Proteomic Study of Esophageal Squamous Cell Carcinoma

Comprehensive profiling of genome and transcriptome has identified myriads of alternations at the level of gene and gene expression, which drive malignant development and progression in context of oncology. As a result, qualitative or quantitative changes of protein expression pattern will inevitably ensue during multi-stage of carcinogenesis. In this sense, the proteome is a functional translation of the genome and is the actual manipulator of cellular behavior. Therefore, proteomic profiling of cellular protein constituents should generate the most relevant marker of the functional state of a cell. On the other hand, lack of correlation between mRNA and protein expression have been documented for a variety of genes. Unlike the genome which is static in certain sense, the proteome of a cell is dynamic and changes over time in terms of protein pattern, protein interactions and modifications triggered by external or internal signals[Kolch et al., 2004; Kolch et al., 2005]. Only dynamic information flow through protein circuitry reflects the course of a disease and allows us to track the pathogenetic mechanisms as well as treatment response[Kolch et al., 2005]. Furthermore, examining DNA sequences and measuring mRNA expression do not specify splicing, post-translational modifications, cleavages, protein subcellular localization and complex formations[Banks et al., 2000; Chambers et al., 2000]. There exists a huge information gulf between RNA transcription and protein expression. Proteome represents a much richer source for the functional description of diseases and the biomarker discovery implicated in cancer. Moreover, most of diagnostic assays currently applied in clinical practice are protein-based immunological methods, which are well adapted to standardization and clinical implementation. Proteomic profiling during disease formation and evolution not only provides an integrated understanding of pathogenesis in context of genome and proteome but also holds greater promise to identify the biomarkers of diagnosis and therapeutic targets for diseases such as cancer.


Introduction
Comprehensive profiling of genome and transcriptome has identified myriads of alternations at the level of gene and gene expression, which drive malignant development and progression in context of oncology.As a result, qualitative or quantitative changes of protein expression pattern will inevitably ensue during multi-stage of carcinogenesis.In this sense, the proteome is a functional translation of the genome and is the actual manipulator of cellular behavior.Therefore, proteomic profiling of cellular protein constituents should generate the most relevant marker of the functional state of a cell.On the other hand, lack of correlation between mRNA and protein expression have been documented for a variety of genes.Unlike the genome which is static in certain sense, the proteome of a cell is dynamic and changes over time in terms of protein pattern, protein interactions and modifications triggered by external or internal signals [Kolch et al., 2004;Kolch et al., 2005].Only dynamic information flow through protein circuitry reflects the course of a disease and allows us to track the pathogenetic mechanisms as well as treatment response [Kolch et al., 2005].Furthermore, examining DNA sequences and measuring mRNA expression do not specify splicing, post-translational modifications, cleavages, protein subcellular localization and complex formations [Banks et al., 2000;Chambers et al., 2000].There exists a huge information gulf between RNA transcription and protein expression.Proteome represents a much richer source for the functional description of diseases and the biomarker discovery implicated in cancer.Moreover, most of diagnostic assays currently applied in clinical practice are protein-based immunological methods, which are well adapted to standardization and clinical implementation.Proteomic profiling during disease formation and evolution not only provides an integrated understanding of pathogenesis in context of genome and proteome but also holds greater promise to identify the biomarkers of diagnosis and therapeutic targets for diseases such as cancer.
with diverse functions, such as vulnerable genes to chemicals, tumor-related genes, tumor suppressor genes, metastasis genes, apoptosis gene, proliferation genes, etc [Enzinger & Mayer, 2003;Greenawalt et al., 2007;Kwong, 2005;Lin et al., 2009].Moreover, epigenetic alterations, chromosomal changes and transcriptional changes have also been found to play crucial roles in the pathogenesis of ESCC [Abnet et al., 2010;Greenawalt et al., 2007;Wang et al., 2010].Although these findings improve our general understanding about the molecular biology of ESCC, the appropriate biomarkers for high-risk population screening, for clinical diagnosis and prognosis, for evaluation of treatment efficiency have not been identified yet.Therefore, it is imperative to search more effective biomarkers for such purposes.

