ELISA

Direct binding ELISA was employed to assess the binding of ABBV-916 to different Aβ peptides. High-binding 96-well ELISA plates were coated with recombinant human AβpE3-42 fibrils, human Aβ1–40, human AβpE11-40 or murine Aβ1–40 peptides. After the plates were blocked with 2% BSA, dilutions of ABBV-916 were added to the plates. Anti-human IgG-HRP (Thermo Fisher Scientific) was used for ABBV-916 detection. The plates were incubated with TMB substrate (Life Technology). Reaction was stopped by 2 N sulfuric acid (VWR). Plates were read on a ClarioStar plate reader (Serial number 430–0469) to obtain the absorbance at 450 nm. Binding (EC50) of ABBV-916 to Aβ peptides was calculated using nonlinear regression (4-parameter dose-response curve model) in GraphPad Prism 8.4.3.

Binding to unfixed brain tissue

Unfixed human AD (n = 6) and non-AD tissue (n = 1) from Frontal Cortex were acquired from Folio Biosciences (Ohio, USA). AD brain samples were Braak Stage VI, with an age of individuals between 60 and 78 years old; 2 females and 4 males. All of the AD samples were confirmed as amyloid positive using anti-Aβ IHC. The non-AD tissue was from a 79 years old male at Braak Stage I. There were no plaques in the non-AD tissue as confirmed by 3D6 IHC (Supplementary Table 1). Twenty µm tissue sections were prepared using a Leica CM3050 S cryostat and were thaw-mounted onto glass slides. Sixteen µm unfixed brain sections from a 10-week-old Sprague Dawley rat and 10 µm unfixed brain sections from a 4.5-year-old cynomolgus monkey were acquired from Zyagen (California, USA). For unfixed rat brain sections, ABBV-916 was used as primary antibody and rabbit anti-human IgG-biotin (Southern Biotech) was used as secondary antibody. For unfixed human and cynomolgus monkey brain tissue, biotinylated ABBV-916 primary antibody was used. The immunoreactivity (IR) was visualized by incubation in ABC Elite (Vector labs) followed by diaminobenzidine (DAB, Vector labs). To calculate the EC50, unfixed AD tissue was incubated in 0, 0.5, 1.5, 4, 12, 35, and 100 µg/mL of biotinylated ABBV-916 primary antibodies. The images were acquired using a slide scanner (3DHistech, Pannoramic 250). The IR of biotinylated ABBV-916 in the AD samples (n = 6) were quantified using the Area Quantification module in HALOTM image analysis software v3.1.1076.423 (Indica Labs). A nonlinear regression analysis was carried out using GraphPad Prism Version 9.1.0 to determine the EC50.

Binding to human brain tissue homogenates and CSF

Brain samples from the middle frontal gyrus were obtained from the Netherlands Brain Bank (Amsterdam, NL). Brain cortex homogenates were prepared in TBS buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 20 mM NaF, 1 mM Na3VO4, 0.5 mM MgSO4 with protease inhibitor and phosphatase inhibitor) using a glass rod homogenizer and stored at − 80 °C. For formic acid (FA) treatment, brain homogenate and CSF samples were mixed with FA to 70% final concentration and sonicated using a Sonoplus Sonifier (Bandelin) for 24 cycles, 1 second each with 1 second pause and 60% amplitude. Samples were frozen on dry ice and stored at − 80 °C. Directly before use, FA treated samples were thawed and neutralized using 1 M Tris with 0.5 M Na2HPO4, and immediately diluted using assay specific sample diluent. AD brain samples reflected Braak stage V to VI and amyloid/neuritic plaque score C for neuropathologic change, with an age of individuals between 55 and 82 years old; 3 females and 2 males. Control brain samples reflected Braak stage zero to I and amyloid/neuritic plaque score zero to A, ages 60 to 93 years; 4 females and 1 male.

CSF samples were purchased from Precision Med LLC (CA, USA). Three distinct human CSF pools were used in this study, corresponding to: AD Aβ-positive individuals, ages 57–75 years, 3 females and 15 males; MCI Aβ-negative individuals with ages between 56 and 76 years, 9 females and 22 males; and control Aβ-negative individuals, ages 50-–77, 4 females and 18 males. Aβ-positivity was confirmed internally via CSF Aβ42/Aβ40 ratio. Samples were stored at −80˚ C and thawed directly before use.

