Bifunctional TGF-β trap/IL-15 protein complex elicits potent NK cell and CD8⁺ T cell immunity against solid tumors

Abstract

Advances in immunostimulatory and anti-immunosuppressive therapeutics have revolutionized cancer treatment. However, novel immunotherapeutics with these dual functions are not frequently reported. Here we describe the creation of a heterodimeric bifunctional fusion molecule, HCW9218, constructed using our soluble tissue factor (TF)-based scaffold technology. This complex comprises extracellular domains of the human transforming growth factor-β (TGF-β) receptor II and a human interleukin-15 (IL-15)/IL-15 receptor α complex. HCW9218 can be readily expressed in CHO cells and purified using antibody-based affinity chromatography in a large-scale manufacturing setting. HCW9218 potently activates mouse natural killer (NK) cells and CD8⁺ T cells in vitro and in vivo to enhance cell proliferation, metabolism, and antitumor cytotoxic activities. Similarly, human immune cells become activated with increased cytotoxicity following incubation with HCW9218. This fusion complex also exhibits TGF-β neutralizing activity in vitro and sequesters plasma TGF-β in vivo. In a syngeneic B16F10 melanoma model, HCW9218 displayed strong antitumor activity mediated by NK cells and CD8⁺ T cells and increased their infiltration into tumors. Repeat-dose subcutaneous administration of HCW9218 was well tolerated by mice, with a half-life sufficient to provide long-lasting biological activity. Thus, HCW9218 may serve as a novel therapeutic to simultaneously provide immunostimulation and lessen immunosuppression associated with tumors.

Graphical abstract

A bifunctional fusion protein, HCW9218, was created using tissue factor-based scaffold technology. HCW9218 comprises the extracellular domains of human transforming growth factor-β receptor II and human interleukin-15 (IL-15)/IL-15 receptor α. This fusion protein represents a potent anti-cancer immunotherapeutic to simultaneously provide immune stimulation and lessen immunosuppressive activities associated with tumors.

Introduction

Immunotherapeutics, either as immunostimulants or inhibitors of tumor-induced immunosuppression, are promising anti-neoplastic agents and have revolutionized cancer therapy.1, 2, 3 Additionally, combination approaches with these agents have been demonstrated to additively or synergistically enhance antitumor activities.^4,5 Bifunctional fusion protein approaches, which eliminate the need for complicated multi-agent treatment regimens, are also gaining considerable interest.⁶ We recently reported a soluble tissue factor (TF)-based protein scaffold used to create a heterodimeric multi-functional fusion complex comprising cytokines, interleukin-12 (IL-12), IL-18, and IL-15.⁷ The fusion complex was able to generate cytokine-induced memory-like natural killer (NK) cells for adoptive cell therapies and is currently in clinical development for relapsed/refractory acute myeloid leukemia. We postulated that we could employ this protein scaffold to develop a bifunctional fusion protein complex to simultaneously provide immunostimulation while inhibiting immunosuppression caused by tumors.

IL-15 is a critical factor for the development, proliferation, and activation of NK cells and CD8⁺ T cells.^8,9 IL-15 is co-expressed with its α chain receptor (IL-15Rα) by antigen-presenting cells, and the two proteins form a complex on the cell surface, which is transpresented to NK cells and T cells bearing the IL-2 receptor βγ_c complex.⁹ We previously reported the construction of an IL-15 superagonist complex, ALT-803, using an IL-15 variant (N72D) and the extracellular IL-15Rα sushi domain (IL-15RαSu) fused to immunoglobulin G1 (IgG1) Fc.¹⁰ ALT-803 is a potent immunostimulant that is capable of activating the innate and adaptive immunity against tumors and infectious agents.11, 12, 13, 14, 15 ALT-803 is well tolerated and provides clinical benefit to cancer patients either alone or in combination with other therapies.^4,16,17 Thus, IL-15/IL-15RαSu was selected as the immunostimulatory component of our fusion complex.

Transforming growth factor-β (TGF-β) isoforms are members of an evolutionarily conserved cytokine superfamily that mediates a diverse range of signaling functions to provide tissue-specific control of cell differentiation and proliferation. They also promote or protect against cell death, promote extracellular matrix protein expression, cell motility, and invasion, and control cell metabolism.18, 19, 20 TGF-β1 triggers epithelial-mesenchymal transitions (EMT) through induction of the expression of specific transcription factors.²¹ EMT provides migratory and invasive behaviors to the cells due to cell adhesion modifications, and this process involves a loss of epithelial features and acquisition of features leading to motility and invasive properties. EMT represents an important process leading to the progression and metastasis of cancer cells.²¹ As an immunosuppressive cytokine, TGF-β1 also inhibits the function and development of innate and adaptive immune systems including macrophages, NK cells, dendritic cells, and T cells. TGF-β1 also stimulates regulatory T cells, which suppress the function of other lymphocytes.22, 23, 24 Overall, TGF-β1 produced by tumor cells, tumor-associated immune cells, and stroma cells is one of the most important agents promoting immunosuppression in the tumor microenvironment (TME). Thus, we targeted TGF-β neutralization and employed the extracellular domain of human TGF-β receptor II (TGFβRII) as a component of our fusion protein complex, which binds and sequesters TGF-β.

Herein, we detail the construction, purification, and characterization of the heterodimeric bifunctional fusion complex HCW9218, comprising soluble TGF-βRII and IL-15/IL-15RαSu domains. Our data support its clinical development as a novel immunostimulant and inhibitor of immunosuppression for patients with cancer.

