Introduction
Immunotherapy by blocking the axis of the immune checkpoint molecule programmed cell death protein (PD1) or its ligand PDL1 has presented remarkable survival benefit and thus became a frontline treatment for metastatic or unresectable melanoma [
1,
2]. While the survival in advanced melanoma has improved substantially since FDA approved PD1/PDL1 blockade therapy, objective response rates approach only 50% at best with checkpoint inhibitors combination therapy [
3]. Combination of anti-PD1 antibody and the antibody to the cytotoxic T cell lymphocyte-associated protein 4 (CTLA-4) slightly increased the response rate compared to anti-PD-1 monotherapy, however with significantly increased autoimmune toxicity [
3]. Therefore, the need to increase the response rate with new combination therapy without increased toxicity is still imperative.
Identifying tumor-derived targetable factors that may impact patients’ response to PD1/PDL1 blockade would rationalize a potential combination therapy to improve the clinical outcome. In recent clinical studies, the presence of circulating soluble NKG2D ligands was shown to be negatively correlated with clinical outcome to anti-PD1/PDL1 response [
4,
5]. In humans, there are two major family members of NKG2D ligands, the MHC I-chain-related molecules MICA and MICB and the viral HCMV UL16-binding proteins ULBP1-6 family. NKG2D ligands are rarely expressed by normal tissues unless under stress insults, such as infection [
6], but induced in most tumor cells in part through activation of DNA damage response pathway or oxidative stress [
7‐
9]. Although the MICA and MICB family molecules are better characterized and more prevalently expressed than the ULBP family proteins, these ligands often co-exist in one tumor type, presumably through host-viral co-evolutionary processes [
10]. While levels of the NKG2D ligand MIC have been correlated with survival benefits in the early stages of several cancers, the opposite has been demonstrated with more invasive tumors [
11‐
14]. In most invasive human tumors, NKG2D ligands also exist as a soluble form through proteolytic shedding or exosome secretion [
15]. Soluble human NKG2D ligands have been shown to subvert antitumor immunity through multiple mechanisms, including but not limited to, perturbing NK cell homeostatic maintenance and function, impairing CD8 T cell function by destabilizing CD3ζ [
16], and expanding myeloid-derived suppressive cells (MDSCs) in the tumor microenvironment [
17]. These mechanistic understandings along with the reported clinical observations prompt us to test the hypothesis that co-targeting tumor-derived soluble NKG2D ligands would enhance melanoma tumor response to PD/PDL1 blockade therapy.
We have previously described that clearance of tumor-derived soluble NKG2D ligands, sMIC, with a monoclonal antibody (mAb) B10G5 restores NK cell homeostatic renewal, enhances NK cell and antigen-specific CD8 T cell function, immobilizes NK and CD8 T cell to the tumors, and re-modulates tumor microenvironment by eliminating MDSCs and tumor-associated macrophages [
17,
18]. In this study, we demonstrate that antibody targeting sMIC increases the IL-2 sensing receptor IL-2Rα on NK cells, reprograms NK cell homeostatic maintenance, and enhances the therapeutic response of melanoma tumors to PD1/PDL1 blockade therapy. Our current study provides a new mechanistic understanding to accentuate the significance of co-targeting tumor-derived soluble NKG2D ligand, sMIC, to enhance the therapeutic efficacy of PD1/PDL1 checkpoint blockade therapy for melanoma patients.
Materials and methods
Mice and cell lines
Mice were bred and housed under specific pathogen-free conditions in the animal facility of the Medical University of South Carolina and Northwestern University in accordance with institutional guidelines with approved IACUC protocols. All mice used in this study were male rPB-MICB mice on the B6 background as previously described and thereafter defined as MICB/B6 mice [
19]. sMIC-expressing B16F10-sMICB cell line was developed by transduction of B16F10 cells (ATCC) with an IRES-GFP retroviral vector containing the construct for recombinant soluble MICB, as described previously [
20]. sMIC
+ B16F10 cells were selected by puromycin and further by flow cytometry sorting for GFP-positive cells.
