Gene Expression Profiling of Oral Squamous Cell Carcinoma by Differential Display RT-PCR and Identification of Tumor Biomarkers

Abstract

Oral squamous cell carcinoma (OSCC) is the sixth most common cancer worldwide. Despite progress in therapeutic and surgical treatments, its survival period at 5 years is the lowest among major cancers, and remains unchanged in the last two decades. The growing epidemiological relevance of oral cancer emphasizes the need to better understand the molecular mechanisms underlying this disease and identify predictive tumor markers and therapeutic targets. To this end, we have used the DDRT-PCR analysis to profile the oral tumor transcriptome and identify differentially regulated genes that may be used as potential biomarkers and therapeutic targets. Our DDRT-PCR analysis identified 51 differentially expressed fragments, of which 25 were revalidated by reverse Northern analysis. Northern blot analysis further corroborated these findings for a few genes. In order to ascertain the utility of some of the identified genes as molecular markers and therapeutic targets, semi-quantitative RT-PCR analysis was carried out in a panel of matched oral normal and tumor samples, that confirmed GLTP, PCNA, RBM28, C17orf75 and DIAPH1 as significantly upregulated, whereas TNKS2, PAM and TUBB2C showed significant downregulation in tumor samples. Taken together, our DDRT-PCR analysis has revealed several genes, belonging to diverse cellular pathways, that have been associated with OSCC for the first time. Thus, these genes could be investigated as biomarkers and therapeutic targets for OSCC.

Introduction

Oral squamous cell carcinoma (OSCC) or oral cancer is the sixth most common human malignancy and a major cause of cancer related mortality and morbidity worldwide [1]. In India, it is the leading cause of cancer in males and the third most common cancer in females with tobacco, alcohol and viruses, implicated as the major etiological factors [2–4]. Despite rapid advances in therapy, its 5-year survival rate is the lowest among all major cancers [5, 6]. Thus, finding a biological tumor marker(s) that will increase an early diagnosis and treatment monitoring rates assumes considerable importance [6]. The lack of success in effectively treating oral cancer is primarily due to a gap in the understanding of its etiology and a lack of drugable targets [7]. Addressing this issue requires the large-scale analysis of gene expression profiles. DDRT-PCR (differential display reverse transcription-polymerase chain reaction) is an important tool in this regard, as it can be used to randomly sample the transcriptome, and is not restricted to a pre-defined set of genes as in the case of microarrays [8].

To identify cellular genes that could potentially serve as predictive molecular markers for oral cancer, Chang et al. [9] used DDRT-PCR analysis and identified seven genes as differentially expressed. Other studies employing DDRT-PCR have also identified novel transcripts in oral tumors [10, 11]. Investigations by Arora et al. [12] identified genes belonging to diverse functional groups and signaling pathways such as RABGAP1L (KIAA0471), YWHAZ (14-3-3-zeta), SPA17 (SP17) and RRAS2 (TC21) by DDRT-PCR and reverse Northern analysis. Application of high throughput screening technologies and construction of oral cancer specific cDNA libraries have provided evidence of the existence of a large repertoire of genes that may be dysregulated in oral cancer and contribute to its progression [13, 14]. The identification and evaluation of such genes may thus conceivably foster the discovery of molecules that are directly involved in tumor development [14].

In this study, we have employed DDRT-PCR analysis to identify those genes and pathways that could have roles in initiation, development or progression of oral cancer. Using this technique, we have identified several genes that exhibit striking dysregulation in oral cancers for the first time and may thus provide new biomarkers and therapeutic targets.

Methods

Sample Collection

A total of 16 OSCC (oral cancer) samples were ascertained at Bangalore Institute of Oncology, Bangalore and R.L. Jalappa Hospital and Research Center, Kolar [15]. All tumor samples were from the tongue and cheek areas of the mouth. This study was performed with informed consent from the patients and approval from the ethics committees of the Bangalore Institute of Oncology and Indian Institute of Science. The specimens were obtained as biopsy or surgical samples from oral cancerous lesions and adjacent normal mucosa (taken from the farthest margin of the surgical resection). No treatment has been given at the time of biopsy/surgery [15]. Tumors were classified according to TNM (Tumor, Node and Metastasis) criteria [16].