A ligand binding assay for AβpE3 peptide was developed on the Singulex Erenna platform (EMD Millipore). For this purpose, ABBV-916 antibody was biotinylated using PierceTM Antibody Biotinylation Kit for IP (Thermo ScientificTM) and subsequently coated onto magnetic beads using SMCTM Capture Labeling Kit (Merck) at 25 µg antibody/mg beads according to manufacturer’s instructions. Anti-Aβ17–24 antibody clone 4G8 (BioLegend) was labeled with Alexa Fluor® 647 as detection antibody according to manufacturer’s instructions (Abcam). Biological samples and AβpE3-42 peptide (Bachem) used as calibrator were diluted using Standard Diluent (Merck). Immunoassay analysis was carried out using the Immunoassay Development kit (Merck). In brief, samples were diluted in Standard Diluent prior to incubation for 2 hr with the Capture antibody beads and for 1 hr with the Detection antibody at room temperature. In between, washing steps were performed using an automated plate washer (Tecan) with Wash Buffer plus Proclin (Merck). Prior to elution, beads were transferred to a new 96-well plate using the Viaflo96 pipetting robot (Integra) to minimize background signals. Elution of detection antibodies from the beads was done with Elution Buffer B (Merck) for 20 min at 25 °C. To increase robustness, neutralization and transfer of the elute to a 384-well scanning plate was performed in a single tip using the Viaflo96 pipetting robot. Signals were measured using the Erenna (Singulex) platform.

Phagocytosis assay

Unfixed, 20 μm coronal brain tissue sections were prepared from 21 month-old APPPS1-21 mice (B6.Cg-Tg(Thy1-APPSw,Thy1-PSEN1*L166P)21 Jckr)) [24] using a CM3050 S cryostat (Leica). Tissue sections were treated with 0, 1, 5, or 10 µg/mL ABBV-916 diluted in Dulbecco’s phosphate buffered saline [-Ca++-Mg++] (DPBS), incubated overnight at 4 °C (n=6), and washed with DPBS. Human induced pluripotent stem cell (hiPSC)-derived phagocytes, generated as described previously [14, 33], were directly cultured on ABBV-916-treated tissue and incubated for 72 hr at 37 °C/5% CO2. Tissue sections were then fixed in 4% paraformaldehyde and fibrillar plaques were labelled with 0.025% thioflavin S (ThioS, Sigma-Aldrich). Fluorescent images were acquired with a Pannoramic 250 Slide Scanner (3DHISTECH) and analyzed using HALO Image Analysis Software v3.1.1076.423 (Indica Labs). For measurement of fibrillar plaque area, fluorescent images from control-treated tissue sections were thresholded to highlight dense plaques, and subsequent tissue sections were thresholded with identical parameters. Data are reported as the measured area occupied by dense plaques (thresholded area) expressed as a percentage of total tissue area per section.

Animals

APPPS1-21 [24] mice were utilized under a licensing agreement with Koesler, Rottenburg, Germany. Cx3cr1-tdTomato mice were generated for AbbVie by GenOway (Lyon, France) as follows: CreERT2 and tdTomato were inserted at the Cx3xr1 locus, generating a recombined locus. The chimera mice were then intercrossed with Cre deleter mice, to excise the neomycin selection cassette, creating a fluorescent reporter mouse that stably expresses tdTomato in microglia under control of the endogenous Cx3cr1 locus. Cx3cr1-tdTomato mice were crossed with APPPS1-21 mice to generate Cx3cr1-tdTomato/APPPS1-21 mice.

For in vivo target binding, mice were bred at Charles River Laboratories (Sulzfeld, Germany) and delivered to the vivarium at AbbVie Deutschland GmbH and aged for experiments. For all other studies, breeding was conducted at Charles River Laboratories, Wilmington, DE, USA and mice were shipped to AbbVie Bioresearch Center Worcester, MA for aging before being transferred to the AbbVie Cambridge Research Center (CRC) for experiments. All animals were housed in a climate-controlled vivarium, where they were treated according to the standards recommended by AAALAC International and the NIH Guide for the Care and Use of Laboratory Animals (NIC publication No. 86–23, revised 1985). All animal experimental procedures conducted in the US were reviewed and approved by the AbbVie Institutional Animal Care and Use Committee (IACUC). All animal studies conducted in Germany were approved by the government of Rhineland Palatinate (Landesuntersuchungsamt) and conducted in accordance with the directive 2010/63/EU of the European Parliament and of the Council on the protection of animals used for scientific purpose, the ordinance on the protection of animals used for experimental or scientific purposes (German implementation of EU directive 2010/63; BGBl. I S. 3125, 3126), the Commission Recommendation 2007/526/EC on guidelines for the accommodation and care of animals used for experimental and other scientific purposes, the German Animal Welfare Act (BGBl. I S. 1206, 1313) amended by Art. 1 G from 17 December 2018 I 2586.