Antibodies, peptides, and tetramers
InVivoMAb anti-mouse PDL1 (clone 10F.9G2) was purchased from BioXCell. Generation of the anti-MIC mAb B10G5 was previously described [
19]. B10G5 is a mouse IgG1 isotype, recognizing both MICA and MICB. B10G5 binds to free sMIC but does not block the interaction of sMIC with the receptor NKG2D [
18]. B10G5 was produced and purified from the hybridoma culture by BioXCell (West Lebanon, NH). In vivo NK cell-depleting anti-NK1.1 (clone PK136) and CD8 T cell-depleting antibody anti-CD8α (clone 2.43) were purchased from BioXcell. Peptide gp100
25-33 (KVPRNQDWL) was synthesized by GenScript. H-2D
b/ gp100
25-33 tetramer was produced by NIH Tetramer Core Facility at Emory University.
Tumor inoculation and in vivo experiments
For subcutaneous study, B16F10-sMICB cells were implanted subcutaneously into the right flank of cohorts of syngeneic MICB/B6 male mice (4 × 105 cells/mouse) at ages 8–10 weeks old. When tumor volume reached approximately 75–100 mm3, animals were randomized into four therapy groups (n = 5 to 7 per group): (1) control mouse IgG (3.0 mg/kg BW), (2) anti-MIC mAb B10G5 (3.0 mg/kg BW), (3) anti-PDL1 mAb (3.0 mg/kg BW), and (4) B10G5 and anti-PDL1 mAb. All antibodies were given via I.P. injection every 3 days. For survival studies, tumor volume of 1800 mm3 was defined as survival endpoint. For mechanistic studies, animals were euthanized after 9 days of treatment. After euthanization, the spleens and two inguinal draining lymph nodes (dLN) and tumors were harvested. Partial of the tumors were formalin fixed, paraffin embedded, and sectioned for histology and immunohistochemistry staining (IHC). The remaining tumors were used for single-cell suspension preparation by the method of mincing, mechanically processing, and passing through a 70-μm filter. Single-cell suspension of splenocytes, dLN, and tumors was used for ex vivo stimulation and flow cytometry analyses.
For lung metastasis, B16F10-sMICB cells were injected into the lateral tail vein of syngeneic B6/MICB male mice (2 × 105 cells/mouse) at ages 8–10 weeks old. At day 10 post-tumor inoculation at which time point tumor nodules were visible on the surface of the lung by random examination of three animals, mice were randomized into four therapy groups (n = 5 per group): (1) control mouse IgG (3.0 mg/kg BW), (2) anti-MIC mAb B10G5 (3.0 mg/kg BW), (3) anti-PDL1 mAb (3.0 mg/kg BW), and (4) B10G5 and anti-PDL1 mAb. All antibodies were given via I.P. injection every 3 days. Animals were euthanized at day 21 following tumor inoculation. Spleens, inguinal draining lymph nodes, and lungs were harvested for analyses.
Ex vivo cytokine re-stimulation assay
For general re-stimulation, single-cell suspensions of splenocytes and draining lymph nodes were stimulated at 37°C for 6 h with 50 ng/ml phorbol myristate acetate (PMA) and 500 ng/ml ionomycin. To assess melanoma antigen-specific T cell function, single-cell suspension of bulked splenocytes or tumor-draining lymph nodes was stimulated with 1 μg/ml of melanoma antigen gp10025-33 peptides overnight. IFNγ production was assayed by intracellular staining with BD IFNγ staining Kits following the manufacturer’s instruction.
