Preclinical Characterization of HPN536, a Trispecific, T-Cell–Activating Protein Construct for the Treatment of Mesothelin-Expressing Solid Tumors

Redirection of cytotoxic T cells with bispecific antibody constructs for cancer therapy has been validated in the clinic (1–6). Blinatumomab is the first and thus far the only bispecific T-cell engager (BiTE) approved by the FDA (7). T-cell–engaging biologics function by forming an immunologic cytolytic synapse between cancer target cells and T cells, which leads to target cell lysis independent of T-cell receptor (TCR) specificity, peptide antigen presentation by HLA, and T-cell costimulation. Despite the clinical success of blinatumomab for treating relapsed and refractory acute lymphoblastic leukemia, other molecules, including BiTE antibodies, showed only limited activity in the treatment of solid tumors (8, 9). Their short plasma half-life required continuous intravenous infusion limiting their utility for most solid tumor indications. Novel designs for T-cell–engaging antibodies aim at overcoming limitations of the first generation and are already being tested in clinical trials (10).

The Trispecific T-cell–Activating Construct (TriTAC) design has been specifically developed to treat solid tumors (11). TriTACs consist of a single polypeptide chain aligning three humanized, antibody-derived binding domains: a single-domain antibody (sdAb) specific for a tumor antigen, a sdAb specific for serum albumin for half-life extension, and a single-chain fragment variable (scFv) specific for the CD3ϵ subunit of the TCR complex (11). Their molecular size of 53 kDa is about one-third of that of an IgG. Binding of TriTACs to tumor antigen and CD3ϵ is monovalent, which minimizes off-target CD3ϵ clustering that can potentially lead to nonspecific T-cell activation. The absence of an Fc-gamma domain for half-life extension is functionally compensated by an albumin-binding domain. HPN424 (11) and HPN536, the first two TriTACs are in phase I clinical testing in hormone refractory prostate cancer and mesothelin (MSLN)-overexpressing solid tumors, respectively.

Human MSLN is produced as a 71-kDa precursor of 628 amino acids, which is expressed as a glycophosphatidylinositol-linked cell surface glycoprotein. Its 31-kDa N-terminal domain is released as a soluble protein, termed as the megakaryocyte potentiating factor (MPF), while the 40-kDa C-terminal domain remains attached to the plasma membrane as mature MSLN (12–14). MSLN expression on normal tissue is confined to the single-cell mesothelial layer covering the surface of tissues and organs of the pleural, pericardial, and peritoneal cavities (13, 15). MUC16/CA125 is a binding partner for MSLN, implicating a role for MSLN in cell adhesion (16, 17). However, the precise physiologic functions of MSLN have not been defined, and MSLN-knockout mice exhibit no detectable phenotype or developmental abnormality (18).

MSLN is overexpressed in many malignancies, including ovarian cancer (13, 15, 19), pancreatic cancer (20, 21), non–small cell lung cancer (22–26), triple-negative breast cancer (26, 27), and mesothelioma (28, 29). In triple-negative breast cancer (25) and in lung and pancreatic adenocarcinomas (22, 23, 30), overexpression of MSLN correlates with poor prognosis. Differential expression of MSLN in cancer versus normal tissue has made it an attractive target for MSLN-directed imaging agents and therapeutics (10, 31–33). A challenge in developing MSLN-directed therapeutics is the expression of MSLN on normal mesothelial cells, potentially leading to dose-limiting toxicities.

HPN536 specifically redirects T cells for potent redirected lysis of MSLN-expressing cancer cells with concomitant T-cell activation. In three different mouse xenograft models, HPN536 induced durable antitumor activity at very low doses. In cynomolgus monkeys, HPN536 was well tolerated, showed a long serum half-life, and elicited signs of target engagement on mesothelial structures.