Results

Creation of bifunctional protein complexes containing TGF-β antagonist and IL-15 immunostimulatory domains

We previously reported the generation of novel multi-functional immunotherapeutics using a human TF-based protein scaffold.⁷ In this study, we genetically fused two human TGF-β receptor II extracellular domains (TGFβRII) in tandem to the extracellular domain of TF linked to human IL-15. We also genetically fused two tandem TGFβRII domains to the extracellular human IL-15Rα sushi domain (IL-15RαSu). When co-expressed in CHO cells, these fusion proteins formed a soluble heterodimeric complex (HCW9218, Figure 1A) that was purified using anti-TF antibody (Ab)-based affinity chromatography (see the Materials and methods). Based on reducing SDS-polyacrylamide gel electrophoresis (PAGE) analysis, purified HCW9218 comprises two polypeptides that migrated at ∼70 kDa and ∼40 kDa after deglycosylation, consistent with the expected molecular weights (2∗TGFβRII-TF-IL-15, 69 kDa; 2∗TGFβRII-IL-15RαSu, 39 kDa; Figure 1B, lane 2), whereas a ≥25 kDa increase in molecular mass was observed in the native glycosylated forms of these proteins (lane 1). Size exclusion chromatography (SEC) showed a MW of ∼500 kDa for the native form of HCW9218 and ∼370 kDa for the deglycosylated complex (Figure 1C), suggesting that HCW9218 is in a trimeric form. Previous studies have indicated that the TF and IL-15/IL-15Ra domains can self-dimerize, which may potentially contribute to the observed trimer.25, 26, 27 ELISAs verified that the IL-15 and TGFβRII domains of HCW9218 are intact (Figure S1). Characterization of the purity/impurity profile of HCW9218 generated from a 50-L engineering run in chemically defined media is detailed in Table S1. Thus, large-scale production of this fusion complex under GMP manufacturing is highly feasible to support future clinical development.

Figure 1.

Biochemical characteristics of HCW9218

(A) Cartoon models of the heterodimeric bifunctional fusion protein complex, HCW9218, comprising 2∗TGFβRII-TF-IL-15 and 2∗TGFβRII-IL-15RαSu polypeptides. (B) The polypeptides of HCW9218 are highly glycosylated. To examine the molecular weight characteristics, we ran deglycosylated and untreated protein samples on 4%–12% SDS-PAGE Bis-Tris gels under reducing conditions and stained with InstantBlue. Lane 1, non-deglycosylated HCW9218; Lane 2, deglycosylated HCW9218; Lane M, Mark12 unstained molecular weight markers. (C) High-performance liquid chromatography (HPLC)-SEC analysis of purified native and deglycosylated HCW9218 samples. Characteristics of purified HCW9218 generated in an engineering run (50 L) scale process are shown in Table S1 (also see Figure S1).

HCW9218 retains TGF-β-binding and IL-15 biological activities

The TGF-β growth factor is a disulfide-linked homodimer that interacts with dimeric TGF-β receptors.²⁸ Previous studies have shown that tandemly linked TGFβRII domains exhibited a 100- to 1,000-fold higher in vitro ligand binding and neutralization activity compared with monovalent TGFβRII and a similar TGF-β1 binding affinity as TGF-βRII-Fc fusions (TGFβRII-Fc), which contains two TGF-β receptor II extracellular domains.²⁷ Using a cell-based TGF-β reporter assay, we found that HCW9218 containing two TGFβRII dimers was also capable of neutralizing TGF-β1 activity (Figure 2A). Additionally, HCW9218 exhibited a 10-fold stronger TGF-β1 neutralizing activity compared to TGFβRII-Fc (IC₅₀ = 43 pM HCW9218 versus 470 pM TGFβRII-Fc). This enhanced activity appears to be due to increased valency of HCW9218 since similar fusion protein complexes containing a single TGFβRII dimer had an IC₅₀ of >700 pM in this assay (data not shown). HCW9228, an IL-15D8N mutant form of HCW9218 lacking IL-15 activity, also exhibited TGF-β1 neutralizing activity similar to HCW9218 (IC₅₀ = 45 pM), indicating that an active IL-15 domain was not required for TGF-β1 neutralizing activity.

Figure 2.

HCW9218 inhibits TGF-β activity and binds TGF-β and LAP

(A) Dose response curves of HCW9218, HCW9228, and TGFβRII-Fc showing concentration-dependent inhibition of human TGF-β1 activity on HEK-Blue TGF-β reporter cells. Data (mean ± SEM) are from duplicate measurements. (B) TGF-β1, -β2 and -β3 inhibited IL-4 dose-dependent stimulation of CTLL-2 cell proliferation. Addition of HCW9228 reduced the immunosuppressive activity of TGF-β1 and -β3 on IL-4-stimulated CTLL-2 cells. Data (mean ± SEM) are from at least duplicate measurements. (C) Binding of HCW9218, HCW9228, and TGFβRII-Fc to plate-immobilized SLC or CD39 (control) was determined following detection with anti-TGFβRII Ab. (D) Comparative binding is shown for plate-immobilized HCW9218, HCW9228, and TGFβRII-Fc to TGF-β1 or LAP detected with anti-TGF-β1 or anti-LAP Ab, respectively. Data (mean ± SEM) are from at least duplicate measurements.

TGF-β antagonist activity of the dimeric TGFβRII domains was also assessed using a CTLL-2 cell proliferation assay. Since IL-15 promotes the growth of CTLL-2 cells, HCW9228 was used in this study. Human TGF-β isoforms can inhibit IL-4-induced CTLL-2-cell growth in a concentration-dependent manner (Figure 2B). The suppressive activities of TGF-β1 and -β3, but not TGF-β2, were effectively reversed by HCW9228 (Figure 2B). Thus, collectively, these studies demonstrated that the functionality of the TGFβRII domains of HCW9228 are fully retained with greater antagonist activity for TGF-β1 and -β3 isoforms than for TGF-β2.

Previous studies suggested that binding of the latency-associated peptide (LAP) to TGF-β to form the small latency complex (SLC) inhibited interactions with TGFβRII.²⁰ Surprisingly, HCW9218, HCW9228, and TGFβRII-Fc were found to bind to SLC, with HCW9218 and HCW9228 showing greater SLC binding affinity than TGFβRII-Fc (Figure 2C). HCW9218, HCW9228, and TGFβRII-Fc not only bound to human TGF-β1 but also to LAP alone (Figure 2D), indicating interactions with both components of the SLC.

