Adeno-associated virus production and in vivo transduction

Single-stranded adeno-associated virus (ssAAV) vector backbones with AAV2 inverted terminal repeats (ITRs) were kindly provided by B. Schneider (EPFL). Herein, expression of the monomeric NeonGreen fluorophore was driven by the human Synapsin I (hSynI) promoter. A P2A sequence (GSGATNFSLLKQAGDVEENPGP) was introduced between mNG and PrPC for bi-cistronic expression. For mPrPR207A and mPrP2cys expression, a synthetic gene block (gBlock, IDT, full sequence is given in the Supplementary Information) was cloned between the BsrGI and HindIII site of the vector replacing the wild-type PrPC sequence. Recombination of plasmids was tested using SmaI digestion prior to virus production. The viral vectors and viral vector plasmids were produced as hybrid AAV2/6 (AAV6 capsid with AAV2 ITRs) by the Viral Vector Facility (VVF) of the Neuroscience Center Zurich. The identity of the packaged genomes was confirmed by Sanger DNA-sequencing (identity check). Quantification of mNG-positive cells from confocal images was done using the Spots function in Imaris (Bitplane).

Neurotropic AAV variants for scFv antibody expression were constructed from a synthetic gene fragment, NheI-IL2-scFv-Myc-EcoRV (produced by Genscript Biotech), which contained hcY104A sequences preceded by the signal peptide from interleukin-2 (IL-2)23. NheI and EcoRV restriction-enzyme digestion was performed on NheI-IL2-scFv-Myc-EcoRV synthetic gene fragments, which were then inserted into a ssAAV vector backbone. ScFv expression was under the control of the strong, ubiquitously active CAG promoter. A WPRE (woodchuck hepatitis virus post-transcriptional regulatory element) sequence was also included, downstream of the transgene, to enhance transgene expression. Production, quality control and determination of vector titer was performed by ViGene Biosciences). Rep2 and CapPHP.B plasmids were provided under a Material Transfer Agreement (MTA). Further details about packaging and purification strategies can be found on the company’s website (http://www.vigenebio.com).

Allen Mouse Brain Atlas data

Images from in situ hybridization for calbindin 1 and synapsin 1 expression were taken from the Allen Mouse Brain Atlas (www.brain-map.org). The first dataset retrieved by the R package allenbrain (https://github.com/oganm/allenBrain) with the closest atlas image to the center of the region (regionID = 512, settings: planeOfSection = ‘coronal’, probeOrientation = ‘antisense’) was downloaded (dataset ID nos. for calb1 = 71717640, syn1 = 227540).

Animals and in vivo experiments

We conducted all animal experiments in strict accordance with the Swiss Animal Protection law and dispositions of the Swiss Federal Office of Food Safety and Animal Welfare (BLV). The Animal Welfare Committee of the Canton of Zurich approved all animal protocols and experiments performed in this study (animal permits 123, ZH90/2013, ZH120/16, ZH139/16). Genetically modified mice from the following genotypes were used in this study: Zurich I mice homozygous for disrupted Prnp genes (Prnpo/o, denoted as PrnpZH1/ZH1)5, Zurich III Prnpo/o (denoted as PrnpZH3/ZH3)15 and tga20 (ref. 16).

For in vivo transduction with the neurotropic AAV-PHP.B construct, mice received a total volume of 100 µL (1 × 1011 total vector genomes) by intravenous injection into the tail vein. Fourteen days after AAV transduction, the left hemispheres of Tga20 mice were innoculated with 30 μL of 0.1% RML6 brain homogenate, corresponding to 3 × 105 median lethal dose (LD50; 3.6 μg of total brain homogenate). Brain homogenates were prepared in 0.32 M sucrose in PBS at a concentration of 10% (wt/vol). Protein analysis of mouse brains is described below.

After fixation with 4% paraformaldehyde for 1 week, tissues were treated with concentrated formic acid for 60 minutes, fixed again in formalin and eventually embedded in paraffin. HE staining and SAF84 immunohistochemistry were performed as previously described24. For immunohistochemical detection of Myc-tag, tissue was deparaffinized and incubated in citrate buffer (pH 6.0) in a domestic microwave for 20 minutes. Unspecific reactivity was blocked using blocking buffer (10% goat serum, 1% bovine serum albumin, 0.1% Triton X-100 in PBS) for 1 hour at room temperature. Primary rabbit anti-Myc-tag antibody (1:500, ab9106, Abcam, overnight at 4 °C) was detected with Alexa Fluor 594 rabbit anti-goat (IgG) secondary antibody (1:1,000, 1 hour at room temperature), diluted in staining buffer (1% bovine serum albumin, 0.1% Triton X-100 in PBS). Tissue was counterstained with DAPI (5 µg/ml, 15 minutes at room temperature).