DDRT-PCR Analysis

DDRT-PCR analysis was performed according to Liang et al. [17] with a few modifications. Five arbitrary (AP1: 5′-TACAACGAGG-3′, AP2: 5′-TGGATTGGTC-3′, AP3: 5′- CTTTCTACCC-3′, AP4: 5′-TTTTGGGCTCC-3′ and AP5:5′-GGAACCAATC-3′) and three two-base anchored primers (T₁₁AC, T₁₁GT and T₁₁CG) were used in this study [18]. Five hundred nanograms of DNAse I treated total RNA was converted into 1st-strand cDNA with one of the anchored primers in a 20 μl reaction volume and the Revertaid™ H Minus First Strand cDNA Synthesis Kit (MBI Fermentas, Burlington, ON, Canada) according to the manufacturer’s instructions. PCR amplification of each 1st-strand cDNA product was carried out in combination with one of the five arbitrary primers. Amplification was carried out in a 25 μl reaction volume containing 2 μl of the 1st-strand cDNA product as a template, 60 ng each of the arbitrary and anchored primers, 50 μmol of each dNTP, 0. 25 μl α-³²P dCTP (3,000 ci/mmole; NEN, USA) and 1 unit of Taq DNA polymerase (Bangalore Genei®, Bangalore, India) in a standard 1× buffer supplied by the manufacturer. Amplification was performed in a PTC100™ Programmable Thermal Controller (MJ Research® Inc, Waltham, MA) under the following conditions: 94°C for 30 sec, 42°C for 2 min, 72°C for 30 sec for 40 cycles and finally 72°C for 5 min. Aliquots of PCR products were run on a 6% polyacrylamide gel with 8 M urea at 1,700 V using the Hoefer™ SQ3 Sequencer system (Amersham Pharmacia Biotech, San Francisco, CA). The gel was dried and bands were visualized by X-ray film autoradiography. Different combinations of anchored and arbitrary primers were used in separate reactions. The bands that showed consistent and differential expression were excised from the gel, eluted in distilled water and re-amplified with the same pair of primers used in the initial reaction. DNA fragments were either purified by gel extraction using the Genelute^TM Gel Extraction Kit (Sigma-Aldrich, St. Louis, MO) or cloned directly into a T/A cloning vector using the InsT/Aclone^TM PCR Product Cloning Kit (MBI Fermentas, Burlington, ON, Canada). Plasmid DNAs were isolated using a standard alkaline lysis method and were checked for the right sized inserts by restriction enzyme digestion and comparing with the PCR products used initially for cloning.

Reverse Northern Blot Analysis

In order to screen for the cDNA fragments (T/A clones) that were truly differential, reverse Northern analysis was carried out in accordance with Zhang et al. [19] with a few modifications. Plasmids were isolated from all the clones that were identified as differentially expressed by DDRT-PCR. Five hundred ng of each plasmid was denatured in 0.4 M NaOH at 100°C for 5 min, snap chilled on ice and spotted in duplicates on two replicas of the N+ Biodyne nylon membrane (LifeTechnologies, Gaithersburg, MD) using a 96-well dot-blot manifold (Bio-Rad, Hercules, CA). Nylon membranes were neutralized by 1 M Tris-HCl pH 8.0, rinsed with 6xSSC (Sodium Saline Citrate: 3 M sodium chloride, 0.3 M sodium citrate, pH 7.0) and treated with a UV cross linker (Stratagene, La Jolla, CA). cDNA probes for RNA samples from normal and tumor tissues were prepared separately using 10 μg total RNA by reverse transcription in a 40 μl reaction that consisted of 50 mM Tris-HCl pH 8.3, 50 mM KCl, 4 mM MgCl₂, 10 mM DTT, 500 μM each of dTTP, dATP and dGTP, 0.5 μg oligo (dT)₁₈ primer and 50 μCi α-³²P dCTP (3,000 Ci/mmol; NEN, USA). After 5 min incubation at 70°C, samples were shifted to 37°C and 1,000 U of MMLV reverse transcriptase (MBI Fermentas, Burlington, ON, Canada) was added, followed by continued incubation at 42°C for 1 h. RNA was then hydrolysed by adding equal volume of 0.6 N NaOH and further incubated at 70°C for 30 min. After reverse transcription, the QIAquick^R Nucleotide Removal kit (Qiagen, Hilden, Germany) was used to remove unincorporated radionucleotide α-³²P dCTP according to the manufacturer’s instructions. Equal counts (5–10 × 10⁶ c.p.m) of cDNA probes, made from total RNA samples from either the normal or tumor oral tissues, were heat-denatured separately and used to probe duplicate membranes. Membranes were hybridized with either of the labeled probes for 14–16 h in 6xSSC, 0.5% SDS and 5× Denhardt’s reagent. Both membranes were then given stringent washes in 5xSSC, 0.5% SDS (3 × 15 min) and 0.1xSSC, 0.5% SDS (3 × 15 min). The membranes were wrapped in plastic sheets and exposed to BAS 2040 cassettes and scanned on a FLA 2000 Phosphor Image System (Fujifilm, Tokyo, Japan). Signal intensities of the clones were quantified using the Kodak Digital Science Image Station imaging software, version 3.6.1. (Kodak, Rochester, NY). Signal intensities were normalized against GAPDH and β-actin that remained unchanged in expression in normal and tumor tissues. The empty T/A vector pTZ57R and human genomic DNA samples were also used as negative controls. A >1.8 fold differential cut-off was used to designate the differential expression [12].

