All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) protocol at Medical University of South Carolina and Northwestern University. All mice are maintained at respective Institutional animal facility under specific pathogen-free conditions. Generation and characterization of the TRAMP/MICB mice have been previously described [27]. Briefly, the human MICB was overexpressed in the prostate under the hormone-sensitive rat probasin (rPB) promoter to generate MICB/B6 mice. The MICB/B6 mice were bred with TRAMP mice to generate TRAMP/MICB mice which have been shown to recapitulate the NKG2D-mediated onco-immune dynamics of MIC⁺ cancer patients [27]. TRAMP/MICB male mice of 26 to 28 week-old were assigned into four cohorts with similar distribution of serum sMIC (Additional file 1: Figure S1) receiving therapy with intraperitoneal (i.p.) injection of sMIC-targeting nonblocking monoclonal antibody B10G5 [28] or anti-PD-L1 (clone 10F.9G2, BioXCell) antibody or isotype control IgG (cIgG) respectively at a respective dose of 4.0 mg/kg body weight twice weekly. For NK depletion, anti-NK1.1 (PK136) mAb (100 μg/mouse) were given 1 day before therapy and thereafter twice weekly as well together with therapeutic antibodies. All animals were treated for 8 weeks before euthanization which was designed for the study end point. Each study was repeated for three times unless otherwise specified.

Single cell suspension from spleen, draining lymph nodes (dLN), non-dLN or tumor were prepared as previous description [28]. Combination of the following antibody was used for cell surface or intracellular staining to define populations of NK, CD8, and subsets of CD4 T cells: CD3e (clone 145-2c11), CD8a (clone 53–6.7), CD4 (clone GK1.5), NK1.1 (clone PK136), NKG2D (clone CX5), CD44 (clone eBio4B10), CD11c (clone N418), MHCII (clone M5/114.15.2), CD80 (clone 16-10A1), CD86 (clone PO3) and CD40 (clone 1C10). For ex vivo re-stimulation, single cell suspension of freshly isolated splenocytes, LNs or TILs were cultured in complete RPMI 1640 medium containing 50 ng/mL PMA and 500 ng/mL Ionomycin for 6 h and analyzed by intracellular staining with antibodies specific to IFNγ (XMG1.2). All antibodies and the corresponding isotype controls were fluorochrome conjugated and were purchased from Biolegend, eBioscience or BD Biosciences. Multicolored Flow cytometry analyses were performed on an LSR II (BD). Data were analyzed with FlowJo X software (Tree Star).

Single suspended splenocytes from TCR-I transgenic mice were injected i.v. into animals (1x10⁶cells/mouse) that were received B10G5, anti-PD-L1 antibody, antibody cocktail, or control IgG therapy at 4.0 mg/kg body weight for each mouse. Animals were sacrificed at indicated time points to assess TCR-I T cell in vivo frequency with TCR-I-specific H-2D^b/TAg epitope I-tetramer (D^b/I-tetramer) [29]. To assay antigen-specific CD8⁺ T cell response, single cell suspension of splenocytes, tumor-draining lymph nodes (dLN) and tumor infiltrated lymphocytes (TILs) were stimulated overnight with 0.5 μM TAg epitope I peptide (SAINNYAQKL) and assayed by intracellular IFNγ staining of CD8⁺ or D^b/I-tetramer⁺ T cells.

For in vivo proliferation assays, splenocytes from TCR-I transgenic mice were suspended at 1 × 10⁷/ml in PBS/0.1% BSA and labeled with 5 μM CFSE (Biolegend, San Diego, CA, USA) for 10 min at 37 °C. Cells were then washed for three times in PBS, finally resuspended in PBS, and injected by i.v at a dose of 5 × 10⁶ cells per mouse. After 14 days, isolation of spleens, dLNs and TILs from recipient mice were harvested, and the intensity of CFSE staining was measured among CD8⁺ D^b/I-Tetramer⁺ T cells by flow cytometry.

Mouse blood was collected via tail bleeding before therapy or via cardiac puncture after euthanization. Serum was separated from blood by centrifugation. Splenocytes, draining lymph nodes (dLN), non-draining lymph nodes and partial of prostate tumors were directly meshed for isolation of TILs were collected for immunological analyses. Partial of prostate, lung, liver, kidney, pancreas, and colons were collected and fixed in 10% neutral fixation buffer followed by paraffin embedment for pathological and histological analyses.

Serum levels of sMICB from experimental mice were assessed using Duoset MICB Sandwich ELISA kit (Cat. DY1599) from R&D Systems according to manufacturer’s instruction. Serum was diluted 1:20 in PBS. Each assay was run in triplicates.

