N-803 was generously provided by NantBioScience under a Cooperative Research and Development Agreement with the National Cancer Institute. αPD-L1 (10F.9G2), CD8 (2.43), and CD4 (GK1.5) depletion antibodies were from BioXcell. The NK depletion antibody (anti-asialo-GM1) was from Wako Chemicals.

Six- to 10-week old female Balb/c and C57BL/6 mice transgenic for carcinoembryonic antigen (CEA) (designated C57BL/6-CEA) [24] were obtained from the NCI Frederick Cancer Research Facility and maintained under specific pathogen-free conditions in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) guidelines. All studies were approved by the NIH Intramural Animal Care and Use Committee (IACUC).

4T1 TN breast carcinoma and Yac-1 lymphoma cells were obtained from American Type Culture Collection (ATCC). MC38 colon carcinoma cells expressing human CEA (MC38-CEA) were generated as previously reported [25]. All cell lines were cultured according to providers’ instructions [25], determined as mycoplasma free by MycoAlert Mycoplasma Detection Kit (Lonza), and used at low passage number.

For anti-tumor studies, 4T1 tumor cells (5 × 10⁴, s.c.) were orthotopically implanted into the mammary fat pad of female Balb/c mice on day 0. In select studies, the primary tumor was surgically excised at day 15. MC38-CEA (5 × 10⁵, s.c.) tumor cells were implanted into the right flank of female C57BL/6-CEA mice. Tumors were measured biweekly using calipers, and volumes were determined as (length² × width)/2. Mice were randomized based on tumor size and treatment initiated when tumors reached 50-100 mm³. Mice received three doses of 200 μg αPD-L1 i.p. (10 mg/kg), a clinically relevant dose [21], and/or two doses of N-803 s.c. at 1 μg [9]. Quantification of 4T1 lung metastasis was performed as previously described [26, 27].

CD4 or CD8 depletion antibodies (100 μg, i.p.) were administered on days 6, 7, and 8 post-tumor implant followed by once weekly. NK depletion antibody (25 μl in 100 μl PBS, i.p.) was administered on days 6 and 8 post-tumor implant, then every 3 days. Due to toxicity, depletions were terminated after day 19. Weekly depletion efficiency was determined in the blood (~ 50 μl) by flow cytometry. Percent reduction of CD8⁺ T, NK, or CD4⁺ T cells was determined versus undepleted N-803 + αPD-L1-treated mice (set to 0%).

Homogenized and filtered spleens underwent ACK lysis. Tumors and lungs were digested using the gentleMACS Dissociator according to the manufacturer’s instructions (Miltenyi Biotec). Tumor-infiltrating lymphocytes (TILs) and lung-resident immune cells were enriched using a 44%/67% Percoll Plus (GE Healthcare) gradient. Cell counts were performed using 123count eBeads (ThermoFisher Scientific).

Antibody labeling of cells for flow cytometry (1–10 × 10⁶ immune cells) was performed using the BD Cytofix/Cytoperm Kit (BD Biosciences) according to the manufacturer’s instructions. Antibodies (Additional file 1: Table S1) and matched isotypes were obtained from the listed manufacturers. Live/Dead Fixable Dead Cell Stain was from Invitrogen. Flow cytometry (≥1 × 10⁵ events) was performed on a BD FACSVerse or LSRFortessa flow cytometer (Becton Dickinson) and analyzed with FlowJo FACS Analysis Software v9.9.6 (Treestar). Cell populations were identified as listed (Additional file 1: Table S2). Expression of phenotypic proteins was determined by subtracting the respective isotype, set between 1 and 5% of the population.

To discriminate the presence of lung parenchymal versus intravascular immune cells, mice were injected with anti-CD45.2 (3 μg, i.v.) as previously described [28]. See Additional file 1: Figure S1.

Isolated spleen, lung, and tumor immune cells were stimulated for 4 h in the presence of 2 μg/ml GolgiPlug (BD Biosciences) with nothing, for the CD8⁺ T cell restimulation 1 μg/ml αCD3 (2C11, BD Biosciences) + 1 μg/ml αCD28 (37.51, BD Biosciences), or for the NK cell restimulation 50 ng/ml PMA (Sigma) + 500 ng/ml ionomycin (Sigma). Frequencies of stimulated IFNγ⁺ and/or TNFα⁺ cells were calculated by subtracting the non-stimulated controls.

