Calnexin Impairs the Antitumor Immunity of CD4⁺ and CD8⁺ T Cells Free

Although T-cell immunity plays a critical role in mediating antitumor immunity, the molecular mechanisms underlying impaired antitumor T-cell immunity are not fully understood. Immune checkpoint blockade with mAbs directed against the inhibitory immune receptors CTLA-4, PD-1, and PD-L1 has emerged as a successful treatment approach that has shown striking antitumor activity in a variety of cancer types (1–3). Currently, there are more than 10 CTLA4, PD-1, and PD-L1 antibodies in different stages of clinical trial in tumors. Despite the durable response rates observed with cancer immunotherapies, the majority of patients do not benefit from the treatment. A “hot” tumor microenvironment, typified by an increased number of CD8⁺ CTLs and PD-L1–positive cells, has been identified to be a reliable predictive biomarker of response to immune checkpoint blockade (4–6). Some new strategies have been developed to convert immunologically “cold” tumors into “hot” tumors. Examples of these strategies include metabolic reprogramming of T cells or modulation of the gut microbiome (7, 8). Still another approach needs to be developed based on novel insights into T-cell responses and immune systems.

Endoplasmic reticulum (ER) chaperones, including BiP/GRP78, calreticulin, calnexin, GRP94, and ERP57, are a large family of proteins that have been discovered to have many important roles in maintaining ER homeostasis and contributing to cancer cell survival and progression. The correlation between ER chaperone expression and tumorigenesis has been extensively studied in various cancers, and most reports have indicated that these proteins promote the proliferation, migration, and attachment of cancer cells (9–13). The ER chaperone calnexin, which is an ER-specific type I transmembrane protein, regulates the folding and quality control of newly synthesized glycoproteins (14). Calnexin can escape from the ER and be transported to the plasma membrane or released into the extracellular space by interacting with glycoproteins (15–17) or via its phosphorylation at two serine residues (Ser554/564) by protein kinase CK2 (18). Some studies have found that extracellular calnexin could be involved in innate and adaptive immunity in non-mammals (19–21), but the impact of surface or secreted calnexin on the human immune system has not been reported. The results from a lung cancer patient cohort provided evidence that calnexin may be a serodiagnostic marker for lung cancer (15).

In this study, we found that calnexin inhibits the infiltration and effector functions of T cells and enhances the expression of PD-1 on T cells. Membrane expression of calnexin was positively correlated with poor prognosis of patients with oral squamous cell carcinoma (OSCC). Our work thus uncovers a mechanism by which antitumor responses are regulated by calnexin expressed in tumor cells and suggests that calnexin may serve as a target for antitumor immunotherapy. Elucidation of these mechanisms will reveal clues as to the next steps that need to be taken to potentially overcome resistance to immunotherapy.

For Western blot analyses, 8 pairs of primary OSCC samples and corresponding normal oral epithelial tissues were obtained during surgeries at the Hospital of Stomatology, Sun Yat-sen University (Guangzhou, Guangdong, P.R. China). For real-time PCR analyses, 33 pairs of primary OSCC samples and corresponding normal oral epithelial tissues were obtained during surgeries at the Hospital of Stomatology, Sun Yat-sen University (Guangzhou, Guangdong, P.R. China). For IHC and immunofluorescence analyses, the expression of calnexin was investigated using a tissue microarray (TMA) containing samples from 357 patients with primary OSCC who were treated at the Hospital of Stomatology, Sun Yat-sen University (Guangzhou, Guangdong, P.R. China), between January 2007 and January 2009. All specimens were confirmed histologically by hematoxylin and eosin staining, and tumor tissue was present in more than 80% of the specimens. The follow-up interval was calculated from the date of surgery to the date of death or last clinical evaluation. The detailed information of these patients is described in Supplementary Table S2. This study protocol was approved by the Institutional Review Board of the Hospital of Stomatology, Sun Yat-sen University (Guangzhou, Guangdong, P.R. China) and was conducted in agreement with the Helsinki Declaration, and written informed consent was obtained from all study participants.

