Secreted proteins can serve as excellent biomarkers for the early detection, prognostication and management of ESCC. To identify potential biomarkers from the ESCC secretome, a quantitative proteomic analysis based on a SILAC strategy was carried out. A non-neoplastic esophageal epithelial cell line (Het-1A) was labeled with ‘heavy’ arginine plus lysine, while 7 ESCC cell lines representing various stages of tumor differentiation were labeled with ‘light’ arginine plus lysine. The conditioned medium containing secreted proteins was processed according to the work flow displayed in . The protein sample was resolved in duplicate by SDS-PAGE and stained with colloidal coomassie. Twenty-four bands from both lanes were excised and in-gel trypsin digestion was then carried out.
Quantitative mass spectrometric analysis of the ESCC secretome.
From quantitative proteomic analyses of the ESCC secretome using high-resolution tandem mass spectrometry coupled with liquid chromatography (LC-MS/MS), 17,310 MS/MS spectra were acquired. Using an FDR cutoff of 1%, 950 unique peptide spectrum matches (PSMs) were obtained, leading to the identification of 441 proteins (Mascot and Spectrum Mill). Proteins identified are summarized in Supplemental Table 1
. In , the Venn diagram shows the distribution of proteins identified by either the Mascot or Spectrum Mill search engines. Among the identified proteins, 45% were common to both search engines, while 32 and 23% were unique to the Mascot or Spectrum Mill searches, respectively. and C
summarize the distribution of these proteins based on their fold-changes.
Figure 2 Protein profiled using the SILAC strategy. (A) Venn diagram showing proteins identified by Mascot and Spectrum Mill search algorithms. (B) Distribution of proteins identified by Spectrum Mill plotted against the log2 ratios as indicated. Proteins for (more ...)
From Spectrum Mill searches, 35 proteins were not detected in the normal cell secretome and 5 were not detected in the ESCC secretome. 100 proteins were upregulated ≥ 2-fold in the ESCC vs. the normal cell secretome. 106 proteins exhibited values between <2 and 0.5; these were considered essentially unchanged between the two secretomes. 52 proteins were downregulated < 0.5-fold in the ESCC vs. normal secretome.
Quantitative analysis using Mascot Distiller identified 340 proteins. 10 proteins were not detected in the normal cell secretome, while 1 was not detected in the ESCC secretome. 144 proteins fell between < 2 and 0.5-fold change i.e., their expression was essentially unchanged between the normal and ESCC secretomes. 101 proteins were upregulated ≥ 2-fold in the ESCC secretome vs. that of normal cells, while 84 proteins were downregulated < 0.5-fold in the ESCC secretome.
Details of peptide quantitation corresponding to the identified proteins using Spectrum Mill and Mascot distiller are summarized in Supplemental Tables 2 and 3
respectively. Included in this Table are fold-changes for differential expression between the ESCC and normal cell secretomes, peptide sequence, parent charge, modifications, parent m/z, parent mass and delta parent mass. Representative MS and MS/MS spectra of the selected peptides for known, novel and upregulated proteins are shown in .
Figure 3 MS and MS/MS spectra of selected differentially expressed proteins. MS and MS/MS spectra of peptide from representative differentially expressed proteins identified in this study. (A) Matrix metalloproteinase 1 (MMP1); (B) TGFbeta induced, 68 KD (TGFBI (more ...) Biological function, pathway analysis and cellular localization analysis.
We carried out a bioinformatics analysis to classify proteins based on subcellular localization and biological function. Classification was carried out based on annotations in the Human Protein Reference Database (HPRD; www.hprd.org
in compliance with Gene Ontology (GO) standards. This summary includes fold-changes for protein expression between the secretomes of ESCC and non-neoplastic cells, along with biological domains and motifs obtained from HPRD. We also searched for previous reports describing detection of these proteins in any biological fluids, using HPRD and the Human proteinpedia (HUPA; www.humanproteinpedia.org
Of the 441 proteins identified, 72 contained signal peptides (SP), 11 contained a transmembrane (TM) domain and 15 contained both a TM domain and an SP motif. The MS and MS/MS spectra of representative known and novel proteins are displayed in . In the current study, 343/441 (77.8%) of proteins had been previously reported in biological fluids, including urine, semen, plasma, serum, tear, saliva, synovial fluid, cerebrospinal fluid, bronchoalveolar fluid, blood, milk, colostrum, pancreatic fluid, cerebrospinal fluid, aqueous humor or vitreous humor. Further analysis revealed that 75% of upregulated proteins had been previously reported in one or more biological fluids in normal or diseased conditions.
Known and overexpressed proteins in ESCC.
