focal point review
IWAN W. SCHIE,
CENTER FOR BIOPHOTONICS, SCIENCE, AND TECHNOLOGY,
UNIVERSITY OF CALIFORNIA–DAVIS, SACRAMENTO, CA 95817, USA
THOMAS HUSER
DEPARTMENT OF PHYSICS AND CENTER FOR BIOPHOTONICS,
UNIVERSITY OF CALIFORNIA–DAVIS, SACRAMENTO, CA 95817, USA
Methods and Applications of
Raman Microspectroscopy to
Single-Cell Analysis
Raman spectroscopy is a powerful biochemical
analysis technique that allows for the dynamic
characterization and imaging of living biological cells in the absence of fluorescent stains. In
this review, we summarize some of the most
recent developments in the noninvasive biochemical characterization of single cells by
spontaneous Raman scattering. Different instrumentation strategies utilizing confocal detection optics, multispot, and line illumination
have been developed to improve the speed and
sensitivity of the analysis of single cells by
Raman spectroscopy. To analyze and visualize
the large data sets obtained during such
experiments, sophisticated multivariate statistical analysis tools are necessary to reduce the
data and extract components of interest. We
highlight the most recent applications of single
cell analysis by Raman spectroscopy and their
biomedical implications that have enabled the
noninvasive characterization of specific metabolic states of eukaryotic cells, the identification
and characterization of stem cells, and the
rapid identification of bacterial cells. We
conclude the article with a brief look into the
future of this rapidly evolving research area.
Received 20 December 2012; accepted 6
March 2013
* Authors to whom correspondence should be
sent. E-mail: [email protected]; trhuser@
ucdavis.edu.
DOI: 10.1366/12-06971
Index Headings: Intralipid; Temperature stability; Raman scattering; Optical spectroscopy;
Optical microscopy; Inverse adding–doubling;
Optical properties.
INTRODUCTION
T
he ability to analyze single cells
for their biochemical composition in terms of small molecule
content, their proteomic and genomic
content, or possibly even their epigenome has become increasingly more
important as we continue to improve
our understanding of cellular biology.
Furthermore, because most diseases
initially start at the level of single cells
(the initial cell that stopped responding
to apoptosis signals, the initial point of a
bacterial or viral infection, the initial
macrophage that has begun to accumulate excess lipids and begins to turn into
a foam cells, etc.), understanding events
that lead to changes in the way cells
respond to external events, without
harming or modifying those cells, has
become ever more important. Few tools
provide a truly noninvasive and nondestructive analytical capability at the
scale of single cells. Optical microscopy,
the oldest form of imaging at the
microscopic scale, has many of these
attributes: It is typically noninvasive
when combined with fluorescent labeling and highly sensitive detectors, it
provides single-molecule sensitivity,
and it inherently provides a spatial
resolution that enables the study of
living cells, which, with novel approaches invented during the last decade, can
even be improved down to the molecular
scale. Such manifold noninvasive analytical capabilities are simply not found
in any other microscopic imaging technique. Electron microscopy has a simil a r l y w i d e r an g e of a na l yt i c a l
capabilities and achieves even higher
spatial resolution, but a lack of a
similarly wide range of contrast agents
and typical sample preparation steps
have limited it to providing snapshots
of cells, rather than supporting extended
live-cell imaging.1 Secondary ion-mass
spectrometry, another analytical imaging
technique, destroys cells during the
imaging process.2 Another emerging
technique is based on soft X-ray imaging, which provides higher spatial resolution than standard optical microscopy,
but again, providing contrast for multiple species of interest simultaneously is
tricky, and cells typically have to be
APPLIED SPECTROSCOPY OA
813
focal point review
fixed in order to be imaged.3 This pretty
much leaves optical microscopy unmatched where the analysis of single
cells is concerned.
Extensive research during the last
decade has lead to other significant
advances in optical microscopy. The
ability to detect single fluorescent molecules with very high spatial resolution,
as well as the discovery of fluorescent
proteins and the possibility to incorporate them into cells through genetic
fusion with other proteins of interest,
has lead to novel single-cell genomics
and proteomics studies. Fluorescence
microscopy and especially super-resolution optical microscopy4–6 now enable
the visualization of subcellular structures and macromolecular complexes at
an unprecedented level, and is now
expanding to dynamic studies of subcellular structures, e.g., vesicle trafficking, gene transfer, cytoskeletal
rearrangements, viral transfer, etc.,
which will broaden our knowledge of
cell biology drastically. All of these
developments, however, are currently
based on our ability to fluorescently
label cellular structures. This often
necessitates workarounds and tricks that
avoid impacting the biological activity
of the cell. Fluorescent proteins are large
constructs that have an effect on the
overall diffusion of the protein of
interest that they are supposed to label.7
When attached to the wrong part of
another protein, they will also significantly interfere with biological activity,
i.e., by changing association and/or
dissociation constants or sterically hindering binding activity. Similarly, many
organic fluorophores are toxic to cells at
high concentrations. Similar issues can
be raised for many other biological
processes, especially when such small
molecules as sugars, lipids, or ions, are
involved. The specific identification of
these molecules by fluorescent probes
significantly interferes with the biological process of interest. In addition,
fluorescence photobleaching limits the
observation time window of such samples. All these issues have led researchers around the world to try to explore
other contrast mechanisms to overcome
these problems while at the same time
allowing us to assess the biochemical
composition and state of cells. Here, we
814
Volume 67, Number 8, 2013
review the most recent applications of
Raman microspectroscopy to the analysis of single cells.
Raman microspectroscopy is based on
the dispersion and detection of photons
that are inelastically scattered by molecular vibrations. When a monochromatic
laser is used to probe a sample, most of
the interacting photons undergo elastic
scattering, while just a small fraction of
photons scatters inelastically off molecular bonds and thereby exchanges energy with the vibrational energy levels of
the molecules in the sample. This
process can result in a loss in energy
of the scattered photon, termed Stokesshifted Raman scattering, as well as a
gain in energy of the scattered photon,
termed anti-Stokes–shifted Raman scattering. For applications at the single-cell
level, the overall weak nature of the
inelastic scattering process typically
makes the acquisition of anti-Stokes–
scattered photons unpractical, which is
why we only discuss Stokes-shifted
Raman scattering here. The dispersion
and acquisition of these scattered photons results in a spectrum that contains
discrete peaks, where each peak corresponds to a distinct molecular vibration.
Since molecular vibrations are strongly
dependent on the conformation of a
molecule and its chemical environment,
spectral analysis of the scattered photons
can be used to not only identify a
molecular bond, but also to assess the
chemical micro-environment in which it
is found.
Although Raman spectroscopy can
provide substantial chemical information about single cells, the signals it
provides are very weak when compared
with fluorescence, with typical scattering cross-sections of approximately
10-30 cm2. Several techniques have
been developed to ‘‘boost’’ this signal,
including resonance Raman scattering,
surface-enhanced Raman scattering,8
and coherent Raman scattering.9–12 Each
one of these techniques has its own pros
and cons in the analysis of single cells,
which could be detailed in a separate
review of its own, but for brevity, we
limit our discussion of Raman scattering
in this review solely to the use of
spontaneous Raman scattering in the
analysis of single cells. We briefly
summarize recent advances in instru-
ment development, multivariate statistical analysis of complex hyperspectral
data cubes containing Raman spectra,
and specific applications that demonstrate the power of Raman spectroscopy
for single-cell analysis.
RECENT DEVELOPMENTS IN
INSTRUMENTATION FOR
SINGLE-CELL RAMAN
SPECTROSCOPY
Confocal and Laser Tweezers Raman Spectroscopy. The instrumentation side of Raman spectroscopy has
undergone a quantum leap since Sir
C.V. Raman first discovered and reported on it in A New Type of Secondary
Radiation13 in 1928, which was later
named Raman scattering in his honor.
Sir Raman observed the first spectra of
light that was scattered by molecular
bonds by focusing sunlight into a
sample by using a telescope with an 18
cm–diameter large lens. For filters, he
used colored glasses. Even though in
further experiments he began using
mercury vapor lamps and more compact
spectrometers to observe the inelastic
light-scattering properties of gases and
liquids,14 it would have been impossible
to acquire Raman spectra of microscopic
biological samples, i.e., cells, with such
a setup.
The basic components, however,
remain the same. The performance of
the components, however, has changed
remarkably. Thanks to the development
of lasers, high-quality optical filters, and
modern charge-coupled device (CCD)
cameras and similar detectors, Raman
spectra can now be acquired directly
from almost any biological sample, with
little difficulty.15–17 To acquire Raman
spectra of biological samples, there are
some key requirements that have to be
satisfied. Autofluorescence from the
sample and the sample substrate has to
be minimized or creative ways for
isolating the different signals have to
be developed, and sample damage has to
be avoided. The majority of Raman
microspectroscopy systems are designed
in a confocal-like optical arrangement,
which helps acquire Raman spectra
from very small volumes, on the order
of 1 femtoliter (fL), and allows us to
reject signals that are not generated in
the very focus of the laser beam. By
appropriately choosing the excitation
source, one can not only minimize
residual sample autofluorescence, but
also reduce potential sample damage.
Depending on the type of application, a
variety of different wavelengths have
been used in the literature, the most
common being 632, 647, and 785, as
well as 532 nm and very recently, 1064
nm. The choice of the proper excitation
wavelength has to be weighted carefully, because there are advantages and
disadvantages on either side. Signal
intensity and detector performance increase with shorter wavelengths, but so
do sample autofluorescence and sample
damage. For example, it has been shown
that excitation at a wavelength of 532
nm is, generally speaking, suitable for
single-cell measurements, but it will
lead to rapid cell damage, and it also
cannot be used for tissue characterization due to strong autofluorescence.
Since the early days of single-cell
Raman spectroscopy, Puppels et al.18
have shown that individual chromosomes and cells degraded when probed
at 514.5 nm wavelength, but can
withstand extended exposure to light
above 660 nm. Min and colleagues19
also found that reasonably backgroundfree Raman spectra from fresh human
lung tissue can be acquired with 1064
nm excitation wavelength, but not with
the more commonly used 785 nm
wavelength. Similarly, Ando et al.20
have shown that it is possible to acquire
spectra of carotenoid, chlorophyll a, and
phycocyanin in highly fluorescent cyanobacterial cells by using 1064 nm
excitation. The drawback of using
wavelengths above 1 lm is, however,
the significantly lower quantum efficiency of the detectors (e.g., an InP–
InGaAsP near-infrared detector from
Hamamatsu has a quantum efficiency
of only about 4% for [0.9–1.4] lm) and
the significantly reduced scattering
cross-section. Thus, excitation with
inexpensive sources at 633 nm or, more
commonly 785 nm, is a compromise
that many researchers seek.