In vivo target binding

A single intravenous (IV) dose of ABBV-916 at 1, 5, 25, or 50 mg/kg was administered (n=2~3) to APPPS1-21 male and female mice. Seventy-two hrs post-dosing, animals were euthanized with Ketamine/Xylazine, perfused with cold PBS containing 1000 U/L heparin, and the brains rapidly dissected. The right hemibrains were drop-fixed in 10% formalin and processed for paraffin sectioning. ABBV-916 in the brain was detected via immunohistochemistry (IHC) using biotinylated donkey anti-human F(ab`)2 fragment detecting human IgG (H+L) (Jackson Immuno). The staining was processed using BOND RX autostainers (Leica Biosystems, Germany), and the slides were scanned using a Pannoramic 1000 scanner (3DHistech). On adjacent sections, total ABBV-916 target in plaques was determined by IHC using ABBV-916 as primary antibody. The % area covered by IR was quantified using HALOTM software (Indica Labs). Then the total target area occupied by injected ABBV-916 was calculated by (anti-hIgG IR/ABBV-916 IR) ×100%.

The level of ABBV-916 was measured in tissue lysates from the cerebral cortex of the left hemibrain using an electrochemiluminescence (Meso Scale Discovery) immunoassay detecting human IgG1. Briefly, the diluted sample was loaded on a streptavidin plate pre-coated with biotinylated antibody ab99757 (Abcam) capturing human IgG1, and human IgG1 was detected with sulfo-tagged EB89 antibody. ABBV-916 was used for the construction of standard curves and quality control samples which comprised the same concentration of biological matrix as the samples. A nonlinear regression analysis of normalized IR signal as a function of cortex antibody concentrations was carried out using GraphPad Prism 9.1.0.

Effects on Aβ pathology in APPPS1-21 mice

APPPS1-21 mice (4 months of age) were administered with the mouse chimeric IgG2a form of ABBV-916 (ABBV-916 chi, n=51, mixed gender) or a mIgG2a negative control antibody (n=52, mixed gender) via intraperitoneal (IP) injection at 40 mg/kg once weekly for 3 months. A group of untreated mice were euthanized at 4 months of age (n = 20, mixed gender) to obtain baseline pathology. At the end of dosing, mice were euthanized with pentobarbital, perfused transcranially with PBS, and the brain removed. The left hemisphere was drop-fixed in 10% formalin and processed for paraffin sectioning. Total amyloid plaques were stained using an anti-Aβ N-terminus antibody 3D6, and fibrillary plaques were stained using ThioS. The slides were scanned using Pannoramic 250 (3DHistech) and the % IR area in the cortex was quantified using HALOTM software (Indica Labs).

Two-photon microscopy study in APPPS1-21 mice

Cranial windows were implanted in 9–10 months old APPPS1-21 mice and 16 months old Cx3cr1-tdTomato×APPPS1-21 double transgenic mice. Briefly, mice were anesthetized with 2% (vol/vol) isoflurane and placed in a stereotactic apparatus. A 6 mm diameter piece of skull over the somatosensory cortex was drilled, removed, and replaced with an 8 mm diameter glass coverslip. An anodized aluminum chamber plate for mice with an observation hole (Tritech research) was bonded onto the head to enhance head stabilization during the imaging sessions. The body temperature was maintained at 37 °C throughout the procedure with a heating pad. Mice were given Meloxicam (4 mg/kg) the day of the surgery. Healing from the surgery was allowed for 6–7 weeks before the start of the imaging.

For Aβ plaque and cerebral amyloid angiopathy (CAA) imaging over time, mice with implanted windows were first dosed IV with Alexa Fluor 568 conjugated ABBV-916 antibodies to identify ABBV-916 targeted plaques. Only plaques pre-labeled with Alexa Fluor 568 ABBV-916 antibodies were quantified over the 8 weeks study. After the pre-label session, Methoxy-X04 (10 mg/kg) was injected IP to label all amyloid β plaques [17]. 72 hr after Methoxy injection, mice were reimaged in the exact same areas as for baseline (week 0). After week 0 imaging, mice were subjected to weekly IP injection of 40 mg/kg unlabeled ABBV-916 chi or negative control mIgG2a antibody. Mice were imaged at week 2, 4, 6 and 8. Methoxy-X04 (10 mg/kg) was injected IP before each imaging time point. For microglia recruitment over time, Methoxy-X04 (10 mg/kg) was injected IP 72 hr before the first imaging session to label Aβ plaques. After baseline imaging, either ABBV-916 chi or control mIgG2a were injected IP at 40 mg/kg, and images were acquired after 24 hr, 48 hr and 1 week. Then a second dose was injected IP at 40 mg/kg, and mice were imaged 24 hr, 48 hr and 1 week after the second dose.

