Oral Oncology (2005) 41, 183–199

http://intl.elsevierhealth.com/journals/oron/

Proteome-wide analysis of head and neck

squamous cell carcinomas using laser-capture
microdissection and tandem mass spectrometry
Haven Bakera, , Vyomesh Patelb, , Alfredo A. Molinolob,
Edward J. Shillitoec, John F. Ensleyd, George H. Yooe,
Abelardo Meneses-Garcı́af, Jeffrey N. Myersg, Adel K. El-Naggarh,
J. Silvio Gutkindb,*, William S. Hancocka,*

a
Chemistry and Chemical Biology Department, Barnett Institute, Northeastern University,
341 Mugar Building, 360 Huntington Avenue, Boston, MA 02115, USA
b
Oral and Pharyngeal Cancer Branch, National Institute of Craniofacial and Dental Research,
National Institutes of Health, 30 Convent Drive, Building 30, Room 212, Bethesda, MD 20892-4330, USA
c
Department of Microbiology and Immunology, SUNY College of Medicine, Syracuse, NY 13210, USA
d
Department of Internal Medicine, Wayne State University/Karmanos Cancer Center,
4201 St Antoine, Detroit, MI 48201, USA
e
Department of Otolaryngology—Head and Neck Surgery, Wayne State University,
5E University Health Center, 4201 St Antoine, Detroit, MI 48201, USA
f
Department of Pathology, National Institute of Cancerology, Mexico. Av. San Fernando 22, Mexico, D.F.
g
Department of Head and Neck Surgery, 1515 Holcombe Blvd, U.T.M.D. Anderson Cancer Center,
Houston, TX 77030, USA
h
Department of Pathology, 1515 Holcombe Blvd, U.T.M.D. Anderson Cancer Center, Houston, TX 77030, USA

Received 2 June 2004; accepted 18 August 2004

KEYWORDS Summary Remarkable progress has been made to identify genes expressed in squ-
amous cell carcinomas of the head and neck (HNSCC). However, limited information
Oral cancer; is available on their corresponding protein products, whose expression, post-trans-
Microdissection; lational modifications, and activity are ultimately responsible for the malignant
Proteome; behavior of this tumor type. We have combined laser-capture microdissection
Biomarkers; (LCM) with liquid chromatography–tandem mass spectrometry (LC–MS/MS) to iden-
Drug targets; tify proteins expressed in histologically normal squamous epithelium and matching
Mass spectrometry

*
Corresponding authors. Tel.: +1 301 496 6259; fax: +1 301 402 0823 (J.S. Gutkind); tel.: +1 617 373 4881; fax: +1 617 373 2855
(W.S. Hancock).
E-mail addresses: sg39v@nih.gov (J.S. Gutkind), wi.hancock@neu.edu (W.S. Hancock).
 
These authors contributed equally to this work.

1368-8375/$ - see front matter Published by Elsevier Ltd.

doi:10.1016/j.oraloncology.2004.08.009
184 H. Baker et al.

SCC. The protein fraction from approximately 10,000–15,000 normal and tumor
cells was solubilized, digested with trypsin, and the resulting peptides were ana-
lyzed by LC–MS/MS. Database searching of the resulting sequence information iden-
tified 30–55 proteins per sample. Keratins were the most abundant proteins in both
normal and tumor tissues. Among the proteins differentially expressed, keratin 13
was much lower in tumors, whereas heat-shock (Hsp) family members were highly
expressed in neoplastic cells. Wnt-6 and Wnt-14 were identified in both normal
and tumor tissues, respectively, and placental growth factor (PIGF) was detected
only in tumors. Immunohistochemical analysis of HNSCC tissues revealed lack of ker-
atin 13 in tumor tissues, and strong staining in normal epithelia, and high expression
of Hsp90 in tumors. Our study, by combining LCM and proteomic technologies,
underscores the advantages of this approach to investigate complex changes at
the protein level in HNSCC, thus complementing existing and emerging genomic
technologies. These efforts may likely result in the identification of new biomarkers
for HNSCC that can be used to diagnose disease, predict susceptibility, and monitor
progression in individual patients.
Published by Elsevier Ltd.