A simple confocal Raman spectroscopy setup is shown in Fig. 1a. In this
example, the excitation beam is being
delivered to the sample by a singlemode fiber, which produces a clean
Gaussian-beam profile. The excitation
source is a 785 nm continuous-wave,
fiber-coupled laser with an optical
output power of 50 mW, after the fiber.
The beam is recollimated after exiting
the fiber by using a 25 mm focal length
achromatic lens and expanded to a
diameter of 8 mm. To remove spectral
side-wing contributions from the source
and to ensure monochromatic illumination, the collimated beam is then passed
through a narrow 785 nm bandpass
filter. The beam is reflected by a 785
nm dichroic filter (DC) to the objective
lens. Signals generated in the focal plane
are collected by the same objective lens
and separated from the excitation source
by DC and a sharp-edge long-pass filter.
To enable confocal detection, the signal
is then focused into a multimode fiber
with 105 lm core diameter so that the
fiber entrance acts as a pinhole and
ensures confocality. The multimode
fiber is connected to an imaging spectrograph. In our experiments, we typically use a 300 groove/mm grating to
disperse the signal onto a back-illuminated, deep-depletion CCD camera. A
deep-depletion CCD camera is needed
for excitation wavelengths at 785 nm
due to its enhanced quantum efficiency
in the near-infrared, and to avoid the
occurrence of interference fringes due to
reflections that appear at these wavelengths in regular back-illuminated CCD
cameras.
As mentioned earlier, there are several other obstacles to the application of
Raman spectroscopy to single cells that
need to be overcome; autofluorescence
in the substrate, e.g., glass coverslips,
being one of them. Although such issues
can be avoided by using magnesium
fluoride or quartz coverslips, such
coverslips have to be custom made and
are very expensive and extremely brittle. An alternative way of reducing
substrate autofluorescence and of improving the signal-to-noise ratio is the
operation of the confocal microscopy
setup in the laser-tweezers mode. The
setup shown in Fig. 1 can also be used
for laser tweezers Raman spectroscopy
(LTRS), when using an objective lens
with a high numerical aperture (typically either a water- or oil-immersion
objective lens).
When an excitation beam with a
Gaussian profile is focused tightly by
such an objective lens, optical gradient
forces and momentum transfer lead to
the optical manipulation of small, micron-sized dielectric objects, which can
be optically trapped under the right
conditions. This allows for a wide range
of possibilities to manipulate and characterize not just one, but many cells in
parallel.21–23 Figure 1b illustrates the
different effects occurring during the
optical trapping process on a polystyrene bead with a diameter larger than the
focal-spot diameter. When this object is
illuminated by a beam that exhibits
differences in its intensity distribution
(as indicated in the figure by the
thickness of the arrows), it will experience a gradient force that is dependent
on the photon flux through that area.
Photons change their momentum due to
a change in the refractive index between
the medium, e.g., water, and the bead.
Due to the conservation of momentum,
the bead also experiences a momentum
that compensates for the change in the
direction of the photons. Hence, there is
a stronger force acting on the right-hand
side of the particle, moving it toward the
higher beam intensity. When the sample
is illuminated by a tightly focused beam
with a radially symmetric Gaussian
intensity distribution, the bead is pulled
into the center of the excitation due to
the presence of the higher beam intensities in the center of the Gaussian beam
(Fig. 1c).
The main disadvantage of point
spectroscopy is that cells are inherently
microscopically small biological systems with a complex substructure, and
Raman spectra from individual cellular
compartments or organelles will not
necessarily reflect the overall state of
the cell. Cell nuclei will show a rich
nucleic acid signature in their Raman
spectra, and spectra collected in the
cytosol will exhibit high protein, lipid,
and metabolite contents. Other cellular
organelles such as lipid droplets will
also produce their own signature, e.g.,
that of esterified fatty acids. Because of
this heterogeneity, it is important to
acquire Raman spectra not just an
individual point within the cell, but also
to either probe multiple spots within the
cell or the entire cell. This is achieved by
Raman microspectroscopy in pointscanning or line-scanning mode.
APPLIED SPECTROSCOPY OA
815
focal point review
FIG. 1. (A) Typical confocal microscope arrangement for point-scanning Raman
spectroscopy. A detailed discussion of the setup is provided in the main text. (B) A balance
between optical gradient forces and the transfer of photon momentum permit the optical
trapping of single cells, as is typically used in laser tweezers Raman spectroscopy. ES =
excitation source; CL = cleanup filter; L1 = lens; LP = long-pass filter; DC = dichroic filter;
Obj. = objective lens; TS = translational stage; L2 = coupling lens.
Point-Scanning and Line-Scanning
Raman Microscopy. The setup of a
point-scanning Raman microspectrometer is very similar to the Raman
spectroscopy setup shown in Fig. 1a.
The Raman signal is, however, acquired
by raster scanning the beam across the
sample. Raster scanning can be achieved
in multiple ways, e.g., by moving the
sample through the excitation beam, by
moving the objective lens relative to the
sample, or by scanning the excitation
816
Volume 67, Number 8, 2013
beam by a pair of galvanometric mirrors
across the sample. The main disadvantage of point-scanning Raman microspectroscopy is that it is intrinsically
slow, because a Raman spectrum is
acquired individually from every single
point within the sample. For example, a
40 3 40 pixel image consists of a total of
1600 Raman spectra, and with an
acquisition time of 5 s per spectrum,
the entire image acquisition time is 133
min, which, of course, is not generally
suitable for living samples. One way to
accelerate the signal acquisition of
Raman spectra is to multiplex the beam
geometrically. Instead of recording a
Raman signal from a single point, one
can, e.g., generate a line spot in the focal
plane, and excite and detect Raman
scattered signals from the entire line.
The general layout of a line-scanning
Raman microspectrometer is shown in
Fig. 2a. Here, we use a high-power laser
system with an excitation wavelength of
785 nm, with a single-mode fiber output
option. After exiting the fiber, the beam
is collimated and it passes a narrow 785
nm laser-line bandpass filter to ensure
monochromatic excitation. An achromatic cylindrical lens reshapes the
Gaussian beam with the lowest fundamental transverse mode into a line
profile. The cylindrical lens sits on a
rotation stage to ensure that the line is
properly imaged onto the entrance slit of
the spectrometer. The line profile is
imaged by an achromatic lens to the
back aperture of an objective lens, which
focuses the beam in the sample plane.
The sample is held on a motorized flat
top XY translation stage for inverted
microscopes. The Raman signal generated from the line focus is collected by
the same objective lens and separated
from the excitation source by a 785 nm
dichroic long-pass filter. The Raman
signal then passes a long-pass filter and
is ultimately imaged onto the entrance
slit of the spectrometer. A grating is
used to disperse the Raman signal from
the line focus, and to image it in its
entirety onto a back-illuminated, deepdepletion CCD camera.
Figure 2b shows images of the CCD
camera inside the spectrometer for both
the point-scanning mode and the linescanning mode. In the upper image, the
spectrum obtained from a single point at
the sample is dispersed horizontally
across the CCD chip. In the lower
image, obtained in line-scanning mode,
the signal is distributed along a line in
the vertical direction, which is then
dispersed by the grating horizontally
across the CCD chip. After the image
acquisition, the information is represented as a hyperspectral data cube, as
shown in Fig. 2c, where each pixel
element is an individual Raman spectrum.
FIG. 2. (A) Expansion of a laser beam along one axis by a cylindrical lens leads to the illumination of an entire line across the sample. When
imaged onto the entrance slit of an imaging spectrometer, the entire line can be dispersed across a CCD chip, and tens of Raman spectra
can be acquired simultaneously. (B) Differences between the CCD images of Raman-scattered photons obtained by point-scanning and linescanning Raman scattering. (C) By scanning an expanded line across the sample, the rapid acquisition of hyperspectral data cubes
containing spectral as well as temporal information on the biochemical state of a cell is facilitated. ES = excitation source; F = fiber; CL =
cleanup filter; L1–5 = lens; LP = long = pass filter; CL = cylindrical lens; DC = dichroic filter; Obj. = objective lens; TS = translational stage;
L2 = coupling lens.
MULTIVARIATE DATA
ANALYSIS
Raman spectra provide a wealth of
information about the chemical composition of a cell, which can be used to
reliably identify major macromolecules,
such as nucleic acids, proteins, carbohydrates, and lipids. This ability makes
Raman spectroscopy a valuable tool for
the investigation of various biochemical
aspects of living cells and their dynamic
changes. Chemical information from
volumes as small as 1 fL can be
obtained, and this can be used to identify
even minute changes in the chemical
composition on the tissue-, cellular-, and
subcellular-length scales.24–28 Raman
spectra from complex samples such as
single cells are, however, difficult to
analyze due to the simultaneous presence of prodigious amounts of organic
molecules that can exhibit a large
number of sometimes overlapping molecular vibrations that are hard to
distinguish. This fact is further complicated when large numbers of spectra are
collected by point or line scanning
across a cell. This is where multivariate
statistical data analysis shows its
strength and provides us with a handle
to better visualize the information contained within the spectra.
Figure 3a shows the averaged Raman
spectrum of a T-lymphocyte that was
probed by line-scanning Raman micro-
spectroscopy. Some Raman peaks of
interest are indicated in the figure. The
assignment of the peaks highlighted in
the figure, their associated molecular
bonds, and the corresponding biomolecules are summarized in Table 1.29–31
Figure 3b shows images of the Tlymphocyte, corresponding to the intensity of each highlighted Raman peak.
Because of significant intensity variations
between the Raman peaks, each image
was intensity scaled to better visualize the
corresponding cellular distribution of the
biomolecule. With this image, it is
possible to investigate the presence and
the location of particular molecular bonds
and compare their relative signal intensities. For example, one could determine
APPLIED SPECTROSCOPY OA
817
focal point review
FIG. 3. (A) Raman spectrum of a living T-lymphocyte obtained in line-scanning mode.
