Aggregation-dependent epitope sequence and modification fingerprints of anti-Aβ antibodies
- Rudolf Virchow Center for Integrative and Translational Bioimaging, Julius-Maximilians-Universität (JMU), Germany
- Department of Neurology, University Hospital Wuerzburg (UKW), Germany
- Department of Psychiatry and Psychotherapy, University Medical Center Göttingen (UMG), Germany
- Neuroproteomics Group, Department of Molecular Neurobiology, Max Planck Institute for Multidisciplinary Sciences, Germany
- Center of Neurology, Molecular Cell Biology, University Hospital Bonn, Germany
Figures
Figure 1 with 7 supplements
Microarray workflow and amyloid-beta antibod (Aβ-Ab) epitope pre-screening.
(a) Schematic of the microarray workflow. The library comprised the amyloid precursor protein (APP) residues 649–722 (APP770 numbering), corresponding to Aβ residues –23–51, as a library of 15-mer peptides overlapping by 14 AAs, copied in dozens, and probed for Aβ-Abs binding followed by analysis in MARTin. (b) The epitope screening binding data are summarized as a heatmap. Blue and red shades indicate the relative binding strength of the related arrayed peptides to the left and cognate Aβ-Ab on top. The blue-red scale is visualizing the normalized binding intensity for each antibody individually. The applied antibody concentrations ranged between 0.5–1 µg/mL. Note: Among the 20 tested antibodies, 1E4E11, 5H11C10, 2–48, Aβ-pE3, and Donanemabbs have known PTM selectivity, which could not be fully addressed with the initial microarray consisting of unmodified Aβ fragments only.
Figure 1—figure supplement 1
Titration series of N-terminal specific Anti-amyloid-beta s (Aβs).
(a) Titration of monoclonal Aβ-Ab 3D6, 82E1, and 7H3D6 at concentrations ranging from 2 to 0.00041 µg/mL to study the selectivity for free-Asp(1) over Asp(1) within an N-terminal elongated peptide. The signals were normalized internally and presented as a heat map with blue-to-red shades ranging from 0 to 1. (b) Monoclonal Aβ-Ab 11H3D6 and 80C2 exhibited high selectivity towards free-Asp(1) across the chosen concentration range of 2 µg/mL down to 0.0041 µg/mL.
Figure 1—figure supplement 2
µSPOT (l)-Asp 1–15 LC-MS QC.
Figure 1—figure supplement 3
µSPOT (d)-Asp 1–15 LC-MS QC.
Figure 1—figure supplement 4
µSPOT (l)-iso-Asp 1–15 LC-MS QC.
Figure 1—figure supplement 5
µSPOT (d)-iso-Asp 1–15 LC-MS QC.
Figure 1—figure supplement 6
µSPOT pyro-Glu 3–17 LC-MS QC.
Figure 1—figure supplement 7
µSPOT pSerine 1–15 LC-MS QC.
Figure 2 with 4 supplements
Deep mutational scans define amyloid-beta-antibody (Aβ-Ab) core sequences.
(a) Schematic summary of the three deep positional scans in the N-terminal region. Seven reported amyloid precursor protein (APP) mutations are indicated as yellow boxes. From each 18-mer epitope, positional substitutions were generated (342 variants for each epitope) and tested semi-quantitatively in microarray format. The wild-type (WT) sequence was sequentially scanned from N to C-terminal by turning each position into every proteinogenic amino acid. Consequently, the contribution of each residue is depicted in blue-white-red shades, with white corresponding to no variation from WT (Colvin et al., 2016) and blue-to-red representing loss or gain of binding strength. (b) Fingerprint analysis for the first peptide group covering the sequence KTEEISEVKMDAEFRHDS. The Aβ-Abs tested here are 101-1-1 and 14-2-4. (c) Fingerprint analysis for the second peptide group covering the sequence VKMDAEFRHDSGYEVHHQ. The Aβ-Abs tested here are IC16 and 1E8. (d) Fingerprint analysis for the third peptide group covering the sequence DAEFRHDSGYEVHHQKLV. The Aβ-Abs tested here are Aducanumab, Donanemab and Lecanemab biosimilars. In addition, commercially available Aβ-Abs, such as 3D6, 82E1, 11H3, 80C2, 7H3D6, and 1E4E11 were included. Note: The antibodies, 1E4E11 and Donanemabbs have known reported PTM preference, which will be addressed in the next paragraphs.