Northern Analysis

Twenty micrograms of RNA samples from both normal and tumor tissues were size fractionated on a 1% agarose formaldehyde denaturing gel and subsequently transferred to a nylon N+ membrane (Biodyne LifeTechnologies, Gaithersburg, MD). α-³²P labeled probes were generated for the cDNA of interest using the RadPrime labeling system (Promega, Madison, WI) according to the manufacturer’s instructions. Northern hybridization was performed according to Sambrook and Russell [20]. The membrane was subsequently stripped off and probed with β-actin to check for equal loading. Radiolabelled β-actin probe was prepared as described in Sambrook and Russell [20].

DNA Sequencing and Computational Analysis

Plasmid clones, found to be true differentials by reverse Northern screening, were expanded in liquid culture and plasmid DNAs were extracted using the Genelute^TM Plasmid miniprep columns (Sigma-Aldrich, St. Louis, MO). These were sequenced on an ABIprism® A377-AZXautomated sequencer (Applied Biosystems, Foster City, CA). Nucleotide sequences of the inserts were compared with the NCBI GenBank database using BLASTN searches (http://www.ncbi.nlm.nih.gov).

Semi-Quantitative RT-PCR

Total RNA was isolated from 16 paired normal and tumor samples using the TRI REAGENT^TM (Sigma-Aldrich, St. Louis, MO). Semi-quantitative RT-PCR and quantification of signal intensities were performed as described earlier [15]. The details of primers and their specific conditions for amplification are available from the authors on request. A fold difference ≥1.8 was used as a cutoff to determine whether a gene showed upregulation or downregulation in a particular sample. The significance of difference in mRNA levels between normal and tumor samples for a gene was assessed by Student’s t-test and the results are expressed as mean±SEM [21]. A probability value of p < 0.05 was considered to be significant. PCR amplification for each gene was repeated once.

Results

DDRT-PCR Analysis of Differential Gene Expression in Oral Tumors

DDRT-PCR analyses of normal and tumor tissues were carried out using 15 primer combinations (three anchored and five arbitrary primers). Each primer combination yielded around 60–65 bands and a total of approximately 975 different cDNAs were screened in this study. Of these, 51 fragments that showed marked difference in expression between the normal and tumor sample were chosen for further analysis. These ranged in size from 60 to 800 bp (data not shown). A representative differential display profile is shown in Fig. 1.

Fig. 1.

