Glutamic acid-valine-citrulline linkers ensure stability and efficacy of antibody-drug conjugates in mice

Abstract

Valine-citrulline linkers are commonly used as enzymatically cleavable linkers for antibody-drug conjugates. While stable in human plasma, these linkers are unstable in mouse plasma due to susceptibility to an extracellular carboxylesterase. This instability often triggers premature release of drugs in mouse circulation, presenting a molecular design challenge. Here, we report that an antibody-drug conjugate with glutamic acid-valine-citrulline linkers is responsive to enzymatic drug release but undergoes almost no premature cleavage in mice. We demonstrate that this construct exhibits greater treatment efficacy in mouse tumor models than does a valine-citrulline-based variant. Notably, our antibody-drug conjugate contains long spacers facilitating the protease access to the linker moiety, indicating that our linker assures high in vivo stability despite a high degree of exposure. This technology could add flexibility to antibody-drug conjugate design and help minimize failure rates in pre-clinical studies caused by linker instability.

Fig. 1

Structures and plasma stability of cathepsin-responsive cleavable peptides. a VCit and EVCit-based ADC linkers. VCit linkers are unstable in mouse plasma due to susceptibility to the extracellular carboxylesterase Ces1c. This instability often triggers premature release of payload in circulation. This study presents that VCit-based tripeptide sequences with an acidic side chain such as EVCit are responsive to cathepsin-mediated cleavage but highly stable in mouse plasma. b Structures of pyrene-based small-molecule probes containing a VCit (1a), SVCit (1b), EVCit (1c), DVCit (1d), KVCit (1e) or HO-GVCit (1f) cleavable sequence. c Stability of probes (1a–f) in undiluted BALB/c mouse plasma at 37 °C. (1a) blue circle; (1b) orange triangle; (1c) green square; (1d) magenta hexagon; (1e) red inversed triangle; (1f) gray cross. EVCit and DVCit probes (1c, d) showed great plasma stability while highly responsive to cathepsin B-mediated cleavage (see Supplementary Fig. 2). All assays were performed in triplicate. Error bars represent s.e.m. and values in parentheses are 95% confidential intervals

Fig. 2

Construction and characterization of ADCs 3a–c. a Construction of ADCs (3a–c) by MTGase-mediated branched linker conjugation and following strain-promoted azide–alkyne cycloaddition (cyan cylinder: PEG spacer–XVCit–PABC module; yellow spark: MMAF). b Deconvoluted ESI-mass spectra. Top panel: N297A anti-HER2 mAb. Second panel: antibody–branched linker conjugate. Third–fifth panels: highly homogeneous ADCs (3a–c). Asterisk (*) indicates a fragment ion detected in ESI-MS analysis. c SEC traces of ADCs (3a–c). d HIC analysis of ADCs (3a–c) under physiological conditions (phosphate buffer, pH 7.4). e Overlay of the three HIC traces (VCit ADC 3a: blue; SVCit ADC 3b: orange; EVCit ADC 3c: green). DAR, drug-to-antibody ratio; MTGase, microbial transglutaminase; PABC, p-aminobenzyloxycarbonyl; PEG, polyethylene glycol

Fig. 3

Plasma stability and in vitro cytotoxicity. a Stability in human plasma. b Stability in mouse plasma. Cell killing potency in the breast cancer cell lines KPL-4 (c), JIMT-1 (d), and MDA-MB-231 (e). We tested unconjugated N297A anti-HER2 mAb (black cross), VCit ADC (3a, blue circle), SVCit ADC (3b, orange triangle), EVCit ADC (3c, green square), non-cleavable ADC (4, red inversed triangle), and isotype control ADC containing EVCit (5, magenta diamond, non-targeting control). All assays were performed in quadruplicate. Error bars represent s.e.m

Fig. 4

In vivo pharmacokinetics (PK) and antitumor activity. a, b PK of unconjugated N297A anti-HER2 mAb (black cross), VCit (blue circle), SVCit (orange triangle), and EVCit (green square) ADCs (3a–c) in female BALB/c mice (n = 3). At the indicated time points, blood was collected to quantify total antibody (conjugated and unconjugated, a) and ADC (conjugated only, b) by sandwich ELISA. c, d Antitumor activity of anti-HER2 ADCs (3a, c) in the JIMT-1 (c) and KPL-4 (d) xenograft tumor models (female NCr nude mice, n = 3 for vehicle in the KPL-4 model; n = 5 for vehicle in the JIMT-1 model and ADCs in both models). A single dose of VCit ADC (3a, 3 mg kg^–1, blue circle), EVCit ADC (3c, 3 mg kg^–1, green square; 1 mg kg^–1, magenta triangle), or vehicle control (gray inversed triangle) was administered to mice when a mean tumor volume reached ~100 mm³ (indicated with a black arrow). Error bars represent s.e.m. e, f Changes in the percentages of surviving mice over time in the JIMT-1 (e) and KPL-4 (f) models. The curve of ADC (3c) at 1 mg kg^–1 (magenta, e) is slightly shifted upward for clarity. Mice were euthanized at the pre-defined endpoint (see the Method). * P < 0.025, ** P < 0.01, *** P < 0.005 (PK analysis: Welch’s t test; tumor volume on Day 27: Mann–Whitney U test; survival curve: log rank test). The vehicle control groups were not used for statistical analysis