This study examines peripheral blood and plasma samples prospectively collected as part of two clinical trial protocols at City of Hope National Medical Center, in which treatment-naive patients with metastatic renal cell carcinoma treated with nivolumab and ipilimumab were enrolled. Blood samples were collected at baseline and week 12. Response assessment was performed in 12-week intervals by the principal investigator per RECIST 1.1 criteria. Patients with complete or partial response were considered responders, while those with the best response of stable disease or progressive disease were considered non-responders. Both protocols were approved by the City of Hope Institutional Review Board. All study procedures were undertaken in accordance with the Declaration of Helsinki. Demographic and patient responses are shown in Table 1 .

Patients who received a first-line combination of nivolumab and ipilimumab in two separate clinical trials at City of Hope were identified, and peripheral blood samples were collected at baseline or at week 12 from 37 RCC patients representing responders (n = 16) or non-responders (n = 21). We elected to use samples obtained at baseline before treatment initiation and at week 12 ( ± 4 weeks) as typical times used in assessing the initial patients’ immune response. A total of 70 samples obtained from all patients were included in the final analysis. To assess cytokine/chemokine concentration, plasma was separated from peripheral blood mononuclear cells (PBMCs) by centrifugation and stored at −80°C until analysis using a panel of 30 cytokines including IL-1RA, IL-1b, IL-2, IL-2R, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12, IL-13, IL-15, IL-17, eotaxin, EGF, FGF, G-CSK, IFN-α, IFN-γ, CXCL9, CXCL10, CCL2, CCL3, CCL4, CCL5, TNF-α, and VEGF on the Luminex Flexmap 3D system (Biotechne, Minneapolis, MN, USA).

PBMCs were thawed at 37°C for 5 min, washed in 10% FBS/RPMI 1640, resuspended in the matched patient’s plasma from the same draw (20% plasma/RPMI 1640), and incubated for a minimum of 2 h at 37°C. Following incubation, cells were washed and incubated with a combination of 1 µM of sodium orthovanadate (Sigma-Aldrich, St. Louis, MO, USA) and DNase I (Roche, Basel, Switzerland), before staining with a viability dye and fluorescent antibodies for surface immune markers. Extracellular staining was performed using fluorochrome-labeled antibodies (from BD, Franklin Lakes, NJ unless stated otherwise) to CD3 (#612940), CD19 (#565697), HLA-DR (#565073), CD14 (#563561), CD8 (#612889), PD-1 (#329920), CD4 (# 2500492), CD69 (#562989), CD33 (Thermo-Fisher Scientific, Irwindale, CA, USA, #47033841), PD-L1 (#563742), CD15 (#747426), and CD56 (#565139). For intracellular staining, cells were fixed before permeabilization (Thermo-Fisher Scientific, #00-5523-00) and immunostained for arginase-1 (R&D Systems, Minneapolis, MN, USA, #IC5868F), phosphotyrosine (Y705) STAT3 (#557815), and FoxP3 (#560852).

T-cell proliferation studies were performed as described before with minor modifications (24, 26). Briefly, whole blood from patients was cultured for 2 h at 37°C. Next, samples were washed and centrifuged at 250×g for 5 min. Cell pellets were then resuspended in red blood cell lysis buffer for 10 min/4°C, treated using DNase for 5 min/37°C, then filtered through a 70-µm nylon mesh filter and washed. CD15⁺ cells were enriched using a positive selection kit from Stemcell Technologies, Vancouver, Canada (#18651) following the manufacturer’s protocol, then cultured in the presence of 20% matched patient’s plasma and 500 nM of CpG-STAT3ASO or control CpG-scrON or with PBS. The following day, CD3⁺ T cells were then enriched using a T-cell enrichment kit (Stemcell Technologies, #17951) from healthy donor PBMCs obtained using density centrifugation over Histopaque-1077 at 1,500 rpm/20 min. T cells were then labeled with CFSE dye according to the manufacturer’s protocols (Thermo-Fisher Scientific, #C34554) and co-cultured with CD15⁺ cells and CD3/CD28 beads in round-bottom 96-well plates (Thermo Fisher, #11131D) at a ratio of 1:6 of T cells to CD15⁺ myeloid cells. After 3 days, flow cytometric analysis was performed to assess T-cell proliferation using CFSE dilution using antibodies specific to CD8a (RPA-T8, #25-0088-42).