Figure 2—figure supplement 1
N-terminally elongated deep mutational scans of Aducanumabbs and 6E10 with core motifs summary.
(a) Fingerprint analysis for the first peptide group covering the sequence KTEEISEVKMDAEFRHDS. The amyloid-beta-antibody (Aβ-Ab) tested here is Aducanumabbs. (b) Fingerprint analysis for the second peptide group covering the sequence VKMDAEFRHDSGYEVHHQ. The Aβ-Ab tested here is Aducanumabbs. (c) Fingerprint analysis of 6E10 on the sequence KTEEISEVKMDAEFRHDS. (d) Summary of the antibody core motifs, ‘X’ within the sequence stands for non-conserved residue. When an amino acid variation led to a decrease in binding below 50% of wild-type (WT), it was marked as dominant negative. If more than 10 amino acid substitutions were dominant negative, the entire position was designated as part of the core motif (represented in single-letter coding). (e) Aβ-Ab species preferences. Amino acid positions which are different in human and rat/mouse are highlighted in red. These specific positions were probed. Antibodies showing altered binding strength after replacing R5 or Y10 in the human Aβ sequence by G5 or F10 (rat/mouse Aβ sequence) are highlighted in blue. Human: Homo sapiens; Dog: Canis lupus familiaris; Pig: Sus scrofa; Rabbit: Oryctolagus cuniculus; Sheep: Ovis aries; Rat: Rattus norvegicus; Mouse: Mus musculus.
Figure 2—figure supplement 2
Aβ1–15 carrying the English mutation (H6R) shows reduced affinity for Aducanumabbs.
(a) Dot blot from the purified 6xHis peptides DAEFRHDSGYEVHHQ and DAEFRRDSGYEVHHQ were printed using Slide Spotter robot. IgG signals were detected using Goat anti-human DL650, in square brackets the µg/mL of Aducanumabbs. (b) Biolayer interferometry responses of Aducanumabbs (10 nM) over the NTA-loaded DAEFRHDSGYEVHHQ and DAEFRRDSGYEVHHQ peptides are shown.
Figure 2—figure supplement 3
Purified 1–15 H6R.
Figure 2—figure supplement 4
Purified 1–15 and p1-15 peptides LC-MS QC.
Figure 3 with 4 supplements
Multiplex analysis of amyloid-beta-antibodies (Aβ-Abs) with preference for post-transcriptionally modified Aβ variants.
The signals were normalized internally and presented as a heat map ranging from 0 to 1. Blue and red shades indicate the binding strength over the related arrayed post-transcriptionally modified peptide shown on top and cognate anti-Aβ antibody to the left. The antibodies were tested under optimal dilutions: 6E10, IC16, Aβ-pE3, 1E411, 101-1-1, 5H11C10, 2–48 at 0.25 µg/mL; Lecanemabbs, Donanemabbs at 0.0012 µg/mL; Aducanumabbs at 0.02 µg/mL; 3D6, 11H3, 82E1 at 0.037 µg/mL; 7H3D6 at 0.012 µg/mL. The dataset is split over different sections (a-f) to allow proper visualization. Each antibody was screened in parallel against all modifications. (a) The control peptide Aβ(−1–14) was used alongside Asp(1) isomers and stereoisomers. (b) The region overlapping with unmodified Serine 8. (c) The region overlapping with modified phospho Serine 8 (represented as ‘2’). (d) The control unmodified 3–17 was used alongside pE3. (e) The region overlapping with unmodified Serine 26. (f) The region overlapping with modified phospho Serine 26 (represented as ‘2’).
-
Figure 3—source data 1
A10–EC50 values of 1E4E11 and Aducanumabbs on microarray.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig3-data1-v1.xlsx
Figure 3—figure supplement 1
Antibody 1E4E11 shows high preference for phosphorylated pSer8 Aβ1–15 in µSPOT and Biolayer Interferometry (BLI) assays.