Flow cytometry analysis
Single-cell suspensions were incubated on ice for 30 min with a combination of antibodies specific to cell surface markers for identification of lymphocyte subsets. These antibodies are anti-NK1.1 (clone PK136), anti-CD3 (clone 145–2C11), anti-CD8α (clone 53–6.7), anti-NKG2D (clone CX5), anti-CD44 (clone IM7), anti-CD25 (clone PC61), anti-Gr1 (clone RB6-8C5), and anti-CD11b (clone ICRF44). All antibodies used for flow cytometry analyses were purchased from Biolegend (San Diego, CA, USA). Tetramer staining was performed with 2 μg/ml of PE-labeled H-2Db/ gp10025-33 tetramer at 37 °C for 20 min and followed by surface marker staining. For intracellular staining, cells were stained with surface markers followed by fixation and permeabilization with BD Perm/Fix kits and antibodies specific to intracellular molecules. Cells were analyzed using the BD Fortessa. Data were analyzed using the FlowJo software (Tree Star).
Histological and immunohistochemistry staining
Five micrometers of formalin-fixed paraffin-embedded sections were stained with H&E for pathological evaluation and used for immunohistochemistry (IHC) staining. Mouse tumor sections were also stained with the following: (a) anti-NKp46/NCR1 (rabbit IgG; 1:200; Abcam); (b) anti-CD8 (BD biosciences); (c) anti-arginase 1 (rabbit IgG; 1:200; Santa Cruz Biotechnology); (d) anti-CD31 (rabbit IgG; 1:100; Abcam); and (e) anti-Ki67 (rabbit IgG, Abcam, 1 μg/ml). Human tissue microarray (TMA) sections containing 62 cases of malignant melanoma, 21 metastatic malignant melanoma, and other control tissues were purchase from US Biomax (Cat. ME1004g) and were stained with the mouse monoclonal anti-MIC antibody D4H3 (
Supplement Material and Methods). Sections were deparaffinized and incubated for 10 min in 10 mM citrate buffer (pH 6.0) at 95 °C for antigen retrieval. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide. After quenching endogenous peroxidase activity and blocking nonspecific binding, slides were incubated with specific primary antibody overnight at 4 °C followed by subsequent incubation with the appropriate biotinylated secondary antibody: goat anti-rabbit IgG or goat anti-mouse (Vector) at a 1:1000 dilution for 20 min at 37 °C. Immunoreactive antigens were detected using the Vectastain Elite ABC Immunoperoxidase Kit and DAB. All slides were counterstained with hematoxylin (Vector) and mounted with Permount (Fisher Scientific).
RNAseq and data analyses
Single-cell suspension from the spleens from SCID mice was prepared as described [
20]. After removal of adherent cells for 2 h in complete media, splenocytes were cultured in media containing 1000 U/ml IL-2 for 3 days. NK cells were negatively selected with EasySep™ mouse NK isolation kits (StemCell Technologies). A 99% purity was obtained. Purified NK cells were cultured with purified recombinant sMIC(B)-His (Sino Biologicals) or sMIC(B)-Fc (R&D) for 12 h, with or without the presence of the anti-sMIC mAb B10G5. Total RNA was prepared with RNeasy kit (Qiagen). RNAseq library were constructed with Illumina TruSeq Stranded mRNA Library Prep Kit. Twenty to 32 million of RNAseq reads were obtained for each sample using single-end 50 bp sequencing. Trim Galore (
https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) was used to validate the quality of the reads and to remove ones with low quality by default parameters. STAR program was used to align the reads against the mouse reference genome mm10 with the transcriptome annotation GTF file from ENSEMBL (GRCm38.82) via the default parameters [
21]. FeatureCounts was used to calculate gene expression, represented as transcript per million (TPM) values [
22]. Only genes showing TPM value greater than 4 in at least one sample were included in the downstream differential gene expression analyses. DAVID online tool (
http://david.abcc.ncifcrf.gov/) was used for the gene enrichment analysis. Genes showing consistent differential expression (> 2 fold in two biological replicates) upon sMIC treatments as compared to the control sample were selected for the Gene Ontology (GO) analysis and for the heatmap.
