Dynamic Article LinksC<Analyst

Cite this: Analyst, 2011, 136, 1620

www.rsc.org/analyst PAPER

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7DView Article Online / Journal Homepage / Table of Contents for this issue

Detection of acute brain injury by Raman spectral signature

Li-Lin Tay,*a Roger G. Tremblay,b John Hulse,a Bogdan Zurakowski,b Michael Thompsond andMahmud Bani-Yaghoub*bc

Received 11th November 2010, Accepted 8th February 2011

DOI: 10.1039/c0an00897d

Brain injury can lead to irreversible tissue loss and functional deficit along with significant health care

costs. Raman spectroscopy can be used as a non-invasive technique to provide detailed information on

the molecular composition of diseased and damaged tissues. This technique was used to examine acute

mouse brain injury, focusing on the motor cortex, a region directly involved in controlling execution of

movement. The spectral profile obtained from the injured brain tissue revealed a markedly different

signature, particularly in the amide I and amide III vibrational region when compared to that of healthy

brain tissue. Most noticeably, there was a significant reduction of the amide I vibration at the acute

injury site and the appearance of two distinct features at 1586 and 1618 cm�1. Complementary

immunohistochemical analysis of the injured brain tissue showed an abundant expression of Caspase 3

(a cysteine protease marker used for apoptosis), suggesting that the injury-induced specific Raman

shifts may be correlated with cell death. Taken together, this study demonstrates that Raman

spectroscopy can play an important role in detecting the changes that occur in the injured brain and

provide a possible technology for monitoring the recovery process.

Introduction

Recent years have seen significant advances in chemical detection

science, especially in medicine and the field of clinical diagnos-

tics.1 Techniques involved range from methods for the early

detection of disease to devices designed for point of care, rapid

analyses of biomarkers. A particularly important challenge is

posed by the assessment of brain damage. Injury to this organ is

among the primary causes of death and disability, and is a direct

consequence of irreversible neuronal loss, functional impairment

and limited repair.2,3 The management of brain injury requires

a comprehensive approach in which efficient diagnosis in

combination with an appropriate care and rehabilitation

program may lead to improved recovery. Currently, diagnosis

relies on various imaging techniques in order to complement the

clinical and physiological information acquired at the bedside.

Magnetic resonance imaging (MRI), computerized axial

tomography (CAT) and scanning by positron emission tomog-

raphy (PET) provide significant information concerning changes

aInstitute for Microstructural Sciences, National Research CouncilCanada, Ottawa, Ontario, Canada K1A 0R6. E-mail: lilin.tay@nrc.cabNeurogenesis and Brain Repair, Neurobiology Program, Institute forBiological Sciences, National Research Council Canada, 1200 MontrealRd., Bldg. M-54, Ottawa, Ontario, Canada K1A 0R6. E-mail: mahmud.bani@nrc.cacDepartment of Cellular and Molecular Medicine, Faculty of Medicine,University of Ottawa, Ottawa, Ontario, Canada K1H 8M5dDepartment of Chemistry, University of Toronto, 80 St. George Street,Toronto, Ontario, Canada M5S 3H6

1620 | Analyst, 2011, 136, 1620–1626

in brain structure, physiology and metabolic activity caused by

brain damage.4,5 However, the high cost and strict regulation

associated with these systems result in the fact that most hospi-

tals and clinics around the world do not have access to these

technologies. Accordingly, the design and implementation of

a rapid, sensitive, non-invasive and less expensive technique

represent an attractive alternative for diagnostic application.

Moreover, such an advance, even in modern facilities, could lead

to reduced waiting periods and enhancement of the ability of the

physician to reach an early and accurate diagnosis.

To complement the clinically established imaging systems

mentioned above, recent technological advances in optical

techniques provide a promising strategy to obtain valuable

information concerning the brain structure and composition. In

particular, vibrational spectroscopic techniques, such as Fourier

transform infrared absorption spectroscopy (FTIR) and Raman

spectroscopy, are well established methods capable of probing

the distribution and location of biomolecules (proteins, lipids,

nucleic acids) present in the brain and other tissues.6–8 These

techniques provide researchers with a means to study differences

in structure, morphology and chemical composition in tissues

and cells without the need of labels at a sub-cellular spatial

resolution, while being minimally invasive to the cells and tissues

probed. It has been shown that both methods are capable of

discerning healthy from diseased tissues based on biochemical

differences and have been applied to tissue samples from breast,

colon, cervix, skin, liver, lung, brain and from bodily fluids.9–17 In

addition, Raman spectroscopy and FTIR have made contribu-

tions to the field of stem cells by providing a non-invasive

This journal is ª The Royal Society of Chemistry 2011

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7D

View Article Online

approach that complements the label-based histological methods

to determine different cell types.18–21 This is not surprising, since

most samples with displaying disease show fundamental changes

in the biochemistry of the tissue. With respect to the two vibra-

tional techniques, Raman spectroscopy has the added advantage

of higher spatial resolution and insensitivity towards water or

biological buffers making it a particularly suitable method

for probing biological tissues. Raman spectroscopy using mini-

aturized fiber-optic probes could provide clinicians with a

non-invasive means to locate or screen for diseased tissue during

surgical procedures and to monitor post-treatment recovery.