(a) Titration of monoclonal amyloid-beta-antibody (Aβ-Ab) 1E4E11 at concentrations ranging from 0.25 to 0.001 µg/mL to achieve high selectivity for Aβ1–15 phosphorylated at Ser(8). The modified pSer is represented as ‘2.’ The signals were globally normalized against DAEFRHDpSGYEVHHQ and presented as a heat map with blue-to-red shades ranging from 0 to 1. (b) BLI dose-response curve of mAb 1E4E11 binding to the NTA-loaded DAEFRHDSGYEVHHQ-6xHis-amide peptide. (c) Biolayer interferometry dose-response curve of mAb 1E4E11 binding to the NTA-loaded DAEFRHDpSGYEVHHQ-6xHis-amide peptide. (d) BLI steady-state analysis revealed KDs of 4.1 nM and 1.3 µM for Aβ1–15 with phosphorylated and unphosphorylated Ser (8), respectively (1:1 model).
Figure 3—figure supplement 2
Phosphorylation aids Aducanumabbs binding to synthetic monomeric Aβ1–15.
(a) Titration of Aducanumabbs at different concentrations to achieve high selectivity for phosphorylated pSer8-Aβ1–15 over unphosphorylated Aβ1–15. The signals were internally normalized and fitted as a sigmoidal curve (Boltzmann). The phospho signal for pSer8-Aβ1–15 is shown in red (r^2 0.98) and the signal for unphosphorylated Aβ1–15 is shown in blue (r^2 0.96). (b) Biolayer Interferometry (BLI) steady-state analysis revealed KDs of 1.4 nM and 8.7 nM for phosphorylated and unphosphorylated Aβ1–15, respectively (1:1 model). (c) Biolayer interferometry dose-response curve of Aducanumab binding to the NTA-loaded DAEFRHDSGYEVHHQ-6xHis-amide peptide. (d) Biolayer interferometry dose-response curve of Aducanumab binding to the NTA-loaded DAEFRHDpSGYEVHHQ-6xHis-amide peptide.
Figure 3—figure supplement 3
Molecular dynamics (MD) simulation of Aducanumab with pSer(8)-Aβ2–8.
(a) Sticks and cartoon depiction of the MD simulations between Aducanumab and AEFRHDS. (b) Electrostatic surface potential for the non-phosphorylated sequence interaction (blue and red represent electropositve and electronegative regions, respectively). (c–d) RMSD analysis of the backbone atoms of the unphosphorylated candidate (average conformation over 600 ns). (e) Sticks and cartoon depiction of the MD simulations between Aducanumab and AEFRHDpS. (f) Electrostatic surface representation for the pSer8 sequence interaction. (g–h) Root mean square deviation (RMSD) analysis of the backbone atoms of the post-translational modification (PTM) candidate (average conformation over 600 ns).
Figure 3—figure supplement 4
pSer8 but not Ser8-Aβ recapitulates the binding mode resolved by crystallography.
Comparison of molecular dynamics (MD) simulations with structural data. The MD simulation was conducted with the resolved epitope peptide EFRHDS, but in the absence of additional ions, and predicts a largely altered binding mode characterized by over 10 Å movements of R317 and R318. In contrast, the crystal structure (PDB 6CO3) harbors a sulfate ion that neutralizes the positive charge introduced by R317 and R318. Strikingly, the MD of the pSer(8) modified peptide recapitulates the resolved X-ray binding mode to monomeric amyloid-beta (Aβ). Aducanumab targets aggregated Aβ, while still engaging with low micromolar affinity a monomeric Aβ.
Figure 4 with 2 supplements
IP-MS signature of different biosimilar Anti-Aβ antibody ies (Aβ-Abs).
(A) A mixture of synthetic Aβ peptides was immunoprecipitated with the following Aβ-Abs (from top to bottom): 4G8 (pan-antibody), Donanemabbs (Don-bs), Lecanemabbs (Lec-bs), Aducanumabbs (Adu-bs), and 1E4E11 (phospho-selective control). Mass spectra were generated by MALDI-TOF-MS in reflector mode from TOPAC matrix and are representative of two independent experiments. We noted that the signal pattern slightly varied between experiments, which we mainly attributed to some immunoprecipitation of co-aggregated Aβ species, an effect that has to be considered under the experimental conditions applied. (B) IP-MS from pooled human cerebrospinal fluid (CSF) using MALDI-TOF-MS from conventional CHCA matrix. Mab 4G8 (positive control) and Lecanemabbs, but not Aducanumabbs immunoprecipitated authentic biological Aβ peptides from pooled human CSF under the tested conditions. Major peaks were observed for Aβ1–40 and Aβ1–38. The spectra shown are representative for two replicate measurements each from two independent immunoprecipitations except for the positive control 4G8 which was processed only once.
