Efficacy and safety of a therapeutic humanized FSH-blocking antibody in obesity and Alzheimer's disease models

Abstract

There is growing evidence for direct actions of follicle-stimulating hormone (FSH) on tissues other than the ovaries and testes. Blocking FSH action, either genetically or pharmacologically, protects against bone loss, fat gain, and memory loss in mice. We thus developed a humanized FSH-blocking antibody, MS-Hu6, as a lead therapeutic for 3 diseases of public health magnitude - osteoporosis, obesity, and Alzheimer's disease (AD) - that track together in postmenopausal women. Here, we report the crystal structure of MS-Hu6 and its interaction with FSH in atomistic detail. Using our Good Laboratory Practice platform (21 CFR 58), we formulated MS-Hu6 and the murine equivalent, Hf2, at an ultra-high concentration; both formulated antibodies displayed enhanced thermal and colloidal stability. A single injection of 89Zr-labeled MS-Hu6 revealed a β phase t½ of 79 and 132 hours for female and male mice, respectively, with retention in regions of interest. Female mice injected subcutaneously with Hf2 displayed a dose-dependent reduction in body weight and body fat, in the face of reduced free (bioavailable) FSH and unperturbed estrogen levels. Hf2 also rescued recognition memory and spatial learning loss in a context- and time-dependent manner in AD-prone 3xTg and APP/PS1 mice. MS-Hu6 injected into African green monkeys (8 mg/kg) intravenously, and then subcutaneously at monthly intervals, was safe, and without effects on vital signs, blood chemistries, or blood counts. There was a notable approximately 4% weight loss in all 4 monkeys after the first injection, which continued in 2 of the monkeys. We thus provide Investigational New Drug-enabling data for a planned first-in-human study.

Keywords: Adipose tissue; Drug therapy; Endocrinology; Neurodegeneration; Therapeutics.

Figure 1. X-ray crystal structure of the Fab domain of MS-Hu6 and docked complex with FSH.

(A and B) Structural comparison between the Hu6-Fab:FSHβ homology model (A) and the Hu6-Fab:FSHβ complex based on crystal structure of Hu6-Fab (B). Shown are complementarity-determining regions (CDRs), as well as the 13-mer epitope of FSHβ against which MS-Hu6 was raised. Also shown is the orientation of the FSHβ epitope with respect to the epitope-binding region of MS-Hu6, including CDR-H3. (C) An overlay of Fv regions of modeled and crystal-based Hu6-Fab:FSHβ complexes. (D) Atomistic interactions between Hu6-Fab and FSHβ. Heavy and light chains are shown in blue and pink, respectively. Variable domains of heavy chain (V_H) and light chain (V_L) are shown in light blue and light pink, respectively. CDRs are shown in dark blue for V_H (CDR-H1, CDR-H2, CDR-H3) and magenta for V_L (CDR-L1, CDR-L2, CDR-L3). FSHα and FSHβ chains are shown in light green and yellow, respectively. Also shown is a close-up of Hu6-Fab:FSHβ interfaces and interaction network of the 13-mer epitope (light orange) with CDRs (V_H and V_L). Please refer to Supplemental Tables 1 and 2. The atomic coordinates and structure factors for Hu6-Fab have been deposited in the Protein Data Bank (PDB) with PDB ID 8VZW.

Figure 2. Therapeutic formulation of Hf2 prevents fat mass accrual and weight gain in mice on a high-fat diet.

(A) Protein thermal shift assay to evaluate thermostability of MS-Hu6 and Hf2 (100 mg/mL) in therapeutic formulation versus phosphate-buffered saline (PBS). Melting curves are shown as first derivatives [(change in fluorescence)/(change in temperature)]; melting temperature (T_m) is calculated from the second derivative. The notably higher T_m values for both Fc and Fab domains of both antibodies in formulation suggest greater thermostability (also see Table 1). n = 8 replicates. (B) Colloidal stability was compared using dynamic light scattering, which relies on scattering of light caused by Brownian motion of particles. Data were collected in terms of hydrodynamic radius (R_h) and polydispersity index (PDI). Particle size distribution curve, Z-average of R_h values, and volume and size of main and other peaks are shown. n = 3 replicates. Formulated MS-Hu6 and Hf2 exhibited a dominant peak volume greater than 99%, with an average R_h of 4–5 nm — aligning with industry standards (<10 nm). Both formulated MS-Hu6 and Hf2 also displayed acceptable PDI values — collectively indicating the maintenance of the monomeric state (also see Table 2). (C–E) Effect of formulated Hf2 on food intake (C), fat mass and body weight (D), and tissue weights of renal white adipose tissue (rWAT), subcutaneous WAT (sWAT), mesenteric WAT (mWAT), gonadal WAT (gWAT), and brown adipose tissue (BAT) (E). Groups of male C57BL/6J mice, matched for body weight and fed ad libitum on a high-fat diet, were injected with formulated Hf2 at different doses (10, 50, or 100 μg/d, 5 d/wk) or formulation buffer for 8 weeks. Net food intake was measured every other day, and body weight and quantitative nuclear magnetic resonance (qNMR) measurements were made weekly. Upon sacrifice, fat depots were collected and weighed. Statistics: Two-tailed Student’s t test vs. vehicle; n = 10 mice per group; mean ± SEM; *P ≤ 0.05, **P ≤ 0.05, ^#0.05 < P ≤ 0.1.