ESCC analysis by proteomics 2.1 Advantages of proteomics compared with genomics
The completion of human genome sequence did not ensure panacea solutions to all problems related to biological deregulation.In fact, human proteome is far more complex and dynamic than genome sequence.It is estimated that the human genome contains about 32 000 protein coding genes, which code for 100 000 to 10 million proteins due to alternative RNA splicing, overlapping of transcription units, post-translational processing and modifications [Lander et al., 2001;Venter et al., 2001].Thus, a big disparity between genome and proteome exists, which indicates that the combinatorial diversification of regulatory networks lead to functional evolution of proteins.Through detecting the functioning units, proteomic studies generate a protein fingerprint, which reflects both the intrinsic genetic programme of the cell and the impact of its immediate environment.Therefore, proteomics is valuable for biomarker discovery since its application provides higher opportunity to identify genuine determinants or causal factors involved in biological functions or the pathogenesis of disease.

Two-dimensional electrophoresis-based proteomic findings of ESCC
Two-dimensional electrophoresis (2DE) has been used for over 30 years now due to its high resolution for the separation of complex protein mixtures.In combination with mass spectrometry, 2DE has been so far the most commonly used method for analyzing protein expression and identity.Our laboratory used 2DE to profile the proteome from ESCC tumors and matched adjacent non-cancer mucosa, and proteome from immortalized esophageal cell line and cancer cell lines.Comparative analysis and MS for protein identification showed that the over-expressions of four proteins were common in ESCC tissues and cancer cell lines, which include tropomyosin isoform 4 (TPM4), prohibitin, peroxiredoxin (PRX1) and manganese superoxide dismutase (MnSOD); the expressions of another three proteins, i.e. stratifin, prohibitin, squamous cell carcinoma antigen 1 (SCCA1), were correlated inversely with dedifferentiation of ESCC [Qi et al., 2005;Qi et al., 2008].Immunohistochemistry (IHC) analysis showed that loss of expressions of annexin A2 and stratifin were 45% and 64% in ESCC, respectively [Qi et al., 2007a;Qi et al., 2007b;Ren et al., 2010].Differential expressions of ten proteins including TPM1, SCCA1, stratifin, peroxiredoxin 2 isoform a, alpha B-crystalline, annexin A2, heterogeneous nuclear ribonucleoprotein L (hnRNP L), triosephosphate isomerase1 (TPI), laminA/C, and cyclophilin A (CypA) can be observed as well.Our findings may suggest that these differential proteins contribute to the multistage process of carcinogenesis, tumor progression, and invasiveness of ESCC.Published in the same issue, Zhou et al found 28 www.intechopen.comProteomics -Human Diseases and Protein Functions 260 proteins aberrantly expressed in ESCC cancer cells with at least three-fold difference between ESCC and normal epithelial cells [Zhou et al., 2005].The overlap between these two studies was quite small.Only expression of SCCA1 was commonly down-expressed in ESCC, but transgelin showed increased expression in tumor in our study and decreased expression in Zhou's study.The disparity of proteins identified between these two studies may be due to different sample source, different methods used by these two groups, such as laser capture microdissection vs. bulk tissues, 2D-DIGE vs. silver staining.Later, five groups reported proteomic signatures associated with ESCC using ESCC samples collected from different regions of China, including high risk areas for ESCC such as Linzhou, Xinjiang and low risk areas like Beijing and Guangdong, but only four reports displayed details of identified proteins.Interestingly, more overlap of the identified proteins came from Fu's study and ours, both of which used ESCC samples from Linzhou, one of the highest areas for ESCC adjacent to Taihang Mountain [Fu et al., 2007].The commonly identified proteins with the same change direction included alpha enolase, TPM, tubulin, prohibitin and PRX2.Although the prevalence of ESCC in Xinjiang is comparable to Linzhou, the protein signatures were unique to sample origin, indicative of more important roles of environmental, ethnic or hereditary factors in the carcinogenesis of ESCC [Liu et al., 2011].It seems that hsp27 was a general molecular events involved in ESCC since four out of five studies observed down-expression in ESCC except ours.Only one among seven studies performed survival assay after identifying the candidate proteins by ESCC proteomic profiling.Du et al. reported that over-expression of calreticulin and GRP78 could predicate poor prognosis of ESCC [Du et al., 2007].Although 2DE is indeed a very useful method for biomarker discovery, more examinations of the biological functions and the clinical relevance of biomarker candidates involved in ESCC are necessary to verify its clinical value.Two reports described the proteomic signatures of ESCC with samples from Japan.Nishimori et al. used the agarose IEF gel in the first dimension, which not only allows for large-scale quantitative comparisons of protein expression but also is able to resolve high molecular mass proteins larger than 150 kDa [Nishimori et al., 2006].As a result, a different protein pattern was revealed, including a few protein candidates with MW > 70 kDa.Western blot and IHC verified the different expression of a 195 kDa