Cancer Cell–Intrinsic Expression of MHC Class II Regulates the Immune Microenvironment and Response to Anti–PD-1 Therapy in Lung Adenocarcinoma

Abstract

MHC class II (MHCII) expression is usually restricted to APC but can be expressed by cancer cells. We examined the effect of cancer cell–specific MHCII (csMHCII) expression in lung adenocarcinoma on T cell recruitment to tumors and response to anti–PD-1 therapy using two orthotopic immunocompetent murine models of non–small cell lung cancer: CMT167 (CMT) and Lewis lung carcinoma (LLC). We previously showed that CMT167 tumors are eradicated by anti-PD1 therapy, whereas LLC tumors are resistant. RNA sequencing analysis of cancer cells recovered from tumors revealed that csMHCII correlated with response to anti-PD1 therapy, with immunotherapy-sensitive CMT167 cells being csMHCII positive, whereas resistant LLC cells were csMHCII negative. To test the functional effects of csMHCII, MHCII expression was altered on the cancer cells through loss- and gain-of-function of CIITA, a master regulator of the MHCII pathway. Loss of CIITA in CMT167 decreased csMHCII and converted tumors from anti–PD-1 sensitive to anti–PD-1 resistant. This was associated with lower levels of Th1 cytokines, decreased T cell infiltration, increased B cell numbers, and decreased macrophage recruitment. Conversely, overexpression of CIITA in LLC cells resulted in csMHCII in vitro and in vivo. Enforced expression of CIITA increased T cell infiltration and sensitized tumors to anti–PD-1 therapy. csMHCII expression was also examined in a subset of surgically resected human lung adenocarcinomas by multispectral imaging, which provided a survival benefit and positively correlated with T cell infiltration. These studies demonstrate a functional role for csMHCII in regulating T cell infiltration and sensitivity to anti–PD-1.

Lung cancer is the leading cause of cancer-related deaths, with an overall 5-year survival rate of <18% (1). Non–small cell lung cancer (NSCLC) represents 85% of all lung cancers (2) and can be divided into adenocarcinomas and squamous cell carcinomas. During the past 10 years, a paradigm shift has changed the view of cancer from an autonomous cellular disease to a complex system of interactions between cancer cells and the tumor microenvironment (TME) (3, 4). Interactions between cancer cells and T cells can facilitate or hinder tumor progression. In general, increased tumor T cell infiltration correlates with positive clinical outcome (4, 5), indicating that T cells can inhibit tumor progression. However, cancer cells can mobilize multiple mechanisms to evade immune attack, and immunoevasion is one of the hallmarks of cancer progression (6). Binding of programmed cell death ligand-1 (PD-L1), expressed on cancer cells and other cells of the TME, to programmed cell death protein-1 (PD-1) expressed on T cells results in inhibition of TCR signaling. Prolonged PD-1/PD-L1 engagement results in a hypo-functional, “exhausted” T cell state that fails to contain tumor progression. Abs against either PD-1 or PD-L1 disrupt this interaction, reinvigorating T cell function, and can potentially result in tumor elimination (7, 8). These agents are Food and Drug Administration approved for multiple malignancies, including NSCLC (9, 10). However, even in patients with high PD-L1 expression, less than half respond in the first line of therapy (11). Whereas associations with mutational and antigenic burden, an inflamed TME, and levels of PD-1/PD-L1 expression have been described, definitive mechanisms underlying tumor responsiveness or resistance to anti–PD-1/anti–PD-L1 targeted therapies remain highly sought after (12–14).

One well-established mechanism of immunoevasion is the failure of cancer cells to present tumor Ags. For example, loss of MHC class I (MHCI) contributes to immune evasion by decreasing Ag presentation to CD8⁺ cytotoxic T cells, which recognize and directly kill tumor cells (15). Whereas MHCI is expressed on all cells, MHC class II (MHCII) expression is usually restricted to APCs. Peptide-loaded MHCII molecules are constitutively expressed on APCs, and MHCII Ag presentation is essential for CD4⁺ helper T cell–dependent immune responses (16). MHCII can also be induced on non-APCs in response to an inflammatory environment and inflammatory cytokines such as IFN-γ (17). Among non-APCs, MHCII expression by cancer cells could potentially have an important role in antitumor immunity, as it would afford the potential for direct recognition and engagement of cancer cells presenting tumor neoantigens in the context of MHCII to CD4⁺ helper T cells. In this study, we designate expression of MHCII on cancer cells as cancer cell–specific MHCII (csMHCII) to distinguish it from expression on other cells in the TME. In colon cancer, high csMHCII expression is associated with increased survival (18, 19). Similarly, melanomas with high expression of csMHCII respond better to immune checkpoint therapy blockade (20). In triple-negative breast cancer, csMHCII expression was also associated with better progression-free survival (21, 22). Although these studies describe a positive correlation between csMHCII and survival or response to therapy, whether csMHCII is functionally important in Ag presentation, regulating the TME, or promoting the response to immunotherapy remains unclear.

Our laboratory has developed an immunocompetent orthotopic murine model of lung adenocarcinoma in which murine lung cancer cells are directly implanted into the lungs of syngeneic mice (23–26). This allows the tumors to develop in the appropriate microenvironment. We previously compared the response of two KRAS mutant lung cancer cell lines, CMT167 and Lewis lung carcinoma (LLC), to immune checkpoint inhibitors (24). CMT167 tumors were strongly inhibited by anti–PD-1 or anti–PD-L1 Abs, whereas LLC tumors were resistant. The mechanisms underlying this differential response are not well understood. In this study, we present evidence that csMHCII expression positively correlates with response to PD-1/PD-L1 targeted immunotherapy and test the functional contribution of csMHCII on antitumor immunity under baseline and immunotherapy conditions. We further demonstrate that csMHCII expression varies between human lung adenocarcinomas and is positively correlated with T cell infiltration and a survival benefit. These studies emphasize the unique contribution of csMHCII on shaping the TME and influencing the response to immunotherapy.

Materials and Methods

Cell lines and culture conditions

CMT167 cells were stably transfected with firefly luciferase, as previously described (24). Luciferase-expressing LLC cells were purchased from Caliper Life Sciences (LL/2-luc-M38). The CMT167 cells harbor a Kras^G12V mutation, and the LLC cells harbor a Kras^G12C mutation (24). Both cell lines were maintained in DMEM with 4.5 g/l glucose (no. 10–017-CV; Corning) containing 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 500 μg/ml G418 at 37°C in a humidified 5% CO₂ atmosphere. Cells were periodically tested for mycoplasma infection, maintained as frozen stocks, and cultured for only 2–4 wk before use in experiments. Authentication of cell lines based on morphology, growth curve analysis, and metastatic phenotype was performed regularly. Human NSCLC cell lines were maintained in RPMI 1640 (no. 10–040-CV; Corning) containing 10% FBS at 37°C in a humidified 5% CO₂ atmosphere. All were KRAS mutant.