A representative differential display profile from a single patient using normal (N) and tumor (T) samples. T₁₁AC was used as the anchored primer. AP1, AP2 and AP3 were used as the arbitrary primers. The arrows point to bands showing differential expression pattern between the normal and tumor sample

Validation of DDRT-PCR Data by Reverse Northern, Northern and Sequence Analyses of the Fragments

The truly differential nature of the fragments identified by DDRT-PCR was re-confirmed by reverse Northern analysis. The signal intensities of the various clones were normalized against GAPDH and β-actin that remained unchanged in expression in normal and tumor tissues (Fig. 2). Empty T/A vector clones and genomic DNA samples were also used as controls (data not shown). Following two rounds of reverse Northern screening, 25 clones were confirmed as differentially expressed (Fig. 2). Of them, 12 were upregulated and 13 were downregulated in tumor tissue (Fig. 2). All the clones were sequenced and their identity was established by computer searches against the GenBank database. Fourteen clones did not show homology with any known gene in the database and therefore may represent non-specific genomic DNA sequences or genes that have not yet been identified (Table 1). The remaining 11 clones showed homology to known genes such as DIAPH1, C17orf75 (NJMU-R1), RBM28, PCNA, GLTP, MT-ATP6, ZKSCAN1, TNKS2, PAM, TUBB2C and C14orf154 (Table 1). TNKS2, PAM, TUBB2C and C14orf154 showed downregulation and the remaining seven genes were upregulated in oral tumor sample (Table 1). To further validate the results, Northern blot analysis was carried out on matched normal and tumor samples for a few genes. In keeping with the earlier observation, PCNA, C17orf75 and ZKSCAN1 showed upregulation, while TUBB2C showed downregulation in the tumor sample (Fig. 3a).

Fig. 2.

Clones identified as true differentials after second round of reverse Northern screening. Twelve clones were found to be upregulatedand and 13 were downregulated in tumor. ß-actin and GAPDH were used as equal loading controls. As expected, no change was observed in signal intensities for ß-actin and GAPDH spots probed with cDNA probes from normal or tumor tissues. ß-actin and GAPDH spots contain RT-PCR products. A >1.8 fold differential cut-off was used to designate the differential expression. Numbers in parentheses indicate the fold difference in expression in tumor in comparison to normal tissue

Table 1.

Details of differentially expressed genes identified by DDRT-PCR and reverse Northern analyses

Serial no.	Clone ID	Gene/EST/Clone	Fragment size (bp)	Chromosome location	Unigene/GenBank accession no.	Function	Expression in oral tumors
1	DD39C	Diaphanous homolog 1 (Drosophila) (DIAPH1)	375	5q31	Hs.529451	Regulation of actin polymerization	UP
2	DD42A	Chromosome 17 open reading frame 75 (C17orf75)/NJMU-R1*	313	17q11.2	Hs.655257	Molecular and cellular function unknown	UP
3	DD28A	RNA binding motif protein 28 (RBM28)	236	7q32.2	Hs.274263	Unknown	UP
4	DD5A	Proliferating cell nuclear antigen (PCNA)	320	20pter-p12	Hs.147433	DNA replication, repair, cell proliferation, regulation of cell cycle	UP
5	DD5B	Glycolipid transfer protein (GLTP)	274	12q24.12	Hs.381256	Transfer of certain glycosphingolipids and glyceroglycolipids	UP
6	DD43A	ATP synthase 6 (MT-ATP6)	212	mtDNA	AF271776	Involved in oxidative phosphorylation, acts as a receptor for angiostatin	UP
7	DD9A1	Zinc finger with KRAB and SCAN domains 1 (ZKSCAN1)	206	7q21.3-q22.1	Hs.615360	DNA dependent transcription	UP
8	DD31F	Tankyrase, TRF1-interacting ankyrin-related ADP-ribose polymerase 2 (TNKS2)	350	10q23.3	Hs.329327	TRF1interacting, overexpression causes rapid cell death.	DOWN
9	DD2B	Myc binding protein 2 (MYCBP2)/PAM	225	13q22	Hs.591221	Binds to the transactivation domain of c-Myc	DOWN
10	DD9A3	Tubulin beta 2C (TUBB2C)	172	9q34	Hs.433615	GTPbinding, GTPase activity, cytoskeleton.	DOWN
11	DD33B	SET domain containing 3 (SETD3)/C14orf154*	372	14q32.2	Hs.510407	Unknown	DOWN
12	DD 27A	Clone RP11-30L4	374	9q22.1-q22.33	AL353629.22	Unknown	UP
13	DD19D	Clone B76A21	233	21q21.1-q21.2	AP000402.2	Unknown	UP
14	DD37B	Clone RP11-576F1	326	2	AC008179.2	Unknown	UP
15	DD13A1	PAC clone I2280	665	21q22.1	AF165147.1	Unknown	UP
16	DD 32C	Clone RP3-440O21	238	X	Z84481.1	Unknown	UP
17	DD24A2	cDNA clone MDSDJA06	498	10	AV762652.1	Unknown	DOWN
18	DD 1C	Clone RP11-171A24	228	9	AL355674.10	Unknown	DOWN
19	DD21B	Clone RP11-413G15	645	1	AL360000.8	Unknown	DOWN
20	DD 41B	Clone RP11-68D16	768	4	AC092593.2	Unknown	DOWN
21	DD 18D	Clone RP11-171A24	366	9	AL355674.10	Unknown	DOWN
22	DD38C	Clone RP11-68D16	768	4	AC092593.2	Unknown	DOWN
23	DD 33C	Clone RP5-1022E24	440	20	AL121581.41	Unknown	DOWN
24	DD10D	Clone RP11-626B22	574	5	AC113425.2	Unknown	DOWN
25	DD15B	Clone RP11-328J14	236	16	AC009075.7	Unknown	DOWN

