A Rationally Designed Humanized Antibody Selective for Amyloid Beta Oligomers in Alzheimer's Disease

Abstract

Advances in the understanding of Alzheimer's disease (AD) suggest that pathogenesis is not directly related to plaque burden, but rather to soluble toxic amyloid-beta oligomers (AßO). Therapeutic antibodies targeting Aß monomers and/or plaque have shown limited efficacy and dose-limiting adverse events in clinical trials. These findings suggest that antibodies capable of selectively neutralizing toxic AßO may achieve improved efficacy and safety. To this end, we generated monoclonal antibodies against a conformational Aß epitope predicted by computational modeling to be presented on toxic AßO but not monomers or fibrils. The resulting lead antibody, PMN310, showed the desired AßO-selective binding profile. In vitro, PMN310 inhibited AßO propagation and toxicity. In vivo, PMN310 prevented AßO-induced loss of memory formation and reduced synaptic loss and inflammation. A humanized version (huPMN310) compared favorably to other Aß-directed antibodies showing a lack of adverse event-associated binding to Aß deposits in AD brains, and greater selective binding to AßO-enriched AD brain fractions that contain synaptotoxic Aß species. Systemic administration of huPMN310 in mice resulted in brain exposure and kinetics comparable to those of other therapeutic human monoclonal antibodies. Greater selectivity for AßO and the potential to safely administer high doses of huPMN310 are expected to result in enhanced safety and therapeutic potency.

Figure 1

Aβ residues 13–16, HHQK, held in a constrained turn represent a predicted AβO-specific epitope. (a) Representative molecular dynamics simulated conformation of the cyclic HHQK epitope, with the side chains oriented into solvent. Yellow dot is sulfhydryl on terminal cysteine, used for conjugation to carrier proteins. (b) Skeletal structural formula of c[CGHHQKG].

Figure 2

Selective binding of the muPMN310 antibody to cyclized CGHHQKG peptide and Aβ oligomers. (a) SPR measurements of muPMN310 binding to cognate cyclized peptide epitope c[CGHHQKG], and unstructured linear CGHHQKG peptide. Representative sensorgrams from two independent experiments are shown. (b) SPR measurements of muPMN310 binding to Aβ42 monomers and oligomers (Aβ42O). Means and SEM of three identical experiments are shown. Differences were determined by two-way ANOVA with Sidak’s multiple comparison test: muPMN310 binding to Aβ42O is significantly higher than all other measurements (p ≤ 0.0001); muPMN310 binding to Aβ42 monomers is not different from isotype control (mIgG1) by two-way ANOVA.

Figure 3

PMN310 does not react with Aβ plaque or vascular deposits in AD brain sections. Sections from the frontal cortex of human AD brain were stained with 1 μg/ml of indicated antibodies. Detection of bound antibody with secondary anti-human or anti-mouse IgG appears in brown, and nuclear counterstaining with hematoxylin in blue. The images are representative of 3 or more independent experiments with four individual AD brains. Scale bar = 100 μm, applicable to all images. White arrows - characteristic plaque staining; black arrows - Aβ vascular deposit staining.

Figure 4

muPMN310 binding inhibits Aβ42 aggregation and Aβ42O toxicity in vitro. (a) β-sheet formation was tracked for 24 hours in vitro with a Thioflavin T fluorescent assay after addition of Aβ monomers alone (red line), or in the presence of muPMN310 (blue line) or isotype control mIgG (green line) at a molar ratio of 1:5 (Antibody:Aβ). Data are representative of three independent experiments. (b) Thioflavin T fluorescent assay samples were collected at end-point and fractionated into soluble Aβ (Supernatant; monomers, small oligomers) or insoluble Aβ (Pellet; large aggregates) by centrifugation, then run on a denaturing SDS-PAGE gel for monomerization and immunoblotting with pan-Aβ antibody, 6E10. Shown are representative data from one of two identical experiments. (c) MTT assay of viability of neurons treated with vehicle (Veh.) or AβO and increasing molar ratios of muPMN310. Antibody:AβO ratios of 1:10 (0.1), 1:3 (0.3), 2:1 (2). Mean ± SEM shown. Differences were determined with a Students t-test: (*) Veh. vs AβO p ≤ 0.0001, (#) AβO vs muPMN310 + AβO p = 0.0112. Similar results were observed in 3 independent assays.

