Impaired SNX9 Expression in Immune Cells during Chronic Inflammation: Prognostic and Diagnostic Implications Free

Chronic inflammation is a common factor underlying various pathologies that differ in their etiology and physiology such as cancer and autoimmune and infectious diseases. As these diseases progress, immune suppression of T cells and NK cells is evident in association with downregulation of the TCR CD247 subunit. This protein is essential for T and NK cell activation and function (1, 2). Various studies, including ours, have shown that chronic inflammation leads to immunosuppression mediated by myeloid-derived suppressor cells (MDSCs). MDSCs comprise a heterogeneous immature cell population coexpressing Gr-1⁺CD11b⁺ in mice and CD11b⁺CD14⁻CD33⁺, LIN⁻HLA-DR⁻CD33⁺ or CD14⁺CD11b⁺ in humans (3–7). In most cases, chronic inflammation leads to complications such as immunosuppression and tissue damage (8), which predispose the host to cancer, opportunistic infections, and limited success of immune-based therapies (2, 9, 10). Currently, the lack of biomarkers that enable distinction between acute (beneficial) versus chronic (harmful) inflammation prevents accurate diagnosis and improved treatment of the above-described diseases. It is therefore of great interest to identify sensitive biomarkers that could predict developing chronic inflammation and immunosuppression to allow better clinical management of these conditions.

In searching for new biomarkers that could indicate an impaired host immune function associated with chronic inflammation, we discovered sorting nexin 9 (SNX9) as a potential candidate that meets this criterion. SNX9 is a ubiquitously expressed protein sharing common structural and functional characteristics with SNX18 and SNX33 (11–13). It participates in early stages of clathrin-mediated endocytosis, coordinates actin polymerization with vesicle release (14), binds the GTPase dynamin, adaptor protein 2 clathrin (15), and activates the actin regulator neural Wiskott–Aldrich syndrome protein (N-WASp) (16). Moreover, SNX9 plays a role in CD28 endocytosis after T cell stimulation and in coupling WASp to p85 and CD28, thus linking CD28 engagement to its internalization, actin remodeling, and cosignaling (16).

In this study, we describe novel characteristics and expression pattern of SNX9 in various cell types during the course of chronic inflammation. To this end, we employed mouse models for human diseases such as chronic infection (Leishmania donovani), cancer (melanoma and colon carcinoma), and autoimmunity (rheumatoid arthritis [RA]). Importantly, we demonstrate the clinical relevance of our observations indicating decreased SNX9 levels in the blood of colorectal cancer carcinoma (CRC) patients, as compared with healthy donors. Additionally, we show the wide range of effects generated by the inflammatory milieu on different cell types and processes, which determine the general host immune status. We suggest that SNX9 could be a useful biomarker for chronic inflammation-induced immunosuppression.

Female/male BALB/c, DBA/1, B10.A, and C57BL/6 mice, 6–8 wk of age, were bred at the Hebrew University specific pathogen-free facility. Animal use followed protocols approved by the Hebrew University–Hadassah Medical School Institutional Animal Care and Use Committee.

Female C57BL/6 mice repeatedly exposed to heat-killed Mycobacterium tuberculosis (bacillus Calmette–Guérin [BCG]) (Difco Laboratories) were used as the mouse model for a pathology-free chronic inflammation as previously described (17). Unless stated otherwise, splenocytes were collected 2 d after the last BCG injection (see Fig. 1A). For experiments analyzing the immune status of mice subjected to a Th2 chronic stimulus, the same protocol was maintained but a chicken OVA (Sigma-Aldrich) was used as an Ag in conjunction with Al(OH)₃ (alum; Thermo Scientific) as an adjuvant (18). The type of the generated immune response (Th1 versus Th2) was analyzed by determining OVA-specific IgG2a and IgG1 isotype concentrations in serum using ELISA according to the manufacturer’s instructions (SouthernBiotech).

Male DBA/1 mice were injected intradermally at the tail base with 200 μg type II collagen purified from bovine articular cartilage emulsified in CFA, and a booster of 200 μg type II collagen emulsified in CFA was injected 3 wk after the first dose (19). The mice were inspected daily with microcalipers, and each animal with erythema and/or swelling in one or more limbs was randomly assigned to one of the following groups: sacrificed at the day of disease onset (day 0), disease peak (day 5), first disease extinguishing (disease peak plus 16 d) or at the second disease extinguishing (disease peak plus 30 d). Unless otherwise indicated, splenocytes were collected from the sacrificed mice for flow cytometry and Western blot analysis. Arthritis was monitored during a 10- to 21-d treatment period in terms of the criteria of the disease (19).

