Colorectal cancer development is affected by the ECM molecule EMILIN-2 hinging on macrophage polarization via the TLR-4/MyD88 pathway

Colorectal cancer is one of the most frequent and deadly tumors. Among the key regulators of CRC growth and progression, the microenvironment has emerged as a crucial player and as a possible route for the development of new therapeutic opportunities. More specifically, the extracellular matrix acts directly on cancer cells and indirectly affecting the behavior of stromal and inflammatory cells, as well as the bioavailability of growth factors. Among the ECM molecules, EMILIN-2 is frequently down-regulated by methylation in CRC and the purpose of this study was to verify the impact of EMILIN-2 loss in CRC development and its possible value as a prognostic biomarker. The AOM/DSS CRC protocol was applied to Emilin-2 null and wild type mice. Tumor development was monitored by endoscopy, the molecular analyses performed by IHC, IF and WB and the immune subpopulations characterized by flow cytometry. Ex vivo cultures of monocyte/macrophages from the murine models were used to verify the molecular pathways. Publicly available datasets were exploited to determine the CRC patients’ expression profile; Spearman’s correlation analyses and Cox regression were applied to evaluate the association with the inflammatory response; the clinical outcome was predicted by Kaplan-Meier survival curves. Pearson correlation analyses were also applied to a cohort of patients enrolled in our Institute. In preclinical settings, loss of EMILIN-2 associated with an increased number of tumor lesions upon AOM/DSS treatment. In addition, in the early stages of the disease, the Emilin-2 knockout mice displayed a myeloid-derived suppressor cells-rich infiltrate. Instead, in the late stages, lack of EMILIN-2 associated with a decreased number of M1 macrophages, resulting in a higher percentage of the tumor-promoting M2 macrophages. Mechanistically, EMILIN-2 triggered the activation of the Toll-like Receptor 4/MyD88/NF-κB pathway, instrumental for the polarization of macrophages towards the M1 phenotype. Accordingly, dataset and immunofluorescence analyses indicated that low EMILIN-2 expression levels correlated with an increased M2/M1 ratio and with poor CRC patients’ prognosis. These novel results indicate that EMILIN-2 is a key regulator of the tumor-associated inflammatory environment and may represent a promising prognostic biomarker for CRC patients.


Background
Colorectal cancer (CRC) is the third most common malignancy and one of the major causes of cancer-related deaths [1]. Resection of both primary and metastatic lesions ensures the best prognosis for these patients, however post-intervention recurrence is very common due to disseminated latent or therapy-resistant tumor cells [2]. Accumulating evidence suggest that tumor progression and recurrence are governed not only by genetic changes intrinsic to cancer cells, but also by microenvironmental cues [3]. The role of the immune system in modulating CRC growth and progression has been extensively demonstrated [4]. Inflammatory responses play a decisive role in different stages of tumor development, including initiation, promotion, malignant conversion, invasion and metastasis [5]. Recent observations highlighted that the immune landscape of CRC impacts also on the efficacy of conventional chemotherapies and of targeted therapies and affects the onset of possible mechanisms of resistance [6][7][8][9][10][11].
It is currently accepted that the immune system plays a dual role in cancer. On one side, it can suppress tumor growth by destroying cancer cells or inhibiting their outgrowth, a process that is known as cancer immunosurveillance [12], chiefly mediated by myeloid derived suppressor cells (MDSC) [13]. On the other side, it can also promote tumor progression, either by selecting for tumor cells that are more adaptable to survive in an immunocompetent host (immunoselection), or by establishing conditions within the tumor microenvironment that facilitate tumor growth (immunosubversion) [4]. All these mechanisms are referred to as tumor immunoediting [14]. One of the main tumor-infiltrating immune cell types involved in immunoediting are tumor-associated macrophages. Two functionally contrasting macrophage subtypes have been described, namely classical activated M1 macrophages and alternatively activated tumor-promoting M2 macrophages. Interestingly, a high degree of plasticity characterizes these two cell subtypes, which can easily shift from one phenotype to the other, despite the microenvironmental changes responsible for this phenomenon have not been completely uncovered [15].
One vital microenvironmental component is the extracellular matrix (ECM) since it plays both direct and indirect roles in modulating tumor cell behavior. In the tumor microenvironment, the ECM exerts pleiotropic effects regulating cell adhesion and migration, controlling tumor infiltration by macrophages or other leukocytes, affecting tumor angiogenesis and regulating growth factor availability, as well as receptor signaling [16][17][18][19][20][21][22][23][24]. EMILIN-2 is an ECM molecule belonging to the EDEN protein family [25,26] and exerts a suppressive function in a number of tumor types [27,28]. In the tumor microenvironment, EMILIN-2 affects the activation of the Wnt/β-catenin signaling pathway [29], a key regulator of CRC development, thus down-modulating cell proliferation and migration. Like other members of this family such as Multimerin-2 [30][31][32], EMILIN-2 also influences angiogenesis [27,33,34], triggering IL-8 expression via the EGFR/EGF pathway, thus affecting vessel development and perfusion [35]. EMILIN-2 expression is often down-regulated during tumor progression [36,37], however the impact of this loss has not been explored thus far. We verified that the expression of EMILIN-2 was variable among the CRC patients, thus we sought to evaluate the effects of EMILIN-2 loss in CRC development, also considering its role in modulating the Wnt/ β-catenin signaling pathway [29]. To this end, we took advantage of the Emilin-2 null mouse model. Contrary to our expectations, we found that loss of EMILIN-2 did not affect the activation of the Wnt/β-catenin signaling pathway, instead we unveiled a previously undescribed immunomodulatory function for this molecule. In fact, low EMILIN-2 levels correlated with a low M1/M2 macrophage ratio infiltrating the tumor tissue, and with poor CRC patient prognosis. Based on these results we envision that EMILIN-2 may represent a valid prognostic marker for CRC patients.