Purified splenic NK cells (NK Cell Isolation Kit, Miltenyi Biotec) were co-cultured for 18 h with ¹¹¹In-labeled Yac-1 target cells at a 100:1 effector-to-target ratio. Supernatant ¹¹¹In radioactivity was quantified (WIZARD² γ counter, PerkinElmer), and percent lysis was determined as previously described [9].

Serum cytokines were quantified using the V-PLEX Proinflammatory Panel I Mouse Kit and MESO QuickPlex SQ 120 (Meso Scale Diagnostics, LLC). Limits of detection were: IFNγ: 0.04 pg/ml, IL-6: 0.61 pg/ml, IL-10: 0.95 pg/ml, TNFα: 0.13 pg/ml.

Statistical analyses were performed in Prism 7.0a (GraphPad Software). Unless otherwise stated, data presented in bar graphs or scatter plots were analyzed using one-way ANOVA with Tukey’s multiple comparisons. Two-way ordinary ANOVA was used to analyze tumor growth curves. Survival was analyzed using Log-rank (Mantel-Cox) test. Outliers were identified using ROUT test. Statistical significance was set at p < 0 .05. *p < 0.05, **p < 0.01, ***p < 0.001.

N-803 reduces tumor burden in multiple murine solid carcinoma models. However, in many cases, it was not curative [5, 9]. It has been shown that IL-15 increases PD-L1 expression on immune cells in vitro [22], but the effect of rIL-15 or N-803 on PD-L1 expression in vivo has not been reported. Thus, it is possible that N-803 enhances PD-L1 expression in tumor-bearing mice, reducing the efficacy of N-803 monotherapy. Indeed, s.c.-delivered N-803 enhanced PD-L1 expression on CD45⁺ hematopoietic cells in the primary tumor, lung parenchyma and vasculature, and spleen of 4T1 TN breast tumor-bearing mice (Fig. 1a). In addition, granulocytic myeloid derived suppressor cell (G-MDSC)/granulocyte and monocytic MDSC (M-MDSC)/monocyte populations had increased expression of PD-L1 (Fig. 1b,c). PD-L1 expression on non-hematopoietic, tumor-associated CD45⁻ cells did not change (Fig. 1a). These results demonstrate that N-803 increases PD-L1 expression on immune cells in vivo and provide the rationale for combining N-803 with a PD-L1–targeted antibody for the treatment of solid carcinomas.Fig1Fig. 1

N-803 treatment increases PD-L1 expression on total CD45⁺ cells and MDSC populations in the primary tumor, lung, and spleen. 5 × 10⁴ 4T1 tumor cells were orthotopically implanted into female Balb/c mice. When tumor volumes reached ~50mm³, mice were treated at days 9 and 13 with 1 μg N-803 (s.c.). Twenty-four hours after the last treatment, PD-L1 expression (geometric mean fluorescence intensity (gMFI)) in the primary tumor, lung parenchyma and vasculature, and spleen was determined by flow cytometry on total CD45⁺ or CD45⁻ cells (a), G-MDSC/Granulocytes (b), or M-MDSC/Monocytes (c). Graphs show mean ± SD. Data combined from 2 independent experiments, n = 5 mice/group per experiment

Given the increase in PD-L1 expression upon N-803 treatment, we examined the combination of N-803 with an αPD-L1 monoclonal antibody for anti-tumor efficacy. Successful combination of these agents has not been reported [9], and previous studies using N-803 with checkpoint blockade did not use subcutaneous delivery of N-803, the route of administration used in the clinic [7, 8, 23].

First, we examined the efficacy of N-803 plus αPD-L1 (N-803 + αPD-L1) combination therapy in mice bearing 4T1 TN breast tumors, which are highly metastatic [29] and have previously demonstrated resistance to αPD-L1 monotherapy [30]. Initial studies determined that 1 μg s.c. N-803 in combination with 200 μg (10 mg/kg) αPD-L1, a clinically relevant dose [21], promoted greater reduction in lung tumor burden than higher doses of N-803 (Additional file 1: Figure S2A-C). Furthermore, significant reduction in body weight was observed after treatment with higher doses of N-803 (Additional file 1: Figure S2D-E). Thus, all studies were performed using 1 μg N-803 s.c. plus 200 μg αPD-L1 i.p.