Introduction estimated to be over 1 million proteins in a single

Annually, it is estimated that there are close to
500,000 cancer-related deaths in the United States
alone, and of these approximately 13,000 are
attributed to squamous cell carcinomas of the head
and neck (HNSCC), making it the sixth most com-
mon cause of cancer deaths.1 Even though risk fac-
tors for HNSCC, such as the use of tobacco and
alcohol, are well documented, a distinctive lack
of suitable pre-malignant markers for early detec-
tion and risk assessment is clearly reflected by
the fact that more than 50% of all HNSCC
patients have advanced disease at the time of diag-
nosis.2–4 Indeed, the five year survival rate of
HNSCC patients is in general poor, less than 50%,
and the prognosis of the advanced HNSCC cases
have not changed much over the past three dec-
ades.5 This limits the treatment options and renders
management of HNSCC extremely challenging.6
Thus, the ability to identify and confidently predict
malignant progression of HNSCC lesions will result
in a reduction in mortality, by aiding in early diag-
nosis and treatment of this disease.
Expression profiling using various microarray
platforms, and large-scale cDNA sequencing pro-
jects, such as CGAP (cancer genome anatomy pro-
ject), have led to a plethora of publicly available
information on gene transcripts, which has proven
to be fundamental for research efforts related to
understanding both human biology and disease
states.7–9 Despite this, there is only limited infor-
mation available on the gene products, currently
cell, which play vital roles in most key cellular
processes.10 Until recently, the analysis of a cell
proteome using two-dimensional gels (IEF and
SDS-PAGE) and mass spectrometry, was deemed
technologically challenging.11 For instance, im-
proved instrumental advances and the coupling of
HPLC to electrospray mass spectrometry combined
with the rapid growth in genomic databases ame-
nable to searching with mass spectrometry data,
now affords the opportunity to develop high-
throughput proteomic approaches to identity min-
ute amounts (typically femtomoles) of proteins
present in complex samples.10,12,13 In that context,
comparative analysis of the proteome in disease
and normal cells, selectively procured by the use
of laser-capture microdissection (LCM), is a critical
step in the validation of the results because of the
inherent clonal heterogeneity of most human can-
cers and the presence of host cells (fibroblast,
endothelial and inflammatory cells).14
In this study, we have used LCM to isolate
10,000–15,000 normal and tumor epithelial cells
from clinical samples of HNSCC, combined with
mass spectrometry, to explore the feasibility of
establishing a pattern of expressed cancer-related
proteins for HNSCC. Our findings indicate that these
approaches generate large proteomic datasets from
minimal clinical samples that is likely to lead to the
identification of novel HNSCC protein biomarkers.
Indeed, some of the emerging protein information
has already provided evidence of the expression
of molecules that might be involved in tumor
Proteome-wide analysis of head and neck cancer
progression, as well as clinically useful markers
defining the margins of the neoplastic lesions.
Materials and methods
Tissue samples
Histologically squamous mucosa and tumor spec-
imens from primary resected HNSCC from patients
who provided written informed consent for
planned studies and approved by Institutional Re-
view Board, were immediately harvested by a head
and neck pathologist, embedded in OCT (Tissue Tek
compound, Sakura Finetechnical, CA) and stored at
-70
C until use. Eight-micrometer cyrosections
were cut on to standard RNAase free glass slides,
which were stored at 70
C and used for down-
stream applications (see below) within a two-week
period. Prior to use, all tissue sections were
stained with hematoxylin and eosin, and confirmed
by a pathologist (A.A.M) as being either dysplasia,
malignant, invasive or normal epithelium (non-
malignant).
Laser-capture microdissection (LCM) and
protein extraction of HNSCC cells
Frozen tissue section slides were stained prior
to LCM, by briefly fixing in 70% ethanol (30 s),
washed in purified dH2O, placed in MayerÕs hema-
toxylin for 30 s and subsequently rinsed in 70% eth-
anol, the slides were counterstained with eosin
(10 s) and dehydrated with 95–100% ethanol and
SAFECLEAR II (xylene substitute; Fisher Diagnos-
tics, Middletown, VA, USA), and thoroughly air
dried. For LCM, stained uncovered slides were
viewed and after locating the cells of interest, a
CapSure LCM Cap (Arcturus, Mountain View, CA,
USA) was placed over the target area and pulsed
with laser to adhere cells to the cap, which after
sufficient capture (5000 cells) was transferred to
a 0.5 mL sterile microfuge tube for immediate pro-
cessing. For each sample, approximately 10,000–
15,000 cells were captured on multiple caps,
which were subsequently processed for protein
extraction. The procedure of protein extraction
from CapSure caps was essentially as described.15
Briefly, 30–50 lL PicoPure (Arcturus) protein
extraction buffer was placed in each tube and with
the enclosed caps, the tubes were inverted ensur-
ing the buffer was adequately dispersed to cover
the surface of the caps, followed by vortexing
185
for 10 s and incubation at 65
C for approximately
3 h. The extraction solution was then recovered
by centrifugation (14,000 rpm) for 2 min, pooled
and stored at
80
C until ready to be processed
for analysis.
Preparation of protein extracts
The protocol used to prepare protein extracts
for mass spectrometry was adopted from the
one used by Zhang et al.16 and the reagents used
included sequence grade trypsin (Promega,
Madison, WI), dithiothreitol (DDT), iodacetamide,
ammonium bicarbonate (NH4HCO3), and formic
acid (ICN Biomedical Inc., Aurora, OH). HPLC
grade solvents (methanol, chloroform and aceto-
nitrile) were purchased from Fisher Scientific
(Hanover Park, IL). Briefly, the frozen protein
extracts were thawed to room temperature,
followed by denaturation (25 mM DDT) and alkyl-
ation (300 mM iodoacetamide). The salts and
detergent were then removed from the samples
by methanol/chloroform precipitation. After re-
moval of the supernatant, the protein precipitates
were rinsed once in methanol, air dried, and dis-
solved in 15 lL of 100 mM NH4HCO3 at pH 8.0, with
agitation for up to 45 min. Dissolved proteins were
digested with trypsin (1:20 w/w) at 26
C initially
for 12 h, and then fresh trypsin was added at
the same ratio and the digestion continued for
an additional 12 h. The final volume of each sam-
ple was 17 lL.
Liquid chromatography (LC)–tandem mass
spectrometry (MS/MS) analysis of samples
LC–MS/MS was performed on a ThermoFinnigan
Deca-XP coupled to a Surveyor LC system (Thermo-
Finnigan, San Jose, CA). Fifteen microliters of the
analyte solution was injected on a Thermo-Hypersil
100 · 0.18 mm C18 reversed phase microcolumn. A
210 minute 0.1% formic acid, ACN/0.1% formic
acid, H2O gradient was followed by analysis on
the orthogonal microspray ion trap mass spectrom-
eter. The flow rate was 150 lL a minute, and after
splitting this was maintained at 1.5 lL a minute.
The dynamic exclusion mode on the mass spec-
trometers was used. To obtain peptide fragmenta-
tion spectra, each MS spectra was followed by 3
MS/MS scans and for analysis, precursor ion
+/
2 Da was excluded for 1.5 min if it was analyzed
twice in the previous 30 s, and normalized collision
energies were set at 35%.
186
Database search and protein identification
Data analysis and protein identification of the
emerging spectra was performed with the TurboSE-
QUEST search engine using default parameters and
the SWISS-PROT human protein database, and the
bioinformatics of the resulting information was as
previously described.17 Briefly, the algorithm com-
pares certain parameters for instance, similarities
between theroretical peptides derived from the
database and those from experimental MS/MS
scans, and subsequently assigns a unified ranking
score (combination of delta CN and X corr without
a bias against small peptides), which requires a
minimum of 2400 to discriminate between back-
ground noise and true identification. The search
parameters included modification of cysteine by
carbamidomethylation. The assignments of low-
level peptide sequences were then manually con-
firmed by comparing the acquired MS/MS spectra
to the theoretical fragmentation patterns. The
presence of multiple peptides of differing mass
and sequence of the same protein also indicated
favorably to the identity of the protein.
Immunohistochemistry
Archival HNSCC tissue sections were deparaffi-
nized in Safeclear II (Fisher Scientific, USA) and hy-
drated through graded alcohols, distilled water and
PBS 1x. Cytokeratin 13 was retrieved by incubating
with 0.25 mg/mL trypsin (Invitrogen Corporation,
Carlsbad, CA, USA) in PBS, for 30 min at 37
C.
Endogenous peroxidase activity was quenched by
incubation in 3% hydrogen peroxide in 96% alcohol.
Each incubation step was performed and followed
by three sequential washes in PBS for 5 min each.
Slides were incubated in blocking solution (5%
horse serum) for 30 min at room temperature and
reacted with the indicated primary antibodies
(Cyotkeratin 13, Novacastra Laboratories Ltd,
Newcastle upon Tyne, UK;Hsp90, Stressgen Bio-
technologies, Victoria, Canada) diluted 1:100 or
1:40, respectively, in blocking solution at 4
C,
overnight. Sections were then washed with PBS,
and incubated with biotinylated secondary anti-
body (Vector Laboratories, Burlingame, CA, USA)
for 30 min, washed again and reacted with the
ABC complex, prepared according to the manufac-
turer’s instructions (Vector Stain Elite, ABC kit,
Vector Laboratories) for 30 min at room tempera-
ture. The peroxidase was visualized with 3,3-diam-
inobenzidine (Sigma FASTDAB tablets with metal
enhancer, Sigma Chemical, St. Louis, MO, USA) as
chromogen substrate and closely monitored the
H. Baker et al.
staining under microscope. The reaction was
stopped with distilled water. Cytokeratin 13, were
counterstained with hematoxylin, washed in tap
water, dehydrated and mounted with glass cover
slips. Hsp90 slides were not counterstained, as a
faint positive nuclear reaction was evident at
developing time that may have been masked by
hematoxylin.
Results
HNSCC patient sets
For the proteomic analysis of HNSCC, we initially
selected five cases of matched histologically nor-
mal squamous epithelium and carcinoma from each
patient. As a criterion for selection, we elected to
focus on SCC from the tongue, the most fre-
quent anatomical location of the primary HNSCC
lesions.18 As indicated in Table 1, these lesions
were diagnosed as poorly (Case 1), moderate to
well-moderate (Case 2–4) or well-differentiated
(Case 5) tongue carcinomas. Normal mucosa was
defined tissue with squamous epithelium histologi-
cally lacking hyperplastic or dysplastic features by
light microscopic examination. Additionally, the
age of the HNSCC patients who underwent resec-
tion ranged from 48 to 74 years and all were of
male gender, which correlated well with the pa-
tient population at the highest risk of developing
this disease.5
Laser-capture microdissection (LCM)
We have previously applied LCM to successfully
isolate pure populations of normal and tumor cells
from HNSCC tissues for the subsequent gene
expression analysis.19 However, a similar applica-
tion to interface LCM with the analysis of the pro-
teome of specific cell populations could be
limited primarily due to sample availability and
the absence of in vitro amplification steps for pro-
tein identification. Indeed, to succeed, this ap-
proach is likely to depend on the use of highly
sensitive protein detection systems, together with
the accuracy of LCM for the isolation of specific cell
types, and the quality and preservation of the tis-
sue samples. To this end, LCM generally ensures
approximately 95% purity of the correct cell popu-
lation from tissues with optimum tissue integrity
intact.19 Thus, before proceeding with the analysis,
we first confirmed the quality and the histopathol-
ogy of all the samples by microscopic visualization
Proteome-wide analysis of head and neck cancer 187