Patients
For this study, 23 patients with colorectal cancer were consecutively enrolled. Patients were examined before chemotherapy or surgical intervention. The bioptic samples were obtained with macrobiopy (COOK Medical, Ireland) at the end of examination. The clinical parameters are reported in Table S1.

RT-qPCR, cell transfection and recombinant protein production
Total RNA was isolated from homogenized tissues or cell lines with Trizol and reverse transcribed using AMV-RT (Promega, Milan, Italy). Semi-quantitative endpoint reactions were performed with GoTaq DNA polymerase (Promega, Milan, Italy) and Real-time PCRs with iQ ™ SYBR ® Green Supermix (Bio-Rad, Milan, Italy). The oligonucleotide sequences are listed in Table S2.
Cell transfection was carried out using the FuGene6 reagent (Promega, Milan, Italy). For recombinant EMI-LIN-2 production, E293 cells were transfected with the pCEP-Pu-EMILIN-2 or the empty vector and were then selected with 250 μg/ml G418 and 0.5 μg/ml puromycin. Confluent cells were then incubated in serum-free medium for 48 h and the conditioned medium collected; the His-tagged protein was then purified with Ni-NTA beads (Qiagen GmbH, Hilden, Germany).

MTT and TUNEL assay
10 4 HT-29 or HCT116 cells were plated in 96 well plate, let adhere and incubated with 5 μg/ml EMILIN-2 or PBS for 24, 48 and 72 h. Cells were incubated for 3 h with 5 mg/ml MTT and absorbance detected at 560 nm. The results were reported as % of cell proliferation. The apoptotic rate was determined using the Cell Death Detection ELISA PLUS TUNEL assay (Roche Diagnostics S.p.a., Milan, Italy).

Soft agar colony assays
HCT-116 and HT-29 cells, 5 × 10 3 cells/well, were included in 0.4% low melting-point agarose, placed on top of a 0.6% agarose layer, challenged with EMILIN-2 (5 μg/ml) or PBS and pictures were taken after 10 days. The colony number and area of clones formed by more than 10 cells were measured using Image Tool software.

Migration assay
To evaluate cell migration, Transwell ® chambers (Corning, Inc., New York, NY, USA) were used. 20 × 10 3 HCT-116 and HT-29 cells were seeded into the upper chamber in serum free RPMI containing EMILIN-2 (5 μg/ml) or PBS; serum free RPMI was added to the lower chamber.
The cells were allowed to migrate for 6 h at 37 °C in 5% CO 2 . The migrated cells were fixed with 4% paraformaldehyde and stained with crystal violet (0.5%) and counted.

Adhesion assay
Naive and TPA activated U937 cells (50 ng/ml for 72 h) were kept in serum free RPMI medium overnight. Wells were pre-coated overnight with 5 μg/ml EMILIN-2 or Fibronectin or PBS, followed by incubation in blocking buffer (2% BSA in PBS) for 1 h. Cells were let adhere for 4 h. Unattached cells were washed away with PBS and adhered cells were stained with crystal violet and counted.

Co-immunoprecipitation
THP-1 cells were activated with TPA (50 ng/ml), treated with recombinant EMILIN-2 (2 μg/ml), LPS (1 mg/ml) or PBS. After stimulation, whole-cell extracts were prepared with HNTG lysis buffer and incubated overnight with the anti-MyD88 antibody, together with Protein A/G beads (Santa Cruz). Beads were then washed with PBS and immunoprecipitates were eluted with sodium dodecyl sulfate (SDS) loading buffer, resolved in 4 to 20% Criterion Precast Gels (Bio-Rad, Milan, Italy), transferred onto Hybond-ECL nitrocellulose membranes and further incubated with the appropriate antibodies.

Isolation of bone marrow-derived macrophages (BMmacs)
BMmacs were isolated and differentiated from murine tibias and femurs as previously described [40]. Red blood cells were lysed in ammonium chloride potassium buffer for 4 min and 2 × 10 6 cells/ml were differentiated with 30 ng/ml recombinant M-CSF for 7 days. Following differentiation, the medium was supplemented with IFN-γ and TNF-α (for M1 polarization) or with IL-4 and IL-10 (for M2 polarization) in the presence of recombinant EMILIN-2 (2 μg/ml) or PBS. After 24 h mRNA was harvested with Trizol and processed for expression analyses.