The HSC-3 cell line was purchased from the cell bank of the Japanese Collection of Research Bioresource (JCRB). SCC15, SCC25, CAL27, B16F10, and MB49 cells were purchased from the ATCC. The normal keratinocyte cell line (NOK-SI) was kindly provided by J. Silvio Gutkind (UCSD, San Diego, CA). Cell lines used in these experiments were passaged a maximum of four times before the experiments. Cells were tested for Mycoplasma contamination and identified by short tandem repeat.

C57BL/6 mice were purchased from the experimental animal center of Sun Yat-sen University (Guangzhou, Guangdong, P.R. China). NOD-Prkdcem26Cd52Il2rgem26Cd22/Nju (NCG) mice were purchased from Nanjing Biomedical Research Institute of Nanjing University (Nanjing, China). All experiments were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University (Guangzhou, Guangdong, P.R. China) and performed following local rules.

Single-cell preparations were stained with antibodies purchased from eBioscience, BD Biosciences, and BioLegend. Isotype-matched control mAbs were used. Intracellular staining was done using a Foxp3/Transcription factor staining kit or Intracellular fix & Perm Set according to the manufacturer's instructions (BD Biosciences). Briefly, cells were stimulated with 50 ng/mL PMA (Sigma-Aldrich) and 5 μg/mL ionomycin (Sigma-Aldrich) in the presence of GolgiPlug (BD Biosciences). After 4 hours, cells were stained for dead cells using a FVD dye (eBioscience) and surface markers then fixed prior to intracellular staining for cytokines. Data were analyzed using a FACSVerse flow cytometer (BD Biosciences) and using FlowJo software (Tree Star). The following antibodies were used for flow cytometry: anti-mouse CD4 (clone GK1.5); anti-mouse CD8a (clone 53-6.7); anti-mouse CD3 (clone 17A2); anti-mouse/rat Foxp3 (clone FJK-16s); anti-mouse CD45.2 (clone 104); anti-mouse ki-67 (clone 16A8); anti-mouse TNFα (clone MP6-XT22); and anti-mouse IFNγ (clone MOB-47).

CNX-knockdown (sh-CNX) cells or control cells (6 × 10⁶) were injected subcutaneously into right flank of NCG mice, and 1 × 10⁷ human peripheral blood mononuclear cells (PBMCs) were injected via the tail vein after tumor implantation. The animals were monitored for tumor formation every 2 days and euthanized 3 weeks later. Tumor length (L) and width (W) were measured at the end of the experiment, and tumor volume was calculated by the formula (L × W²)/2. Serum cytokines were analyzed at the indicated time points, and human CD3⁺ T cells were counted after tumor dissociation.

B16F10 tumor cells were retrovirally transduced with sh-CNX or a control and selected with puromycin (3 μg/mL). For tumor vaccination, naïve C57BL/6 mice were immunized with 1 × 10⁶ irradiated B16F10 (1 × 10⁴ rad) cells that were inoculated subcutaneously into the left flank. On day 14, the vaccinated mice were challenged with live transduced tumor cells that were inoculated subcutaneously into the right flank. A CNX-Ig fusion protein or Flag-Ig (200 μg) was injected intraperitoneally (i.p.) into each mouse once a week. Tumor growth was monitored every 2 days. The mice were euthanized when the tumor size reached 15 mm diameter.

Tumor-infiltrating leukocytes (TILs) from the xenograft tumors were prepared according to the protocol described previously (22, 23). Briefly, tumors were dissected and homogenized using a GentleMACS dissociator (Miltenyi Biotec), digested with 0.05% collagenase IV (Sigma-Aldrich), 0.002% DNase I (Roche) at 37°C for 1 hour prior to centrifugation on Percoll density gradient (40%–80%), and the TILs were washed and resuspended in RPMI.