Table 1 Clinical and histological information on HNSCC matched patient sets

Case Tissue Age Sex Site Grade
1 N 58 M Tongue Squamous epithelium
T Poorly differentiated
2 N 74 M Tongue Squamous epithelium
T Moderately differentiated
3 N 69 M Tongue Squamous epithelium
T Moderately differentiated
4 N 68 M Tongue Squamous epithelium
T Well-moderately differentiated
5 N 48 M Tongue Squamous epithelium
T Well differentiated
Each case consisted of non-malignant (N) squamous epithelium and tumor counterpart (T). Age and the sex of the HNSCC patients,
tumor grade, and the site within the oral cavity that the tissues originated from, is indicated.

of frozen tissue sections (8 lm thickness), stained
with hematoxylin-eosin. Oral squamous epithelium
lacking hyperplastic and/or dysplastic cellular fea-
tures were localized carefully for microdissection.
Similarly, all tumor biopsies were assessed to be
squamous cell carcinomas with variable histologi-
cal grades, which ranged from poorly to well-dif-
ferentiated lesions. After this initial assessment,
all the samples were considered suitable for LCM
and proteome analysis. Successful microdissection
of a representative sample (tumor) is depicted in
Figure 1, whereby tumor cells once identified (a)
are captured by laser onto caps (b) and after con-
firming the cell purity (c–d), processed for protein

Figure 1 Laser-capture microdissection of HNSCC: Tumor cryosections were fixed, stained with hematoxylin and
eosin (H&E), dehydrated and analyzed under the microscope. Once tumor cells of interest were identified (a),
microdissection was performed (b), and the cell purity confirmed by inspecting the area of tissue that underwent
procurement (c) and the captured cells (d). Bar represents 100 lm.
188 H. Baker et al.

extraction. Using this procedure, cells from each more traditional techniques like 2-D gels.20 LC–
tissue sample were rapidly captured on multiple MS/MS is now sensitive enough to analyze only a
caps.19 few cells, making it compatible with LCM and thus
suitable for proteome-wide profiling of cancer
LC–MS/MS analysis of microdissected tissues. To this end, protein extracts of cells micro-
HNSCC cells dissected from HNSCC patient sets underwent glo-
bal proteolysis with trypsin, and the resulting
A major challenge for proteomics research to- complex peptide mixture were subjected to liquid
day, is the identification of all the proteins in a gi- chromatography, and the separated molecules
ven biological system, as they exist in vivo. This is were fed directly into ESI (electrospray ionization)
due primarily to the vast complexity and the dy- and tandem mass spectrometer (MS/MS). Briefly,
namic range of proteins, and concurrent with tech- ESI ionizes peptides by passing the solution through
niques that lack the desired sensitivity of a high-voltage nozzle. Analyzers arranged in tan-
detection. Nonetheless, recent studies have indi- dem use radiofrequency and direct-current volt-
cated that the LC–MS/MS approach may signifi- ages to analyze ions based on their mass and
cantly improve the dynamic range of protein charge. The resulting full scan mass spectra was
detection and identification in comparison with collected in real time in a fully automated fashion

C:\Haven\oral_cancer_101003\run_8228N 10/11/200308:41:56PM

RT : 0.34-119.99
77.67 NL:
100
1.29E7
90 BasePeak
54.77 MS
34.10 44.41 run_8228N
80
Relative Abundance

47.00
70 25.89
56.39
60 26.12 36.98
58.37
61.45
50 78.24
30.11
68.14 76.30
40 90.24 94.01 97.17 117.92
66.64 82.95 105.79
30 24.30
108.37
22.55 111.08
20
21.52
19.07
10 0.66
1.94 13.45 15.12
0
10 20 30 40 50 60 70 80 90 100 110
Time(min)

run_8228N #1288 RT: 35.25 AV: 1 NL: 1.39E5 R.DAEEWFHAK.S

T: + c ESI d Full ms2 567.62@35.00 [ 145.00-1150.00]
817.2
100
688.2

465.0
80
Relative Abundance

60 502.2
818.2

509.5 689.2
40
915.2
187.0 946.2
631.0
20 218.1 316.0 445.0 799.5
558.4 887.2
736.5
1004.2
0
200 300 400 500 600 700 800 900 1000 1100
m/z

Figure 2 MS analysis of HNSCC: Whole cell protein extracts of LCM procured cells underwent global proteolysis with
trypsin and the resulting peptide fragments were first separated by liquid chromatography, followed by electrospray
ionization and analysis in a single MS mode, giving a mass spectrum of the peptides eluted with time (upper panel).
Next, fragments underwent tandem mass spectrometric sequencing and identification of selected peptides by
matching against protein sequence database (lower panel).
Proteome-wide analysis of head and neck cancer 189

Table 2 Proteins expressed in LCM-procured normal oral epithelium

Proteins identified Samples
Keratin, type I cytoskeletal 13 5
Keratin, type I cytoskeletal 17 5
Keratin, type II cytoskeletal 6F 5
Keratin, type II cytoskeletal 5 5
Keratin, type I cytoskeletal 14 5
Keratin, type II cytoskeletal 1 5
Keratin, type II cytoskeletal 6E 5
Keratin, type I cytoskeletal 10 5
Keratin, type II cytoskeletal 4 4
Annexin I (lipocortin I) (calpactin II) (chromobindin 9) 4
Actin, cytoplasmic 2 (gamma-actin) 4
Keratin, type I cytoskeletal 19 4
Keratin, type II cytoskeletal 2 epidermal 4
Serum albumin precursor 3
Hemoglobin alpha chain 3
Fibrinogen gamma chain precursor 3
Keratin, type II cytoskeletal 6A 3
Junction plakoglobin (desmoplakin III) 3
IGS 3
Annexin II (lipocortin II) (calpactin I heavy chain) 2
Elongation factor 1-alpha 1 (EF-1-alpha-1) 2
Heat shock 27 kDa protein (HSP 27) (stress-responsive protein 27) 2
Spectrin beta chain, brain 1 (spectrin, non-erythroid beta chain 1) 2
WNT-6 protein precursor 2
Rod CGMP-specific 30 ,50 -cyclic phosphodiesterase alpha-subunit 2
Hemoglobin beta chain 2
Histone H2A.L (H2A/L) 2
Collagen alpha 2(I) chain precursor 2
Keratin, type II cytoskeletal 6C 2
Histone H4 2
Histone H3.3 (H3.A) (H3.B) (H3.3Q) 2
Keratin, type I cytoskeletal 9 2
Thioredoxin (ATL-derived factor) (ADF) (surface associated sulphydry) 1
Keratin, type I cytoskeletal 15 1
Elongation factor 2 (EF-2) 1
Calnexin precursor (major histocompatibility complex class I antigen) 1
Hypothetical zinc finger protein KIAA0296 1
T-complex protein 1, delta subunit (TCP-1-delta) 1
Splicing factor U2AF 65 kDa subunit 1
Acylamino-acid-releasing enzyme (acyl-peptide hydrolase) 1
Keratin, type I cytoskeletal 12 (cytokeratin 12) 1
Natural resistance-associated macrophage protein 1 (NRAMP 1) 1
Heat shock 70 kDa protein 1 (HSP70.1) (HSP70-1/HSP70-2) 1
EZRIN (p81) (cytovillin) (villin 2) 1
Serine/threonine protein phosphatase 2B catalytic subunit, alpha is 1
Nonsyndromic hearing impairment protein 5 1
Triple functional domain protein (PTPRF interacting protein) 1
C-terminal binding protein 1 (CTBP1) 1
Desmoplakin (DP) (250/210 kDa) 1
Lim domain kinase 2 (LIMK-2) 1
80 kDa MCM3-associated protein (ganp protein) 1
Nucleolar phosphoprotein P130 1
Transaldolase 1
Hypothetical protein KIAA0064 (HA1355) 1
Propionyl-COA carboxylase alpha chain, mitochondrial precursor 1
CTD-binding SR-like protein RA4 1
(continued on next page)
190 H. Baker et al.