The left hind footpads of female BALB/c mice were inoculated with 10 × 10⁶ stationary phase culture of L. donovani promastigotes in 40 μl PBS i.p. (20). Mice were sacrificed 3 mo after infection and splenocytes were collected for quantifying parasite burden and for flow cytometry and Western blot analysis.

Female C57BL/6 mice were treated with azoxymethane (AOM) as previously described (21), followed by three cycles of dextran sulfate sodium (DSS) treatment. Body weight was measured weekly, and the animals were sacrificed 3 wk after the third cycle of DSS for macroscopic inspection, histological analysis, and assessment of changes in the immune cells.

Mouse melanoma B16 cells (clone MO5, 10⁶/mouse) (22, 23) were injected s.c. into the right rear flank of male and/or female C57BL/6 mice. Animals were weighed twice a week and observed daily. Twenty days after inoculation of the B16-MO5 cells, or when the tumor reached a size of 10 mm, mice were sacrificed, tumors were dissected from the implantation site, the actual tumor size was measured, and splenocytes were analyzed.

Whole-blood samples obtained from healthy donors and CRC patients were analyzed for SNX9, CD247, CD3ε expression, and for MDSC levels by flow cytometry. The protocol for patients’ blood sample collection for research studies was approved by the Hebrew University–Hadassah Medical School Institutional Review Board. The patients were randomly selected by their physicians, based on availability and willingness to participate in this study, and written informed consent was obtained from each donor in accordance with the Declaration of Helsinki protocol. All patients (n = 13) had not been treated at the time of blood donation for this study. Healthy volunteers (n = 13) were recruited among blood donors in the Hadassah University Hospital blood bank facility after signing a written informed consent. In all experiments, patient and control blood samples were processed and handled in the same way, and all samples were always tested at the same time.

For in vivo MDSC depletion, purified anti–Gr-1 (anti-Ly6G/Ly6C rat mAb, clone RB6-8C5, a gift from Prof. Alberto Mantovani, University of Milan, Milan, Italy) were injected (0.5 mg/mouse) i.p. to the inflamed mice 2 d and 1 d before the mice were sacrificed. Splenic MDSC depletion was verified by flow cytometry.

In vivo proliferation of T cells was performed as previously described (17). In brief, CD45.2⁺ C57BL/6 mice were injected with 50 mg CMV-OVA plasmid into the ear pinna by a 31-gauge needle. Seven days later, chronic inflammation was induced in some of the groups and at day +2, after the last BCG injection, CFSE⁺ OT-I CD45.1⁺ splenocytes were i.v. injected into the recipient CD45.2⁺ mice. Four days after cell transfer, mice were sacrificed, spleens were harvested, and CFSE dilution was determined in CD8⁺CD45.1⁺ cells by flow cytometry. For ex vivo proliferation assays, splenocytes labled with CFSE (Invitrogen) were activated with 0.5 mg/ml CD3ε (145-2C11; BioLegend) and CD28 (37.51; BioLegend) Abs as previously described (21). The number of cell divisions of Thy1.2⁺ cells was determined by flow cytometry.

BCG-treated inflamed mice received three i.p. injections of 5-fluorouracil (5FU) (21). The first injection (1 mg/mouse) was given at the day of the second BCG treatment, and the second 5FU injection (1 mg/mouse) was given 4 d later. The third 5FU injection (0.5 mg/mouse) was administered at the day of the third BCG treatment.

MDSCs and T cells from control or inflamed mice were isolated with a magnetic column separation system (Miltenyi Biotec), as previously described (17). For separation of naive B cells, the same selection protocol was applied using CD43 magnetic microbeads. The purity of cell populations was >95%.

Isolated mouse splenocytes and PBLs were subjected to cell surface staining as previously described (17), using the following Abs (BioLegend): FITC-labeled anti–Gr-1, anti–Thy-1.2, PE-labeled anti-CD3ε and anti-CD45R/B220, biotinylated anti-CD11b, and anti-TCRαβ detected with streptavidin-Cy5. Intracellular staining of CD247 was performed as previously described (17) using FITC-labeled or biotinylated anti-CD247 (clone H146), the latter detected with streptavidin-Cy5. For human whole-blood cell phenotyping, intracellular staining of CD247 cells was performed by fixing the cells with 2% paraformaldehyde followed by permeabilization with 0.1% saponin. Allophycocyanin-labeled anti-CD11b and anti-CD3, PE-labeled anti-CD33, FITC-labeled anti–HLA-DR, and anti-CD247 were used (all purchased from BD Pharmingen). After surface staining, cells were treated with one-step Fix/Lyse solution (eBioscience) according to the manufacturer’s instructions. Intracellular staining of SNX9 was performed as for CD247 staining, using PE-labeled anti-SNX9 Ab (mouse mAb 5C7 [IgG2a]). All samples were analyzed in a FACSCalibur apparatus using CellQuest software (BD Biosciences) and analyzed by FCS Express software (De Novo Software).