Among overexpressed proteins in ESCC cell lines, we found a number of proteins that have been previously described in the context of ESCC, confirming the validity of our quantitative proteomic approach. A partial list of known and upregulated proteins is shown in . Proteins previously reported as overexpressed in ESCC include matrix metallopeptidase 1 (MMP1
enolase 1 (ENO1
) and isocitrate dehydrogenase 1 (IDH1
). In our earlier mRNA profiling study of ESCC tissues, MMP1
was 11-fold more abundant in cancer vs. adjacent normal tissues.41
In the current study, at the protein level, MMP1
was also ~24-fold upregulated in the ESCC secretome.
Partial list of overexpressed proteins that were previously reported in esophageal squamous cell carcinoma
Enolase 1 (ENO1
) is widely expressed in varying types of tissues. In our study, ENO1
was 2.3-fold upregulated in the ESCC secretome. ENO1
has also been previously reported as 1.6-fold upregulated in ESCCs vs. adjacent normal epithelia in a proteomic study using 2-dimensional gel electrophoresis (2-DE).21
We identified HSP70
as 2.4-fold upregulated in the ESCC secretome. In another previous study of ESCC patients, an autoantibody against HSP70
was identified by MALDI-TOF/TOF-MS in sera.23 HSP90B1
, also known as tumor rejection antigen (gp96
), was 3.8-fold upregulated in the current study and was found to be 5-fold upregulated in an earlier study on ESCC. Keratin 1 (KRT1
) is another protein that was ~11-fold upregulated in our study; KRT1
was also 12.1-fold upregulated in a previously published report.25
Finally, our study provides validation at the protein level for several biomarkers previously reported only at the mRNA level in ESCC including transferrin receptor (TFRC
) and transforming growth factor, beta-induced, 68 kDa (TGFBI
). Transferrin receptor has been described as an independent prognostic factor based on mRNA expression analysis in ESCC.42
In the current study, TFRC
were 4.3- and 14.9-fold upregulated in the ESCC secretome.
) is a cytoplasmic peripheral membrane protein that functions as a protein-tyrosine kinase substrate in microvilli. EZR
serves as an intermediate between the plasma membrane and the actin cytoskeleton. EZR
plays an important role in cell surface structure adhesion, migration and organization.43
In earlier studies of ESCC, the expression of EZR
protein was studied by western blotting, IHC labeling or RT-PCR.44–46
In the current study, EZR
was 2.5-fold upregulated in the ESCC secretome. Zeng et al. studied and reported an association of EZR
overexpression with poor survival in ESCC using IHC labeling.46
Heat shock protein HSP90
was earlier studied in the context of ESCC, but no significant differences were reported in its expression levels between normal and ESCC subjects.47,48
In our study, it was 3.9-fold upregulated in the ESCC vs. normal cell secretomes.
Neutrophil gelatinase-associated lipocalin (NGAL
) or lipocalin 2 (LCN2
), is a member of the lipocalin family, which is involved in transport of small lipophilic substances. LCN2
may play an important role in breast cancer in vivo by protecting MMP9
from degradation, thereby enhancing its enzymatic activity and facilitating angiogenesis and tumor growth. Clinically, these published data suggest that the detection of LCN2/MMP9
in urine may be useful in non-invasively predicting the disease status of breast cancer patients.49
Enzymatic levels of the LCN2/MMP9
complex in ESCC have been reported to correlate significantly with depth of tumor invasion.50
Moreover, hypomethylation of LCN2
was reported in ESCC tissues and cell lines, giving rise to the hypothesis that NGAL
may play an important role in ESCC.51
In the same study, LCN2
overexpression was found to be positively correlated with cell differentiation in ESCC.51
In our study, LCN2
was not detected in the normal cell secretome, in contrast to the ESCC secretome. We verified the overexpression of LCN2
using western blotting, finding that overexpression of LCN2
was observed in the ESCC secretome but not detectable in the normal secretome (), consistent with the aforementioned earlier study on ESCC.50
Cathepsin D (CTSD
) is an aspartic protease involved in tumor progression and other biological processes including cell proliferation, angiogenesis and apoptosis. CTSD
overexpression has been reported in cholangiocarcinoma,52
and it has been shown as an independent indicator of poor prognosis in breast cancer.53
In our study, CTSD
was 4.8-fold upregulated in the ESCC secretome. CTSD
has been tested for its prognostic value in ESCC, but there has been no reported association with clinical factors.54
Downregulated proteins in ESCC.
120 proteins were downregulated ≥ 2-fold in the ESCC secretome vs. the esophageal epithelial cell line secretome. Profilin 2 (PFN2
), an actin monomer binding protein, has been shown to be downregulated in nasopharyngeal carcinoma,12
and breast cancer.56
In our study of the ESCC secretome, PFN2
was 0.3-fold downregulated. PFN2
regulates the structure of the cytoskeleton, but its role in ESCC has not yet been explored.