Here, all spectra were averaged to produce the average spectrum of the cell. Molecular
vibrations associated with different macromolecules are indicated in the spectrum. (B)
Scaled Raman images of the cell, where the intensity value of the indicated molecular
vibrations highlighted in (A) is represented by the color of each pixel of the image.
the DNA distribution throughout the cell,
based on the intensity of the 785 cm-1
ring-breathing mode or density variations
of proteins within the cell, based on the
1003 cm-1 phenylalanine peak. This
method works well for isolated peaks
that are well characterized, but a Raman
spectrum of cells typically consists of a
large number of peaks that often overlap
or partially overlap. In this case, it can be
very challenging to trace minute spectral
changes during an experiment in which
thousands of spectra are acquired within
a short time. Thus, multivariate statistical
analysis has become increasingly important in the analysis of Raman spectra.
To understand multivariate statistics,
we have to remind ourselves that a
Raman spectrum consisting of 1000þ
data points (e.g., pixels on the CCD
detector) can also be represented as a
818
Volume 67, Number 8, 2013
vector that is pointing to a specific
location within an n-dimensional vector
space, where n is the number of data
points. Despite its complexity, a set of
Raman spectra has a low-rank structure,
meaning it can be explained by fewer
uncorrelated variables, because peaks
are spread across several data points,
and groups of peaks can be assigned to
the same molecule, so they will scale
together. This transformation to a reduced set of variables is commonly done
by multivariate statistical analysis
(MVA) and chemometrics. Depending
on the particular problem, different
MVA methods can be utilized. Frequently found methods used in the
analysis of Raman spectra are various
ways of least-squares fitting procedures,
principal component analysis (PCA),
cluster analysis, and blind source sepa-
ration (BSS). Here, we describe some of
these methods and provide examples of
their applicability.
Principal Component Analysis. A
widely used MVA method to analyze
Raman spectra is PCA, because it is
readily implemented in most standard
software packages for data acquisition
and data analysis.32–34 A comprehensive
review of PC analysis of Raman spectra
can be found elsewhere.35 PCA applies a
set of linear–orthogonal transformations
to the data set, which identifies correlated variables in the data and assigns them
to new, uncorrelated variables, the socalled PCs. This reduces the dimensions
of the original data set by removing
redundancy of information.36 Moreover,
the new data set is structured based on
its overall contribution (importance) to
the original data: The first PC accounts
for the largest variance within the data
set; the second PC accounts for the
second-largest variance within the data
set, etc. Only the first few PCs contain
important information about the data;
the rest is typically noise. This results in
a simplified way to analyze and visualize complicated multidimensional data
sets such as Raman spectra.
An example of three common ways to
display information extracted by PCA is
shown in Fig. 4. A scatter plot of the
scores of the first two PCs is shown in
Fig. 4a, where the first two components
together account for more than 37% of
the total data variation. The amount of
variance accounted for by the first few
PCs is lower in Raman imaging due to
the larger data sets and the noise, which
can increase the intraspectrum variability.35 There are a total of 1600 points in
the scatter plot, with each point corresponding to a single Raman spectrum
from the same data set, as in Fig. 3.
Immediately, one can see that the data is
distributed in three distinct clusters—
this is critical information that would
have been extremely difficult to extract
from the original data by peak analysis.
Through further analytical steps (e.g., by
analyzing the ‘‘loading factors,’’ see
below) it becomes clear that the data
clusters correspond to background spectra, as well as Raman spectra due to
lipids, proteins and nucleic acids. A very
typical application of such scattering
plots is the distinction of spectra from
TABLE I.
Raman shift
(cm-1)
785
1003
1090
1260
1303
1342
1444
1571
1668
1739
2854
2930
a
Peak assignments for molecular Raman vibrations and related biomolecules.
Molecular
bond
O–P–O
d(C–H)
PO2CH2
d(CH2)
CH
CH2
C=C
C=O
CH2
CH3
Biomoleculea
DNA backbone, DNA–RNA (U, T, C) (ring-breathing mode)
Phenylalanine (protein)
Phosphate stretching (DNA)
Lipids, protein, amide III (in-plane deformation)
Twisting, wagging, protein, fatty acids (twisting, wagging)
DNA–RNA (G) (deformation)
Fatty acids, lipids, proteins (bending and scissoring)
DNA–RNA (G, A) (ring-breathing mode)
Protein, a-helix of amide I, unsaturated fatty acids (stretching)
Esters, phospholipids, esterified fatty acids (stretching)
Fatty acids, lipids, proteins (asymmetric stretch)
Proteins (symmetric stretch)
A = adenine; C = cytosine; G = guanine; T = thymine.
different classes of sample, e.g., cell
type,37 drug exposure,38 or cell cycle
stage.39
One way to determine differences
between the PCs is an analysis of the
loading factors of the corresponding PC,
Fig. 4b. The first spectrum in this figure
shows the mean Raman spectrum of the
T-lymphocyte, as a visual aid. The other
two ‘‘spectra’’ show the loading factors
of the first and second PC. The loadings
can be understood as weighting functions that correlate the original values of
the variables to the new values in the
reduced data set, the scores. Thus, the
first loading factor clearly shows that
positive score values of the first PC are
associated with the actual Raman spectrum of the cell. Accordingly, negative
scores of the first PC are associated with
background spectra. This can also be
seen in the scatter plot Fig. 4a, where the
background Raman spectra have negative score values for PC1, and the
Raman spectra of the cell have positive
score values for PC 1. The second
loading shows the locations of peaks
that discriminate between macromolecules in the original spectra. In this case,
the presence of peaks that can be
assigned to lipid vibrations in a spectrum are associated with positive score
values along PC2, and peaks assigned to
nucleic acid and protein vibrations are
associated with negative score values
along PC2. This can, again, be compared
with the distribution of the score values
in the scattering plot Fig. 4a.
It is also possible to apply PCA to
visualize the data in the form of an
image. However, instead of reconstruct-
ing the images based on individual
peaks and their respective intensity
values, the image is constructed based
on the PCs of the spectra for every pixel.
Figure 4c shows the score image of the
first and second PCs of the same data set
as in the previous examples. This image
provides a good visual representation of
the information that was extracted in
Figs. 4a and 4b. The first PCs show the
location of the cell versus the background, i.e., it basically shows the
outline of the cell, while the second
PCs shows the distribution of cellular
macromolecules, i.e., lipid, protein, and
nucleic acid. The typical kidney-shaped
nucleus of a T-cell can be seen in this
image as the object shown in blue
(negative PC2 values) on the right-hand
side of the cell, while lipid-containing
structures are shown in red and yellow
on the left-hand side of the cell. These
images match the previously shown
images based on the peak intensity
values of the DNA–RNA peak at 785
cm-1 and that of the lipid peak at 1441
cm-1 quite well.
Cluster Analysis. Cluster analysis
provides an alternative way of reconstructing information contained in a
hyperspectral data cube obtained by
Raman imaging of cells.40,41 In cluster
analysis, data clusters are established
based on spectral similarity, e.g., Euclidean distances. The clusters found by
this method will be assigned a color,
which can then be used to display the
location of the constituents of each
cluster in the image. Two clustering
methods are frequently found in Raman
spectroscopy, k-means cluster analysis
and agglomerative hierarchical cluster
analysis (AHCA).
In traditional k-means cluster analysis,
the algorithm chooses at random a
defined number k of initial starting
locations and subsequently compares
the spectral distance between the centroid spectra and spectra assigned with
the initial cluster. After minimizing the
sum of distances, a new centroid
spectrum is determined, and spectra not
falling into the cluster are assigned to a
different cluster. This process is performed iteratively and stops when no
more spectra can be reassigned to a
different cluster. Once all clusters are
established, a color code is applied to
represent the cluster groups in an image.
AHCA, on the other hand, is a
bottom-up clustering approach in which
every initial spectrum is a starting
cluster. On every iteration, the spectra
are grouped together until only one
cluster remains. The cluster groups can
be displayed in form of a dendrogram
(or tree); the user can then choose the
number of clusters of interest.
As an example for k-means clustering
applied to Raman imaging, we used the
same data set obtained by Raman lineimaging of the T-cell that was used in
the previous examples. Figure 5a shows
the pseudocolored image of the cell,
based on three, five, seven, and nine
clusters. It can be seen that each
additional cluster reveals new details
about the cell, but it is important to keep
in mind that too many clusters can
obscure important information. Figure
5b shows the mean spectra for the
indicated clusters for the case of k = 5.
Here, we can again see that the different
clusters can be attributed to differences
in the distribution of the major cellular
macromolecules. The mean spectrum of
cluster 1 shows distinct peaks that can
be assigned to DNA–RNA and protein
vibrations, and it is most likely outlining
the cell’s nucleus, which is shown by the
red in the cluster image. This observation is consistent with information
extracted by PCA. The mean spectrum
of cluster 2 shows a decrease in peak
intensities of DNA–RNA modes, while
Raman peaks that can be assigned to
protein modes are still highly present.
The mean spectra of clusters 3 and 4
emphasize the phosphatidylcholine peak
APPLIED SPECTROSCOPY OA
819
focal point review
FIG. 4. PCA applied to a Raman image data set obtained by line scanning of a human lymphocyte. (A) Scatter plot of the score values of
each spectrum for the first and second PC. The first PC separates the data, based on association with cell spectra and background spectra.
The second PC separates the spectra, based on the signature of specific macromolecules, i.e., lipids, nucleic acids, and proteins. (B)
Reference Raman spectrum from the cell Fig. 3a and loadings for the first and second PC. The loading for PC1 contains Raman peaks that
are positively associated with the cell; the loading for PC2 reveals Raman peaks that are positively associated with lipids and negatively
associated with protein and nucleic acids. (C) False-colored, reconstructed images of the cell based on the score values for the first and
second PC. Here, again, the PC1 image clearly distinguishes the cell from the surrounding background, while the PC2 image shows a clear
separation between lipid-rich areas, and protein- and nucleic acid-rich areas.
even more than spectra of clusters 1 and
2, highlighting a phospholipid class that
is a part of the cellular plasma membrane. This can also be seen by the
location of the signal in the figure,
which occurs mostly on the circumference of the cell. Cluster 5 exhibits the
strongest resemblance of a lipid Raman
spectrum. This cluster indicates mostly
neutral lipid deposits, i.e., triglycerides
that are found in cellular lipid droplets,
which can be seen by the strong signal
of the triglyceride C=O peak at 1739
cm-1.
Blind Source Separation. The capability of Raman spectroscopy to provide high chemical specificity is due to
820
Volume 67, Number 8, 2013
the fact that every molecule has a
unique, identifiable Raman signature,
even if it is present as part of a complex
mixture—the kind of environment that
is typically found in a biological cell.