protein, periplakin, between cancer and adjacent non-cancer tissues.Not only was the expression of periplakin significantly down-regulated in ESCC but also translocation of periplakin was observed, which localized at cell-cell boundaries in normal epithelium and dysplastic precursor lesions, and disappeared from cell boundaries and shifted to cell cytoplasm in early cancers.The other research group from Japan used unsupervised classification to analyze the 2D-DIGE protein spots and procured the protein signatures most relevant to clinical parameters with progression of ESCC [Hatakeyama et al., 2006].The authors developed the largest protein database relevant to ESCC, which identified 240 proteins with expression level associated with carcinogenesis, histological differentiation and the number of lymph node metastases.A significant overlapping was observed between the proteins identified in ESCC with other different types of tumor.In addition, Jazii et al did proteomic profiling using ESCC samples from Iran, another high incidence area for ESCC like northern China, and identified six over-expressed proteins and six under-expressed proteins associated with ESCC [Jazii et al., 2006].However, the authors only used RT-PCR to verify the loss oftropomyosin in ESCC.The functions of identified proteins associated with the development and progression of ESCC include cytoskeletal/structural organization, transport, chaperon, oxioreduction, proliferation, glycolysis, cell motility, transcription, signal transduction, suggesting multiple dysregulated pathways involved in ESCC.For better understanding the pathogenesis of ESCC and development of biomarkers, integrated and comprehensive studies on these protein candidates are needed.An alternative approach to identify novel tumor biomarkers is the assessment of immune response elicited by tumor antigen since the humoral immune response to cancer in humans has been evidenced by the identification of autoantibodies to a variety of intracellular and surface antigens in cancer patients with different types of tumors [Chen et al., 2007;Disis et al., 1997;Hong et al., 2004;Soussi, 2000].In ESCC, a number of reports have documented the presence of autoantibodies in serum against various proteins, including p53, cytokeratins, myomegalin, TRIM21, peroxiredoxin VI proteins, Hsp70, and CDC25B [Bergqvist et al., 2001;Fujita et al., 2006;Fujita et al., 2008;Liu et al., 2008;Shimada et al., 2007;Shimada et al., 2005;Veale et al., 1988].The proteomic-based approach to identify panels of tumor antigens and related autoantibodies was introduced by Brichory et al. in 2001, which identified antiannexin I and II antibodies in sera from patients with lung cancer [Brichory et al., 2001].There have been four articles published by two research groups, which reported the existence of autoantibodies in sera of ESCC patients.The first report was published by Fujita et al from Japan, who used 2DE to resolve protein extracts from ESCC cell line TE-2 as tumor antigens and then probed the blot with sera of ESCC patients, healthy controls and patients with other cancers [Fujita et al., 2006].One positive spot was identified as PRX VI by MALDI TOF/TOF MS.The frequency of autoantibody against PRX VI was 50% (15/30) in ESCC, only 6.6% (2/30) in health controls and 3.3% (1/30) in colon cancer.Two years later, the same research group discovered augmented concentration of Hsp70 autoantibody in the serum of ESCC patients, which was significantly higher in ESCC patients than gastric and colon cancer, healthy controls [Fujita et al., 2008].On the other hand, Liu et al. used ESCC tissue protein extracts and autologous sera to search for autoantibodies in ESCC patients and identified autoantibody CDC25B [Liu et al., 2008].Furthermore, CDC25B expression was significantly higher in ESCC tissues with positive autoantibody CDC25B and significantly correlated with tumor stage.The sensitivity and specificity of autoantibody CDC25B for ESCC detection was 56.7% and 91%, respectively [Dong et al., 2010].The autoantibodydriven research is indeed a promising approach for the identification of novel serum biomarkers present in ESCC and for the tumor antigen itself, which may aid the diagnosis of ESCC and development of more effective immunotherapies.Similar to other cancers, development of multiple drug resistance in ESCC is one of major causes of failure to chemotherapy treatment.Furthermore, recent studies have shown that there exists intrinsic sensitivity and resistance to chemotherapy and/or radiotherapy in malignant cells of ESCC, which may predict clinical outcome of ESCC patients receiving neoadjuvant chemotherapy.Prior stratification of ESCC patients according to reliable biomarkers could not only save patients unnecessary adverse effects of chemotherapeutic agents but also render patients more chance to access to alternative curative treatment options.Therefore, it is imperative to define new diagnostic indicators that can reliably predict response to chemotherapy and radiotherapy in advance.A recent study compared the 2DE gels of parental esophageal cancer cell line EC109 and its resistant sub-cell line EC109/CDDP to determine the different proteins spots and identified 44 proteins with potential contribution to chemotherapy resistance [Wen et al., 2010].In another study, radioactive 2DE proteomic comparative analysis was performed using protein extracts of biopsies from 34 patients with locally advanced EAC receiving neoadjuvant chemotherapy.