Cell lines

CAD5 is a subclone of the central nervous system catecholaminergic cell line CAD showing particular susceptibility to prion infection10. Generation of the CAD5 Prnp–/– clone no. C12 was described before, as was overexpression of murine, full-length PrPC in CAD5 Prnp–/– by cloning the open reading frame of Prnp into the pcDNA3.1(+) vector, Prnp expression was driven by a constitutively expressed cytomegalovirus promoter (yielding pcDNA3.1(+)-Prnp) as described earlier13. For stable expression of mPrP2cys, pcDNA3.1(+)-Prnp vector was modified using Quikchange II Site-Directed Mutagenesis Kit (Agilent), according to the manufacturer’s guidelines. We first introduced a mutation leading to p.R207C (primers (5′→3′): mutagenesis forward: GTG-AAG-ATG-ATG-GAG-TGC-GTG-GTG-GAG-CAG-A, reverse: TCT-GCT-CCA-CCA-CGC-ACT-CCA-TCA-TCT-TCA-C) which was then followed by the p.I138C mutation (mutagenesis forward: AGT-CGT-TGC-CAA-AAT-GGC-ACA-TGG-GCC-TGC-TCA-TGG, reverse: CCA-TGA-GCA-GGC-CCA-TGT-GCC-ATT-TTG-GCA-ACG-ACT). For stable expression of mPrPR207A, pcDNA3.1(+)-Prnp was mutated correspondingly (mutagenesis forward: TGT-GAA-GAT-GAT-GGA-GGC-CGT-GGT-GGA-GCA-GAT-G, reverse: TCT-GCT-CCA-CCA-CGC-ACT-CCA-TCA-TCT-TCA-C).

Cell vacuolation assay

Mouse hypothalamic Gt1 neuronal cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) in the presence of 10% fetal bovine serum (FBS), 1% penicillin–streptomycin and 1% glutamax (all obtained from Invitrogen). For prion infection of the cells, Gt1 cells grown in DMEM were incubated with either Rocky mountain laboratory strain of prion (RML6) prions (0.1%) or non-infectious brain homogenate (NBH; 0.1%) for 3 days in 1 well of a 6-well plate. This was followed by splitting the cells at a 1:3 ratio every 3 days for at least 10 passages. The presence of infectivity in the cells was monitored by the presence of PK-resistant PrP, as described below. At 70 days postinfection (dpi), the cells started developing vacuoles, which were visualized by phase-contrast microscopy. Antibody treatment with hcY104A was administered on 70–75 dpi at a concentration of 180 nM.

Cerebellar organotypic slice cultures

Mice from C57BL/6, tga20, PrnpZH1/ZH1 and PrnpZH3/ZH3 strains were used for preparation of COCS, as described14. Herein, 350-µm-thick COCS were prepared from 9- to 12-day-old pups. Free-floating sections of COCS were infected with 100 µg prions per 10 slices of RML6 (passage 6 of the Rocky Mountain Laboratory strain mouse-adapted scrapie prions) or 22L (mouse-adapted scrapie prions) brain homogenate from terminally sick, prion-infected mice. Brain homogenate from CD1-inoculated mice was used as non-infectious brain homogenate (NBH). Sections were incubated with brain homogenates diluted in physiological Grey’s Balanced Salt Solution for 1 hour at 4 °C and washed, and 5–10 slices were placed on a 6-well PTFE membrane insert. Analogously, for AAV experiments, COCS were incubated with AAV at a final concentration of 5.2 × 1010 total vector genomes diluted in physiological Grey’s Balanced Salt Solution for 1 hour at 4 °C, then washed and placed on PTFE membrane inserts. Antibody treatments were given thrice weekly, e.g. with every medium change. In naive slices, antibody treatments were initiated after a recovery period of 10–14 days.

For testing of innocuity of pomologs (Fig. 2c, Supplementary Fig. 7c and Supplementary Fig. 10), POM1 and pomolog antibodies were added at 400 nM for 14 days. Supplementary Figures 7c and 10 represent aggregated data from multiple experiments with COCS from mice of identical genotype and age; compounds were administered at identical timepoints and dosage. When added to RML-infected tga20 COCS (Fig. 2d and Supplementary Fig. 7d), hcY57A was added from 20 to 45 dpi, hcY104A was added from 21 to 45 dpi and both antibodies were given at 400 nM. Antibody treatment with hcY57A and hcY104A of RML-infected tga20 COCS used for determination of PrPSc by western blot, see detailed protocol below, was initiated and stopped at 21 dpi and 45 dpi, respectively. hcD55A was added to RML-infected tga20 COCS at either 1 (800 nM, Supplementary Fig. 12d) or 21 (400 nM, Fig. 2d and Supplementary Fig. 7d) dpi. When added to C57BL/6 COCS (Fig. 2e and Supplementary Fig. 7e), hcY104A was added from 1 dpi at 400 nM until 45 dpi. In 22L-inoculated COCS, hcY104A was administered at 21 dpi, and slices were collected at 44 dpi. Phage-derived Fabs were added to RML-infected COCS (Fig. 4b–g) from 1 dpi until 45 dpi at 550 nM.