Table 2 (continued)
Proteins identified Samples
Kinesin-like protein KIF3C 1
Keratin, type II cytoskeletal 8 1
Histone deacetylase 4 (HD4) (HA6116) 1
Interleukin-14 precursor (IL-14) 1
Opioid binding protein/cell adhesion molec 1
Transcription initiation factor TFIID 135 1
Jumonji protein 1
NG,NG-dimethylarginine dimethylaminohydrol 1
Eukaryotic translation initiation factor 3 1
Peripheral plasma membrane protein cask 1
Insulin-like growth factor I receptor precu 1
Myosin heavy chain, skeletal muscle, adult 2 1
Pregnancy zone protein precursor 1
Alpha-1-antitrypsin precursor (alpha-1 protease inhibitor) 1
Fibrinogen alpha/alpha-E chain precursor 1
14-3-3 protein sigma (stratifin) (epithelial cell marker protein 1) 1
Cadherin-19 precursor 1
Neuropeptide Y receptor type 4 (NPY4-R) (pancreatic polypeptide rec) 1
Alpha-2-macroglobulin precursor 1
Eosinophil peroxidase precursor (EPO) 1
Ornithine decarboxylase antizyme 2 (ODC-AZ 2) (AZ2) 1
Nitric-oxide synthase, brain (NOS, type I) 1
Glutathione synthetase (glutathione synthase) 1
Vitronectin precursor (serum spreading factor) (S-protein) 1
Tenascin precursor (TN) (hexabrachion) 1
Caspase-9 precursor (CASP-9) (ice-like apoptotic protease 6) 1
Keratin, type II cytoskeletal 2 oral 1
DNA topoisomerase I 1
Keratin, type I cuticular HA3-II 1
DNA-repair protein XRCC1 1
Neurofilament triplet H protein 1
Surfeit locus protein 6 1
Desmocollin 3A/3B precursor 1
Alpha-actinin 1 1
Peripheral plasma membrane protein cask 1
Zinc finger protein 222 1
Cholinephosphate cytidylyltransferase A 1
Versican core protein precursor (large fibroblast proteoglycan) 1
Fukutin precursor (fukuyama-type congenital muscular dystrophy pro) 1
Isoleucyl-TRNA synthetase, cytoplasmic (isoleucine–TRNA ligase) 1
Trichohyalin 1
Period circadian protein 1 (circadian pacemaker protein RIGUI) 1
Developmentally regulated GTP-binding protein 1 (DRG 1) 1
Protein tyrosine phosphatase, non-receptor type 14 1
Myosin VB (myosin 5B) 1
Tubulin beta-4 chain (tubulin beta-III) 1
RAS-related protein RAB-35 (RAB-1c) (GTP-binding protein ray) 1
DNA ligase III (polydeoxyribonucleotide synthase [ATP]) 1
Proteins detected in combined Normal samples. Number of samples in which the protein was detected is indicated.

to identify the constituent peptides, followed by a steady stream of peptide and fragment masses
tandem mass spectrometry and sequencing of se- to be fed to a computer for off-line database
lected peptides in each time interval. This MS to search and protein identification. A representative
MS–MS cycle spans approximately 0.2 s, enabling ion chromatogram mass spectrum for one of the
Proteome-wide analysis of head and neck cancer 191

Table 3 Proteins expressed in LCM-procured tumor squamous oral epithelium

Proteins identified in tumor Samples
Keratin, type II cytoskeletal 1 5
Keratin, type II cytoskeletal 6F 5
Keratin, type I cytoskeletal 14 5
Keratin, type I cytoskeletal 17 5
Keratin, type II cytoskeletal 5 5
Keratin, type II cytoskeletal 6E 5
Keratin, type I cytoskeletal 10 5
Serum albumin precursor 5
Keratin, type II cytoskeletal 2 epidermal 5
Actin, alpha skeletal muscle 4
Glyceraldehyde 3-phosphate dehydrogenase, liver 3
Hemoglobin alpha chain 4
Actin, cytoplasmic 2 3
Fibrinogen gamma chain precursor 3
Keratin, type II cytoskeletal 6A 3
Keratin, type I cytoskeletal 9 3
Elongation factor 1-alpha 2 (EF-1-alpha-2) 3
Desmoplakin (DP) (250/210 kDa) 3
Tubulin alpha-4 chain 3
Keratin, type I cytoskeletal 16 2
Keratin, type II cytoskeletal 3 2
Keratin, type II cytoskeletal 8 2
Heat shock protein HSP 90-beta (HSP 84) (HSP 90) 2
Elongation factor 1-alpha 1 (EF-1-ALPHA-1) 2
Heat shock 27 kDa protein (HSP 27) 2
Histone H4 2
Heat shock 70 kDa protein 1 (HSP70.1) 2
Pregnancy zone protein precursor 2
Histone H2A.L (H2A/L) 2
Annexin I (lipocortin I) (calpactin II) 2
Hemoglobin beta chain 2
Spectrin beta chain, brain 1 (spectrin, non-erythroid beta chain 1) 1
Protein-tyrosine phosphatase-like N precursor 1
Insulin receptor precursor (IR) 1
Fatty acid-binding protein, epidermal (E-FABP) (psoriasis-associated) 1
WNT-14 protein 1
Zinc finger protein 43 (zinc protein HTF6) 1
DNA topoisomerase I 1
Cholinephosphate cytidylyltransferase A (phosphorylcholine transfer) 1
Vacuolar ATP synthase subunit C (V-ATPase C subunit) 1
T-cell differentiation antigen CD6 precursor (T12) (TP120) 1
Vimentin 1
Histone H3.3 (H3.A) (H3.B) (H3.3Q) 1
Glutamate [NMDA] receptor subunit epsilon 1
Collagen alpha 2(VI) chain precursor 1
Camp and camp-inhibited CGMP 30 ,50 -cyclic 1
Signal recognition particle 68 kDa protei 1
Inositol polyphosphate 5-phosphatase OCRL 1
Complement decay-accelerating factor precur 1
Solute carrier family 2, facilitated glucos 1
ACTG_human actin, cytoplasmic 2 1
Calgranulin A (migration inhibitory factor-related protein 8) 1
Galectin-7 (HKL-14) 1
Annexin II (lipocortin II) (calpactin I heavy chain) 1
14-3-3 Protein gamma (protein kinase C inhibitor protein-1) 1
Ephrin-B2 precursor (EPH-related receptor tyrosine kinase ligand 5) 1
(continued on next page)
192 H. Baker et al.