Human CD8 T cells were seeded in anti-CD3 (1 μg/ml, BD Biosciences) pre-coated 96-well plates and cultured with conditions where indicated with the following reagents: 1) 1 μg/ml soluble anti-CD28 antibody (Biolegend); 2) 100 ng/ml of soluble recombinant MICB (Sino Biologicas); 3) 100 ng/ml of B10G5. IFNγ production was assayed by intracellular staining after 24 h of culture (BD IFNγ staining Kits).

For assessing antigen-specific CD8 T cell response, human tyrosinase-specific HLA-A₂-restricted TIL13831 was co-cultured o/n with the HLA-A2⁺ T2-A2 cells (Generous gifts of Dr. Rubinstein at the Medical university of South Carolina) under indicated condition before functional assay. The tyrosinase peptide_369–377 was purchased from AnaSpec (Fremont, CA). After overnight culture, activation of TIL13831 was assessed by intracellular staining for IFNγ, TNFα, and CD107a (degranulation).

All results are expressed as the mean ± SEM. Mouse and sample group were n > 5, unless otherwise indicated. Data were analyzed using unpaired t-test, and treatment differences were considered significant at P values< 0.05. Kaplan-Meier survival curves were generated using GraphPad Prism software.

Tumor-derived sMIC suppresses anti-tumor immunity through impairing NK and CD8 T cell function and facilitating the expansion of MDSCs in tumor microenviroment [17, 20, 22, 27]. High levels of serum sMIC either at baseline or during therapy correlate with poor response to PD1/PD-L1 blockade therapy [26, 30]. We thus sought to investigate whether targeting sMIC can enhance tumor response to PD1/PD-L1 blockade therapy in pre-clinical models. With the knowledge that rodents do not express orthologs of human MIC and that human MICB serves as a functional ligand for mouse NKG2D, we generated a bi-transgenic TRAMP/MICB mouse that recapitulates the onco-immunological characteristics of human MIC⁺ cancer patients in that: i) MIC is specifically expressed in a given organ and concurrently expressed with oncogenic insults; ii) tumor releases sMIC during disease progression; iii) elevated levels of circulating sMIC correlate with more immune suppressive phenotype and more aggressive diseases [27].

To investigate whether targeting sMIC enables or enhances the response of MIC⁺ tumors to PD1/PD-L1 blockade therapy in TRAMP/MICB mice, we evaluated serum levels of sMIC in animals of ages between 27 to 29-weeks old when tumors are easily palpable in the abdominal. In accordance with our previous work, more than 50% of TRAMP/MICB mice would have invasive tumors with distant metastases in draining lymph nodes or lungs at 27 weeks of age [27]. These animals represented a subject population with heterogeneous diseases and varied levels of circulation serum sMIC [27, 31]. Given that higher levels of serum sMIC generally reflect more advanced diseases in TRAMP/MICB mice [27], we assigned these animals into four therapeutic groups, with the consideration that each group consists of animals with similar distribution of serum sMIC (Additional file 1: Figure S1). The four therapies include: an anti-PD-L1 antibody, a well-described sMIC-targeting monoclonal antibody (mAb) B10G5, an antibody cocktail composed of anti-PD-L1 mAb and B10G5, and control IgG (Fig. 1a). Consistent with our previous findings [28], all animals responded to the sMIC-targeting antibody B10G5, as demonstrated by the significantly decreased tumor weight (reflected as prostate organ weight) and reduced incidence of distant metastasis as compared to control IgG-treated animals (Fig. 1b and c). Animals that received anti-PD-L1 antibody therapy generally did not elicit a signficant beneficial response (Fig. 1b and c). Strikingly, animals subjected to combination therapy of the anti-PD-L1 mAb and B10G5 showed a significant decrease in tumor weight as compared to all other groups (Fig. 1b). Combination treatment also significantly decreased the incidence of lung metastasis and improved overall survival (Fig. 1c and d). These data demonstrate that antibody targeting sMIC resulted in the response of tumors, otherwise unresponsive, to PD1/PD-L1 blockade therapy.Fig1Fig. 1

