4. Biomarker Discovery for CRC Early Diagnosis by Secretome Proteomics
As carcinogenesis is a long process, early tumor detection requires highly sensitive and specific screening tests whose efficacy relies on the proper biomarkers, mainly in the case of asymptomatic tumors such as CRC [
88]. In this context, carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA 19-9) are the most adopted blood-based biomarkers in current CRC clinical practice, although they are not suitable for early diagnosis [
88,
89]. Hence, the pre-selection of potential circulating biomarkers is still an ongoing task and secretome proteomic studies fulfill this requirement, thereby overcoming the technical limitations underlying the direct search in clinically relevant biofluids [
9].
As stated above, secreted protein biomarkers from CM of CRC cell lines could be detectable candidates in blood. The results from the most recent secretome proteomic studies, including protein biomarkers already related to CRC detection, are shown in and and discussed below.
We previously reported the secretome signatures of two CRC cell lines (CaCo-2 and HCT-GEO), and among the identified proteins, more than a half were classified as secretory by in silico analysis [
34]. Based on the results, we suggested that HCT-GEO CM is enriched by pro-invasive factors, whereas CaCo-2 CM was characterized by adhesion proteins [
34].
Likewise, Greening and colleagues compared the protein profile of LIM1215 CM with or without sulindac treatment (a chemopreventive nonsteroidal antinflammatory drug) [
35]; downregulated proteins in the dataset represented sulindac-sensitive species that were enriched in the CRC secretome. Among the downregulated extracellular matrix (ECM)-remodeling-associated proteins, the authors identified collagens, the basement membrane laminin receptors and several proteoglycans [
35]. In another report, the same authors analyzed LIM1215 and LIM1863 secretopeptidomes, and identified, after in silico analyses, secreted proteins implicated in tumor progression and angiogenesis; cell–cell recognition and signaling; and tumor invasiveness and metastasis [
36].
A wider secretome proteomic analysis was performed by Fanayan and colleagues, who considered 21 CRC cell lines (LIM1215, LIM1863, LIM1899, LIM2099, LIM2405, LIM2463, LIM2537, LIM2550, LIM2551, HCT-15, HCT-116, HT-29, CaCo-2, HCA-7, LOVO, LS174T, SW480, SW620, SW1222, SW1463 and T84). These authors identified about 2500 non-redundant proteins, and using bioinformatic analysis they selected 15 putative CRC-related protein biomarkers, including carcinoembryonic antigen-related cell adhesion molecule 6 (CEACAM6/CD66c) [
45]. Interestingly, CEACAM6/CD66c was tested as a CRC stemness marker in CRC patients [
90], and recently, the
CEACAM6 transcript was validated as a blood marker for CRC screening, along with the transcripts of
LGALS4,
TSPAN8 and
COL1A2, also to discriminate false-positive fecal immunochemical test (FIT) subjects [
91].
Karagiannis and colleagues analyzed and compared the secretomes of 12 different CRC cell lines (SW1116, SW480, LS174T, LS180, WiDR, SW620, RKO, LoVo, HCT-116, DLD1, Colo320HSR and Colo205), thereby identifying secreted proteins involved in cell adhesion machinery during cancer progression, especially during phenotypic reprogramming such as epithelial-to-mesenchymal transition (EMT) [
46]. A profile of enriched biomarkers of CRC progression was obtained by using a multi-step bioinformatic analysis [
46]. Among the selected biomarkers, the levels of olfactomedin-4 (OLFM4) showed a significant correlation with CRC when measured in the sera of CRC patients versus controls. Interestingly, among the proposed biomarkers, there was also carbonic anhydrase IX (CA9), a well-known high expressed isoenzyme in solid tumors [
92,
93].
By comparing the secretomes of HCT-8 and HCT-116 cell lines with a previous characterized CRC tissue proteome [
94,
95,
96], Shin and colleagues selected 11 secreted candidate biomarkers (). Interestingly, melanotransferrin (TRFM) plasma levels were found significantly higher in CRC patients compared with healthy controls by using Western blot [
37]. This striking result was then confirmed by ELISA in a wider CRC patient group, highlighting for TRFM a very high positive predictive value and very high specificity, and suggesting its suitability for early-stage diagnosis in CRC [
37].