Sequences of TriTACs, sdAbs, and extracellular domains of target proteins fused to an Fc domain or a hexahistidine tag were cloned into mammalian expression vector, pcDNA 3.4 (Invitrogen), preceded by a leader sequence. Expi293 Cells (Life Technologies) were maintained in suspension in Optimum Growth Flasks (Thomson) between 0.2 and 8 × 10⁶ cells/mL in Expi293 media. Purified plasmid DNA was transfected into Expi293 cells in accordance with Expi293 Expression System Kit (Life Technologies) protocols and cultured for 4–6 days after transfection. Alternatively, HPN536 was produced in CHO-DG44 DHFR–deficient cells (34). The amount of expressed proteins in conditioned media was quantitated using an Octet RED96 instrument with Protein A Tips (ForteBio/Pall) using appropriate purified control proteins for a standard curve. Conditioned media from either host cell were filtered and purified by protein A affinity and desalted or subjected to preparative size exclusion chromatography (SEC) using an AKTA Pure Chromatography System (GE Healthcare). Protein A purified TriTAC proteins were further purified by ion exchange and formulated in a buffered solution containing excipients. Final purity was assessed by SDS-PAGE by resolving 2.5 μg/lane on TRIS-Glycine gels and visualized with Simply Blue Stain (Life Technologies). Native purity was also assessed by analytic SEC using a Yarra SEC150 3 μm 4.6 × 150 mm Column (Phenomenex) resolved in an aqueous/organic mobile phase buffered at neutral pH on a 1290 LC system and peaks were integrated with OpenLab ChemStation Software (Agilent Technologies).

Affinities of HPN536 analyte for albumin, CD3ϵ, and MSLN ligands were measured by biolayer interferometry using an Octet RED96 instrument with Streptavidin Tips (ForteBio/Pall). Experiments were performed at 27°C in PBS plus casein in the absence or presence of 15 mg/mL has, as described in Results section and figure legends. Binding sensograms generated from empirically determined ligand loads, appropriate serial dilutions of known analyte concentrations, and association and dissociation times were then fit globally to a one-to-one binding model using Octet DataAnalysis 9.0 software.

T cells from healthy donors were purified from leukopaks (leukapheresis samples, StemCell Technologies) using EasySep Human T Cell Isolation Kits (StemCell Technologies, 17951) following the manufacturer's instructions. All cancer cell lines were obtained from the ATCC, with the exception of OVCAR8 cells, which were obtained from the NCI (Bethesda, MD). Cell lines were passaged a maximum of 36 times after being received from the ATCC. Cell line authentication and Mycoplasma testing were not performed. T-cell–dependent cell cytotoxicity (TDCC) assays were performed as described previously (35). Briefly, luciferase-expressing target cells and purified human T cells were seeded per well of a 384-well plate at a 10:1 T cell-to-target cell ratio. Target cell killing was assessed following incubation for 48 hours at 37ºC and 5% CO₂. Target cell viability was assessed by incubation with the SteadyGlo Reagent (Promega). Luminescence was measured using a PerkinElmer EnVision Detection System. Activated T cells were identified by CD69 and CD25 surface expression (BD Biosciences). Samples were analyzed on a FACSCelesta Flow Cytometer (BD Biosciences). Flow cytometry data were processed using FlowJo v10 Software (FlowJo, LLC).

Cultured cells were incubated with 1 μg/mL HPN536 or anti-GFP TriTAC (control) for 1 hour. Binding was detected using Alexa647-anti-TriTAC antibody using a FACSCelesta Flow Cytometer (BD Biosciences). The QIFIKIT (Dako) was used according to the manufacturer's instructions to estimate the number of MSLN molecules expressed per cell.

To measure the cytokines, AlphaLISA Kits were used (PerkinElmer) per the manufacturer's instructions, except that the assays were performed in 384-well plates instead 96-well plates. Plates containing conditioned media from TDCC assays were used for analysis. Plates were read on a PerkinElmer EnVision Plate Reader equipped with an AlphaLISA module.