UP upregulated in oral tumors; DOWN downregulated in oral tumors; and mtDNA mitochondrial DNA

Fig. 3.

Expression profiling of differentially expressed genes. a Northern blot hybridizations showing upregulation of PCNA, C17orf75 and ZKSCAN1, and downregulation of TUBB2C in the tumor sample. ß-actin was used as a control to equalize RNA loading. b Semi-quantitative RT-PCR analysis of genes identified as differentially expressed by DDRT-PCR analysis in 16 matched normal and tumor samples. Squares and triangles represent relative expression values for normal and tumor samples respectively, adjusted according to the internal control GAPDH. Each square or triangle corresponds to data from one sample. Horizontal lines represent mean values of mRNA expression across normal or tumor samples

Expression Profiling of the Differentially Expressed Genes

We further validated the expression of eight differentially expressed genes (viz., GLTP, PCNA, RBM28, C17orf75, DIAPH1, TUBB2C, PAM and TNKS2) in a panel of 16 matched normal and tumor samples by semi-quantitative RT-PCR analysis (Fig. 3b; Table 2). The mean mRNA expression levels in normal vs. tumor samples were as follows: GLTP (0.75 ± 0.11 vs. 1.351 ± 0.20), PCNA (0.41 ± 0.01 vs. 0.77 ± 0.10), RBM28 (0.35 ± 0.07 vs. 0.57 ± 0.08), C17orf75 (0.59 ± 0.13 vs. 1.24 ± 0.02) and DIAPH1 (0.24 ± 0.34 vs 0.41 ± 0.07), which were significantly greater in tumor samples than in matched normal samples (Fig. 3b). For the genes that showed downregulation, the mean expression levels in normal vs. tumor samples were as follows: TUBB2C (1.79 ± 0.21 vs. 1.23 ± 0.17), PAM (0.85 ± 0.14 vs. 0.44 ± 0.09) and TNKS2 (0.78 ± 0.10 vs. 0.48 ± 0.07) (Fig. 3b).

Table 2.

Clinicopathological characteristics and gene expression variation in folds for analyzed patient samples

