A challenge in achieving optimal management of cancer is the discovery of secreted biomarkers that represent useful surrogates for the disease and could be measured noninvasively. Because of the problems encountered in the proteomic interrogation of plasma, secretomes have been proposed as an alternative source of tumor markers that might be enriched with secreted proteins relevant to the disease. However, secretome analysis faces analytical challenges that interfere with the search for true secreted tumor biomarkers. Here, we have addressed two of the main challenges of secretome analysis in comparative discovery proteomics. First, we carried out a kinetics experiment whereby secretomes and lysates of tumor cells were analyzed to monitor cellular viability during secretome production. Interestingly, the proteomic signal of a group of secreted proteins correlated well with the apoptosis induced by serum starvation and could be used as an internal cell viability marker. We then addressed a second challenge relating to contamination of serum proteins in secretomes caused by the required use of serum for tumor cell culture. The comparative proteomic analysis between cell lines labeled with SILAC showed a number of false positives coming from serum and that several proteins are both in serum and being secreted from tumor cells. A thorough study of secretome methodology revealed that under optimized experimental conditions there is a substantial fraction of proteins secreted through unconventional secretion in secretomes. Finally, we showed that some of the nuclear proteins detected in secretomes change their cellular localization in breast tumors, explaining their presence in secretomes and suggesting that tumor cells use unconventional secretion during tumorigenesis. The unconventional secretion of proteins into the extracellular space exposes a new layer of genome post-translational regulation and reveals an untapped source of potential tumor biomarkers and drug targets.

Purpose: The study of the cancer secretome suggests that a fraction of the intracellular proteome could play unanticipated roles in the extracellular space during tumorigenesis. A project aimed at investigating the invasive secretome led us to study the alternative extracellular function of the nuclear protein high mobility group A1 (HMGA1) in breast cancer invasion and metastasis. Experimental Design: Antibodies against HMGA1 were tested in signaling, adhesion, migration, invasion, and metastasis assays using breast cancer cell lines and xenograft models. Fluorescence microscopy was used to determine the subcellular localization of HMGA1 in cell lines, xenograft, and patient-derived xenograft models. A cohort of triple-negative breast cancer (TNBC) patients was used to study the correlation between subcellular localization of HMGA1 and the incidence of metastasis. Results: Our data show that treatment of invasive cells with HMGA1-blocking antibodies in the extracellular space impairs their migration and invasion abilities. We also prove that extracellular HMGA1 (eHMGA1) becomes a ligand for the Advanced glycosylation end product-specific receptor (RAGE), inducing pERK signaling and increasing migration and invasion. Using the cytoplasmic localization of HMGA1 as a surrogate marker of secretion, we showed that eHMGA1 correlates with the incidence of metastasis in a cohort of TNBC patients. Furthermore, we show that HMGA1 is enriched in the cytoplasm of tumor cells at the invasive front of primary tumors and in metastatic lesions in xenograft models. Conclusions: Our results strongly suggest that eHMGA1 could become a novel drug target in metastatic TNBC and a biomarker predicting the onset of distant metastasis.

The unconventional secretion of proteins is generally caused by cellular stress. During the tumorigenesis, tumor cells experience high levels of stress, and the secretion of some theoretically intracellular proteins is activated. Once in the extracellular space, these proteins play different paracrine and autocrine roles and could represent a vulnerability of cancer. One of these proteins is the high mobility group A1 (HMGA1), which is frequently overexpressed in tumors and presents a low expression in normal adult tissues. We have recently described that HMGA1 establishes an autocrine loop in invasive triple-negative breast cancer (TNBC) cells. The secretion of HMGA1 and its binding to the receptor for advanced glycation end products (RAGE) mediates the migration, invasion, and metastasis of TNBC cells and predicts the onset of metastasis in these patients. In this review, we summarized different strategies to exploit the novel tumorigenic phenotype mediated by extracellular HMGA1. We envisioned future clinical applications where the association between its change in subcellular localization and breast cancer progression could be used to predict tumor aggressiveness and guide treatment decisions. Furthermore, we proposed that targeting extracellular HMGA1 as monotherapy using monoclonal antibodies, or in combination with chemotherapy and other targeted therapies, could bring new therapeutic options for TNBC patients.

Purpose: The lack of secreted biomarkers measurable by noninvasive tests hampers the development of effective targeted therapies against cancer. Our hypothesis is that cetuximab (an anti-EGFR mAb) induces a specific secretome in colorectal cancer cells that could be exploited for biomarker discovery. Experimental Design: Considering the strong correlation between mutated KRAS and a lack of response to cetuximab therapy, we addressed whether performing secretome-based proteomics on isogenic colorectal cancer cells sharing the KRAS mutations found on patients would yield candidate-secreted biomarkers useful in the clinical setting. Because 2D culture did not optimally model the sensitivity/resistance to cetuximab observed in colorectal cancer patients, we moved to 3D spheroids, developing a methodology for both cell-based assays and quantitative proteomics. Results: A large comparative quantitative proteomic analysis of the 3D secretomes of colorectal cancer isogenic cells treated with cetuximab uncovered an EGFR pathway-centric secretome found only when cells grow in 3D. The validation of the secretome findings in plasma of colorectal cancer patients, suggests that phosphorylated-EGFR (pEGFR) is a candidate-secreted biomarker of response to cetuximab. Conclusions: We have proved that 3D spheroids from colorectal cancer cells generate secretomes with a drug-sensitivity profile that correlates well with patients with colorectal cancer, illustrating molecular connections between intracellular and extracellular signaling. Furthermore, we show how the secretion of pEGFR is associated with the sensitivity of colorectal cancer cells to cetuximab and the response of patients with colorectal cancer to the drug. Our work could allow the noninvasive monitoring of anti-EGFR treatment in patients with colorectal cancer. Clin Cancer Res; 20(24); 6346–56. ©2014 AACR.

Drug tolerant persister (DTP) cells enter into a reversible slow-cycling state after drug treatment. We performed proteomic characterization of the breast cancer (BC) DTP cell secretome after eribulin treatment. We showed that the growth differentiation factor 15 (GDF15) is a protein significantly over-secreted upon eribulin treatment. The biomarker potential of GDF15 was confirmed in 3D-cell culture models using BC cells lines and PDXs, as well as in a TNBC in vivo model. We also found that GDF15 is required for survival of DTP cells. Direct participation of GDF15 and its receptor GFRAL in eribulin-induction of DTPs was established by the enhanced cell killing of DTPs by eribulin seen under GDF15 and GFRAL loss of function assays. Finally, we showed that combination therapy of eribulin plus an anti-GDF15 antibody kills BC-DTP cells. Our results suggest that targeting GDF15 may help eradicate DTP cells and block the onset of acquired resistance.