The CpG oligonucleotide conjugates were synthesized in the DNA/RNA Synthesis Core (COH) as previously described (24). The resulting oligonucleotide (ON) conjugates are shown below (x = a single C3 unit; underline = 2′O-methyl-modification; asterisk = phosphorothioation site):

CpG(D19)-human STAT3ASO:

5′ G*G*TGCATCGATGCAG*G*G*G*G*G-xxxxx-C*A*G*C*A*G*A*T*C*A*A*G*T*C*C*A*G*G*G*A 3′.

STAT3 ASO (human STAT3 ASO targeting sequence):

5′ C*A*G*C*A*G*A*T*C*A*A*G*T*C*C*A*G*G*G*A 3′.

CpG(D19)-scrambled oligonucleotide (scrON):

5′ G*G*TGCATCGATGCAG*G*G*G*G*G-xxxxx-A*G*A*G*C*C*T*A*A*C*G*G*A*A*G*G*C*A*C*T 3′.

CpG (1668)-mouse STAT3ASO:

5′ T*C*C*A*T*G*A*C*G*T*T*C*C*T*G*A*T*G*C*T-xxxxx-G*A*C*T*C*T*T*G*C*A*G*G*A*A*T*C*G*G*C*T 3′.

CpG (1668)-scrambled oligonucleotide (scrON):

5′ T*C*C*A*T*G*A*C*G*T*T*C*C*T*G*A*T*G*C*T-xxxxx-A*G*A*G*C*C*T*A*A*C*G*G*A*A*G*G*C*A*C*T* 3′.

Mouse kidney (Renca) and bladder (MB49) cancer cells were purchased from the American Type Culture Collection (ATCC). Renca and MB49 cells were cultured in RPMI 1640 or DMEM media, respectively, supplemented with 10% FBS, 1% penicillin/streptomycin, and 1% GlutaMAX (Thermo-Fisher). All cell lines were cultured for less than 3 months prior to the experiments and were tested to be free of Mycoplasma infection.

Balb/C and C57BL/6 mice, aged between 6 and 8 weeks, were purchased from the Jackson Laboratory. Mouse care and experimental conditions were performed under pathogen-free conditions and in accordance with established institutional guidance and approved protocols from the Institutional Animal Care and Use Committees. For efficacy studies with the subcutaneously (SC) implanted Renca tumors, 5 × 10⁵ Renca cells were resuspended in a 1:1 ratio with Matrigel (Corning, Corning, NY, USA, #356231) and 1× PBS and injected SC into female Balb/C mice. Mice were then treated intraperitoneally (IP) with the anti-PD-1 antibody on day 3 and day 5 (200 µg, BioXCell, Lebanon, NH, USA, #BE0273) before treatment every other day with CpG-STAT3ASO (5 mg/kg) in combination with anti-PD-1, or alternatively, mice were treated with CpG-STAT3ASO (5 mg/kg) alone or IgG or PBS control, all injected intravenously (IV) via retro-orbital venous sinus injection. Tumor size was monitored and measured every other day using calipers. For the SC MB49 mouse model, 5 × 10⁵ MB49 cells were resuspended in 1× PBS and injected SC into male C57BL/6 mice. When tumors reached approximately 100 mm³, mice were treated twice using IP injections of anti-PD-1 antibody (200 µg) (BioXCell, Cat #BE0273) 2 days apart, before treatment 2 days later with CpG-STAT3ASO (5 mg/kg) in combination with anti-PD-1, or alternatively, mice were treated with CpG-STAT3ASO (5 mg/kg) alone or IgG or PBS control injected IV. In the animal study using subcutaneous administration of oligonucleotides, mice were first treated IP with anti-PD-1 antibody on day 3 and day 5 (200 µg, BioXCell, #BE0273) before treatment on days 7, 9, 11, 13, and 15 with or without SC-injected CpG-STAT3ASO (10 mg/kg) on days 7, 9, 11, 13, and 15.