Sample (Patient No.)	Age/Sex	TNM Classification	Metastasis	Histology	Treatment	Tobacco use	DIAPH1 (U)	C17orf75 (U)	RBM28 (U)	PCNA (U)	GLTP (U)	TNKS2 (D)	PAM (D)	TUBB2C (D)
54	40 M	T4N1M0	N/A	WDSCC	Surgery	Yes	7.17	2.16	1.8	3.31	2.36	2.24	14.04	1.30
39	55 F	T4N1M0	Yes	PDSCC	Surgery/RT	Yes	5.81	2.10	2.19	1.89	2.58	1.62	1.59	1.44
19	48 F	T4N2bM0	No	WDSCC	Surgery/RT	Yes	1.62	3.41	1.88	2.07	1.85	1.14	2.82	1.48
15	55 F	T3N1M0	No	WDSCC	Surgery/RT	Yes	0.84	1.33	1.34	3.79	2.27	1.74	1.85	1.01
20	50 F	T4N1M0	Yes	MDSCC	Surgery/RT	Yes	2.35	1.39	0.84	1.58	1.17	7.53	2.92	1.27
8	50 F	T3N1M0	No	WDSCC	Surgery/RT	Yes	0.35	2.50	1.80	0.87	2.24	1.59	1.14	2.07
11	60 F	T4N1Mx	Yes	MDSCC	Surgery/RT	Yes	1.07	2.18	1.75	1.35	1.63	1.60	1.65	1.03
40	50 F	T2N1M0	Yes	WDSCC	Surgery/RT	Yes	1.36	2.79	2.27	10.91	1.03	0.99	2.68	2.23
50	40 M	T4N2bM0	Yes	WDSCC	Surgery/RT	Yes	1.03	4.69	1.71	1.06	1.30	2.00	1.96	3.46
52	70 M	ED	N/A	N/A	No RT	No	1.57	1.55	1.44	2.24	2.17	3.77	0.54	2.30
32	35 M	T4N0Mx	No	WDSCC	Surgery/RT	Yes	1.92	3.04	12.19	2.04	0.96	1.60	2.14	1.72
53	38 F	T2N0M0	No	WDSCC	Surgery/RT	Yes	1.11	3.71	8.18	1.61	2.04	1.54	0.35	2.20
55	40 F	T4N1M0	Yes	WDSCC	Surgery/RT	Yes	1.96	2.74	1.51	2.75	2.62	0.94	19.34	1.51
57	62 F	T1N0M0	Yes	WDSCC	Surgery/RT	Yes	3.77	3.57	1.89	2.61	1.97	1.50	3.39	0.55
59	70 F	T4N2bM0	Yes	WDSCC	Surgery/RT	Yes	1.72	1.69	0.71	1.25	2.04	0.44	2.09	1.48
60	40 F	T4aN1Mx	Yes	PDSCC	Surgery/RT	Yes	1.49	1.22	2.99	1.08	1.42	0.81	0.64	1.59

M male; F female; TNM tumor, node, metastasis; ED epithelial dysplasia; N/A not known; WDSCC well-differentiated squamous cell carcinoma; PDSCC poorly-differentiated squamous cell carcinoma; MDSCC mildly-differentiated squamous cell carcinomas; and, RT, radiation therapy.‘Yes’ in tobacco use refers to addiction to tobacco, bidi and cigarettes for at least 15–20 years. (U) denotes genes whose mean expression is significantly upregulated and (D) denotes genes whose mean expression is significantly downregulated across 16 tumor samples

Discussion

Our present approach using DDRT-PCR analysis has identified a total of 25 differentially expressed genes (cDNA fragments) in oral cancer. However, except PCNA to the best of our knowledge, the 24 other genes identified in the present study have not been associated with oral cancers earlier [22]. This could be due to the use of different primer combinations for DDRT-PCR, that is key to the population of mRNAs that can be screened, and also restriction of microarrays to a defined set of genes. As DDRT-PCR is a technique fraught with a lot of false positives, the truly differential nature of the clones was re-validated by reverse Northern analysis [19]. In the present study, there was a consistent correlation between the results obtained from DDRT-PCR, reverse Northern, Northern and semi-quantitative RT-PCR analyses.

The C17orf75 (NJMU-RI) gene was found to be upregulated in an oral tumor sample by Northern analysis as well as in 11/16 tumor samples by semi-quantitative RT-PCR. There is no literature available on the molecular biology and functional relevance of this gene. However, it has been shown to be significantly upregulated in retinoblastoma [23].

RBM28 is a common nucleolar component of the spliceosomal ribonucleoprotein complexes, possibly coordinating their transition through the nucleolus [24]. It specifically associates with U1, U2, U4, U5, and U6 snRNAs [24]. There is no report on the involvement of this gene in any form of cancer and our study has shown the upregulation of this gene in 9/16 tumors.