Orthotopic murine model and anti–PD-1 treatment

Wild-type C57BL/6J and GFP-expressing mice (C57BL/6–132Tg [UBC-GFP] 30Scha/J) were obtained from The Jackson Laboratory (Bar Harbor, ME). Mice were bred and maintained in the Center for Comparative Medicine at the University of Colorado Denver. Experiments were performed in 7- to 12-wk-old male mice. All procedures were performed under protocols approved by the Institutional Animal Care and Use Committee at the University of Colorado Denver. Cancer cells were directly implanted into the left lobe of the lung, as previously described (23, 24). For experiments with checkpoint inhibitors, tumor-bearing mice were i.p. injected (8–10 mg/kg) with either an IgG2a isotype control Ab (no. BE0089; BioXCell) or an anti–PD-1 Ab (no. BE0146; BioXCell) twice weekly starting 7–10 d after injection of cancer cells.

Transcriptome profiling by RNA sequencing of lung cancer cells from tumor-bearing GFP-expressing transgenic mice

Cancer cells were injected into GFP-expressing transgenic mice as described above. Four weeks after CMT167 and 2 wk after LLC tumor implantation, mice were sacrificed, and single-cell suspensions were prepared as described below. Lung cancer cells were recovered by sorting for GFP⁻ cells, and RNA sequencing (RNA-seq) was performed as previously described (23). Cell sorting was performed at the University of Colorado Cancer Center Flow Cytometry Shared Resource using a MoFlo XDP cell sorter equipped with a 100-μm nozzle (Beckman Coulter).

Generation of CIITA knockdown cells and CIITA overexpression cells

CMT67 cells were stably transfected using lentivirus with short hairpin RNA (shRNA) constructs purchased from the University of Colorado Functional Genomics core: nontargeting control (NT-ctrl) (SHC001V), shRNA targeting CIITA (shCIITA) 48 (TRCN0000086448), shCIITA49 (TRCN0000086449), and shCIITA50 (TRCN0000086450). Selection of stable transductants was carried out with puromycin (2 μg/ml; Sigma-Aldrich). Knockdown efficiency was confirmed by quantitative real-time PCR (qRT-PCR). LLC-luc cells were stably transfected with CIITA overexpression vector pcDNA3-myc-tagged CIITA (plasmid no. 14650; Addgene) or control pcDNA3.1 (V79020; Thermo Fisher Scientific) using Lipofectamine 2000 Reagent (11668–027; Invitrogen). Cells were selected with G418 (500 ng/ml; Corning), and individual clones were screened for Ciita mRNA expression and MHCII expression by flow cytometry.

Immunofluorescence

Tissues were fixed in 4% paraformaldehyde. Paraffin-embedded samples were cut into 5-μm sections. Sections were dehydrated, immersed in 0.1% Sudan black B (199664–25G; Sigma-Aldrich) in 70% ethanol for 20 min, washed in TBST, incubated in citrate Ag retrieval solution at 100°C for 2 h, washed in 0.1 M glycine/TBST (G-8898; Sigma-Aldrich) for 10 min, and placed in 10 mg/ml sodium borohydride in HBSS (14175–095; Life Technologies). After blocking with 10% goat serum in equal parts of 5% BSA in TBST and Superblock (AAA999; ScyTek Laboratories) at 4°C overnight, sections were incubated with primary anti-CD3 (MA5–14524; Thermo Fisher Scientific), anti-CD4 (no. 14–9766-82; eBiosciences), or anti-CD8 (no. 14–0808-82; eBiosciences) Ab in equal parts solution of 5% BSA in TBST and Superblock for 1 h at room temperature; washed in TBST; and incubated with secondary goat anti-rabbit IgG Alexa Fluor 488 (no. A11034; Invitrogen) or goat anti-rabbit IgG Alexa Fluor 594 (no. A11037; Invitrogen). Slides were mounted with Vectashield mounting medium with DAPI (no. H-1200; Vector Laboratories). Positive staining was determined by count per high-power field (HPF) (×40), and each tumor count was an average of five different views of the tissue, of three or more separate tumors, and averaged between two different observers.

Preparation of single-cell suspension and flow cytometry

Mice were sacrificed at 4 wk (CMT167 tumors) or 3 wk (LLC tumors), and left lungs containing tumors were excised. Single-cell suspensions were prepared as previously described (23–25). Cells were stained with three panels of Abs (Supplemental Table I). For intracellular markers, cells were treated with brefeldin A (no. 420601; BioLegend), Monensin (no. 420701; BioLegend), and a cell stimulation mixture (PMA/ionomycin) for 5 h at 37°C. Cells were then permeabilized using the FOXP3/Transcription Factor Staining Buffer Set (no. 00–5523-00; eBioscience) and stained with intracellular Abs. Cells were analyzed at the University of Colorado Cancer Center Flow Cytometry Core using a Gallios flow cytometer (Beckman Coulter). All flow cytometry experiments included single-stain cell controls and isotype controls for intracellular stains and were subjected to postacquisition compensation using VersaComp Ab Capture Bead Kit (no. B22804; Beckman Coulter). Data were analyzed using Kaluza Software (Beckman Coulter). Each measurement represents three separate isolations from lungs of three pooled mice. Gating strategy is shown in Supplemental Fig. 2F.

Coculture of CD4⁺ T lymphocytes and cancer cells and analysis by flow cytometry

CD4⁺ T cells were isolated from the lungs of CMT-implanted mice using the EasySep Mouse CD4⁺ T Cell Isolation Kit (Stemcell Technologies). The CD4⁺ T cells were then added in a 1:1 ratio to cultured CMT167 cells, either NT-ctrl or cells silenced for CIITA (CMT shCIITA) that were pretreated with IFN-γ (100 ng/ml) (no. 315–05; Peprotech) for 48 h. The cells were centrifuged at 100 × g for 5 min and incubated overnight at 37°C.