Several reports have been published on the use of the technique

in the study of healthy brain tissues, Alzheimer’s diseased tissue,

and brain tumors.7,17,18,22–29 However, there have been no reports

of the analysis of brain injury by vibrational spectroscopy. In this

paper we present the first Raman spectroscopic analysis of the

brain, using mouse motor cortex injury as a model system. The

spectral fingerprint obtained from injured regions of the brain

has been further complemented by neurobiological measure-

ments in an attempt to provide a correlation between Raman

spectral data and changes in cellular apoptosis observed after

injury.

Experimental procedures

Materials and methods

Tissue preparation. Animal studies were approved by the

Animal Care Committee at the National Research Council

Canada-Institute for Biological Sciences (accredited by the

Canadian Council on Animal Care). Briefly, a total of 16 three-

month-old CD1 mice (Charles River Labs, St Constant, QC)

were anesthetized, using isoflurane (Aerrane, Baxter, Montreal,

QC). The animals were placed into equal ‘‘control’’ (un-injured)

and ‘‘injured’’ groups. To perform an injury, animals were placed

in a stereotaxic frame and the skull was exposed. The injury site

was marked on the bone, using specific coordinates (from ‘‘AP

�0.25 mm to�1.0 mm, Lat +0.7 mm’’, to ‘‘AP +1.25 mm to +3.0

mm, Lat +2.4 mm’’) with respect to Bregma. The bone was

removed with a dental drill, and the motor cortex was injured by

using a sterile needle to remove the neural tissue at a depth of 1

mm. Animals were sacrificed right after injury (along with their

control counterparts), and the whole brain was removed, washed

and maintained in Hank’s Balanced Salt Solution (HBSS, Invi-

trogen, Mississauga, ON), and immediately examined by Raman

spectroscopy. In parallel experiments, horizontal sections were

prepared and deposited on a microscope slide (8 mm thick

sections for immunohistochemistry) according to previously

established procedure.30

Immunohistochemistry. Mouse brains were fixed with 10%

formalin for 16 hours, processed (Thermo Shandon, Citadel

1000, Pacific Southwest Lab Equipment Inc., Vista, CA, USA)

and paraffin-embedded (Thermo Shandon Histocentre 2, Pacific

Southwest Lab Equipment Inc.). Each brain sample was

sectioned into 8 mm slices (Leica RM2165 Microtome, Ger-

many), after which the sections were de-waxed with xylene (5

minutes in each of the three sequential baths), hydrated in

ethanol (100%, 90%, 70%, and 50% for 5 minutes each), and

This journal is ª The Royal Society of Chemistry 2011

washed with phosphate buffered saline (PBS, three times for 5

minutes).30 Sections were incubated with polyclonal Caspase 3

antibody (Santa Cruz, CA, USA, diluted 1 : 50 in 1% BSA) and

monoclonal microtubule associate protein 2a + 2b (MAP2, clone

AP-20, 1 : 200, Sigma, Toronto, ON, Canada) or glial fibrillary

acidic protein (GFAP, clone GA-5, 1:100, Neomarkers, Fre-

mont, CA, USA) for 1 hour at room temperature, washed with

PBS (three times, 10 minutes each) and labeled with Alexa-

conjugated anti-rabbit IgG (Invitrogen, Mississauga, ON, Can-

ada, diluted 1 : 600 in 1% BSA) and rhodamine-conjugated anti-

mouse IgG (1 : 600; Jackson Immunoresearch Laboratories) for

1 hour. Sections were washed again with PBS (three times, 10

minutes each), stained with Hoechst (Invitrogen, Eugene, Ore-

gon, USA, 1 mg ml�1) for 10 minutes, rinsed with water and

covered with Vectashield mounting medium (Vector Laborato-

ries, Burlingame, CA). The immunoreactivity was examined with

an Axiovert 200M fluorescence microscope, AxioCam and Axi-

ovision 4.7.2 (Zeiss, Germany). The images were processed with

Adobe PhotoShop (Adobe Systems Incorporated, Mountain

View, CA) and CorelDraw (Corel Corporation, Ottawa, ON).