ELISA

PrPC levels were measured by ELISA using monoclonal anti-PrPC antibody pairs POM19/POM3 or POM3/POM2 (all as holo-antibodies), as described previously25. First, 384-well SpectraPlates (Perkin Elmer) were coated with 400 ng mL−1 POM19 (POM3) in PBS at 4 °C overnight. Plates were washed 3 times in 0.1% PBS-Tween 20 (PBS-T) and blocked with 80 µL 5% skim milk in 0.1% PBS-T per well for 1.5 hours at room temperature. Blocking buffer was discarded, and samples and controls were dissolved in 1% skim milk in 0.1% PBS-T for 1 hour at 37 °C. Twofold dilutions of rmPrP23–231, starting at a dilution of 100 ng/mL in 1% skim milk in 0.1% PBS-T, were used as a calibration curve. Biotinylated POM3 (POM2) was used to detect PrPC (200 ng/mL in 1% skim milk in 0.1% PBS-T), and biotinylated antibodies were detected with streptavidin-HRP (1:1,000 in 1% skim milk in 0.1% PBS-T, BD Biosciences). Chromogenic reaction and reading of plates were performed as described in ref. 25. Unknown PrPC concentrations were interpolated from the linear range of the calibration curve using linear regression (GraphPad Prism, GraphPad Software).

ELISA screening of phage display

Single colonies were picked and cultured in a 384-well plate (Nunc) in 2YT, ampicillin and 1% glucose medium overnight at 37 °C, 80% humidity, 500 r.p.m. These precultures were used to prepare glycerol stock master plates. Expression plates were prepared from the master plates by inoculating corresponding wells with 2YT, carbenicillin and 0.1% glucose medium, followed by induction with 1 mM IPTG. After 4 hours at 37 °C, 80% humidity, cultures were lysed for 1.5 hours at 400 r.p.m., 22 °C in borate-buffered saline, pH 8.2, containing EDTA-free protease inhibitor cocktail, 2.5 mg/mL lysozyme and 40 U/mL benzonase. Fab-containing bacteria lysate was blocked with Superblock and used for ELISA screening, and the reactivity to four different antigens was assessed in parallel. The following antigens were coated on separate 384-well ELISA plates: anti-Fd antibody (The Binding Site) 1:1,000 in PBS, to check the expression level of each Fab clone in bacteria; rmPrP23–231 at 87 nM in PBS, to identify candidate PrPC binders; mPrP2cys at 87 nM in PBS, to check for cross reactivity with mPrP2cys; neutravidin at 87 nM as a control for specificity. Antigen-coated ELISA plates were washed twice with PBS-T and blocked with Superblock for 2 hours. Fab-containing bacteria lysates from the expression plate were transferred to corresponding wells of the ELISA plates. After 2 hours of incubation, ELISA plates were washed 3 times with PBS-T, and anti-human-F(ab′)2-alkaline-phosphatase-conjugated antibody (1:5,000 in PBS-T) was added. After 1 hour of incubation at room temperature, followed by 3 washings with PBS-T, pNPP substrate was added and, after 5 min incubation, the ELISA signal was measured at 405 nm. Fabs from bacteria lysates producing an ELISA signal 5 times higher than the technical background, which was calculated as the average of the coated well containing un-inoculated medium, and negative for neutravidin were considered as PrPC binder candidates. For hit selection, we considered only anti-PrPC Fabs whose ELISA signal for rmPrP23–231 was at least two times higher than for mPrP2cys. All the identified hits were checked in a confirmatory ELISA screening. Bacterial cultures of the selected clones were used for DNA minipreps followed by Sanger sequencing using the following sequencing primers: HuCAL_VH (5′-GATAAGCATGCGTAGGAGAAA-3′) and M13Rev (5′-CAGGAAACAGCTATGAC-3′).