Figure 5

muPMN310 inhibits AβO toxicity in vivo. Wild-type mice (n = 12 per group) were injected i.c.v. with either vehicle (Veh.), Aβ42O + vehicle (AβO), muPMN310 + vehicle (muPMN310) or muPMN310 + AβO at a molar ratio of 2:1. (a) NOR discrimination indices (n = 7–10 evaluable mice per group). Mean ± SEM shown. Data were not normally distributed. Differences were compared using Kruskal-Wallis ANOVA with Dunn’s multiple comparison test. AβO is significantly less than all other samples (*p ≤ 0.05 vs Vehicle, muPMN310, muPMN310 + AβO). No other comparisons were significant, including Veh.:PMN310 + AβO. Hippocampal levels of TNF-alpha (b), PSD-95 (c), SNAP25 (d) (n = 11–12 mice per group). Mean ± SEM shown. Differences were compared using one-way ANOVA with Tukey’s multiple comparison test for normally distributed data in (b: *p ≤ 0.05) and (c). (d) Difference determined with Kruskal-Wallis ANOVA with Dunn’s multiple comparison test for non-normally distributed data. In (c,d) *p ≤ 0.05 for AβO vs Veh., muPMN310, muPMN310 + AβO.

Figure 6

huPMN310 selectively recognizes soluble, low molecular weight, Aβ aggregate species in AD brains. (a) SEC fractionation chromatogram of pooled soluble AD human brain extracts. A representative chromatogram from 9 independent fractionations is shown (red line). MW markers are superimposed for reference (blue line). (b) SPR binding response of indicated antibodies to pooled LMW and HMW fractions. Four measurements were made under identical conditions: statistical significance of differences was determined by two-way ANOVA with Sidak’s multiple comparison test: LMW binding by huPMN310 was greater than all other measurements (p < 0.0001), huIgG1 (LMW and HMW) binding was significantly less than all other measurements (p ≤ 0.0012). No significant difference between LMW and HMW binding for aducanumab vs bapineuzumab. (c,d) SPR binding response of indicated antibodies to LMW AD brain extract and subsequent aducanumab (c) or huPMN310 (d) detection of the analyte captured. Fifteen capture and three detection measurements under identical conditions were performed. (e) c[CGHHQKG] peptide was pre-injected over indicated immobilized antibody surfaces to compete with LMW brain analyte. Data are representative of two experiments with similar results.

Figure 7

CNS penetrance of huPMN310. (a,b) Aged WT C57Bl/6 mice injected i.p. with 30 mg/kg of aducanumab (n = 3), huPMN310 (n = 4) or PBS (n = 2). (a) Plasma and brain levels of human IgG at 24 h. No statistically significant difference between aducanumab and huPMN310 in plasma or brain by two-way ANOVA. (b) CNS penetrance - Percent of human IgG in brain compared to plasma. No statistically significant difference between aducanumab and huPMN310 by Mann-Whitney test. (c,d) Aged APP/PS1 mice injected i.p. with 30 mg/kg of huPMN310. (c) Plasma and brain levels of huPMN310 at days 1–21 (n = 4–6 mice/time point). Two-way ANOVA with Sidak’s post-test test shows statistically significant differences in plasma levels between day 1 and days 7 (p = 0.0016), 14 and 21 (p < 0.0001) with no significant differences in brain levels at the different time points. (d) Brain:plasma ratios of huPMN310 at days 1–21. Kruskal-Wallis ANOVA with Dunn’s post- test shows a statistically significant difference only for Day 1 vs 14 (p = 0.043).