All mouse studies were performed in accordance with the policies of the Institutional Animal Care and Use Committee (IACUC) at Harpoon Therapeutics and Charles River Laboratories. For TOV21G and HPAFII experiments, NCG (NOD-Prkdc^em26Cd52Il2rg^em26Cd22/NjuCrl) mice received subcutaneous coimplants of human cancer cells (5 × 10⁶) and human T cells (5 × 10⁶) in 50% Matrigel (BD Biosciences) on day 0. Human T cells were expanded before implantation using Human T Cell Activation/Expansion Kit (Miltenyi Biotec) according to the manufacturer's instructions. Mice were dosed on days 1–15 (HPAFII, Fig. 4A and TOV21G, Fig. 4C) or days 7–16 (HPAFII, Fig. 4B) via intraperitoneal injection. For NCI-H292 experiments, NCG mice received subcutaneous coimplants of human cancer cells (1 × 10⁷) and human peripheral blood mononuclear cells (PBMC; 1 × 10⁷). Mice were administered HPN536 daily for 10 days starting on day 6 via intravenous injection. Tumor size was measured twice weekly and calculated using the following formula: tumor volume (mm³) = (w² × l)/2. Percent tumor growth inhibition (%TGI) was defined as the difference between the mean tumor volume (MTV) of the control group and the MTV of the treated group, expressed as a percentage of the MTV of the control group.

The pharmacology, pharmacokinetics, and toxicity of HPN536 were evaluated after a single intravenous bolus dose of 0.1, 1.0, or 10 mg/kg HPN536 in one male and one female cynomolgus monkey per group followed by either a 1- or 3-week postdose recovery period. The study followed the protocol and standard operating procedures of the testing facility (Charles River Laboratories) and was approved by their IACUC. Pharmacologic activity was evaluated by clinical observations, cytokine assessments, flow cytometry, and evidence of target engagement by histology. Two research electrochemiluminescence assays, a functional assay and an anti-idiotype assay, were used for measuring HPN536 levels in serum. For the functional assay, HPN536 was captured with biotinylated CD3ϵ and was detected with a sulfo-tagged MSLN. For the anti-idiotype assay, HPN536 was captured with an anti-idiotype antibody recognizing the anti-albumin domain and was detected with a sulfo-tagged CD3ϵ. Toxicokinetic parameters were estimated using Phoenix WinNonlin pharmacokinetic software. A noncompartmental approach, consistent with the intravenous bolus route of administration, was used for parameter estimation. Toxicity endpoints included daily morbidity and mortality, daily clinical observations, weekly body weights, daily food consumption, clinical pathology (hematology, clinical chemistry, and coagulation), and anatomic pathology (gross necropsy, organ weights, and histopathology).

Recombinant HPN536 has a molecular weight of approximately 53 kDa. A humanized llama sdAb specific for human MSLN is placed at its N-terminus (Fig. 1A). A humanized llama sdAb specific for human serum albumin (HSA) is placed in the middle of the molecule. The C-terminal end contains a humanized scFv specific for the human CD3ϵ subunit of the TCR complex. GGGGSGGGS linkers connect the three binding domains.

HPN536 is produced by eukaryotic cell culture and secreted as a single, nonglycosylated polypeptide. Stability studies subjecting HPN536 to various stress conditions, including multiple freeze thaw cycles and storage at 4°C and 40°C for 2 weeks, suggest the protein is stable and stress resistant (Supplementary Fig. S1). The high stability of HPN536 ensures limited aggregation, which would otherwise lead to CD3 clustering and nonspecific T-cell activation in the absence of target cells, and potential off-target side effects.

Biolayer interferometry analysis demonstrated that HPN536 binds with high affinity to recombinant or purified human and cynomolgus monkey MSLN, CD3ϵ, and albumin in the range of 0.21–6.7 nmol/L (Fig. 1B). The K_D values for binding to human and cynomolgus monkey orthologs were found to be within 6-fold of each other for all three target antigens. HPN536 bound to mouse MSLN and albumin with equilibrium binding constants of 210 and 170 nmol/L, respectively, but did not detectably bind mouse CD3ϵ (Fig. 1B). As demonstrated by flow cytometry analysis, HPN536 exhibited surface binding to human MSLN-expressing OVCAR cells and purified human T cells, confirming that HPN536 recognized native MSLN and CD3ϵ expressed on cells (Fig. 1C).