Expression and purification of selected anti-PrP Fabs

Chemically competent BL21(D3) cells (Invitrogen) were transformed with selected pPE2-Fab plasmids and grown on plates with LBagar, kanamycin and 1% glucose. A single colony was inoculated into 20 mL of 2×YT, kanamycin and 1% glucose pre-culture medium and incubated for at least 4 hours at 37 oC, 220 r.p.m. One liter of 2YT medium containing kanamycin and 0.1% glucose was inoculated with 20 mL pre-culture, and Fab expression was induced by 0.75 mM IPTG followed by incubation overnight at 25 oC, 180 r.p.m. The overnight culture was centrifuged at 4,000g at 4 oC for 30 minutes, and the pellet was frozen at −20 oC. For Fab purification, the thawed pellet was resuspended into 20 mL lysis buffer (0.025 M Tris pH 8; 0.5 M NaCl; 2 mM MgCl2; 100 U/mL benzonase (Merck); 0.25 mg/mL lysozyme (Roche), EDTA-free protease inhibitor (Roche)) and incubated for 1 hour at room temperature at 50 r.p.m. The lysate was centrifuged at 16,000g at 4 oC for 30 minutes, and supernatant was filtrated through 0.22-μM Millipore Express Plus Membrane. Fab purification was achieved via the His6-Tag of the heavy chain by IMAC. Briefly, after equilibration of the Ni-NTA column with running buffer (20 mM Na-phosphate buffer, 500 mM NaCl, 10 mM imidazole, pH 7.4), and the bacteria lysate was loaded and washed with washing buffer (20 mM Na-phosphate buffer, 500 mM NaCl, 20 mM Imidazole, pH 7.4). The Fab was eluted with elution buffer (20 mM Na-phosphate buffer, 500 mM NaCl, 250 mM imidazole, pH 7.4). Buffer exchange was performed using PD-10 columns, Sephadex G-25M (Sigma), whereby the Fab was eluted with PBS.

Förster resonance energy transfer

Europium (Eu3+) donor fluorophore was coupled to POM1 (yielding POM1-Eu3+) and allophycocyanin (APC) acceptor fluorophores was coupled to holoantibody POM3 (yielding holo-POM3-APC) as previously described26. Full-length, recombinant mouse prion protein (rmPrP23–231) was added at a final concentration of 1.75 nM, followed by addition of holo-POM3-APC at a final concentration of 5 nM and subsequent incubation at 37 °C for 30 minutes with constant shaking at 400 r.p.m. Pomologs were then added in serial dilutions from 0 to 3 nM and were again incubated at 37 °C for 60 minutes with constant shaking at 400 r.p.m., followed by addition of POM1-Eu3+ at a final concentration of 2.5 nM). Net Förster resonance energy transfer (FRET) was calculated as described previously26.

Determination of binding constants from Förster resonance energy transfer

The dependence of the FRET signal on POM1 concentration was modelled by a simple competitive binding model. The binding constant of the FRET-labeled POM1-Eu3+ was defined as:

$$K_F = \frac}} \right]\times[F_}]}}}]}} = \frac}} \right] - \left[ }} \right] - \left[ }} \right]} \right)\times\left( }} \right] - [F_}]} \right)}}}]}}$$

where square brackets denote concentration, Ftot, Ffree and Fb denote total, free and bound POM1-Eu3+, PrPtot and PrPfree denote the total and free PrP, Atot, Afree and Ab denote total, free and bound single-chain fragment variables (scFvs) and KF is the binding constant of POM1-Eu3+. The righthand equality is obtained by imposing conservation of mass. An equivalent equation defines the binding constant of the scFvs:

$$K_} = \frac}} \right]\times[A_}]}}}]}} = \frac}} \right] - \left[ }} \right] - \left[ }} \right]} \right)\times\left( }} \right] - [A_}]} \right)}}}]}}$$

This system of equations is solved to give Fb as a function of Atot. To relate the concentration of bound POM1-Eu3+, Fb, to the FRET measurements, this equation was rescaled to 100 for the fully bound and 10 for the fully unbound limit. An additional complication in interpreting the experimental data stems from the fact that a FRET signal will appear only if both a POM1-Eu3+and holo-POM3-APC are bound to the same PrP. We assume that the binding of POM1 and POM3 is independent, so we can approximate the concentration of PrP bound to a holo-POM3-APC as the effective PrP concentration, PrPtot in the above equations. The binding constant of holo-POM3-APC was determined to be 0.23 nM, giving an effective concentration of PrP of 1.64 nM (compared with the total PrP concentration of 1.75 nM). To verify the robustness of these results, we fitted the data assuming a much weaker binding of holo-POM3-APC, with a binding constant of 1 nM. The obtained KD values of the single-chain fragments were within the error of the ones determined with a holo-POM3-APC binding constant of 0.23 nM.