The identified proteins with different expression between responders and non-responders were classified into two major families, cytoskeleton proteins and molecular chaperon proteins.Further validation by IHC and RT-PCR showed that weak expression of HSP27 at protein level and mRNA level were associated with non-response to platin-based chemotherapy [Langer et al., 2008].As serum represents a rich source for biomarker discovery, proteomic spectra were examined using 27 and 12 serum samples of responders and non-responders, respectively, to preoperative chemoradiotherapy in a training set by surface-enhanced laser desorption and ionization coupled with mass spectrometry analysis.A proteomic classifier comprising four mass peaks, at 7 420, 9 112, 17 123 and 12 867 m/z was identified with 93.3% predicative accuracy in the validation set [Hayashida et al., 2005].Since chemotherapy resistance is a complex and multi-factorial event, proteomic-based studies enable comprehensive characterization of resistance phenotype of malignant cancers, which may lead to identification of potential distinguishing biomarkers between responders and non-responders and lay foundation for further molecular mechanism studies.In addition of 2DE gel for proteomic studies, surface enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF-MS) is an alternative proteomic tool to profile the serum or other body fluids and define potential protein pattern with diagnostic potential.By profiling of the serum proteome with SELDI-TOF-MS combined with bioinformatics tools, a number of highly sensitive and specific potential diagnosis markers have been revealed in various types of cancers.Wang et al. used weak cation exchange (WCX2) protein chips and SELDI-TOF-MS to profile 130 symptom-free serum samples collected from high-incidence area of ESCC in northern China, Linzhou, which included 63 subjects with normal esophageal mucosa, 40 subjects with basal cell hyperplasia, 27 subjects with dysplasia and 30 ESCC patients.Biomarker pattern's software identified four protein features at m/z of 9 306.61, 13 765.9, 2 942.15 and 15 953.4,which could distinguish normal esophageal epithelium, basal cell hyperplasia, dysplasia and ESCC with satisfactory diagnostic accuracy [Wang et al., 2006].Xinjiang is one of the high-incidence areas for ESCC and comprise different ethnic peoples including Han decent.Using CM10 protein chips to capture targets from serum, SELDI-TOF-MS and bioinformatics analysis resulted in identification of six protein peaks (m/z 5667, 5790, 5876, 5979, 6043 and 6102) with diagnostic power with sensitivity and specificity of 91.43% and 88.89%, respectively [Xu et al., 2009].In the case of ESCC profiled by SELDI-TOF-MS, further purification and identification of discriminatory peaks is necessary for development of simple methods for wider clinical application, and to enhance our understanding of the molecular mechanisms of esophageal carcinogenesis as well.

Clinical relevance of potential protein biomarkers in ESCC
To answer clinical questions, the protein biomarkers identified by proteomic techniques with potential diagnosis and therapeutic targets for ESCC need to be translated into clinical scenario, which is realized by using clinical samples, such as biopsy samples, resected tissue samples, plasma or serum samples, urine samples, saliva samples, etc.The methods used for validation generally comprise Western blot, IHC and ELISA at protein level, and RT-PCR at transcription level.Using 2DE-and SILAC-based quantitative proteomic approaches, we have identified a total of 78 non-redundant proteins with aberrant expression associated with ESCC, suggesting that these proteins may play functional roles in carcinogenesis of ESCC and may have clinical values.Afterwards, Western blot analysis verified the decreased expressions of three proteins, i.e.SCCA1, TPM1 and B-Cryst in cancer, in accordance with 2DE quantitative results.At transcription level, SCCA1 mRNA was downregulated in tumor as well.More importantly, the expression of SCCA1 decreased step by step as a function of precancer lesions progression, which suggests that SCCA1 may take part in the multi-stage transformation of ESCC, even in the earliest stages [Qi et al., 2005].In the 2DE-based comparative proteomic study using immortalized and cancer cell model, we Stepwise decrease in annexin A2 protein expression was observed when epithelial cell was transformed malignantly.In poorly-differentiated squamous carcinoma, 46% (5/11) of cancer tissue sample lost annexin A2 protein and 36% (4/11) expressed at weak intensity [Qi et al., 2007b].In a separate study, IHC was used to determine 14-3-3σ in 60 cases of ESCC, nearby matched normal esophageal epithelium and a variety of ESCC precursor lesions.High level of 14-3-3σ expression was found ubiquitously in