Bioinformatics and statistical analysis
Transcriptomics data from Cancer Genome Atlas of Colorectal Cancer type were downloaded from GDC Data portal (https:// portal. gdc. cancer. gov/ accessed on 12 March 2021) and proteomics data were downloaded from Vasaikar et al. using cBioportal (The cBio Cancer Genomics Portal: An Open Platform for Exploring Multidimensional Cancer Genomics Data, accessed on 12 March 2021) [43]. The EMILIN-2 gene expression profiles were analyzed using a dataset of colon cancer patients from the GEO database (GSE35834) [44]. This dataset reports expression estimations, obtained with Affymetrix Human Genome U133 Plus 2.0 arrays. Cox regression and Kaplan-Meier survival curves were computed using R (version 3.6.1) with survival and survminer packages. Survival curves were compared with the log-rank test. Survival analyses were performed according to Liu et al. [45] from the TCGA-COAD cohort.
The composition and density of immune cells in the tumor microenvironment (TME) was assessed using the immunedeconv tool as described by Sturm et al. [46] and the timer2 tool [47]. EMILIN-2 methylation profile on colorectal cancer was obtained using MEXPRESS (http:// mexpr ess. be) [48].
Statistical analyses were performed with the Sigma-Plot and GraphPad Prism 6 software (Graphpad) and the values represent the mean ± standard deviation obtained with not less than three measurements on randomized samples. The statistical significance of the differences was determined by the two-sided Student's t test for the comparisons between two groups; for more than two groups, the ANOVA 1-way analysis of variance was used, according to the Bonferroni method. For the in vivo experiments, mice were randomly assigned to treatment groups. Differences were considered statistically significant when P ≤ 0.05.

EMILIN-2 levels are down-regulated in the CRC microenvironment
To outline the profile of EMILIN-2 in the context of CRC, we used different omic approaches. The epigenetic modulation of gene expression has a pivotal role in CRC development and is emerging as an important biomarker to guide the therapeutic choices [49]. Thus, based on the finding that epigenetic modifications of EMILIN-2 gene occur in a number of solid tumors [36], we first evaluated the levels of methylation of the EMILIN-2 gene in the TCGA COAD cohort of 477 CRC patients and found that the EMILIN-2 gene carries many methylation sites (Fig.  S1). Furthermore, the MEXPRESS analyses indicated that a higher methylation of the EMILIN-2 gene was associated with higher rate of braf mutations (p = 0.040), higher number of colon polyps (p = 5.7e10 − 6 ), increased lymphatic invasion (p = 0.040) as well as with the pathologic M status (p = 0.015) (Fig. S1). The down-regulation of EMILIN-2 expression in CRC patients was also confirmed at the mRNA level, as assessed in a cohort of 121 patients (GEO ID: GSE3629 [44]), as well as at the protein level in a cohort of 110 patients available at the cBioportal public dataset [50] (Fig. 1 A,B, respectively). Despite overall reduced, we observed a variability of EMILIN-2 among the CRC samples analyzed ( Fig. 1C and Fig. S2). The same dataset was interrogated to verify if the expression of EMILIN-2 associated with known CRC molecular and clinically relevant parameters. We found that the expression of EMILIN-2 did not correlate with the gender of the patients, the stage, histological type, MSI status or stage N of the disease. However, among the various markers taken into account we found that low EMILIN-2 expression associated with high levels of the calcium binding protein S100P (P = 0.0271), whose expression is linked to poor overall survival in CRC patients [51].

Emilin-2 null mice are more susceptible to AOM/DSS induced CRC
To investigate the effects of EMILIN-2 loss in CRC development, we took advantage of the Emilin-2 −/− mouse model [35]. First, we assessed the expression of EMI-LIN-2 in the colonic mucosa and, in wt animals, we found a strong EMILIN-2 signal within the lamina propria surrounding the crypts ( Fig. 2A). As expected, no signal was found in Emilin-2 −/− mice.
To verify if the lack of EMILIN-2 could affect CRC development, we chemically-induced CRC in wt and Emilin-2 −/− mice by means of the AOM/DSS protocol. The AOM/DSS treatment dramatically impacted on . B EMILIN-2 protein levels in healthy (normal; n = 100) and neoplastic (colorectal cancer; n = 97) human samples according to the cBioportal public dataset. C Evaluation of molecular and clinical parameters in relation to the levels of EMILIN-2 from the CRC patients' cohort reported in B. Graphs in A and B represent the mean ± SD; P values were obtained using the paired Student's t-test; ***P < 0.001 the survival of Emilin-2 −/− mice due to excessive intestinal dysfunction resulting in chachexia (Fig. 2B). Mice were monitored weekly and the disease activity index (DAI) was calculated based on weight loss, intestinal bleeding and stool consistency. The DAI indicated that Emilin-2 −/− mice displayed more severe signs of suffering compared to wt animals (Fig. 2C). To assess tumor growth over time, mice were monitored by endoscopy; mice. Graphs in C, F and g represent the mean ± SD; P values were obtained using the paired Student's t-test; **P < 0.01, ***P < 0.001, n.s.: P > 0.05 small polyps were detectable 10 weeks after the beginning of the treatment (Fig. 2D). At 25 weeks from the AOM injection, neoplastic lesions were macroscopically visible in the terminal part of the intestinal tract (Fig. 2E). In agreement with our hypothesis that the absence of EMILIN-2 may result in an increased activation of Wnt/β-catenin signaling pathway, one of the major drivers of CRC growth [29], Emilin-2 −/− mice developed a significant higher number of tumors compared with the wt littermates (Fig. 2D,F). Notably, only 65% of the wt mice developed CRC, while all the Emilin-2 −/− animals displayed one or more tumors (Fig. 2F). The tumor burden was comparable between the two mouse models (Fig. 2G), even if the Emilin-2 −/− mice were characterized by a slightly increased number of lesions with a diameter higher than 0.3 cm, despite the differences were not significant (Fig. 2H).