The in vitro potency of HPN536 was evaluated in a TDCC assay. Primary resting human T cells were combined with target cells at a ratio of 10:1, incubated for 48 hours, and the viability of target cells was determined. In cocultures of MSLN-expressing OVCAR3 ovarian cancer cells and resting T cells from five different donors, HPN536 efficiently directed T cells to lyse OVCAR3 target cells with EC₅₀ values ranging from 1.3 to 2.5 pmol/L (Fig. 2A; Supplementary Table S1). No lysis of OVACR3 cells was observed with a control TriTAC specific for GFP, showing that sole binding to CD3ϵ on T cells was not sufficient to mediate cytotoxicity. HPN536 also redirected T cells for lysis of other ovarian cancer cell lines expressing MSLN, including Caov3, Coav4, and OVCAR8, but it was inactive against MSLN-negative cell lines MDAPCa2b and NCI-510A, demonstrating strict specificity for MSLN (Fig. 2B; Supplementary Table S1). Because HPN536 contains an anti-albumin binder for half-life extension, TDCC assay was conducted in the presence of physiologic levels of serum albumin. Serum albumin did not inhibit the ability of HPN536 to redirect human T cells to kill MSLN-expressing target cells and had a minimal impact on potency (Supplementary Fig. S2).

Lysis by HPN536-engaged T cells was further explored with MSLN-expressing tumor cell lines of different histologic origins, including three pancreatic (Hs766T, Capan-2, and HPAFII; Fig. 2C; Table 1), three non–small cell lung cancer (NCI-H596, NCI-H292, and NCI-H1563; Fig. 2D; Table 1), and two mesothelioma cell lines (NCI-H2052 and NCI-H2452; Table 1). EC₅₀ values of HPN536 for redirected lysis ranged from 2.3 to 15 pmol/L across the various cell lines. HEK 293 cells, genetically engineered to express either human or cynomolgus monkey MSLN, were killed with EC₅₀ values of 0.9 and 0.7 pmol/L, respectively. PBMCs from cynomolgus monkeys were also able to kill MSLN-expressing tumor cells in the presence of HPN536 (Supplementary Fig. S3). These data support the use of cynomolgus monkey as a relevant species for toxicology studies with HPN536. Table 1 summarizes the EC₅₀ values across all cell lines of different histopathologic origins, as well as an estimate of the number of MSLN molecules expressed per cell as determined by flow cytometry.

T-cell activation by HPN536 was first assessed by induction of CD69 and CD25 surface expression on T cells. T cells from four normal donors were cocultured with the MSLN-positive cancer cell line, OVCAR8, at various concentrations of HPN536. Within 48 hours, HPN536 mediated a dose-dependent increase in the percentage of CD69- and CD25-positive T cells in coculture with MSLN-positive OVCAR8 cells (Fig. 3A and B). A GFP-specific TriTAC control had no effect on new CD69 or CD25 expression on T cells. Depending on the T-cell donor, between 5% and 20% of the T cells in culture upregulated CD69 and CD25 surface expression. Similar effects on T-cell activation were observed in coculture with MSLN-positive tumor cell line, Caov3 (Supplementary Fig. S4). The EC₅₀ values of HPN536 for upregulation of CD69 and CD25 in the presence of MSLN-positive cancer cells ranged from 0.14 to 9.0 pmol/L (Supplementary Table S2).

T-cell activation was also assessed by the release of inflammatory cytokines by T cells that were cocultured with MSLN-expressing OVCAR8 cancer cells. As shown in Fig. 3C and D, HPN536 mediated a dose-dependent secretion of TNFα and IFNγ into coculture media. A GFP-specific control TriTAC did not cause cytokine release by T cells. Comparable results were observed with Coav3, Caov4, and OVCAR3 as target cells and with multiple T-cell donors (Supplementary Tables S3 and S4). Of note, induction of CD69 and CD25 and release of cytokines by T cells were not observed in coculture with MSLN-negative tumor cell line, MDAPCa2b (Supplementary Fig. S5).