The Raman spectrum of a mixture of
biomolecules is a linear combination of
Raman spectra of individual biomolecules mixed in an unknown ratio. It is
possible to determine the relative
amount of a single compound in that
mixture by fitting Raman spectra from
pure components to the experimental
spectrum by using least-squares optimization. This approach has been
widely applied to determine the composition of biomolecules in cells and
tissue by spontaneous Raman spectroscopy.25,42–44
The composition of the sample is,
however, not always entirely known,
and at times, it might be cumbersome or
impossible to perform an a priori
chemical analysis to determine the exact
composition of the sample. Several
methods have been developed that can
extract the ‘‘pure’’ components of a
sample by spectral unmixing, also
widely known as BSS. Two methods
that have frequently been applied to
Raman spectroscopy are independent
component analysis45 and vertex component analysis (VCA).46 PCA is, strictly speaking, also a method of BSS. At
mentioned before, the end members
determined by VCA do not necessarily
result in real pure components—in this
case, nucleic acids, proteins, and esterified fatty acids. They can also reflect a
mixture of all components that is
weighted toward a specific component.
For example, the VCA spectrum reflecting cell nuclei contains not only the
signature of nucleic acids, but also that
of proteins. Similarly, the extracted end
member–reflecting proteins also contain
an RNA signature. The only end member that is very close to the pure
component is the spectrum of the
esterified fatty acids, mostly because
fatty acids can be found in cellular lipid
droplets at high local concentrations,
resulting in a few pure spectra in the
original data set.
BIOLOGICAL APPLICATIONS
FIG. 5. k-Means clustering analysis of the same lymphocyte shown previously. (A)
Examples of cluster-based images obtained for an increasing number of clusters used for kmeans clustering. Different, spatially distinct cellular structures emerge with an increasing
number of clusters. (B) Mean spectra obtained for the differently colored cluster groups
indicated by numbers for the case of five clusters. The biochemical components that are
identified by cluster analysis of the spectra are indicated, and their spatial distribution
within the cell can be seen in the five-cluster image in (A).
this point, however, we only consider
methods that can extract pure component spectra that represent spectra of
biomolecules.
Vertex component analysis is an
unsupervised spectral unmixing method
that can be used to determine pure
components from a data set. The method
is based on the assumption that if a large
number of measurements have been
acquired, that there will be a few data
points in the set that contain information
of pure components. The algorithm
determines end members (pure components) that span a subspace in the data
vector space. The algorithm then iteratively projects the data orthogonal to
that subspace. The new end members are
the extreme points in the projection.47
One has to keep in mind that VCA
cannot always extract pure component
spectra. In particular, this is the case
when the data set is relatively small and
does not contain pure component spectra. In this case, the extracted spectra
will not necessarily reflect the real pure
components. Once VCA has determined
the pure components, they can be fitted
in a least-squares fashion to the projected data set and then used to reconstruct
an RGB image.
Figure 6a shows the VCA image of
the same T-cell that we used in our
previous examples. The three pure
components that VCA extracted were
used to reconstruct the VCA image in
Fig. 6b. The resulting image has a more
pleasant appearance than the stark
contrast provided in the PCA image,
Fig. 4c, or the cluster analysis image,
Fig. 5a. As before, the extracted end
members have high degree of similarity
to spectra of nucleic acids, proteins, and
esterified fatty acids. As has been
Cell Cycle Analysis by Raman
Spectroscopy. The life cycle of a cell
constitutes the most fundamental chain
of events in any living cell, characterizing the phases and processes that lead to
replication or cell division in prokaryotes and eukaryotes alike (see Fig. 7).
Currently, mostly fluorescence microscopy or flow cytometry are utilized to
follow the cell cycle. This does, however, require fluorescent staining or
transfection with constructs that encode
for fluorescent proteins to render cellular
structures visible that is not applicable
for cells in vivo, e.g., in an animal.
Raman spectroscopy can provide a
viable alternative to these established
fluorescence-based methods, requiring
no sample preparation, and providing
localized information about a cell’s
nucleic acid, protein, and lipid contents.
Raman spectroscopy has previously
been shown to provide the ability to
distinguish between highly proliferating
cells such as neoplastic tumor cells and
normal cells, based on differences in
their DNA and protein content. The cell
cycle itself can, on the other hand, have
significant implications for the results of
Raman measurements. This can go as far
as camouflaging subtle cellular changes
or changes in a cell’s metabolism, which
could be mistaken for increased proliferation. In this section, we review the
current state of the art on analyzing the
cell cycle by Raman microspectroscopy.
APPLIED SPECTROSCOPY OA
821
focal point review
FIG. 6. Image and major component spectra obtained by vertex component analysis of the same data set used in the previous figures.
Short et al.48 investigated tumorigenic
and nontumorigenic suspension cells in
two proliferating states; plateau cells
were defined as cells with low proliferation index, and exponential cells defined as rapidly proliferating cells,
modeling both normal tissue and solid
tumors, respectively. By fitting highly
abundant biomolecules, i.e., lipid, protein, DNA, and RNA signatures into
their spectra, the group was able to show
that changes in the relative abundance of
these biochemical compounds were
associated with differences in their
proliferation. A thorough Raman spectroscopic investigation of osteosarcoma
cells synchronized in different phases of
the cell cycle, i.e., G0–G1, S, and G2–M,
was performed by Swain et al.49 By
applying linear discriminant analysis
(LDA) to Raman spectra acquired at
different phases of the cell cycle, they
were able to discriminate between cells
in G0–G1 phase and cells in G2–M
phase, with 97% specificity. Matthews
et al.,39 on the other hand, conducted a
study on the variability of Raman
spectra due to the influence of the cell
cycle and the confluence of a cell
culture. Raman spectra of resuspended
adherent cells were acquired, and the
resulting data were subjected to PCA
analysis. This enabled the authors to
show that the first PC reflected changes
mostly related to the variability in
nucleic acid and protein content relative
822
Volume 67, Number 8, 2013
to the lipid content. Those changes were
correlated to cell cycle phase-to-phase
transitions, i.e., G1–S and G2–M, and
validated by flow cytometry. Huang and
coworkers50 investigated cell cycle–related changes in molecular composition,
and molecular distribution in yeast cells
in time-lapsed Raman-based imaging.
Time-lapsed images based on Raman
peaks associated with lipid and protein
vibrations were used to show the
distribution of biochemical components
during the G2–M phase and G1–S phase.
Single-cell Raman spectroscopy is
now frequently found in combination
with other optical methods, especially
multiphoton microscopy. For example,
Pully et al.51 developed a hybrid twophoton excitation fluorescence (TPEF)
microscope that can also acquire local
Raman spectra and used this versatile
device to investigate nucleotides in
human bone marrow stromal stem cells.
By combining these two techniques, the
group was able to visualize cellular
organelles by using TPEF microscopy,
such as the nucleus stained with a
nucleic acid stain (DAPI or Hoechst
33 342), to then sub-sequentially acquire
Raman spectra from the stained nuclei.
This enabled them to make correlative
statements about the connection between fluorophore-stained regions within the cell and their chemical
composition by Raman spectroscopy.
Another technical extension to investi-
gate the cell cycle was developed by
Pliss et al.,52 who, in addition to the
hybrid TPEF microscopy and Raman
spectroscopy system, added the ability
to perform coherent Raman imaging by
coherent anti-Stokes Raman-scattering
microscopy. This modification allowed
them to investigate changes in macromolecular organization (fluctuations in
local density, structural modifications)
in HeLa cells in different phases of the
cell cycle. Matthäus et al.,53 on the other
hand, combined Raman spectroscopy
with infrared absorption spectroscopy
to image the DNA and protein distribution, allowing them to make direct
comparisons between the techniques
and visualize different stages of mitosis
without fluorophores. Such and similar
studies are currently being pursued by
an ever-increasing number of groups
because of a growing interest in the
label-free characterization of cellular
changes leading to the evolution of
cancer, as well as in their potential use
as a means to distinguish and characterize stem cells for therapeutic applications.
Studies of Drug–Cell Interaction.
Since Raman spectroscopy provides
chemical specificity in volumes as
small as 1 fL in living cells, it is also
a prominent tool in the evaluation of the
effects of small molecules, i.e., pharmaceutical drugs, on cells. Not only can
the presence of the drug in the cell be
FIG. 7. This diagram shows a simplified representation of the eukaryotic cell cycle. During
the different phases of the cycle the cells undergo a variety of significant changes, such as
protein increase in the G1 phase, chromosome duplication in the S phase, ensuring that the
cell is ready for division in G2 phase, and process of cell division in the M phase.
probed, but also the resulting effects of
the drug on the cell—again, without the
need for additional, potentially interfering fluorophores—to facilitate the investigation of drug–cell interactions in
vitro as well as in vivo. In recent years,
Raman scattering–based spectroscopy
and microscopy has been extensively
used to determine the effects of chemotherapeutic agents on living eukaryotic
cells, as well as antibiotics and other
small molecules on prokaryotic cells.
Raman spectroscopy is particularly
interesting in this regard, because it
enables one to follow the long-term
effects of drug treatments on cells,
potentially identifying cells that avoid
apoptosis and determining the minimum dose needed to avoid minimal
residual disease, a particular problem in
the fight against the recurrence of
cancer. Here, we summarize some of
the most recent and most significant
work in this area.
Nawaz et al.54 applied Raman spectroscopy to determine the response of
lung adenocarcinoma cells to the chemotherapeutic drug cisplatin and investigated the effects of different drug
concentrations for an incubation time
of up to 96 h. In order to assess the
potential of Raman spectroscopy for this
application, the group was able to
establish a correlation between a classic
MTT [3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide]–based cytotoxicity assay and Raman measurements of cells under the same treatment
conditions by applying partial least
squares regression to the spectra. The
effects of cisplatin treatment on carcinoma cells in conjunction with Raman
spectroscopy has also been studied by
Huang et al. 55 This group treated
nasopharyngeal carcinoma cells with
cisplatin by using different drug concentrations and obtained spectra at
multiple time points within 24 h. The
results of the analysis showed that the
measured cellular changes were correlated with a decrease in Raman vibrational-band intensities associated with
nucleic phosphodiester bonds and DNA
bases, indicating a direct degrading
effect of the drug on the genome.