qRT-PCR

Cells were harvested and homogenized in 350 μl of RLT buffer (Qiagen). Total RNA was isolated using an RNeasy Kit (Qiagen), and reverse transcription was performed on 1 μg of total RNA using qScript cDNA SuperMix (95048–025; Quanta BioSciences). Real-time PCR analysis was conducted in triplicate in a C1000 Touch Thermal Cycler (Bio-Rad) using Power SYBR Green PCR Master Mix (4367659; Applied Biosystems). The relative message levels of each target gene were normalized to β-actin. β-Actin was as follows: forward (Fwd): 5′-GGCTGTATTCCCCTCCATCG-3′, reverse (Rev): 5′-CCAGTTGGTAACAATGCCATG-3′. Murine Ciita was as follows: Fwd: 5′-TGCGTGTGATGGATGTCCAG-3′, Rev: 5′-CCAAAGGGGATAGTGGGTGTC-3′. H2-Aa was as follows: Fwd: 5′-ATCGTGGTGGGCACCATCTTCA-3′, Rev: 5′-AAGAGGGACACACGCCTTCCTT-3′. H2-Dma was as follows: Fwd: 5′-CTCGAAGCATCTACACCAGTG-3′, Rev: 5′-TCCGAGAGCCCTATGTTGGG-3′. Cd74 was as follows: Fwd: 5′-CGCGACCTCATCTCTAACCAT-3′, Rev: 5′-ACAGGTTTGGCAGATTTCGGA-3′. H2-k1 was as follows: Fwd: 5′-GCTGGTGAAGCAGAGAGACTCAG-3′, Rev: 5′-GGTGACTTATCTTCAGGTCTGCT-3′. H2-d1 was as follows: Fwd: 5′-CTCCGTCCACTGACTCTTAC-3′, Rev: 5′-GAGAACTGAGGGCTCTGGATG-3′. Primers are from IDT.

Western blot analysis

Proteins were harvested in lysis buffer (50 mM β-glycerophosphate, pH 7.2, 0.5% Triton X-100, 5 mM EGTA, 100 μM sodium orthovanadate, 1 mM DTT, 2 mM MgCl₂) with protease inhibitor mixture (Sigma-Aldrich). Proteins were fractionated by SDS-PAGE, electrophoretically transferred to PVDF membranes, and blocked for 1 h in 5% BSA and TBST (G-8898; Sigma-Aldrich). Abs were as follows: Anti-Mouse pSTAT1 (Y701) (no. 9167s; Cell Signaling), Anti-Mouse STAT1 (no. 9172S; Cell Signaling), Anti-Mouse CIITA (no. ab49132; abcam), β-actin (no. A5441; Sigma-Aldrich).

Spectral flow cytometry Ab staining

Single-cell suspensions were stained with a panel of Abs (see Supplemental Table I). Zombie NIR Viability dye was diluted based upon recommended protocols from BioLegend and Cytek, before staining samples at room temperature for 15 min (no. 423105; BioLegend) in 1.5-ml Eppendorf tubes. After incubation, samples were resuspended and subjected by Fc receptor blockade (CD16/CD32 no. 14–0161-86; eBioscience) followed by surface Ab staining with fluorescently labeled Abs. All samples were covered in foil and incubated for a total of 30 min at room temperature in the dark. Cells were then permeabilized using the FOXP3/Transcription Factor Staining Buffer Set (no. 00–5523-00; eBioscience) at 4°C for 18–24 h and then subjected to a 2-h incubation at 4°C for intracellular staining. All samples were washed and filtered through 35-μm cell strainer tubes (5-ml 12 × 75 mm Polystyrene Round-Bottom Tube no. 352235; Falcon) before instrument analysis. Experiments included single-stain compensation beads for unmixing purposes (Ultra Comp Beads no. 01–2222-42; Invitrogen, eBioscience) as well as single-stain cell controls for validation and gating strategies. Samples were analyzed at the University of Colorado Cancer Center Flow Cytometry Core on the Cytek Aurora spectral flow cytometer using SpectroFlo, with data visualized using Kaluza Software (Beckman Coulter) and clustering analysis done with R using Rphenograph.

Patient studies

Tumor tissues were collected through a protocol approved by the Mayo Clinic Institutional Review Board and obtained from the Mayo Clinic Lung Cancer Repository. Written informed consent was obtained from all patients in accordance with the Declaration of Helsinki. Patients were identified from the tissue repository that underwent curative surgical resection of lung adenocarcinoma between 2004 and 2007 and had available residual tumor specimens. The electronic medical record was reviewed, and pertinent clinical data, including age at time of surgery, gender, smoking status, pack years, and last follow-up date or date of death, were extracted. Patients were staged based on surgical findings using the seventh edition of the American Joint Committee on Cancer Tumor, Node, Metastasis system for NSCLC. For cancer-specific survival, patients were censored at the date of last follow-up or death from causes other than lung cancer. Formalin-fixed paraffin-embedded tissue blocks were sectioned into 5-mm slides by the Pathology Research Core (Mayo Clinic, Rochester, MN).

Tissue multiplex immunohistochemistry

Sections from formalin-fixed paraffin-embedded blocks were stained using Opal multiplex according to the manufacturer’s protocol (PerkinElmer) for DAPI, HLA-DR, cytokeratin (CK), CD3, and CD8 at the Human Immune Monitoring Shared Resource at CU Anschutz Medical Campus. Slide scanning was performed on the Vectra 3.0 instrument. Three to five multispectral regions of interest were selected and analyzed using inForm Software 2.4. Images were spectrally unmixed, evaluated for staining intensities and morphology, tissue segmented based on tissue markers, cell segmented based on nuclear and membrane markers, and phenotypically scored.

RNA-seq data acquisition

RNA-seq data for 15 human KRAS mutated lung adenocarcinoma was obtained from the Cancer Cell Line Encyclopedia (27).

Statistical analysis

Statistical analyses were performed using the GraphPad Prism 7 or R software. Data are presented as means ± SEM. A one- or two-way ANOVA was used to compare differences in more than two groups. A Student t test was used to compare differences between two groups in data with a normal distribution. A Pearson test was used to test correlation. A log-rank test was used to compare survival. In all circumstances, p values ≤0.05 were considered significant.

Results

NSCLC cells differentially upregulate an Ag presentation pathway signature

We hypothesized that the differential response to anti-PD-1 therapy of CMT167 tumors (sensitive) versus LLC tumors (resistant) that we previously described (24) was an inherent property of the cancer cells and how they respond to signals from the TME. To define these differences, we analyzed changes in gene expression, comparing cancer cells recovered from in vivo tumors by FACS with identical cells grown in vitro. CMT167 and LLC cells were injected into separate groups of C57BL/6 GFP⁺-transgenic mice, and the GFP⁻ population, which represents purified cancer cells, was subjected to RNA-seq (28). KEGG pathway analysis determined CMT167 cells harvested from primary tumors showed an enhanced Ag presentation signature relative to LLC cells harvested from primary tumors. Among these genes, CMT167 cells recovered from tumors selectively induce MHCII genes and cofactors necessary for processing and presentation, such as CIITA, the master transcriptional regulator of the MHCII pathway, whereas the LLC cells did not (Fig. 1A). To validate the induction of csMHCII at the protein level, separate single-cell suspensions of tumor-bearing lungs grown from C57BL/6 GFP⁺ mice were analyzed by flow cytometry. Flow cytometric analysis demonstrated that the GFP⁻ CMT167 cancer cells showed a significant increase of surface MHCII compared with cells cultured in vitro (Fig. 1B). Importantly, csMHCII expression was not observed in LLC tumors (Fig. 1B). For both tumors, we detected MHCII expression on the GFP⁺ population, which reflects expression on APCs of the TME (Supplemental Fig. 1A, 1B).