Immunohistochemical stainings and analysis of immunofluorescence

COCS were washed twice in PBS and fixed in 4% paraformaldehyde for at least 2 days at 4 °C and were washed again twice in PBS prior to blocking of unspecific binding by incubation in blocking buffer (0.05% Triton X-100 vol/vol, 0.3% goat serum vol/vol in PBS) for 1 hour at room temperature. For visualization of neuronal nuclei, the monoclonal mouse anti-NeuN antibody conjugated to Alexa Fluor 488 (clone A60, Life Technologies) was dissolved at a concentration of 1.6 µg mL−1 into blocking buffer and incubated for 3 days at 4 °C. Further primary antibodies used were recombinant anti-calbindin antibody (1 µg mL−1, ab108404, Abcam), anti-glial fibrillary acidic protein (1:500, Z0334, DAKO) and anti-F4/80 (1 µg mL−1, MCAP497G, Serotec). Unconjugated antibodies were dissolved in blocking buffer and incubated for 3 days at 4 °C. After 3 washes with PBS for 30 minutes, COCS were incubated for 3 days at 4 °C with secondary antibodies Alexa-Flour-594-conjugated goat anti-rabbit-IgG (Life Technologies) or Alexa-Fluor-647-conjugated goat anti-rat-IgG (Life Technologies) at a dilution of 1:1,000 in blocking buffer. Slices were then washed with PBS for 15 minutes and incubated in DAPI (1 µg mL−1) in PBS at room temperature for 30 minutes to visualize cell nuclei. Two subsequent washes in PBS were performed, and COCS were mounted with fluorescence mounting medium (DAKO) on glass slides. NeuN, GFAP, F4/80 and calbindin morphometry was performed by image acquisition on a fluorescence microscope (BX-61, Olympus), and analysis was performed using gray-level auto thresholding function in ImageJ (www.fiji.sc). Cell numbers in Figure 2f were determined using the ‘Spots’ function in Imaris (Oxford Instruments). Morphometric quantification was done on unprocessed images with identical exposure times and image thresholds between compared groups. Representative fluorescent micrographs in the main and supplementary figures have been processed (linear adjustment of brightness and contrast) for better interpretability.

For immunohistochemistry of CAD5 cells, cells were seeded on 18-well µ-slides (Ibidi) and fixed with 4% paraformaldehyde for 5 minutes at room temperature. Unspecific reactions were blocked using 3% goat serum in PBS for 1 hour at room temperature. Mouse monoclonal anti-PrPC antibodies POM1, POM5, POM8 and POM19 (all holo-antibodies) were established before25; POM antibodies were incubated at 4 µg mL−1 in 3% goat serum in PBS at 4 °C, followed by 3 washes in PBS. Antibodies were detected using Alexa-Fluor-488-conjugated goat anti-mouse-IgG at 1:250 dilution, followed by nuclear counterstain with DAPI (1 µg mL−1 in PBS) for 5 minutes at room temperature. Image analysis was performed using SP5 confocal microscope (Leica) with identical exposure times across different experimental groups.

In vitro toxicity assessment

Quantification of POM1 toxicity on CAD5 Prnp–/– cells stably transfected with mPrPC, mPrPCR207A or empty control vector, as described above, was measured as percentage of PI-positive cells using flow cytometry, as described before11.

CAD5 cells were cultured with 20 mL Corning Basal Cell Culture Liquid Media–DMEM and Ham’s F-12, 50/50 Mix, supplemented with 10% FBS, Gibco MEM Non-Essential Amino Acids Solution 1×, Gibco GlutaMAX Supplement 1× and 0.5 mg/mL of Geneticin in T75 Flasks (Thermo Fisher) at 37 °C, 5% CO2. Sixteen hours before treatment, cells were split into 96-well plates at 25,000 cells/well in 100 μL.

POM1 alone was prepared at 5 μM final concentration in 20 mM HEPES pH 7.2 and 150 mM NaCl, and 100 μL of each sample, including buffer control, was added to CAD5 cells, in duplicates.

After 48 hours, cells were washed 2 times with 100 μL MACS buffer (PBS + 1% FBS + 2 mM EDTA) and resuspended in 100 μL MACS buffer. Thirty minutes before FACS measurements, PI (1 μg/mL) was added to the cells. Measurements were performed using a BD LSRFortessa. The percentages of PI-positive cells were plotted in columns as mean with s.d. The gating strategy is depicted in Extended Data Figure 3a.

In vivo toxicity assessment

The in vivo toxicity assessment was performed as previously described9. In brief, mice where i.c. injected using a motorized stereotaxic frame (Neurostar) at the following bregma coordinates (AP –2 mm, ML ±1.7 mm, DV 2.2mm, angle in ML/DV plane 15°). Antibodies (2 μL) were injected at a flow rate of 0.5 μL/minute. After termination of the injection, the needle was left in place for 3 minutes.

Twenty-four hours after stereotactic injection, mice were placed on a bed equipped with a mouse whole-body radio frequency transmitter coil and a mouse head surface-coil receiver and then transferred into the 4.7 Bruker Pharma scan. For DWI, routine gradient echo sequences with the following parameters were used: TR: 300 ms TE: 28 ms, flip angle: 90°, average: 1, matrix: 350 × 350, field of view: 3 × 3 cm, acquisition time: 17 minutes, voxel size: 87 × 87 μm3, slice thickness: 700 μm, isodistance: 1,400 μm3 and b values: 13,816 s/mm2. Finally, mice were euthanized after 49 hours, and the brains were fixed in 4% formalin. Coronal sections from the posterior cortex were paraffin-embedded (4 mm) and 2-μm coronal step sections (standard every 100 μm) were cut, deparaffinized and routinely stained with hematoxylin and eosin.