Given the positive correlation between immunoglobulin-like cell adhesion receptor L1 (L1CAM) expression and CRC tumorigenesis [
97], Basu and colleagues recently found, in L1CAM-overexpressing CRC cells, more than 10-fold increased levels of nine secreted proteins [
41]. Among these, they confirmed that increased levels of cathepsin D (CTSD) enhance the motility, tumorigenesis and liver metastasis of CRC cells by activation of Wnt/β-catenin signaling. In particular, high levels of CTSD were found in the invasive front of CRC tumor tissues, thereby suggesting CTSD as a promising biomarker for CRC progression [
41].
Barderas and colleagues compared the secretome of the KM12C cell line with that of its metastatic counterpart KM12SM, and after bioinformatic analysis they selected 80 potentially secreted and differentially expressed proteins linked to metastatic processes in CRC. Among them, eight proteins were validated in metastatic cells by western blot [
32]. The authors also demonstrated by functional experiments that some of them were implicated in cellular adhesion, migration/invasion and metastatic spreading towards the liver in vivo. Interestingly, neuroserpin (SERPINI1), growth/differentiation factor 15 (GDF15) and calcium-binding protein A8/A9 (S100A8/A9) showed potential as biomarkers for CRC diagnosis. In fact, these candidates were validated by ELISA in serum, and they allowed to discriminate CRC patients from healthy controls with high sensitivity and specificity [
32].
To investigate the influence of an in vitro-mimicked tumor microenvironment on the protein expression profile, Zeng and colleagues compared the secretomes of HT-29 cell line and of a normal human colon mucosal epithelial cell line (NCM460) alone and in co-cultures [
33]. They selected insulin-like growth factor-binding protein 6 (IGFBP6), vimentin (VIM) and acrogranin (GRN) as proteins potentially implicated in the CRC progression and validated them by Western blot, ELISA and immunofluorescence (). Indeed, the authors pointed out that the changes in the levels of these three proteins correlated with different NCM460/HT29 co-culture ratios resembling different stages of CRC [
33]. Interestingly, VIM, besides being a key regulator of cell adhesion and cell–cell interactions [
98], has a well-known role in the EMT process in many types of cancer [
99], while IGFBP6 is particularly implicated in the IGF1-mediated EMT [
33,
99,
100]. Additionally, Bukhari and colleagues proposed VIM as CRC circulating biomarker by using an affinity proteomic approach as described above. This approach allowed the identification of potential plasmatic cancer biomarkers that might be responsible for CRC development and progression; among these proteins, very high levels of VIM were found in CRC patient sera compared to in healthy controls, and there was a significant correlation of these levels with CRC, thereby supporting the predictive value of VIM for monitoring subjects at increased risk of CRC [
44].
Until now, there have been few reports on CRC glycosecretome that represent suitable sources of biomarkers using glycoprotein enrichment to detect low abundance secreted proteins in the CM of CRC cell lines. In this context, Bernhard and colleagues analyzed the glycosecretome in LIM1215 CM and in the TIF of tumors derived by the same cell line xenografted in immunodeficient mice. These authors found three secreted species shared between the two type of samples (). Among these proteins, cadherin-17 (CDH17) was also validated by Western blot in cell lysates, CM, TIFs and in the plasma of xenografted mice, thereby suggesting its potential as a CRC biomarker [
31]. A different set of glycoproteins was highlighted by Lin and colleagues when they compared the glysecretome of HCT-116 cell line with its metastatic derivative. A special focus was set on laminin β-1 (LAMB1), over-secreted in the metastatic cell line, which was significantly higher in the sera of CRC patients compared with healthy controls. Moreover, ROC analyses showed that LAMB1 discriminated CRC patients from controls better than CEA [
38]. In a recent glycosecretome study, Link-Lenczowski and colleagues highlighted the N-glycosylation changes between two differentiation stages of CaCo-2 cell line. In particular, they found an enhanced fucosylation in differentiated cells and suggested that H4N5F1 glycans might be a biomarker of intestinal epithelial cell differentiation [
42].
