Introduction

Interleukin (IL)-2 was the first Food and Drug Administration approved immunotherapy for use in patients with metastatic melanoma and renal cell carcinoma.1 2 However, the use of recombinant IL-2 (Proleukin/aldesleukin) has been problematic due to low response rates, off-target effects, and severe dose-dependent toxicities. Due to its potent T-cell growth factor activity and ability to promote development and activity of cytotoxic T lymphocytes and natural killer (NK) cells, IL-2 has been re-engineered for cancer immunotherapy to increase its half-life, decrease its toxicity, and to drive selectivity toward memory-phenotypic (MP) CD8+ T and NK cells.3 4 Since these cells express high amounts of the intermediate-affinity IL-2R, consisting of CD122 (IL-2Rβ) and CD132 (γc), one major strategy has been to derive IL-2 biologics with reduced or absent binding to CD25 (IL-2Rα), which is essential to form the high-affinity IL-2R. The rationale behind these CD122/CD132-selective IL-2s is to promote antitumor immunity while reducing non-specific toxicities, such as vascular leak syndrome,5 and by minimizing undesired immunosuppressive regulatory T cells (Tregs).

High-affinity IL-2R-selective compounds have also been formulated primarily for use at low doses to boost Tregs in the context of autoimmunity and inflammatory disorders.6–8 This class of IL-2 analogs has not been extensively explored for anticancer effects due to the concern of increased off-target Tregs. In the context of cancer immunotherapy, these high-affinity IL-2R biologics are expected to also react with recently activated tumor antigen-specific CD4+ and CD8+ T effector (Teff) cells that have upregulated CD25 to express the high-affinity IL-2R. Whether such an effect on tumor-responsive Teff cells is diminished by an accompanying polyclonal increase in Tregs remains to be determined. In contrast, the ability of CD122/CD132-selective IL-2s to expand tumor-reactive T cells depends on their frequency within the MP CD8+ T-cell population. Another potential limitation of CD122/CD132-selective IL-2s is that tumor-reactive T cells may need to compete with the global response to the IL-2 agonist by MP CD8+ T cells. Furthermore, CD122/CD132-selective IL-2s may not optimally stimulate tumor-reactive CD4+ T cells, as these are not characterized by expression of the intermediate affinity IL-2R,9 yet these are increasingly appreciated to significantly contribute to antitumor responses.10 11

We have developed an IL-2 fusion protein comprised of mouse IL-2 and mouse CD25 (mIL-2/CD25) that has a longer half-life than IL-2 and preferentially targets cells that express the high-affinity IL-2R.12 The selectivity toward high-affinity IL-2R stems from its unique mechanism of action, whereby mIL-2/CD25 predominantly forms biologically inactive head-to-tail transdimers that slowly dissociate into low concentrations of active mIL-2/CD25 monomers that readily stimulate cells bearing the high-affinity IL-2R. In practical terms, mIL-2/CD25 provides a continuous source of an IL-2 agonist which at low doses, selectively targets Tregs to limit autoimmunity13 and at higher doses, also stimulates Teff cells with minimal stimulation of MP CD8+ T and NK cells. In previous tumor immunotherapy studies, we showed that the transient application of HD mIL-2/CD25 in the context of tumor neoantigen peptide-based vaccines readily supported antitumor immunity to the poorly immunogenic B16.F10 melanoma,14 15 validating the application of this IL-2R agonist for cancer immunotherapy.

In this study, we tested the effectiveness of HD mIL-2/CD25 monotherapy or with anti-PD-1 checkpoint blockade to limit the growth of immunogenic tumors. When compared with IL-2/anti-IL-2 immune-complexes (IL2/IC) directed toward the intermediate-affinity IL-2R, HD mIL-2/CD25 monotherapy more effectively supported antitumor responses, with optimal tumor-free survival when co-administered with anti-PD-1. HD mIL2/CD25 enhanced the expansion of tumor-reactive T cells that showed a less exhausted phenotype and supported the development of an antitumor memory response. Thus, these preclinical studies support stimulating the high-affinity IL-2R in conjunction with checkpoint blockade to optimize antitumor activity.