HCW9218 exhibited IL-15 activity by supporting proliferation of the IL-15-dependent 32Dβ cell line, although this activity was significantly lower than that of IL-15 (Figure 3A). This is consistent with our previous observations for other protein domain fusions to the N terminus of IL-15.²⁹ Incubation of human peripheral blood mononuclear cells (PBMCs; Figure 3B) or purified NK cells (Figure S2A) with HCW9218 resulted in a dose-dependent increase in granzyme B expression, as well as enhanced cytotoxicity against human K562 leukemia cells. HCW9218 also promoted release of interferon-γ (IFN-γ) in the culture media. HCW9228 elicited similar, though reduced, immune stimulation (Figure 3B; Figure S2A), presumably through inhibiting endogenous TGF-β activity. This was verified by addition of 10 ng/mL TGF-β1 to the cultures resulting in further suppression of immune responses, which were reversed in a dose-dependent manner by HCW9218 or HCW9228 (Figure 3B). Incubation of mouse splenocytes with HCW9218 or HCW9228 also resulted in an increase in IFN-γ release and elevated cell surface activation markers (granzyme B, CD69) in NK cells (Figure 3C) and CD8⁺ T cells (Figure S2B). In addition to stimulating immune effector cell cytotoxicity against target cells, IL-15 is known to enhance NK cell-mediated antibody-dependent cellular cytotoxicity (ADCC) against tumors.^30,31 Similarly, HCW9218 was found to significantly enhance human NK cell cytotoxicity and rituximab-directed ADCC against human Daudi B-lymphoma cells (Figure 3D).

Figure 3.

HCW9218 stimulates human and mouse immune cells in vitro

(A) Dose response curves of HCW9218, HCW9228, and IL-15 showing concentration-dependent activity on proliferation of 32Dβ cells. Data (mean ± SEM) are from duplicates. (B) Stimulation of human PBMCs (n = 6 donors) was determined following 3-day incubation in increasing concentrations of HCW9218 and HCW9228 in the absence or presence of 10 ng/mL human TGF-β1. Cytotoxicity was measured based on a 4 h killing assay using K562 target cells (E:T = 20:1). Expression of granzyme B by stimulated PBMCs and levels of IFN-γ released into the culture supernatant (based on mean fluorescence intensity [MFI]) is shown based on flow cytometry. (C) Expression of granzyme B and CD69 of mouse splenic NK cells (n = 3/group) is shown based on flow cytometry following 4-day incubation with HCW9218 or HCW9228. Release of IFN-γ into the mouse splenocyte culture supernatant after 2-day incubation was also assessed. (D) Cytotoxicity of human NK cells against Daudi tumor cells in the presence or absence of 10 nM rituximab and/or 50 nM HCW9218 was measured based on a 20 h killing assay using the indicated E:T ratios. Data represent mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 versus control based on one-way ANOVA (also see Figure S2).

Human TF is an integral membrane protein, which requires an appropriate position on the surface of a phospholipid bilayer for its procoagulant activity.^32,33 The TF domain of HCW9218 consists of the extracellular domain that is not expected to have procoagulant activity.³² This was confirmed in prothrombin time (PT) assays using soluble and cell membrane-bound HCW9218 at concentrations up to 620 nM (Table 1). This was also consistent with the absence of changes in hemostasis in mice when HCW9218 was administered at up to 200 mg/kg (Tables S2 and S3).

Table 1.

TF activity of HCW9218

Condition	PT time (s)	Relative TF activity (%)
6.3 nM Lipidated recombinant TF	8.5	100
0.06 nM Lipidated recombinant TF	25	1
PT buffer	181	0
1 × 106/mL CTLL-2 cells	162	7e−5
62 nM HCW9218	179	5e−6
185 nM HCW9218	168	4e−5
620 nM HCW9218	166	5e−5
185 nM HCW9218 + 1 × 106/mL CTLL-2 cells	155	4e−5

Taken together, these findings demonstrated that HCW9218 provides optimal immune responses through the combination of TGF-β binding and antagonistic activity and the immunostimulatory activity of IL-15.

Toxicity and pharmacokinetics of subcutaneous administration of HCW9218 in mice

To assess the dose-dependent toxicological effects of HCW9218, we administered female C57BL/6 mice (n = 3/group) one or two (every 2 weeks) subcutaneous doses of PBS or HCW9218 at 3, 10, 50, and 200 mg/kg (Figure 4A). Animals were monitored for signs of study drug-related toxicities, changes in body weight during the study period, and hematology and serum chemistry parameters at day 7 post-dosing. Mice receiving 200 mg/kg HCW9218 exhibited significant body weight loss beginning 4 days after the first injection (study day [SD] 0) and reached a nadir between SD6–9 before returning to pre-dose levels by SD11 (Figure 4B). Mortality was observed in one mouse of the 200 mg/kg group on SD9. There were no apparent treatment-mediated effects on body weight or other clinical signs in any other dose group or after the second HCW9218 dose at 200 mg/kg. Spleen weights increased in a dose-dependent manner following one or two doses of HCW9218 (Figure 4C). Compared to the PBS group, mice also exhibited a 25-fold increase in White blood cell (WBC) counts 7 days after a single 200 mg/kg dose of HCW9218, which remained 5-fold higher 7 days after the second 200 mg/kg dose (Figure 4D; Tables S2 and S3). WBC subset analysis showed a 16-fold increase in absolute lymphocyte counts and >50-fold increase in neutrophil, monocyte, eosinophil, and basophil counts at SD7 in the 200 mg/kg group. These changes were not observed at lower HCW9218 dose levels but were similar to those reported for C57BL/6 mice treated subcutaneously with IL-15/IL-15Rα complexes.³⁴ Other hematology and serum chemistry parameters were similar in the HCW9218 and PBS-treated animals and were generally within expected ranges for C57BL/6 mice (Tables S2 and S3). HCW9218-mediated effects were greatest 7 days after the first dose and were reduced after the second dose, consistent with previous studies showing decreased immune responses in mice following repeat dosing with IL-15/IL-15Rα.^35,36 Overall, HCW9218 was well tolerated by C57BL/6 mice at dose levels up to 50 mg/kg.