HSC3 cells were transfected with CNX shRNA or empty vector (Genechem) using Polybrene. The cells were trypsinized and replated in 0.5 μg/mL puromycin 48 hours after transfection. Two months later, the puromycin-resistant stable line was established and maintained in medium with 1 μg/mL puromycin. The transfected cells were incubated for 24 hours and harvested for real-time PCR and Western blot analysis.

OSCC tumor cells were isolated from fresh specimen; single-cell suspensions were obtained as described above. Human PBMCs from 6 healthy donors and 8 patients with OSCC were density-enriched by Ficoll (TBD). Tumor antigens were prepared as described previously (24). Briefly, 2 × 10⁷ tumor cells were subjected to four freeze (liquid nitrogen) and thaw (37°C water bath) cycles to obtain a crude lysate as tumor antigen. After removal of large particles by centrifugation and sterilization by filtering (0.22 μm), the protein concentration in the supernatant was measured (Coomassie blue protein assay kit, Thermo Scientific) and aliquots stored at −80°C until use. A total of 1 × 10⁵ PBMCs were stimulated for 48 hours in the presence or absence of tumor lysate from the same patient and 1 μg/mL PHA-M (Sigma Aldrich). GolgiPlug was added for 5 hours to the cells in culture. After 5 hours, cells were stained for viability using a FVD dye (eBioscience), and surface markers were then fixed prior to intracellular staining for cytokines using Transcription Factor Buffer Set and Fixation/Permeabilization Solution Kit (BD Biosciences) as per the manufacturer's instructions (25). The culture supernatants were collected for examination by a human Th1/Th2 cytometric bead assay (BD Biosciences) to quantitatively measure IL2, IL4, IL5, IL10, TNFα, and IFNγ protein levels. Briefly, test samples (50 μL) and PE detection antibody were incubated with capture bead reagent for 3 hours in the dark at room temperature. All unbound antibodies were washed (1.0 mL wash buffer) and resuspended in 300 μL before acquisition on FACSVerse Flow Cytometer (BD Biosciences).

Tissues were deparaffinized and rehydrated prior to antigen retrieval in citrate buffer. Tissues were stained with anti-calnexin (C5C9, Cell Signaling Technology), anti-CD3 (17A2, eBioscience), anti-CD4 (OKT4, eBioscience), anti-CD8 (HIT8A, eBioscience). Horseradish peroxidase staining was visualized with 3-39 diaminobenzidine (Gene Company), and fluorescence staining was visualized with Alexa Fluor–conjugated secondary antibody (Thermo Scientific). Slides were counterstained, cleared, and mounted.

Each TMA staining result was confirmed by a tissue slide from the same patient. The slides were scanned using an Aperio Scan Scope AT Turbo for digital image analysis. The images were blindly reviewed and scored by two certified anatomic pathologists. Calnexin expression was defined as cytoplasmic or membranous based on its immunoreactivity. As previously described (26–28), cytoplasmic calnexin staining in tumor cells was evaluated using the staining–intensity distribution (SID) score. The staining intensity of positive tumor cells was categorized into three grades by comparing the intensity with that of internal controls: 0, negative staining; 1 (weak), lighter than skeletal muscle; 2 (moderate), equal to skeletal muscle; and 3 (strong), more intense than skeletal muscle. The distribution of positive tumor cells was graded as follows: 0, no stained cells; 1, <25% stained cells; 2, 25%–50% stained cells; and 3, >50% stained cells. After multiplying the distribution score by the intensity score in eight different high-powered image fields, the average of the eight fields was the SID score for the sample. Cytoplasmic calnexin expression was categorized into low and high expression groups using the median of SID scores of the total patients. Although upon analysis, calnexin membranous immunoreactivity correlated with overall survival, we defined extensive staining as positive and no staining or sporadic staining as negative.