The exacerbated CRC development in Emilin-2 −/− mice does not hinge on an altered Wnt/β-catenin signaling
Since we had previously shown that EMILIN-2 attenuates the Wnt/β-catenin signaling pathway in breast cancer cells [29], we next verified if tumors from Emilin-2 −/− mice displayed an increased activation of this pathway. Unexpectedly and in contrast to our hypothesis, in tumors from wt and Emilin-2 −/− mice the levels of active β-catenin, as well as the expression of the target genes c-Myc and cyclin D1, were comparable (Fig. S3A, B). Consistently, cell proliferation was similar in tumors developed in the two mouse models (Fig. S3C).
We next hypothesized that EMILIN-2 could influence colorectal cancer cell behavior impinging on other molecular mechanisms. To verify this hypothesis, we used human HT-29 and HCT-116 CRC cells, which do not express EMILIN-2 ( Fig. S4A, B). Contrary to our hypothesis, the treatment with soluble recombinant EMILIN-2 did not affect their viability or apoptotic rate (Fig. S4C, D), nor their capability to form colonies in 3D settings (Fig. S4E, F). These results excluded the possibility of a direct effect of EMILIN-2 on colon cancer cell growth. However, EMILIN-2 inhibited the migration of HCT-116 cells, despite the migration rate of these cells was modest (Fig. S4G), whereas HT-29 cells failed to migrate with or without the treatment (data not shown). These results suggested that EMILIN-2 could directly affect the migration of colon cancer cells rather than the proliferation, and may impinge on metastatic dissemination.

EMILIN-2 impacts on CRC formation affecting the inflammatory response
Ruled out the possibility of a direct action of EMI-LIN-2 in CRC cell growth, based on the fact that the protocol to induce CRC carcinogenesis included the pro-inflammatory agent DSS, we next hypothesized that the increased tumor growth in Emilin-2 −/− mice upon AOM/DSS treatment could depend on an altered inflammatory response. We thus dissected the contribution of the single components administrating either AOM alone or, alternatively, DSS alone to establish a model of sporadic or chronic DSS-induced inflammatory CRC, respectively (Fig. 3).
During AOM treatment, the animals did not show signs of systemic disease, in fact loose stools or bleeding were not detected and their weight increased physiologically (Fig. 3A). At 16 weeks from the first AOM injection, colorectal samples from wt and Emilin-2 −/− mice displayed no significant difference in the number of tumor lesions (Fig. 3B). Both mouse models developed tubular adenomas, tubulo-villous adenomas and adenocarcinomas. Nonetheless, the presence of adenomas displaying villous histological features or severe epithelial dysplasia, both associated to a greater propensity to develop carcinomas, were more frequent in Emilin-2 −/− mice (Fig. 3C).
On the contrary, DSS administration alone failed to induce the formation of CRC lesions in both wt and Emilin-2 −/− mice; however, the treatment was stopped at 12 weeks following the first DSS treatment, instead of the scheduled 16 weeks, due to severe suffering of the Emilin-2 −/− mice. As observed upon AOM/DSS administration, following prolonged DSS treatment Emilin-2 −/− animals were characterized by a much higher DAI compared to wt littermates (Fig. 3D). After 12 weeks, the colonic mucosa of both wt and Emilin-2 −/− mice showed massive and diffuse inflammation with ulcerative lesions, crypt abscesses and glandular architectural distortion (Fig. 3E). In addition, the overall condition of the colonic mucosa was poorer in Emilin-2 −/− mice; the ulcerative lesions were wider, occasionally resulting in necrotichemorrhagic foci of the lamina propria, and in large tracts of the mucosa, the crypt epithelium was replaced by granulation tissue (Fig. 3E). In few Emilin-2 −/− mice, the colon progressively swelled and became gangrenous developing toxic megacolon.
In order to determine if EMILIN-2 expression could be impaired not only during CRC growth, but also in the context of chronic inflammation, we interrogated a cohort of 363 patients affected by chronic intestinal inflammation (GEO ID: GSE83687) [52]. These analyses indicated that EMILIN-2 gene is down-modulated also in the context of inflammatory bowel diseases (IBD) (Fig.  S5). This result suggests that loss of EMILIN-2 may also occur before the IBD-associated CRC onset, thus creating a favorable soil for CRC development.