Quantitative RT-PCR
Total RNA was prepared as described above. Complementary DNA (cDNA) was synthesized using the SuperScript II kit (Invitrogen). A volume of 1 μl of cDNA was mixed with Power SYBR Green qPCR SuperMix (Bio-Rad, USA), and specific primer sets were added to a final concentration of 400 nM in 20 μl of reaction mixture. The reaction was performed on a Bio-Rad CFX96 Touch Real-Time PCR Detection System. Data were analyzed using CFX Maestro Software (BioRad). Each sample was assayed in triplicates. Target mRNA levels were normalized against mouse GAPDH. Gene expression level in control NK cells was used as a reference for calculating expression fold changes. The primers used are listed in Supplement Table S
1.
Statistics
All statistical data were expressed as mean ± SEM. Difference between means of populations was compared by standard Student’s t test using one-way ANOVA. Survival was determined via Kaplan-Meier analysis with a comparison of curves using the Mantel-Haenszel log-rank test. A P value of 0.05 or less was considered significant. GraphPad Prism software was used for all analyses.
Discussion
With syngeneic subcutaneous and metastatic tumor models, we demonstrate in this study significantly enhanced and cooperative therapeutic effects against melanoma with the combination of the sMIC-clearing antibody B105 and PDL1/PD1 pathway blockade. We have shown that combination therapy with B10G5 and anti-PDL1 resulted in significantly more inhibition of primary tumor growth and higher degree of clearance of lung metastases than respective monotherapy. We demonstrated that the enhanced therapeutic effect with combination therapy is associated with augmented NK and CD8 T cell activation and anti-tumor potential. We also demonstrated that combination therapy cooperatively reduces expression of immunosuppressive arginase I in the tumor microenvironment and inhibits tumor angiogenesis. Interestingly, we demonstrate that expression of the receptor to enable NK cell to sensitize low dose IL-2, the IL-2R
a, on NK cells was upregulated with B10G5 therapy and further synergistically upregulated with the combination therapy. With RNAseq analyses, we demonstrated that clearance of sMIC with B10G5 rescues the impairment of NK cell survival and proliferation pathway caused by sMIC. Finally, we demonstrate the abundance or prevalence of MIC/sMIC in metastatic melanoma tumors. Given recent clinical findings of an association between high levels of circulating NKG2D ligand sMIC and reduced response to PD1 blockade therapy in melanoma patients [
4], our current study offers a viable new combination therapy to improve the response of melanoma patients to PD1/PDL1 inhibitors.
Our study demonstrates that targeting sMIC increases the IL-2 sensing receptor IL-2Rα on NK cells in vivo and that sMIC/antibody complex reprograms NK cell for homeostatic survival in vitro, although targeting sMIC enhances IL-2Ra expression on NK cells warrants further investigations. The increased IL-2Ra expression may be important for the cross-talk of NK cells and the adaptive immune response in the response to PD1/PDL1 blockade therapy. Clinical correlative studies have revealed the significance of NK cell in association with response to PD1/PDL1 blockade therapy. In metastatic melanoma patients, the expression of CD25 on NK cells has been associated with clinical response to anti-PD1 therapy [
35], presumably in part due to increased NK cell sensitivity to IL-2 and thus increased survival and function. A higher density or frequency of peritumoral NK cells was found to be associated with response to anti-PD1 therapy in metastatic melanoma patients [
36,
37]. A concurrent upregulation of NK cell activity related genes and MHC I in tumors that responded to PD1/PDL1 blockade therapy in melanoma patients [
36]. A clustering of NK cells and stimulatory dendritic cells (DCs) was in tumors of melanoma patients who responded to anti-PD1 therapy and had prolonged survival [
37]. In preclinical melanoma models, it was found that NK cells, not T cells, are required for sustaining stimulatory DC in the tumors [
37]. Together, these studies underscore the significance of NK cells in mediating tumor response to PD1/PDL1 blockade therapy.
