| Colorectal cancer(CRC) is one of the most common malignancies worldwide.In American,the incidence and mortality of CRC still ranked the 3rd among all the cancers in both the male and the female in 2008.In China,with the changes of life style and diet structure,the incidence and mortality of CRC continues to increase in recent decades. Despite considerable refinement in therapeutic modalities,almost half of the CRC patients treated with "curative" surgery undergo recurrence within 5 years-mostly with metastatic lesions.Notably,extensive metastasis renders the current treatment ineffective and accounts for most fatalities caused by CRC.Therefore,to predict the metastatic potential of CRC can provide valuable prognostic information as well as opportunities for enhanced intervention,leading to an improved prognosis and increased survival rate.It is widely accepted that the best ways to predict cancer metastasis is to use cancer-specific biomarkers.Over the past several decades,enormous efforts have been made to screen and characterize useful cancer-specific biomarkers.Some important molecules such as CEA,AFP and PSA are identified as cancer biomarkers and commonly applied in clinical practice.Unfortunately,most currently available cancer biomarkers are not satisfactory because of limited specificity and/or sensitivity, stressing the need to discover clinically valuable biomarkers.In a post-genome era,proteomic approaches have been introduced to seek novel cancer biomarkers.Nevertheless,the conventional tissue proteomics approaches have been less striking for biomarker discovery.The reason lies in the fact that proteins identified from tissues are not necessarily detectable in serum or plasma.It is conceivable that a biomarker to be useful in cancer screening and monitoring should be measurable in body fluid samples.Accordingly,mining cancer biomarkers from blood proteome is of particular interest.However,the prospects of blood proteomics are challenged by the fact that blood is a very complex body fluid,comprising an enormous diversity of proteins with a large dynamic range.The abundant blood proteins may mask the less abundant proteins,which are usually potential biomarkers.Several procedures have been made to remove these more abundant proteins before proteomic analysis.Nevertheless,these methods may sacrifice other proteins by nonspecific binding and thus lower the screen efficiency.Above-mentioned major limitations of blood proteomics emphasize the need to seek other approaches for cancer biomarker discovery.More recently,there has been an increasing interest in a newly emerging approach-cancer secretome analysis,which is a promising tool for the identification of cancer biomarkers.The term "secretome" was first proposed by Tjalsma et al.in a genome-based global survey on secreted proteins of Bacillus subtilis.In a broader sense,the secretome harbors proteins released by a cell,tissue or organism through various mechanisms including classical secretion,nonclassical secretion,and secretion through the release of exosomes.It is known that cancer cells interact with their microenvironment by secreting a variety of proteins,including growth factors,extracellular matrix-degrading proteinases,cell motility factors and immunoregulatory cytokines or other bioactive molecules.These cancer-secreted proteins are essential in the processes of differentiation,invasion,and metastasis.More importantly,these cancer secreted proteins or their fragments always enter body fluids such as blood or urine and can be measured via non-invasive assays.Thus,cancer secretome may reflect a broad variety of pathological conditions and represents a more reliable source of biomarkers.To date,only a minority of studies has analyzed cancer secretomes,however,the results regarding biomarker discovery are exciting.Of note,no differential secretome analysis on CRC metastasis has been reported.In this study we used a liquid chromatography-tandem mass spectrometry(LC-MS/MS) based label-free quantitative shotgun proteomics approach to investigate the secretomes of two human CRC cell lines(SW480 and SW620),in order to seek novel metastasis-associated serum biomarkers of CRC.First,we used SW480 and SW620 as our model system because this pair of CRC cell lines derived from the same individual but with different metastatic potentials.The secretome samples collected from SW480 and SW620 were digested with trypsin and analyzed by a FinniganTMLTQTMMS coupled with a HPLC system.MS/MS spectra were automatically searched against the non-redundant International Protein Index(IPI) human protein database(Version 3.26) using the TurboSEQUESTTMprogram.The stringent protein identification criteria were based on Delta Cn(≥0.1) and cross-correlation scores(Xcorr,one charge≥1.9,two charges≥2.2,three charges≥3.75).BuildSummary,a in-house tool,was used to combine the peptide sequences into proteins and delete redundant proteins.Finally,a total of 910 nonredundant proteins based on the identification of two or more unique peptides were identified,which,to our knowledge,represents one of the largest protein profiles identified for CRC.In SW480,588 proteins were identified.Of these,451 proteins were identified in all three replicates,showing a protein identification reproducibility of 77%.The similarly high protein identification reproducibility(74%) was