Cell activation and immunoprecipitations were performed as previously described (24). Immunoprecipitations were done using rabbit polyclonal Abs directed against intracellular epitope of CD247 bound to protein A beads (Amersham Biosciences). For immunoblotting assays, splenocytes, blood cells, or the EL4 T cell line (25) were lysed, and proteins were resolved on 12% reduced SDS-PAGE and subjected to immunoblotting using the following specific Abs: anti-CD247 (H146), anti-SNX9 (a gift from our collaborator Dr. Stefan F. Lichtenthaler, Munich University), anti-SNX27 (Novus Biologicals), and anti-CD3ε (Santa Cruz Biotechnology), followed by protein A (Amersham Biosciences), anti-rat, or anti-goat Abs conjugated to HRP (Jackson ImmunoResearch Laboratories). Detection was performed by ECL using a blotting reader from Bio-Rad.

Isolated T cells were activated with Alexa Fluor 488–labeled anti-CD3ε and purified anti-CD28 for 15 min at 37°C, fixed, permeabilized, and processed as described (26). TCR and SNX9 colocalization was analyzed by confocal microscopy. Splenocytes from normal or inflamed mice were activated with anti-CD3ε Ab alone, fixed, and processed as above for surface labeling. TCR clustering was detected by anti-TCRβ Abs (BioLegend) and phalloidin (Life Technologies).

Normal splenocytes, isolated splenic T cells, or EL4 T cells were coincubated for 18 h at 37°C with an MDSC-enriched population purified from the spleen of inflamed mice (4) at the indicated ratios (see Fig. 3C). Cells were then harvested, lysed, resolved on SDS-PAGE, and subjected to immunoblotting with anti-SNX9, anti-CD247, and CD3ε Abs.

Statistical analyses were performed using GraphPad Prism 5.04. Averaged values are presented as the mean ± SEM. When comparing two groups, statistical significance was determined using a two-tailed Student t test. A p value <0.05 was considered statistically significant.

While searching for new biomarkers that could validate the impaired host immune status under chronic inflammation, we used a mouse model (described in 2Materials and Methods) (Fig. 1A) characterized by elevated levels of MDSCs in the spleen and blood (Fig. 1B). The induced immunosuppression in these mice results in T and NK cell impaired function (17) associated with CD247 downregulation both in the spleen (Fig. 1C) and blood (Fig. 1D). Expression of other TCR subunits such as the CD3ε was unchanged (Fig. 1C, 1D), ensuring that the observed reduction in CD247 expression was not the result of TCR internalization following T cell activation. Although CD247 downregulation is common to functionally impaired immune cells (18), we discovered that SNX9 displays an even more dramatic downregulation in splenic and blood T cells of chronically inflamed mice (Fig. 1C, 1D). Importantly, when comparing SNX9 expression in B and T cells that have been separated from noninflamed control mice (Fig. 1E), a remarkable decline of SNX9 expression was observed in the entire splenic and blood cell populations that were isolated from the inflamed mice (Fig. 1C, 1D), suggesting a bystander effect of chronic inflammation. Interestingly, SNX9, which is a ubiquitously expressed protein, was not detected in MDSCs (Fig. 1E), which are abundant in the spleen and blood of chronically inflamed mice (Fig. 1B). Note that the immunoblotting analysis in the present study was performed using total spleen or blood cells to avoid recovery of SNX9 expression upon cell separation. SNX9 expression is different in various immune cell types, that is, abundant in B cells and absent in MDSCs (Fig. 1E). Therefore, comparison of SNX9 expression in splenocyte or blood cells from normal or inflamed mice is based on the number of cells loaded on the SDS-PAGE corrected according to the percentage of CD3⁺ cells in each sample. To assess whether chronic inflammation exclusively affects SNX9 expression, we measured the expression of SNX18, which is another member of this protein family that shares structural and functional characteristics with SNX9 and SNX33 (27). We found that SNX18 is also downregulated in the spleen and blood of chronically inflamed mice, similarly to SNX9 (Fig. 1C, 1D), suggesting a common regulation of these proteins in immune cells under chronic inflammation.