In the adjuvant setting, without resection of the primary tumor, N-803 + αPD-L1 reduced primary 4T1 tumor growth (Fig. 2a) and decreased the number of lung metastases by 78% compared to phosphate buffered saline (PBS) treatment (Fig. 2b,c). Neither αPD-L1 nor N-803 monotherapy significantly reduced primary or lung tumor burden compared to PBS treatment (Fig. 2a-c). Combination therapy further reduced lung metastasis burden by 75 and 62% compared to αPD-L1 and N-803 monotherapies, respectively (Fig. 2c).Fig2Fig. 2

Combination of N-803 and αPD-L1 decreases 4T1 lung metastasis and improves survival. a-c Mice were implanted with 4T1 tumors as in Fig. 1 and treated on days 8 and 14 with 1 μg N-803 (s.c.) and/or 200 μg αPD-L1 (i.p.) on days 8, 11, and 14. Graphs of tumor volumes of individual animals at day 25 post-tumor implant in (a) and number of lung metastases in individual mice 26 days after tumor implant (b) show mean ± SD. c Table denotes the distribution of number of lung metastases and % reduction in mean versus PBS. Data are representative of 2 independent experiments, n = 20–23 mice. d-g Mice were implanted with 4T1 tumors as in Fig. 1 and treated at days 9 and 13 with N-803 and/or αPD-L1 on days 9, 11, and 13. The primary tumor was surgically resected at day 15. Graphs of tumor volumes of individual animals at day 14 post-tumor implant (d) and number of lung metastases in individual mice at day 28 (e) show mean ± SD. f Table denotes the distribution of number of lung metastases per mouse and % reduction in mean versus PBS. Data are representative of 2 independent experiments, n = 15–19 mice. g Survival curves (inset: mOS) show % survival. Data are representative of 2 independent experiments, n = 13–19 mice

To begin elucidating the mechanism by which N-803 + αPD-L1 mediated anti-tumor efficacy, we first determined the immune cells required to reduce tumor burden in 4T1 tumor-bearing mice. As CD8⁺ T and NK cells were necessary for the anti-tumor efficacy of N-803 monotherapy in the 4T1 model [9], we performed CD8⁺ T cell and NK cell depletions in N-803 + αPD-L1–treated mice. Vascular CD8⁺ T cells and NK cells were depleted by ~ 98% and ~ 90%, respectively, over the first 3 weeks of the experiment (Additional file 1: Figure S4A-C). N-803 + αPD-L1 significantly reduced primary tumor growth and lung metastasis versus PBS treatment (Fig. 3a-c). While CD8⁺ T cell depletion did not significantly increase the number of lung metastases versus N-803 + αPD-L1 therapy, it did increase mean lung tumor burden by 175% (Fig. 3b,c). Furthermore, there was no significant difference in the number of lung metastases in CD8-depleted mice compared to PBS treatment (Fig. 3b,c). The NK cell depletion, on the other hand, significantly increased lung tumor burden by 589% versus N-803 + αPD-L1 therapy (Fig. 3b,c). Depletion of both CD8⁺ T cells and NK cells mimicked the results seen with NK cell depletion (Fig. 3a-c). Of note, 99% depletion of CD4⁺ T cells had no effect on the anti-tumor efficacy of N-803 + αPD-L1 (Additional file 1: Figure S4D-G). These results indicate NK cells, and to a lesser extent CD8⁺ T cells, are required for the anti-tumor efficacy of N-803 + αPD-L1 therapy in 4T1 tumor-bearing mice.Fig3Fig. 3

CD8⁺ T cells and NK cells contribute to the anti-tumor efficacy of N-803 + αPD-L1 combination. Mice were implanted with 4T1 tumors as in Fig. 1 and treated on days 13 and 17 with N-803 and αPD-L1 on days 13, 15, and 17. CD8-expressing cells and NK cells were depleted on days 10, 11, 12, 16, and 19 using 100 μg anti-CD8 and/or 25 μl anti-asialo-GM1 (i.p.). Graphs show tumor volumes of individual mice (a) or number of lung metastases in individual mice at day 23 post-tumor implant (b) as mean ± SD. c Table denotes the distribution of the number of lung metastases per mouse and % reduction in mean versus PBS. Data combined from 2 independent experiments, n = 23–25 mice/group total

