Search

CD38 is highly expressed on plasma cells and is an attractive target for multiple myeloma (MM) therapies. Several anti-CD38 antibodies including daratumumab and SAR650984 show promising results in clinical development, though such antibodies are not able to stimulate T cell-mediated killing of myeloma cells. To exploit a T cell immunotherapy mechanism while retaining the favorable drug properties of therapeutic antibodies, we designed bispecific antibodies that recruit T cells to CD38+ MM cells. Such bispecifics act via redirected T cell-cytotoxicity (RTCC) to stimulate T cell-mediated target cell killing regardless of T cell receptor antigen specificity. These anti-CD38 × anti-CD3 antibodies possess a full Fc domain and spontaneously form stable heterodimers that are readily manufactured. Their Fc domain lacks binding to Fcγ receptors to minimize nonselective T cell activation, yet retains binding to FcRn to maintain long serum half-life. We have previously reported that XmAb13551, a humanized and affinity-optimized anti-CD38 × anti-CD3 antibody, stimulates killing of the CD38+ MM cell line RPMI8226 by human T cells and suppresses human Ig levels in SCID mice engrafted with human PBMCs, showing much greater efficacy than daratumumab in these models (Blood 2014 124:4727). We also investigated efficacy of XmAb13551 in monkeys given a single dose of 2, 5, and 20 μg/kg. Within 1 hr after dosing, CD25 and CD69 activation markers were upregulated on T cells and within 8 hr, circulating CD38+ cells were depleted by > 95% at the 20 μg/kg dose. However, depletion of peripheral CD38+ cells was not sustained, suggesting that a large antigen sink was limiting drug exposure. Although higher dosing might overcome an antigen sink, higher doses of XmAb13551 (0.2 mg/kg or higher) resulted in a T cell-mediated cytokine release syndrome (CRS) in monkeys. We reasoned that an anti-CD38 × anti-CD3 antibody with reduced CD3 affinity would stimulate sufficient RTCC to deplete MM cells, yet would attenuate the acute T cell activation (and associated CRS) induced by high-affinity coengagement of T cells with CD38+ target cells. Using XmAb13551 as a starting point, we engineered a series of bispecifics retaining the same high-affinity (0.2 nM) binding to CD38, but with reduced affinity to CD3. We selected two antibodies - XmAb15426 and XmAb14702 - that have significantly reduced CD3 affinity. As expected, these molecules showed reduced potency in RTCC assays using T cells to kill RPMI8226 cells, with potency correlating with CD3 affinity (XmAb13551 > XmAb15426 >> XmAb14702). We next tested XmAb15426 and XmAb14702 at single doses of 0.5 mg/kg and 3 mg/kg, respectively, in cynomolgus monkeys. Both antibodies were well-tolerated at these higher doses, consistent with the moderate levels of IL6 observed in serum from the treated monkeys. Moreover, XmAb15426, with intermediate CD3 affinity, more effectively depletes CD38+ cells at 0.5 mg/kg compared to the original high-affinity XmAb13551 dosed at 2, 5 or 20 µg/kg. Depletion by XmAb15426 was more sustained compared to the highest dose of XmAb13551 in the previous study (7 vs. 2 days, respectively). Notably, although target cell depletion was greater for XmAb15426, T cell activation (CD69, CD25 and PD1 induction) was much lower in monkeys treated with XmAb15426 even dosed 25-fold higher than the 20 µg/kg XmAb13551 group. XmAb14702, with very low CD3 affinity, had little effect on CD38+ cells and T cell activation. Our results demonstrate that modulating T cell activation by attenuating CD3 affinity is a promising method to improve the therapeutic window of T cell-engaging bispecific antibodies. This strategy has potential to expand the set of antigens amenable to targeted T cell immunotherapy by improving tolerability and enabling higher dosing to overcome antigen sink clearance with targets such as CD38. We have shown that by reducing affinity for CD3, XmAb15426 effectively depletes CD38+ cells while minimizing the CRS effects seen with comparable doses of its high-affinity counterpart XmAb13551. Our preclinical data for XmAb15426 provide a rationale for clinical testing of this bispecific antibody in patients with multiple myeloma and other CD38+ malignancies. Disclosures Moore: Xencor, Inc.: Employment, Equity Ownership. Lee:Xencor, Inc.: Employment, Equity Ownership. Schubbert:Xencor, Inc.: Employment, Equity Ownership. Miranda:Xencor, Inc.: Employment, Equity Ownership. Rashid:Xencor, Inc.: Employment, Equity Ownership. Pong:Xencor, Inc.: Employment, Equity Ownership. Phung:Xencor, Inc.: Employment, Equity Ownership. Chan:Xencor, Inc.: Employment, Equity Ownership. Chen:Xencor, Inc.: Employment, Equity Ownership. Endo:Xencor, Inc.: Employment, Equity Ownership. Ardila:Xencor, Inc.: Employment, Equity Ownership. Bernett:Xencor, Inc.: Employment, Equity Ownership. Chu:Xencor, Inc.: Employment, Equity Ownership. Leung:Xencor, Inc.: Employment, Equity Ownership. Muchhal:Xencor, Inc.: Employment, Equity Ownership. Bonzon:Xencor, Inc.: Employment, Equity Ownership. Szymkowski:Xencor, Inc.: Employment, Equity Ownership. Desjarlais:Xencor, Inc.: Employment, Equity Ownership.

Both rHis6-BoNT/E and rBoNT/E-His6 were produced from a codon optimized gene within Escherichia coli. Similar to native BoNT/E isolated from Clostridium botulinum, the recombinant proteins were isolated as single-chain molecules. The His6 affinity tag facilitated a two-step IMAC-IEX purification. While reasonable yields were obtained, single-chain rHis6-BoNT/E was produced less efficiently than was rBoNT/E-His6. Treatment with recombinant trypsin efficiently converted both purified rBoNT/E constructs to the activated dichain forms, with concomitant removal of the His6 tag, and the nicking profile for each rBoNT/E molecule matched that of native BoNT/E. Following nicking with trypsin, it was observed that the resultant rHis6-BoNT/E dichain displayed reduced in vivo potency (mouse potency assay) compared to activated native BoNT/E dichain. In contrast, activated rBoNT/E-His6 displayed a specific potency equivalent to that of trypsin-nicked native BoNT/E in the mouse potency assay. These results demonstrate that rBoNT/E, containing a C-terminal His6 affinity tag, can be expressed and purified in high yield, with the activated molecule displaying equivalent in vivo potency to activated native BoNT/E.