Binding affinities and cytotoxicity of HER2-XPAT and its metabolites
The effect of XTEN polypeptide masking on binding and in vitro activity was investigated using a HER2-targeted XPAT protein. Kinetic analysis by surface plasmon resonance evaluated the affinity of HER2-XPAT and its active metabolites for human HER2 and CD3ε (Fig. 2a). The presence of the XTEN masks reduced target affinities tenfold for HER2 and approximately sixfold for CD3 compared with unmasked HER2-TCE. The affinity (KD) of HER2-XPAT for human HER2 was 24.9 ± 4.3 nM and for human CD3 was 160.0 ±19.5 nM, whereas the affinity for the unmasked HER2-TCE for HER2 was 2.5 ± 0.9 nM and for CD3 was 26.3 ±2.2 nM. The affinities of HER2-XPAT and its metabolites were similar for cynomolgus monkey HER2 and CD3 (see Supplementary Material, Supplementary Fig. 1), supporting the relevance of assessing toxicity and pharmacodynamics (PDs) of HER2-XPAT in NHPs.
The in vitro cytotoxicity of HER2-XPAT and its metabolites was investigated using co-cultures of human peripheral blood mononuclear cells (huPBMCs) with HER2-expressing human tumor cell lines. The proteolytically activated, unmasked HER2-TCE demonstrated highly potent cytotoxicity against high HER2-expressing tumor cells, SKOV3 (~650,000 HER2 receptors per cell; Fig. 2b) and BT-474 (~975,000 HER2 receptors per cell; Fig. 2c), showing nearly complete target cell killing with 50% effective concentrations (EC50) in the single-digit picomolar range. Cytotoxicity was strongly attenuated with fully masked HER2-XPAT (average EC50 values shifted by ~4 logs), demonstrating the significant functional protection provided by the two XTEN polypeptide masks. Unmasked HER2-TCE was also cytotoxic in the medium-low HER2-expressing cell line, MCF7 (~12,500 receptors per cell), although with a 7- to 16-fold higher EC50 than seen with high HER2-expressing cells (Fig. 2d). With MCF7, a similar >4-log attenuation of cytotoxicity was observed with masked HER2-XPAT versus unmasked HER2-TCE. A single mask on HER2-XPAT (either at the C- or N-terminal end of the TCE) was associated with an intermediate level of cytotoxicity in the BT-474 cell line (Fig. 2e).
The low level of cytotoxicity with HER2-XPAT in the HER2-expressing tumor cell lines was due, in part, to a low degree of proteolytic cleavage occurring during the in vitro assay. This is supported by the further reduction in cytotoxicity seen with the HER2-XPAT molecule lacking one or both of its protease-cleavage sites (HER2-XPAT-NoClvSite variants) (Fig. 2f).
Target-dependent T-cell activation with HER2-XPAT
T-cell activation by HER2-XPAT and its proteolytically activated unmasked HER2-TCE was characterized utilizing primary huPBMCs co-cultured with SKOV3 tumor cells (HER2 high). Incubation with unmasked HER2-TCE for 72 hours led to dose-dependent upregulation of the activation marker CD69 on T cells (Fig. 2g) and secretion of the pro-inflammatory cytokine interleukin (IL)-2 (Fig. 2h). CD69 expression and IL-2 secretion were significantly attenuated with HER2-XPAT versus unmasked HER2-TCE.