Human CD3⁺ T cells were isolated from PBMCs of healthy donor using pan T-cell isolation kits (Stem Cell Inc.). Then, labeling was performed by incubating 10⁶ cells/mL at 37°C for 15 minutes with 5 μmol/L CFSE in PBS. Carboxyfluorescein diacetate succinimidyl ester (CFSE) was quenched by adding twice the volume of complete media, followed by three washes in complete media before stimulation. 96-well flat-bottom plates were coated with 1 μg/mL anti-CD3 (clone OKT3) at 4°C overnight. The wells were washed three times with PBS to remove unbound antibody and coated with another 5 μg/mL (ratio 1:5) calnexin-Ig or control-Ig protein in PBS at 4°C overnight. Wells were washed three times with PBS before adding cells. Replicate cultures (1 × 10⁵ cells per well) were maintained in complete RPMI1640 medium supplemented with 10% FBS, 10 mmol/L HEPES, 50 μmol/L β-mercaptoethanol, and penicillin/streptomycin/l-glutamine. The cultures were analyzed for CFSE profiles according to a time course as indicated.

Total RNA was derived from cultured cells with Rnaiso Reagent (Takara). Quantitative real-time PCR was done using SYBR Green I Dye (Roche) and according to the protocol of Light Cycle 480 kits (Roche). The primer sequences of CNX were 5′-CATGATGGACATGATGATGACAC-3′ (forward) and 5′-CTAGAGGCTTGGT GTATAC-3′. Results were normalized to the expression of GAPDH (forward, 5′-AACTTTGGCATTG TGGAAGG-3′; reverse, 5′-ACACATTGGGGGTAGGAACA-3′).

Cells and tissues were lysed with radioimmunoprecipitation assay (RIPA) buffer containing proteases inhibitor cocktail (Sigma Aldrich) and ultrasonication. Protein quantification was performed using BCA Protein Assay Reagent (Thermo Fisher Scientific), and 45 mg protein per sample was loaded into SDS-PAGE and sequentially immunoblotted with calnexin mAb (C5C9, Cell Signaling Technology). The proteins were using GAPDH antibodies (D16H11, Cell Signaling Technology) as loading controls.

Tumor antigen–specific CD8⁺ human T-cell clones were generated from PBMCs from a healthy donor by in vitro stimulation using dendritic cells loaded with corresponding peptide epitopes (irradiated HSC3 cells were used as tumor antigens). CNX-overexpressing HSC3 cells or control cells were labeled with CFSE and cocultured with CTLs at an effector-to-target ratio (E/T) of 5:1 and 10:1 for 4 hours. Then, 0.1 μg of DAPI was added to each sample, and the samples were immediately analyzed by flow cytometry. CTL killing (%) = CFSE⁺DAPI⁺ cells/total CFSE⁺ cells × 100%.

Baseline characteristics were described by mean and SD for continuous variables or described by numbers and percentages for categorical variables. To compare the baseline characteristics between different groups, Student t test was used for continuous variables, whereas χ² tests were used for categorical variables. Overall survival was calculated and described by Kaplan–Meier method. The difference of survival curves was tested by log-rank test. Univariate and multivariate Cox proportional models were used to analyze the associations between baseline characteristics and overall survival, and the HRs with 95% confidence interval (CI) were calculated. All statistical analyses were performed by GraphPad Prism 7.0 and Stata/MP 14.0. All tests were two-sided, and a P value of less than 0.05 was considered significant.

iTRAQ-coupled 2D LC-MS/MS technique was used to study the protein expression patterns of OSCC tumor tissue and control normal tissue. In total, 6 pairs of tissue lysates were analyzed. When the protein patterns of the primary tumor and its corresponding normal tissue were compared, multiple proteins were found to be differentially expressed. Supplementary Table S1 shows 43 proteins that are significantly upregulated (>2-fold). Calnexin exhibited a higher expression in cancer tissues (2.7-fold elevation) when compared with the corresponding normal tissues. We further confirmed its expression by qRT-PCR (Fig. 1A), Western blotting (Fig. 1B), and IHC staining (Fig. 1C) in OSCC cells; paired cancer and adjacent noncancerous tissues were derived from the same patients with OSCC. We found that calnexin was significantly upregulated in OSCC tissues.