To further elucidate the mechanism of N-803 + αPD-L1–induced anti-tumor efficacy, we performed a comprehensive characterization of immune cell populations in the primary tumor, in the lung as a site of metastasis, and in the peripheral immune compartment in the spleen of 4T1 tumor-bearing mice 24 h after the last N-803 and/or αPD-L1 treatment. The effect of N-803 + αPD-L1 combination on the immune system in murine solid carcinoma models has not been previously reported.

While we were particularly interested in NK and CD8⁺ T cells, given the results of the depletion studies, we first investigated the effect of N-803 + αPD-L1 therapy on regulatory T cell (T_reg) and MDSC populations, as these immunosuppressive cells play important roles in the progression of 4T1 tumors [31, 32]. T_reg numbers were significantly decreased in the lung vasculature, but not in other immune compartments, with all treatments (Additional file 1: Figure S6). In addition, N-803 + αPD-L1 treatment decreased the number of G-MDSC/granulocytes in the lung parenchyma and vasculature (Additional file 1: Figure S7A). αPD-L1 and N-803 also reduced G-MDSC/granulocytes in the lung parenchyma or vasculature, respectively (Additional file 1: Figure S7A). There was no effect on the number or frequency of the M-MDSC/monocyte population (Additional file 1: Figure S7B). Thus, N-803 + αPD-L1 combination decreases immunosuppressive T_reg and G-MDSC/granulocyte populations in the lung.

Given the necessity of NK cells for N-803 + αPD-L1 anti-tumor efficacy, we examined the effect of combination therapy on NK cell phenotype. Furthermore, while N-803 is known to greatly increase the number and activation of NK cells in the tumor and spleen of tumor-bearing mice [9–11], the effect of subcutaneous N-803 plus checkpoint blockade on NK cell phenotype and function has not been reported. In the primary tumor, N-803 + αPD-L1 did not affect NK cell number (Fig. 4a), frequency (Additional file 1: Figure S8A), or expression of activation-associated receptors NKG2D (Fig. 4b) or NKp46 (Additional file 1: Figure S8B). There was, however, a significant increase in the proportion of Ki67⁺ NK cells (Fig. 4c), indicating an increased proliferative potential.Fig4Fig. 4

N-803 monotherapy and N-803 + αPD-L1 combination promote an activated NK cell phenotype and increase function. Mice were implanted with 4T1 tumors as in Fig. 1 and treated on days 9 and 13 with N-803 and/or αPD-L1 on days 9, 11, and 13. a-d NK cells were examined by flow cytometry in the primary tumor, lung vasculature, and spleen 24 h after the last treatment. Graphs show NK cell number (a) and frequencies of NKG2D⁺ (b), Ki67⁺ (c), and Granzyme B⁺ (d) NK cells. e Purified splenic NK cells were co-cultured with ¹¹¹In-labeled Yac-1 target cells at a 100:1 effector-to-target (E:T) ratio for 18 h. ¹¹¹In release was measured to determine cytotoxic function. All graphs show mean ± SD. Data combined from 2 to 3 independent experiments, n = 4–5 mice/group per experiment