At the time of imaging, fluorescein dextran (70,000 MW, 12.5 mg/mL in PBS, Molecular Probes) was retro-orbitally injected to create a fluorescent angiogram. In vivo imaging was performed on anesthetized mice (2% vol/vol isoflurane). Mice were placed in a stereotactic instrument with a chamber holder plate (Tritech Research), and their body temperature was maintained at 37 °C with a heating pad. Images of Aβ plaques, CAA, microglia and vessels were obtained using an Olympus FVMPE-RS Multiphoton Laser Scanning Microscope and an Olympus 25 x dipping objective (NA = 1.05). A Mai Tai HP DS-OL (Mai Tai-DeepSee 690–1040 nm 100 fs) and an Insight X3-OL (Insight Laser 1 line 680–1300 nm) generated two-photon excitation at 800 nm and 1100 nm respectively. Four photomultiplier tubes (PMTs) collected emitted light in the range of 410–460nm, 495–540nm, 575–645 nm and 660–750 nm. PMT settings and laser power remained unchanged throughout the imaging sessions. Laser power was kept below 50 mW to avoid phototoxicity. Four to six randomly chosen cortical volumes (z-series, 2 µm step size, 127 µm x 127 µm, 100–250 µm depth) were acquired per mouse (512 x 512 pixels, 1x or 3x zoom). To avoid visual artifacts induced from surgery and window placement, analyzed plaques were at a depth of 50 µm or deeper from the pial surface.

Images were exported and the 3-dimensional (3D) reconstructed confocal z-stacks were processed using IMARIS version 9.3.1 (Bitplane). For plaque volume quantification over time, the different time point images (prelabeled, week 0 to week 8) were added up and aligned. Background subtraction was selected for thresholding, and the split touching object function was enabled. 10–20 plaques were selected per field of view (FOV), and 3–4 FOV were included per mouse. The “track surfaces” function was used, and the volume of every individual plaque was followed over time. Only plaques prelabeled with Alexa Fluor 568 conjugated ABBV-916 antibodies were selected for quantification. For CAA quantification, a tubular surface was first created around the artery. Then, the methoxy channel was masked within the tubular surface, to avoid any non-CAA labeling being quantified. Background subtraction was selected for thresholding, and the seed point diameter function was enabled. For microglia engagement study, microglia volume around a plaque and microglia total volume were quantified using the “surfaces” function. Briefly, the default Imaris Surfaces function was used to create individual Region of Interest (ROI) of 20–30 µm from the edge of the plaque, and the volume occupied by microglia within the ROI was quantified at every time point. Absolute intensity was used for thresholding. 1–5 plaques were included in the FOV. Images represented in the figures are either single slices or maximum intensity image projections of the 3D volumes.

Acute microhemorrhage study in APPPS1-21 mice

The microhemorrhage study was performed in APPPS1-21 mice at the age of 15–17 months. At this age, AβpE3 is detectable in the cerebral vasculature in this mouse model. Mice (n = 4, mixed gender) were given a single IP injection of the mouse IgG2a precursor of ABBV-916 at 40 mg/kg. A non-targeted mIgG2a (n=4, mixed gender) was used as negative control and murine precursor mIgG2a antibody representing bapineuzumab, 3D6 mIgG2a, was used as positive control (n=5) at the same dose. Three days after injection, or as soon as severe adverse effects were observed (whichever came first), mice were euthanized with sodium pentobarbital, perfused with PBS, and the brains were drop-fixed in 10% formalin for 24 hr. Once fixed the brains were cryoprotected in 30% sucrose and serial coronal 50 µm sections were collected from the right hemisphere of the brain using a freezing sliding microtome. For each type of histology analysis, every 12 th section (600 µm interval) was mounted on glass slides. For red blood cell quantification, a Hematoxylin counterstain was performed. The hemosiderin accumulated in the brain after microhemorrhage was detected by Prussian Blue staining with Congo-Red counterstain. Briefly, after incubation in 80% ethanol containing saturated NaCl and 0.01M NaOH, the sections were incubated in 80% ethanol containing saturated NaCl, 0.01M NaOH and 0.5% Congo Red. The slides were then washed in 80% ethanol and normal saline followed by incubation in 2% potassium ferrocyanide in 0.12M HCl. After washing in PBS, the slides were dehydrated, coverslipped, and scanned using Pannoramic 250 (3DHistech) and quantified using HALOTM software (Indica Labs).