Raman spectroscopy. Raman spectroscopy was performed

with a commercial microRaman system (LabRAM HR, Horiba

Jobin Yvon, Edison, NJ) equipped with a software controlled

XY stage and a thermal-electric cooled CCD detector. Samples

(whole mouse brain) were excited with 785 nm single mode diode

laser (XTRA, Toptica Photonics Inc., Victor, NY) at a power

density of �104 W cm�2 without having problems such as auto-

fluoresence.29 Incident radiation was coupled into an Olympus

BX51 optical microscope and focused to �4 mm diameter spot

through a Leica 50� long working distance objective. Whole

brain samples were placed on a microscope slide and irradiated

through the microscope objective (NA: 0.55). Backscattered

Raman emission was collected by the same objective with an

8 second integration time and dispersed with a 300 groove per

mm grating. Rayleigh rejection was achieved with a notch filter.

For the un-fixed tissue samples, all Raman measurements were

started immediately following the removal of the brains from the

skull and completed within one hour of the surgery.

Results and discussions

Although Raman spectroscopy is a particularly well-established

technique, it is an emerging diagnostic tool when it comes to the

early detection of chemistry associated with diseased tissues. The

method displays considerable potential due to its sensitivity to

molecular composition, and to structural conformational

changes associated with pathological features that cannot be

detected easily by conventional morphological methods. Most

previous Raman studies of brain tissue have been performed

either on frozen or fixed tissue cross-sections. While frozen

tissues are commonly employed in vibrational studies, the

complicated handling process can significantly reduce the

experimental flexibility. On the other hand, the fixation process

needs to be carefully performed to facilitate the proper

interpretation of experimental results.31 Real-time chemical

information can be obtained from living cells by performing

in situ monitoring of samples.20,29 In the latter report, bleeding

Analyst, 2011, 136, 1620–1626 | 1621

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7D

View Article Online

artifacts were avoided partially by injecting cancer cells through

the carotid artery to form tumors in the mouse brain.

In our study the injury was mechanically introduced to the

motor cortex through a cranial window (and meninges),

a phenomenon observed in the patients with brain injury caused

by motor vehicle accidents, penetration of foreign bodies into the

neural tissue or tumor excision during surgical procedures.32 We

performed Raman spectroscopy on the healthy and injured

motor cortices immediately after the whole brain was removed

from the skull. Additionally, parallel immunohistology experi-

ments were also done on the fixed brain samples to enable critical

comparison with the Raman spectral features introduced by

injury. With prior expertise established in neuroregeneration,32

we have used a focal brain injury mouse model system that

Fig. 1 The images of control and injured adult mouse brains. The high

resolution digital photographs (A and D) and optical images (B and E)

show significant differences between the healthy (A–C) and injured (D–F)

brains. The injury was mechanically introduced to the left motor cortex

(D, circled). Corresponding immunohistochemical images reveal intact

neurons (arrowheads) in the healthy motor cortex (C). In contrast,

neurons are significantly affected by injury, as evidenced by morpho-

logical features and MAP2 immunoreactivity (F). Cb: cerebellum, Ncx:

neocortex, and OB: olfactory bulb. Scale bar: A and D (1.6 mm), B and

E (400 mm), C and F (70 mm).

1622 | Analyst, 2011, 136, 1620–1626

mimics human conditions33 and allows controlling the injury size

in a reproducible and quantifiable manner (Fig. 1). The injury

was introduced to the left motor cortex (Fig. 1D) and compared

to its counterpart in the right hemisphere (as internal control in

the same mouse) and the left motor cortex in the control mice

(Fig. 1A). Optical images of the healthy (Fig. 1B) and injured

(Fig. 1E) motor cortices as seen under the optical microscope

showed significant discolouration in the injured sites (Fig. 1E) as

a result of mechanical damage to the tissue. The regions exam-

ined spectroscopically were further evaluated by fluorescence

microscopy (Fig. 1C and F) to correlate the changes in the

injured area at the cellular level. Fig. 1C shows well-organized

MAP2 positive neuronal cell bodies and dendrites in the healthy

motor cortex, whereas, the injured motor cortex (Fig. 1F)

exhibited significant tissue loss as well as damage to the cell

bodies and neurite extensions in the vicinity of the injury core.

Raman spectroscopy of fresh and fixed brain samples

Fig. 2 presents a series of typical Raman spectra from eight intact

motor cortices. Spectra 1–4 were obtained from fixed tissue

samples, whereas, spectra 5–8 were acquired from fresh tissues.

All freshly prepared (un-fixed) samples were maintained in the

physiological buffer (HBSS) and applied to spectroscopic studies

immediately after the surgical procedure.