Though it has been established that
Raman spectroscopy is sensitive enough
to determine drug-induced changes in
cells at high drug concentrations, there is
also an increasing interest in research
exploring the sensitivity of Raman
spectroscopy for treatments at low or
even microdose levels. For example,
Draux et al.56 investigated the effects of
noncytotoxic doses of the anti-cancer
drug gemcitabine. In their study, human
lung carcinoma cells were exposed to
drug concentrations as low as 1 nM at
time points of 24 h and 48 h. The
corresponding drug response was determined based on the comparison of the
Raman signature of the intact cells, as
well as chemically extracted biomolecules such as DNA, RNA, and proteins.
Their analysis showed that incubation
with 100 nM gemcitabine for 48 h
resulted in a measurable DNA decrease
and measurable protein increase. A
different study investigated the apoptotic
effects of fluorouracil on gastric carcinoma, showing that major drug-induced
changes are related to reduced intensities
of vibrational bands associated with
lipids, proteins, and nucleic acids.57
In contrast to these imaging-based
studies, Moritz et al.38 applied laser
LTRS to probe non-adherent human
cells, which allowed them to not only
assess these cells in a native state, but
also to significantly reduce background
interference from any substrates onto
which the cell would have adhered
otherwise. They utilized this technique
to follow the effects of doxorubicin
treatment on T-lymphocytes. By obtaining spectra of T-cells exposed to doxorubicin at multiple time points over a
period of 72 h, they were able to
determine distinct cellular spectral
changes occurring at different stages of
drug-induced apoptosis. Similarly, Lin
and colleagues58 also used doxorubicin
APPLIED SPECTROSCOPY OA
823
focal point review
to induce apoptosis in cells and studied
the effects on extracted nuclei by using
Raman spectroscopy. While more invasive, this technique has the advantage of
relating specific changes occurring in
chromatin structure to their corresponding Raman spectra, which also allowed
them to identify effects occurring at
even lower drug concentrations. Conceptually, however, both techniques
probe the cell’s nuclei, because LTRS
traps human lymphocytes by their nucleus.59 In contrast, Buckmaster et al.60
developed a biosensor that utilizes
patterned single-cell arrays in conjunction with confocal Raman spectroscopy
to repeatedly probe large numbers of the
same cells over an extended period. This
format enabled them to show that in
living human medulloblastoma cells
treated with the chemotherapeutic drug
etoposide, Raman peaks associated with
DNA and protein vibrations decreased
with time.
There is, however, a particular problem that occurs when studying cellular
changes based on Raman scattering in a
purely spectroscopic mode, because
most studies address specific subcellular
structures with diffraction-limited spot
sizes rather than the entire cell. Cells,
however, are highly complex biological
systems, and drug-induced changes in
cells are heterogeneous in nature. This
means that spectral changes that are
observed in Raman spectra obtained
from one organelle, e.g., the nucleus
can be significantly different from
changes observed in the cytoplasm.
Thus, point spectroscopy can lead to
potential problems and significant variations between results obtained from the
same treatment groups, and this can
easily obscure specific changes that
could be assessed by either analyzing
or imaging the entire cell. While approaches to averaging Raman scattering–based data over the entire volume of
a cell, such as spatial dithering of the
laser beam or shaping the beam to focus
to larger volumes have been used, a
more common approach to overcome
such problems is the use of Raman
scattering in the microscopic imaging
mode, resulting in the acquisition of
Raman spectra from every point within
the cell. In early work along these lines,
Uzunbajakava et al.61 developed a
824
Volume 67, Number 8, 2013
confocal Raman imaging system and
used it to investigate the effects of taxol
on HeLa cells. Zoladel et al.62 built a
point-scanning Raman microspectrometer and used it for time-lapse imaging of
living human breast cancer cells undergoing apoptosis after exposure to 300
lM etoposide. Hartmann et al.63 showed
that Raman images of MCF-7 breast
cancer cells correlate very well to the
contrast revealed by hematoxylin and
eosin staining of cells, the classical tools
used in pathology, and found that
Raman spectra from the cell nuclei
could be used to distinguish between
treated and untreated MCF-7 cells.
Identification of Cells by Raman
Spectroscopy. There are several challenges in modern medical diagnostics
that can severely affect the outcome of
medical treatment or success of a
surgical procedure, in particular when
it comes to resecting tumors. A particular challenge is the in vivo assessment
of the boundary between benign and
malignant tissue. In vivo Raman spectroscopy could provide a unique opportunity to guide surgeons during a
surgical procedure to minimize the
possibility of recurring cancers and also
to minimize the excision of otherwise
healthy tissue. As such, a number of
studies have evaluated the potential of
Raman spectroscopy as a means to
identify and distinguish between different tumorigenic and normal eukaryotic
cells.15,64,65 Moreover, due to their high
specificity, Raman-based methods could
also be potentially used in the future as a
means of label-free cell sorting66 or to
distinguish between different states of
differentiation of embryonic stem
cells.67
In early work, Nothinger et al.68 used
Raman spectroscopy to discriminate
between different bone cell types that
are commonly used in tissue engineering
of bone. By applying LDA to Raman
spectra of tumor cell lines, the authors
found that that they could discriminate
tumor cells from normal cells, with 96%
specificity. In another early study, Raman spectroscopy was applied to randomly mixed cell populations of human
leukemia and human breast cancer cells,
and human sarcoma and MCF-7 human
cancer cells.69 This study indicated that
based on their spectra, a good separation
between unmixed cell types can be
obtained, but surprisingly the study fell
short of truly separating cells within
mixed populations. After these studies,
in 2006, Rösch et al.70 showed that
Raman spectra of yeast cells provide
sufficient information to distinguish
between different types of yeast cells.
Chan et al.37 then used Raman spectroscopy to discriminate between fixed
normal human lymphocytes and transformed Jurkat and Raji lymphocyte cell
lines, showing that they were able to
detect cancer cells, with a sensitivity of
98.3% and classifying cells belonging to
normal or transformed type, with a
sensitivity of 97.2%. Similarly, Notingher et al.71 used Raman spectroscopy to
discriminate between healthy and tumor
bone cells, with an accuracy of 95%.
After their earlier success, Chan et al.72
used LTRS in a clinical study to
discriminate between T- and B-lymphocytes from four healthy individuals and
cells from three leukemia patients,
showing a positive classification of
95% of cells from the healthy individuals and a 90% positive classification of
the patients’ cells.
Starting in the late 2000s and after
earlier work on mouse stem cells,
Raman spectroscopy was also shown
able to distinguish between cells derived
from embryonic stem cells, progenitor
cells, and differentiated cells. In early
work on human stem cells, it was shown
that human embryonic stem cells
(hESCs), hESC-derived cardiomyocytes
(CM), and human fetal left-ventricular
CMs, can be distinguished based on
their Raman signature, with accuracies
of 98, 66, and 96% respectively.73 In
extensive work further analyzing the
distinctive differences between human
stem cells and their derivatives, Konorov et al.74–76 were able to show a major
discriminant between the different cell
types is their glycogen content. In very
recent work, Ghita et al.77 investigated
cytoplasmic RNA in undifferentiated
neural stem cells (NSC) as a potential
marker to assess their degree of differentiation. By applying LDA, Ghita and
colleagues were able to discriminate
between differentiated and undifferentiated NSC with a sensitivity of 89.4%
and specificity of 96.4%. This work has
significant potential because of the
difficulty of obtaining fresh human
neural cells and their importance in
pharmaceutical studies related to the
development of novel drugs for treating
neural disorders or tumors.
If the throughput and specificity of the
Raman-based evaluation of single cells
can be further improved, then rapid cell
sorting in the absence of fluorescent cell
markers or in applications where such
markers need to be avoided appear
likely. In a recent demonstration of this
potential, Duchow et al.78 combined
optical trapping with cell sorting in
microfluidic channels to discriminate
between erythrocytes, leukocytes, acute
myeloid leukemia cells, and two types of
breast tumor cells, with an overall
accuracy of 92.2% by using 785 nm
wavelength excitation and a specificity
of 94.9%, after switching to the excitation of Raman scattering at 514 nm.
Raman Spectroscopy of Prokaryotic Cells. After the initial demonstration
of single-cell Raman spectroscopy on
eukaryotic cells, and after extensive
work in infrared absorption spectroscopy, it was quickly shown that Raman
scattering is also impressively sensitive
when it comes to the identification of
prokaryotic cells and its ability to
distinguish between different prokaryotic cells types. Similar to its potential for
assessing mammalian cells, this offers a
wide range of possibilities, e.g., for the
rapid identification of contaminations on
medical devices or potential contaminants in drugs produced in bacterial cell
culture or in food processing plants. A
number of studies have shown that
Raman scattering–based methods allow
for the assessment of the type of
contamination, as well as the degree of
contamination in such applications.79–84
In early work, Choo-Smith et al.85
investigated different clinically relevant
colonies of microorganisms by using
Raman and Fourier transform infrared
spectroscopy at different time points and
at different imaging depths. They were
able to show that the Raman signatures
from the various colonies were very
homogenous when probed at 6 h after
initiating the colony, in contrast to
Raman signals obtained from colonies
at 12 and 24 h. A more careful analysis
of the results obtained at the two later
time points revealed that the signals
were position and depth dependent, with
the surface layer showing high glycogen
content and deeper layer showing increased RNA content. Escoriza et al.86
used Raman spectroscopy to quantify
filtered waterborne pathogenic microbial
contamination, and Pasteris et al.87 used
Raman spectroscopy to identify sulfur
inclusions in sulfur-oxidizing marine
bacteria. Recently, Palchaudhuri et al.88
investigated the uptake and metabolism
of the five-carbon sugar alcohol xylitol
by gram-positive viridian group Streptococcus and two gram-negative Escherichia coli strains. In assessing Bacillus
strains, Hutsebaut et al.89 characterized
the influence of culture conditions on the
ability of Raman spectroscopy to discriminate between Bacillus cereus, Bacillus pumilus, and Bacillus
licheniformis. They were able to show
that the identification of bacterial cells
by Raman spectroscopy at the species
level is not sensitive to culture conditions. Grouping the bacteria into strains,
however, was shown to provide sensitivity to specific culture conditions. A
similar study was performed by Harz et
al.,90 who investigated various culture
conditions on different Staphylococcus
strains and compared them with strains
under constant growth conditions. This
study showed that strains under different
growth conditions could be readily
distinguished with 94.1% accuracy.
Fourier transform Raman spectroscopy
was separately used to differentiate six
different microorganisms growing on
whole apples, and to differentiate between pathogenic strains and nonpathogenic strains.91 Xie et al.92 used LTRS
to identify single bacterial cells in an
aqueous solution and discriminate between six species, based on PCA.