Figure 3. Hf2 protects against domain-specific and time-dependent memory loss in AD-prone mice.

(A) Groups of ovariectomized or aged AD-prone mice were injected with Hf2 and subjected to the novel object recognition (NOR) test for recognition memory and the Morris water maze (MWM) test for learning and spatial memory acquisition (training phase), and retrieval of consolidated spatial memory (probe trial). (B) A prevention protocol involved injection of Hf2 into ovariectomized 3xTg mice, initially for 8 weeks, and then for a further 12 weeks for a total of 20 weeks. (C–E) With Hf2, there was a significant (P = 0.048) increase in novel object interactions at 8 weeks, but not at 20 weeks (C), a significant (P = 0.002) reduction in latency to platform at 20 weeks, but not at 8 weeks (D), and no difference in spatial memory retrieval (time in the 40 cm platform zone) (probe trial) at either time point (E) (n = 7 mice per group). (F–I) In a treatment protocol, Hf2 was injected into 15-month-old 3xTg mice for 12 weeks (F); this revealed no effect on object interactions in the NOR test (G), a marked (P = 0.06) reduction in latency to platform (H), and no effect on the probe trial in the MWM test (I) (n = 6–7 mice per group). (J–L) A complementary treatment protocol in which Hf2 was injected into 18- to 22-month-old APP/PS1 mice for 12 weeks (J) revealed a significant (P = 0.03) increase in novel object interaction (K), but no difference in latency to platform (L) (n = 10–12 mice per group). Statistics: Two-tailed Student’s t test; mean ± SEM. AUC, area under the curve.

Figure 4. Acute and long-term safety of MS-Hu6 in African green monkeys.

Retired 18- to 23-year-old female monkeys (n = 4; IDs: 1132, 1136, 1139, 1241) were infused i.v. with MS-Hu6 (8 mg/kg), given over 1 minute, and subsequently received 4 further s.c. injections (8 mg/kg) 30 days apart (SQ1–SQ4). Vital signs (A) were monitored over 20 minutes after injection, and blood chemistries (B) and blood cell counts (C) were evaluated over 16 weeks. Age-appropriate reference ranges are shown as shaded areas (35). ALP, alkaline phosphatase; ALT, alanine aminotransferase; BUN, blood urea nitrogen; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; MCV, mean corpuscular volume.

Figure 5. Pharmacokinetics and biodistribution of MS-Hu6 in C57BL/6J mice.

(A) Pharmacokinetics of ⁸⁹Zr-labeled MS-Hu6 injected as a single s.c. dose of 250 μCi into groups of male and female C57BL/6J mice (n = 5 mice per group). Blood counts were corrected for decay and expressed as percentage of the injected dose per gram of blood. Mean pharmacokinetic parameters calculated by non-compartmental analysis after extravascular input using PK Solver v2.0 included β phase t_½, time to peak concentration (T_max), peak serum concentration (C_max), mean residence time (MRT), area under the curve up to 72 hours (AUC), volume of distribution (Vz/F), and apparent clearance (Cl/F) (also see Table 3). (B) For biodistribution studies, a single dose of ⁸⁹Zr-MS-Hu6 (250 μCi) was injected s.c. into groups of male C57BL/6J mice (n = 5 mice for each time point). Blood was drawn and tissues, including bone, subcutaneous and visceral WAT (sWAT and vWAT), brown adipose tissue (BAT), brain, kidney, liver, muscle, lung, heart, spleen, testis, adrenal gland, and pancreas, were isolated at 24, 48, and 72 hours. γ-Counts were corrected for decay and expressed as percentage of the injected dose per gram of tissue or blood.