Table 3 (continued)
Proteins identified in tumor Samples
Fructose-bisphosphate aldolase A (muscle-type aldolase) (lung cancer) 1
Cortistatin precursor [contains: cortistatin-29;cortistatin-17] 1
Calnexin precursor (major histocompatibility complex class I antigen) 1
Peroxiredoxin 1 (thioredoxin peroxidase 2) 1
Gamma-interferon inducible lysosomal thiol reductase precursor 1
CENP-F kinetochore protein (centromere protein F) 1
Cell division cycle 2-related protein kinase 7 1
RAS GTPASE-activating-like protein IQGAP1 (P195) 1
Placenta growth factor precursor (PlGF) 1
40S ribosomal protein S26 1
Origin recognition complex subunit 2 1
Cleavage stimulation factor, 64 kDa subunit 1
GMP synthase [glutamine-hydrolyzing] 1
Fibrinogen gamma chain precursor 1
Trans-1,2-dihydrobenzene-1,2-diol dehydrogenase 1
Keratin, type I cytoskeletal 15 1
Lysosome-associated membrane glycoprotein 1 precursor (LAMP-1) 1
Endothelial actin-binding protein (ABP-280) 1
Alpha-2-macroglobulin precursor (ALPHA-2-M) 1
78 kDa glucose-regulated protein precursor (GRP 78) 1
Myosin heavy chain, skeletal muscle, adult 2 1
Cyclic-AMP-dependent transcription factor ATF-6 1
Keratin, type I cytoskeletal 13 (cytokeratin 13) (K13) (CK 13) [MAS] 1
Keratin, type II cytoskeletal 4 1
Keratin, type I cytoskeletal 19 (cytokeratin 19) 1
Keratin, type I cuticular HA6 (hair keratin, type I HA6) 1
Keratin, type I cuticular HA3-II (hair keratin, type I HA3-II) 1
D-site-binding protein (albumin D box-binding protein) 1
Down syndrome critical region protein 5 (Down syndrome critical reg) 1
Alpha-1 catenin (cadherin-associated protein) (alpha E-catenin) 1
P55-c-FOS proto-oncogene protein (cellular oncogene c-FOS) 1
Interstitial collagenase precursor (matrix metalloproteinase-1) 1
Serine/threonine protein phosphatase 2A, 72/130 kDa regulatory 1
CENP-F kinetochore protein (centromere protein F) (mitosin) 1
Inward rectifier potassium channel 2 1
Trifunctional enzyme beta subunit, mitochondrial precursor (TP-BETA) 1
Proteins detected in combined tumor samples. Number of samples in which the protein was detected is indicated.

HNSCC cases is shown in Figure 2, whereby a com-
plex mixture of tryptic digested peptides, were
separated and a peptide for cytokeratin 13 was se-
quenced from its MS/MS fragmentation pattern.
The data indicate that protein identification can
be achieved from small sample material.
Identification of proteins expressed in
HNSCC
Although a list of candidate proteins expressed
in both normal oral epithelium and tumor speci-
mens can be obtained using 2-D gels, the enhanced
sensitivity needed for analyzing LCM samples re-
quires LC–MS/MS and the tryptic digestion of the
samples. This generates several dozen peptides
per protein and leads to complex proteome profil-
ing due to the large number of redundant peptides.
Thus, data confidence to identify true proteins
from background noise is primarily facilitated by
the use of computational algorithms. For each sam-
ple analyzed in this study, a list containing the
identity of numerous detected molecules was gen-
erated, but only those demonstrating high confi-
dence (unified score >2400 and multiple peptide
coverage) were selected.15,17 In this regard, be-
tween 21–53 distinct proteins were readily identi-
fied in each sample (data not shown). Although
many proteins were noted to be common to most
or all normal or tumor tissues, a significant number
Proteome-wide analysis of head and neck cancer 193

Table 4 Protein frequency in normal (panel A) and tumor (panel B) oral epithelium
Panel A: Abundant proteins in normal Samples
Keratin, type I cytoskeletal 13 5
Keratin, type I cytoskeletal 17 5
Keratin, type II cytoskeletal 6F 5
Keratin, type II cytoskeletal 5 5
Keratin, type I cytoskeletal 14 5
Keratin, type II cytoskeletal 1 5
Keratin, type II cytoskeletal 6E 5
Keratin, type I cytoskeletal 10 5
Keratin, type II cytoskeletal 4 4
Annexin I (lipocortin I) (calpactin II) (chromobindin 9) 4
Actin, cytoplasmic 2 (GAMMA-ACTIN) 4
Keratin, type I cytoskeletal 19 4
Keratin, type II cytoskeletal 2 epidermal 4
Serum albumin precursor 3
Hemoglobin alpha chain 3
Fibrinogen gamma chain precursor 3
Keratin, type II cytoskeletal 6A 3
Junction plakoglobin (desmoplakin III) 3
IGS 3
Annexin II (lipocortin II) (calpactin I heavy chain) 2
Elongation factor 1-alpha 1 (EF-1-ALPHA-1) 2
Heat shock 27 kDa protein (HSP 27) (stress-responsive protein 27) 2
Spectrin beta chain, brain 1 (Spectrin, non-erythroid beta chain 1) 2
WNT-6 protein precursor 2
ROD cGMP-specific 30 ,50 -cyclic phosphodiesterase alpha-subunit 2
Hemoglobin beta chain 2
Histone H2A.L (H2A/L) 2
Collagen alpha 2(I) chain precursor 2
Keratin, type II cytoskeletal 6C 2
Histone H4 2
Histone H3.3 (H3.A) (H3.B) (H3.3Q) 2
Keratin, type I cytoskeletal 9 2
Heat shock 70 kDa protein 1 (HSP70.1) (HSP70-1/HSP70-2) 1
Desmoplakin (DP) (250/210 kDa) 1
Glyceraldehyde 3-phosphate dehydrogenase, liver 1
Keratin, type II cytoskeletal 8 1
Panel B: Abundant proteins in tumor Samples
Keratin, type II cytoskeletal 1 5
Keratin, type II cytoskeletal 6F 5
Keratin, type I cytoskeletal 14 5
Keratin, type I cytoskeletal 17 5
Keratin, type II cytoskeletal 5 5
Keratin, type II cytoskeletal 6E 5
Keratin, type I cytoskeletal 10 5
Serum albumin precursor 5
Keratin, type II cytoskeletal 2 epidermal 5
Actin, alpha skeletal muscle 4
Hemoglobin alpha chain 4
Glyceraldehyde 3-phosphate dehydrogenase, liver 3
Actin, cytoplasmic 2 3
Fibrinogen gamma chain precursor 3
Keratin, type II cytoskeletal 6A 3
Keratin, type I cytoskeletal 9 3
Elongation factor 1-alpha 2 (EF-1-alpha-2) 3
(continued on next page)
194 H. Baker et al.