Both fresh and fixed samples exhibit a strong band at 1660

cm�1, that is attributed to the amide I vibration of the protein and

C]C stretching of lipids. Three broad features at 1274, 1301 and

1346 cm�1 are also obvious in both sets and are attributed to the

amide III vibration of protein and CH deformation of protein

Fig. 2 Raman spectroscopy of fresh and fixed healthy adult mouse

brains. Consistent peaks (1002, 1274, 1450 and 1660 cm�1) are present in

the spectra obtained from both fixed (1–4) and fresh (5–8) brain samples.

All spectra were obtained with 785 nm excitation radiation.

This journal is ª The Royal Society of Chemistry 2011

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7D

View Article Online

and lipid components in the probed tissue. The sharp and narrow

feature at 1002 cm�1 is associated with the ring breathing mode of

phenylalanine in the protein. The series of bands at 1033, 1090

and 1130 cm�1 are attributed to the C–N stretch of phenylala-

nine, the PO2� stretch of phospholipids and nucleic acid, and the

C–C stretch, respectively. The sharp and strong band centered at

1450 cm�1 is assigned to the CH2 deformation mode from both

lipids and proteins. In addition, in the un-fixed samples (spectra

5–8, Fig. 2), a small band at 1740 cm�1 due to the carbonyl

stretch of the ester group of the lipids is also observed. These

observations are in agreement with the earlier studies of mouse,

rat and human brains.22,23,26,27,29 A comparative analysis shows

several common spectral signatures (such as 1002, 1090, 1130,

1274, 1450 and 1660 cm�1) in the fresh and fixed tissues.

However, there are intriguing differences between the two sets of

data, likely due to fixation and its consequent delipidation.

Specifically, the C–N stretch of phenylalanine (1033 cm�1) and

the lipid carbonyl stretch at 1740 cm�1 are consistently different

between the fresh and fixed samples. Spectra obtained from the

fresh tissues generally show a 1033 cm�1 band, but upon fixation,

this band weakened. The fixed samples also consistently show

a decrease or disappearance of the 1740 cm�1 band compared to

the fresh tissues possibly due to delipidation during the fixation

process.24 Since the brain tissue, in particular white matter,

contains high (5–15%) lipid content2 and various structural

proteins, the adjunct Raman spectroscopy of fresh and fixed

samples provided an appreciation of the differences between the

two methods used to prepare brain samples. Such comparison is

essential with respect to an accurate discernment of the spectral

fingerprint of brain injury from changes caused by sample

preparation.

Spectral signatures of healthy and injured brain tissues

Through the comparative studies of fresh and fixed brain

samples, we were able to establish that the fixation process did

not impact the major vibrational signatures. Therefore, we used

both fresh and fixed brain tissues to examine the effect(s) of

Fig. 3 Comparative Raman spectroscopy of healthy and injured adult mo

cortices demonstrate consistent Raman shifts following injury.

This journal is ª The Royal Society of Chemistry 2011

injury by Raman spectroscopy. Two sets of control brain

samples were used to compare with the injured region: the

healthy brains obtained from the uninjured mice and the

contralateral (opposite) motor cortex from the injured mice.

The latter serves as the most appropriate control for organs such

as the brain, which contain two structurally symmetric hemi-

spheres. Fig. 3 shows marked differences in the Raman spectra

obtained from the control (Fig. 3A) and injured (Fig. 3B)

hemisphere of six independent mice. Most notably, the charac-

teristic amide I vibration at 1660 cm�1, one of the most prom-

inent features observed from all control brain samples, was much

weaker in the spectra obtained from the injured brain samples.

Instead, two sharp bands located at 1586 and 1618 cm�1 appear.