Therapeutic efficacy of antibody cocktail composed of the sMIC-targeting antibody B10G5 and an anti-PD-L1 antibody in the autochthonous bi-transgenic TRAMP/MICB mice. a, Depiction of the therapy schema. Cohorts of 27 to 29-week-old TRAMP/MICB mice (male) were assigned to four therapy groups according to similar distribution of serum levels of sMIC for the four defined therapy. All therapies were given twice weekly for 8 weeks. b, Prostate organ weight, which reflects tumor weight in situ, of the animals at necropsy after receiving an 8-week duration of specific therapy. c, Incidence of lung metastasis in animals of each therapeutic group after a duration of 8-week therapy. Due to multiple focal micrometastasis in distant organs, quantitation of micrometastasis in each organ is not achievable. d, Kaplan-Meier survival at the designated study point (end of 8-week therapy). ns, not statistically significant

Combination therapy of anti-PD-L1 mAb and B10G5 targeting sMIC significantly augmented CD8 T cell mediated anti-tumor immunity. Although combination therapy did not significantly impact the population of CD8 T cells in peripheral lymphoid tissues, such as spleen, CD8 T cell population in the tumor-draining lymph nodes (dLN) and tumor was significantly increased (Fig. 2a, c). Consistent with our previous findings [28], B10G5 also augmented antigen-specific CD8 T cell effector functions in peripheral lymphoid tissue and tumor site (Fig. 2b, d). Monotherapy with anti-PD-L1 mAb did not significantly enhance antigen-specific CD8 T cell function as measured by ex vivo response to stimulation by TRAMP-specific SV40TAg peptide. However, combination therapy of the anti-PD-L1 mAb and B10G5 remarkably enhanced the antigen-specific responsiveness of CD8 T cells as compared to B10G5 single agent therapy (Fig. 2b, d). The augmented therapy effect of combined reagents was further demonstrated by a significant increase in the CD44^hi CD8 T cell compartments in the spleen, dLN, and tumors and the significant increase in their cell intrinsic ability to produce IFNγ (Fig. 2e-h). These results suggest a potential synergistic effect of B10G5 targeting sMIC and the anti-PD-L1 mAb therapy.Fig2Fig. 2

Combination therapy of sMIC-targeting mAb B10G5 and anti-PD-L1 antibody cooperatively enhances anti-tumor potential of CD8⁺ T cells. a and c, Representative dot plots and summary data from all animals present that cocktail therapy significantly enriched CD8 T cell in draining LN (dLN) and tumor infiltrates. b and d, Representative histograms (c) and summary data (d) of CD8 T cell IFNγ production in response to SV40TAg re-stimulation. e and g, Representative histograms (e) and summary data (f) of CD44^Hi CD8 T cell population. f and h. Representative histograms (g) and summary data (h) of IFNγ production by CD44^Hi CD8 T cells in response to PMA/ionomycin stimulation. Spln, spleen. dLN, tumor-draining lymph node, TIL, tumor infiltrated lymphocytes. ns, not significant

To investigate whether anti-PD-L1 mAb and B10G5 combination therapy enhances CD8 T cell anti-tumor response in an antigen-specific manner, we utilized the availability of SV40TAg-specific T cell receptor (TCR) transgenic (TCR-I) mice [29]. Of note, tumors in TRAMP/MICB mice were driven by SV40TAg through interrupting p53 and Rb signaling [33]. CD8 T cells from the TCR-I mice (thereafter TCR-I CD8 T cell) bear SV40TAg-specific TCR and can be detected by SV40TAg peptide I-specific D^b/I-tetramer. We adoptively transferred CFSE-labeled TCR-I CD8 T cell into mice that have received four-week duration of treatment as depicted in Fig. 3a and analyzed the sustainability of these TCR-I CD8 T cells on day 14 post-adoptive transfer. Normally, adoptively transferred antigen-specific TCR-I CD8 T cells fail to be sustained after their initial expansion in TRAMP or TRAMP/MICB mice due to clonal deletion [28, 34]. While monotherapy with anti-PD-L1 mAb presented a marginal effect on sustaining D^b/I-tetramer⁺ TCR-I CD8 T cells in the dLN, tumor, or spleen, B10G5 therapy consistently sustained the adoptively transferred TCR-I CD8 T cells with high frequency in the tumor as we have previously shown [28] (Fig. 3b, c). Remarkably, combination therapy of anti-PD-L1 and B10G5 further significantly enhanced the sustainability of TCR-I CD8 T cells in comparison to B10G5 monotherapy (Fig. 3b, c). The subsequent CFSE dilution assay confirmed that only therapy with mAb B10G5 or combination therapy provoked an expansion of SV40TAg-specific TCR-I CD8 T cells represented by an elevation in the percentage of CFSE^lo D^b/I-tetramer⁺ CD8 T cells. However, combination therapy remarkably enhanced the expansion as compared to B10G5 monotherapy (Fig. 3d, e). Expansion of antigen-specific CD8 T cells in the tumor and draining lymph nodes is considered the hallmark for activation of antigen-specific CD8 T cells. In support of this concept, combination therapy resulted in significantly enhanced IFNγ production by D^b/I-tetramer⁺ CD8 T cells in response to TAg-peptide re-stimulation (Fig. 3f, g). Together, these data demonstrate an important effector mechanism for cooperative therapeutic effect of anti-PD-L1 mAb and B10G5.Fig3Fig. 3