Table 4 (continued)
Desmoplakin (DP) (250/210 kDa) 3
Tubulin alpha-4 chain 3
Keratin, type I cytoskeletal 16 2
Keratin, type II cytoskeletal 3 2
Keratin, type II cytoskeletal 8 2
Heat shock protein HSP 90-beta (HSP 84) (HSP 90) 2
Elongation factor 1-alpha 1 (EF-1-ALPHA-1) 2
Heat shock 27 kDa protein (HSP 27) 2
Histone H4 2
Heat shock 70 kDa protein 1 (HSP70.1) 2
Pregnancy zone protein precursor 2
Histone H2A.L (H2A/L) 2
Annexin I (lipocortin I) (calpactin II) 2
Hemoglobin beta chain 2
Spectrin beta chain, brain 1 (spectrin, non-erythroid beta chain 1) 1
WNT-14 protein 1
Histone H3.3 (H3.A) (H3.B) (H3.3Q) 1
Annexin II (lipocortin II) (calpactin I heavy chain) 1
Keratin, type I cytoskeletal 13 (cytokeratin 13) 1
Keratin, type I cytoskeletal 19 (cytokeratin 19) 1
Collagen alpha 2(VI) chain precursor 1
Histone H3.3 (H3.A) (H3.B) (H3.3Q) 1
Number of samples in which the protein was detected is indicated. Proteins highlighted in italic font are more frequent in
normal oral epithelium, whereas those highlighted in bold font are more frequent in tumor epithelial cells.

of proteins identified were unique to individual
samples. In this manner, approximately 90–105 un-
ique proteins were identified in all normal and tu-
mor samples. Table 2 shows a partial list of the
proteins identified based on a unified score
>2400, and the frequency of detection in normal
samples (n = 5). In this group, the keratins were
readily detected in all five normal samples. These
include keratin 1, 2, 4, 5, 6, 10, 13, 17 and 19.
Some structural (actin, junction plakoglobin, colla-
gen), serum (serum albumin, hemoglobin), histone
(H2A/L, H3, H4), were also identified. Proteins of
interest detected in 2/5 samples, include Wnt-6,
Annexin I and II, Hsp27 and elongation factor 1 al-
pha 1. Hypothetical Zinc finger protein (KIAA0296),
Jumonji, Pregnancy Zone protein precursor, 14-3-3
sigma, caspase 9 precursor, DNA Topoisomerase 1,
DNA repair protein XRCC1 and Rab 35 were present
in at least 1/5 normal samples. In the tumor sam-
ples (Table 3), the keratins were again well repre-
sented, with the exception of keratin 13 and 19
that were much less expressed than in normal squa-
mous tissue. There was some indication that the
keratin 6 isoforms (6C) may be differentially ex-
pressed. Furthermore, Hsps were highly detected
in the tumor cells (Hsp27, 70, 90). Elongation fac-
tor 1 alpha 2 and desmoplakin were present in mul-
tiple tumor samples (3/5). Additional proteins of
interest include Wnt-14, psoriasis-associated fatty
acid binding protein, 14-3-3 gamma, cdc2 related
protein kinase 7, Ras GTPase activating like pro-
tein, placenta growth factor precursor (PIG), alpha
1 catenin, c-Fos, MMP-1 and mitosin, which were
detected in one of five tumor samples. Frequency
of selected proteins detected in all normal and tu-
mor samples are summarized in Table 4 (panels A
and B), which also indicates the differential
expression of keratin (6C, 13, 19) and Hsp (70,
90) family members.
Detection of cytokeratin 13 and Hsp90 by
immunohistochemistry in archival HNSCC
tissues
As our previous data indicated that a sub-group
of proteins might be differentially expressed in
HNSCC tissue samples, we opted to confirm the
presence or absence of cytokeratin 13 and Hsp90
in a representative set of HNSCC tissues. Because
frozen sections offer poor morphology for immuno-
histochemistry, sections from paraffin-embedded
archival HNSCC tissues were used. As seen in Figure
3a, cytokeratin 13 was expressed in normal epithe-
lia, and the staining was preferentially located
within the upper layers of the squamous epithelium
Proteome-wide analysis of head and neck cancer 195

Figure 3 Detection of keratin 13 and Hsp90 by immunocytochemistry in archival HNSCC. Archival oral squamous cell
carcinoma paraffin sections were assessed for cytokeratin 13 (a) and Hsp90 (b), as described in materials and methods
section . Expression of cytokeratin 13 is shown to be present in normal epithelia (left panel), while tumor cells show
reduced levels (middle panel) and H&E staining indicates the boundary between the tumor and stoma (fl), and distal to
non-malignant normal epithelia (right panel). Expression of Hsp90 protein was undetectable in normal tissues (left
panel), while elevated levels were readily observed in cells (right panel). Bar represents 100 lm.