To define the immunological milieu required for SNX9 downregulation, we tested the effects of Th1 versus Th2 sustained responses. To this end, SNX9 expression levels were compared between OVA-BCG–treated (inflamed) mice, which display a Th1-biased response, and OVA-alum–treated mice that mount a Th2 response. OVA-alum sustained treatment induced a strong Th2 response, as indicated by the presence of IgG1, but not IgG2a, anti-OVA Abs (Fig. 1F). Despite the potent and sustained immune response, the OVA-alum treatment did not induce a profound SNX9 downregulation as observed in the OVA-BCG–treated mice (Fig. 1G). These results suggest that downregulation of both CD247 and SNX9 is induced by a combination of sustained exposure to Ag and the ensuing progression of a Th1 inflammatory response.

To test the kinetics of SNX9 downregulation during the development of chronic inflammation, we analyzed spleens and lymph nodes from mice 2 d after the first, second, and third BCG injections. We found that SNX9 is downregulated in the spleen of chronically inflamed mice (Fig. 2A, top, BCG injection III), when high MDSC levels are evident (Fig. 2A, bottom, BCG injection III). Moreover, upon withdrawal of the BCG stimulus, SNX9 and CD247 expression recovered, along with a reduction in the MDSC levels (Fig. 2A). These results show the reversible features of SNX9 and CD247 downregulation upon recovery from the chronic inflammatory and immunosuppressive stage. Interestingly, in contrast to spleen (Figs. 1C, 2A, top) and blood (Fig. 1D), lymph nodes maintained a constant level of SNX9 (Fig. 2B, top) in correlation with the barely changed MDSC levels (Fig. 2B, bottom), suggesting a protective environment prevailing in the lymph nodes. Note that although MDSC levels increased up to 4.82% in lymph nodes after the third BCG injection, this was insufficient for the induction of CD247 and SNX9 downregulation. The SNX9 expression pattern in the lymphatic organs during chronic inflammation was similar to that of CD247 (Fig. 2B, top) (3). These results suggest that despite their functional disparity, SNX9 and CD247 are similar in their expression pattern during chronic inflammation, yet with a different sensitivity to the generated environment. Moreover, the effect on both proteins inversely correlates with the elevation of MDSCs during the course of chronic inflammation.

The correlation between elevated levels of MDSCs and downregulated SNX9 levels suggests that SNX9 expression during chronic inflammation is regulated by MDSCs. To test this possibility, inflamed mice were depleted of MDSCs by injecting them with anti–Gr-1 Abs (17, 28). The in vivo depletion of MDSCs, verified by flow cytometry analysis (Fig. 3A), correlated with the recovery of SNX9 and CD247 expression in the spleen (Fig. 3B), indicating an improved immunological status. We next assessed whether SNX9 expression levels correlate with T cell function as determined by Ag-induced cellular proliferation in vivo. To this end, mice were immunized with a CMV-OVA plasmid and then subjected to a treatment that induces chronic inflammation (Fig. 1A). At the peak of the inflammatory response the mice were injected with CFSE-labeled OT-I splenocytes (17) followed by depletion of MDSCs by injection of anti–Gr-1 Abs. Four days later, the in vivo proliferative response of splenic CD8⁺ OT-I cells was analyzed. As seen in Fig. 3C, the impaired proliferative response observed in inflamed mice was MDSC mediated, because in vivo depletion of MDSCs completely restored the ability of CD8⁺ OT-I cells to proliferate (Fig. 3C). To determine whether MDSCs directly affect SNX9 expression in spleen cells, we performed ex vivo coincubation experiments, taking advantage of the fact that MDSCs do not express SNX9 (Fig 1E). We observed that MDSCs isolated from spleens of inflamed mice and coincubated with total spleen cells (Fig. 3D), splenic T cells (Fig. 3E), or EL4 cells (Fig. 3F) at different ratios induced massive SNX9 downregulation. The same results were obtained when MDSCs were coincubated with either LK B cell line (29) or B cells isolated from normal spleens (data not shown). We also examined the effect of MDSCs on the expression of SNX27, a protein that belongs to the “sorting nexins family” but is unrelated to the subgroups of the “SNX9 family.” We observed that contrary to SNX9, SNX27 expression is not downregulated but slightly increases in the presence of MDSCs, most likely due to its expression in the latter cells (Supplemental Fig. 1). These data reinforce our conclusion that CD247 and SNX9 are both affected by the MDSC-induced immunosuppressive environment, and that SNX9 is more sensitive than CD247 to the generated inflammatory miliue as depicted in Figs. 1C, 1D, 2A, and 3.