Krause et al. combined Raman spectroscopy with fluorescence microscopy and
were able to show that staining of
bacterial cells with fluorophores does
not affect the acquisition of Raman
spectra from the same cells93 and also
allows one to easily locate and identify
bacteria.94 Harz et al.95 used Raman
spectroscopy as a sensitive assay to
analyze bacteria in the cerebrospinal
fluid of patients with bacterial meningitis, concluding that the cerebrospinal
fluid does not mask the Raman spectrum
of a bacterial cell, and that Raman
spectroscopy could be applicable to
clinical use in assessing meningitis and
after its treatment. Tripathi et al.96
determined some key experimental parameters for bacterial imaging by Raman
spectroscopy, such as determining the
laser-induced photodamage threshold,
the composition of the water matrix, as
well as assessing organism aging in
water. Schmid et al.97 analyzed a data
set consisting of 3642 spectra from 29
different strains of bacteria and utilized
Gaussian mixture discriminant analysis
for their classification, achieving an
average specificity of 86.6%. Analysis
by support vector machines, on the other
hand, showed an average specificity of
87.3%. These examples should demonstrate that single-cell Raman spectroscopy has proven a viable tool in the
analysis of bacteria, their response to
different environmental conditions, and
their identification in complex mixtures.
CONCLUSION
In this article, we reviewed some of
the most recent and most significant
developments in the analysis of single
cells by spontaneous Raman spectroscopy. The fact that molecularly specific
information can be gained from single
living cells, without the need for exogenous stains, opens up exciting new
prospects for real-time and long-term
biochemical research at the level of
single cells in vitro and in vivo. The
physiological state of individual cells
can be assessed, e.g., by characterizing
the rate at which they process metabolites or by characterizing translational
activity of RNA within the cell and
specific cellular compartments. The
continuous development of optical technology such as more sensitive redenhanced, faster detectors; more compact and affordable, tunable fiber-laser
sources; and robust and compact spectrometers will lower the cost and
complexity of these instruments, while
enhancing their utility, such that they
will see widespread use in biomedical
research in the near future. Already,
pioneer studies have demonstrated the
potential of Raman spectroscopy for in
vivo cancer demarcation that can guide
surgical procedures or for cell sorting
without contaminating antibodies or
fluorophores.78 We believe that combi-
APPLIED SPECTROSCOPY OA
825
focal point review
nations of current technology, e.g.,
microfluidic sample delivery, liquid
chromatography, or super-resolved optical approaches will soon be combined
with Raman spectroscopy to enable even
faster sample analysis or the localization
of chemical products at subwavelengthlength scales. Raman spectroscopy has
evolved from an exotic technique that
only a handful of select research groups
were able to perform to an entire
research area by itself, enriching and
complementing other analytical approaches. Such multimodal imaging
and analysis tools are currently in the
early phases of making their way into
the market place, but based on ever
increasing demands for minimally invasive experiments and diagnostics in the
life sciences, laser Raman microspectroscopy is one of the most powerful
tools currently available to label-free
analyze and characterize living cells and
their components.
ACKNOWLEDGMENTS
We thank our colleagues at the University of
California–Davis, specifically, at the National
Science Foundation (NSF) Center for Biophotonics Science and Technology and the University of
Bielefeld for their help and contributions during
the last several years. In particular, we thank Drs.
James Chan, Deborah Lieu, Sebastian Wachsmann-Hogiu, Stephen Lane, Dennis Matthews, and
Rod Balhorn. We are grateful for the continued
support of this work by the NSF Center for
Biophotonics Science and Technology, an NSF
Science and Technology Center, which is managed
by the University of California–Davis, under
cooperative agreement no. PHY 0120999 with
the NSF.
1. A.J. Koster, J. Klumperman. ‘‘Electron Microscopy in Cell Biology: Integrating Structure and Function’’. Nat. Rev. Mol. Cell Biol.
2003. 4: SS6-SS9.
2. S.G. Boxer, M.L. Kraft, P.K. Weber. ‘‘Advances in Imaging Secondary Ion Mass
Spectrometry for Biological Samples’’. Annu.
Rev. of Bioph. 2009. 38: 53-74.
3. J. Kirz, C. Jacobsen, M. Howells. ‘‘Soft X-ray
Microscopes and Their Biological Applications’’. Q. Rev. Biophys. 1995. 28(1): 33-33.
4. M.G. Gustafsson. ‘‘Extended Resolution Fluorescence Microscopy’’. Curr. Opin. Struct.
Biol. 1999. 9(5): 627-628.
5. J. Lippincott-Schwartz, S. Manley. ‘‘Putting
Super-Resolution Fluorescence Microscopy to
Work’’. Nat. Methods. 2008. 6(1): 21-23.
6. B. Huang, M. Bates, X. Zhuang. ‘‘Super
Resolution Fluorescence Microscopy’’. Annu.
Rev. Biochem. 2009. 78: 993-1016.
7. P. Schwille, E. Haustein. ‘‘Fluorescence Correlation Spectroscopy: An introduction to Its
826
Volume 67, Number 8, 2013
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
Concepts and Applications’’. Spectrosc. 2009.
94(3): 52-67. doi:10.1002/lpor.200910041.
K. Kneipp, M. Moskovits, H. Kneipp. Surface-Enhanced Raman Scattering: Physics and
Applications. New York, NY: Springer, 2006.
Vol. 103.
J.-X. Cheng, X.S. Xie. ‘‘Coherent Anti-Stokes
Raman Scattering Microscopy: Instrumentation, Theory, and Applications’’. J. Phys.
Chem. B. 2004. 108(3): 827-840.
C.W. Freudiger, W. Min, B.G. Saar, S. Lu,
G.R. Holtom, C. He, J.C. Tsai, J.X. Kang,
X.S. Xie. ‘‘Label-Free Biomedical Imaging
with High Sensitivity by Stimulated Raman
Scattering Microscopy’’. Science. 2008.
322(5909): 1857-1861.
B.G. Saar, C.W. Freudiger, J. Reichman, C.M.
Stanley, G.R. Holtom, X.S. Xie. ‘‘Video-Rate
Molecular Imaging In Vivo with Stimulated
Raman Scattering’’. Science. 2010. 330(6009):
1368-1370.
W. Min, C.W. Freudiger, S. Lu, X.S. Xie.
‘‘Coherent Nonlinear Optical Imaging: Beyond Fluorescence Microscopy’’. Annu. Rev.
Phys. Chem. 2011. 62: 507. Abstract.
C. Raman, K. Krishnan. ‘‘A New Type of
Secondary Radiation’’. Nature. 1928. 121:
501-502.
C. Raman, K. Krishnan. ‘‘A New Class of
Spectra Due to Secondary Radiation. Part I’’.
Indian J. Phys. 1928. 2: 399-419.
J. Chan, S. Fore, S. Wachsmann-Hogiu, T.
Huser. ‘‘Raman Spectroscopy and Microscopy
of Individual Cells and Cellular Components’’.
Laser Photon. Rev. 2008. 2(5): 325-349.
I.W. Schie, T. Huser. ‘‘Label-Free Analysis of
Cellular Biochemistry by Raman Spectroscopy and Microscopy’’. Compr. Physiol. 2013.
2: 941-946.
M. Diem, M. Romeo, S. Boydston-White, M.
Miljkovic, C. Matthaus. ‘‘A Decade of Vibrational Microspectroscopy of Human Cells and
Tissue (1994–2004)’’. Analyst. 2004. 129(10):
880-885.
G. Puppels, J. Olminkhof, G. Segers-Nolten,
C. Otto, R. De Mul, J. Greve. ‘‘Laser
Irradiation and Raman Spectroscopy of Single
Living Cells and Chromosomes: Sample
Degradation Occurs with 514.5 nm but not
with 660 nm Laser Light’’. Exp. Cell Res.
1991. 195(2): 361-367.
Y.-K. Min, T. Yamamoto, E. Kohda, T. Ito,
H.-O. Hamaguchi. ‘‘1064 nm Near-Infrared
Multichannel Raman Spectroscopy of Fresh
Human Lung Tissues’’. Raman Spectrosc.
2005. 36(1): 73-76.
M. Ando, M. Sugiura, H. Hayashi, H.-o.
Hamaguchi. ‘‘1064 nm Deep Near-Infrared
(NIR) Excited Raman Microspectroscopy for
Studying Photolabile Organisms’’. Appl.
Spectrosc. 2011. 65(5): 488-492.
H. Zhang, K.-K. Liu. ‘‘Optical Tweezers for
Single Cells’’. J. R. Soc. Interface. 2008.
5(24): 671-690.
P.Y. Chiou, A.T. Ohta, M.C. Wu. ‘‘Massively
Parallel Manipulation of Single Cells and
Microparticles Using Optical Images’’. Nature.
2005. 436(7049): 370-372.
G.P. McNerney, W. Hiibner, B.K. Chen, T.
Huser. ‘‘Manipulating CD4þ T-cells by Optical Tweezers for the Initiation of Cell–Cell
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
Transfer of HIV-1’’. J. Biophotonics. 2010.
3(4): 216-223.
C.J. Frank, D.C. Redd, T.S. Gansler, R.L.
McCreery. ‘‘Characterization of Human Breast
Biopsy Specimens with Near-IR Raman
Spectroscopy’’. Anal. Chem. 1994. 66(3):
319-326.
T.J. Romer, J.R. Brennan III, M. Fitzmaurice,
M.L. Feldstein, G. Deinum, J.L. Myles, J.R.
Kramer, R.S. Lees, M.S. Feld. ‘‘Histopathology of Human Coronary Atherosclerosis by
Quantifying its Chemical Composition with
Raman Spectroscopy’’. Circulation. 1998.
97(9): 878-885.
H. Tang, H. Yao, G. Wang, Y. Wang, Y.-Q.
Li, M. Feng. ‘‘NIR Raman Spectroscopic
Investigation of Single Mitochondria Trapped
by Optical Tweezers’’. Opt. Express. 2007.
15(20): 12708-12716.
A. Bankapur, E. Zachariah, S. Chidangil, M.
Valiathan, D. Mathur. ‘‘Raman Tweezers
Spectroscopy of Live, Single Red and White
Blood Cells’’. PLoS One. 2010. 4: el0427.
R.D. Snook, T.J. Harvey, E.C. Faria, P.
