically relevant PCs are overlaid on one another: PC2 (yellow), PC3 (magenta), PC4 (red), PC5 (green) and PC6 (cyan). D) The result PCA
image of the middle region map. Scale bar: 20 m. The area of the nematode mapped is as indicated by the red box in the middle in (A). Seven biochemically relevant PCs
are overlaid on one another: PC2 (yellow), PC3 (magenta), PC4 (green), PC5 (white), PC6 (red), PC7 (cyan) and PC8 (blue). (For interpretation of the references to colour in
this gure legend, the reader is referred to the web version of the article.)

Table 1
Peak positions and tentative assignments of major vibrational bands observed in the PC loadings from the nematode Raman maps [2224,26,27,30].
Peak position (cm1 )
1674
1659
1589
1542
1462
14391440
1393
1370
1355
1347
1332
1301
12581265
1255
12401242
1209
1199
1158
11241129
10311032
10281029
1003
972
956
938
934
922
895
875
864
853
826
784

Protein assignments

Lipid assignments

(C O) Amide I
(C O) Amide I

(C C) lipid, unsaturated fatty acids

Tyr
(CH2 ), protein
CH2 deformation, protein
CH rocking, protein
Globin
Trp, myoglobin

(CH2 ) twisting/wagging, collagen

(CN), (NH) amide III (-helix) collagen, Trp

(CH2 ), lipid
CH2 deformation, lipid
CH rocking, lipid

(C H) deformation, phospholipids
CH2 twisting/wagging, ( C H) phospholipids,
lipid
( C H), lipid

Others

Nucleic acids
Carotenoid
Deoxy-ribose, (CH2 )

Guanine
Unassigned
Guanine
Adenine, cytosine

Adenine, cytosine, (CC) carotenoid

Amide III (collagen/disordered)
Tyr, Phe

(C N) protein
C C skeletal stretching band (protein), ring
breathing mode (Phe)
Proline/Phe
Ring breathing mode, Phe
(C C) backbone, protein

(C C), lipid

Carotenoid
( CC ) -car, carotenoids
(C O), glucose

( C H), lipid

(C C) backbone, collagen backbone,

hydroxyproline
(C C), protein (-helix), myosin
(C C), proline ring
(CH2 ) rocking, hydroxyproline
(C C), hydroxyproline
Collagen, Trp, Tyr
Tyr

, stretching; , bending; Trp, tryptophan; Phe, phenylalanine; Tyr, tyrosine; -car, -carotene.

Carotenoid
(C O C), glycogen

Glucose
DNA backbone

1,4-Glycosidic link, glycogen

(O P O), DNA
Cytosine, uracil

Author's personal copy

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

39

Fig. 3. Detailed Raman images of the tail section of the nematode (also shown in Fig. 2C). The distribution of the scores per PC is laid out separately in their assigned colours:
PCs 2 (yellow), 3 (magenta), 4 (red), 5 (green) and 6 (cyan). The PC loadings are displayed on the right, in colours corresponding to their PCs. (For interpretation of the
references to colour in this gure legend, the reader is referred to the web version of the article.)
Table 2
The main peaks found in the principal component (PC) loadings for each PC shown in the tail map are listed below.
PC

Colour

PC loadings description

Denotes

2
3

Yellow
Magenta

Main worm body constituents.

Pure lipid domains

Red

Green

Cyan

Positive: protein and lipid contribution at 1003, 1032, 1260, 1302, 1443 and 1658 cm1 .
Positive: lipid bands at 1303, 1437 and 1657 cm1 , Negative: protein bands at 1003, 1031 and
1240 cm1 .
Positive: amino acid band 1003 cm1 , glycogen band at 1027 cm1 , glucose band at 1127 cm1 ,
carotenoid band at 1155 cm1 , globin band at 1350 cm1 .
Positive: proline/hydroxyproline (collagen components) bands at 864, 895 and 1028 cm1 , and
collagen amide III band at 1241 and 1249 cm1 .
Positive: protein amide I and unsaturated fatty acid band at 1657 cm1 , nucleic acid band at
1585 cm1 , lipid band at 1441 and 1434 cm1 , myoglobin band at 1355 cm1 , and amide III band at
1258 cm1 .