ResultsHD mIL-2/CD25-induced antitumor activity correlates with tumor immunogenicity

Several mouse tumor models that vary in immunogenicity were used to test the efficacy of HD mIL-2/CD25 as a monotherapy. Tumor-bearing mice were treated biweekly for 2 weeks with mIL-2/CD25 (half-life of 16–18 hours) to maintain persistent IL-2R signaling. This treatment supported robust antitumor responses in mice bearing the highly immunogenic MC38 or CT26 colon carcinomas (figure 1A, online supplemental figure 1A,B). The mIL-2/CD25 exhibited less effective antitumor responses against the poorly immunogenic B16.F10 melanoma (figure 1A, online supplemental figure 1C) and was ineffective in mice bearing the non-immunogenic 4T1 mammary carcinoma (figure 1A, online supplemental figure 1D).

Figure 1

Immunogenic tumors are susceptible to HD mIL-2/CD25 therapy. (A) C57BL/6 mice bearing MC38 (n=9–10/group pooled from two experiments) or B16F10 (n=15/group pooled from three experiments) and BALB/c mice bearing CT26 (n=32–33/group pooled from five experiments) or (4T1 n=6–7/group pooled from two experiments) received PBS or HD mIL-2/CD25 (100 µg) two times a week for 10 (MC38) or 14 days (CT26, B16.F10, 4T1) and tumor growth was followed over time. Day 0 represents the start of administration of PBS or mIL-2/CD25. (B) CT26-bearing mice (n=17–18/group pooled from four experiments) treated with HD mIL-2/CD25 that showed tumor-free survival (‘cured’) were rechallenged with CT26 90 days post-tumor clearance. (C) AH1-tetramer+ CD8+ T cells were detected in the PBMCs (n=3–4/group) after rechallenge with CT26. Data is shown for D22 (representative flow plots, left; quantitative data, right). Survival data (A,B) were analyzed by a log-rank test. AH1 tetramer data (C) are shown as mean±SD and analyzed by a Welch’s t-test of the area under the curve. HD, high-dose; mIL-2, mouse interleukin-2; PBS, phosphate-buffered saline; PBMCs, peripheral blood mononuclear cells.

The mIL-2/CD25-treated mice that cleared CT26 were rechallenged with CT26 tumor cells approximately 3 months after the primary tumor was cleared. Compared with the rapid growth of CT26 in naïve mice, none of the 17 HD mIL-2/CD25-‘cured’ mice showed signs of tumor growth for 60 days post-tumor rechallenge (figure 1B, online supplemental figure 1E). After the rechallenge, an increase in AH1-specific CD8+ T cells, an immunodominant tumor antigen of CT26,16 17 was noted when compared with naïve CT26-bearing mice (figure 1C). These data together suggest that persistent IL-2R signals expand pre-existing tumor-reactive Teff cells to effectively reject highly immunogenic tumors while supporting the development of an effective pool of tumor-reactive memory T cells.

Targeting the high-affinity IL-2R generates more effective antitumor responses than the intermediate-affinity IL-2R