The antitumor activity of HPN536 was assessed in three xenograft mouse models established from cancer types that express MSLN: HPAFII (pancreatic cancer), TOV21G (ovarian cancer), and NCI-H292 (non–small cell lung cancer). A daily dosing regimen was selected on the basis of the reported half-life for mouse serum albumin of approximately 1 day (36, 37). Moreover, HPN536 bound mouse serum albumin with a lower affinity than HSA (K_D values of 170 vs. 6.3 nmol/L, respectively). Consistent with this, the serum half-life of HPN536 in mice was determined to be 24 hours (Supplementary Fig. S6; Supplementary Table S5). In a first study, HPAFII tumor cells were subcutaneously coimplanted with T cells from normal human donors at an effector-to-target cell ratio (E:T) of 1:1 (Fig. 4A). Daily doses of HPN536, 100, 20, or 4 μg/kg, were administered over 15 days by intraperitoneal injection starting 1 day after implantation. Doses of 20 and 100 μg/kg HPN536 caused HPAFII tumor eradication, while 4 μg/kg only slightly delayed tumor outgrowth (Fig. 4A) compared with a GFP-specific control TriTAC at 100 μg/kg. In a second experiment, HPAFII cancer cells and T cells from a healthy donor at an E:T ratio of 1:1 were coimplanted under the skin, and tumors were allowed to establish for 7 days until they reached an average volume of 170 mm³ before a 10-day treatment with HPN536 was initiated. On day 24 after tumor implantation, both 500 and 100 μg/kg doses of HPN536 led to eradication of the established HPAFII tumors in mice (Fig. 4B).

For the TOV21G tumor model, mice were treated with 500, 100, and 20 μg/kg HPN536 1 day following tumor coimplantation with human T cells. On day 35 postimplantation, 500, 100, and 20 μg/kg HPN536 treatment groups showed TGI of 65.1%, 68.6%, and 52.3%, respectively, and delayed tumor outgrowth with high statistical significance (Fig. 4C). For the NCI-H292 model, tumor cells were coimplanted with human PBMCs and tumors were allowed to grow for 6 days, at which point the tumor volume reached about 27 mm³, before treatment was initiated. On day 6, HPN536 was administered at daily doses of 500, 100, and 20 μg/kg for 10 days. A highly significant tumor inhibition was seen in both the 100 and 500 μg/kg HPN536-treated groups compared with the vehicle-treated group (70.9% and 77.1% TGI, respectively; Fig. 4D). These results demonstrate activity of HPN536 in controlling growth of MSLN-expressing tumors derived from different histologic origins.

On the basis of the species cross-reactivity of HPN536 to cynomolgus monkey MSLN, CD3, and albumin (see Fig. 1), the pharmacokinetics of a single dose of 0.1, 1, or 10 mg/kg was evaluated in cynomolgus monkeys (two males and two females per dose group). An anti-idiotype assay (anti-anti-albumin used for capture and anti-anti-CD3 for detection) and a functional assay (recombinant CD3 and MSLN proteins used for capture and detection, respectively) were used for detection of HPN536. The serum concentration–time profile for HPN536 exhibited a biphasic decline for the dose range 0.1–10 mg/kg over the time course of the study (Fig. 5A). Noncompartmental analysis of HPN536 exhibited a dose proportional increase in maximum serum concentration (C_max) and area under the concentration–time curve (AUC_0–inf). In addition, the volume of distribution at steady state was independent of dose. The clearance rate was in the range of 0.57–1.39 mL/hour/kg and the mean terminal half-life was between 49.0 and 113 hours, supporting the hypothesis that the HSA binding domain engaged its target HSA to extend the serum half-life of HPN536 (Supplementary Table S6). The time–concentration profiles determined with the two different assays overlaid (Fig. 5A), suggesting that HPN536 had retained its structural and functional integrity in cynomolgus monkeys over the course of the 3-week study. To examine in vivo stability and biological activity, HPN536-containing serum samples collected after 168 hours of post-HPN536 administration were tested for redirected tumor cell lysis in the TDCC assay. As shown in Fig. 5B, HPN536 contained in the 168-hour serum samples from monkeys was as potent in the TDCC assay as the HPN536 reference stored at −80°C.