Figure 4.

Repeat dose subcutaneous administration of HCW9218 is well-tolerated by C57BL/6 mice

(A) Female C57BL/6 mice (3/group) were injected with PBS or increasing dose levels of HCW9218 on study day (SD)0 or on SD0 and SD14. Mice were subsequently sacrificed on SD7 (single dose) and SD21 (two doses) for toxicological assessments. (B) Changes in body weights are shown through SD21. (C and D) Spleen weights (C) and blood cell counts and differentials (D) are indicated for mice at SD7 after one dose and SD21 after two doses of HCW9218. Data represent mean ± SEM. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 versus control based on one-way ANOVA. Additional details on hematologic and serum chemistry are shown in Tables S2 and S3.

Pharmacokinetic (PK) analysis of 3 mg/kg HCW9218 administered subcutaneously to C57BL/6 mice showed a maximal serum concentration of 92 μg/mL 7 h after dosing and a terminal serum half-life of 18 h (Table S4).

Activity of HCW9218 in mice

Based on the toxicity profile, we further focused on in vivo characterization of HCW9218 in C57BL/6 mice at dose levels up to 10 mg/kg. Subcutaneously administered HCW9218 at as little as 3 mg/kg resulted in increased spleen weight (Figure S3A) and expansion of CD8⁺ T cells and NK cells in the spleen (Figure 5A) and blood (Figure S3B) of mice 4 days post-treatment. These changes were not observed with HCW9228, indicating HCW9218 immune stimulation was mediated by its IL-15 domain. Time course studies showed increased spleen weights (Figure S3A) and splenic CD8⁺ T cell and NK cell percentages (Figure 5B) over 92 h post-HCW9218 treatment when compared to PBS controls. HCW9218 promoted CD8⁺ T cell and NK cell proliferation in vivo based on upregulation of the Ki67 proliferation marker starting 24 to 48 h post-treatment (Figure 5B). Consistent with its in vitro activity, HCW9218 administration also induced expression of granzyme B and CD69 on mouse CD8⁺ T cells and NK cells and increased immune cell cytotoxic activity against NK-sensitive Yac-1 target cells (Figure 5B; Figure S3B). HCW9218 treatment also elevated CD44 expression on both CD4⁺ and CD8⁺ T cells and at 50 mg/kg, induced activation and proliferation of CD4⁺ T cells in mice (Figure S3B). HCW9218 was found to significantly decrease plasma TGF-β1 and TGF-β2 levels in C57BL/6 mice 2 days after treatment (Figure 5C), consistent with the activity of the TGFβRII domains. The levels of these isoforms returned to baseline 7 days post-dosing. We were unable to assess treatment effects on TGF-β3 due to its low baseline levels (<60 pg/mL). Overall, these findings verify that HCW9218 treatment can increase immune cell proliferation, activation, and effector functions while sequestering TGF-β isoforms in vivo.

Figure 5.

HCW9218 induces CD8⁺ T cells and NK cells proliferation and activation and reduces TGF-β levels in mice

(A) Female C57BL/6 mice were injected subcutaneously with PBS or increasing dose levels of HCW9218 or HCW9228 and the percentages of splenic CD8⁺ T cells and NK cells were determined by flow cytometry on day 4 post-dosing. (B) Mice treated subcutaneously with 3 mg/kg HCW9218 were assessed through 92 h post-dosing for changes in splenic CD8⁺ T cell and NK cell percentage and expression of Ki67 (proliferation) and granzyme B (activation) markers by flow cytometry. Cytotoxicity of in vivo activated splenocytes was measured based on a killing assay using Yac-1 target cells (E:T = 10:1). (C) Plasma levels of TGF-β1 and TGF-β2 were determined in mice following subcutaneous treatment with PBS, 3 mg/kg HCW9218, or 3 mg/kg HCW9228. Data (mean ± SEM) are from at least duplicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 versus control based on one-way ANOVA (also see Figure S3).

Metabolism plays a key role in immune cell proliferation and activation. IL-15 has been shown to impact glycolysis and oxidative phosphorylation in human and mouse immune cells.^37,38 To assess treatment mediated effects on immune cell metabolism, we performed extracellular flux assays on splenocytes isolated from mice 4 days after HCW9218 or HCW9228 administration. As expected, HCW9218 and IL-15 increased the rate of glycolytic capacity (ECAR; Figure 6A and 6B) and mitochondrial respiratory capacity (OCR; Figure 6C and 6D) of the isolated splenocytes in a dose-level-dependent manner. Surprisingly, in vivo HCW9228 treatment also increased ECAR and OCR of splenocytes. This phenomenon was not observed when splenocytes from untreated C57BL/6 mice were incubated 4 days with HCW9228 in vitro. Only HCW9218 (but not HCW9228) was capable of increasing splenocyte ECAR and OCR in vitro at physiologically relevant concentrations (Figure S4). This suggests that both the IL-15 and TGFβRII domains of HCW9218 have a role in stimulating immune cell metabolism in vivo.

Figure 6.