Current IL-2-based approaches for antitumor immunity favor targeting the intermediate affinity IL-2R due to its selectivity toward MP CD8+ T and NK cells over Tregs.3 4 To test the antitumor efficacy of IL-2 biologics with preferential activity toward the high-affinity versus intermediate-affinity IL-2R, HD mIL-2/CD25 was compared with IL-2Rβγ-directed IL-2/anti-IL-2 (S4B6) complexes, S4B6-IL2/IC. Consistent with other studies,6 18–24 we also found that that a dose of 1.5 µg IL-2/15 µg S4B6 yielded substantial expansion of BALB/c and C57BL/6 CD8+ T and NK cells while exerting minimal effects on Tregs (online supplemental figure 2A), leading to optimal CD8+ T cell:Treg and NK:Treg ratios (online supplemental figure 2B). At the higher doses tested, this selectivity diminishes toward cells expressing only the intermediate-affinity IL-2R as Tregs increased. Thus, 1.5 µg IL-2 was complexed with 15 µg anti-IL-2 S4B6 for subsequent studies examining antitumor effects. In contrast, HD mIL-2/CD25 (100 µg) increased Tregs that express the high-affinity IL-2R but did not expand CD8+ T cells or NK cells, which express high amounts of the intermediate-affinity IL-2R.

Mice bearing MC38, CT26, or B16.F10 received the IL-2 agonists for 2 weeks. As the half-life of S4B6-IL2/IC is shorter than mIL-2/CD25,6 12 21 22 it was administered every other day to provide persistent IL-2Rβγ signaling in parallel to HD mIL-2/CD25. HD mIL-2/CD25 was more effective in supporting antitumor responses in each of these models (figure 2A). To verify the selectivity of the IL-2s during therapy, expansion of T and NK cells was assessed weekly in PBMCs. As expected, CD8+ T cells and NK cells were readily expanded with S4B6-IL2/IC whereas Tregs were expanded with HD mIL-2/CD25 in CT26-bearing mice (figure 2B). Similar results were found for mice bearing MC38 and B16.F10 (online supplemental figure 3A,B). Importantly, these IL-2R agonists differentially supported CD8+ T cell expansion in the TME, where greater expansion was found after treatment with HD mIL-2/CD25 (figure 2C). These data suggest that the driver for the improved antitumor responses by HD mIL-2/CD25 is due to more effective accumulation of CD8+ T cells in the TME when compared with S4B6-IL2/IC.

Figure 2

Targeting the high-affinity IL-2R with HD mIL-2/CD25 yields more productive antitumor responses than targeting the intermediate affinity IL-2R. (A) The indicated tumor-bearing mice (see legend to figure 1) received PBS, mIL-2/CD25 or S4B6-IL2/IC (1.5 µg mIL-2/15 µg S4B6); the latter was administered every other day for 12 (MC38, n=9/group pooled from two experiments) or 14 (CT26, 12–15/group pooled from three experiments; B16.F10, n=5–8/group pooled from two experiments) days. Mouse survival was assessed over time and analyzed by a log-rank test. (B) Frequencies of the indicated cells (n=10/group pooled from three experiments) for PBMCs collected from CT26-bearing mice. Data (mean±SD) were analyzed by the Kruskal-Wallis multiple comparison test. (C) CD8+ T cells from the indicated tissues were determined 7 days after initiation of the indicated treatments. Data (n=9/group pooled from three experiments) are shown as mean±SD and analyzed via one-way analysis of variance with Tukey’s multiple comparison test. HD, high-dose; IC, immune-complexes; mIL-2, mouse interleukin-2; NK, natural killer; PBS, phosphate-buffered saline; PBMCs, peripheral blood mononuclear cells; TDLN, tumor-draining lymph node; TME, tumor microenvironment; Tregs, regulatory T cells.

HD mIL-2/CD25 promotes antitumor responses by acting on pre-existing tumor-reactive CD8+ T cells