Single doses of HPN536 were well tolerated in nonhuman primates (NHP) up to 10 mg/kg and no dose-limiting toxicities were observed. HPN536 administration resulted in transient, nonadverse changes in clinical pathology parameters (Fig. 5C,–E; Supplementary Table S7) and cytokines. Consistent with T-cell activation by HPN536, transient mild to moderate decreases in circulating lymphocytes were observed between days 1 and 2 after drug administration that trended back to control or baseline values by days 3 or 8 (Supplementary Table S7). They included Th lymphocytes, cytotoxic T lymphocytes, natural killer cells, and B lymphocytes as observed by flow cytometry. Figure 5C exemplifies a transient reduction in T lymphocytes that recovered by 168 hours after drug administration. A small fraction of T cells showed upregulation of T-cell activation marker, CD69, in response to HPN536 that was not seen in vehicle-treated animals (Fig. 5D). T-cell activation in response to HPN536 was also evident from a dose-dependent increase in the inflammatory cytokine, IL6, in serum (Fig. 5E). At 10 mg/kg, IL6 levels peaked between 4 and 8 hours after HPN536 administration and had declined by 24 hours after dose. Increases in serum levels of IFNγ and IL2 were also observed between 4 and 8 hours after dose in a subset of animals treated at 10 mg/kg (Supplementary Fig. S7). No consistent changes in serum levels were observed for other cytokines, including IL4, IL5, IL10, and TNFα (Supplementary Fig. S7).

Single doses of HPN536 did not elicit gross pathologic changes or dose-limiting toxicities. The main histopathologic finding was a moderate, dose-dependent mesothelial hypertrophy accompanied by a mixed immune cell infiltration and extracellular matrix deposition. This effect was most pronounced in animals treated at the highest dose level of 10 mg/kg HPN536. Figure 6 compares a microscopic section through the pulmonary mesothelial layer of an animal receiving a 10 mg/kg dose (middle) with a section from a vehicle-treated animal (left). These HPN536-related mesothelial changes were observed in fewer tissues and at a lower incidence and/or severity 3 weeks after dosing when compared with findings at 1-week after dosing, possibly reflecting reversibility of the histopathologic effect (Fig. 6, middle and right).

Multiple MSLN-targeted therapies for the treatment of MSLN-expressing malignancies have entered clinical development in recent years (38, 39). These include the antibody amatuximab (40), MSLN-based vaccines (41, 42), chimeric antigen receptor (CAR)-T cells (32), immunotoxins (33), and antibody–drug conjugates (43). Some clinical benefit has been reported for the immunotoxin, SS1P (33, 44, 45), and the antibody–drug conjugate, DMOT4039A (46). Likewise, MSLN-specific CAR-T cells have shown some benefit after intraperitoneal administration to patients with mesothelioma in combination with an anti-PD-1 antibody (32, 47, 48). Challenges of current MSLN-targeted therapies are low response rates and narrow therapeutic windows. To overcome these, we have generated HPN536, which currently is the only T-cell–engaging antibody construct in clinical testing in patients with MSLN-expressing cancers.

While many T-cell engager formats in development rely on an Fc-gamma domain for half-life extension, HPN536 utilizes an sdAb binding to HSA. Here, we show that the sdAb can confer a prolonged serum half-life of the 53-kDa TriTAC, for up to 113 hours in NHPs. In contrast, the 55-kDa BiTE antibody, blinatumomab, had a serum half-life of only 2 hours in patients and, therefore, required continuous intravenous infusions (7, 49). A weekly or perhaps less frequent dosing schedule may be sufficient to maintain appropriate plasma drug concentration of HPN536 for antitumor activity in the clinic, which would overcome a limitation of the canonical BiTE format. Of note, the in vitro biological activities of HP536 were only slightly affected by the presence of physiologic concentrations of serum albumin, which seems to be owed to positioning of the albumin-binding sdAb in the middle of the TriTAC molecule.