The colonic mucosa from Emilin-2 −/− mice treated with DSS displays an altered myeloid cell infiltration
Since the AOM/DSS treatment associated with an increased number of neoplastic lesions in Emilin-2 −/− mice, we hypothesized that the lack of EMILIN-2 could lead to poor tumor immunosurveillance, leading to the escape of a higher number of transformed cells. To verify this hypothesis, we analyzed the acute phase of the colonic inflammatory response. Similarly to what observed under chronic DSS treatment (Fig. 3D), during the acute phase of the DSS-induced inflammation, the Emilin-2 −/− mice were characterized by higher weight loss, increased presence of loose stools and bleeding, compared to wt animals (Fig. 4A). Additionally, the Emilin-2 −/− mice displayed reduced colon length, a feature associated to extensive phlogosis (Fig. 4B). Accordingly, inflammation and edema were more pronounced in Emilin-2 −/− mice and also infiltrated the submucosa and the adventitia (Fig. 4C).
Since IFN-γ, IL-12 and IL-6 are key regulators of the inflammatory response [53,54], we verified the expression of these molecules in the colonic mucosa of the  mice treated or not with DSS for 7 days; n = 8. Graphs represent the mean ± SD; in A, B, D, E and F, P values were obtained using the paired Student's t-test; in G, P values were obtained with the One Way Anova Test; * P < 0.05, **P < 0.01, ***P < 0.001, n.s.: P > 0.05 treated mice. Notably, the expression of IFN-γ and IL-12 was lower in Emilin-2 −/− mice compared to wt littermates, whereas that of IL-6, which associates with poor CRC prognosis [55], was higher (Fig. 4D).
Having determined an alteration of the inflammatory cytokines in the two mouse models, we next verified if this associated with an altered recruitment of inflammatory cells. First, we examined the leukocytes content and found that the total number of leukocytes infiltrating the intestinal tract was slightly, but not significantly lower in Emilin-2 −/− mice (Fig. 4E). Next, we hypothesized that the increased number of tumor lesions observed in Emilin-2 −/− mice could be accounted for the escape of tumor cells from immunosurveillance. Indeed, the number of myeloid derived suppressors cells (MDSC), whose presence associates with poor patient prognosis [56,57], was significantly higher in Emilin-2 −/− mice (Fig. 4F). Taken together, these results indicate that the absence of EMI-LIN-2 associates with the onset of an immunosuppressive microenvironment.
To verify if the altered recruitment of inflammatory cells within the tumors was due to unbalanced levels of circulating immune cells, the number of circulating monocytes, lymphocytes and neutrophils was analyzed. Remarkably, as opposed to wt animals, the AOM/DSStreatment did not induce any increase of monocytes, lymphocytes and neutrophils in the peripheral blood of Emilin-2 −/− mice (Fig. S6A). To assess if this was due to an altered maturation of the hematopoietic subpopulations, the cells were isolated from the bone marrow of wt and Emilin-2 −/− mice. These analyses excluded this possibility, since the bone marrow subpopulations were comparable (Fig. S6B).
Hence, we next determined if the abnormal immune cell recruitment could depend on altered cytokines/ chemokines serum levels. In fact, the induction of acute inflammation in wt animals resulted in the increase of key inflammatory cytokines such as IFN-γ, GM-CSF, IL-4, IL-1β, and TNF-α (Fig. S7A). On the contrary, in Emilin-2 −/− mice the treatment did not alter the levels of IL-4, GM-CSF and TNF-α, instead it associated with lower levels of IFN-γ and IL-1β (Fig. S7A). Notably, the expression of the Granulocyte Colony Stimulating Factor (G-CSF), known to be associated with myeloid cell infiltration and linked to a worse outcome in AOM/ DSS-treated mice [58], was strikingly higher with a 6-fold increase in Emilin-2 −/− mice (Fig. 4G). These alterations were also associated with a different pattern of CCL2 and CCL3 (Fig. S7B).

EMILIN-2 loss associates with an altered inflammatory tumor microenvironment
Overall, our observations indicated that tumors developed in Emilin-2 −/− mice were not only more frequent but were also characterized by histological features indicating a higher propensity to progress into adenocarcinomas. Emilin-2 −/− tumors displayed cellular pleomorphism and a poorly differentiated phenotype (Fig. 5A).
To examine if this pattern could be linked to an altered tumor-associated inflammation, we analyzed the inflammatory component associated with established tumors. These analyses indicated that the absence of EMILIN-2 did not affect the recruitment of cytotoxic T cell and Treg cells (Fig. 5B); however, tumors developed in Emilin-2 −/− mice displayed a significant higher number of mononuclear cells mainly consisting of macrophages ( Fig. 5C and Fig. S8A). The massive presence of macrophages was observed in the AOM/DSS (colitis-associated CRC) model, as well as in the AOM-induced (sporadic) model (Fig. S8B), suggesting that EMILIN-2 affects not only the immunosurveillance in the early phases of CRC development, but also the immune environment in established tumors ( Fig. 5C and Fig. S8B).
A rich macrophage infiltrate normally associates with a more favorable outcome [59], in contrast tumors grew more efficiently in Emilin-2 −/− mice, despite they displayed an extensive macrophage infiltrate. We thus and this was in contrast with our preclinical observations, we examined the quality of the macrophage infiltrate and indeed we verified that the balance tilted towards the tumor-promoting M2 phenotype. In fact, the macrophages infiltrating the tumors of Emilin-2 −/− mice were characterized by an altered M1/M2 ratio and displayed a lower number of cells expressing high levels of the M1 marker MHC II (Fig. 5D). The prevalence of M2 macrophages in the tumor tissues from Emilin-2 −/− mice was consistent with a 100% increase of IL-10 (Fig. 5E), one of the main M2-associated cytokines [60].