To further evaluate the effect of HD mIL-2/CD25 monotherapy, the distribution of major T cell subsets was determined in the TME, spleen, and tumor-draining lymph node (TDLN) in CT26-bearing mice. In the TME, HD mIL-2/CD25 supported a substantial increase in frequency of CD3+ T cells (figure 3A) and this was largely due to an increased frequency and number of CD8+ T cells (figure 3B). The frequency of conventional CD4+ Foxp3− T (Tconv) cells was low and unaffected and strikingly, the frequency of Tregs decreased (figure 3C). In contrast, the number of CD8+ T cells in the spleen and TDLN were unchanged after HD mIL-2/CD25 (figure 3D); as expected, a large increase in Tregs was found in the spleen, with a much lesser effect in the TDLN (figure 3D). Targeting the high-affinity IL-2R with HD mIL2/CD25, therefore, selectively led to a very high CD8+ T cell to Treg ratio in the TME but not the spleen and TDLN (figure 3E). Overall, these data indicated that HD mIL-2/CD25 effectively supported the expansion of Tregs in secondary lymphoid tissues, but only CD8+ T cells showed expansion in the TME, leading to a high CD8+ T cell:Treg ratio, which is a good prognostic factor for antitumor responses.

Figure 3

HD mIL-2/CD25-induced antitumor activity is CD8+ T cell-dependent. (A–E) 7 days post treatment initiation, the indicated tissues from CT26-bearing BALB/c mice were examined for the frequency and number of the indicated T cell populations. Data (mean±SD) are pooled from three experiments (n=11/group for TME and spleen; n=6–7/group for TDLN) and analyzed by Mann-Whitney test. (A) CD3+ T cell frequencies, (B) CD8+ T cell frequencies and numbers, (C) CD4+ Tconv and CD4+ Treg frequencies. (D) The numbers of CD8+ T cells and Tregs in the spleen and TDLN. (E) The CD8+ T cell to Treg ratio in the TME, spleen, and TDLN. (F) CT26-bearing mice (n=8–11/group pooled from two experiments) received anti-CD8 to deplete CD8+ T cells 3 days prior (D−3,−2, −1) to administrating HD mIL-2/CD25, represented as D0. Data were analyzed by a log-rank test. HD, high-dose; mIL-2, mouse interleukin-2; PBS, phosphate-buffered saline; TDLN, tumor-draining lymph node; TME, tumor microenvironment; Tregs, regulatory T cells.

To directly test that CD8+ T cells importantly contribute to the HD mIL-2/CD25-mediated antitumor response, CD8+ T cells were depleted prior to treatment with HD mIL-2/CD25. The effectiveness of CD8+ T cell depletion prior to and during treatment with HD IL-2/CD25 was verified for PBMCs (online supplemental figure 4A) for the same mice antitumor responses were evaluated. In the absence of CD8+ T cells, HD mIL-2/CD25-induced antitumor responses were completely abrogated (figure 3F, online supplemental figure 4B). These results indicate that CD8+ T cells are the major cellular driver of the antitumor activity induced by HD mIL-2/CD25.

Increased neutrophils in the TME by HD mIL-2/CD25 do not contribute to the anticancer response

The effect of HD-mIL-2/CD25 was also assessed for other cell types in the TME of CT26-bearing mice. We were particularly interested in dendritic cells (DCs) as previous studies found that IL-2 and IL2/ICs indirectly activate and promote the expansion of type-I and type-II conventional DCs (cDC1s and cDC2s)19 25 26 and in one study they confirmed that this occurred through FLT3L, CSF-2, and TNF and that the increase in DCs correlated with favorable anticancer responses.26 The frequency (online supplemental figure 5A) and number (online supplemental figure 5B) of cDC1s and cDC2s as well as natural killer T (NKT) cells, and NK cells, was unchanged in the TME, where the latter result further demonstrates that HD mIL-2/CD25 is less effective in targeting the intermediate-affinity IL-2R. We do not know why we did not detect an increase in either cDC1 or cDC2 subsets but this may reflect some differences in the duration or cellular targeting of IL-2R signals by mIL-2/CD25.