In addition, immunofluorescence analysis suggested that although most of the calnexin protein expression was observed in the cytosol of OSCC cells, a considerable fraction of calnexin colocalized with WGA in the cell membrane of tumor cells. Membrane localization of calnexin has been found in several tumors, including OSCC and melanoma (Fig. 1D), suggesting that calnexin may exert its biological or immunologic regulation functions as either a secreted or membrane-bound form. Because T cells play a critical role in mediating antitumor immunity, and impaired T-cell infiltration is positively correlated with poorer prognosis of tumor patients (29), we first determined whether there was any correlation between membranous expression of calnexin and T-cell infiltration in tumor tissues. Cell surface calnexin expression was determined by flow cytometry. We found that higher calnexin membranous expression was negatively correlated with the numbers of infiltrated CD4⁺ and CD8⁺ T cells in OSCC tumor tissues (Fig. 1E and F).

Given that calnexin membranous expression was upregulated in OSCC tissues, we next determined whether calnexin expression in the cytosol and cell membrane was correlated with tumor prognosis. To address this, IHC staining of calnexin was performed to assess the expression of calnexin in samples from 357 patients with primary OSCC. Representative images of the intensity stages are shown in Fig. 2A. Overall, cytoplasmic calnexin expression was categorized into low (178 of the 357 tumor samples, 49.86%) and high (179 of the 357 tumor samples, 50.14%) expression groups using the cutoff point 5.04 based upon the median of SID scores of the total patients. In addition, 71 of the 357 tumor samples (19.89%) showed apparent calnexin expression at the plasma membrane, whereas 286 of the 357 tumor samples (80.11%) showed negative staining of calnexin at the plasma membrane (Supplementary Table S2). Representative images are shown in Fig. 2A. The expression of calnexin was assessed for association with a number of clinicopathologic variables (Supplementary Table S2).

Kaplan-Meier survival curves show, overall patient survival was not significantly different when compared between low and high cytoplasmic expression of calnexin (P = 0.405), whereas patients with positive calnexin membranous expression had a significantly reduced overall survival than patients with negative expression [3-year OS: 47.89% (35.93%–58.88%) vs. 66.43% (60.64%–71.58%), P = 0.016; Fig. 2B]. In the univariate analysis, calnexin membranous expression (P = 0.018) along with nodal stage (P < 0.001), clinical TNM stage (P = 0.030), and radiotherapy (P = 0.042) were significantly associated with overall survival. Adjusted for nodal stage and radiotherapy, patients with positive membranous expression of calnexin was significantly associated with reduced overall survival, compared with patients with negative expression (HR, 1.59; 95% CI, 1.10–2.30; P = 0.013; Fig. 2C). There was no significant association between cytoplasmic expression of calnexin and overall survival among patients with OSCC (Supplementary Tables S3 and S4).

Because upregulation of calnexin in OSCC tumor tissues was correlated with reduced infiltration of CD4⁺ and CD8⁺ T cells, we hypothesized that calnexin impairs the antitumor immunity of effector T cells. A calnexin-Ig fusion protein (CNX-Ig) was generated to examine the regulatory roles of calnexin in T-cell responses. Indeed, we found that when immobilized on a microplate, calnexin-Ig, but not control-Ig, suppressed the proliferation of bulk purified CD4⁺ and CD8⁺ T cells in response to anti-CD3 stimulation (Fig. 3A) and inhibited the production of effector molecules such as IFNγ, TNFα, and IL2 (Fig. 3B). Furthermore, calnexin inhibited the antitumor cytolytic functions of CD8⁺ T cells against HSC3 tumor cells (Fig. 3C). These data collectively suggested that calnexin inhibited the proliferation and antitumor effector functions of CD4⁺ and CD8⁺ T cells. Although the receptor for calnexin is unknown, we speculated that the engagement of calnexin-R on T cells suppresses T-cell receptor (TCR) signaling. To test this hypothesis, proximal TCR signaling events were examined using calnexin-Ig. LAT is a proximal signaling adaptor that is phosphorylated upon TCR stimulation and forms a complex with multiple signaling molecules, including SH2 domain containing a leukocyte protein of 76 kDa (SLP76) and phospholipase C (PLC)-γ1 (30). Immobilized calnexin-Ig substantially reduced the amount of SLP76 recruited to the CD3 complex, as well as its phosphorylation. When total cell lysates were examined, the phosphorylation of several downstream signaling molecules, such as Akt and Erk1/2 was also impaired (Fig. 3D).