FIGURE 1.

NSCLC cells differentially express an Ag presentation signature. (A) Heatmap of RNA-seq of MHC-associated genes from the Ag processing presentation signature using KEGG pathway analysis of LLC or CMT167 cells derived from in vitro or in vivo conditions. (B) The percentage of MHCII⁺ staining measured by flow cytometry gated on singlets of live cells of CMT167 or LLC cells cultured in vitro and/or CMT167 or LLC tumor cells from C57BL/6 GFP⁺ mice gated on singlets of live cells of GFP⁻ in vivo (n = 5, one-way ANOVA). (C) Ciita mRNA expression in LLC and CMT167 cells measured by qRT-PCR (n = 3, unpaired t test). (D) Percentage of MHCII⁺ expression measured in LLC and CMT167 cells measured by flow cytometry gated on singlets of live cells (n = 3, unpaired t test). (E) Ciita mRNA expression in CMT167 cells expressing an NT-ctrl shRNA or a shCIITA untreated or treated with IFN-γ for 48 h measured by qRT-PCR (n = 3, one-way ANOVA). (F) The percentage of MHCII⁺ staining from live, singlet cells of CMT167 cells expressing an NT-ctrl shRNA or a shCIITA untreated or treated with IFN-γ for 48 h measured by flow cytometry (n = 3, unpaired t test). (G) Western blot of CIITA, p-STAT1, and total STAT1 in cells cultured in the presence or absence of IFN-γ. (H) Flow cytometric analysis of percentage of MCHII⁺ events focused on live, singlet cells that were GFP⁻ (n = 3, unpaired t test). *p < 0.05, **p < 0.01, ****p < 0.0001.

Despite their divergent MHCII induction in vivo, both CMT167 and LLC cells expressed low basal levels of csMHCII expression in vitro. Because MHCII expression can be induced via IFN-γ signaling, we examined whether IFN-γ can induce MHCII genes and surface MHCII expression in vitro. CMT167 cells treated with IFN-γ for 48 h had increased levels of the MHCII master regulator CIITA, as assessed both by mRNA and protein, as well as increased mRNA expression of other MHCII genes, including H2-Aa, H2-Dma, and Cd74 (Fig. 1C, Supplemental Fig. 1C, 1D). This was confirmed by increased levels of surface MHCII protein by flow cytometry (Fig. 1D). Although LLC cells express IFN-γ receptors and upregulate some genes following IFN-γ treatment, LLC cells failed to induce MHCII upon IFN-γ treatment (Supplemental Fig. 1C, 1D).

Loss of CIITA decreases MHCII induction

To define the functional role of csMHCII expression, CMT167 cells were transduced with lentiviral constructs encoding three different shCIITAs, the MHCII transactivator, or an NT-ctrl shRNA (NT-ctrl). Because CIITA is a known transcriptional activator of multiple MHCII genes, silencing should disable the whole MHCII Ag presentation pathway while having minimal impact on other pathways (29, 30). In NT-ctrl CMT167 cells, IFN-γ induced Ciita mRNA similar to the parental cell lines. However, this induction was impaired in CMT167 transduced with three distinct CIITA shRNAs (Supplemental Fig. 1E). Similar decreases in cells subjected to knockdown for Ciita were also detected in other MHCII pathway genes, such as H2-Aa (Supplemental Fig. 1E). We isolated single-cell clones from the pools of one of the knockdowns, shCIITA (31), to achieve a better CIITA knockdown. Clone2 of shCIITA (31) had decreased CIITA expression, other MHCII genes, and surface expression of MHCII protein (Fig. 1E–G, Supplemental Fig. 1F), but showed no alterations in STAT1 signaling or changes in induction of MHCI genes in response to IFN-γ (Fig. 1G, Supplemental Fig. 1G). In addition, in vitro knockdown of CIITA did not alter Cd274 (PD-L1) mRNA induction in response to IFN-γ (Supplemental Fig. 1H). All subsequent in vivo experiments were therefore performed with this clone (designated CMT167-shCIITA). To validate knockdown of MHCII on the cancer cells in vivo, CMT167–NT-ctrl or CMT167-shCIITA cells were injected into C57BL/6 GFP⁺–transgenic mice. The CIITA knockdown tumor cells (gated on GFP [-negative] events) showed a decrease in csMHCII expression in vivo (Fig. 1H, Supplemental Fig. 1I). The CIITA knockdown did not affect MHCII expression in the TME GFP⁺ population or MHCI expression on the cancer cells (Supplemental Fig. 1J, 1K). Thus, CIITA knockdown in CMT167 cells reduces expression of csMHCII with minimal impact on MHCII expression in noncancer cells of the TME.

Loss of MHCII decreases CD4⁺ and CD8⁺ tumor infiltration and activation

To examine how csMHCII expression influenced T cell recruitment, infiltrating T cells in CMT167–NT-ctrl tumors compared with CMT167-shCIITA tumors were quantified by immunostaining. CMT167-shCIITA tumors, with reduced csMHCII, had a lower CD3⁺ T cell density (NT-ctrl 139.8 cells/HPF ± 17.96 versus shCIITA 53.50 cells/HPF ± 8.82), CD4⁺ T cell density (NT-ctrl 83.66 cells/HPF ± 9.903 versus shCIITA 38.43 cells/HPF ± 7.050), and CD8⁺ T cell density (NT-ctrl 54.53 cells/HPF ± 9.968 versus shCIITA 27.1 cells/HPF ± 2.605) (Fig. 2A). Representative images are shown in Fig. 2B. To determine potential mechanisms to account for the decrease in T cells, we isolated RNA from CMT167–NT-ctrl or CMT167-shCIITA whole tumors and performed qRT-PCR for T cell chemoattractants. The shCIITA tumors trended toward decreased levels of Cxcl9 and Ifnγ (Supplemental Fig. 2A, 2B). There were no changes in Cxcl10 and Cxcl11 and a slight decrease in Cxcl12 (Supplemental Fig. 2C–E). To determine which cells were producing Cxcl9, we injected CMT167–NT-ctrl or CMT167-shCIITA into GFP⁺ mice and then harvested the tumors and recovered the GFP⁺ and GFP⁻ populations by FACS. By qRT-PCR, there was a significant decrease in Cxcl9 production in GFP⁺ cells isolated from CMT-shCIITA tumors relative to NT-ctrl, with no change in Cxcl10, and a slight decrease in Ifnγ, although it did not reach statistical significance (Fig. 2C–E). There were no significant changes in Cxcl9, Cxcl10, or Ifnγ in the GFP⁻ population representing the cancer cells (Fig. 2F–H). This suggests the cells in the TME are decreasing their Cxcl9 production, leading to a decrease in T cell infiltration.