A significant increase, however, was noted for neutrophils (online supplemental figure 5A,B). To assess whether this increase in neutrophils contributes to the antitumor response, anti-Ly6G was used to deplete these cells27 prior to treatment of CT26-bearing mice with HD mIL-2/CD25. Examination of PBMCs revealed that anti-Ly6G effectively depleted neutrophils (online supplemental figure 6A). Nevertheless, the HD mIL-2/CD25-driven antitumor response to CT26 was largely unaffected (online supplemental figure 6B). Thus, the increase in neutrophils is not necessary for this antitumor response and may reflect the increased CD8+ tumor-infiltrating lymphocytes (TILs) whose effector response recruits neutrophils to the TME.

HD mIL-2/CD25 enhances the number and function of tumor-antigen specific CD8+ T cells

Given that IL-2R signaling upregulates CD25 on antigen-activated Teff cells, we evaluated CD25 expression on CD8+ T cells in peripheral immune tissues and the TME during HD mIL-2/CD25 therapy. A significant increase in CD25+ CD8+ T cells was noted for the TME, spleen, and TDLN (figure 4A). This finding suggests that in a tumor-burden setting, HD mIL-2/CD25 upregulates CD25 expression on CD8+ tumor-reactive T cells.

Figure 4

HD mIL-2/CD25 supports increased number of tumor antigen-specific CD8+ T cells. (A) 7 days post treatment initiation, the indicated tissues from CT26-bearing mice were examined for CD25 expression by flow cytometry (left, representative flow plot; right, quantitative data). Data (n=6/group pooled from two experiments) are shown as mean±SD and analyzed by Mann-Whitney test. (B) PBMCs from CT26-bearing BALB/c mice were evaluated by flow cytometry for the frequency and number of AH1-specific CD8+ T cells during treatment with PBS or mIL-2/CD25. (representative flow plots of D7, left; quantitative data, right). Data (n=5/group pooled from two experiments) are shown as the mean±SD and analyzed by an unpaired t-test with Welch’s correction of the area under the curve. (C) 7 days post treatment initiation of CT26-bearing mice, the number of AH1 tetramer-specific CD8+ T cells was determined for the indicated tissues. Data (n=9/group pooled from three experiments) are shown as the mean±the SD and analyzed by Mann-Whitney test. HD, high-dose; mIL-2, mouse interleukin-2; PBS, phosphate-buffered saline; PBMCs, peripheral blood mononuclear cells; Spl, spleen; TDLN, tumor-draining lymph node; TME, tumor microenvironment.

To directly investigate the effect of HD mIL-2/CD25 on tumor-reactive T cells, the number of CD8+ T cells specific for AH1 was determined. A time course study revealed a striking increase in percentage and number of AH1-specific CD8+ T cells in the PBMCs (figure 4B). Although some variability was noted, an increase in AH1+ CD8+ T cells was also detected in the TME and spleen, but not the TDLN (figure 4C) in CT26-bearing mice after treatment with HD mIL-2/CD25.

With respect to functional activity, when CD45-enriched cells from CT26-bearing mice were stimulated ex vivo with the AH1 peptide, HD mIL-2/CD25 supported increased numbers of tumor-specific interferon (IFN)-γ-secreting T cells in the TME, spleen, and TDLN (figure 5A), but with similar mean spot size between phosphate-buffered saline (PBS)-treated and HD mIL-2/CD25-treated mice (online supplemental figure 7), indicative of similar IFN-γ secretion per cell. After re-stimulation in vitro, CD8+ TILs showed an increase in frequency of granzyme B (GrzB)+ cells after stimulation with PMA and ionomycin (figure 5B, top) and trended upward after stimulation with tumor antigen AH1 (figure 5B, bottom). In both settings, these CD8+ TILs produced greater amounts of GrzB, as assessed by the mean fluorescence intensity (MFI) (figure 5B). In addition, HD mIL-2/CD25 supported a greater percentage of polyfunctional IFNγ+ TNF+ CD8+ TILs after rechallenge in vitro under both conditions (figure 5C). Collectively these data indicate that HD mIL-2/CD25 targets the high-affinity IL-2R on tumor-specific T cells to increase their number and function.