In addition to demonstrating the inhibitory effect of calnexin on peripheral blood T cells from healthy donors, we also evaluated its role in circulating blood T cells from patients with OSCC. Tumor lysates from the same patients were used as tumor antigens. We found that PBMCs from OSCC patients cocultured with calnexin-Ig showed inhibitory effects on the proliferation of CD8⁺ T cells (Fig. 4A) and reduced the number of functional CD8⁺ T cells producing IFNγ by nearly half (Fig. 4B). These findings were confirmed by our cytometric bead assay (CBA) results, which showed decreased production of IFNγ and TNFα but increased production of IL10 by T cells (Fig. 4C). These changes were more significant in the tumor antigen–experienced cells. Because an increase in IL10 production by T cells was observed, the induction of Treg was examined. However, calnexin could not promote Treg production in an antigen-specific manner (Supplementary Fig. S1). These data collectively suggested that calnexin inhibited the proliferation and antitumor effector functions of CD8⁺ T cells in patients with OSCC in an antigen-dependent manner.

Because calnexin inhibits the proliferation and effector functions of CD4⁺ and CD8⁺ T cells, we next determined whether calnexin-mediated impairment of antitumor T-cell responses contributes to tumor growth. Because OSCC tumor cells do not grow well in wild-type mice, a humanized mouse model was used (31, 32). HSC3 tumor cells expressing shRNA-CNX (sh-CNX) or shRNA-control (sh-NEG) were inoculated into NCG mice, and the mice were engrafted with human PBMCs after tumor implantation. The mice were euthanized before experiencing weight loss, a symptom of graft-versus-host disease (GVHD) that occurs in this humanized mouse model (Fig. 5A). In this model, the tumor growth in the sh-CNX group was lower than that in the sh-NEG group (P = 0.047; Fig. 5B). We also detected increased frequencies of multifunctional CD3⁺ T cells producing IFNγ in the sh-CNX group compared with the sh-NEG group upon PBMC engraftment (Fig. 5C and D). We also examined control mice without PBMC injection in HSC3 tumor model and found that in contrast to the results from humanized mice, calnexin silencing promoted tumor growth in the immunodeficient mice (Fig. 5E and F), indicating that calnexin might have another tumor-intrinsic role that is independent of its function on T cells. These data indicated that calnexin suppressed antitumor immunity and promoted OSCC tumor growth via inhibiting the proliferation and effector functions of CD4⁺ and CD8⁺ T cells.