Muscular domains
Outer cuticle
Globins

Author's personal copy

40

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

Fig. 4. Detailed Raman images of the middle section of the nematode (also shown in Fig. 2D). The distribution of the scores per PC is laid out separately in their assigned
colours: PCs 2 (yellow), 3 (magenta), 4 (green), 5 (white), 6 (red), PC7 (cyan) and PC8 (blue). The PC loadings are displayed on the right, in colours corresponding to their PCs,
except for PC5 loadings (black). (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of the article.)
Table 3
The main peaks found in the principal component (PC) loadings for each PC shown in the mid-section map are listed below.
PC

Colour

Yellow

Magenta

Green

White

Red

Cyan

Blue

PC loadings description

Denotes
1

Positive: lipid and protein bands at 972, 1264, 1301, 1400, 1439 and 1658 cm
Negative: Phe bands at 1001 and 1031 cm1 ; proline band at 1031 cm1 ; amide I
bands at 1635 and 1675 cm1
Positive: lipid band at 1436 cm1 , unsaturated fatty acids at 1657 cm1
Negative: nucleic acids and nucleotides bands at 784, 826, 898, 1100, 1238, 1318,
1339, 1464, and 1558 cm1 ; Phe bands at 1003, 1031, 166 and 1617 cm1 ; Tyr and Trp
bands at 855, 880, 1206, 1339, 1606 and 1617 cm1 ; protein at 1127, 1517, 1318 and
1450 cm1 ; amide I/III at 1288/1638 and 1673 cm1
Positive: proline/hydroxyproline bands at 863, 876, 892, 922, 940 and 1029 cm1 ;
amide III (collagen) at 1244 cm1 ; amide I (-sheet) at 1680 cm1
Negative: Phe at 1003 cm1
Positive: amide I and unsaturated fatty acid at 1659 cm1 ; nucleic acid band at 784,
827, 1486 and 1578 cm1 ; carotenoid at 1155 cm1 ; glucose at 1124 cm1
Negative: amide III at 1252 and 1271 cm1 ; proline/hydroxyproline at 870, 875, 938
and 1028 cm1
Positive: glycogen at 853, 934, 1080, 1128 and 1333 cm1 ; muscle bre/myosin at
934 cm1 ; carotenoid/-carotene at 1158, 1199 and 1542 cm1 ; glycoprotein at
1311 cm1 ; globins at 1353, 1370, 1505 and 1585 cm1 ; amide I (-helix)/unsaturated
fatty acid at 1649 cm1
Positive: proline/hydroxyproline at 875, 935, 940, 1029 and 1205 cm1 ; collagen at
1071 cm1 ; amide I (collagen) at 1663 cm1 ; carotenoid at 956 and 1154 cm1
Positive: nucleic acid at 784, 810, 1178, 1255, 1337, 1396, 1482, 1575 and 1658 cm1 ;
amide III at 1236 and 1270 cm1 ; amide I at 1658 and 1683 cm1