Figure 5

HD mIL-2/CD25 enhances the effector function of tumor antigen-specific CD8+ T cells. (A) 7 days post treatment initiation, the indicated tissues from CT26-bearing mice were enriched for CD45+ immune cells and examined for IFN-γ-secreting cells by ELISpot (Enzyme-Linked ImmunoSpot) assay (representative wells, left; quantitative data of spot-forming cells (SFC), right) 19–24 hours post in vitro restimulation with BMDCs in the presence or absence of AH1 peptide (2 µg/mL). Data (n=15–18/group pooled from five experiments) are mean±SD and were analyzed by Mann-Whitney test. (B–C) CD45-enriched tumor-associated immune cells were also restimulated in vitro in the presence of PMA and ionomycin (P+I) or BMDCs in the presence (DCs+AH1) or absence (DCs only) of AH1 peptide (10 µg/mL). Cytokine production was evaluated by flow cytometry 4 hours after restimulation. (B) Frequency of and corresponding MFIs of granzyme B-expressing CD8+ T cells is shown (representative histograms, left; quantitative data, right). Data (n=8/group pooled from three experiments) are mean±SD and analyzed by Mann-Whitney test. (C) Frequency of IFN-γ+ TNF+ double cytokine-producing CD8+ T cells is shown (representative flow plots, left; quantitative data, right). Data (n=5/group are representative of one experiment) are mean±SD and were analyzed by Mann-Whitney test within each condition. BMDCs, bone marrow-derived DCs; DCs, dendritic cells; GrzB, granzyme B; HD, high-dose; IFN, interferon; MFI, mean fluorescence intensity; mIL-2, mouse interleukin-2; PBS, phosphate-buffered saline; TDLN, tumor-draining lymph node; TILs, tumor-infiltrating lymphocytes; TME, tumor microenvironment.

HD mIL-2/CD25 reduces exhaustion of CD8+ T cells in the TME

Persistent antigen stimulation that occurs in the TME leads to T cell exhaustion.28 To test whether HD mIL-2/CD25 might lower T cell exhaustion, the expression of PD-1 and TOX, which are associated with exhausted T cells,29 30 was examined for CD8+ T cells after treatment of CT26-bearing mice with HD mIL-2/CD25 or PBS. In the spleen and TDLN, HD mIL-2/CD25 supported a modest increase in PD-1+TOX− CD8+ T cells, a finding consistent with an increase in antigen-experienced cells31(figure 6A). In contrast, most CD8+ TILs expressed PD-1 and TOX (figure 6B). However, HD mIL-2/CD25 supported lower levels of both exhaustion markers, consistent with less exhausted CD8+ T cells (figure 6B,C). In addition, the co-expression of PD-1 with LAG-3, another immune checkpoint associated with exhaustion,32 33 was selectively lower after HD mIL-2/CD25 treatment (figure 6D).

Figure 6

CD8+ T cells from HD mIL-2/CD25-treated CT26-bearing mice are less phenotypically exhausted. (A–D) 7 days post treatment initiation, the indicated tissues from CT26-bearing mice were examined for markers characteristic of CD8+ T cell exhaustion. (A) The frequency of PD-1+TOX− cells of total CD8+ T cells in the spleen and TDLN is shown in addition to (B) the frequency of PD-1+TOX+ of total CD8+ T cells in the TME (representative flow plots, left; quantitative data, right). Data (n=6–7/group pooled from two experiments) are mean±SD and analyzed by Mann-Whitney test. (C) The normalized MFI of PD-1 and TOX are shown for CD8+ T cells in the TME. Data (n=11/group pooled from four experiments) are mean±SD and analyzed via Mann-Whitney test. (D) The frequency of PD-1+LAG-3+ of total CD8+ T cells in the TME is shown (representative flow plots, left; quantitative data, right). Data (n=11/group pooled from four experiments) are mean±SD and analyzed by Mann-Whitney test. (E) The frequency, number, and normalized PD-1 MFI of PD-1+ TCF1hi CD8+ T cells in the TME is shown (representative flow plots, left; quantitative data, right). Data (n=8/group, pooled from two experiments) are mean±SD and analyzed via Mann-Whitney test. mIL-2, mouse interleukin-2; MFI, mean fluorescence intensity; PBS, phosphate-buffered saline; TDLN, tumor-draining lymph node; TILs, tumor-infiltrating lymphocytes.