The observed differences appear to be weakly correlated with the

sharp phenylalanine modes at 1002 cm�1. In the spectra taken

from the injured site (Fig. 3B), the band at 1002 cm�1 was either

weak or absent, as compared to the strong feature observed in

the undamaged hemisphere (Fig. 3A). Furthermore, the peak

structure in the region of the amide III vibration changes

significantly and two additional bands at 1175 and 1227 cm�1 are

observed in spectra obtained from the injured site. These features

resemble the vibrational features that are characteristic of

cholesterol and phospholipids.34 There is evidence that addi-

tional bands (such as minor Raman bands at 750, 1002, 1212,

1546, 1605 and 1619 cm�1) can be produced due to the resonance

excitation of the haemoglobin, resulting in increased Raman

cross-sections of heme-associated vibrations.29 In our study, we

have taken extra care to ensure that robust and reproducible data

are obtained from multiple brain samples. A similar ex vivo

approach has been taken by others to establish Raman maps of

brain specimen. In particular, many of the spectral features

obtained from the brain in our study are similar to those

previously reported,7 further validating the use of ex vivo brain

samples for Raman spectroscopy. In addition, Raman spec-

troscopy of ex vivo brain samples has been sufficient to demon-

strate that human intracranial tumours such as gliomas have

more haemoglobin and lower lipid to protein rations, and

meningeomas contain more collagen than normal brain tissue.7

use brains. Raman spectra from the control (A) and injured (B) motor

Analyst, 2011, 136, 1620–1626 | 1623

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7D

View Article Online

One of the techniques that can play a major role in the

improvement of detection sensitivity is surface-enhanced Raman

scattering (SERS).35–37 It is well-known that living cells can

uptake and transport nanoparticles through an endosomal

pathway through which nanoparticle aggregates can produce

large spectroscopic enhancements used for analysis by SERS.38

In fact, multi-color SERS nanotags have been used to detect

tumour cells expressing epidermal growth factor receptor

(EGFR) or other cell surface receptors as well as identification of

pathogenic microorganisms.36,39–44 The SERS nanotags are

typically made of Raman reporter molecules, with gold nano-

particles encapsulated in a protective SiO2 or polymer shell,

functionalized with antibodies of interest for targeted molecular

imaging. In addition, others have used SERS to distinguish the

spectra between healthy and tumour samples,45 by sudden

freezing of tissue in liquid nitrogen followed by crashing and

mixing the sample with a concentrated silver colloidal suspen-

sion. More recently, SERS has been used to monitor the chemical

changes that occur during differentiation of human adipose

derived adult stem cells.20 However, despite its above-mentioned

applications, the delivery of SERS active nanostructures into the

injury site (without altering the local biochemistry) remains

a great challenge.46–48 SERS holds amazing promises to all forms

of Raman spectroscopy and it can certainly be integrated in this

study, if the presence of metallic nanostructures does not pose

changes such as photothermal reactions to the injury state of the

tissue.

Since the motor cortex is composed of six distinctive layers

(layers I to VI), it is possible that removal of the outermost

portion of the cortex (layers I and II) allows the detection of

molecular signature of lower layers (layers III and IV) by Raman

spectroscopy. However, if this is the case, cross-sectional studies

of brain tissues should reveal such differences. The second

possible explanation for the observed spectral difference in our

study is the change in biochemical composition of the probed site

induced by the injury itself. To evaluate such a possibility, we

carried out immunohistology in the tissue sections in order to

understand the observed spectral differences.

Fig. 4 Caspase 3 activity in the control and injured adult mouse brains.

(A) Histological analysis of healthy brains shows almost no Caspase

3 immunoreactivity in the control motor cortex. (B) A corresponding

phase contrast image confirms a normal cortical structure. (C) The

injured motor cortex reveals an abundant level of Caspase 3 protein

within the cell bodies. (D) The abnormal cell bodies (arrowheads) can be

also detected in the corresponding phase contrast image. (E) MAP2

immunostaining shows that most Caspase 3 positive cells (C) are neurons.

(F) Bar graph depicts the percentage of Caspase 3 positive cells in

neurons and astrocytes in comparison to the control samples. Scale

bar: 15 mm; p < 0.001, one-way ANOVA.