We next assessed whether change in SNX9 expression is a hallmark of pathologies characterized by chronic inflammation and, thus, may reflect the immune status of the host. To this end, we measured SNX9 expression levels at different stages of RA using the collagen-induced arthritis (CIA) mouse model. A dramatic SNX9 downregulation was observed already at the onset of disease (Fig. 4A, top). This reduction was maintained at the peak of the disease followed by a recovery in expression only 30 d afterward (Fig. 4A, top). In contrast, CD247 expression (an internal control) was gradually downregulated and then recovered earlier than SNX9 (16 d after the peak) (Fig. 4A, bottom). In all disease stages CD3ε expression level was stable (Fig. 4A, middle), pointing to a unique downregulation of SNX9 induced by chronic inflammation, which also affects CD247 but to a lesser extent. These results indicate that the inflammatory response generated in CIA leads to SNX9 downregulation much earlier and more extensively than that of the CD247. Moreover, we demonstrate that SNX9 downregulation is reversible because it recovers upon disease regression. SNX9 expression levels inversely correlate with MDSC levels; it recovers upon recuperation of the inflammatory environment, as indicated by the decreased MDSC proportions found in spleens of the experimental mouse groups (Fig. 4B).

During active leishmaniasis, high levels of IFN-γ and TNF-α are detected in the serum, suggesting a strong immunoinflammatory response (20). We have previously reported that these two cytokines play a key role in MDSC generation and recruitment (IFN-γ) (18) as well as in their differentiation arrest and increased suppressive features (TNF-α) (17). Accordingly, we hypothesized that parasite persistence is associated with chronic inflammation and immunosuppression. We therefore investigated whether SNX9 expression levels fluctuate in mice inoculated with L. donovani promastigotes. The results revealed that under the pathological conditions generated during 3 mo of L. donovani infection, SNX9 expression levels were significantly downregulated as compared with the pronounced but less dramatic reduction of CD247 expression (Fig. 4C). In contrast, CD3ε expression levels remained unchanged (Fig. 4C), similarly to RA, and downregulation of both proteins was inversely correlated with MDSC levels (Fig. 4D). The total loss of SNX9 expression observed in the spleen (Fig. 4C) suggests a bystander effect of L. donovani on various immune cell types as seen in the aforementioned model characterized by chronic inflammation (Fig. 1).

Our previous studies demonstrated that chronic inflammation leads to immunosuppression, as observed in humans and mice with developing tumors; MDSCs accumulate at the tumor site and later in the periphery, leading to T and NK cell dysfunction associated with CD247 downregulation (3, 4, 21). To investigate whether SNX9 expression levels are also affected during tumor development, we used mouse models for CRC and melanoma. To establish a model of CRC, mice were injected twice with AOM followed by three cycles of DSS administration in the drinking water, as described in 2Materials and Methods. Repeated DSS administration instigates chronic inflammation, thereby mimicking inflammatory bowel disease, which greatly enhances the incidence of AOM-induced tumors (30). Splenocytes from mice bearing colorectal tumors were first analyzed for SNX9, CD247, and CD3ε expression levels. SNX9 expression is dramatically decreased whereas CD247 is only slightly downregulated (Fig. 4E). Spleens of mice with colorectal carcinoma had higher percentages of MDSCs as compared with normal animals (Fig. 4F), inversely correlating with the affected SNX9 expression levels. We next assessed SNX9 expression in mice inoculated with B16 melanoma cells (31, 32). When analyzing their splenocytes for SNX9, CD247, and CD3ε expression, we found that in mice with B16 melanoma tumors weighing an average of 0.1 g, a decrease in SNX9 expression levels was evident (Fig. 4G), which inversely correlated with MDSC levels (Fig. 4H). Importantly, note that in the latter animal model although the increase in MDSC levels was the lowest as compared with the various models used in this study, it was sufficient to induce a significant decrease of SNX9 expression, whereas CD247 expression was yet unaffected. These results indicate that SNX9 is much more sensitive than CD247 to the MDSCs in the chronic inflammatory environment.