For the analysis of tumor-associated immune cells, single-cell suspensions were prepared from whole tumors or tumor-draining lymph nodes using short collagenase-IV/DNase-I treatment (20 min/37°C) followed by mechanical dispersion by pipetting through a 70-µm mesh filter. Viable cells were enriched using density centrifugation over Histopaque-1083 at 1,500 rpm/20 min. Extracellular staining was performed with fluorochrome-labeled antibodies (from BD unless stated otherwise) to CD45 (#564279), CD86 (#741285), CD11c (#612797), CD80 (#562611), CD11b (Thermo Fisher, #48-0112-82), CD8 (#563063), CD206 (#141721), Gr1 (BioLegend, San Diego, CA, USA, #50-604878), MHC II (Thermo Fisher, #11-5321-82), CD3 (#560527), CD4 (R&D Systems, #FAB554S-100), and FoxP3 (#566881). Fluorescence data were analyzed on BD Fortessa LSR II and the Cytek Aurora Spectral Cytometer and using FlowJo v10 software (TreeStar, Ashland, OR, USA) or Cytobank (Beckman-Coulter, Brea, CA).

Peripheral blood samples were obtained from tumor-bearing mice using tail vein bleed on day 14. Blood was centrifuged at 1,200 rpm/5 min to separate plasma which was then stored at −80°C until use. IFNγ and IL-6 concentrations in plasma were assessed using ELISA assays (Thermo Fisher, #KMC4021 or #KMC0061, respectively) according to the manufacturer’s instructions.

For the patients’ sample analysis and animal studies, comparisons of groups were performed using the Wilcoxon signed-rank test, one-way ANOVA, or two-way ANOVA with Bonferroni’s multiple comparisons test. Data were presented as mean ± SEM. Statistical significance was ranked and indicated as follows: *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

To assess the effects of combination immune checkpoint therapy (ICB) on plasma immune mediators, peripheral blood was obtained from all patients at baseline and at week 12 when the first indications of response to clinical immunotherapies often occur. We assessed immune alterations in 30 plasma cytokines and chemokines in samples from a total of 37 patients ( Figure 1 and Supplementary Figure S1 ). As shown in Figure 1 , the refractory patients/non-responders showed a significant increase in immunoregulatory cytokines such as IL-6, IL-8, IL-10, and IL-1RA. In contrast, only IL-10 was weakly upregulated in responding patients. IL-6, IL-8, and IL-10 are important STAT3 activators and are often associated with the recruitment or expansion of tolerogenic MDSCs (14, 28, 29). Protein markers of ongoing immune response such as IL-12, soluble IL-2R, and CXCL9 and CXCL10 were significantly elevated after combined ICB therapy in both groups of patients. However, the levels of IFN-inducible CXCL10 were significantly higher in responding than in non-responding patients after 12 weeks of ICB therapy, which is likely indicative of IFN-driven immune responses critical for long-term antitumor effects. We also observed a significant increase in proinflammatory, innate immune regulator, CCL11/eotaxin, in some of the responding patients.