FIGURE 2.

Loss of MHCII decreases CD4⁺ and CD8⁺ tumor infiltration and decreases T cell effector function. (A) Average count of immunofluorescence of CD3⁺, CD4⁺, and CD8⁺ T cells per HPF (original magnification ×40) in CMT167–NT-ctrl tumors and CMT167-shCIITA tumors (n = 6, n = 5, respectively; unpaired t test). (B) Representative pictures of CMT167–NT-ctrl and CMT167-shCIITA tumors immunofluorescence: DAPI (blue), CD3 (red), and CD4 or CD8 (green). qRT-PCR analysis of CMT167–NT-ctrl and CMT167-shCIITA of GFP⁺ cells of (C) Cxcl9, (D) Cxcl10, and (E) Ifnγ (n = 5, unpaired t test). qRT-PCR analysis of CMT167–NT-ctrl and CMT167-shCIITA of GFP⁻ cells of (F) Cxcl9, (G) Cxcl10, and (H) Ifnγ (n = 3, unpaired t test). Flow cytometry on tumor-bearing lung events gated on singlets that are live, CD45⁺, and MHCII⁻ that were either CD4⁺ or CD8⁺ (I) IFN-γ⁺ CD4⁺ T cells and (J) TNF-α⁺ CD4⁺ T cells (n = 3, unpaired t test). Percentage of positive (K) IFN-γ⁺CD8⁺ T cells in CMT167–NT-ctrl and CMT167-shCIITA tumors (n = 3, unpaired t test). *p < 0.05, **p < 0.01.

To understand the functional consequence of csMHCII on effector CD4⁺ T cell function, we examined different CD4⁺ T cell populations and function by flow cytometry gating strategy shown in Supplemental Fig. 2F. We observed a statistically significant decrease in CD4⁺ Th1 effectors, defined by their expression of IFN-γ and TNF-α, in CMT167-shCIITA tumors compared with CMT167–NT-ctrl (Fig. 2I, 2J). csMHCII had no discernable impact on either IL4⁺ Th2 (Supplemental Fig. 2G) or IL17⁺ Th17 cells (Supplemental Fig. 2H). T regulatory cells, defined by Foxp3 expression, showed a trend toward being increased, but this did not reach statistical significance (Supplemental Fig. 2I). There were also no changes observed in CD69⁺ and CD44⁺ activation markers among CD4⁺ T cells from tumor-bearing lungs (Supplemental Fig. 2J, 2K). CD8⁺ T cells also had a decrease in effector function as defined by a decrease in IFN-γ production but no significant change in TNF-α production (Fig. 2K, Supplemental Fig. 2L) or expression of CD69⁺ and CD44⁺ activation markers (Supplemental Fig. 2M, 2N). There was no significant change in CTLA4 expression on CD4⁺ or CD8⁺ T cells or PD-1 expression on CD8⁺ T cells. However, CD4⁺ T cells had an increase in PD-1 in the CIITA knockdowns (Supplemental Fig. 2O–R). These data demonstrate that loss of csMHCII is associated with reduced antitumor Th1 responses and suggest that csMHCII influences the effector functions of the antitumor CD4⁺ T cell response.

NSCLC cells expressing MHCII directly induce CD4⁺ T cell cytokine production of a Th1 phenotype in vitro

To determine if csMHCII expression is functional and can directly activate CD4⁺ T cells, CD4⁺ T cells isolated from lungs of CMT167 tumor-bearing mice were cocultured overnight with CMT167–NT-ctrl or CMT167-shCIITA cells that had been pretreated with IFN-γ for 48 h to induce MHCII expression. CD4⁺ T cell activation was analyzed by flow cytometry. CD4⁺ T cells cocultured witht CMT167–NT-ctrl cells showed a significant increase in IFN-γ and TNF-α production, characteristic of a Th1 phenotype, that was no observed in either CD4⁺ T cells cultured alone or CD4⁺ T cells cocultured with CMT167-shCIITA cells (Fig. 3A, 3B, Supplemental Fig. 2S). CD4⁺ T cells cocultured with the CMT167–NT-ctrl cells did not show changes in IL-10 production associated with a Th2 phenotype or expression of Foxp3 (Fig. 3C, 3D). In addition, csMHCII did not influence the frequency of CD69⁺ events (Fig. 3E). These data provide direct evidence that csMHCII is functional, allowing direct interaction with antitumor CD4⁺ T cells to induce Th1 effector function using an in vitro coculture system.

FIGURE 3.

NSCLC cells expressing MHCII directly activate CD4⁺ T cells in vitro. Flow cytometric analysis of CD4⁺ T cells isolated from a CMT167 tumor-bearing lung and cocultured for 24 h with CMT167–NT-ctrl or CMT167-shCIITA, defined as singlets that are live, CD45⁺, or CD4⁺. (A) Percentage of IFN-γ⁺ CD4⁺ T cells, (B) percentage of TNF-γ⁺ CD4⁺ T cells, (C) percentage of IL-10⁺ CD4⁺ T cells, (D) percentage of Foxp3⁺ CD4⁺ T cells, and (E) percentage of CD69⁺ CD4⁺ T cells (n ≥ 3, one-way ANOVA). ***p < 0.001, **p < 0.01.