(left panel). In malignant tumors the expression of
cytokeratin 13 is lost; adjacent non-malignant epi-
thelial cells were positively stained (middle panel).
The highlighted area (arrows) distinguishes the
margin of the tumor from the adjacent stroma,
and the distal non-malignant epithelium (middle
and right panels). For Hsp90 (Fig. 3b), non-neoplas-
tic squamous epithelium (left panel) gave a uni-
formly mild cytoplasmic staining with scattered
nuclear reaction; in malignant tumors (right panel)
196
the immunoreactivity was exclusively cytoplasmic
and moderate to strong in intensity.
Discussion
Direct and rapid analysis of low abundance pro-
teins in complex mixtures by mass spectrometry is
a compelling approach to comprehensively identi-
fying protein components. It provides a list of ac-
tual proteins present in a purified complex instead
of a descriptive visualization of the components
that must be individually identified.21 Typically,
1D or 2D gel electrophoresis, a time- and labor-
intensive process with limited molecular mass or
pI ranges, is used to resolve complicated protein
mixtures into individual bands or spots.11 Peptides
from digested proteins must be recovered from
the stained gel or an electroblot of the gel. Auto-
mation of this process requires expensive robotics
to isolate and process the spots. Directly identify-
ing proteins from the digest of a complex sample
bypasses the potential limitations of gel electro-
phoresis, including protein insolubility, limited
fractionation ranges, and limited recoveries of
material, and provides the ability to perform pro-
teomic analyses on samples containing a few thou-
sand cells that could open the possibility of
performing individual patient studies.22,23 How-
ever, the identification of proteins in a few cells
has been often associated with technical chal-
lenges such as the solubilization, extraction and
separation of whole cellular proteins. A recent
improvement in buffers and separation tech-
niques, such as liquid chromatography coupled
directly to MS/MS, makes it possible to generate
high-quality data from a very little starting
material.
In this regard, a recent study using LCM to col-
lect a small number of homogeneous cells for pro-
teomic analysis by the LC–MS/MS was evaluated,
which helped establish extraction procedures en-
abling the subsequent solubilization of the majority
of the protein complexes in a single step.15
Although the use of cultured cells can serve as a
model system for clinical studies, the biggest chal-
lenge in cancer proteomics is to unravel the molec-
ular complexity of the tissue microenvironment,
and thus the proteomic analysis of cells cultured
in vitro may not truly reflect the expression pattern
of cells in vivo.24 To this end, we opted to analyze a
representative set of HNSCC patient samples, in or-
der to identify proteins that may provide insights to
the pathogenesis of the disease or potential bio-
H. Baker et al.
markers. It is noteworthy that two-dimensional
chromatography is typically used for separation,
but for this study, all the tissue samples underwent
a single dimension RPLC, primarily to maximize the
recovery of small samples of complex protein mix-
tures, 1–5 lg of total protein, and to minimize con-
taminants from sample handling. Altogether, this
novel method enabled the identification of approx-
imately 91–103 proteins expressed in either nor-
mal or tumor tissues.
From the proteins detected, an assessment was
made to determine whether any of these molecules
were differentially detected. Down-regulated pro-
teins in tumors included several cellular structural
molecules such as cytoskeletal keratins 13, 19, and
6c. These were all found in the normal samples but
at a much lower levels and frequency in the tumor
samples. Studies have shown that down-regulation
of specific cytokeratins in squamous cell carcino-
mas correlates in part with the general loss of dif-
ferentiation.25 Specifically, cytoskeletal keratins
13 and 19 are often less expressed in tumor than
normal cells.26,27 Our findings of both keratins in
the normal and not in corresponding malignant tis-
sues support these studies. We further confirmed
the expression of keratin 13 in archival tissues,
and noted that protein levels were primarily con-
fined to the upper layer of the normal squamous
epithelia, and not to the tumor counterpart. These
observations were further corroborated during
gene array analysis of HNSCC samples from an inde-
pendent study, in which we readily observed the
presence of keratin 13 mRNA in normal samples,
but its absence in tumors (data not shown).
Although the mechanism leading to the loss in
expression of keratin 13 in HNSCC is unclear, this
protein might play a role in the HNSCC carcinogen-
esis, possibly by acting as a tumor suppressor gene
product, and thus its level of expression may be
relevant for the diagnosis and prognosis of this can-
cer type. Additionally, our findings suggest that
keratin 13 (and/or19) might be used as a clinical
marker, whereby the presence or absence of this
protein could aid to define the surgical margin.28
Of the up-regulated proteins, the Hsps were the
most notable in the tumor samples. Hsp70 and
Hsp90 kDa both appeared in two tumor samples,
but not in the controls. While the presence of
Hsp may indicate a general response to biological
stress experienced by the tumor cells, the precise
role of these proteins in HNSCC is not clear.29,30
Interestingly, there are reports indicating that
Hsp70 may promote proliferation and survival of
oral tumor cells, possibly by interacting and inhib-
iting apoptosis mediators, such as apoptosis-induc-
Proteome-wide analysis of head and neck cancer
ing factor (AIF) and Bag-1.31,32 Furthermore, over-
expression of Hsp70 on the cell membrane of
HNSCC cells was evaluated as a potential target
for natural killer (NK) cells on tumor material and
control tissues of HNSCC patients.33 On the other
hand, Hsp90, which is elevated in HNSCC, may rep-
resent a novel target for cancer treatment, as 17-
allylamino-17-demethoxygeldanamycin (17-AAG),
a benzoquinoid ansamycin antibiotic currently in
Phase I clinical trial, has been recently shown that
by binding to Hsp90 can cause the destabilization
of various Hsp90-dependent kinases important in
oncogenesis, and consequently can exert a desired
anticancer effect.34,35 For example, Hsp90 forms a
complex with Akt, and the use of 17-AAG has
been demonstrated to reduce not only the expres-
sion levels of the kinase but also its intrinsic activ-
ity, which is elevated in HNSCC, thus providing a
rational for evaluating 17-AAG in this tumor
type.36,37
Proteins usually expressed at very low level and
generally difficult to detect, were also identified in
our analysis. These include Wnt-6, Wnt-14 and, pla-
cental growth factor (PIGF). The fact that these
proteins could be detected at all in such limited
starting material supports the sensitivity of our
experimental approach, and the likely high levels
of these molecules in normal or tumor tissues. In
the case of the Wnt proteins, previous studies using
gene arrays, including ours, have implicated the
Wnt pathway in HNSCC pathogenesis.19,38 More re-
cently, evidence was obtained that enhanced Wnt
expression and signaling accelerate the progression
of carcinomas by activating the epithelial–mesen-
chymal transition and local invasiveness.39 None-
theless, very limited information is currently
available specifically about Wnt-6 and Wnt-14. In
regard to the Wnt-6, its expression in normal oral
epithelium matches its pan-epithelial expression
in normal epithelium in chicken, but it is also ex-
pressed at certain levels in cervical and colon can-
cer cells.40,41 Thus, further work will be required to
study whether Wnt-6 is indeed expressed primarily
in normal versus tumor squamous epithelium. Wnt-
14 is expressed in joint-forming regions of the
skeleton during development, and its ectopic
expression has been documented at the mRNA level
in a variety of cancer cells, including those derived
from brain, breast, pancreas, esophageal, gastric
and cervical tumors.42,43 Thus, our findings suggest
that Wnt-14 protein may correlate or even play a
causative in carcinogenesis in HNSCC, an exciting
possibility that warrants further investigation. PIGF
was also detected in tumor samples and its potent
pro-angiogenic function, through the activation of
197
the flt-1 vascular endothelial growth factor (VEGF)
receptor (VEGFR-1) has been well documented, but
very limited information is available on its putative
role in tumor-induced angiogenesis. Nonetheless,
available information suggests that PIGF is ex-
pressed in a variety of tumors, including meningio-
mas, prostate cancer, and renal cell cancer.44,45 Of
more direct relevance to our present study, high
expression levels of PIGF have been recently docu-
mented in mouse models of skin SCC.46 Thus, these
observations raise the exciting possibility that this
poorly studied angiogenic factor may play an unex-
pected role in the process of neovascularization
that characterizes both experimental as well as hu-