Immune cell metabolism is enhanced following in vivo activation with HCW9218 and HCW9228

Metabolic parameters were examined using an XFe96 Extracellular Flux Analyzer on mouse splenocytes isolated 4 days after in vivo treatment with PBS or increasing dose levels of HCW9218, HCW9228, or IL15SA. (A) Representative splenocyte glycolysis stress test tracing with measurement of extracellular acidification rate (ECAR). (B) Summary data from (A) showing glycolytic capacity (measured after 2 μM oligomycin). (C) Representative donor mitochondrial respiration tracing of oxygen consumption rate (OCR). (D) Summary data from (C) showing basal respiration (without drugs) and spare respiratory capacity (after 0.5 μM rotenone). Data represent the mean ± SEM from 8–9 independent measurements. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus control based on one-way ANOVA (also see Figure S4).

Antitumor efficacy of HCW9218 against B16F10 melanoma in C57BL/6 mice

We used the syngeneic B16F10 melanoma tumor model in C57BL/6 mice to assess the antitumor efficacy of HCW9218 as a monotherapy. B16F10 cells were subcutaneously implanted into C57BL/6 mice followed by subcutaneous injection of PBS, HCW9218 (at 3 or 20 mg/kg), or HCW9228 (at 3 or 20 mg/kg) on the first and fourth days after tumor implantation. Tumor volume and animal survival (based on tumor volume < 4,000 mm³) were assessed. When compared through SD15 (i.e., prior to animal mortality), treatment with HCW9218 or HCW9228 at 20 mg/kg resulted in significantly slower tumor growth than was observed in the PBS-treated mice (Figure 7A; Figure S5A). Tumor-bearing mice treated with 20 mg/kg HCW9218 also showed prolonged survival when compared to the 3 mg/kg HCW9218 and PBS treatment groups (Figure 7B). These results indicate that HCW9218 and HCW9228 have antitumor activity against solid B16F10 melanoma tumors with the bifunctional HCW9218 complex exhibiting the greater efficacy. Thus, both the TGFβRII and IL-15/IL-15RαSu domains play a role in HCW9218-mediated activity against B16F10 tumors.

Figure 7.

HCW9218 is efficacious against B16F10 melanoma tumors in mice

(A and B) C57BL/6 mice (n = 16/group) were injected subcutaneously with 5 × 10⁵ murine B16F10 melanoma tumor cells followed by subcutaneous treatment with PBS or 3 or 20 mg/kg of HCW9218 or HCW9228 on day 1 and 4 post-tumor cell inoculation. Tumor volume of individual animals (A) and mouse survival (based on tumor volume < 4,000 mm³) (B) were assessed. HCW9218 treatment at 3 and 20 mg/kg significantly reduced tumor growth by day 15 post-tumor cell inoculation and at 20 mg/kg prolonged animal survival when compared to controls. (C and D) Mice (9/group) were intraperitoneally treated with anti-CD8, anti-NK, or anti-CD8⁺ and anti-NK Abs for 1 week to deplete immune cells prior to injection with B16F10 melanoma tumor cells as in (A). Tumor-bearing mice were then treated with PBS or 20 mg/kg HCW9218 on day 1 and 4 post-tumor cell inoculation. Tumor volume of individual animals (C) and mouse survival (based on tumor volume < 4,000 mm³; D) were assessed. (E and F) B16F10 tumor-bearing mice (5/group) were treated with PBS or 20 mg/kg of HCW9218 on day 1 and 7 post-tumor inoculation. On day 11 post tumor inoculation, tumors were collected and tumor-infiltrating NK1.1⁺ cells and CD8⁺ T cells were quantitated by flow cytometry. (E) Representative flow cytometry plots of TILs isolated from tumor-bearing mice treated with PBS or HCW9218. (F) NK1.1⁺ and CD8⁺ TILs in tumor-bearing mice following PBS and HCW9218 treatment. ∗p < 0.05, ∗∗∗p < 0.001 versus control based on log-rank (Mantel-Cox) test for survival and one-way ANOVA for TIL analysis (also see Figures S5 and S6).

HCW9218 treatment is capable of significantly increasing the number of NK and T cells in vivo. To determine whether these immune cells were responsible for HCW9218-mediated antitumor efficacy, we conducted Ab immunodepletion of CD8⁺ T cells and NK1.1⁺ cells in tumor-bearing mice prior to HCW9218 treatment. It was found that NK1.1⁺ cell depletion (alone or in combination with CD8⁺ T cell depletion) eliminated the antitumor effects of HCW9218 in B16F10 tumor-bearing mice during the first 2 weeks post-treatment (Figure 7C; Figure S5B), whereas either CD8⁺ T cell or NK1.1⁺ cell depletion reduced the survival benefit seen with HCW9218 (Figure 7D). Consistent with these findings, HCW9218 treatment also promoted an increase in NK cell and CD8⁺ T cell infiltration into B16F10 tumors (Figures 7E and 7F). These results support our conclusion that both CD8⁺ T cells and NK cells play a major role in HCW9218-mediated activity against melanoma tumor cells in C57BL/6 mice.

Discussion

In this study, we show that the heterodimeric bifunctional fusion molecule HCW9218 comprising soluble TGFβRII and IL-15/IL-15Rα domains is an immunotherapeutic agent with potent immunostimulatory and TGF-β antagonist activities. HCW9218 is capable of promoting proliferation and metabolic activity of NK cells and CD8⁺ T cells and enhancing their cytotoxicity with tumor targets. In the syngeneic B16F10 melanoma model, HCW9218, as a single agent, was effective in reducing tumor burden and prolonging animal survival. Mechanism of action studies demonstrated that this HCW9218-associated antitumor activity was mediated by NK cells and CD8⁺ T cells and correlated with increased immune cell infiltration into tumors. In addition to its use as a single agent, HCW9218 was found to enhance ADCC of therapeutic Abs against tumor targets due to its potent NK cell stimulatory activity.