STAT3 is commonly activated in the microenvironment of human tumors including RCC (30). Given the elevated plasma levels of several STAT3 activators, we assessed STAT3 phosphorylation across circulating immune cell populations using high-parameter spectral flow cytometry. As shown in Figure 2 , t-distributed stochastic neighbor embedding (t-SNE), a non-linear dimensionality reduction algorithm, indicated STAT3 activation (pSTAT3) and the expression of STAT3 downstream target and immune checkpoint molecule PD-L1 in overlapping myeloid cell clusters, likely polymorphonuclear MDSCs (PMN-MDSC: CD15⁺CD33⁺HLA-DR⁻) and monocytic MDSCs (M-MDSCs: CD14⁺CD33⁺HLA-DR⁻) at the treatment initiation as well as after 12 weeks of ICB therapy. We also detected modest STAT3 within a cluster of circulating B cells; however, the percentages of B cells did not significantly change during the ICB therapy and did not show a positive correlation with STAT3-inducing cytokines ( Supplementary Figure S2 ). The analysis of both MDSC subsets confirmed the upregulation of pSTAT3 at baseline in both responders and non-responders ( Figures 3A, B , gating strategy). While there was no significant change in the total percentage of M-MDSCs before and after ICB therapy, a subset of patients responding to therapy showed an increase in the percentage of PMN-MDSCs ( Figure 3C ). Importantly, responding but not the refractory patients showed a significant reduction in the activity of STAT3 in M-MDSCs between baseline and week 12. However, we did observe a decrease in pSTAT3 in M-MDSCs in patients who responded to therapy between baseline and W12 ( Figure 3D ). Correspondingly, we observed a strong correlation between the level of STAT3 activity in M-MDSCs and plasma concentrations of IL-6 ( Figure 3E ; r = 0.7, p = 0.0113) as well as IL-8 ( Figure 3F ; r = 0.6, p = 0.0477) in non-responding patients. We did not find such a correlation between pSTAT3 and plasma cytokine levels in patients responding to therapy or in PMN-MDSCs ( Figures 3E, F ). Our observations suggest that nivolumab/ipilimumab immunotherapy has a partial effect on the tolerogenic myeloid cells accumulating in RCC patients, especially on the subset of M-MDSCs. However, the presence of both MDSC populations after 12 weeks of therapy in both non-responding and responding patients suggests that STAT3-driven immunosuppression may limit the clinical efficacy of combined immune checkpoint inhibition. In fact, we have not observed an increase in the percentages of CD8⁺ T-cell subsets in ICB-treated RCC patients. Instead, there was a modest but significant increase in the percentage of regulatory T cells (Tregs) specifically in non-responders but not in responding patients ( Supplementary Figure S3 ).

The persistent presence of tolerogenic MDSCs in the blood of nivolumab/ipilimumab-treated RCC patients suggested a potential mechanism of STAT3-driven therapeutic resistance. Targeting STAT3 signaling in RCC-derived MDSCs could alleviate their immunosuppressive effects and restore T-cell proliferation and activity. To test this hypothesis, we used a myeloid cell-selective CpG-STAT3ASO strategy. CpG-STAT3ASO is selectively internalized by TLR9⁺ myeloid cells, including human MDSCs but not by T lymphocytes (24). Primary CD15⁺ myeloid cells, including PMN-MDSCs, were enriched from treatment-refractory RCC patients’ PBMCs and incubated with CpG-STAT3ASO to knockdown STAT3 or with non-targeting but immunostimulatory CpG-scrON oligonucleotide ( Figures 4A, B ). Next, we co-cultured the treated MDSCs with healthy donor T cells for 3 days and assessed T-cell proliferation using the flow cytometric assay. As shown in Figure 4C , CpG-STAT3ASO but not control CpG-scrON almost completely abrogated the inhibitory effect of RCC-derived MDSCs on T-cell proliferation. The control CpG-scrON, lacking STAT3 inhibitory activity but activating TLR9, had only limited and non-significant stimulatory effect on T-cell proliferation (p = 0.008). These results imply that CpG/TLR9 stimulation alone without STAT3 inhibition is not sufficient for disrupting the tolerogenic effects of RCC-associated myeloid cells.

Next, we set out to assess whether STAT3 inhibition/TLR9 stimulation using CpG-STAT3ASO will prove effective alone or together with immune checkpoint blockade against genitourinary cancers, such as Renca and MB49 tumors, two commonly used models of mouse kidney and bladder cancers ( Figure 5A ). We first tested the effect of systemic administration of CpG-STAT3ASO with or without prior PD-1 blockade on the growth of Renca tumors. As shown in Figure 5B , repeated IV injections of CpG-STAT3ASO alone significantly reduced kidney tumor progression in comparison to all control groups. Our control study ( Supplementary Figure S4A ) and the published results in prostate tumor models (24) confirmed that concomitant TLR9 activation and STAT3 inhibition are necessary for the generation of effective antitumor effects. In contrast, Renca tumors were completely resistant to PD-1 blockade alone, even after four more intensive treatments ( Supplementary Figure S4B ), while the combination of anti-PD-1 with CpG-STAT3ASO showed modest improvement in the antitumor effect ( Figure 5B ). We also confirmed that the subcutaneous route of CpG-STAT3ASO delivery, at a site distant from the tumor, also resulted in an antitumor effect, suggesting an alternative systemic route of administration for this oligonucleotide ( Supplementary Figure S4C ).