Effects of silencing csMHCII on the TME

To investigate PD-L1 expression in the TME, we injected CMT167–NT-ctrl and CMT167-shCIITA cells into GFP⁺ mice, recovered the GFP⁺ and GFP⁻ populations, and analyzed PD-L1 expression by flow cytometry. The GFP⁻ cancer cells had a decrease in PD-L1 expression when harvested from primary tumors (i.e., in vivo); however, there was no change in PD-L1 expression on the GFP⁺ cells in the TME as a whole (Fig. 4A, 4B). To further characterize the TME, we used spectral flow cytometry on the Cytek Aurora to look at 22 markers. We harvested mice with CMT167–NT-ctrl tumors and CMT167-shCIITA tumors. When we analyzed MHCII expression on all cells in the TME, MHC II did not change (Supplemental Fig. 1J); however, we were interested to see how MHCII was expressed on specific APC subsets. We analyzed the Cytek data with unsupervised clustering analysis, shown in a t-distributed stochastic neighbor embedding plot (Fig. 4C). There were changes in the B cell clusters and some macrophage clusters, but nothing reached statistical significance. To further analyze the data, we used biaxial gating to characterize the populations in the TME, as previously defined (25, 28). There was an increase in B cells in the CIITA knockdown tumors, and these B cells had increased expression of PD-L1 and MHCII (Fig. 4D–F). In the CIITA knockdowns we detected no change in resident alveolar macrophages, but there was a decreased frequency of recruited macrophages. The recruited macrophages had a slight increase in MHCII expression and no changes in PD-L1 expression (Fig. 4G–J). There was no change in dendritic cell or neutrophil recruitment: however, dendritic cells showed a slight increase in MHCII expression, although they did not reach statistical significance (Fig. 4K–N). These data suggest that in the setting of less MHCII Ag presentation by cancer cells (i.e., lack of csMHCII), there may be a compensatory increase in B cells and a decrease in recruited macrophages.

FIGURE 4.

Effects of silencing csMHCII on the TME. Flow cytometric analysis of percentage of PD-L1 of (A) GFP⁻ and (B) GFP⁺ cells from CMT167–NT-ctrl and CMT167-shCIITA tumors injected into GFP⁺ mice and FACs sorted into GFP⁻ and GFP⁺ populations (n = 3, unpaired t test). (C) t-Distributed stochastic neighbor embedding (tsne) plot of unsupervised clustering analysis of spectral flow cytometry–based analysis of CMT167–NT-ctrl and CMT167-shCIITA. Biaxial-based analysis of spectral flow cytometry–based analysis of singlets that are live to identify (D) B cells (percentage of CD45⁺, CD19⁺), (E) percentage of B cells (percentage of CD45⁺, CD19⁺) that are MHCII⁺, (F) percentage of B cells (percentage of CD45⁺, CD19⁺) that are PD-L1⁺, (G) alveolar macrophages (CD45⁺, SiglecF⁺, Ly6G⁻, CD11c⁺), (H) recruited macrophages (CD45⁺, SiglecF⁻, Ly6G⁻, CD11b⁺), (I) percentage of recruited macrophages (CD45⁺, SiglecF⁻, Ly6G⁻, CD11b⁺) that are MHCII⁺, (J) percentage of recruited macrophages (CD45⁺, SiglecF⁻, Ly6G⁻, CD11b⁺) that are PD-L1⁺, (K) dendritic cells (CD45⁺, MHCII⁺, SiglecF⁻, Ly6G⁻, CD11c⁺), (L) percentage of dendritic cells (CD45⁺, MHCII⁺, SiglecF⁻, Ly6G⁻, CD11c⁺) that are MHCII⁺, (M) percentage of dendritic cells (CD45⁺, MHCII⁺, SiglecF⁻, Ly6G², CD11c+) that are PD-L1+, and (N) neutrophils (CD45+, SiglecF², Ly6G+) (n = 3, unpaired t test). *p < 0.05.

Overexpression of CIITA in LLC cells induces MHCII and increases CD4⁺ and CD8⁺ tumor infiltration

Because loss of csMHCII in CMT167 resulted in decreased T cell infiltration into tumors, we sought to determine if induction of csMHCII in LLC cells, a cell line that does not express MHCII, would be sufficient to enhance T cell detection and retention within the tumor. LLC cells were transfected with a lentiviral expression vector (LLC-CIITAOE) or an empty vector control (LLC–EV-ctrl). CIITA overexpression induced Ciita mRNA expression and surface MHCII expression at baseline and expression of MHCII-associated genes (Fig. 5A, 5B, Supplemental Fig. 3A, 3B). Tumor-bearing lungs were stained for tumor-infiltrating T cells. LLC-CIITAOE tumors had increased infiltration of CD3⁺ (EV-ctrl 36.35 cells/HPF ± 12.3 versus CIITAOE 89.93 cells/HPF ± 8.766), CD4⁺ (EV-ctrl 19.48 cells/HPF ± 9.595 versus CIITAOE 37.41 cells/HPF ± 7.881), and CD8⁺ T cells (EV-ctrl 17.26 cells/HPF ± 3.014 versus CIITAOE 59.75 cells/HPF ± 22.46) (Fig. 5C). Representative images for LLC–EV-ctrl and LLC-CIITAOE tumors are shown in (Fig. 5D).

FIGURE 5.

Overexpression of CIITA in LLC cells induces MHCII and increases CD4⁺ and CD8⁺ tumor infiltration. (A) Fold change of Ciita mRNA expression for LLC–EV-ctrl and LLC-CIITAOE (n = 3, unpaired t test). (B) Flow cytometry analysis of MHCII on LLC–EV-ctrl and LLC-CIITAOE cells gated on live singlets (n = 3, unpaired t test). (C) Average count of immunofluorescence of CD3⁺, CD4⁺, and CD8⁺ T cells per HPF (original magnification ×40) in LLC–EV-ctrl tumors and LLC-CIITAOE tumors (n = 3, n = 4, respectively; unpaired t test). (D) Representative pictures of LLC–EV-ctrl and LLC-CIITAOE tumors immunofluorescence: DAPI (blue), CD3 (red), and CD4 or CD8 (green). *p < 0.05, **p < 0.01, ****p < 0.0001.

Effects of csMHCII expression on tumor progression and response to anti–PD-1

To determine the effect of silencing csMHCII on tumor progression and response to immunotherapy, equal numbers of CMT167–NT-ctrl or CMT167-shCIITA cells were injected into the lungs of wild-type mice, and primary tumor size was measured at 4 wk. Compared with CMT167–NT-ctrl tumors, the CMT167-shCIITA tumors had an increase in tumor size (NT-ctrl 15.07 mm³ ± 1.674 versus shCIITA 21.37 mm³ ± 1.821) (Fig. 6A). Separate groups of mice were treated starting 7 d after tumor implantation with anti–PD-1 Ab or IgG isotype control, as previously described (24). Anti–PD-1 therapy resulted in almost complete ablation of CMT167–NT-ctrl tumors, similar to what we have previously shown with parental CMT167 tumors (24). In contrast, CMT167-shCIITA tumors had a diminished response to anti–PD-1 therapy, with only a 50% inhibition of tumor growth in response to anti–PD-1 (Fig. 6B). These data demonstrate that csMHCII is necessary for optimal response to anti-PD-1 therapy. Thus, loss of csMHCII expression increases tumor progression and at least partially reverses responsiveness to anti–PD-1.