Some authors focus their attention on CSCs, and hence, on their secretome to fish for the proper CRC biomarkers. At this regard, when Emmink and colleagues compared the secretomes of CSCs and of isogenic differentiated tumor cells (DTCs), isolated from three different metastatic CRCs, most CSC secreted proteins were involved in cell survival, antioxidant activities and proteome integrity maintenance processes [
56]. Among them, aldehyde dehydrogenase 1 (ALDH1A1) and bleomycin hydrolase (BLMH) conferred to CSCs resistance against maphosphamide and bleomycin, respectively [
56].
In the search for CRC stemness markers, De Boeck and colleagues compared the secretomes of CAFs from two CRC patients with mesenchimal stem cells (MSCs) from bone marrow of healthy individuals [
53]. The in silico analysis of the unique species, exclusively identified in CRC CAFs, suggested their involvement in the regulation of cellular movements and cell-to-cell signaling and interactions, and in the inflammatory responses and cancer progression (). The authors evaluated also the MSC secretome after TGF-β1-induced differentiation, a treatment triggering the conversion of MSCs into CAF precursors: accordingly, they found 16 secreted species shared with CAF secretome including chemokines and growth factors [
53].
Qiao and colleagues analyzed the secretomes of five CRC CAFs and of three CRC cell lines and focused on the proteins which were exclusively secreted by CAFs. Among these proteins, COL6A3 was also analyzed by immunohistochemistry (IHC) analysis on a tissue microarray (TMA) containing 90 pairs of CRC and normal counterpart tissues [
39]. This analysis showed that COL6A3 protein levels were significantly higher in the stromal cells of CRC tissues than in the normal counterparts (). Moreover, COL6A3 expression in cancer stroma was correlated to a poor outcome with a significant prognostic value. An ELISA assay also permitted a comparison of COL6A3 plasma concentrations in CRC patients and healthy individuals. This analysis showed COL6A3 upregulation in the CRC patients and demonstrated a very high prediction value, sensitivity and specificity of COL6A3 as plasmatic biomarker of CRC [
39].
As for the tissue samples, de Wit and colleagues analyzed the secretomes of CRC tissues paired with their normal counterparts from four CRC patients and the secretomes of five CRC cell lines (). By performing a multistep bioinformatic analysis where previously reported datasets were taken into consideration [
101,
102], 21 potential biomarkers were selected for early detection of CRC [
49]. This selection included many nuclear proteins involved in DNA replication, cell division and other CRC-related processes. Among them, CRC overexpression of minichromosome maintenance complex component 5 (MCM5) was validated by immunohistochemical staining in a TMA containing 82 colon adenomas and 82 CRCs [
49].
Wang and colleagues, instead, generated an inflammation-related CRC mouse model and analyzed the protein profile of TIFs from mouse colon tissues during four stages of CRC development. They focused on 11 tumor growth-related proteins analyzed by multiple reaction monitoring (MRM) in the TIF samples (). Among these proteins, leucine-rich alpha-2-glycoprotein 1 (LRG1), tubulin beta-5 chain (TUBB5) and immunoglobulin J chain (IGJ) were confirmed to be stage-dependently increased in the serum of the mice [
50]. Moreover, LRG1 and TUBB5 were also verified as potential biomarkers in the sera of patients with different CRC stages [
50].
Similarly, Xie and colleagues used a widely accepted Apc−/+ mouse model of CRC whose pathological phenotype largely overlaps with human familial adenomatous polyposis and sporadic CRC [
103]. The authors compared the TIF secretomes from Apc−/+ and WT mice of the different ages and identified 46 proteins with tumor progression-dependent expression profiles [
51]. As reported in , the authors particularly focused their attention on six serine proteases and validated their levels by MRM assay, both in TIFs and in mouse sera. Importantly, levels of chymotrypsin-like elastase 1 (CELA1) and chymotrypsin-like protease (CTRL) were also validated by IHC in human TMAs containing 80 pairs of CRC tissues and their normal counterparts. Most importantly, the levels of CELA1, CTRL, chymotrypsin-like elastase 2A (CEL2A) and trypsin 2 (TRY2) were measured by MRM in the sera of CRC patients versus healthy controls and these proteins showed significantly higher expression in CRC sera. In particular, the best diagnostic performance was showed by the combination of CELA1 and CTRL which demonstrated a very high sensitivity and specificity [
51].