normal esophageal epithelium with an immuonstaining score of 8.22 in expression.Protein 14-3-3σ was down-regulated stepwise during the multi-stage development of ESCC.Sixty-four percent of poorly-differentiated squamous cancer lost 14-3-3σ expression with a score of 0.45 [Qi et al., 2007a].In agreement with our results, Ren et al. documented that the level of 14-3-3σ in terms of mRNA and protein was markedly downregulated in ESCC compared with nearby matched non-cancer tissues.Furthermore, decrease of 14-3-3σ expression was correlated with tumor infiltration depth, lymph node metastasis, distant metastasis and lymphovascular invasion and shorter 5-year survival rate [Ren et al., 2010].Among the different proteins identified by SILAC-based quantitative analysis using immortal cell and cancer cell model, the clinical values of MIF in tumorigenesis of ESCC was determined as well.Not only the increased expression of MIF was detected in cellular protein but also in the conditioned medium of esophageal cancer cell lines EC1, EC109 and EC9706 compared with immortal cell lines NE3 and NE6.Low frequency and very weak expression of MIF was detected predominantly in basal cells in normal esophageal epithelium, with an immunostaining score of 1.13.Pronouncedly upregulated expression of MIF occurred in severe dysplasia compared with weak immunostaining in mild and moderate dysplasia.In ESCC, high frequency of intense expression of MIF was observed with a score of 5.46.Furthermore, high expression of MIF was significantly correlated with advanced clinical stages.ELISA tests revealed that there was an increase trend in serum level of MIF in clinically advanced stage IV compared to stage I-III.Functional studies on MIF indicated that MIF knockdown resulted in decrease in proliferation, clonogenicity, non-adherent growth and invasive potential.Our findings indicate that MIF may play crucial roles in malignant transformation of pathogenesis of EC and MIF could become a potential biomarker for high-risk population screening, assessment of therapeutic efficiency, prognostic evaluation, and molecular targets of developing novel therapeutic regimen as well.In addition of our proteomic results in ESCC, several other reports have looked at the clinical value of potential biomarkers, including cytokeratin 14, Annexin I, SCCA1/2, calgulanulin B and HSP 60, alpha-actinin 4 and 67 kDa laminin receptor, cathepsin D and PKM2, periplakin, calreticulin and GRP78, galectin-7, anti-CD25B antibody [Dong et al., 2010;Du et al., 2007;Fu et al., 2007;Hatakeyama et al., 2006;Liu et al., 2011;Nishimori et al., 2006;Zhu et al., 2010].Nevertheless, further extensive studies are still necessary to determine the clinical utility of the identified proteins in tumorigenesis and progression of ESCC.

Conclusions
Nowadays, the dilemma for cancer control and management is not due to lack of efficient treatment options but diagnosis at late stages.In the case of ESCC in China, five-year survival rate for early stage tumor reaches around 90% [Hu et al., 2001].Obviously, to detect tumor as early as possible is the key for reducing the mortality and morbidity of ESCC.It is believed that development of ESCC from normal esophageal epithelium takes at least about 10 years, during which diseased epithelium manifests as basal cell hyperproliferation, dysplasia, carcinoma in situ in terms of morphology and finally evolves to malignant neoplasms.As such, carcinogenesis of ESCC is a multi-stage and dynamic process which accumulates ongoing changes at the level of both gene and protein expression.Proteomic studies from various research groups worldwide have identified distinct dysregulated protein expression pattern associated with ESCC.The discrepancy might reflect the different etiology, different stages of disease and diverse pathways involved, which makes identification of biomarkers for ESCC difficult.In light of a wealth of potential biomarkers associated with ESCC identified so far in the exploratory phase, future largescale validation studies involving symptom-free patients with precursor lesions in highincidence area and ESCC patients compared with controls are essential toward clinical application.Therefore, ultimate translation from laboratory into bedside for ESCC biomarkers will require close collaboration and cooperation between researchers and clinicians to look into the clinical utility in diagnosis at early stage, prognosis and monitoring treatment efficiency for ESCC.

Acknowledgement
This work was supported in part by National Natural Science Founding of China (No. 30700366 and No. 81072039) and Cancer Research UK (to Yi-Jun Qi).

Table 2 .
Differential proteins between immortalized and cancer cell lines derived from ESCC identified by SILAC-based proteomics selected Annexin A2 for validation by Western blot and IHC.