FIGURE 6.

MHCII mediates the tumor progression and response to checkpoints. (A) Primary tumor volume of CMT167–NT-ctrl or CMT167-shCIITA tumors at 4 wk (n ≥ 17, unpaired t test). (B) Primary tumor volume of CMT167–NT-ctrl or CMT167-shCIITA 4.5 wk postimplantation treated with IgG Ab or PD-1 Ab (n ≥ 11, two-way ANOVA). (C) Primary tumor volume of LLC–EV-ctrl or LLC-CIITAOE treated with IgG or PD-1 Ab at 3 wk (n = 5, two-way ANOVA). *p < 0.05, **p < 0.01.

To determine if induction of csMHCII in LLC cells would be sufficient to sensitize the tumors to anti–PD-1 in vivo, LLC–EV-ctrl or CIITAOE cells were injected into C57BL/6 wild-type mice. Tumors were allowed to develop for 7 d, and then mice were treated with either an IgG control or anti–PD-1 Ab for 2 wk. In mice injected with control Ab, LLC–EV-ctrl and LLC-CIITAOE tumors had comparable growth, indicating that enforced csMHCII expression was not sufficient to alter tumorigenesis under baseline conditions in vivo. However, whereas anti–PD-1 Ab had no impact on tumorigenesis in mice injected with LLC–EV-ctrl cells, we observed an ~70% decrease in primary tumor volume in mice implanted with LLC-CIITAOE tumor cells (Fig. 6C). In total, these data demonstrate that csMHCII can regulate responses to anti–PD-1 therapy in mouse models of NSCLC.

Adenocarcinoma lung cancer expresses HLA-DR (MHCII) and is correlated with enhanced CD3⁺, CD4⁺, and CD8⁺ tumor infiltrating lymphocytes in human lung tumors

To extend our findings on csMHCII to human lung cancer, we quantified MHCI and MHCII expression in a panel of human NSCLC cell lines using RNA-seq data from the Cancer Cell Line Encyclopedia across 15 KRAS mutant lung cancer cell lines (27). MHCI mRNA expression (HLA-A, HLA-B, HLA-C) and MHCI accessory genes were high across almost all lung adenocarcinoma cell lines (Supplemental Fig. 3C). However, MHCII mRNA expression (HLA-DR and HLA-DQ) and MHCII accessory genes were only expressed in a subset of cell lines (Supplemental Fig. 3D). We validated HLA-DR, HLA-DQ, and CD74 protein expression in a subset of these human lung cancer cell lines using flow cytometry (Supplemental Fig. 3E). The cell lines could be divided into three subsets: 1) high basal levels (H1373), 2) low basal levels that were not altered by IFN-γ (H1155), and 3) low levels with the ability to induce MHCII in response to IFN-γ (A549). H1155 had low basal levels and did not induce HLA-DR, A549 had low basal levels and induced HLA-DR, and H1373 had high basal levels with the ability to induce HLA-DR (Supplemental Fig. 3F).

In addition to characterizing MHCII expression in human lung cancer cell lines, we quantified csMHCII in primary human lung adenocarcinomas. To do this, we applied multispectral imaging using the Vectra 3.0 system to quantify csMHCII in a panel of 90 resected human lung adenocarcinomas patients. To define csMHCII expression, we stained tissue sections for CK, marking the cancer cells and HLA-DR, a human MHCII molecule, and analyzed the incidence of double-positive staining within the tumor. Using an arbitrary cutoff, we defined csMHCII⁺ tumors as those exhibiting ≥10% double-positive cells (CK⁺, HLA-DR⁺). Cancer-specific positive HLA-DR (≥10%) was observed in 51 of the 90 tumors investigated (57%), consistent with previous findings (32). Patients with HLA-DR⁺ tumors (≥10%) had a 5-y overall survival benefit (Fig. 7A). Tumor cell expression of HLA-DR⁺ events was positively correlated with CD4⁺ events in the tumor (Fig. 7B). In addition, HLA-DR–high (with ≥10% HLA-DR⁺) tumor cells had an increase in CD4⁺ T cell infiltration and a slight increase in CD8⁺ T cell infiltration, although it did not reach statistical significance, recapitulating our results in the murine model (Fig. 7C, 7D). Representative images show DAPI (blue), CK (magenta), HLA-DR (yellow), CD3 (green), and CD8 (red) of an HLA-DR–low tumor and an HLA-DR–high tumor (Fig. 7E, 7F, respectively). These data demonstrate that csMHCII is a feature of many human lung adenocarcinomas and that csMHCII is positively associated with T cell infiltration.

FIGURE 7.

Adenocarcinoma lung cancer expresses HLA-DR (MHCII) and is correlated with enhanced CD3⁺, CD4⁺, and CD8⁺ tumor infiltrating lymphocytes. (A) Overall 5-y survival curve stratified by HLA-DR–low (<10%) and HLA-DR–high (≥10%) patients (n = 90, log-rank (Mantel–Cox) test). (B) The correlation of the percentage of the number of counts of HLA-DR⁺ on cancer cells and the percentage of the number of counts of CD8 (r² = 0.0069, p = 0.439) and CD4 (CD3⁺, CD8⁻) (r² = 0.1227) in the tumor. (C) Percentage of CD4⁺ and (D) CD8⁺ T cell infiltration in the tumor in patients stratified by HLA-DR low (<10%) and HLA-DR high (≥10%) (n = 90, unpaired t test). Representative images of the composite image after spectral unmixing. DAPI nuclear marker (pseudocolored blue), CK (membrane, Cy3, pseudocolored magenta), HLA-DR (membrane, Cy5, pseudocolored yellow), CD3 (membrane, FITC, pseudocolored green), and CD8 (membrane, Texas Red, pseudocolored red) of (E) HLA-DR–low tumor of patient 28 and (F) HLA-DR–high tumor of patient 3. *p < 0.05, **p < 0.001, ***p < 0.0001.

Discussion

The ability of the immune system to recognize Ags is critical for tumor elimination by T cells. Classically, this is mediated through presentation of Ags via MHCI expressed on cancer cells and MHCII expressed on professional APCs. However, non-APCs, such as endothelial cells and epithelial cells (including lung epithelial cells), can also express MHCII (33–36). These cells induce MHCII expression in response to IFN-γ, with the capability to directly present Ag to CD4⁺ T cells (37, 38). Although the potential impact of MHCII expression on antitumor immunity has been appreciated for years, recent studies have provided new evidence for the ability of MHCII to influence the mutational landscape of diverse tumor types (39). Furthermore, MHCII expression on cancer cells has been observed in a variety of cancers and in general correlates with improved outcomes. In colon cancer, high csMHCII expression is associated with increased survival (18). In melanoma, csMHCII expression is associated with a better response to checkpoint therapy blockade with an increase in progression-free and overall survival in anti–PD-1/PD-L1–treated patients stratified by csMHCII positivity (20). In breast cancer, csMHCII expression is downregulated as a consequence of decreased CIITA in highly metastatic cells, and in triple-negative breast cancer, tumor-specific MHCII is associated with increased progression-free survival (22, 40).