EMILIN-2 affects the M1/M2 macrophage polarization by triggering TLR-4 activation
To further explore the effects of EMILIN-2 on macrophages' behavior we employed the human monocytic cell line U937. Cells were used under basal conditions or upon TPA-induced differentiation into a macrophage-like phenotype. EMILIN-2 did not affect the TPA-induced activation of U937 cells, their apoptotic rate, nor proliferation (Fig. S9A-D) and did not represent a substrate for U937 cell adhesion or migration (Fig. S9E, F).
To shed light on the role of EMILIN-2 in macrophage polarization, monocytes were isolated from wt and Emilin-2 −/− mice, differentiated into macrophages and stimulated towards the M1 or M2 phenotype (Fig. 6). Upon differentiation, the expression of the M1 markers Nos-2 and TNF-α was lower in macrophages derived from Emilin-2 −/− mice, suggesting that these cells were less prone to polarize towards the M1 phenotype (Fig. 6A). Importantly, treatment with recombinant EMILIN-2 partially improved the M1 polarization in macrophages derived from Emilin-2 −/− mice (Fig. 6A). On the contrary, upon treatment with IL-4 and IL-10, macrophages from Emilin-2 −/− mice expressed higher levels of the M2 makers Chil3 and Arg-1; however, treatment with recombinant EMILIN-2 did not further increase the levels of M2 markers in these cells (Fig. 6B). Similar results were obtained verifying the protein levels of NOS-2 and Arg-1 (Fig. 6 C-E).
To investigate the molecular mechanisms by which EMILIN-2 could affect macrophage polarization we  10) relative to that of GAPDH, in tumors from wild type (wt) and Emilin-2 −/− (E2 −/− ) mice; n = 6. Graphs represent the mean ± SD; P values were obtained using the paired Student's t-test; * P < 0.05, **P < 0.01, n.s.: P > 0.05 higher levels of the M1 marker TNF-α (Fig. 7A), whereas, upon M2 stimulation, the expression levels of the M2 marker CD163 did not change (Fig. 7B), suggesting that H Western blot (left) and quantification (right) of the down-regulation of TLR-4 upon lentiviral transduction with two specific shRNA constructs; pLKO served as a shRNA control (shCTRL). I qPCR analyses of TNF-α expression relative to that of GAPDH in TPA-activated THP-1 cells treated or not with recombinant EMILIN-2 (rE2) upon downregulation of TLR-4. Graphs represent the mean ± SD; P values were obtained using the paired Student's t-test (A, B, D and E) or the One Way Anova Test (F, G, H and I); * P < 0.05, **P < 0.01, ***P < 0.001, n.s.: P > 0.05 also in this cell line EMILIN-2 tilts macrophages towards the M1 phenotype. We next analyzed the pathways involved in this response, assessing the activation of key downstream modulators of macrophage polarization, i.e. NF-κB and STAT1 for M1, and STAT3 for M2 [61]. EMI-LIN-2 stimulation resulted in increased pNF-κB levels (Fig. 7C, D), whereas the levels of pSTAT1 and pSTAT3 did not change (Fig. 7C, E and Fig. S10). TLR-4 is a chief regulator of M1 polarization [61] and, upon engagement, it recruits MyD88 in the intracellular region, and the formation of this complex activates a signaling cascade leading to NF-κB phosphorylation. Thus, to verify if the increased EMILIN-2-dependent NF-κB phosphorylation hinged on TLR-4 activation, the TLR4/MyD88 complex was immunoprecipitated from TPA-activated THP-1 cells challenged with recombinant EMILIN-2 or the LPS positive control (Fig. 7F). In line with our hypothesis, EMILIN-2 significantly increased the levels of TLR-4 co-precipitating with MyD88, similarly to the effect of LPS, despite EMILIN-2 did not significantly alter TLR-4 expression (Fig. S11A, B). These results were corroborated by the use the TLR-4 and NF-κB inhibitors TAK-242 and JSH-23, respectively, which abolished the EMILIN-2-induced polarization towards the M1 phenotype, as assessed through the expression of the TNF-α marker (Fig. 7G). Overlapping results were obtained upon down-modulation of the TLR-4 expression by shRNA (Fig. 7H, I).