Lipid and protein rich domains

Pure lipid domains

Outer cuticle

Unassigned

Muscular domains

Collagen and carotenoids

Lysed cells of an unassigned origin

Author's personal copy

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

nematode C. elegans, two of which have been transcribed and their

oxygen binding properties characterised [29]. Lastly, PC6 (cyan) is
likely to represent globins as, according to the loadings, PC6 contains a strong protein component, and the peak at 1355 cm1 is
tentatively assigned to myoglobin [30].
3.3.2. Mid-section Raman image
Seven biologically relevant PCs were resolved from the
mid-section Raman data (Fig. 4, PC2PC8). PC1 represents the
background (data not shown). PC2 (yellow) represents the lipid
and protein rich domains and PC3 (magenta) denotes pure lipid
domains and it is positively characterised by lipid Raman peaks,
but negatively characterised by protein Raman peaks in a similar fashion to that observed for the analysis of the tail region. PC4
(green) is rich in collagen and its components, and it is found brightest on the lower rim (Fig. 4), it therefore denotes the outer cuticle
layer. It is also found internally (but not as abundant) since collagen is one of the major components of nematodes. PC5 (white) has
abundant protein including a sharp band originating from phenylalanine, fatty acids, nucleic acids, carotenoid and glucose. Its origin
is currently unassigned but appears to be isolated in the centre
of the nematode. PC6 (red) is characterised by glycogen, myosin,
carotenoid/-carotene, glycoprotein, globins and other proteins.
Similar to PC4 in the tail region, mid-section PC6 is likely to denote
muscular regions. PC7 denotes collagen and carotenoids but does
not show the same distribution as the collagen-only component of
PC4. PC8 is rich in nucleic acids and protein. It is likely to represent
lysed cells, though their origin is as yet unknown.
4. Discussion
Both FTIR and Raman spectroscopies have been applied as labelfree minimally invasive imaging techniques in this study. The use of
these techniques resulted in images that correlate closely with the
white light images of the samples. Furthermore, this study has successfully demonstrated that vibrational spectroscopy can be used to
extract biochemical information from nematodes, which are popular model organisms for gene studies and human diseases [1,2],
and that the information obtained by FTIR and Raman imaging is
complementary.
The FTIR images were collected using a FPA detector, which
enabled much faster data collection compared to conventional FTIR
mapping. This dramatic improvement in speed enabled FTIR spectra of high signal-to-noise ratio to be collected over the entire
organism in a relatively short time, of the order of an hour. FPAFTIR imaging is extremely useful for providing a rapid overview of
the presence and distribution of biologically interesting substances
throughout the nematode. Biochemical information such as the
nature and levels of lipid, protein, nucleic acid and carbohydrates,
can be resolved through spectral deconvolution. HCA was applied
to the FTIR data in our study as it is the most suitable multivariate
analysis method for this type of data according to our experience.
While FTIR imaging provided biomolecular information from
the full thickness of a dehydrated nematode, it is possible to collect
biochemical information on a selected plane using confocal Raman
imaging. Using a high confocal Raman imaging mode, and a 532 nm
laser, diffraction limited spatial resolution was attained. The lateral resolution was below 1 m, whereas the axial resolution was
approximately 2 m. Using a water immersion objective and rehydrating the sample in saline solution, regained the original volume
and shape. The ability to image in a highly spatially resolved manner
enables images to be created similarly to cross-sectioning, except
no physical sectioning is required.
The Raman analysis on S. kraussei in this report agreed with
the FPA-FTIR imaging data. Both techniques showed that high lipid

41

content was found in the centre of the nematode, surrounded by

higher protein contents towards the outer layers. According to the
full length nematode Raman image (Fig. 1B), the head region and
the dorsal side were found to contain a high level of nucleic acids.
The Raman data showed that the ventral side and the tail contained mainly collagen and carotenoid, but such information was
not apparent in the FTIR data.
The discrepancy may be explained by the different spatial resolutions of the techniques. Since water is an extremely strong IR
absorber, the nematode was dried for the FTIR analysis whereas
the nematode was rehydrated for the Raman analysis. For FTIR,
the IR absorption through the entire thickness was investigated
as the measurements were taken in transmission mode; hence
each pixel in the map contains information from several anatomical layers. For Raman analysis, only the inelastic scattering from
one plane was collected. The lateral resolution of FTIR and Raman
spectroscopy is diffraction limited (310 m for FTIR versus submicrometre resolution for Raman). It is therefore conceivable
that a thin collagen-containing cuticle layer may not be resolved
by FTIR imaging, but resolved by Raman imaging. Furthermore,
the typical collagen triplet in the amide III region [5] in the
IR spectrum overlaps with phosphate signal between 1200 cm1
and 1300 cm1 . The presence of this phosphate signature in the
blue and grey clusters closest to the edge of the nematode
would therefore hinder the unambiguous detection of collagen by
FTIR.
A higher number of biologically relevant PCs were resolved from
the partial section Raman maps compared with the full organism
Raman map. This can be explained by the area of the full organism map itself containing more background area than the partial
section maps: a higher number of PCs in the PCA represented the
background. Also, the spatial resolution is higher for the partial section maps (0.5 m step size) than for the full organism map (1 m
step size), therefore an increased amount of detail was obtained in
the partial section maps.