Anti-PD-L1 antibody in combination with sMIC-targeting mAb B10G5 cooperatively enhances antigen-specific CD8 T cell anti-tumor responses. a, Depiction of experiment schema. CFSE-labeled tumor antigen SV40TAg-specific TCR-I CD8 T cells were transferred into TRAMP/MICB mice that have received 4-week duration of therapy, which was continued post transfer of TCR-I CD8 T cells. Data shown are 14-day post-transfer of CFSE-labeled TCR-I CD8 T cells. b and c, Representative dot plots (b) and summary data (c) demonstrating the percentage of D^b/I-tetramer⁺ SV40TAg-specific CD8 T cells in dLN, tumor infiltrates, and spleen. d and e, Representative histograms (d) and summary data (e) demonstrating proliferation of SV40Tag-specific CD8 T cells represented by CFSE^lo population. f and g, Representative histograms (f) and summary data (g) demonstrating the response of D^b/I-tetramer⁺ SV40TAg-specific CD8 T cells to ex vivo SV40TAg peptide re-stimulation as measured by IFNγ production

We have previously shown that single agent B10G5 enhances DC activation in tumor draining lymph nodes and increases the expression of DC co-stimulatory molecule CD80 and CD86 [28]. Combination therapy of anti-PD-L1 mAb and B10G5 further significantly increased DC surface expression of the co-stimulatory molecule CD80 and CD86 as well as the DC activation molecule CD40 (Fig. 4a and b). Given that CD80 and CD86 engagement with CD28 on CD8 T cells amplifies TCR/CD3 signaling, these data indicate that inhibition of the PD1/PD-L1 pathway with co-targeting sMIC could potentially facilitate a more vigorous CD28-mediated co-stimulatory signal delivered to antigen-specific CD8 T cells for the sustained anti-tumor activity.Fig4Fig. 4

Combination therapy of anti-PD-L1 antibody and sMIC-targeting mAb B10G5 cooperatively increase DC activation (CD40) and the expression of co-stimulatory molecules CD80 and CD86 in tumor sites. a, Representative histograms from flow cytometry analyses of CD40, CD80 and CD86 expression on DCs from tumor-draining lymph nodes and tumor beds. Gray filled profiles, control isotype staining. Open dark profiles, antibody to specific DC surface molecules. b, Summary data of the increase in mean fluorescence intensity (MFI) of CD80, CD86, and CD40 on DCs

NKG2D is a co-stimulatory molecule on CD8 T cells that functions non-redundantly of CD28 [12, 35, 36]. Tumor-derived sMIC was shown to downregulate NKG2D expression in cancer patients and subvert NKG2D-co-stimulation to CD8 T cells [20]. NKG2D is constitutively expressed by all human CD8 T cells; however, it is expressed only by activated murine CD8 T cells [8]. Consistent with our previous report [28], targeting sMIC with B10G5 increased the frequency of NKG2D⁺ CD8 T cells in the dLN and in tumors (Fig. 5a and b). Although anti-PD-L1 mAb single agent only presented a marginal effect on NKG2D expression on CD8 T cells, combination therapy of anti-PD-L1 mAb and B10G5 resulted in an increase in the frequency of NKG2D⁺ CD8 T cells in draining LN and tumor infiltrates as compared to B10G5 single agent (Fig. 5a and b). These data suggest potential synergistic effects of B10G5 targeting MIC and anti-PD-L1 mAb in restoring and sustaining NKG2D expression on activated CD8 T cells.Fig5Fig. 5

Targeting sMIC increases co-stimulatory molecules on CD8 T cells in tumor draining LN and tumor infiltrates. a and b, Representative histograms (a) and summary data (b) demonstrating NKG2D expression on CD8 T cells. c and d, Representative histogram overlay (c) and summary data of mean fluorescence intensity (MFI) (d) demonstrating CD28 expression on CD8 T cells. Grey profiles in (c) are CD28 expression in CD8 T cells from cIgG treated animals. Black profiles in (c) are CD28 expression in CD8 T cells from animals receiving respective therapy