T-cell activation by HER2-XPAT and its proteolytically activated metabolites was also characterized using Jurkat NFAT-luciferase reporter T cells co-cultured with BT-474 tumor cells (Fig. 2i). Unmasked HER2-TCE activated Jurkat NFAT-luciferase reporter T cells in the presence of BT-474 cells with EC50 values in the sub-nanomolar range (accurate EC50 values could not be obtained due to incomplete saturation of the response). A 4-log differential in the concentrations required to initiate signaling was observed between HER2-XPAT and unmasked HER2-TCE. The maximal activation achieved by HER2-XPAT at 3 mM was only ~10% of that observed with unmasked HER2-TCE, confirming the ability of XTEN polypeptide masking to reduce T-cell receptor (TCR) activation. The presence of a single mask (either at the C- or N-terminal end of the TCE) showed intermediate activation, consistent with results of other functional assays. HER2-XPAT-NoClvSite induced no detectable T-cell activation, indicating that the minimal response observed with HER2-XPAT was likely driven by cleavage resulting from active proteases released during the assay.
T-cell activation by unmasked HER2-TCE was insignificant in the absence of HER2-expressing BT-474 tumor cells, demonstrating that monovalent engagement of CD3 was insufficient for activation (see Supplementary Material, Supplementary Fig. 2).
Anti-tumor activity in HER2-positive human tumor xenograft models
In vivo anti-tumor activity of HER2-XPAT was assessed in the HER2-high BT-474 human breast tumor model (~975,000 HER2 receptors per cell) and HER2-low HT-55 colorectal model (~25,000 HER2 receptors per cell) inoculated subcutaneously (SC) into immunodeficient mice, and then engrafted with huPBMCs as a source of T cells.
Equimolar doses of HER2-XPAT (2.1 mg/kg) or unmasked HER2-TCE induced robust and complete tumor regression within 35 days of dosing (P <0.001 for both versus vehicle control at day 35) (Fig. 3a) in HER2-high BT-474 bearing mice. Tumor regression was dependent on protease activity, evidenced by the lack of efficacy with HER2-XPAT-NoClvSite versus vehicle.
Similarly, in mice harboring HER2-low HT-55 xenografts equimolar doses of HER2-XPAT and unmasked HER2-TCE treatment led to significant tumor growth inhibition (TGI) (P <0.01 versus vehicle control for both) (Fig. 3b). HER2-XPAT at 2.1 mg/kg resulted in an intermediate anti-tumor response (P <0.05 versus vehicle control). All eight mice in the vehicle-treated group developed large tumors >1,000 mm3 by day 39, while only 38% (3 of 8 mice) and 0% (0 of 8 mice) in the 2.1 mg/kg and 5.1 mg/kg HER2-XPAT groups, respectively, had tumors >500 mm3. HER2-XPAT-NoClvSite had no impact on tumor growth relative to the vehicle.
T-cell activation in BT-474 tumor-bearing mice was assessed on day 18 post-treatment with equimolar doses of HER2-XPAT and unmasked HER2-TCE. HER2-XPAT and unmasked HER2-TCE promoted similar activation of T cells in the TME, indicated by increased expression of activation marker CD25 on both helper and cytotoxic T cells (Fig. 3c). T cells were not activated in peripheral blood where human HER2 is not expressed, which is consistent with the dual requirement for engagement of both HER2 and CD3 to activate T cells and redirect their killing.
In vivo cleavage of XPAT proteins in patient-derived cancer xenografts versus healthy organs
Two fluorophore-labeled XPAT proteins, XPAT(DyLight 800 [DyL800]) and XPAT(Alexa Fluor 680 [AF680]), were prepared to quantify unmasking in vivo in immunodeficient mice bearing patient-derived xenografts (PDX) (Fig. 4a). Two days after dosing with XPAT(DyL800), tumors and other organs were excised and prepared, with XPAT(AF680) added to monitor cleavage during sample processing, followed by analysis.
A typical sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of fluorophore-labeled cleavage products from HER2-XPAT is shown in Fig. 4b. Four bands representing HER2-XPAT and its three unmasked forms are clearly visible in the tumor sample (XPAT[DyL800] green channel) while the other (XPAT[AF680] red channel) showed minimal unmasking, suggesting that limited cleavage occurred ex vivo during sample processing. Limited unmasking was detected in healthy organs, with fully unmasked HER2-TCE undetectable in healthy tissues.