Expression of TCF1 and PD-1 marks subpopulations of exhausted CD8+ T cells, where progenitor-like PD-1+TCF1hi CD8+ T cells give rise to short-lived terminally exhausted PD-1+TCF1lo CD8+ T cells.34–36 A high percentage of progenitor-like PD-1+TCF1hi CD8+ T cells were found in the TME of PBS-treated and HD mIL-2/CD25-treated mice (figure 6E). However, the number of such cells after HD mIL-2/CD25 were on average 6.5-fold greater, although some variability was noted, and the MFI of PD-1 on these cells was approximately 40% lower (figure 6E). This phenotype in conjunction with lower TOX (figure 6B,C) is consistent with less exhausted CD8+ T cells that are expected to be more responsive to checkpoint blockade.34 37 38 These changes along with lower LAG-3 likely represent effects of HD mIL-2/CD25 on CD8+ TILs, although we cannot rule out that decreased tumor burden will provide lower antigenic stimulation, and this may also contribute to these effects. These phenotypic data in conjunction with the increased number and function of AH1 tumor-specific T cells (figures 4 and 5) and enhanced antitumor responses (figures 1 and 2) are consistent with a role for targeting CD8+ T cells expressing the high-affinity IL-2R by HD mIL-2/CD25 to promote less exhausted CD8+ T cells with high antitumor activity.

Antitumor immunity by HD-mIL-2/CD25 is enhanced by co-administering anti-PD-1

While the frequency of PD-1+ CD8+ T cells remained high, the relative PD-1 expression on tumor-infiltrating CD8+ T cells was decreased with treatment (figure 6B,C). This is a good prognosis for the therapeutic efficacy of PD-1 blockade39 and therefore led us to investigate the effects of PD-1 blockade in conjunction with HD mIL-2/CD25 therapy. HD mIL-2/CD25 alone, but not anti-PD-1, supported tumor-free survival in approximately 50% of the mice, as previously observed (figure 1A and figure 7A). However, the combination of HD mIL-2/CD25 plus anti-PD-1 was much more effective, where approximately 90% of the mice were long-term survivors (figure 7A, online supplemental figure 8A). When these latter mice were rechallenged with CT26, all mice remained tumor-free, indicative of the induction of immune memory (figure 7B). In additional cohorts of mice, unlike the synergistic activity between HD mIL-2/CD25 and anti-PD-1, combining anti-PD-1 with S4B6-IL2/IC (figure 7C, online supplemental figure 8B) did not support vigorous antitumor responses. Thus, targeting the high-affinity IL-2R is essential for increasing antitumor immunity when applied with anti-PD-1 checkpoint blockade.

Figure 7

HD mIL-2/CD25 and PD-1 blockade generate highly effective antitumor activity. (A) CT26-bearing mice were treated with PBS, anti-PD-1 (200 µg), MOPC-21 IgG isotype control (200 µg), mIL-2/CD25 (100 µg), or combination therapy as indicated (n=9–18/group pooled from three experiments) and survival was determined. Data were analyzed by a log-rank test. (B) Naïve mice or mice that cleared CT26 after combination therapy (mIL-2/CD25+anti-PD-1) were rechallenged with CT26 90 days after primary tumor clearance and survival was determined. Data (n=4/group pooled from two experiments) were analyzed by a log-rank test. (C) CT26-bearing mice were treated with PBS, anti-PD-1 (200 µg), S4B6-IL2/IC (1.5 µg mIL-2/15 µg S4B6), or combination therapy as indicated and survival was determined. Data (n=8–9/group pooled from two experiments) were analyzed by a log-rank test. (D) 7 days post treatment initiation in the experiments described in A, TILs from CT26-bearing mice were examined for expression of PD-1 and TOX (representative flow plots, left; quantitative data, right). Data (n=6/group pooled from two experiments) are mean±SD and analyzed by one-way analysis of variance with Tukey’s multiple comparison test. HD, high-dose; IC, immune-complexes; mIL-2, mouse interleukin-2; PBS, phosphate-buffered saline; TILs, tumor-infiltrating lymphocytes.