We next assessed the efficacy of CpG-STAT3ASO and anti-PD-1 as single treatments and the combination in the MB49 bladder cancer model ( Figure 5C ). Compared with the Renca model, CpG-STAT3ASO resulted in a more pronounced antitumor effect against MB49 tumors with significant growth inhibition and occasional complete tumor eradication. PD-1 blockade alone demonstrated significant although less consistent oligonucleotide treatment antitumor efficacy against MB49 tumors. Finally, the combination of CpG-STAT3ASO and anti-PD-1 resulted in augmented antitumor efficacy leading to complete tumor regression in the majority of mice. These results suggested potential differences in the cellular mechanisms of antitumor effects likely related to the different compositions of the tumor microenvironment in both kidney and bladder cancer models.

To assess cellular mechanisms of the observed antitumor effects against Renca and MB49 tumors, we characterized the immunophenotype of the major tumor-associated myeloid cell populations, such as macrophages and DCs. As shown in Figure 6A (with gating strategy in Supplementary Figure S6 ), both CpG-STAT3ASO and CpG-STAT3ASO/anti-PD-1 combination treatment dramatically increased the percentage of M1-like macrophages in Renca tumors, four- to five-fold higher than in anti-PD-1 or control treatment groups. In addition, both CpG-STAT3ASO and CpG-STAT3ASO/anti-PD-1 treatments seemed to reduce the percentage of tumor-associated M2-like macrophages compared with PD-1 blockade alone, although this effect did not reach significance in relation to negative controls ( Figure 6A and Supplementary Figure S7 ). This is consistent with the direct effect of PD immune blockade on T cells and not myeloid cells. Our previous studies using genetic STAT3 deletion in tumor-infiltrating myeloid cells or using CpG-STAT3ASO in prostate cancer models suggested that TLR9 activation/STAT3 inhibition results in the recruitment of CD11b⁺Gr1⁺ myeloid cells representing not MDSCs but neutrophils, which can contribute to antitumor effects (24, 31). In fact, CpG-STAT3ASO alone and, to a lesser extent, the combination treatment elevated the percentage of CD11b⁺Gr1⁺ cells in Renca tumors ( Figure 6B ). Corresponding to these potential immunostimulatory effects at the tumor site, we observed strong recruitment of activated and antigen-presenting (MHC-II^HI/CD86⁺ or MHC-II^HI/CD80⁺) M1-like macrophages into Renca tumor-draining lymph nodes by CpG-STAT3ASO and the combination treatment resulting in three- to four-fold increase of the activated macrophages compared with PBS- or antibody-treated controls ( Figure 6C ). We also assessed the recruitment of activated DCs given their important role in antigen presentation and T-cell-mediated antitumor immunity. Interestingly, unlike in the case of macrophages, the combination CpG-STAT3ASO/anti-PD-1 treatment but not CpG-STAT3ASO or anti-PD-1 alone led to a significant recruitment of activated (MHC-II^HI and CD86⁺ or CD80⁺) DCs into tumor-draining lymph nodes ( Figure 6D and Supplementary Figure S5 ). This effect suggested a potentially indirect role of PD-1 blockade in DC activation through IFNγ-dependent mechanism as recently suggested by others (32). In fact, the plasma levels of IFNγ were significantly higher in Renca tumor-bearing mice after the combination CpG-STAT3ASO/anti-PD-1 treatment compared with CpG-STAT3ASO or PD-1 blockade alone with the latter showing only a minor effect. In contrast, the plasma levels of IL-6 were decreased by both CpG-STAT3ASO and the combination treatment but not by PD-1 inhibition alone ( Figure 6E ) and correlated with the previously observed decrease of tolerogenic M2-like macrophages, which are the potential sources of this cytokine in the TME ( Figure 6A ).