Gardner. ‘‘Raman Tweezers and Their Application to the Study of Singly Trapped
Eukaryotic Cells’’. Integrat. Biol. 2009. 1(1):
43-52.
Z. Movasaghi, S. Rehman, D.I.U. Rehman.
‘‘Raman Spectroscopy of Biological Tissues’’.
Appl. Spectrosc. Rev. 2007. 42(5): 493-541.
I. De Gelder, K. De Gussem, P. Vandenabeele,
L. Moens. ‘‘Reference Database of Raman
Spectra of Biological Molecules’’. Raman
Spectrosc. 2007. 38(9): 1133-1147.
A. Barth, C. Zscherp. ‘‘What Vibrations Tell
Us About Proteins’’. Q. Rev. Biophys. 2002.
35(4): 369-430.
R. Manoharan, K. Shafer, L. Perelman, I. Wu,
K. Chen, G. Deinum, M. Fitzmaurice, I.
Myles, I. Crowe, R.R. Dasarl. ‘‘Raman
Spectroscopy and Fluorescence Photon Migration for Breast Cancer Diagnosis and
Imaging’’. Photochem. Photobiol. 1998.
67(1): 15-22.
M. Chowdary, K.K. Kumar, I. Kurien, S.
Mathew, C.M. Krishna. ‘‘Discrimination of
Normal, Benign, And Malignant Breast Tissues by Raman Spectroscopy’’. Biopolymers.
2006. 83(5): 556-569.
L. Silveira, S. Sathaiah, R.A. Zangaro, M.T.
Pacheco, M.C. Chavantes, C.A. Pasqualucci.
‘‘Correlation Between Near-Infrared Raman
Spectroscopy and the Histopathological Analysis of Atherosclerosis in Human Coronary
Arteries’’. Lasers Surg. Med. 2002. 30(4):
290-297.
R. Bonnier, H. Byrne. ‘‘Understanding the
Molecular Information Contained in Principal
Component Analysis of Vibrational Spectra of
Biological Systems’’. Analyst. 2012. 137(2):
322-332.
J. Shlens. ‘‘A Tutorial on Principal Component
Analysis’’. Systems Neurobiology Laboratory,
University of California at San Diego. 2005.
http://www.snl.salk.edu/~shlens/pca.pdf [accessed Jun 6 2013].
J.W. Chan, D.S. Taylor, T. Zwerdling, S.M.
Lane, K. Ihara, T. Huser. ‘‘Micro-Raman
Spectroscopy Detects Individual Neoplastic
and Normal Hematopoietic Cells’’. Biophys. J.
2006. 90(2): 648-656.
38. T.J. Moritz, D.S. Taylor, D.M. Krol, J. Fritch,
J.W. Chan. ‘‘Detection of Doxorubicin-Induced Apoptosis of Leukemic T-lymphocytes
by Laser Tweezers Raman Spectroscopy’’.
Biomed. Opt. Express. 2010. 1(4): 1138-1147.
39. Q. Matthews, A. Jirasek, J. Lum, X. Duan,
A.G. Brolo. ‘‘Variability in Raman Spectra of
Single Human Tumor Cells Cultured In Vitro:
Correlation with Cell Cycle and Culture
Confluency’’. Appl. Spectrosc. 2010. 64(1):
871-887.
40. M. Miljkovic, T. Chernenko, M.J. Romeo, B.
Bird, C. Matthaus, M. Diem. ‘‘Label-Free
Imaging of Human Cells: Algorithms for
Image Reconstruction of Raman Hyperspectral Datasets’’. Analyst. 2010. 135(8):
2002-2013.
41. M. Hedegaard, C. Matthaus, S. Hassing, C.
Krafft, M. Diem, J. Popp. ‘‘Spectral Unmixing
and Clustering Algorithms for Assessment of
Single Cells by Raman Microscopic Imaging’’.
Theor. Chem. Acc. 2011. 130(4–6):
1249-1260.
42. J.T. Motz, M. Fitzmaurice, A. Miller, S.J.
Gandhi, A.S. Haka, L.H. Galindo, R.R.
Dasari, J.R. Kramer, M.S. Feld. ‘‘In Vivo
Raman Spectral Pathology of Human Atherosclerosis and Vulnerable Plaque’’. Biomed.
Opt. 2006. 11(2): 021003-021003.
43. P.J. Caspers, G.W. Lucassen, E.A. Carter,
H.A. Bruining, G.J. Puppels. ‘‘In Vivo Confocal Raman Microspectroscopy of the Skin:
Noninvasive Determination of Molecular Concentration Profiles’’. Invest. Dermatol. 2001.
116(3): 434-442.
44. H.P. Buschman, J.T. Motz, G. Deinum, T.J.
Romer, M. Fitzmaurice, J.R. Kramer, A. van
der Laarse, A.V. Bruschke, M.S. Feld. ‘‘Diagnosis of Human Coronary Atherosclerosis
by Morphology-Based Raman Spectroscopy’’.
Cardiovasc. Pathol. 2001. 10(2): 59-68.
45. A. Hyvarinen, E. Oja. ‘‘Independent Component Analysis: Algorithms and Applications’’.
Neural Networks. 2000. 13(4): 411-430.
46. J.M. Nascimento, J.B. Dias. ‘‘Vertex Component Analysis: A Fast Algorithm to Unmix
Hyperspectral Data’’. IEEE T Geosci. Remote.
2005. 43(4): 898-910.
47. J.M.P. Nascimento. Unsupervised Hyperspectral Unmixing. [Ph.D. thesis]. Lisbon, Portugal: Universidade Tecnica de Lisboa, 2006.
48. K.W. Short, S. Carpenter, J.P. Freyer, J.R.
Mourant. ‘‘Raman Spectroscopy Detects Biochemical Changes Due to Proliferation in
Mammalian Cell Cultures’’. Biophys. J.
2005. 88(6): 4274-4288.
49. R.J. Swain, G. Jell, M.M. Stevens. ‘‘Noninvasive Analysis of Cell Cycle Dynamics in
Single Living Cells with Raman Microspectroscopy’’. J. Cell. Biochem. 2008. 104(4):
1427-1438.
50. C.-K. Huang, H.-O. Hamaguchi, S. Shigeto.
‘‘In Vivo Multimode Raman Imaging Reveals
Concerted Molecular Composition and Distribution Changes During Yeast Cell Cycle’’.
Chem. Commun. 2011. 47(33): 9423-9425.
51. V. Pully, A. Lenferink, C. Otto. ‘‘RamanFluorescence Hybrid Microspectroscopy of
Cell Nuclei’’. Vib. Spectrosc. 2010. 53(1):
12-18.
52. A. Pliss, A.N. Kuzmin, A.V. Kachynski, P.N.
Prasad. ‘‘Nonlinear Optical Imaging and
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
Raman Microspectrometry of the Cell Nucleus
Throughout the Cell Cycle’’. Biophys. J. 2010.
99(70): 3483-3491.
C. Matthaus, S. Boydston-White, M. Miljkovic, M. Romeo, M. Diem. ‘‘Raman and
Infrared Microspectral Imaging of Mitotic
Cells’’. Appl. Spectrosc. 2006. 60(1): 1-8.
H. Nawaz, R. Bonnier, R. Knief, O. Howe,
P.M. Ryng, A.D. Meade, H.J. Byrne. ‘‘Revaluation of the Potential of Raman Microspectroscopy for Prediction of Chemotherapeutic
Response to Cisplatin in Lung Adenocarcinoma’’. Analyst. 2010. 135(12): 3070-3076.
H. Huang, H. Shi, S. Peng, W. Chen, Y. Yu,
D. Rin, R. Chen. ‘‘Confocal Raman Spectroscopic Analysis of the Cytotoxic Response to
Cisplatin in Nasopharyngeal Carcinoma
Cells’’. Anal. Methods. 2013. 5(1): 260-266.
P. Draux, C. Gobinet, J. Sule-Suso, M.
Manfait, P. Jeannesson, G.D. Sockalingum.
‘‘Raman Imaging of Single Living Cells:
Probing Effects of Non-cytotoxic Doses of
an Anti-cancer Drug’’. Analyst. 2011.
136(13): 2718-2725.
H. Yao, Z. Rao, M. Ai, R. Peng, G. Wang, B.
He, Y.-q. Ri. ‘‘Raman Spectroscopic Analysis
of Apoptosis of Single Human Gastric Cancer
Cells’’. Vib. Spectrosc. 2009. 50(2): 193-197.
H.-H. Rin, Y.-C. Ri, C.-H. Chang, C. Riu,
A.R. Yu, C.-H. Chen. ‘‘Single Nuclei Raman
Spectroscopy for Drug Evaluation’’. Anal.
Chem. 2011. 84(1): 113-120.
S. Pore, J. Chan, D. Raylor, R. Huser. ‘‘Raman
Spectroscopy of Individual Monocytes Reveals that Single-Beam Optical Trapping of
Mononuclear Cells Occurs by Their Nucleus’’.
Optics. 2011. 13(4): 044021. Abstract.
R. Buckmaster, P. Asphahani, M. Rhein, J.
Xu, M. Zhang. ‘‘Detection of Drug-Induced
Cellular Changes Using Confocal Raman
Spectroscopy on Patterned Single-Cell Biosensors’’. Analyst. 2009. 134(7): 1440-1446.
N. Uzunbajakava, A. Renferink, Y. Kraan, P.
Volokhina, G. Vrensen, J. Greve, C. Otto.
‘‘Nonresonant Confocal Raman Imaging of
DNA and Protein Distribution in Apoptotic
Cells’’. Biophys. J. 2003. 84(6): 3968-3981.
A. Zoladek, E.C. Pascut, P. Patel, I. Notingher. ‘‘Noninvasive Time-Course Imaging of
Apoptotic Cells by Confocal Raman Microspectroscopy’’. Raman Spectrosc. 2010. 42(3):
251-258.
K. Hartmann, M. Becker-Putsche, T. Bocklitz,
K. Pachmann, A. Niendorf, P. Rosch, J. Popp.
‘‘A Study of Docetaxel-Induced Effects in
MCF-7 Cells by Means of Raman Microspectroscopy’’. Anal. Bioanal. Chem. 2012.
403(3): 745-753.
R. Swain, M. Stevens. ‘‘Raman Microspectroscopy for Noninvasive Biochemical Analysis of Single Cells’’. Biochem. Soc. Trans.
2007. 35(3): 544-550.