We have previously found that fusions of Ab domains to the N terminus of IL-15 resulted in a 10- to 100-fold decrease in IL-15 biological activity.²⁹ This was also the case with the fusion of the TF domain to IL-15 in HCW9218. However, the reduction of IL-15 activity may be advantageous in allowing dosing at higher levels for effective TGF-β neutralization and immune cell stimulation without inducing unwanted immune-related systemic toxicity associated with high dose IL-15 or γ_c cytokines. Studies with an anti-PD-L1 Ab-TGFβRII dimer fusion protein (bintrafusp alfa) indicated that a single dose of 8 mg/kg in mice and 1 mg/kg in patients was sufficient to sequester TGF-β isoforms in the plasma for >7 days.^6,39 This is consistent with our finding that a 3 mg/kg dose of HCW9218 (comprising two TGFβRII dimers) is capable of reducing plasma TGF-β levels in mice. Thus, a potential clinical dose of HCW9218 at as low as 0.3 mg/kg may be sufficient to sequester plasma TGF-β in humans. Although the PK profile of HCW9218 in humans remains to be determined, the IL-15 activity of HCW9218 at this dose level is equivalent to 3 μg/kg of the IL-15 superagonist, ALT-803, which was well tolerated at 20 μg/kg in humans.^4,16,17 We found that HCW9218 was well tolerated by mice following repeat-dose treatment at ≤50 mg/kg. Notably, HCW9218 treatment did not result in cardiovascular toxicity, which has previously been observed in animal toxicity studies of pan-TGF-β Ab therapy.⁴⁰ This suggests that TGF-β neutralization using TGFβRII domains may provide a safer therapeutic moiety than Ab-based approaches. This is also consistent with the recent clinical experience of bintrafusp alfa in cancer patients.^39,41,42

TGF-β is initially translated as an inactive pre-pro-protein homodimer comprising a signal sequence, LAP, and growth factor domain. TGF-β is proteolytically cleaved from LAP in the Golgi complex, but these proteins remain non-covalently associated to form the SLC, which associates with the extracellular matrix or cell membrane proteins.^43,44 In this complex, TGF-β remains inactive since LAP directly covers amino acids of TGF-β that interact with TGF-β receptors.⁴⁴ We show here that HCW9218 independently binds LAP and TGF-β proteins, as well as SLC, indicating HCW9218 interacts with both latent and active forms of TGF-β. This may represent an efficient pathway to sequester TGF-β activity and to out compete membrane-associated receptors for TGF-β binding. Studies are underway to further elucidate the binding mode and dynamics of HCW9218 to SLC and free TGF-β and its effects on TGF-β activity.

It is well established that TGF-β plays a prominent immunosuppressive role in the TME to promote growth and metastasis of tumors. In particular, TGF-β1 contributes to pancreatic ductal adenocarcinoma (PDAC) desmoplasia by enhancing the conversion of fibroblasts or endothelial cells into myofibroblasts and promoting formation of dense extracellular matrix.^45,46 TGF-β also induces proangiogenic factors, such as vascular endothelial growth factor, to support tumor progression, invasion, and metastasis, as well as plasminogen activator inhibitor type 1, a key intermediate of tumor/stroma cross-talk mediating immunosuppression and fibrosis in PDAC.⁴⁷ Thus, the TGF-β neutralization activity of the TGFβRII domains of HCW9218 could simultaneously inhibit EMT and reduce immunosuppression and fibrosis in the PDAC TME. In combination with its ability to increase immune cell activation and infiltration into the tumors, HCW9218 has potential as an immunotherapeutic agent against PDAC and is being advanced to the clinical evaluation phase for this indication.

Material and methods

Mice

Female 6- to 8-week-old C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All animal husbandry procedures performed were approved by the Institutional Animal Care and Use Committee and were in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Mice were maintained in a temperature/humidity-controlled facility with 12-h light -dark cycles, group housed, and had free access to food and water. All mice were treated humanely throughout the experimental period.

Construction and production of bifunctional protein complexes

HCW9218 is a complex of two fusion proteins, one (2∗TGFβRII-TF-IL-15) comprising two TGF-β receptor II extracellular domains linked together with a flexible (G₄S)₃ sequence (2∗TGFβRII), the extracellular domain of human TF, and IL-15, and the other (2∗TGFβRII-IL-15RαSu) comprising the 2∗TGFβRII single-chain protein linked to the soluble sushi domain of IL-15Rα (IL-15RαSu; Figure 1A). The corresponding coding DNA sequences were synthesized (Genewiz, South Plainfield, NJ, USA), cloned into pMSGV-1 modified expression vectors and transfected into CHO.K1 cells (ATCC, Manassas, VA, USA).⁴⁸ Co-expression of the two polypeptides in CHO cells allows for formation of the protein complex via high-affinity interactions between the IL-15 and IL-15RαSu domains and secretion of the complexes into the culture media. HCW9218 expression was then detected with product-specific ELISAs (Figure S1). HCW9228, a derivative of HCW9218, was constructed through the same process using sequences encoding the IL-15 D8N mutation, which was previously described to eliminate binding to IL-2Rβ without affecting IL-15/IL-15Rα interactions.²⁹

Production cell banks for HCW9218 and HCW9228 were generated from stably transfected clonal cell lines following limited dilution cloning. Subsequent fusion protein complex production was conducted using fed-batch methods with chemically defined media in shake flasks or stirred tank bioreactors. HCW9218 and HCW9228 were purified from clarified culture media using immunoaffinity chromatography with anti-TF Ab-conjugated Sepharose resin, and then buffer-exchanged into PBS.⁴⁸ A GMP-suitable manufacturing process (scaled from 2 L to 200 L) was developed for HCW9218, consisting of immunoaffinity chromatography, low pH viral inactivation/depth filtration, multimodal chromatography, nanofiltration, and ultrafiltration/diafiltration steps employing commercially scalable methods. Initial characterization of HCW9218 included deglycosylation using Protein Deglycosylation Mix II (New England Biolabs, Ipswich, MA, USA) followed by evaluation of native and deglycosylated complexes by SDS-PAGE on a 4%–12% Bis-Tris gel (ThermoFisher Scientific, Eugene, OR, USA) and SEC on a TSKgel G3000SWxl column (Tosoh Bioscience, Tokyo, Japan). The purified HCW9218 product was further characterized and released using qualified test methods per established specifications.