In vivo XPAT protein unmasking was assessed in five different PDX tumor models with HER2-XPAT and an additional five PDX models with an XPAT prototype targeting human epithelial cell adhesion molecule (EpCAM; EpCAM-XPAT) (see Supplementary Table 2). uTCE in tumors averaged 23.8% (standard deviation [SD] 20.6%; standard error mean 4.6%; range 2.4% to 61.7%) of all XPAT protein forms; while in healthy tissues, uTCE was below the limit of detection (Fig. 4c). Tumor tissues had significant levels of partially unmasked XPAT protein, (1x-C) and (1x-N), while their levels were lower or undetectable in healthy organs.
Reduced toxicity of HER2-XPAT versus unmasked HER2-TCE in NHP
HER2-XPAT domains bind to human and NHP (cynomolgus monkey) HER2 and CD3 with comparable affinities. An NHP dose-escalation study assessed the pharmacokinetics (PK), tolerability, and maximum tolerated dose (MTD) of HER2-XPAT and its unmasked counterpart. HER2-XPAT showed minimal toxicity up to 42 mg/kg (defined as the MTD, single dose), with severe toxicity (unrelated to cytokine release syndrome [CRS]) evident at 50 mg/kg (Fig. 5a). A second dose of HER2-XPAT 50 mg/kg led to toxicity in one of two NHPs dosed, manifesting as multiple findings, including evidence of decreased cellularity in lymphoid tissues, moderate to severe colonic inflammation associated with infection with intralesional protozoa, and bacterial pneumonia. Lymphoid depletion, resulting in the severe infections, was considered related to HER2-XPAT.
We observed no major clinical symptoms or toxicity typically associated with CRS at any HER2-XPAT dose administered (including the 42 mg/kg MTD). By contrast, the unmasked HER2-TCE at a low dose of 0.3 mg/kg/day induced CRS-associated death (Fig 5b), consistent with literature showing that most TCEs induce significant toxicity in NHPs at doses well below 1 mg/kg9. With HER2-XPAT, we observed transient and dose-dependent evidence of decreased lymphocytes, increased neutrophils, and small increases in C‑reactive protein (CRP) and CD4+ T-cell activation in plasma. With HER2-XPAT doses £42 mg/kg, thymic atrophy was the only adverse histopathologic finding observed. No gross or microscopic findings were found in HER2-expressing tissues, including the heart, an organ in which other HER2-targeted agents have demonstrated toxicity18.
Plasma concentrations over time of HER2-XPAT (single IV infusion) and unmasked HER2-TCE (48-hour IV infusion) are shown in Fig. 5c. HER2-XPAT demonstrated a biphasic PK profile, with dose-proportional increases in exposure observed across all evaluated dose levels. The estimated half-life of HER2-XPAT in an NHP weighing 3 kg was ~3 days, similar to the reported half-life of other proteins fused to XTEN polypeptides in NHPs19. The estimated half-life of unmasked HER2-TCE was ~2 hours in an NHP weighing 3 kg. At their MTDs, peak concentrations of HER2-XPAT (42 mg/kg) were >400-fold higher than the unmasked HER2-TCE (0.2 mg/kg/day).
HER2-XPAT and unmasked HER2-TCE induce different PD responses in NHPs (Fig. 5d to Fig. 5h). NHPs receiving unmasked HER2-TCE beyond the MTD showed all the hallmarks of overt CRS described with other TCEs20,21. CRS symptoms included elevations in inflammatory cytokines (IL-6, tumor necrosis factor alpha [TNF-α], interferon gamma [IFN-ɣ], IL-2, monocyte chemotactic protein-1 [MCP-1]), as well as CRP and blood markers indicating liver and kidney damage (aspartate transaminase/alanine transaminase, bilirubin, blood urea nitrogen). HER2-XPAT, even at the highest doses tested, induced limited systemic T-cell activation (Fig. 5d, 5e) and cytokine responses (Fig. 5f to Fig. 5h). Even at its toxic dose of 50 mg/kg, HER2-XPAT exerted local toxicity, and the systemic release of cytokines remained low.