found in SW620,in which 672 proteins were identified and 497 proteins were present in all three replicates. Furthermore,searches against the sequence-reversed decoy IPI human databases using the same search parameters yielded a FDR of 3.82%at the peptide level.The high reproducibility and low FDR highlights the fidelity of LC-MS/MS analysis.Label-free quantitative comparison between the two cell lines was performed by DeCyderTMMS Differential Analysis Software(Version 1.0).Peptide detection, background subtraction and quantitation were performed on the full scan precursor mass spectra in fully automatic mode.Collectively,145 proteins displayed more than 1.5-fold quantitative alterations(t-test,P<0.01) in the secretomes of SW620 versus SW480.Among the 145 proteins,75 proteins were up-regulated,whereas 70 proteins were down-regulated in SW620 compared with SW480 cells.To date,this is one of the largest qualitative proteome studies for CRC.For the 145 proteins,the number of peptides used for quantitation from each protein varied between 1 and 10.Among these, 13 proteins were comparatively quantified on the basis of change levels of three or more peptides that varied similarly.From these data it is possible to evaluate the confidence of the quantitative approach.The average coefficient of variation(CV) of the fold changes for peptides from these 13 proteins was 21%(range 3.2-48.7%),yielding a reasonable reproducibility of the quantitative data.Subsequently,overall features for the differential secreted proteins were analyzed by various bioinformatics analytic tools.First,cellular localization of identified proteins was analyzed on the basis of Gene Ontology(GO) and Human Protein Reference Database(HPRD).The localization of 42%differential proteins was classified as extracellular and membrane.The ratio is a little higher than previous secretome studies (30%),which demonstrates the advantage of our enrichment method for secreted proteins.Additionally,70 proteins identified in the CM were assigned to intracellular organelles,cytoskeleton,nucleus and cytoplasm(49%).The identification of a large portion of intracellular proteins is due in part to nonspecific liberation of cytoplasmic proteins as a consequence of cell death.However,we believe that the presence of many putative intracelluar proteins in CM may not merely be caused by cell autolysis,but also due in part to the active release through nonclassical secretion pathway and exosomes. Using SecretomeP 2.0 and SignalP 3.0 software,we analyzed the amino acid sequences of differential secreted proteins.The nonclassical secreted protein identification criteria were based on SecP score > 0.5 and the absence of signal peptides.Finally,we successfully identified 23(33%) nonclassical secreted proteins from the 70 intracellular proteins.Biological function classifications were performed with the tools on DAVID Bioinformatics Resources 2008,and pathway analysis was done by searching Kyoto Encyclopedia of Genes and Genome(KEGG) database.The top biological functions of differential proteins were cell differentiation followed by cytoskeleton organization, apoptosis,anatomical structure morphogenesis,cell proliferation,cell adhesion, response to external stimulus and cell motility.The top ranked pathways were those involved in regulation of actin cytoskeleton,ECM-receptor interaction and focal adhesion.These findings suggest that the differential secreted proteins identified in SW480 versus SW620 might be implicated in CRC initiation and progression,and include valuable biomarker candidates for CRC.The relative expression level of proteins determined by DeCyder MS was verified by SYBR Green Q-PCR and Western blot analysis of 7 selected proteins,GDF15,TFF3, AGR2,LASP1,TGM2,LCN2 and IGFBP7.These 7 proteins were selected on the basis of a combination of parameters including level of differential expression in SW620 versus SW480,cellular localization,availability of commercial antibodies,and the potential relevancy with cancer progression.From Q-PCR results,we found that the relative expression level of 6 genes coincided with the results of DeCyder MS;only LASP1 showed reversed fold change between the two cell lines.The expressions of the 7 selected proteins in the CM and in the total cell lysates were assayed by Western blot as well.In the CM harvested from this pair of CRC cells,the changes in protein abundance noted in the Western blot analysis were considerably consistent with the results obtained by proteomic screens,which demonstrates the accuracy of label-free quantitation.For GDF15,TFF3,AGR2,TGM2 and LCN2,similar results were observed in the cell lysates,which suggested the dysregulated secretion of these proteins was likely associated with the altered protein expression between the two cell lines.Notably,the change levels of LASP1 noted in cell lysates were consistent with Q-PCR results,suggesting that the increased expression of this protein in SW620 CM might be the increased secretion.Unfortunately,we could not detect IGFBP7 in cell lysates,although it was clearly present in SW480 CM,indicating rapid and efficient secretion of this protein.Among the 7 proteins confirmed by Q-PCR and Western blot analysis,we chose two up-regulated proteins in SW620 versus SW480,TFF3 and GDF15,for further estimation in terms of its clinical relevancy in cancer