Immunohistology of the injured brain

Raman microspectroscopic mapping of apoptotic cells such as

lung fibroblasts and lung carcinoma epithelial cells has demon-

strated fragmentation of the nucleus, degradation of proteins,

disappearance of lipid bodies and reduction in the intensity of

nucleic acid bands.49,50 Furthermore, other reports have shown

that traumatic brain injury and stroke cause cell death in the

damaged tissue,51 however, the observed injury-induced Raman

spectral features (1586 and 1618 cm�1) have neither been repor-

ted nor assigned. Therefore, it is unclear what biomolecular

activities in the injured region would be responsible for this

observation. One possible hint was provided by the recent

Raman studies from sub-cellular components and other

immortal cell lines, suggesting that the new Raman bands

observed in these cases (1560 to 1640 cm�1) are closely associated

with the mitochondrial activity of cells.52–54 Interestingly, mito-

chondria generate most of the energy of the cell in the form of

adenosine triphosphate (ATP) and are involved in several

processes, such as signaling and cell death. Therefore, it is

1624 | Analyst, 2011, 136, 1620–1626

possible that the new Raman bands observed in this study may

be linked to mitochondrial activity and ultimately to cell death. It

is known that as a natural response to injury, glial cells occupy

the damaged region (gliosis) and form the glial scar. However,

since the formation of glial scars take at least several days, this

process is unlikely to be associated with the observed spectral

difference obtained immediately following the injury. It is well

known that cerebral ischemia triggers two general pathways of

apoptosis: the intrinsic pathway, originating from mitochondrial

release of cytochrome C, which is associated with stimulation of

Caspase 3; and the extrinsic pathway, originating from the

activation of cell surface death receptors, resulting in the stim-

ulation of Caspase 8.55 In particular, Caspase 3 has been shown

to be activated by mitochondria upon cellular apoptosis in the

brain and other tissues.56 Thus, a comprehensive immunohisto-

chemistry analysis was performed to determine the percentage of

Caspase 3 positive cells in the normal and injured tissues. The

Caspase 3 immunofluorescence image of the normal tissue shows

a low level of Caspase 3 protein (Fig. 4A), while the image

This journal is ª The Royal Society of Chemistry 2011

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7D

View Article Online

obtained from injured tissue reveals a significant number of cells

with high Caspase 3 levels (Fig. 4C). Furthermore, Caspase 3 was

used in combination with either a neuronal (MAP2) or astrocytic

(GFAP) marker to determine the identity of Caspase 3 positive

cells. Fig. 4E shows the corresponding MAP2 immunofluores-

cence image of the same injured region also stained with Caspase

3 (Fig. 4C). In addition, phase contrast optical images of the

injured tissue revealed abnormal cell bodies (Fig. 4D, arrow-

heads), which are absent in the normal tissue (Fig. 4B). Using

double staining with either Caspase 3 and MAP2 or Caspase 3

and GFAP, we were able to determine the percentage of the

neurons and astrocytes that were positive for Caspase 3 in

normal and injured tissues and summarize the percentage of

Caspase 3 positive cells. Fig. 4F shows that there is a negligible

percentage (less than 5%) of Caspase 3 positive neurons in the

normal tissue, whereas the majority (>90%) of neurons in the

damaged tissue express Caspase 3. The Caspase 3 expression

levels from the injured tissues suggest that the early apoptotic

events occur mainly in neurons, as most astrocytes in both

normal and damaged tissues are Caspase 3 negative (Fig. 4F).

Thus, neurons appear to be, at least initially, more susceptible to

injury-induced cell death than astrocytic glial cells at this stage.

These results are in agreement with an earlier report, which

demonstrated that multiple Caspases are activated after trau-

matic brain injury.51 Since cortical injuries may quickly lead to

alterations in cerebral energy metabolism including reduced

ATP, GTP and nicotinic enzymes,57 the Raman shifts observed

in this study are thought to correlate to changes in the energy

level following injury associated with the altered mitochondrial

activities in the localized tissues. While an upregulation of

cytochrome C and Caspase 3 may account for some of the

changes observed in the spectra observed after injury, it remains

to be elucidated whether mitochondria are the first target in this

phenomenon or whether they are only affected indirectly. An

alternative approach to resolve this issue would be to examine

other cellular organelles along with mitochondria by electron

microscopy.

Conclusions

In this study, distinct reproducible features observed in the

acquired Raman spectra were compared between control and

injured brains. The Raman spectroscopic signature obtained

from the injured brain consistently reveals significant spectral

differences in the amide I vibrational region with a reduction of

the intensity of amide I band concomitant with the appearance

of additional features at 1587 and 1618 cm�1. The observed

differences show correlations with the sharp phenylalanine

modes at 1002 cm�1, which was either weak or absent from the

spectra obtained from the injury site. We employed Caspase

3 immunohistochemistry to establish a possible link between the

observed Raman signatures from the acute injury tissues and

apoptosis. This study provides the first observation of such

significant difference in the spectroscopic signature from the

acute brain injury and serves as a baseline for future studies on

acute and chronic injuries of the central nervous system. There is

no doubt that our study can further benefit from methods such as

principal component analysis (PCA) and linear discriminant

analysis (LDA) for spectral comparisons.16,18,19,58 Combined

This journal is ª The Royal Society of Chemistry 2011

PCA-LDA has been used by different laboratories to distinguish

cell types such as stem cells, transit amplifying cells and

terminally differentiated cells in the corneal epithelium,18 stem

cell region of human intestinal crypts,19 and distinct zones in

normal and tumour prostate samples.16,58 In our future studies,

we will take advantage of PCA-LDA to further analyze the

biochemical differences associated with injury and to evaluate

the cell types present in the injured region during acute and

chronic phases.

References

1 C. E. Mountford, P. Stanwell, A. Lin, S. Ramadan and B. Ross,Chem. Rev., 2010, 110, 3060–3086.

2 W. B. Deng, Nat. Rev. Neurol., 2010, 6, 328–336.3 Y. Xiong, A. Mahmood and M. Chopp, Curr. Opin. Invest. Drugs,

2010, 11, 298–308.4 J. P. Coles, Br. J. Anaesth., 2007, 99, 49.5 C. N. Gallagher, P. J. Hutchinson and J. D. Pickard, Curr. Opin.