When TILs from CT26-bearing mice were examined during the primary antitumor response, mIL-2/CD25 supported a reduction of PD-1+TOX+ CD8+ T cells and this effect was at best only slightly increased when HD mIL-2/CD25 was combined with anti-PD-1 (figure 7D). Thus, preferential IL-2R signaling through the high-affinity IL-2R is the main driver of CD8+ T cells with a less exhausted phenotype in this combination therapy.

Discussion

Our study demonstrates that stimulating the high-affinity IL-2R on tumor-specific CD8+ Teff cells is essential for effective IL-2-dependent antitumor immunotherapy. HD mIL-2/CD25 monotherapy, which targets the high-affinity IL-2R, supported substantial antitumor responses against highly immunogenic tumors and when combined with anti-PD-1 checkpoint blockade, tumors were eradicated in essentially all mice. These findings contrast with the consensus that targeting the intermediate-affinity IL-2R is the best strategy for IL-2-dependent antitumor immunotherapy due to its ability to limit undesired activation of Tregs. However, even though HD mIL-2/CD25 increases Tregs, it supported more robust antitumor responses than using an IL2/IC that targets the intermediate-affinity IL-2R. The IL2/ICs were used at a dose and administered in a manner that preferentially expanded MP CD8+ T and NK cells over Tregs and generated antitumor immune responses similar to that reported by other studies.18–24 Our findings highlight that a limitation of broadly increasing CD8+ T cells, even with an accompanying expansion of NK cells, is an insufficient increase of tumor antigen-specific CD8+ T cells, which are the critical cells for effective antitumor immunity.

The half-lives and modes of action of mIL-2/CD25 and S4B6-IL2/IC are different,6 12 21 22 complicating the experimental design comparing these two agents. We used an amount of S4B6-IL2/IC that has been commonly used in other studies18–24 exploring antitumor activity by targeting the intermediate affinity IL-2R. The frequency of administration of S4B6-IL2/IC has been somewhat more variable as in some studies it was administered daily for 3–5 days whereas other studies administered the IL2/IC every other day. Based on this concern, we confirmed that 1.5 µg of IL-2 associated with 15 µg S4B6 yielded high expansion of CD8+ T and NK cells while minimally affecting Tregs when administered as a single injection or when treating tumor-bearing mice with a therapeutic regimen. We found selective expansion of CD8+ T and NK cells by our regimen not only after a single administration of S4B6-IL2/IC, but also during therapeutic dosing. We cannot rule out improved anticancer responses if we administered greater amounts of S4B6-IL2/IC. However, we believe we are approaching the maximal dosing and frequency of S4B6-IL2/IC because when we modestly increased the dose of S4B6-IL2/IC to 5 µg IL-2/50 µg S4B6, severe toxicity resulted, likely from a cytokine storm due to high expansion of CD8+ and NK cells. 100 µg of mIL-2/CD25 is a much larger dose of IL-2 than 1.5 µg IL-2/15 µg S4B6. However, this fusion protein is administered largely as an inactive dimer, where we have previously estimated that only a small fraction, approximately 1%, at any time is biologically active as mIL-2/CD25 monomers slowly and continually dissociate from the inactive dimer.12 Thus, administration of 1