We have concluded from the above-described data that SNX9 may serve as potential biomarker for evaluating the hosts’ immune status under chronic inflammation. To further elaborate this possibility, we investigated whether changes in the host’s immune environment following treatments with drugs affecting MDSC levels and suppressive function could be sensed by SNX9. 5FU has been demonstrated to have a direct cytotoxic effect on MDSCs and also affect myeloid cell differentiation, thus decreasing their numbers and suppressive features (21, 33). We therefore treated chronically inflamed and immunosuppressed mice with 5FU and analyzed the expression of SNX9 and CD247 in their spleen, in association with the developing chronic inflammatory environment. Our results show that alongside with the decreased MDSC levels (Fig. 5A), 5FU treatment led to the recovery of SNX9 and CD247 expression levels, as compared with their expression in inflamed untreated spleens (Fig. 5B). Moreover, the T cell proliferation capacity following 5FU treatment was restored (Fig. 5C), suggesting a recovery of the host’s immune function. Taken together, these results indicate that SNX9 is a reliable biomarker sensing the immune function under chronic inflammatory environment, as well as for monitoring the efficiency of a given therapy in regulating MDSCs and associated immunosuppression.

To investigate the role of SNX9 in T cells, primary T cells were unstimulated or stimulated in vitro with anti-CD3ε and anti-CD28 Abs. Our results showed that in activated T cells SNX9 colocalizes with the TCR clusters formed upon activation, and polarizes to the same subcellular site (Fig. 6A), whereas it remains evenly distributed throughout the cytoplasm in resting cells. These results are further supported by the image in Supplemental Fig. 2, where SNX9 was coimmunoprecipitated with CD247 in EL4 cells. Furthermore, this interaction was markedly increased after anti-CD3 and anti-CD28 stimulation. As our results pointed to the possible association between SNX9 and the TCR via the CD247, and the expression of these two molecules is impaired in T cells during chronic inflammation, we next tested TCR clustering and immunological synapse formation under chronic inflammatory and immunosuppressive conditions. To this end, splenocytes from healthy and chronically inflamed mice were activated with anti-CD3 Abs and analyzed by immunofluorescence. Whereas clustering of the TCR and the formation of the immune synapse were observed after activation of splenocytes from healthy mice, these were not detected in cells from chronically inflamed mice (Fig. 6B).

To examine whether the decrease in SNX9 expression levels in the above-tested mouse models could be applied to human pathologies characterized by chronic inflammation, we examined expression of SNX9 in patients suffering from CRC, which is known to be associated with inflammation accompanied by elevated levels of MDSCs and CD247 downregulation (21). To this end, we first generated a new mouse mAb directed against human and mouse SNX9 suitable for flow cytometry to establish a quantitative method for testing this potential biomarker. This mAb can detect SNX9 downregulation on isolated splenocytes from inflamed mice as compared with normal, noninflamed mice (Supplemental Fig. 3). We next preformed a flow cytometry analysis using the newly generated Ab, testing whole blood samples obtained from 13 CRC patients and 13 healthy donors (controls) for SNX9 expression. Our results revealed a significant reduction of SNX9 expression in the CD3⁺ T lymphocytes of the tested CRC patients (Fig. 7A) similar to the decrease in CD247 expression (Fig. 7B). We also detected a decreased trend in SNX9 expression in the non–T lymphocyte population from CRC patients compared with the cells from healthy donors (Supplemental Fig. 4). As in the mouse models, CD3 expression did not change significantly between the groups examined (Fig. 7C). Furthermore, parallel analysis of the MDSC levels, gating on HLA-DR⁻CD33⁺CD11b⁺ cells, revealed a significant increase in the percentage of these cells in the CRC patients relative to healthy donors (Fig. 7D). These results indicate that in CRC patients circulating immune cells are negatively affected by the generated chronic inflammatory environment as sensed by the SNX9 expression levels.

T and NK cell immunosuppression associated with CD247 downregulation have been established as the typical outcomes of chronic inflammation in numerous pathologies (1). MDSC accumulation during the course of chronic inflammation accounts for the subsequent immunosuppression induced by a variety of mechanisms, including depletion of l-arginine (34, 35), nitration and nitrosylation of TCR and CD8 (36), and downregulation of CD247 that results in cryptic TCR and natural cytotoxicity receptor structure and function (37–40). Immunosuppression, accompanied with exacerbated inflammatory response, leads to complications such as irreversible tissue damage (8), opportunistic infections (41), and cancer development (21). In many of these cases, patients are subjected to a variety of treatments whose success depends on a functional immune system such as immune-based therapies for cancer. Although such therapeutic strategies are widely used, their success rates are low, most likely due to the evolving immunosuppression (42). The complexity of the harmful milieu generated in the course of chronic inflammation raises many unresolved concerns related to the molecular changes induced in immune cells that lead to their impaired function. Thus, the need for biomarkers that could independently evaluate the host’s immune status and predict therapy-associated complications is vital. These issues are closely related because molecules of affected hosts’ cells could serve as the desired biomarkers. Therefore, expanding our understanding of how broad is the effect of a developing chronic inflammation on the overall hosts’ immune profile and function could provide new targets for diagnosis and therapy.