EMILIN-2 affects the macrophages' infiltrate and polarization in CRC patients
Given the results obtained in vivo and in vitro, we next aimed at verifying if the levels of EMILIN-2 expression associated with the macrophages' infiltrate and polarization also in CRC patients. To this end, we interrogated the TCGA COAD cohort and found that the levels of EMILIN-2 positively correlated with the expression of the M1 markers CCL3 and CD86, and negatively correlated with the M2 marker IL4R (Fig. 8A). It must be pointed out that the tumor immune microenvironment is characterized by a remarkable complexity, thus a more comprehensive analysis was required to better define which parameters could better describe the relation between EMILIN-2 and macrophages. Among the numerous methods used to estimate immune cells from bulk RNAseq samples derived from the whole tumor [46], we took advantage of TIMER 2.0 resource, which allows a systematical analysis of the immune infiltrate using specific gene signatures. Importantly, in agreement with the data obtained in the in vivo mouse model, we identified a positive association between EMILIN-2 and the M1 macrophage infiltrate, whereas the correlation with the M2 macrophages was inverse (Fig. 8B, C). These results were further confirmed on the TGCA COAD cohort using the recently developed immunodecov tool (P < 2.2e-16 for both M1 and M2).
To verify the results obtained with the in-silico analyses, we exploited 23 biopsies from CRC patients diagnosed and treated in our Institute (Table S1). Bioptic samples were collected from normal and neoplastic sites during diagnostic endoscopy and the EMILIN-2 expression levels evaluated by immunofluorescence. Nineteen normal colonic samples were available for the analyses and overall EMILIN-2 was highly expressed and decorated the normal stroma surrounding the crypts (Fig. 9A). On the contrary, the tumor tissues were characterized by a strong reduction of the EMILIN-2 levels (Fig. 9A). Importantly, tumors displaying high EMILIN-2 levels were characterized by a high number of cells expressing the M1 markers CD86 and HLA-DR, as opposed to those displaying low levels of EMILIN-2, where the presence of IL-4R-and CD163-positive M2 cells was more pronounced ( Fig. 9B and Fig. S12). Of these 23 CRC patients, only for 11 the clinical parameters and the complete blood count were available (Fig. 10A). Thus, for these patients we compared the tumor-associated EMILIN-2 expression levels with the blood count. Despite the low number of patients, EMILIN-2 strongly associated with the number of circulating inflammatory cells also in this small cohort (Fig. 10B). Interestingly, the number of circulating monocytes and, notably, the monocyte/lymphocyte ratio (MLR), which is a predictor of the patients' outcome [62], negatively correlated with EMILIN-2 (Fig. 10C). It is known that patients characterized by a MLR < 0.355 have a better prognosis [62]. Consistently, in the 11 patients cohort from our Institute patients displaying a MLR value lower than 0.355 were characterized by higher EMILIN-2 levels, whereas patients with a MLR higher than 0.355 showed a 6-fold reduction of EMI-LIN-2 (Fig. 10D).

Loss of EMILIN-2 correlates with a poor outcome of CRC patients
Given these observations, we finally assessed if the EMI-LIN-2 expression could associate with the outcome of CRC patients. To this end we evaluated the disease free survival (DFS) exploiting the TCGA COAD cohort [45]. Notably, as shown in Fig. 11A, loss of EMILIN-2 associated with a lower DFS, and a moderate expression of EMILIN-2 correlated with intermediate DFS values (Fig. 11A); more importantly, both high and moderate expression of EMILIN-2 associated with a significantly better outcome of the CRC patients (Fig. 11B,  C). Taken together these results pinpoint EMILIN-2 as an important regulator of the tumor inflammatory microenvironment and suggest that EMILIN-2 could represent a promising prognostic marker for CRC patients.