Glycolipid transfer protein (GLTP) that was found to be upregulated in 10/16 tumor samples is a small (23–24 kDa) cytosolic protein that accelerates the intermembrane transfer of various glycosphingolipids (GSLs) [25]. In cancer cells, expression of certain GSLs has been associated with multidrug resistance [26]. GLTPs are potential regulators of cell processes mediated by the GSLs [27].

The ZKSCAN1 cDNA predicts a 325 amino acid protein belonging to the Krüppel family of zinc finger proteins [28]. These proteins often carry a potent repressor domain called the Krüppel Associated Box (KRAB), which is known to effectively repress transcription through interaction with transcriptional intermediary factor 1 beta (TIF1beta) [28]. Recently, array CGH analyses have shown that ZKSCAN1 is overexpressed in adenocarcinomas of the gastroesophageal junction [29]. Our study using Northern analysis has shown that the ZKSCAN1 gene is upregulated in the tumor sample.

Anti-mitotic chemotherapeutic agents target tubulin, a major protein in mitotic spindles. Tubulin isotype composition is diagnostic of tumor progression and a determinant of the cellular response to chemotherapy [30]. Mammalian microtubules are formed from a mixture of α-tubulin and β-tubulin isotype classes and the antimitotics used in chemotherapy interact primarily with β-tubulin [31, 32]. Studies with paclitaxel or other antimitotic agents, such as colchicine or estramustine, indicate that drug interactions with tubulin isotypes differ and might contribute to cell resistance to antimitotics [29, 33, 34]. Our study has identified the differential regulation of one isotype of β-tubulin, TUBB2C. The functional significance of this isotype is unknown, however, we found it to be downregulated in 5/16 tumors and in a tumor sample by Northern analysis.

PCNA, which was first identified as a co-factor of DNA polymerase delta, is an important component of cellular DNA replication process as it acts as a loading clamp for the replication machinery [35]. It has been found to be upregulated in several cancers and regarded as a marker for proliferating cells [36]. Our results corroborate this, as PCNA shows an upregulation in 9/16 tumor samples at the RNA level and also 8/8 tumor samples at the protein level (data not shown).

Another gene that showed significant downregulation in our analysis was Tankyrase 2 (TNKS2), a newly identified member of the poly-(ADP-ribose) polymerase (PARP) family of proteins. It is a multifunctional protein, localizing to several subcellular sites [37]. Recent reports indicate that overexpression of TNKS2 causes a rapid induction of necrotic cell death [38]. Tankyrase 2 overexpression thus seems to be lethal to proliferating cells and this is in keeping with our findings that TNKS2 is significantly downregulated in oral tumors.

The gene Diaphanous homolog 1 (DIAPH1 or mDia1) was upregulated in 6/16 tumors and is the mammalian homolog of the Drosophila Diaphanous [39]. This gene plays a very important role in cytoskeletal remodeling [40]. Regulation of DIAPH1 has been shown to be linked to increased proliferation and invasiveness of melanoma cells [41].

A very interesting finding of this study is PAM (Protein associated with c-Myc) that was found to be downregulated in 10/16 tumors (Table 2). PAM, first identified as a binding partner for c-Myc, encodes a large protein with several motifs and participates in diverse cellular functions [42]. It has been shown to regulate cAMP mediated signaling that controls various cellular responses [43]. Through its ring zinc-finger and zinc-finger motifs, PAM associates with tuberin, the protein product of the tumor suppressor TSC2 gene [44]. This is particularly interesting since we have previously shown that the TSC genes are potential tumor suppressors in oral cancer [15]. Further, PAM has also recently been shown to be an important regulator of the mTOR signaling cascade, which we have demonstrated to being dysregulated in oral cancer [15, 45].

Taken together, our DDRT-PCR analysis has identified several genes that are being associated with oral cancers for the first time. Some of the identified genes have been further validated as potential biomarkers. These findings also emphasize the fact that although several large-scale gene expression analyses have been carried out for oral cancer, a vast repertoire of genes involved in this pathogenesis, still remain undiscovered. Functional relevance of some of the genes identified in the present study is not yet known and their further characterization is warranted to evaluate their potential as molecular markers and/or novel therapeutic targets.