5. Conclusion
To our knowledge, this is the rst report combining FTIR and
spontaneous Raman spectroscopic imaging to resolve biochemical
details from a nematode. The unique point of our work is that, a
full range of biochemical information can be obtained from each
point of the examined area from each single analysis, and provide
spatial information at the same time. Through applying multivariate analysis such as HCA and PCA, spectral variations in the data set
are extracted. Although coherent anti-Stokes spectroscopy (CARS)
imaging has been applied to showing locations of lipid droplets in C.
elegans [28], and it is rapid enough for live nematode analysis, CARS
is mostly restricted to imaging lipids based on the C H stretching
bands in the 28003000 cm1 region. Our Raman and FTIR images
not only show the distribution of lipid in the imaged area, but also
allow simultaneous visualisation of the distribution of other important biochemical components such as proteins, carbohydrates and
nucleic acids.
This work demonstrates the capability of vibrational spectroscopic imaging to obtain chemico-structural details in a
multicellular model organism. However, as a nematode is a three
dimensional object, to analyse its biochemical details in relation to
its internal organisation, 3D confocal Raman imaging is the logical
next step. Future work including 3D Raman mapping of a nematode (collect Raman maps at different depths) would be insightful
and valuable. Ideally, 3D data should be collected as a single data
set and multivariate analysis applied to the spectra from the volume as an entity, to enable volumetric data deconvolution and
visualisation.

Author's personal copy

42

K. Lau et al. / Vibrational Spectroscopy 60 (2012) 3442

Acknowledgments
BL and AH acknowledge nancial support provided by the Austrian research funding association (FFG) under the scope of the
COMET programme within the research network Process Analytical Chemistry (PAC) (contract # 825340).
References
[1] S. Greiss, B. Schumacher, K. Grandien, J. Rothblatt, A. Gartner, BMC Genomics 9
(2008) 334.
[2] B. Alberts, Science 330 (2011) 1724.
[3] A. Valigurova, I. Hodova, R. Sonnek, B. Koubkov, M. Gelnar, Parasitol. Res. 108
(2011) 383394.
[4] R.J. Swain, S.J. Kemp, P. Goldstraw, T.D. Tetley, M.M. Stevens, Biophys. J. 95
(2011) 59785987.
[5] K. Lau, M.A.B. Hedegaard, J.E. Kloepper, R. Paus, B.R. Wood, V. Deckert, J. Dermatol. Sci. 65 (2011) 191198.
[6] C. Krafft, R. Salzer, S. Seitz, C. Ern, M. Schieker, Analyst 132 (2007) 647653.
[7] P. Lasch, W. Haensch, D. Naumann, M. Diem, Biochim. Biophys. Acta: Mol. Basis
Dis. 1688 (2004) 176186.
[8] B.R. Wood, T. Chernenko, C. Matthaus, M. Diem, C. Chong, U. Bernhard, C. Jene,
A.A. Brandli, D. McNaughton, M.J. Tobin, A. Trounson, O. Lacham-Kaplan, Anal.
Chem. 80 (2008) 90659072.
[9] N. Nicolaou, Y. Xu, R. Goodacre, Anal. Chem. 83 (2011) 56815687.
[10] D. Ami, A. Natalello, A. Zullini, S.M. Doglia, FEBS Lett. 576 (2004) 297300.
[11] E. San-Blas, M. Guerra, E. Portillo, I. Esteves, N. Cubilln, Y. Alvarado, Vib. Spectrosc. 57 (2011) 220228.
[12] A.J. Hobro, B. Lendl, Vib. Spectrosc. 57 (2011) 213219.
[13] S. Charan, F.-C. Chien, N. Singh, C.-W. Kuo, P. Chen, Chem. Eur. J. 17 (2011)
51655170.