Figure 4—figure supplement 1
Synthesized pSerine 8 Aβ1–40 LC-MS QC.
Figure 4—figure supplement 2
Purified pSerine 8 Aβ1–40 LC-MS and MALDI-TOF QC.
Figure 5
Aducanumabbs shows strong preference for highly aggregated fibrillar Aβ1–42.
(A) Hexafluoroisopropanol (HFIP)-treated Aβ1–42 monomer (m), oligomer (o), and fibril preparations (f) (1.44 µg each) were separated by 4–12% Bistris SDS-PAGE, blotted onto nitrocellulose and probed with Aducanumabbs (5 min exposure, F 0.84, image display: High: 65535; Low: 0; Gamma: 0.85). After image recording, the blot image recording the blot membrane was reprobed without prior stripping (no removal of primary and secondary antibodies) with mAb4G8 (1 min exposure, image display: High: 54971; Low: 0; Gamma: 0.99). That way, the 4G8 signals were essentially added on top of the initial Aducanumabbs signals and background artifacts. The positions of pre-stained protein marker bands on the blot membrane are shown on the left-hand side. (B) Parallel sections from an Alzheimer’s disease (AD) patient were stained with Aducanumabbs and Aducanumabbs that had been pre-adsorbed with either Aβ1–42 monomers (Aβ1–42 m) or Aβ1–42 fibrils (Aβ1–42 f). Scale bar: 200 µm.
-
Figure 5—source data 1
Aducanumab and Lecanemab blot and marker.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig5-data1-v1.zip
-
Figure 5—source data 2
Aducanumab, Lecanemab and mAb 4G8 probed blots and marker.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig5-data2-v1.zip
Figure 6 with 1 supplement
Lecanemabbs recognizes an epitope located in the N-terminal region.
(A) An example of Capillary Isoelectric Focusing (CIEF)-immunoassay results for Lecanemabbs displayed as lane view (Western blot simulation). The indicated amyloid-beta (Aβ) peptides were separated by isoelectric focusing in microcapillaries on a Peggy Sue device, immobilized photochemically to the inner capillary wall and subjected to immunological detection. Under the tested conditions, Lecanemabbs detected Aβ1–40 (pI 5.31), Aβ2–40 (pI 5.98), Aβ3–40 (pI 5.97), Aβ–3–40 (pI 6.04), Aβ1–42 (pI approx. 5.3), and pSer8Aβ1–40 (amide) (pI 5.13–5.14). (B) Synthetic Aβ2–40 (361 ng), and monomerized (m), oligomeric (o), and fibrillar (f) preparations of Aβ1–42 (361 ng of each) were separated by 4–12% Bistris SDS-PAGE, blotted onto nitrocellulose, and probed with Lecanemabbs (0.5 µg/mL). Exposure: 2 min 7.5 s; F 0.84; image display: high: 47545; low: 0; gamma: 1.0. After blot development, the blot membrane was reprobed with mAb 4G8 without prior stripping. That way, the 4G8 signals were essentially added on top of the initial Lecanemabbs signals and background artifacts. Exposure: 1 min F 0.84; image display: high: 46277; low: 0; gamma: 1.0. (C) The indicated Aβ peptides (50 ng of each) were separated by Bicine Tris (peptide) SDS-PAGE, blotted on PVDF, and probed with Lecanemabbs. Exposure: 5 min F 0.84, image display: high: 65535; low: 0; gamma: 0.85. (D) To confirm loading and successful blotting of all Aβ peptides, the blot membrane was re-probed with mAb 4G8 without prior stripping (see above). Exposure: 5 min F 0.84; image display: high: 65535; low: 0; gamma: 1.0. (E) Parallel sections from an Alzheimer’s disease (AD) patient were stained with Lecanemabbs and Lecanemabbs that had been pre-adsorbed with either Aβ1–42 monomers (Aβ1–42 m), Aβ1–42 oligomers (Aβ1-42o), or Aβ1–42 fibrils (Aβ1–42 f). Scale bar: 200 µm.
-
Figure 6—source data 1
Original blot image files for Figure 6B-D.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig6-data1-v1.zip
-
Figure 6—source data 2
Original blot image files for Figure 6B-D.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig6-data2-v1.zip
Figure 6—figure supplement 1
Hexafluoroisopropanol (HFIP) treated Aβ1–42, but not shorter synthetic amyloid-beta (Aβ)-peptides, blocked Lecanemabbs binding to amyloid plaques.