An IL-15 superagonist complex reagent (IL15SA) comprising IL-15N72D and IL-15RαSu-IgG1 Fc polypeptides was constructed based on published nucleotide sequences and produced by previously described methods.¹⁰

Binding assays

To evaluate production and PK, we captured HCW9218-containing samples on wells coated with anti-TF Ab and detected with biotinylated-anti-IL-15 or -anti-TGFβRII Abs (R&D Systems, Minneapolis, MN, USA) followed by horseradish peroxidase-conjugated streptavidin (HRP-SA; Jackson ImmunoResearch, West Grove, PA, USA). Comparative binding of HCW9218, HCW9228, and TGFβRII-Fc (R&D Systems) was conducted by capturing purified fusion proteins on small latent complex (SLC; Acro Biosystems, Newark, DE, USA) or CD39 (control) coated wells and detected with anti-TGFβRII Ab (R&D Systems). Alternatively, plate-immobilized HCW9218, HCW9228, and TGFβRII-Fc were used to capture TGF-β1 (BioLegend, San Diego, CA, USA) or LAP (R&D Systems) followed by detection with biotinylated-anti-TGF-β1 (R&D Systems) or -anti-LAP Ab (R&D Systems) respectively, followed by HRP-SA.

In vitro characterization

TGF-β1 activity is capable of inducing detectable signaling of HEK-Blue TGF-β cells (InvivoGen, San Diego, CA, USA). TGF-β1 antagonist activity of HCW9218, HCW9228, and TGFβRII-Fc was determined with HEK-Blue TGF-β cells incubated with 0.1 nM TGF-β1 per manufacturer recommendations (InvivoGen). IL-15 activity of HCW9218 and HCW9228 was determined based on IL-15-dependent proliferation of 32Dβ cells as described previously with recombinant IL-15 (PeproTech, Cranbury, NJ, USA) serving as a control.²⁹ Previous reports have shown the human TGF-β isoforms suppress IL-4-induced proliferation of CTLL-2 cells (ThermoFisher Scientific, Eugene, OR, USA). Since the IL-15 domain of HCW9218 promotes CTLL-2 proliferation, TGF-β antagonist activity of the HCW9228 TGFβRII domain was assessed in CTLL-2 cells incubated in increasing concentrations of IL-4 and TGF-β1 (BioLegend), -β2, or -β3 (R&D Systems; 5 ng/mL). HCW9228 was added at a 100-fold molar excess over TGF-β and after 4 days, CTLL-2 proliferation was assessed using PrestoBlue cell viability reagent (ThermoFisher Scientific, Eugene, OR, USA). Procoagulant activity of HCW9218 was assessed based on clotting of normal human plasma (ThermoFisher Scientific) in a PT test measured with a STart PT analyzer (Diagnostica Stago, Parsippany, NJ, USA). Lipidated recombinant TF (Innovin; AAA Wholesale Company, San Francisco, CA, USA) was used as a reference and the percentage of TF activity was determined based on a standard curve of log (TF concentration) versus log (PT time).

Fresh human PBMCs were isolated from buffy coats of healthy donor blood (OneBlood, St. Petersburg, FL, USA) and NK cells were purified using RosetteSep/human NK cell reagent (StemCell Technologies, Cambridge, MA, USA). Mouse splenocytes were obtained from the spleens of healthy C57BL/6 mice. The immune cells were cultured in RPMI-1640 supplemented with 2 mM L-glutamine, antibiotics (penicillin, 100 U/mL; streptomycin, 100 μg/mL; ThermoFisher Scientific), and 10% (v/v) fetal bovine serum (ThermoFisher Scientific) in the presence or absence of HCW9218 or HCW9228 and/or TGF-β1 at concentrations indicated in the figure legends. After 3 days for human PBMCs and NK cells or 2 to 4 days for mouse splenocytes, cells were harvested and stained with fluorochrome-conjugated anti-CD4, -CD8, -NK, and -CD19 Abs (BioLegend). The percentage of CD4⁺ T cells, CD8⁺ T cells, NK cells, and CD19⁺ B cells was analyzed by flow cytometry (FACSCelesta, BD BioScience, San Jose, CA, USA). The cytotoxic potential based on granzyme B expression and activation based on CD69 expression were evaluated by direct or intracellular staining with fluorochrome-conjugated anti-granzyme B and anti-CD69 Abs (BioLegend). The mean fluorescent intensity (MFI) of the stained cells was analyzed by flow cytometry. Supernatants from immune cell cultures were also harvested and IFN-γ levels (based on MFI) were measured with LEGENDplex assay per manufacturer’s instructions (BioLegend). Cytotoxicity of HCW9218- or HCW9228-activated human PBMCs and NK cells against human K562 chronic myelogenous leukemia cells (ATCC) was assessed by mixing immune cells with tumor target cells labeled with CellTrace Violet (ThermoFisher Scientific) at an E:T ratio of 20:1. After a 4 h incubation, target cell viability was assessed by analysis of propidium iodide positive, CellTrace violet-labeled cells using flow cytometry. Similar methods were used to assay human NK cell cytotoxic activity against CellTrace violet-labeled human CD20⁺ Daudi B-lymphoma cells (ATCC) following 20 h incubation in media with or without 10 nM rituximab (Genentech, South San Francisco, CA, USA) and/or 50 nM HCW9218.