We next developed a mouse melanoma model to determine whether calnexin-mediated impairment of T cells contributes to tumor growth. We injected mice subcutaneously with B16F10 cells expressing shRNA targeting calnexin (sh-CNX) or control shRNA (sh-NEG) and monitored tumor growth. To generate protective immunity, naïve mice were vaccinated with irradiated B16F10 tumor cells in advance (Fig. 6A). We found that knockdown of calnexin in melanoma tumor cells significantly inhibited melanoma growth in mice, whereas administration of calnexin-Ig enhanced melanoma growth (Fig. 6B). Furthermore, knockdown of calnexin in melanoma tumor cells increased the infiltration of CD3⁺, CD4⁺, and CD8⁺ T cells in melanoma tumors (Fig. 6C and D) and enhanced the expression of Ki67 in CD4⁺ and CD8⁺ T cells (Fig. 6E). In addition, treatment with calnexin-Ig inhibited this infiltration in melanoma tumors (Fig. 6C and D) and the expression of Ki67 in these T cells (Fig. 6E). Moreover, knockdown of calnexin in melanoma tumor cells enhanced the expression of the antitumor effector molecules IFNγ and TNFα by CD4⁺ and CD8⁺ T cells in melanoma tumors, and this effect was significantly reversed by treatment with calnexin-Ig (Fig. 6F). No differences in Tregs and MDSC frequencies among TILs were found (Supplementary Fig. S2). There were no significant differences in the proliferation and effector functions of CD4⁺ and CD8⁺ T cells located in the spleen, lymph nodes and PBMCs between the groups (Supplementary Fig. S3). To confirm that there is no intrinsic enhancement of tumor growth in the absence of T-cell–mediated antitumor immunity, tumors were inoculated in T-cell–deficient nude mice. As shown in Supplementary Fig. S4, administration of calnexin-Ig no longer enhanced melanoma growth upon T-cell deficiency. Calnexin-silenced B16F10 tumors grew more rapidly than control tumors, which is consistent with our previous observation (Fig. 5E and F) that tumor-intrinsic calnexin itself suppressed tumor growth. Together, these data indicated that calnexin deficiency promotes antitumor immunity and controls tumor growth in a T-cell–dependent manner.

Given that T-cell surface receptors such as TIGIT, CTLA-4, PD-1, and LAG-3 play critical roles in inhibiting T-cell responses, we next determined whether upregulation of calnexin might enhance the expression of these molecules and therefore induce impairment of the proliferation and effector functions of CD4⁺ and CD8⁺ T cells in tumors. To address this, we analyzed the expression of TIGIT, CTLA-4, PD-1H, PD-1, and LAG-3 on CD4⁺ and CD8⁺ T cells derived from melanoma tumor samples. We found that knockdown of calnexin in melanoma tumor cells significantly decreased the expression of PD-1, but not TIGIT, CTLA-4, PD-1H, or LAG-3, in CD4⁺ and CD8⁺ T cells derived from melanoma tumor samples (Fig. 7A). In contrast, calnexin-Ig treatment partly reversed the decrease in PD-1 expression on CD4⁺ and CD8⁺ T cells conferred by knockdown of calnexin in melanoma tumors (Fig. 7A). Similar results were found in an MB49 tumor model with calnexin-Ig treatment (Fig. 7B). In addition, calnexin-Ig enhanced the expression of PD-1 on CD8⁺ T cells in PBMCs derived from patients with progressive OSCC, and this enhancement was more significant in tumor antigen–experienced T cells, as shown in Fig. 7C. Thus, these data suggested that calnexin enhanced the expression of PD-1 on CD4⁺ and CD8⁺ T cells in OSCC in an antigen-dependent manner.

We then determined the mechanism by which calnexin enhanced the expression of PD-1 on T cells in tumors. Because PD-1 promoter CpG island methylation status plays a central role in mediating PD-1 expression (33, 34), we analyzed the methylation of this region using bisulfite sequencing in T cells from OSCC patients’ PBMCs (Fig. 7D). In contrast to control-Ig, T cells treated with calnexin-Ig significantly repressed the methylation of the PD-1 promoter CpG island (Fig. 7E).

In this study, we discovered that the ER chaperone protein calnexin was highly upregulated in OSCC tumor tissues and multiple tumors. Upregulation of membranous calnexin was positively correlated with poor prognosis of patients with OSCC. We found that calnexin played a central role in inhibiting the infiltration and effector functions of T cells and promoting the expression of PD-1 on CD4⁺ and CD8⁺ T cells in tumors, which therefore enhanced tumor growth, demonstrating the potential of calnexin as a new antitumor immunotherapy target.