Monotherapy with B10G5 to clear sMIC was efficacious in controlling tumor growth and eliminating lung metastasis. However, combination therapy presented a cooperative and significantly enhanced effect. sMIC has been to be highly immune-suppressive via perturbing NK cell peripheral maintenance and function [
19], impairing TCR/CD3 signaling by caspase-dependent destabilization of CD3ζ [
28], and facilitating the expansion of MDSCs and tumor-associated macrophages [
17]. These immune-suppressive effects can directly and indirectly impair CD8 T cell activation, and thus negatively impact the response to PD1 blockade therapy. We have previously shown that clearing sMIC with B10G5, a nonblocking anti-sMIC antibody, can restore NK cell homeostatic maintenance and function [
18], alleviate the immune-suppressive tumor microenvironment by eliminating arginase I
+ MDSCs and tumor-associated macrophages [
17], stabilize CD3ζ expression on CD8 T cells [
28], and enhance CD28-NKG2D dual co-stimulation to antigen-specific CD8 T cells [
38]. These profound therapeutic effects elicited by clearing sMIC could potentiate the CD8 T cell response, in particular the tumor antigen-specific CD8 T cell response, and thus cooperatively enhance the response to PDL1/PD1 blockade therapy. Of note, the B16-sMICB tumor cell line in this study is PDL1
+; thus, the cooperative therapeutic effect of B10G5 and anti-PDL1 demonstrated is sound. Interestingly, patients bearing tumors that are initially PDL1 negative still clinically responded to PD-1 blockade therapy [
39‐
43]. Given that we have previously shown that clearing sMIC with B10G5 can induce the release of IFNγ [
18,
28,
44], a significant regulator of PDL1 expression, one might speculate the combination of B10G5 and PD1 blockade may also be effective for tumors initially lacking PDL1 expression. Indeed, we show that clearance of sMIC rescues NK cell survival and function, presumably in part attributing to enhanced sensitivity to IL-2 released by activated antigen-specific CD8 T cells. Other studies have also presented that blocking sMIC release in preclinical models enhances NK cell function [
45]. Considering that NK cells are the major IFNγ producers in active immune responses, the interaction or inter-dependence of NK cells and CD8 T cells through effector cytokine, such as IFNγ and IL-2, may account for one of the mechanisms mediating the synergistic effect of the combination therapy. Based on published studies and our current data, we propose an innate-adaptive cross-talk model that confers the cooperative therapeutic effect of targeting sMIC and PD1/PDL1 blockade (Supplement Figure S
5). How NKG2D signaling and blocking PD1 signaling synergistically enhance NK cell peripheral maintenance and function as we have shown warrants a further investigation.
Checkpoint inhibitors, particularly PD1 pathway blockade, have been approved by the FDA for a number of indications, including advanced melanoma, head and neck cancer, renal cell carcinoma, non-small cell lung cancer (NSCLC), urothelial carcinoma, and metastatic Merkel-cell carcinoma, due to enhanced survival benefits compared to traditional chemotherapy. However, complete clinical responses are still limited to a small percentage of patients. The anti-PD1 antibody nivolumab only demonstrated survival benefit in advanced melanoma patients without the BRAF V600 mutation, with a 72.9% survival rate at 1 year and increase in median progression-free survival by nearly 3 months as a second-line therapy in patients that progressed after receiving ipilimumab and/or BRAF inhibitor therapy [
46]. Overall, PD1/PDL1 blockade only elicited cumulative response rates of 31% in patients with melanoma, 19% in patients with NSCLC, 25% in patients with renal cell carcinoma (RCC), and 13.3% in patients with head and neck cancers [
47‐
50]. PD1/PDL1 therapy currently is also in phase III clinical trials for the indications of BRAF V600-mutated melanoma [
51], RCC [
18,
30], head and neck squamous cell carcinoma (HNSCC) [
31,
52], nasopharyngeal cancer [
53], esophageal carcinoma [
54], mesothelioma [
55], hepatocellular carcinoma [
48], breast cancer [
20], and multiple myeloma [
4]. Elevated levels of sMIC have been reported in almost all of these indications in association with reduced anti-tumor immunity or poor disease prognosis via common immune-suppressive pathways [
19,
56‐
59]. Our current study further accentuates the potential of enhancing the efficacy of PD1/PDL1 therapy in these malignant indications by targeting sMIC.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.