Our interest in csMHCII began with the finding that in our mouse model, csMHCII was correlated with response to immunotherapy, with anti-PD-1–sensitive CMT167 tumors characterized by csMHCII expression and anti–PD-1–resistant LLC tumors failing to express csMHCII. A major goal of these studies was to identify whether csMHCII had a functional consequence on shaping the TME and/or the response of tumors to immunotherapy. Silencing CIITA, and in turn csMHCII, in CMT167 cells resulted in slightly larger tumors under baseline conditions. CIITA knockdown was further associated with reduced CD4⁺ helper T cell infiltration and Th1 cytokine production as well as IFN-γ-expressing CD8⁺ T cells. This was not a consequence of alterations in MHCI on the cancer cells. Using multiparameter spectral flow cytometry analysis, we further demonstrated that silencing CIITA causes more global changes in the TME. We detected decreases in recruited macrophage populations and increases in B cells. Importantly, silencing CIITA in cancer cells resulted in decreased production in the TME of CXCL9, a chemokine critical for T cell recruitment. Consistent with decreased T cell infiltration into the tumors, csMHCII had a pronounced role in regulating the response to anti–PD-1–targeted immunotherapy in CMT167 tumors; inhibiting CIITA and csMHCII expression converted an immunotherapy-sensitive tumor to an immunotherapy-resistant phenotype. Conversely, expression of CIITA and csMHCII in LLC cells increased T cell infiltration into these tumors and converted them from immunotherapy resistant to immunotherapy sensitive.

A critical consequence of csMHCII is the ability of cancer cells to directly present tumor neoantigens to CD4⁺ T cells, providing an APC-independent mechanism for CD4⁺ T cell activation in the TME. The capacity of cancer cells to directly activate CD4⁺ T cells was confirmed by in vitro coculture studies of cancer cells and T cells, in which cancer cells lacking csMHCII expression showed a significant decrease in effector cytokine production of Th1 CD4⁺ T cells compared with cancer cells expressing MHCII. For CD4⁺ T cells to be activated, they require a second signal in addition to Ag presentation by MHCII. A potential second signal from the cancer cells to the CD4⁺ T cells could be via ICAM-1, which is expressed on endothelial and epithelial cells and can activate CD4⁺ T cells by binding to LFA-1, leading to a Th1 phenotype. By immunoblotting, our cancer cells express ICAM-1 (data not shown). The identity of tumor-specific MHCII neoantigens in the current models remains unclear at this time, but one likely source for these peptides is likely to arise from autophagy, a cellular process associated with the generation of MHCII epitopes (41, 42). It is also possible that neoantigens presented by the cancer cells may be distinct from those presented by MHCII on professional APCs in the TME. The unique contribution of csMHCII is very likely contributed by the distinct peptide binding characteristics of MHCII compared with MHCI (43), with MHCII peptide presentation linked to tumor control in different contexts (44). A recent study has examined specific CD4⁺ T cell neoantigens in human lung tumors, and although most CD4⁺ T cells recognized passenger mutations, a subset recognized the oncogenic driver (45). The status of csMHCII was not examined in that study. Another study showed that in sarcoma, MHCII neoantigens at the tumor are necessary for immune checkpoint therapy (46). Further investigation of the diversity of TCRs elicited in the presence or absence of csMHCII will be particularly informative in defining the impact of csMHCII on shaping antitumor immunity.

These data implicate csMHCII as a modifier of the TME, promoting T cell infiltration and/or T cell retention in the TME. Our results are consistent with a model in which direct interactions between cancer cells expressing MHCII and tumor-specific CD4⁺ T cells result in CD4⁺ T cell activation, leading to local IFN-γ expression in the TME. This in turn leads to IFN-γ–dependent changes, including enhanced CXCL9 production and PD-L1 expression, which in turn causes further T cell recruitment. This positive feedback mechanism culminates in a T cell–inflamed, immunotherapy-sensitive tumor phenotype (24). csMHCII may further enhance T cells in the TME through direct CD4⁺ T cell engagement of cancer cells, providing a retention signal for tumor-reactive CD4⁺ T cells. It has been proposed that in some cases cancer immunotherapies achieve limited results because of a lack of successful activation of antitumor CD4⁺ T cells (47, 48). Specifically, Th1 CD4⁺ T cells can play a crucial role in the TME, facilitating optimal CD8⁺ T cell activation and effector function. Th1 cytokines, including IFN-γ and TNF-α, can further induce cancer cell senescence (31). Although other factors, such as mutational burden and presence of neoantigens, contribute to T cell infiltration and response to immunotherapy (49), one major distinction for our studies was the use of isogenic cells in which only the expression of CIITA has been altered. Thus, our data support a model for csMHCII as a biomarker of sensitivity to checkpoint inhibitors independent of other factors, including MHCII expression by other cells in the TME. Although csMHCII expression has been reported on NSCLC and associated with lymphocytic infiltration (32, 50), this was not previously correlated with response to immunotherapy. Our data show that csMHCII is expressed in a subset of unselected human lung adenocarcinomas and is correlated with an increase in T cell tumor infiltration and a 5-y overall survival benefit. However, oncogenic drivers were not defined for many of these patients. Additional studies on a larger cohort in which oncogenic drivers have been defined and response to immunotherapy can be correlated are necessary to better define effects of csMHCII on survival and response to anti–PD-1 immunotherapy. Our studies provide direct evidence for the functional importance of MHCII expression on cancer cells and illustrate that MHCII expression on professional APCs, including dendritic cells, may not be sufficient to confer susceptibility to immunotherapy.

Abbreviations used in this article

CK

cytokeratin

csMHCII

cancer cell–specific MHCII

HPF

high-power field

Fwd

forward

LLC

Lewis lung carcinoma

MHCI

MHC class I

MHCII

MHC class II

NSCLC

non–small cell lung cancer

NT-ctrl

nontargeting control

PD-1

programmed cell death protein-1

PD-L1

programmed cell death ligand-1

qRT-PCR

quantitative real-time PCR

Rev

reverse

RNA-seq

RNA sequencing

shCIITA

shRNA targeting CIITA

shRNA

short hairpin RNA

TME

tumor microenvironment