Two sdAbs were used to engineer HPN536 to achieve high protein stability and to minimize the overall size of the TriTAC. sdAbs were shown to have a higher stability than conventional antibodies and derived fragments that require pairing of heavy and light chain variable domains (50, 51). This may explain the high stability of HPN536 in the circulation of cynomolgus monkeys for up to 7 days, where the TriTAC retained virtually the same biological activity as stored reference material. The inherent stability of sdAbs is retained with alignment of multiple domains on a single polypeptide chain as exhibited by low aggregation propensity. Notably, HPN536 remained monomeric, that is, monovalent, even after incubation for up to 14 days at 40°C (Supplementary Fig. S1). We consider this to be important for HPN536 for reduction of nonspecific T-cell activation and avoidance of off-target T-cell activation in the periphery as can be caused by formation of aggregates, which often is a problem with multivalent anti-CD3 moieties. Finally, the use of a small albumin-binding sdAb for half-life extension in lieu of an Fc-gamma domain allowed reduction of the overall molecular size of TriTACs to a third of an IgG molecule. On theoretical grounds, we hypothesize that the small size and globular shape of HPN536 may improve diffusion into and across tumor tissue (52) for better T-cell engagement and antitumor activity. Although HPN536 binds albumin and possibly soluble MSLN, these interactions are noncovalent. As a result, a small fraction of HPN536 could remain unbound at any given time that can potentially diffuse as an approximately 53-kDa protein and more readily access the tumor microenvironment.

Redirected lysis and T-cell activation by HPN536 was dependent on binding of the TriTAC molecule to MSLN expressed on cancer cells and to CD3ϵ expressed on T cells. EC₅₀ values for MSLN-dependent target cell killing in the range of 0.6–15 pmol/L suggest that formation of a cytolytic synapse between target and T cells required only minute concentrations of HPN536. Nevertheless, target cell lysis and T-cell activation were of highest specificity as they were not observed with cancer cells lacking MSLN expression or with a TriTAC specific for GFP. HPN536 was similarly active across a variety of MSLN-expressing human cancer cell lines from different cancer indications in vitro and in xenograft models and against cell lines engineered to express human or cynomolgus monkey MSLN. Of note, HPN536 was designed to bind to mature MSLN as it is retained on the cell surface after release of MPF. This way, serum levels of MPF will not impact its activity. The high biological activity, stability, and favorable pharmacokinetics properties of HPN536 support studies testing its clinical activity in patients with MSLN-expressing cancers.

Molecules bridging CD3ϵ on T cells with a target antigen on cancer cells can organize formation of synaptic structure at the cells' interface that resembles that of a native immunologic synapse as formed between a TCR complex on T cells and peptide-MHC complexes on antigen-presenting cells. Such cytolytic synapses have been described for an EpCAM/CD3-bispecific BiTE molecule (53). They serve as sites for T-cell signal transduction and activation, and as sites for perforin and granzyme release ultimately leading to target cell lysis and caspase 3- and 7-mediated apoptosis (54). The EC₅₀ values for redirected lysis by HPN536 were several logs lower than the K_D values measured for HPN536 binding to MSLN and CD3ϵ, suggesting that very few TriTAC molecules are needed to connect TCRs on T cells with MSLN on tumor cells for cytolytic synapse formation. Cooperative adhesion based on avidity gain and a limited diffusion of HPN536 out of the synaptic structure may be crucial for forming and stabilizing synapses with very few TriTAC molecules.