man squamous carcinogenesis.
In summary, we provide evidence that the use of
recently developed LC–MS/MS techniques com-
bined with laser-assisted microdissection and cap-
ture of human cells from clinical samples may
enable a proteome-wide analysis of proteins ex-
pressed in normal and cancerous tissues. The
emerging information is expected to provide valu-
able information regarding the still unknown
molecular mechanisms promoting the malignant
conversion of the normal oral epithelium, as well
as may help identify biomarkers of disease devel-
opment and progression, and novel therapeutic tar-
gets for HNSCC.
References
1. Jemal A, Tiwari RC, Murray T et al. Cancer statistics. CA
Cancer J Clin 2004;54(1):8–29.
2. Shiu MN, Chen TH. Impact of betel quid, tobacco and
alcohol on three-stage disease natural history of oral
leukoplakia and cancer: implication for prevention of oral
cancer. Eur J Cancer Prev 2004;13(1):39–45.
3. Ha PK, Benoit NE, Yochem R et al. A transcriptional
progression model for head and neck cancer. Clin Cancer
Res 2003;9(8):3058–64.
4. Bettendorf O, Piffko J, Bankfalvi A. Prognostic and predic-
tive factors in oral squamous cell cancer: Important tools
for planning individual therapy? Oral Oncol.
2004;40(2):110–9.
5. Neville BW, Day TA. Oral cancer and precancerous lesions.
CA Cancer J Clin 2002;52(4):195–215.
6. Bankfalvi A, Piffko J. Prognostic and predictive factors in
oral cancer: The role of the invasive tumour front. J Oral
Pathol Med 2000;29(7):291–8.
7. Alevizos I, Mahadevappa M, Zhang X et al. Oral cancer in
vivo gene expression profiling assisted by laser capture
microdissection and microarray analysis. Oncogene
2001;20(43):6196–204.
8. Strausberg RL, Greenhut SF, Grouse LH, Schaefer CF,
Buetow KH. In silico analysis of cancer through the Cancer
Genome Anatomy Project. Trends Cell Biol 2001;11(11):
S66–71.
198
9. Leethanakul C, Knezevic V, Patel V et al. Gene discovery in
oral squamous cell carcinoma through the Head and Neck
Cancer Genome Anatomy Project: confirmation by micro-
array analysis. Oral Oncol 2003;39(3):248–58.
10. Wasinger VC, Corthals GL. Proteomic tools for biomedicine.
J Chromatogr B Anal Technol Biomed Life Sci 2002;
771(1–2):33–48.
11. Ong SE, Pandey A. An evaluation of the use of two-
dimensional gel electrophoresis in proteomics. Biomol Eng
2001;18(5):195–205.
12. Sickmann A, Mreyen M, Meyer HE. Mass spectrometry–a key
technology in proteome research. Adv Biochem Eng Bio-
technol 2003;83:141–76.
13. Bodnar WM, Blackburn RK, Krise JM, Moseley MA. Exploiting
the complementary nature of LC/MALDI/MS/MS and LC/ESI/
MS/MS for increased proteome coverage. J Am Soc Mass
Spectrom 2003;14(9):971–9.
14. Simone NL, Bonner RF, Gillespie JW, Emmert-Buck MR,
Liotta LA. Laser-capture microdissection: opening the
microscopic frontier to molecular analysis. Trends Genet
1998;14(7):272–6.
15. Wu SL, Hancock WS, Goodrich GG, Kunitake ST. An
approach to the proteomic analysis of a breast cancer cell
line (SKBR-3). Proteomics 2003;3(6):1037–46.
16. Zhang S, Van Pelt CK, Henion JD. Automated chip-based
nanoelectrospray–mass spectrometry for rapid identifica-
tion of proteins separated by two-dimensional gel electro-
phoresis. Electrophoresis 2003;24(21):3620–32.
17. Anderson DC, Li W, Payan DG, Noble WS. A new algorithm
for the evaluation of shotgun peptide sequencing in
proteomics: support vector machine classification of pep-
tide MS/MS spectra and SEQUEST scores. J Proteome Res
2003;2(2):137–46.
18. Lamaroon A, Pattanaporn K, Pongsiriwet S et al. Analysis of
587 cases of oral squamous cell carcinoma in northern
Thailand with a focus on young people. Int J Oral Maxillofac
Surg 2004;33(1):84–8.
19. Leethanakul C, Patel V, Gillespie J et al. Distinct pattern of
expression of differentiation and growth-related genes in
squamous cell carcinomas of the head and neck revealed by
the use of laser capture microdissection and cDNA arrays.
Oncogene 2000;19(28):3220–4.
20. Wu SL, Amato H, Biringer R, Choudhary G, Shieh P, Hancock
WS. Targeted proteomics of low-level proteins in human
plasma by LC/MSn: using human growth hormone as a model
system. J Proteome Res 2002;1(5):459–65.
21. Chaurand P, Schwartz SA, Caprioli RM. Assessing protein
patterns in disease using imaging mass spectrometry. J
Proteome Res 2004;3(2):245–52.
22. Peng J, Gygi SP. Proteomics: the move to mixtures. J Mass
Spectrom 2001;36(10):1083–91.
23. Glanzer JG, Eberwine JH. Expression profiling of small
cellular samples in cancer: less is more. Br J Cancer
2004;90(6):1111–4.
24. Li Zang DPT, Hancock WS, Sgroi DC, Karger BL. Proteomic
analysis of ductal carcinoma of the breast using laser
capture microdissection, LC–MS, and 16O/18O isotopic
labeling. J Proteome Res 2004;3(3):604–12.
25. Real FX, Vila MR, Skoudy A, Ramaekers FC, Corominas JM.
Intermediate filaments as differentiation markers of exo-
crine pancreas. II. Expression of cytokeratins of complex
and stratified epithelia in normal pancreas and in pancreas
cancer. Int J Cancer 1993;54(5):720–7.
26. Morifuji M, Taniguchi S, Sakai H, Nakabeppu Y, Ohishi M.
Differential expression of cytokeratin after orthotopic
H. Baker et al.
implantation of newly established human tongue cancer
cell lines of defined metastatic ability. Am J Pathol
2000;156(4):1317–26.
27. Prasad S, Thraves P, Dritschilo A, Kuettel M. Expression of
cytokeratin-19 as a marker of neoplastic progression of
human prostate epithelial cells. Prostate 1998;35(3):
203–11.
28. Leunig A, Betz CS, Mehlmann M et al. Detection of
squamous cell carcinoma of the oral cavity by imaging 5-
aminolevulinic acid-induced protoporphyrin IX fluores-
cence. Laryngoscope 2000;110(1):78–83.
29. He QY, Chen J, Kung HF, Yuen AP, Chiu JF. Identification of
tumor-associated proteins in oral tongue squamous cell
carcinoma by proteomics. Proteomics 2004;4(1):271–8.
30. Kaur J, Srivastava A, Ralhan R. Expression of 70 kDa heat
shock protein in oral lesions: marker of biological stress or
pathogenicity. Oral Oncol 1998;34(6):496–501.
31. Schmitt E, Parcellier A, Gurbuxani S et al. Chemosensiti-
zation by a non-apoptogenic heat shock protein 70-binding
apoptosis-inducing factor mutant. Cancer Res
2003;63(23):8233–40.
32. Townsend PA, Cutress RI, Sharp A, Brimmell M, Packham G.
BAG-1: a multifunctional regulator of cell growth and
survival. Biochim Biophys Acta 2003;1603(2):83–98.
33. Kleinjung T, Arndt O, Feldmann HJ et al. Heat shock
protein 70 (Hsp70) membrane expression on head-and-neck
cancer biopsy-a target for natural killer (NK) cells. Int J
Radiat Oncol Biol Phys 2003;57(3):820–6.
34. Neckers L, Ivy SP. Heat shock protein 90. Curr Opin Oncol
2003;15(6):419–24.
35. Neckers L. Development of small molecule Hsp90 inhibitors:
utilizing both forward and reverse chemical genomics for
drug identification. Curr Med Chem 2003;10(9):733–9.
36. Solit DB, Basso AD, Olshen AB, Scher HI, Rosen N. Inhibition
of heat shock protein 90 function down-regulates Akt kinase
and sensitizes tumors to Taxol. Cancer Res
2003;63(9):2139–44.
37. Amornphimoltham P, Sriuranpong V, Patel V, Benavides F,
Conti CJ, Sauk J et al. Persistent activation of the Akt
pathway in head and neck squamous cell carcinoma: a
potential target for UCN-01. Clin Cancer Res 2004;10(12):
4029–37.
38. Rhee CS, Sen M, Lu D et al. Wnt and frizzled receptors as
potential targets for immunotherapy in head and neck
squamous cell carcinomas. Oncogene 2002;21(43):
6598–605.
39. Uraguchi M, Morikawa M, Shirakawa M, Sanada K, Imai K.
Activation of WNT family expression and signaling in
squamous cell carcinomas of the oral cavity. J Dent Res
2004;83(4):327–32.
40. Schubert FR, Mootoosamy RC, Walters EH et al. Wnt6 marks
sites of epithelial transformations in the chick embryo.
Mech Dev 2002;114(1–2):143–8.
41. Kirikoshi H, Sekihara H, Katoh M. WNT10A and WNT6,
clustered in human chromosome 2q35 region with head-to-
tail manner, are strongly coexpressed in SW480 cells.
Biochem Biophys Res Commun 2001;283(4):798–805.
42. Hartmann C, Tabin CJ. Wnt-14 plays a pivotal role in
inducing synovial joint formation in the developing appen-
dicular skeleton. Cell 2001;104(3):341–51.
43. Kirikoshi H, Sekihara H, Katoh M. Molecular cloning and
characterization of WNT14B, a novel member of the WNT
gene family. Int J Oncol 2001;19(5):947–52.
44. Autiero M, Luttun A, Tjwa M, Carmeliet P. Placental growth
factor and its receptor, vascular endothelial growth factor
Proteome-wide analysis of head and neck cancer
receptor-1: novel targets for stimulation of ischemic tissue
revascularization and inhibition of angiogenic and inflam-
matory disorders. J Thromb Haemost 2003;1(7): 1356–
70.
45. Matsumoto K, Suzuki K, Koike H et al. Prognostic signifi-
cance of plasma placental growth factor levels in renal cell
cancer: an association with clinical characteristics and
199
vascular endothelial growth factor levels. Anticancer Res
2003;23(6D):4953–8.
46. Larcher F, Franco M, Bolontrade M et al. Modulation of
the angiogenesis response through Ha-ras control, pla-
centa growth factor, and angioprotein expression in
mouse skin carcinogenesis. Mol Carcinog 2003;37(2):83–
90.