metastasis.Using immunohistochemical staining,we measured TFF3 and GDF15 expression in 38 cases of CRC with lymph node metastasis(LNM),in which each case contains both primary tumor and metastatic lymph node,and 31 cases of non-LNM CRC.Comparison of immunoreactive scores of non-LNM CRC,primary tumors of LNM CRC and matched metastatic lymph nodes demonstrated that TFF3 expression in LNM CRC was slightly up-regulated although not significantly different from that in non-LNM CRC;in contrast,metastatic lymph nodes displayed stronger immunoreactivity of TFF3 than matched primary tumors(P<0.0001).A strong immunoreactivity was detected for GDF15 in LNM CRC,which was significantly higher than that in non-LNM CRC(P < 0.0001).In addition,metastatic lymph nodes exhibited stronger GDF15 expression compared to matched primary tumors(P=0.041).This finding indicated that dysregulation of TFF3 or GDF15 expression in CRC tissues was highly associated with CRC metastasis.Furthermore,No appreciable correlation was demonstrated between TFF3 expression and patient's gender,age,tumor size,histological grade and clinical TNM stage.Notably,however,TFF3 expression was site-related,cancers localized in colon mostly exhibited stronger immunoreactivity than cancers in rectum(P=0.007). There was a significant increase in the imunostaining of GDF15 between TNM tumor stageâ… /â…¡andâ…¢/â…£(P<0.0001).The level of GDF15 expression was not statistically associated with gender,age,tumor size and site.As immunohistochemical analysis indicated the association between TFF3 and GDF15 dysregulation and CRC metastasis,we speculated they might be serum marker candidates for predicting CRC metastasis.To test this speculation,the levels of TFF3 and GDF15 in serum samples collected from CRC patients(n=144) and healthy controls (n=156) were measured by double-antibody sandwich ELISA system.The serum levels of both proteins were significantly higher in CRC patients than in healthy controls.For details,the mean levels for serum TFF3 and GDF15 in healthy control were 824±278 ng/ml,and 639±212 pg/ml,respectively,whereas in CRC patients serum samples,the corresponding mean were elevated to 1925±1637 ng/ml and 2030±1122 pg/ml, respectively.Then,for analysis purposes,the 144 cases of CRC serum samples were subdivided into non-LNM CRC(n=68) and LNM CRC(n=76) groups.Compared with non-LNM CRC group,LNM CRC group exhibited significantly increased serum levels of both TFF3(2690±1839 ng/ml versus 1071±727 ng/ml) and GDF15(2667±1067 pg/ml versus 1319±663 pg/ml).These findings indicated that the two proteins might be potential serum biomarkers for CRC.To evaluate the diagnostic performance of serum TFF3 and GDF15,we constructed receiver operating characteristic(ROC) curve by plotting sensitivity versus specificity.The area under the ROC curve(AUC),a commonly used indicator for estimating the diagnostic efficacy of a potential biomarker,was subsequently calculated. For discriminating CRC from healthy controls,the AUC was determined to be 0.730 (95%confidence interval,0.670-0.791) for TFF3 and 0.897(95%confidence interval, 0.856-0.938) for GDF15,respectively.When a cutoff value of 1323 ng/ml was chosen for TFF3,the sensitivity and specificity for differentiating CRC with healthy controls were 53.5 and 97.4%,respectively.As well,using a cutoff value of 1144 pg/ml,GDF15 had a diagnostic sensitivity of 77.8%and specificity of 99.4%in detecting CRC. Notably,the combination of TFF3 and GDF15 showed a higher diagnostic capacity than either marker alone(AUC=0.926;95%confidence interval,0.890-0.961).The ROC curves of serum TFF3 and GDF15 for discerning LNM CRC versus non-LNM CRC were also constructed.The AUC for TFF3 was 0.822(95%confidence interval, 0.750-0.894),at a cutoff value of 1789 ng/ml,its sensitivity and specificity for predicting CRC metastasis reached to 73.7%and 91.2%,respectively.GDF15 had a similar AUC to that of TFF3(0.867,95%confidence interval,0.806-0.928).Setting a cutoff value of 1881 pg/ml,sensitivity and specificity for GDF15 in distinguishing LNM-CRC with non-LNM CRC were 82.9%and 82.4%,respectively.Nevertheless, combining serum TFF3 with GDF15 measurements did not improve the efficacy for predicting CRC metastasis(AUC 0.866,95%confidence interval,0.805-0.927).These results collectively indicated that TFF3 and GDF15 might be valuable serum biomarker for diagnosis and metastasis prediction of CRC.Analysis of the cohort of 144 CRC patients was carried out to define the relationship between TFF3 and GDF15 serum levels and clinicopathologic features of CRC.The rises in TFF3 or GDF15 serum levels were significantly associated with higher histological grade,increasing clinical TNM stage and the presence of distant metastasis of CRC.Serum level of TFF3 and GDF15 was not statistically correlated with gender,age and site,although cancers localized in colon tended to have higher serum TFF3 level than cancers in rectum(2071±1629 versus 1753±1642 ng/ml, P=0.059).In conclusion,our data demonstrated that:Using label-free quantitative shotgun proteomics approach to investigate differential cancer secretomes is a feasible strategy to seek valuable serum biomarkers. TFF3 and GDF15 are associated with the carcinogenesis and metastasis of CRC,and might be clinically useful serum biomarkers for CRC detection and metastasis prediction. |
PDF Full Text Request