We previously showed that CD247 can serve as a biomarker capable of distinguishing between chronic and acute inflammation and sensing the host’s immune status in diseases characterized by chronic inflammation (1, 42). While searching for additional molecules/pathways that could be used as biomarkers for impaired host immune status and cross-validate the generated harmful conditions, we discovered SNX9 as a novel molecule affected by chronic inflammation. SNX9 is known to play key roles in the endosomal sorting system that regulates intracellular trafficking of plasma membrane receptors (27) and thus is required for basic processes common to different cell types. It was recently reported that the SNX9 gene is silenced in several human malignancies (43–45), suggesting that it may also have a tumor-suppressor function. SNX9 was also proposed to play a role in early mitotic stages (46). Moreover, Badour et al. (16) demonstrated SNX9–WASp interactions in T cells and suggested that the formation of WASp/SNX9/p85/CD28 complexes could enable a unique interface of endocytic, actin-polymerizing, and signal transduction pathways required for CD28-mediated T cell costimulation (16). Our results using immunofluorescence studies confirm the possible role of SNX9 in the T cell activation processes. As SNX proteins have been found to bind proteins involved in T cell activation (47, 48), it is possible that SNX9 plays a role in T cell stimulation such as endocytosis of the surface TCR. These findings highlight SNX9 as a critical regulator of key processes in various cell types, including the activation of T cells and the resulting immune responses and inflammation.

In the present study, we demonstrate that SNX9 expression levels in peripheral blood and spleen are dramatically downregulated in the course of a Th1-mediated chronic inflammatory response associated with immunosuppression, as evident in mouse models for inflammation, infection, cancer, and autoimmunity, all resembling human diseases. We further show by in vivo MDSC depletion and ex vivo coincubation experiments that SNX9 downregulation is mediated by MDSCs, which are devoid of SNX9 expression. Our results depict SNX9 as a complementary biomarker to CD247 because it is a more sensitive indicator for the initial stages of immunosuppression. It is downregulated earlier than CD247 even when there is only a slight increase in MDSC levels as evident during the onset of RA, L. donovani infection, and B16 tumor formation. In contrast, CD247 expression is more sensitive to the recuperation of the inflammatory and immunosuppressive environment, which can, in turn, predict disease regression/therapy success.

Regardless of the different etiology of the mouse models used for diseases characterized by chronic inflammation, the same tendency of decreased SNX9 and CD247 expression levels were observed. Additionally, PBLs of CRC patients display low SNX9 expression levels, which inversely correlate with MDSC levels. These observations strengthen the clinical significance of our results.

Our study suggests that a combined monitoring of SNX9 and CD247 expression levels could serve as an “optimal sensing” tool for defining the severity of chronic inflammation and induced immunosuppression, as well as for the recovery features of the host’s immune system from such of the deleterious conditions. The results presented in the present study also highlight the extended effect of chronic inflammation on the expression of two target molecules operating in different, but not mutually exclusive, cellular pathways: CD247 as a TCR and natural cytotoxicity receptor signaling molecule, and SNX9 as a molecule involved in the endocytic pathway of numerous cell surface receptors. These wide effects of chronic inflammation weaken the hosts’ immune system that leads to disease progression or failure of therapies depending on a functional immune system.