[14] D. Himmel, L.C. Maurin, O. Gros, J.-L. Mansot, Biol. Cell 101 (2009) 4354.
[15] P. Bassan, A. Kohler, H. Martens, J. Lee, H.J. Byrne, P. Dumas, E. Gazi, M. Brown,
N. Clarke, P. Gardner, Analyst 135 (2010) 268277.
[16] A. Kohler, J. Sule-Suso, G.D. Sockalingum, M. Tobin, F. Bahrami, Y. Yang, J.
Pijanka, P. Dumas, M. Cotte, H. Martens, Appl. Spectrosc. 62 (2008) 259262.
[17] R.J. Barnes, M.S. Dhanoa, S.J. Lister, Appl. Spectrosc. 43 (1989) 772777.
[18] A.A. Christy, Y. Ozaki, V.G. Gregoriou, Applications of infrared spectroscopy
in basic and industrial research, in: Modern Fourier Transform Infrared Spectroscopy, Elsevier, 2001.
[19] D. Naumann, Infrared spectroscopy in microbiology, in: R.A. Meyers (Ed.), Encyclopedia of Analytical Chemistry, Wiley and Sons, 2001.
[20] M.L. Blaxter, J. Biol. Chem. 268 (1993) 66006609.
[21] M. Jackson, L.-P. Choo, P.H. Watson, W.C. Halliday, H.H. Mantsch, Biochim. Biophys. Acta: Mol. Basis Dis. 1270 (1995) 16.
[22] C.J. Frank, D.C.B. Redd, T.S. Gansler, R.L. McCreery, Anal. Chem. 66 (1994)
319326.
[23] S. Keller, B. Schrader, A. Hoffmann, W. Schrader, K. Metz, A. Rehlaender, J.
Pahnke, M. Ruwe, W. Budach, J. Raman Spectrosc. 25 (1994) 663671.
[24] K.A. Hartman, N.W. Clayton, G.J.J. Thomas, Biochem. Biophys. J. Commun. 50
(1973) 942.
[25] A. Madhadevan-Jansen, M. Mitchell, N. Ramanujam, A. Malpica, S. Thomsen, U.
Utzinger, R. Richards-Kortum, Photochem. Photobiol. 68 (1998) 123132.
[26] M. Pezolet, M. Pigeon-Gosselin, J. Nadeau, J.-P. Caille, Biophys. J. 31 (1980) 18.
[27] I. Siebinga, G.F.J.M. Vrensen, K. Otto, G.J. Puppels, F.F.M. De Mul, J. Greve, Exp.
Eye Res. 54 (1992) 759767.
[28] T. Hellerer, C. Axang, C. Brackmann, P. Hillertz, M. Pilon, A. Enejder, Proc. Natl.
Acad. Sci. 104 (2007) 1465814663.
[29] E. Geuens, D. Hoogewijs, M. Nardini, E. Vinck, A. Pesce, L. Kiger, A. Fago, L.
Tilleman, S. De Henau, M.C. Marden, R.E. Weber, S. Van Doorslaer, J. Vaneteren,
L. Moens, M. Bolognesi, S. Dewilde, BMC Biochem. 11 (2010) 17.
[30] D.C.B. Redd, Z.C. Feng, K.T. Yue, T.S. Gansler, Appl. Spectrosc. 47 (1993) 787791.