Dose–response analysis and the benchmark dose relation were calculated with benchmark dose software (BMDS) 2.4 (United States Environmental Protection Agency).

Molecular dynamics

Experimental structures were used as a basis for MD simulations when available (scPOM1–mPrP complex, Protein Data Bank (PDB) 4H88; free mPrP, PDB 1XYX). The structures of full-length mPrP, mPrP∆90–231 and the pomologs were predicted by homology modeling on the I-Tasser webserver27, on the basis of the experimental structure of the PrP globular domain (aa 120–231), and were further validated with MD.

In all simulations, the system was initially set up and equilibrated through standard MD protocols: proteins were centered in a triclinic box, 0.2 nm from the edge, and filled with SPCE water model and 0.15 M Na+Cl– ions using the AMBER99SB-ILDN protein force field; energy minimization followed. Temperature (298 K) and pressure (1 bar) equilibration steps (100 ps each) were performed. Three independent replicates of 500-ns MD simulations were run with the above-mentioned force field for each protein or complex. MD trajectory files were analyzed after removal of periodic boundary conditions. The overall stability of each simulated complex was verified by root mean square deviation, radius of gyration and visual analysis, according to standard procedures. Structural clusters, atomic interactions and root mean square fluctuation (RMSF) were analyzed using GROMACS28 and standard structural biology tools. RMSF provides a qualitative indication of residue level flexibility, as shown in Figure 1c.

The presence of H-bonds or other interactions between GD residues was initially estimated by visual analysis and then by distance between appropriate chemical groups during the simulation time.

Nuclear magnetic resonance

Spectra were recorded on a Bruker Avance 600 MHz NMR spectrometer at 298 K, pH 7 in 50 mM sodium phosphate buffer at a concentration of 300 µM. In mapping experiments, mPrP was uniformly labeled with 15N (99%) and 2H (approx. 70%); antibodies were unlabeled. PrP and antibody samples were freshly prepared and extensively dialyzed against the same buffer prior to complex formation. The same procedure was followed for CD measurements. Chemical-shift assignment was based on published data (Biological Magnetic Resonance Data Bank entry 16071)29. Briefly, overlay of [15N,1H]-TROSY spectra of free or bound mPrP90–231 allowed for identification of PrP residues for which the associated NMR signal changed upon complex formation, indicating alterations in their local chemical environment19.

Phage display

A synthetic human Fab phagemid library (Novartis Institutes for BioMedical Research) was used for phage display. First, two rounds of selection against PrPC were performed by coating 96-well Maxisorp plates (Nunc) with a decreasing amount of rmPrP23–231 (1 µM and 0.5 µM respectively, in PBS), overnight at 4 °C. PrP-coated plates were washed 3 times with PBS-T and blocked with Superblock for 2 hours. Input of 4 × 1011 phages in 300 µL PBS was used for the first round of panning. After 2 hours of blocking with Chemiblocker (Millipore), the phages were incubated with PrP-coated wells for 2 hours at room temperature. The non-binding phages were then removed by extensive washing with PBS-T, while rmPrP23–231-bound phages were eluted with 0.1 M glycine/HCl, pH 2.0 for 10 minutes at room temperature; the pH was then neutralized by 1 M Tris pH 8.0. Eluted phages were used to infect exponentially growing amber suppressor TG1 cells (Lubio Science). Infected bacteria were cultured in 2YT, carbenicillin and 1% glucose medium overnight at 37 °C, 200 r.p.m., and superinfected with VCSM13 helper phages. The production of phage particles was then induced by culturing the superinfected bacteria in 2YT, carbenicillin and kanamycin medium containing 0.25 mM isopropyl β-d-1-thiogalactopyranoside (IPTG), overnight at 22 °C, 180 r.p.m. Supernatant containing phages from the overnight culture was used for the second panning round. Output phages from the second round were purified by PEG/NaCl precipitation, titrated and used in the following third rounds to enrich phage-displayed Fabs that bound preferentially mPrPC over mPrP2cys.