Using the pathology-free chronic inflammation model, we demonstrated that SNX9 downregulation during a sustained inflammation is a normal outcome of the immune system aimed at controlling an excessive and potentially hazardous immune response. The downregulation of SNX9 could lead to a reduced endocytosis of immune cell surface receptors (e.g., TCR) and receptors affecting T cell metabolism (e.g., transferrin receptor), thus periodically attenuating the response to prevent irreversible tissue damage. As shown for RA in the present study, mice gained normal SNX9 expression levels simply by stimulus withdrawal and a “rest time” of 30 d. Indeed, in some RA patients there is a pattern of a periodic disease presentation with swelling in one or two joints that may last a few days to weeks, then completely recuperating and reappearing in the same or different joints (49). This pattern most likely reflects the cycles of autoimmune cell activity and an inflammatory response, followed by MDSC recruitment and immunosuppression, silencing the response to avoid tissue damage. This is followed by a new burst of autoimmune response and a repeating disease cycle. In contrast, for cases in which chronic inflammation is elicited by continuous presence of a pathogen, autoantigen, or tumor Ags, a recovery will be most likely prevented due to inefficient Ag clearance. In vivo MDSC manipulation by their depletion (17, 21, 33) or clearing the inflammation-inducing stimulus can lead to SNX9 recovery. This type of pathophysiology positions SNX9 as a biomarker sensing the fluctuation of the host’s immune status and the chronic state of the inflammatory response and highlights the reversibility of this phenomenon.

In the course of our studies although focusing on SNX9, we found that the SNX9-related protein, namely SNX18, acts in a similar way under chronic inflammatory conditions, whereas SNX27, which belongs to a different sorting nexin family, was unaffected. These results suggest a general effect of the inflammatory environment on the SNX9 sorting nexin family proteins, which share some redundant properties in modulating endocytic trafficking of essential plasma membrane receptors (50).

The mechanisms by which activated MDSCs confer their immunosuppressive effect on the immune cell network under chronic inflammation remains to be determined. However, note that the SNX9 dramatic downregulation during chronic inflammation and associated diseases was observed in the entire spleen and blood cell populations, which included a high percentage of B cells normally expressing SNX9. This suggests that SNX9 downregulation is not specific only to T and NK cells, but it is a general phenomenon affecting a wide range of immune cells. Interestingly, our results point to a possible role of SNX9 in TCR clustering that could affect immunological synapse formation in T cells and along this features could be a key player in receptor clustering and immune synapses formed by B and NK cells. Moreover, based on Badour et al. (16), SNX9 activates the actin regulator N-WASp and plays a role in CD28 endocytosis after T cell stimulation and in coupling WASp to p85 and CD28, thus linking CD28 engagement to its internalization, actin remodeling, and cosignaling. Linking both Badour et al. results and our observations strongly suggest the involvement of SNX9 in TCR clustering and possibly in immunological synapse formation. Therefore, it is likely that SNX9 plays a crucial role not only in controlling a balanced immune response, but also in keeping the cells “fit” in a temporarily enforced resting state, owing to reduced endocytosis of essential nonimmune receptors (e.g., TfR). This mechanism prevents a waste of systemic energy until the restoration of a normal immune state, as depicted in our models. However, further studies are required to examine these possibilities (47, 48).

The knowledge gained from our studies on the unique properties of SNX9 highlights its use as a biomarker for monitoring chronic inflammation and associated immunosuppression. Currently, diagnosis of an abnormal immune system is done retrospectively, based on developing complications associated with immune suppression. The currently used parameters are measurements of C-reactive protein, erythrocyte sedimentation rate, complete blood count, and levels of autoantibodies and proinflammatory factors (IL-1, IL-6, IL-8, and TNF-α). However, neither can distinguish between acute and chronic inflammation, nor allocate chronic inflammation–induced immunosuppression. Because the success of immune-based therapies depends on a functional host’s immune system, there is an urgent need for biomarkers that will enable evaluation of the host’s immune status prior to treatment and sense the effect of modalities neutralizing the inflammatory environment to increase immune system responsiveness. Based on our data, we propose that a combined measurement of SNX9 and CD247 expression levels could serve as a valuable validation tool to ascertain the immune and disease status of patients suffering from diseases associated with chronic inflammation and enable the development of optimal personalized treatments to alleviate the disease.

We thank the Society of Research Associates of the Lautenberg Center, the Concern Foundation of Los Angeles, and the Harold B. Abramson Chair in Immunology for support. We thank Dr. Stefan F. Lichtenthaler from Munich University for providing the anti-SNX9 Abs; Prof. Alberto Mantovani from the University of Milan School of Medicine for providing the anti–Gr-1 Abs; and Dr. Keren Or Amar, Dr. Lora Ashkar, and Prof. David Naor from the Lautenberg Center for help in setting the RA mouse models system. We thank Prof. Ayala Hubert from the Hadassah Medical Center for providing the blood samples of the CRC patients, Prof. Charls Yaffe for providing the mice infected with L. donovani, Dr. Juan Bonifacino for advice on this project, and Prof. Eitan Yafe-Nof for reviewing this manuscript.

The authors have no financial conflicts of interest.