Calnexin has been reported to play a role in the folding and quality control of newly synthesized glycoproteins (14, 35). A wide variety of important cellular and viral glycoproteins are known substrates of calnexin, including HIV gp120 and gp160, class I MHC heavy chain, and TCR subunits (36, 37). Although several reports have shown that calnexin expression may be associated with the progression of breast cancer, lung cancer, and colorectal cancer, most previous studies of calnexin focused on the relationship between the expression of calnexin and clinical outcome (38–40). Whether calnexin regulates the T-cell response during tumor development is unknown. Here, we first identified that upregulation of calnexin in tumor cells could inhibit the infiltration of T cells in tumors and the proliferation and effector functions of CD4⁺ and CD8⁺ T cells. As increasing evidence has suggested that the infiltration and effector functions of T cells in tumors are critical for antitumor immunity, this finding therefore reveals a mechanism responsible for poor survival of tumor patients.

A finding of this study is the establishment of an immunologic link between calnexin and PD-1 on T cells. We found that knockdown of calnexin in melanoma tumor cells significantly decreased the expression of PD-1. In addition, calnexin-Ig treatment partly reversed the decrease of PD-1 expression on T cells. Calnexin-Ig also enhanced the expression of PD-1 on T cells in PBMCs derived from patients with progressive OSCC by inhibiting the PD-1 promoter CpG island methylation. PD-1 is upregulated on activated T cells. The binding of PD-1 and PD-L1 induces T-cell anergy and cell death (5, 41). This study provides evidence suggesting that PD-1 expression on T cells can be influenced by the expression of calnexins. The detailed mechanism by which calnexin mediates PD-1 expression during tumor development should be investigated in future studies. There are substantial efforts underway to identify reliable predictive biomarkers of response and resistance to immune checkpoint blockade, including total tumor mutational load (42, 43), as well as markers of an effective immune infiltrate within a tumor signifying a “hot” or “cold” tumor microenvironment (44). This study provides a potential target for the improvement of responses to anti–PD-1 immunotherapy. Because the density or distribution of T cells and PD-1/PD-L1 axis activation could affect the differential responses to checkpoint blockade, the effect of calnexin on the enhancement of antitumor responses during PD-1 blockade should be determined in further studies.

Although we found that calnexin expressed in tumor cells limited the infiltration and effector functions of CD4⁺ and CD8⁺ T cells in tumors and therefore promoted tumor cell growth, the specific receptor expressed on CD4⁺ and CD8⁺ T cells that interacts with calnexin remains unknown. In addition, the observations that only membranous calnexin expressed in tumor cells was associated with poorer survival of patients with OSCC indicate that direct contact with PBMCs is required for calnexin to exert its regulatory function. Identification of the receptor that interacts with calnexin expressed in tumor cells will allow us to better understand the mechanism by which calnexin impairs the infiltration and effector functions of CD4⁺ and CD8⁺ T cells in the tumor microenvironment. Although low concentrations of calnexin were isolated from lung cancer patients’ peripheral blood serum (15), the interaction between calnexin and T cells may primarily occur in the tumor site. Thus, the interaction between calnexin and T cells in the circulation and lymphoid tissue may be not sufficient for inhibition of the proliferation and effector functions of T cells. Previous studies from other groups suggest that calnexin could be transported to the plasma membrane to interact with glycoproteins such as clonotype-independent CD3 complexes (16, 45). Further studies are required to determine what protein interacts with calnexin.

L. Chen is a scientific founder of and has ownership interest in NextCure and TAYU Biotech, reports receiving commercial research funding from NextCure, and is a consultant/advisory board member for Pfizer, Vcanbio, and GenomiCare. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Y. Chen, J. Cui, G. Zeng, L. Chen, B. Cheng, Z. Wang

Development of methodology: D. Ma, X. Wang, J. Song

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Chen, D. Ma, X. Wang, J. Fang, Q. Li, Q. Li, S. Wen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Chen, D. Ma, X. Li, X. Ren, L. Chen

Writing, review, and/or revision of the manuscript: J. Cui, G. Zeng, L. Chen, Z. Wang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Wang, X. Liu, L. Luo, J. Xia

Study supervision: B. Cheng, Z. Wang

We thank Luisa A. DiPietro (College of Dentistry, UIC) for the suggestions on this manuscript. This project was supported by the National Natural Science Foundations of China (grant nos. 81772896, 81472524, 81630025, 81602383, and 81602384).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.