Two strategies were used: depletion of binders to recombinant mPrP2cys by subtraction in solid phase and depletion of mPrP2cys binders by competition with rhPrPC23–230-AviTag in liquid phase. In the former setting, purified phages were first exposed to 0.75 µM mPrP2cys (threefold molar excess compared with rmPrP90–231 or rmPrP121–231), and then the unbound phages were selected for rmPrP90–231 or rmPrP121–231 binders. Alternatively, purified phages were first adsorbed on neutravidin-coated wells to remove the neutravidin binders and then exposed to 0.25 µM rhPrPC23–230-AviTag in solution in the presence of 0.75 µM (threefold molar excess) of mPrP2cys. The phage-displayed Fabs binding to rhPrPC23–230-AviTag were captured on neutravidin-coated wells and eluted as described above. For both strategies, a fourth panning round was performed using 0.3 µM mPrP2cys for depletion and 0.1 µM rmPrP121–231 (coated on the plate) or rhPrPC23–230-AviTag (in solution) for positive selection. At the fourth round of selection, DNA minipreps were prepared from the panning output pools by QIAprep Spin Miniprep kit (Qiagen) and the whole anti-PrP Fab enriched library was subcloned in expression vector pPE2 (kindly provided by Novartis). DNA was then used to transform electrocompetent non-amber suppressor MC1061 bacteria (Lubio Science) to produce soluble Fabs and perform ELISA screening.

Production of recombinant proteins and antibodies

Bacterial production of recombinant, full-length mouse PrP23–231, recombinant fragments of human and mouse PrP and recombinant, biotinylated human PrPC-AviTag (rhPrPC23–230-AviTag) was achieved as previously described30,31,32. Production of scFv and the IgG POM1 antibodies used in this manuscript was performed as described before25. Production of holo-hcY104A was performed as follows: POM1 IgG1 heavy chain containing a Y104A mutation and POM1 kappa light chain were ordered as a bicistronic synthetic DNA block (gBlock, IDT) separated by a P2A site. The synthetic gene block (gBlock, IDT, see full sequence in the Supplementary Information) was then cloned into pcDNA 3.4-TOPO vector (Thermo Fisher Scientific), and recombinant expression was achieved using the FreeStyle MAX 293 Expression System (Thermo Fisher Scientific), according to the manufacturer’s guidelines. Glucose levels were kept constant over 25 mM. Seven days after cell transfection, medium supernatant was collected, centrifuged and filtered. A Protein-G column (GE Healthcare) was used for affinity purification of antibodies, followed by elution with glycine buffer (pH 2.6) and subsequent dialysis against PBS (pH 7.2–7.4). Purity was determined by SDS–PAGE, and protein concentrations were determined using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). For generation of POM1 mutants, we performed site-directed mutagenesis on a POM1 pET-22b(+) (Novagen) expression plasmid7 according to the manufacturer’s guidelines (primers (5′→3′): hcW33A: forward (FW): CATTCACTGACTACGCGATGCACTGGGTGAAGC, reverse (REV): GCTTCACCCAGTGCATCGCGTAGTCAGTGAATG. hcD52A: FW: GAGTGGATCGGATCGATTGCGCCTTCTGATAG, REV: CTATCAGAAGGCGCAATCGATCCGATCCACTC. hcD55A: FW GGATCGATTGATCCTTCTGCGAGTTATACTAGTCAC, REVGTGACTAGTATAACTCGCAGAAGGATCAATCGATCC. hcY57A: FW: CCTTCTGATAGTGCGACTAGTCACAATGAAAAGTTCAAGG, REV: CCTTGAACTTTTCATTGTGACTAGTCGCACTATCAGAAGG. lcS32A: FW: CCAGTCAGAACATTGGCACAGCGATACACTGGTATCAGCAAAG, REV: CTTTGCTGATACCAGTGTATCGCTGTGCCAATGTTCTGACTGG. lcY50A: FW: CTCCAAGGCTTATCATAAAGGCGGCTTCTGAGTCTATCTCTGG, REV: CCAGAGATAGACTCAGAAGCCGCCTTTATGATAAGCCTTGGAG. lcS91A: FW: CAGATTATTACTGTCAACAAGCTAATACCTGGCCGTACACGTT, REV: AACGTGTACGGCCAGGTATTAGCTTGTTGACAGTAATAATCTG. lcW94A: FW: GTCAACAAAGTAATACCGCGCCGTACACGTTCGGAGG, REV: CCTCCGAACGTGTACGGCGCGGTATTACTTTGTTGAC. lcY96A: FW: TAATACCTGGCCGGCCACGTTCGGAGGGG, REV: CCCCTCCGAACGTGGCCGGCCAGGTATTA. hcY101A: FW: CTGTTCAAGATCCGGCGCCGGATATTATGCTATGGAG, REV: CTCCATAGCATAATATCCGGCGCCGGATCTTGAACAG. hcY104A: FW: CCGGCTACGGATATGCTGCTATGGAGTACTGGG, REV: CCCAGTACTCCATAGCAGCATATCCGTAGCCGG), followed by subsequent expression and purification as was described for holo-POM1.