The increased anti-tumor efficacy of N-803 + αPD-L1 versus N-803 monotherapy could not be fully explained by differences in NK cell phenotype and function, as both treatments induced similar changes to these cells. While CD8⁺ T cells contributed less to the reduction of 4T1 lung metastasis by N-803 + αPD-L1 than NK cells, they did partially support this anti-tumor efficacy (Fig. 3). It is known that both N-803 and αPD-L1 monotherapies enhance CD8⁺ T cell proliferation, activation, and function [9–12], but the effect of combining N-803 plus checkpoint blockade on CD8⁺ T cells has not been elucidated. Similar to NK cells, there were no significant differences in CD8⁺ T cell number (Fig. 5a) or phenotype (Fig. 5b-d) in the primary tumor with treatment, other than an increase in the frequency of Ki67⁺ CD8⁺ T cells (Fig. 5d), indicating an increased proliferative potential. In the lung, N-803 + αPD-L1 therapy enhanced lung parenchymal-resident CD8⁺ T cell number (Fig. 5a) and the proportion of activated CD44^hi (Fig. 5b) and Ki67-expressing (Fig. 5d) cells. In the lung vasculature, there was also a significant enrichment of the population of activated CD44^hi CD8⁺ T cells (Fig. 5b), specifically of a central memory (T_CM) phenotype (Fig. 5c), after N-803 + αPD-L1 versus PBS treatment. N-803 + αPD-L1 therapy also greatly altered the splenic CD8⁺ T cell population. While N-803 + αPD-L1 did not affect CD8⁺ T cell number (Fig. 5a) versus PBS treatment, combination therapy augmented the frequency of Ki67⁺ (Fig. 5d) and activated CD44^hi CD8⁺ T cells (Fig. 5b), in particular those with a T_CM phenotype (Fig. 5c). Similar to NK cells, the changes in CD8⁺ T cell number and phenotype, with the exception of Ki67, appeared dependent on N-803, but not αPD-L1, monotherapy (Fig. 5).Fig5Fig. 5

Combination of N-803 + αPD-L1 induces an activated CD8⁺ T cell phenotype. Mice were implanted with 4T1 tumors as in Fig. 1 and treated on days 9 and 13 with N-803 and/or αPD-L1 on days 9, 11, and 13. CD8⁺ T cells were examined by flow cytometry in the primary tumor, lung parenchyma and vasculature, and spleen 24 h after the last treatment. Graphs of CD8⁺ T cell number (a), frequency of CD44^hi (b), CD44^hiCD62L^hi T_CM (c), and Ki67⁺ CD44^hi (d) CD8⁺ T cells show mean ± SD. Data combined from 2 to 3 independent experiments, n = 5 mice/group per experiment

Together, the aforementioned results suggest that N-803 + αPD-L1 therapy would induce significantly higher global levels of immunostimulatory cytokines IFNγ and TNFα, driven by CD8⁺ T and NK cell production, as compared to αPD-L1 and N-803 monotherapies. Along with the proinflammatory cytokine IL-6, sustained high levels of IFNγ and TNFα can induce significant clinical toxicities, as was seen in early rIL-15 clinical trials [6–8]. Thus, we examined the kinetics of global serum cytokine levels. Twenty-four hours after the last treatment, N-803 + αPD-L1 therapy induced significantly higher levels of serum IFNγ (Fig. 7a) and TNFα (Fig. 7b) than all other treatments. The anti-inflammatory cytokine IL-10 (Fig. 7c) was also markedly enhanced versus PBS and N-803 therapy. The increased IFNγ, TNFα, and IL-10 levels were transient, as they returned to PBS-treated concentrations by day 21 or 7 days after treatment (Fig. 7a-c). IL-6 was not significantly increased in the serum with N-803 + αPD-L1 treatment (Fig. 7d). Importantly, even with the highly elevated levels of IFNγ with N-803 + αPD-L1 treatment, the combination was well-tolerated. At 1 μg N-803 plus 200 μg αPD-L1, there was no significant loss of body weight observed when compared to PBS treatment (Additional file 1: Figure S2D-E). In all, these results support that N-803 + αPD-L1 combination is uniquely able to induce a transient, immunostimulatory environment that promotes significant anti-tumor efficacy without inducing significant toxicity.Fig7Fig. 7

N-803 + αPD-L1 combination promotes the generation of an immunostimulatory milieu in the serum. Mice were implanted with 4T1 tumors as in Fig. 1 and treated on days 9 and 13 with N-803 and/or αPD-L1 on days 9, 11, and 13. Serum was obtained at days 14 and 21 post-tumor implant and analyzed for level of IFNγ (a), TNFα (b), IL-10 (c), and IL-6 (d). Level of serum cytokines from individual animals at day 14 (left panels) and curves showing kinetics of serum cytokine levels (right panels) show mean ± SD. Data are combined from 2 experiments, n = 4–5 mice/group per experiment