Activation of NKG2D and CD28 can provide non-redundant co-stimulation to CD8 T cells [12, 35]. We thus sought to understand the significance of increased, or at least sustained, NKG2D and CD28 expression in TCR-mediated CD8 T cell function. We stimulated the SV40TAg-specific TCR-I CD8T cells with various conditions as indicated in Fig. 6. Addition of anti-CD28 agonist antibody resulted in modest CD3/TCR-mediated activation and proliferation as indicated by IFNγ production (Fig. 6a). Addition of recombinant sMIC plus the anti-sMIC antibody notably increased the magnitude of CD3/TCR-mediated activation. Intriguingly, anti-CD28 together with sMIC plus B10G5 remarkably amplified CD3/TCR-mediated activation (Fig. 6a and c), suggested a potential synergistic co-stimulation of TCR-dependent effector function of CD8 T cells. The enhanced CD3/TCR signaling by an CD28 agonist together with sMIC plus B10G5 co-stimulation was further substantiated with independent experiments in which the expression of the critical CD3/TCR signaling molecule CD3ζ was upregulated and sustained with prolonged dual co-stimulation (Fig. 6b and d).Fig6Fig. 6

sMIC-targeting antibody and a CD28 agonist antibody provides CD8 TCR with dual co-stimulation and thus optimal and sustained activation. a, SV40TAg-specific TCR-I CD8 T cells were stimulated with the indicated conditions and evaluated for intracellular IFNγ production after 24 h of stimulation. b, Cells were stimulated with indicated conditions for three and 5 days. Expression of the TCR/CD3 signaling molecule CD3ζ was assessed by flow cytometry analyses with intracellular staining. c and d, Representative summary data of (a) and (b) respectively. Data represent results of triplicates of four independent experiments

As we have reported previously [28], targeting sMIC with B10G5 increased NK cell number in the peripheral and tumor infiltrates and augmented NK cell intrinsic function as measured by the ability to produce IFNγ in response to PMA and Ionomycin (PMA/I) stimulation (Fig. 7a–d). Anti-PD-L1 mAb monotherapy significantly increased NK cell numbers or intrinsic cellular function in tumors but not in the spleen (Fig. 7a–d). Combination therapy with B10G5 and the anti-PD-L1 mAb further significantly increased NK cell number and cellular intrinsic function only in tumors as compared to monotherapy of B10G5 or the anti-PD-L1 antibody (Fig. 7a–d). Notably, combination therapy significantly increased NKG2D expression on NK cells as compared to B10G5 single agent therapy, in spite that anti-PD-L1 mAb therapy alone did not impact NKG2D expression on NK cells (Additional file 1: Figure S7). The enhanced NKG2D expression may account in part for the functional enhancement of NK cells in response to the combination therapy. These observations suggest that NK cell may play a role in the synergistic effect of the anti-PD-L1 antibody and B10G5 targeting sMIC at tumor site.Fig7Fig. 7

Antibody cocktail therapy of sMIC-targeting mAb B10G5 and anti-PD-L1 antibody cooperatively enriched NK cell infiltration and enhanced NK cell function in tumors. a and b, Representative dot plots (a) and summary data (b) from all animals present that antibody cocktail therapy significantly enriched NK cell in tumor infiltrates compared with monotherapy. b and d, Representative histograms (b) and summary data (d) from all animals present that antibody cocktail therapy significantly enhanced NK cell responsiveness in tumors compared with monotherapy. Note that anti-PD-L1 antibody monotherapy presented no significant effect on NK cells. TIL, tumor infiltrated lymphocytes. ns, not significant

We have previously shown that depletion of NK cells compromises the therapeutic effect of B10G5 [28]. We thus investigated the impact of NK cell on the therapeutic efficacy of the combination therapy. Depletion NK cells significant impair the therapeutic outcome of the combination therapy of B10G5 and anti-PD-L1 in TRAMP/MICB mice as evaluated by prostate weight at necropsy (Fig. 8a).Fig8Fig. 8

Anti-sMIC therapy with B10G5 up-regulates PD-L1 expression on tumor cells, which is in part NK cell dependent. a, Depletion of NK cells (dpNK) during therapy diminishes the therapeutic effect of co-targeting sMIC and PD-L1 as assessed by prostate weight at necropsy. All treatments were given twice weekly i.p. for 8 weeks. b, Representative histograms from flow cytometry analyses demonstrating the percentage of PD-L1⁺ tumor cells from TRAMP/MICB mice. c, Summary data of the percentage of PD-L1⁺ tumor cells from (a)