A close species cross-reactivity of HPN536 between human and cynomolgus monkey MSLN, albumin, and CD3ϵ provided the basis for a meaningful assessment of pharmacokinetics, pharmacodynamics, and toxicology of HPN536 in the NHP species. While all tested doses led to a transient eclipse of lymphocytes, only the 10 mg/kg dose of HPN536 led to a more robust and transient cytokine release. While transient lymphocyte margination does not need T-cell activation, but may be largely governed by a conformational change of cell adhesion molecule LFA-1 on T cells to a high-affinity variant, referred to as in-side-out signaling (55), cytokine release and CD69 expression in monkeys were likely to have resulted from synapse formation between T cells and MSLN-expressing normal cells leading to new gene expression. The pharmacodynamic findings with HPN536 in NHPs resemble those of other T-cell–engaging molecules (56, 57). A possible target on normal tissue for HPN536 is the mesothelial cell layer lining cavities, the major tissue expressing MSLN. Reversible hyperplasia and immune infiltrates and matrix deposition observed in the mesothelial linings of treated monkeys support this notion. These dose-dependent histopathologic findings suggest that HPN536 can reach the mesothelial layers and recruit T cells. Simultaneous engagement of MSLN and CD3ϵ on the mesothelial cells and T cells, respectively, by HPN536 then resulted in redirected lysis of the MSLN-expressing cell layer. As sequalae, other immune cells get attracted, and the mesothelial layer thickens possibly due to an ensuing fibrosis. Despite changes at the microscopic level, HPN536 was remarkably well tolerated at single doses up to 10 mg/kg without gross macroscopic findings or dose-limiting toxicities. Of note, the C_max at the 10 mg/kg dose exceeded the highest in vitro EC₅₀ value for redirected lysis (i.e., 15 pmol/L observed in the HPAFII cells) by a factor of 4 × 10⁵-fold. The pharmacodynamics and toxicology evaluation of HPN536 in the relevant NHP species suggest the potential for a wide therapeutic window.

In conclusion, the current preclinical characterization of HPN536 has provided (i) a scientific rationale to examine the activity of HPN536 in patients suffering from MSLN-expressing cancer, (ii) a further understanding of its potency and mechanism of action, (iii) a basis for calculating the first-in-human clinical dose for a phase I/IIa clinical study based on the recommendation by Saber and colleagues (58), and (iv) a high safety margin in the absence of malignant tissue as shown in a pharmacologic relevant NHP species. An open-label, phase I/IIa study of HPN536 as monotherapy to assess the safety, tolerability, and pharmacokinetics in patients with advanced cancers associated with MSLN expression is currently ongoing (NTC 03872206).

M.E. Molloy reports a patent for US10543271B2 issued and US20180327508A1 issued. R.J. Austin reports a patent for US10543271B2 issued, US20180327508A1 pending, US20200270362A1 pending, and US9708412B2 issued; reports employment with Harpoon Therapeutics; and has stock and stock options in Harpoon Therapeutics. B.D. Lemon reports a patent for US9920115 issued, US10066016 issued, US10100106 issued, US10543271 issued, US10544221 issued, US10730954 issued, US20180161428A1 pending, US20180162949A1 pending, US20190031749A1 pending, US20190112381A1 pending, US20200270362A1 pending, and US20200289646A1 pending. A. Jones reports other from Harpoon Therapeutics during the conduct of the study. L. Tatalick reports paid nonclinical consultancy for Harpoon. P.A. Baeuerle reports personal fees from Harpoon Therapeutics, Inc. outside the submitted work, as well as a patent 9708412 issued, 20170298149 pending, and 20180162949 pending. C.-L. Law reports full-time employment at Harpoon Therapeutics and owns shares of the company. H. Wesche reports employment with Harpoon and owns shares of Harpoon Therapeutics Inc. No disclosures were reported by the other authors.

M.E. Molloy: Methodology, writing-original draft, writing-review and editing. R.J. Austin: Conceptualization, supervision, project administration, writing-review and editing. B.D. Lemon: Conceptualization, supervision, project administration, writing-review and editing. W.H. Aaron: Formal analysis, investigation, methodology. V. Ganti: Formal analysis, methodology. A. Jones: Investigation. S.D. Jones: Supervision. K.L. Strobel: Investigation, methodology. P. Patnaik: Investigation. K. Sexton: Investigation. L. Tatalick: Formal analysis, methodology, writing-review and editing. T.Z. Yu: Investigation, methodology. P.A. Baeuerle: Conceptualization, writing-review and editing. C.-L. Law: Conceptualization, supervision, writing-original draft, project administration. H. Wesche: Conceptualization, supervision, funding acquisition, project administration, writing-review and editing.

All funding for this work was provided by Harpoon Therapeutics, Inc.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.