Discussion
Patients affected by inflammatory bowel diseases are at higher risk of developing cancer and a vast amount of evidence indicate that spontaneous colon cancer is driven by inflammation, thus making this type of tumor a paradigm for the link between chronic inflammation and tumorigenesis [63][64][65][66]. CRC development is characterized by profound changes of the immune cell infiltrate and inflammatory cytokines expression, as well as alterations of the ECM [67]. However, researchers have only scratched the surface towards the understanding of the role of ECM in affecting the immune response. In this study, for the first time we highlight the prominent role of EMILIN-2 in regulating MDSC recruitment and macrophage polarization during CRC development. EMILIN-2 acts as a tumor suppressor in different cancer types [27][28][29] and the finding that the molecule was down-regulated in CRC patients prompted us to hypothesize that it could play a crucial role in this context. Indeed, in a cohort of CRC patients, poor EMILIN-2 expression associated with increased levels of S100P [51], a marker of poor prognosis whose expression together with that of APOBEC3G, CD133, LIPC correlates with increased CRC hepatic metastasis [68].
Accordingly, Emilin-2 null mice displayed increased susceptibility to colitis-associated tumorigenesis and Fig. 11 High EMILIN-2 expression levels associate with favorable CRC patient outcome. A Kaplan-Meier curves reporting the disease free survival of the COAD TCGA cohort including 477 CRC patients, stratified according to high, medium and low EMILIN-2 expression levels. B and C Disease free survival of the same cohort stratified according to high and low or medium and low EMILIN-2 expression levels, respectively. The Hazard ratios are reported on the right. Kaplan-Meier curves were computed using R (version 3.6.1) with the survival and survminer packages and compared with the log-rank test worse outcome. However, here the effect on tumor growth was not regulated directly, as previously demonstrated in other tumor types, i.e. sarcomas or breast [27,29]. In fact, the proliferation and apoptotic rate of tumor cells grown in Emilin-2 −/− mice was not altered, consistent with the results obtained using human colon cancer cells challenged with EMILIN-2. In accordance with these findings, the increased CRC growth in Emilin-2 −/− mice did not hinge on the Wnt/β-catenin signaling. It is conceivable to hypothesize that the regulation of the Wnt ligands' availability exerted by EMILIN-2 [29] may be irrelevant when the APC and/or CTNNB1 genes, the major targets of AOM [39], are constitutively hyperactive. This further highlights the complex regulations wielded by EMILIN-2, which may lead to different effects depending on the peculiar molecular characteristics of the tumor cells and/or the tumor microenvironment. An indication of the mechanism by which EMILIN-2 affected CRC development in mice was suggested by the treatment itself, which included a pro-inflammatory agent. A more detailed analysis of the immune traits suggested that EMILIN-2 shapes both the early and late phases of the immune response.
Although it was not possible to verify similar changes in CRC patients, our preclinical data indicated that in the early phases of CRC development EMILIN-2 up-regulates the expression of IL-12 and INF-γ, key cytokines supporting a tumor-prohibitive microenvironment [69]. In addition, during inflammation also the levels of G-CSF, a key regulator of MDSC cell recruitment [70], were dramatically increased in Emilin-2 −/− mice. Consistently, ablation of EMILIN-2 associated with a higher infiltration of MDSC cells, negative regulators of the tumor immunosurveillance [71]. This resulted in increased number of tumor lesions, suggesting that a higher expression of EMILIN-2 in the tumors may suppress the recruitment of MDSC cells restraining the escape of transformed tumor cells and the formation of CRC lesions. Escape of tumor cells from immunosurveillance not only allows the development and growth of primary tumors, but also the metastatic spread, which is the main cause of cancer-related death [72]. Despite the AOM/DSS-induced lesions hardly ever progresses to metastatic dissemination, in future studies it will be necessary to verify if EMILIN-2 also plays a role in this process.
Alteration of macrophages in Emilin-2 −/− mice were observed both in the sporadic and in colitis associated CRC models. The higher macrophage infiltration present in tumors developed in Emilin-2 −/− mice may be ascribed to low levels of IFN-γ. This cytokine is known to reduce macrophage motility but also to enhance the expression of pro-inflammatory genes [73]. Extensive macrophage infiltration associates with favorable patients' outcome [74], however, an imbalance towards immune-suppressive and tumor-supporting macrophage phenotype can lead to poor outcome [75,76]. This may explain why the Emilin-2 −/− mice displayed a higher number of tumor lesions despite the rich macrophage infiltrate. In agreement with an exacerbated CRC development, the tumor immune infiltrate in Emilin-2 −/− mice was characterized by an imbalance of the M1/M2 ratio, the polarization tilting towards the M1 phenotype. This finding was also confirmed using macrophages isolated from wt and Emilin-2 −/− mice. Mechanistically, we demonstrated that EMILIN-2 prompts M1 polarization by activating the TLR-4/MyD88/NF-κB signaling pathway. In CRC, the TLR-4 holds a prominent, yet controversial, role [61,77,78]. In fact, its expression in CRC cells associates with poor patients' prognosis [78]. On the other hand, this study suggests that the activation of TLR-4 in tumor associated macrophages, which affects their polarization, is influenced by microenvironmental cues and may significantly impact on tumor growth and progression. Accordingly, database analyses of CRC patients demonstrated that loss of EMILIN-2 associated with an increased M2 macrophage infiltrate. These results represent a novelty since to our knowledge no ECM molecule has been shown to regulate macrophage polarization prior to this study. Nonetheless, it is known that the specific composition of the ECM milieu from different tissues can influence the polarization towards an M1 or M2 phenotype [79]. In CRC, the switch of immune-stimulating M1 cells in favor of the immunesuppressive M2-like cells not only foster tumor growth and progression [80] but also promotes resistance to therapy [57,[81][82][83][84]; this possibility suggests that the expression levels of EMILIN-2 may also affect the efficacy of CRC treatment.

Conclusions
Taken together, these results demonstrate that EMI-LIN-2 plays a dual indirect role in CRC development and growth by modulating the inflammatory microenvironment impacting on both the immunosurveillance and immunosubversion phases [14]. Differently from the cognate EMILIN1 molecule, which indirectly affects the immune microenvironment in CRC by acting on the lymphatic vasculature [85], EMILIN-2 influences immune cell activation via a direct mechanism involving the TLR-4 [86].
It is known that, under inflammatory conditions, the ECM landscape is altered by extensive remodeling/turnover, as well as dysregulated expression [87], processes that may also impact on the EMILIN-2 levels. Since the loss of EMILIN-2 is variable among the patients, the analysis of its expression may aid a patient-tailored prognosis. The potential of EMILIN-2 as a prognostic marker is supported by the finding that low levels of EMILIN-2 associate with poor prognosis, as assessed in a cohort of 477 CRC patients. However, these data are based on the EMILIN-2 RNA levels. To verify the suitability of EMILIN-2 as a prognostic marker, it will be worth analyzing also the protein expression in a larger amount of samples. In addition, given the dual role of EMILIN-2 in angiogenesis and immune response, it would also be interesting to verify the levels of the protein in a cohort of patients undergoing anti-angiogenic therapy in combination with immunotherapy, and verify the correlation with the patients' outcome. Taken together these observations further indicate that EMI-LIN-2 may represent a valuable prognostic biomarker for CRC patients.