Neurol., 2007, 20, 403–409.6 F. L. Martin and N. J. Fullwood, Proc. Natl. Acad. Sci. U. S. A., 2007,

104, E1.7 C. Krafft, L. Neudert, T. Simat and R. Salzer, Spectrochim. Acta,

Part A, 2005, 61, 1529–1535.8 V.-V. Pully and C. Otto, J. Raman Spectrosc., 2009, 40, 473–475.9 J. T. Motz, M. Fitzmaurice, A. Miller, S. J. Gandhi, A. S. Haka,

L. H. Galindo, R. R. Dasari, J. R. Kramer and M. S. Feld, J.Biomed. Opt., 2006, 11, 021003.

10 U. Neugebauer, T. Bocklitz, J. H. Clement, C. Krafft and J. Popp,Analyst, 2010, 135, 3178–3182.

11 A. B. Zoladek, R. K. Johal, S. Garcia-Nieto, F. Pascut,K. M. Shakesheff, A. M. Ghaemmaghami and I. Notingher,Analyst, 2010, 135, 3205–3212.

12 F. Bonnier, P. Knief, B. Lim, A. D. Meade, J. Dorney,K. Bhattacharya, F. M. Lyng and H. J. Byrne, Analyst, 2010, 135,3169–3177.

13 A. Bonifacio, C. Beleites, F. Vittur, E. Marsich, S. Semeraro,S. Paoletti and V. Sergo, Analyst, 2010, 135, 3193–3204.

14 M. S. Bergholt, W. Zheng, K. Lin, K. Y. Ho, M. Teh, K. G. Yeoh,J. B. Y. So and Z. W. Huang, Analyst, 2010, 135, 3162–3168.

15 L. Hartsuiker, N. J. L. Zeijen, L. Terstappen and C. Otto, Analyst,2010, 135, 3220–3226.

16 Patel, II and F. L. Martin, Analyst, 2010, 135, 3060–3069.17 C. Kendall, J. Day, J. Hutchings, B. Smith, N. Shepherd, H. Barr and

N. Stone, Analyst, 2010, 135, 3038–3041.18 O. Grude, A. Hammiche, H. Pollock, A. J. Bentley, M. J. Walsh,

F. L. Martin and N. J. Fullwood, J. Microsc. (Oxford, U. K.),2007, 228, 366–372.

19 M. J. Walsh, T. G. Fellous, A. Hammiche, W. R. Lin, N. J. Fullwood,O. Grude, F. Bahrami, J. M. Nicholson, M. Cotte, J. Susini,H. M. Pollock, M. Brittan, P. L. Martin-Hirsch, M. R. Alison andF. L. Martin, Stem Cells, 2008, 26, 108–118.

20 B. Moody, C. M. Haslauer, E. Kirk, A. Kannan, E. G. Loboa andG. S. McCarty, Appl. Spectrosc., 2010, 64, 1227–1233.

21 J. K. Pijanka, D. Kumar, T. Dale, I. Yousef, G. Parkes,V. Untereiner, Y. Yang, P. Dumas, D. Collins, M. Manfait,G. D. Sockalingum, N. R. Forsyth and J. Sule-Suso, Analyst, 2010,135, 3126–3132.

22 A. Beljebbar, S. Dukic, N. Amharref and M. Manfait, Anal. Bioanal.Chem., 2010, 398, 477–487.

23 M. Kohler, S. Machill, R. Salzer and C. Krafft, Anal. Bioanal. Chem.,2009, 393, 1513–1520.

24 P. Chen, A. G. Shen, W. Zhao, S. J. Baek, H. Yuan and J. M. Hu,Appl. Opt., 2009, 48, 4743–4748.

25 A. Beljebbar, N. Amharref, A. Leveques, S. Dukic, L. Venteo,L. Schneider, M. Pluot and M. Manfait, Anal. Chem., 2008, 80,8406–8415.

26 A. Mizuno, T. Hayashi, K. Tashibu, S. Maraishi, K. Kawauchi andY. Ozaki, Neurosci. Lett., 1992, 141, 47–52.

27 J. Sajid, A. Elhaddaoui and S. Turrell, J. Raman Spectrosc., 1997, 28,165–169.

Analyst, 2011, 136, 1620–1626 | 1625

Dow

nloa

ded

by U

nive

rsity

of

Min

neso

ta -

Tw

in C

ities

on

12/0

5/20

13 0

4:52

:11.

Pu

blis

hed

on 0

3 M

arch

201

1 on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

0AN

0089

7D

View Article Online

28 S. Koljenovic, T. B. Schut, A. Vincent, J. M. Kros and G. J. Puppels,Anal. Chem., 2005, 77, 7958–7965.

29 M. Kirsch, G. Schackert, R. Salzer and C. Krafft, Anal. Bioanal.Chem., 2010, 398, 1707–1713.

30 M. Bani-Yaghoub, R. G. Tremblay, J. X. Lei, D. L. Zhang,B. Zurakowski, J. K. Sandhu, B. Smith, M. Ribecco-Lutkiewicz,J. Kennedy, P. R. Walker and M. Sikorska, Dev. Biol., 2006, 295, 52–66.