Stability of the HER2-XPAT protease-cleavable linker in plasma
In vivo proteolytic stability was evaluated by comparing the PK profiles of HER2-XPAT and HER2-XPAT-NoClvSite (variant lacking the protease-cleavable linker) in NHPs (Fig. 6a). The concentration-time profiles of HER2-XPAT and HER2-XPAT-NoClvSite were similar: dose-normalized area under the curve from day 0 to day 7 [AUC0-7] was 757 ± 158 day•nM/(mg/kg dose) for HER2-XPAT and 804 ± 79 day•nM/(mg/kg dose) for HER2-XPAT-NoClvSite. Because significant cleavage of HER2-XPAT would result in accelerated elimination versus its noncleavable counterpart, their comparable PK curves indicate that HER2-XPAT is predominately stable in vivo, with very low levels of cleavage at its protease-cleavable linkers in systemic circulation.
After HER2-XPAT dosing, concentrations of partially unmasked metabolites, (1x-N) and (1x-C), increased over time with molar concentrations of (1x-N) and (1x-C) reaching only £3.3% and £2.5%, respectively, of total material fused to XTEN polypeptides at all evaluated time points, despite high doses of 25 mg/kg and 42 mg/kg and peaked (Fig. 6b). Fully unmasked HER2-TCE was below the limit of detection (3 nM).
The stability of HER2-XPAT was also evaluated in spiked plasma samples from humans (healthy volunteers and patients with cancer or inflammatory diseases) and NHPs (healthy monkeys and monkeys with drug-induced systemic inflammation). After a 1-week incubation in human plasma at 37°C, fluorophore-labeled HER2-XPAT remained predominantly intact. Metabolites with reduced length suggest cleavage occurs at the protease-cleavable linkers and throughout the XTEN sequence itself. The distribution of metabolites with similar molecular weight to the unmasked HER2-TCE are shown in Fig. 6c. These cleavage products represented only ~2% to 4% of the spiked fluorophore-labeled HER2-XPAT in samples from healthy subjects, cancer patients, and patients with inflammatory diseases, respectively. Importantly, the profile of identified metabolites was similar following incubation in plasma from humans and NHPs, supporting the relevance of NHP as a species for assessing the peripheral cleavage and toxicology of HER2-XPAT.
PD and PK evaluation of an EGFR-XPAT prototype
An EGFR-XPAT prototype was generated based on the variable sequences of panitumumab and humanized SP34 (CD3-binding domain). Strong masking of in vitro T-cell killing by >4 logs was observed with EGFR-XPAT versus unmasked EGFR-TCE (Fig. 7a), with XTEN polypeptide masking showing similar protection to that of HER2-XPAT. EGFR-XPAT showed potent in vivo anti-tumor activity in huPBMC-engrafted mice bearing HT-29 (BRAFmut) human colorectal tumors (Fig. 7b).
A single IV dose of EGFR-XPAT 0.46 mg/kg in NHPs triggered minor toxicity and cytokine release, while a 1mg/kg dose, the MTD, induced a much stronger response (Fig. 7c). A 48-hour continuous IV infusion at 33 mg/kg/day was the MTD for unmasked EGFR-TCE, while a 1-hour IV infusion of 8 mg/kg followed by a continuous infusion of 66 mg/kg/day (terminated after 25 hours) resulted in severe toxicities associated with CRS. The Cmax of EGFR-XPAT at its MTD (1 mg/kg) is ≥200-fold higher than the Cmax of its corresponding unmasked EGFR-TCE administered at its MTD (0.033 mg/kg), demonstrating that XTEN masks can improve the safety margin (and potentially widen the therapeutic window) for TCEs, including those targeting broadly expressed targets.