Immunostimulatory activities in vivo

To evaluate the immunostimulatory activities in vivo, we subcutaneously injected C57BL/6 mice with PBS, HCW9218, or HCW9228. At the times listed in the figure legends, mice were humanely euthanized. Spleens were collected, weighed, and used to prepare splenocytes, and blood was collected for immune cell staining and to prepare plasma. Splenocytes and blood cells were then stained for immune cell subsets (CD4, CD8, NK1.1), as well as CD69 (activation), CD44 (memory), granzyme B (cytotoxicity), and Ki67 (proliferation marker) using fluorochrome-conjugated Abs (BioLegend). The percentage of immune cell subsets and MFI of CD69, CD44, granzyme B, and Ki67 staining were assessed by flow cytometry. Levels of TGF-β isoforms in the plasma were determined using the TGFβ 3-Plex assay (Eve Technologies, Calgary, AL, Canada). Cytotoxicity of in vivo HCW9218-activated splenocytes was measured based on a 20-h killing assay using Yac-1 target cells (E:T = 10:1) (ATCC) as described above.

The metabolic activities of splenocytes activated in vivo were evaluated. 4 days after subcutaneous treatment with PBS or increasing dose levels of HCW9218, HCW9228, or IL15SA. Splenocytes from C57BL/6 mice were isolated and seeded in Cell-Tak-coated Seahorse Bioanalyzer XFe96 culture plates in Seahorse XF RPMI medium, pH 7.4 (Agilent Technologies, Santa Clara, CA, USA) supplemented with 2 mM L-glutamine for the glycolysis stress test. For mitochondrial stress test, cells were seeded in Seahorse XF RPMI medium, pH 7.4 supplemented with 10 mM glucose and 2 mM L-glutamine. Assays were performed following the manufacturer’s instruction (Agilent Technologies). The data were analyzed using Wave software (Agilent Technologies). As a comparative control, splenocytes isolated from untreated mice were stimulated in vitro overnight with HCW9218, HCW9228, or IL15SA and then evaluated for glycolytic and mitochondrial functions. Key parameters measured for glycolytic function include non-glycolytic acidification (without added agents), glycolysis (after glucose), maximal glycolysis induction/glycolytic capacity (after oligomycin), and glycolysis reserve (after 2-DG). Parameters assessed for mitochondrial respiration include basal respiration (without added agents), ATP-linked respiration/proton leak (after oligomycin), maximal respiration (after FCCP), and spare respiration (after rotenone).

Toxicity and PK parameters of HCW9218 in mice

A toxicology study was conducted in female C57BL/6 mice (n = 3/group) where animals were administered one or two (every 2 weeks) subcutaneous doses of PBS or HCW9218 at 3, 10, 50, or 200 mg/kg. Animals were monitored for signs of study drug-related toxicities and changes in body weight during the study period. At day 7 post dosing, animals were sacrificed and spleens were weighed and blood taken for standard hematology and serum chemistry assessments (Comparative Pathology Laboratory, University of Miami, Miller School of Medicine, Miami, FL, USA). Blood was also collected on days 4 and 7 post-dosing and stained for lymphocyte subsets and proliferation and activation markers as described above.

PK analysis of HCW9218 was evaluated in C57BL/6 mice receiving a single 3 mg/kg subcutaneous dose. Blood was collected at various time points and HCW9218 concentration in serum was determined with ELISA as described above. PK parameters were determined based on mean HCW9218 serum concentrations using PK Solutions 2.0 non-compartmental extravascular methods (Summit Research Services, Montrose, CO, USA).

Murine B16F10 tumor model

To evaluate HCW9218 antitumor efficacy, we selected the murine B16F10 tumor model as it is highly aggressive, poorly immunogenic and devoid of immune infiltrates, expresses TGF-β, which plays a role in its growth, and is resistant to cytokine and checkpoint blockade immunotherapies.49, 50, 51 B16F10 melanoma cells (5 × 10⁵ cells; CRL-6475, ATCC) were subcutaneously injected into C57BL/6 mice followed by subcutaneous injection of PBS, HCW9218 (3 or 20 mg/kg), or HCW9228 (3 or 20 mg/kg) on day 1 and 4 after tumor implantation. Tumor volume was measured every other day and mice with tumors ≥4,000 mm³ were sacrificed per IACUC regulation. Mouse survival was also assessed throughout the study period.

To investigate the role of CD8⁺ T cells and NK cells in HCW9218 activity, we treated C57BL/6 mice intraperitoneally with three 500 μg doses of anti-CD8 (clone 53-6.7), anti-NK1.1 (clone PK136), or anti-CD8 plus anti-NK1.1 Abs (HCW Biologics, Miramar, FL, USA) at 1, 4, and 7 days prior to subcutaneous B16F10 melanoma cell injection. Mice were subsequently injected subcutaneously with PBS or HCW9218 (3 or 20 mg/kg) on day 1 and 4 after tumor implantation and were monitored as described above.

Tumor infiltrating lymphocytes (TILs) were also assessed in B16F10 tumor-bearing mice following subcutaneous treatment with PBS or 20 mg/kg HCW9218. Tumors (10 days post implantation) were collected and dissociated into single cell suspensions by collagenase type 1 digestion. TILs were purified by Ficoll-Paque density gradient centrifugation followed by lysis of RBCs in ACK buffer. TILs were stained with fluorochrome-conjugated anti-CD45, -CD3, -CD4, -CD8, and -NK1.1 Abs (BioLegend) and Live/Dead Fixable Violet stain (ThermoFisher Scientific). The percentage of CD45⁺CD3⁺CD8⁺ T cells and CD45⁺CD3^–NK1.1⁺ cells was analyzed by flow cytometry (FACSCelesta; Figure S6).

Statistical analysis

Data were analyzed using Prism 8 (GraphPad, San Diego, CA, USA) and presented as mean ± SEM. Specific tests used are noted in the figure legends. Overall, data were tested for normality and were analyzed via one-way ANOVA followed by Dunnett’s multiple comparison test, two-way ANOVA, or Dunn’s test. Survival curves were compared two at a time via Mantel-Cox analysis. p values < 0.05 were considered to indicate statistical significance.