31 E. O. Faolain, M. B. Hunter, J. M. Byrne, P. Kelehan,H. A. Lambkin, H. J. Byrne and F. M. Lyng, J. Histochem.Cytochem., 2005, 53, 121–129.

32 M. Bani-Yaghoub, R. G. Tremblay, A. Ajji, M. Nzau, S. Gangaraju,D. Chitty, B. Zurakowski and M. Sikorska, Biochem. Cell Biol., 2008,86, 361–369.

33 I. Cernak, Expert Opin. Invest. Drugs, 2006, 15, 1371–1381.34 N. Amharref, A. Beljebbar, S. Dukic, L. Venteo, L. Schneider,

M. Pluot and M. Manfait, Biochim. Biophys. Acta, 2007, 1768,2605–2615.

35 M. Moskovits, Rev. Mod. Phys., 1985, 57, 783.36 M. Moskovits, L. L. Tay, J. Yang and T. Haslett, inOptical Properties

of Nanostructured Random Media, 2002, vol. 82, pp. 215–226.37 S. M. Nie and S. R. Emery, Science, 1997, 275, 1102–1106.38 Q. Y. Hu, L. L. Tay, M. Noestheden and J. P. Pezacki, J. Am. Chem.

Soc., 2007, 129, 14–15.39 P. J. Huang, L. K. Chau, T. S. Yang, L. L. Tay and T. T. Lin, Adv.

Funct. Mater., 2009, 19, 242–248.40 P. J. Huang, L. L. Tay, J. Tanha, S. Ryan and L. K. Chau, Chem.–

Eur. J., 2009, 15, 9330–9334.41 L. L. Tay, J. Hulse, D. Kennedy and J. P. Pezacki, J. Phys. Chem. C,

2010, 114, 7356–7363.42 D. C. Kennedy, L. L. Tay, R. K. Lyn, Y. Rouleau, J. Hulse and

J. P. Pezacki, ACS Nano, 2009, 3, 2329–2339.

1626 | Analyst, 2011, 136, 1620–1626

43 D. C. Kennedy, K. A. Hoop, L. L. Tay and J. P. Pezacki, Nanoscale,2010, 2, 1413–1416.

44 D. C. Kennedy, D. R. Duguay, L. L. Tay, D. S. Richeson andJ. P. Pezacki, Chem. Commun., 2009, 6750–6752.

45 O. Aydin, M. Altas, M. Kahraman, O. F. Bayrak and M. Culha, Appl.Spectrosc., 2009, 63, 1095–1100.

46 B. D. Chithrani and W. C. W. Chan, Nano Lett., 2007, 7, 1542–1550.47 B. D. Chithrani, A. A. Ghazani and W. C. W. Chan, Nano Lett., 2006,

6, 662–668.48 R. S. Norman, J. W. Stone, A. Gole, C. J. Murphy and T. L. Sabo-

Attwood, Nano Lett., 2008, 8, 302–306.49 C. Krafft, K. Thummler, S. B. Sobottka, G. Schackert and R. Salzer,

Biopolymers, 2006, 82, 301–305.50 S. Verrier, I. Notingher, J. M. Polak and L. L. Hench, Biopolymers,

2004, 74, 157–162.51 S. M. Knoblach, M. Nikolaeva, X. L. Huang, L. Fan, S. Krajewski,

J. C. Reed and A. I. Faden, J. Neurotrauma, 2002, 19, 1155–1170.52 Y. S. Huang, T. Karashima, M. Yamamoto, T. Ogura and

H. Hamaguchi, J. Raman Spectrosc., 2004, 35, 525–526.53 Y. S. Huang, T. Nakatsuka and H. O. Hamaguchi, Appl. Spectrosc.,

2007, 61, 1290–1294.54 H. Tang, H. Yao, G. Wang, Y. Wang, Y. Q. Li and M. Feng, Opt.

Express, 2007, 15, 12708–12716.55 B. R. S. Broughton, D. C. Reutens and C. G. Sobey, Stroke, 2009, 40,

E331–E339.56 D. R. Green and J. C. Reed, Science, 1998, 281, 1309–1312.57 R. Vagnozzi, A. Marmarou, B. Tavazzi, S. Signoretti, D. Di Pierro,

F. Del Bolgia, A. M. Amorini, G. Fazzina, S. Sherkat andG. Lazzarino, J. Neurotrauma, 1999, 16, 903–913.

58 F. L. Martin, M. J. German, E. Wit, T. Fearn, N. Ragavan andH. M. Pollock, J. Comput. Biol., 2007, 14, 1176–1184.

This journal is ª The Royal Society of Chemistry 2011