As typical for human bladder cancers, the mouse MB49 tumor microenvironment is dominated by Th2 cytokines, such as IL-10, which leads to the accumulation of tolerogenic macrophages as well as regulatory T cells (31, 33). Thus, we expected potential differences in the antitumor effect of the tested CpG-STAT3ASO and PD-1 inhibition between Renca and MB49 models. Nonetheless, similar to kidney tumors, we did observe a significant reduction of CD206^HI M2-like macrophages by CpG-STAT3ASO/anti-PD-1 combination in comparison to PBS- and IgG-treated groups, with a less robust inhibitory effect of CpG-STAT3ASO alone ( Figure 7A ). In contrast to the Renca model, there was only a minimal increase in the number of CD11b⁺CD11c^–MHCII⁺ M1-like macrophages ( Figure 7A ) or CD11b⁺Gr1⁺ cells ( Figure 7B ). However, the combined CpG-STAT3ASO/anti-PD-1 treatment and to a lesser extent CpG-STAT3ASO alone strongly increased the recruitment of activated, antigen-presenting MHC-II^HI/CD86⁺ macrophages ( Figure 7C ) and DCs ( Figure 7D and Supplementary Figure S5 ) into tumor-draining lymph nodes. Overall, the results of our immunophenotypic analysis of Renca and MB49 tumor microenvironments correspond well with the differences in antitumor efficacy of various treatments and confirm that CpG-STAT3ASO/anti-PD-1 combination immunotherapy in both tumor types results in more consistent and robust immune stimulation of the tumor-associated myeloid cells.

Given the different myeloid cell activation by CpG-STAT3ASO and/or anti-PD-1 treatments, we next assessed the effect on T-cell populations in both tumors. As shown in Figures 8A, B , we found a small population of tumor-infiltrating CD8 (~10% of total T cells on average) together with CD4 helper and regulatory CD4⁺/FoxP3⁺ T cells (<50% of total CD4⁺ T cells) in Renca tumors. The combined CpG-STAT3ASO/anti-PD-1 treatment significantly increased the percentage of tumor-infiltrating CD8 compared with PBS and IgG controls ( Figure 8A ). It also showed a tendency to reduce Treg numbers from an average of 50% to 25% although without reaching statistical significance (p > 0.05). Thus, CpG-STAT3ASO/anti-PD-1 augmented the CD8/Treg ratio indicative of successful adaptive, antitumor immunity. While CpG-STAT3ASO showed a similar tendency to recruit CD8 T cells to Renca tumors, its overall effect was less robust and not significant given the limited number of tested animals. In contrast to kidney tumors, MB49 bladder cancers were infiltrated by a significant number of tumor-resident CD8 T cells (>20% of total CD3⁺ T cells on average), and regulatory T cells were dominating the CD4 T-cell population (>50% of CD4 T cells) ( Figures 8C, D ). Both the combined CpG-STAT3ASO/anti-PD-1 treatment and CpG-STAT3ASO alone dramatically reduced the percentage of Tregs in MB49 tumors to 15% from an average of 56% in PBS controls. Even though neither of the treatments affected the overall numbers of tumor-resident CD8 T cells, the reduction in the number of Tregs benefitted the ratio of CD8 to Treg cells, thus enabling the generation of antitumor immune responses. The effect of myeloid cell-selective CpG-STAT3ASO on Tregs was likely indirect through disruption of the immunosuppressive myeloid tumor microenvironment that is required for sustaining Treg expansion and activity. Finally, unlike in the Renca tumor model, PD-1 inhibition alone had modestly but significantly reduced the percentage of Tregs improving the CD8 to Treg ratio. This unexpected anti-PD-1 activity may explain its antitumor activity in some of the mice bearing bladder tumors and lack of any effect against the renal cancer model.