Protein analysis

COCS were washed twice in PBS and scraped off the PTFE membranes with PBS. Homogenization was performed with a TissueLyser LT (Qiagen) for 2 minutes at 50 Hz. A bicinchoninic acid assay (Pierce BCA Protein Assay Kit, Thermo Fisher Scientific) was used to determine protein concentrations. PrPSc levels were determined through digestion of 20 µg of COCS homogenates with 25 µg mL−1 of proteinase K (PK, Roche) at a final volume of 20 µL in PBS for 30 minutes at 37 °C. PK was deactivated by addition of sodium-dodecyl-sulfate-containing NuPAGE LDS sample buffer (Thermo Fisher Scientific) and boiling of samples at 95 °C for 5 minutes. Equal sample volumes were loaded on Nu-PAGE Bis/Tris precast gels (Life Technologies) and PrPC/PrPSc was detected by western blot using the monoclonal anti-PrP antibodies POM1, POM2 or POM19 at 0.4 µg mL−1 (all holo-antibodies), as established elsewhere8. Further primary antibodies used for western blots in this manuscript are as follows: monomeric NeonGreen (1:1,000, 32F6, Chromotek), phospho-eIF2α (1:1,000, clone no. D9G8, Cell Signaling Technologies), eIF2α (1:1,000, clone no. D7D3, Cell Signaling Technologies), pan-actin (1:10,000, clone no. C4, Millipore), GFAP (1:1,000, clone no. D1F4Q, Cell Signaling Technologies), Iba1 (1:500, catalog no. 019–19741, Wako), NeuN (0.5 µg/ml, catalog no. ABN78, Merck Millipore) and Myc-tag (1:500, catalog no. ab9106, Abcam). After incubation of primary antibodies at 4 °C overnight, membranes were washed and detected with goat polyclonal anti-mouse (1:10,000, 115–035–062, Jackson ImmunoResearch) or goat polyclonal anti-rabbit (1:10,000, 111–035–045, Jackson ImmunoResearch) antibodies for 1 hour at room temperature. For PNGaseF digestion, 20 µg of samples was processed using a commercially available kit (New England Biolabs), and PrPC detection was performed using the monoclonal anti-PrPC antibody POM2, as described above. Western blots were quantified on native photographs (uncropped, naive images are available in the Source Data); representative western blot images in the main and supplementary figures have been processed (linear adjustment of contrast and brightness) for better visualization.

Surface plasmon resonance (SPR)

The binding properties of the complexes between rmPrP, POM1 and pomologs were measured at 298 K on a ProteOn XPR-36 instrument (Bio-Rad) using 20 mM HEPES pH 7.2, 150 mM NaCl, 3 mM EDTA and 0.005% Tween-20 as running buffer. mPrP was immobilized on the surface of GLC sensor chips through standard amide coupling. Serial dilution of antibodies (full IgG, Fab or single-chain versions) in the nanomolar range were injected at a flow rate of 100 µL/min (contact time 6 minutes); the dissociation phase was then observed for 5 minutes. Analyte responses were corrected for unspecific binding and buffer responses by subtracting the signal of both a channel where no PrP was immobilized and a channel with no antibody was added. Curve fitting and data analysis were performed with Bio-Rad ProteOn Manager software (version 3.1.0.6).

Statistical analyses

All biological measurements are taken from distinct samples. Unless mentioned otherwise, the following tests were performed for statistical hypothesis testing: unpaired, two-tailed t-test was used for comparison between two groups, one-way ANOVA with Dunnett’s multiple-comparison test was used for comparison of multiple groups with a control group, and ordinary one-way ANOVA with Šídák’s multiple comparisons test was used for comparison of preselected pairs of groups. Statistical analysis and visualization were performed using Prism 8 (GraphPad). No statistical methods were used to pre-determine sample sizes, but our sample sizes are similar to those reported in previous publications7,8,9,10,11. Except for in vivo prion inoculation experiments and NeuN morphometry, data collection and analysis were not performed blind to the conditions of the experiments.

Synchrotron radiation circular dichroism

Secondary structure content of complexes between rmPrP and POM1 and hcY57 and hcY104A was analyzed with synchrotron radiation circular dichroism (SRCD) spectroscopy.

Experiments were performed using a nitrogen-flushed B23 beamline for SRCD at Diamond Light Source or ChirascanPlus CD spectropolarimeter (Applied Photophysics). With both instruments, scans were acquired at 20 °C using an integration time of 1 second and 1 nm bandwidth. Demountable cuvette cells with a pathlength of 0.00335 cm were used in the far-UV region (180–260 nm) to measure the CD of the protein concentration, varying from 10 to 102 μM protein in 10 mM NaP pH 7 and 150 mM NaCl. Mixtures were prepared to a stoichiometric molar ratio of 1:1. SRCD data were processed using CDApps33 and OriginLab. Spectra have been normalized using an average amino acid molecular weight of 113 for secondary structure estimation from SRCD, and CD spectra were created using CDApps using the Continll algorithm34. For comparison of calculated and observed spectra, the full molecular weight of sample and complex were used. Measurements of free mPrP and free antibodies were taken as a reference.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.