H2A histone family member X (H2AX
) is involved in the DNA damage response and mediates DNA repair; this protein was downregulated in the ESCC vs. normal cell secretome, with a fold-change of 0.2. H2AX
was also downregulated at the mRNA level in our earlier ESCC transcriptomic study.41
In ESCC, phosphorylation of H2AX
has been observed in response to bortezomib drug treatment in an organotypic culture and an in vivo model of ESCC.57
In our study, we did not identify the phosphotyrosine site for H2AX. H2AX
is required for checkpoint-mediated arrest of cell cycle progression in response to low doses of ionizing radiation, UV-light or radiometric agents, as well as for efficient repair of DNA double-strand breaks (DSBs), specifically when modified by C-terminal phosphorylation at Ser139 by ATM.58
However, the complete biological importance and role of H2AX
are still unclear and there is a need to study its functional relevance in ESCC.
Novel and overexpressed proteins in ESCC.
A number of proteins that were identified as overexpressed in ESCC have not been described previously in the context of ESCC. A partial list of those proteins is shown in . Among the novel candidates, PDIA3, YHWAZ, LGALS3BP
were overexpressed in the ESCC secretome as compared to the normal cell derived secretome. Protein kinase c inhibitor which is also called YWHAZ
or tyrosine 3/tryptophan 5-monooxygenase activation protein, zeta polypeptide was reported to be overexpressed in oral squamous cell carcinoma and amplification of 14-3-3 zeta was also observed in head and neck squamous cell carcinoma,59,60
and urothelial carcinoma.61
Partial list of novel proteins identified as overexpressed in esophageal squamous cell carcinoma
Karyopherin beta1 (KPNB1
) is also called importin subunit beta-1. Co-operation of KPNB1
is essential for the nuclear import of proteins containing nuclear localizing signal (NLS).62
Recently, karyopherins have been shown to be responsible for uncontrolled growth and considered as a therapeutic target for cancer.63 KPNB1
was overexpressed in ESCC derived secretome with 2.2-fold. In ESCC, the biological importance of KPNB1
still needs to be explored. Dermicidin (DCD
) expression has been associated with cancer cell survival and growth in breast cancer. DCD
gene has also been reported in relation to cancer cachexia.64 DCD
acts as an oncogene in invasive breast,65
and hepatic cancer cells.66
In our study, DCD
was 28-fold upregulated in ESCC derived secretome.
Validation by immunohistochemical staining and western blot analysis. Proteins of interest identified in the current study were further validated by immunohistochemical labeling and western blot analysis to determine their utility as potential biomarkers for ESCC. We selected TGFBI, PDIA3, LGALS3BP and GDI2 based on biological importance, fold-change (≥ 2-fold change upregulation) and reports in other cancers, for further validation in formalin fixed paraffin embedded tissue sections.
The selected proteins for immunohistochemical labeling were also validated by using western blotting. Western blot validation was also done for one downregulated, upregulated and unchanged protein in the secretome derived from ESCC as compared to normal epithelial cells. Western blot-based validation of selected proteins is shown in with SILAC ratios in the secretome derived from normal and ESCC cell lines (pooled). Among the molecules which were validated only by the western blot were thrombospondin-1 (THBS1
), Hypoxanthine-guanine phosphoribosyltransferase (HPRT
), glutathione S-transferase mu 3 (GSTM3
) and lipocalin 2 (LCN2
). The IHC scoring for all ESCC patients for TGFBI, PDIA3, LGALS3BP
are summarized in the and also provided for individual patient in the Supplemental Table 4
Western blot validation for selected proteins identified in the ESCC secretome. Pooled conditioned media from different ESCC cell lines and normal cell line was tested for expression of the indicated proteins using commercially available antibodies.
Summary of immunohistochemical labeling of different molecules in ESCC cases
is an extracellular matrix and secreted protein. In an earlier study on transcriptomics of ESCC, we observed 7.9-fold upregulation of THBS1
in ESCC as compared to normal epithelium.41
In ESCC, association of THBS1
was associated with short survival rate.67
In ESCC secretome analysis, it was 2.3-fold overexpressed in secretome of ESCC cell lines as compared to normal. Further, western blot data was in agreement with the findings obtained from SILAC based mass spectrometry data on ESCC secretome.
protein encoded by HPRT
gene was used as control because it was unchanged between secretome derived from ESCC and normal epithelial cells and consistent with earlier studies where it has been used as a reference.68
Another molecule, GSTM3
belongs to the category of antioxidant defense proteins. It was reported to play an important role in breast cancer by protecting normal breast epithelial cells against breast carcinogenesis.69
was downregulated in the ESCC-derived secretome with 0.2-fold. These results were confirmed by using western blot analysis and are in agreement with the SILAC ratio that we observed. GSTM3
downregulation indicates the possibility of combat between antioxidants and free radicals in ESCC, but further studies are needed to explore this aspect of ESCC tumorigenesis.