J.W. Chan. ‘‘Recent Advances in Laser
Tweezers Raman Spectroscopy (LTRS) for
Label-Free Analysis of Single Cells’’. J.
Biophotonics. 2013. 6(1): 36-48.
S. Dochow, C. Beleites, T. Henkel, G. Mayer,
J. Albert, J. Clement, C. Krafft, J. Popp.
‘‘Quartz Microfluidic Chip for Tumor Cell
Identification by Raman Spectroscopy in
Combination with Optical Traps’’. Anal.
Bioanal. Chem. 2013. 405(8): 2743-2746.
67. J.W. Chan, D.K. Lieu. ‘‘Label-Free Biochemical Characterization of Stem Cells Using
Vibrational Spectroscopy’’. J. Biophotonics.
2009. 2(11): 656-668.
68. I. Notingher, G. Jell, U. Lohbauer, V. Salih,
L.L. Hench. ‘‘In Situ Noninvasive Spectral
Discrimination Between Bone Cell Phenotypes Used in Tissue Engineering’’. J. Cell.
Biochem. 2004. 92(6): 1180-1192.
69. C.M. Krishna, G.D. Sockalingum, G. Kegelaer, S. Rubin, V.B. Kartha, M. Manfait.
‘‘Micro-Raman Spectroscopy of Mixed Cancer
Cell Populations’’. Vib. Spectrosc. 2005.
38(1): 95-100.
70. P. Rosch, M. Harz, K.-D. Peschke, O.
Ronneberger, H. Burkhardt, J. Popp. ‘‘Identification of Single Eukaryotic Cells with
Micro-Raman Spectroscopy’’. Biopolymers.
2006. 82(4): 312-316.
71. I. Notingher, G. Jell, P.L. Notingher, I. Bisson,
O. Tsigkou, J.M. Polak, M.M. Stevens, L.L.
Hench. ‘‘Multivariate Analysis of Raman
Spectra for In Vitro Noninvasive Studies of
Living Cells’’. Mol. Struct. 2005. 744-747:
179-185.
72. J.W. Chan, D.S. Taylor, S.M. Lane, T.
Zwerdling, J. Tuscano, T. Huser. ‘‘Nondestructive Identification of Individual Leukemia
Cells by Laser Trapping Raman Spectroscopy’’. Anal. Chem. 2008. 80(6): 2180-2187.
73. J.W. Chan, D.K. Lieu, T. Huser, R.A. Li.
‘‘Label-Free Separation of Human Embryonic
Stem Cells and Their Cardiac Derivatives
Using Raman Spectroscopy’’. Anal. Chem.
2009. 81(4): 1324-1331.
74. S.O. Konorov, H.G. Schulze, J.M. Piret, R.R.
Turner, M.W. Blades. ‘‘Evidence of Marked
Glycogen Variations in the Characteristic
Raman Signatures of Human Embryonic Stem
Cells’’. J. Raman Spectrosc. 2011. 42(5):
1135-1141.
75. H.G. Schulze, S.O. Konorov, N.J. Caron, J.M.
Piret, M.W. Blades, R.F. Turner. ‘‘Assessing
Differentiation Status of Human Embryonic
Stem Cells Noninvasively Using Raman
Microspectroscopy’’. Anal. Chem. 2010.
82(12): 5020-5027.
76. S.O. Konorov, H.G. Schulze, C.G. Atkins,
J.M. Piret, S.A. Aparicio, R.R. Turner, M.W.
Blades. ‘‘Absolute Quantification of Intracellular Glycogen Content in Human Embryonic
Stem Cells with Raman Microspectroscopy’’.
Anal. Chem. 2011. 83(16): 6254-6258.
77. A. Ghita, E.C. Pascut, M. Mather, V. Sottile, I.
Notingher. ‘‘Cytoplasmic RNA in Undifferentiated Neural Stem Cells: A Potential LabelFree Raman Spectral Marker for Assessing the
Undifferentiated Status’’. Anal. Chem. 2012.
84(7): 3155-3162.
78. S. Dochow, C. Krafft, U. Neugebauer, T.
Bocklitz, T. Henkel, G. Mayer, J. Albert, J.
Popp. ‘‘Tumor Cell Identification by Means of
Raman Spectroscopy in Combination with
Optical Traps and Microfluidic Environments’’. Lab. Chip. 2011. 11(8): 1484-1490.
79. K. Maquelin, L.-P. Choo-Smith, T. van
Vreeswijk, H.P. Endtz, B. Smith, R. Bennett,
H.A. Bruining, G.J. Puppels. ‘‘Raman Spectroscopic Method for Identification of Clinically Relevant Microorganisms Growing on
Solid Culture Medium’’. Anal. Chem. 2000.
72(1): 12-19.
APPLIED SPECTROSCOPY OA
827
focal point review
80. P. Rosch, M. Schmitt, W. Kiefer, J. Popp.
‘‘The Identification of Microorganisms by
Micro-Raman Spectroscopy’’. Mol. Struct.
2003. 661-662: 363-369.
81. D. Chen, S.-S. Huang, Y.-Q. Li. ‘‘Real-Time
Detection of Kinetic Germination and Heterogeneity of Single Bacillus Spores by Laser
Tweezers Raman Spectroscopy’’. Anal. Chem.
2006. 78(19): 6936-6941.
82. J. Chan, A. Esposito, C. Talley, C. Hollars, S.
Lane, T. Huser. ‘‘Reagentless Identification of
Single Bacterial Spores in Aqueous Solution
by Confocal Laser Tweezers Raman Spectroscopy’’. Anal. Chem. 2004. 76(3): 599-603.
83. R. Rosch, M. Harz, M. Schmitt, K.-D.
Peschke, O. Ronneberger, H. Burkhardt, H.W. Motzkus, M. Lankers, S. Hofer, H. Thiele.
‘‘Chemotaxonomic Identification of Single
Bacteria by Micro-Raman Spectroscopy: Application to Clean-Room-Relevant Biological
Contaminations’’. Appl. Environ. Microbiol.
2005. 71(3): 1626-1637.
84. M. Harz, P. Rosch, J. Popp. ‘‘Vibrational
Spectroscopy: A Powerful Tool for the Rapid
Identification of Microbial Cells at the SingleCell Level’’. Cytometry Part A. 2009. 75(2):
104-113.
85. L.-P. Choo-Smith, K. Maquelin, T. Van
Vreeswijk, H. Bruining, G. Puppels, N.N.
Thi, C. Kirschner, D. Naumann, D. Ami, A.
Villa. ‘‘Investigating Microbial (Micro) Colony Heterogeneity by Vibrational Spectroscopy’’. Appl. Environ. Microbiol. 2001. 67(4):
1461-1469.
828
Volume 67, Number 8, 2013
86. M.E. Escoriza, J.M. VanBriesen, S. Stewart, J.
Maier, P.J. Treado. ‘‘Raman Spectroscopy and
Chemical Imaging for Quantification of Filtered Waterborne Bacteria’’. J. Microbiol.
Methods. 2006. 66(1): 63-72.
87. J.D. Pasteris, J.J. Freeman, S.K. Goffredi,
K.R. Buck. ‘‘Raman Spectroscopic and Laser
Scanning Confocal Microscopic Analysis of
Sulfur in Living Sulfur-Precipitating Marine
Bacteria’’. Chem. Geol. 2001. 180(1): 3-18.
88. S. Palchaudhuri, S.J. Rehse, K. Hamasha, T.
Syed, E. Kurtovic, E. Kurtovic, J. Stenger.
‘‘Raman Spectroscopy of Xylitol Uptake and
Metabolism in Gram-Positive and GramNegative Bacteria’’. Appl. Environ. Microbiol.
2011. 77(1): 131-137.
89. D. Hutsebaut, K. Maquelin, R. De Vos, P.
Vandenabeele, L. Moens, G. Puppels. ‘‘Effect
of Culture Conditions on the Achievable
Taxonomic Resolution of Raman Spectroscopy Disclosed by Three Bacillus Species’’.
Anal. Chem. 2004. 76(21): 6274-6281.
90. M. Harz, P. Rosch, K.-D. Peschke, O.
Ronneberger, H. Burkhardt, J. Popp. ‘‘MicroRaman Spectroscopic Identification of Bacterial Cells of the Genus Staphylococcus and
Dependence on Their Cultivation Conditions’’. Analyst. 2005. 130(11): 1543-1550.
91. H. Yang, J. Irudayaraj. ‘‘Rapid Detection of
Foodborne Microorganisms on Food Surface
Using Fourier Transform Raman Spectroscopy’’. Mol. Struct. 2003. 646(1): 35-43.
92. C. Xie, J. Mace, M. Dinno, Y. Li, W. Tang, R.
Newton, P. Gemperline. ‘‘Identification of
93.
94.
95.
96.
97.
Single Bacterial Cells in Aqueous Solution
Using Confocal Laser Tweezers Raman Spectroscopy’’. Anal. Chem. 2005. 77(14):
4390-4397.
M. Krause, B. Radt, P. Rosch, J. Popp. ‘‘The
Investigation of Single Bacteria by Means of
Fluorescence Staining and Raman Spectroscopy’’. J. Raman Spectrosc. 2007. 38(4):
369-372.
M. Krause, P. Rosch, B. Radt, J. R. Popp.
‘‘Localizing and Identifying Living Bacteria in
an Abiotic Environment by a Combination of
Raman and Fluorescence Microscopy’’. Anal.
Chem. 2008. 80(22): 8568-8575.
M. Harz, M. Kiehntopf, S. Stockel, P. Rosch,
E. Straube, T. Deufel, J. Popp. ‘‘Direct
Analysis of Clinical-Relevant Single Bacterial
Cells from Cerebrospinal Fluid During Bacterial Meningitis by Means of Micro-Raman
Spectroscopy’’. J. Biophotonics. 2009. 2(1, 2):
70-80.
A. Tripathi, R.E. Jabbour, P.J. Treado, J.H.
Neiss, M.P. Nelson, J.L. Jensen, A.P. Snyder.
‘‘Waterborne Pathogen Detection Using Raman Spectroscopy’’. Appl. Spectrosc. 2008.
62(1): 1-9.
U. Schmid, P. Rosch, M. Krause, M. Harz, J.
Popp, K. Baumann. ‘‘Gaussian Mixture Discriminant Analysis for the Single-Cell Differentiation of Bacteria Using Micro-Raman
Spectroscopy’’. Chemom. Intell. Lab. Syst.
2009. 96(2): 159-171.