Parallel sections from an Alzheimer’s disease (AD) patient were stained, from left to right, with Lecanemabbs only (BioXCell #BXC-SIM0032) or pre-adsorbed with either Aβ1–42 monomers (Aβ1–42 m), Aβ1–16 (Aβ1–16), Aβ1–34 (Aβ1–34), and Aβ1–38 (Aβ1–38). Scale bar: 200 µm.
Figure 7
Donanemabbs shows high preference for pyroglutamate-bearing pEAβ3–40.
The upper panel shows lane views (Western blots simulation) of a Capillary Isoelectric Focusing immunoassays (CIEF)-immonoassay for assessing the detection of different amyloid-beta (Aβ) variants by (A) mAb 4G8 (positive control antibody) and (B) Donanemabbs. The indicated Aβ variants were separated by isoelectric focusing in microcapillaries on a Peggy-Sue device, immobilized photochemically to the inner capillary wall and subjected to immunological detection by (A) mAb 4G8 and (B) Donanemabbs. (C) Synthetic Aβ peptides (50 ng per lane) were separated by Bicine Tris (peptide) SDS-PAGE, blotted on PVDF and probed with Donanemabbs (0.5 µg/mL) (upper image). Exposure: 5 min F 0.84; image display: high: 59406; low: 0; Gamma: 0.65. To confirm that all Aβ variants were loaded and blotted, the blot membrane was reprobed with mAb 4G8 without prior stripping. That way, the 4G8 signals were essentially added on top of the initial Donanemabbs signals and background artifacts (lower image). Exposure: 5 min F 0.84; image display: high: 65535; low: 0; gamma: 0.77. (D) AβpE3-40 and Aβ1–42 monomer (m), oligomer (o) and fibril preparations (f) were separated by 4–12% Bistris SDS-PAGE, blotted onto nitrocellulose, and probed with Donanemabbs (image display: High: 64091; Low: 0; Gamma: 0.75), followed by reprobing without prior stripping with mAb4G8 (image display: High: 46984; Low: 0; Gamma: 1.0). The positions of prestained protein marker bands on the blot membrane are shown on the left-hand side. (E) A pre-adsorption experiment employing synthetic Aβ peptides in human Alzheimer’s disease (AD) brain tissue showed that the immunoreactivity of Donanemabbs was not suppressed by pre-adsorption of the antibody with excess amounts of synthetic Aβ1–40 and Aβ3–40 peptides. In contrast, a pre-incubation with AβpE3-40 effectively attenuated the antibody signal in extracellular Aβ deposits as well as in the vasculature.
-
Figure 7—source data 1
Original blot image files for Figure 7C,D.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig7-data1-v1.zip
-
Figure 7—source data 2
Original blot image files for Figure 7C,D.
- https://cdn.elifesciences.org/articles/106156/elife-106156-fig7-data2-v1.zip
Tables
Table 1
Studied Aβ antibody panel together with previously and here reported epitopes.
| Antibody* | Source† | Clonality ‡ | Species § | Immunogen ¶ | Mod.** | Array†† | Refs. |
|---|---|---|---|---|---|---|---|
| 3D6 | Creative Biolabs (#PABL-011) | Mono | Mouse | 1–5 | - | 1–5 | Feinberg et al., 2014; Britschgi et al., 2009; Klafki et al., 2022a; Zampar et al., 2020 |
| pAb77 | Dept. of Psych., Göttingen | Poly | Rabbit | 2–14 | - | 3–16 | Savastano et al., 2016 |
| 4G8 | Biolegend (#800701) | Mono | Mouse | 17–24 | - | 17–21 | Wirths et al., 2024; Baghallab et al., 2018 |
| 11H3 | Nanotools (#bA4N-11H3) | Mono | Mouse | 1 x | - | 1–4 | |
| 82E1 | IBL (#10323) | Mono | Mouse | 1–16 | - | 1–4 | Zampar et al., 2020; Bader et al., 2023 |
| 14-2-4 | IBL | Mono | Mouse | (−3)–5 | - | (−2)–3 | Wirths et al., 2024 |
| 7H3D6 | Center of Neurol., Bonn | Mono | Rat | 1–14 | - | 1–10 | Kumar et al., 2013 |
| 1E4E11 | Abcam (#ab219817) | Mono | Mouse | p1-14 | pSer8 | 3–8 | Kumar et al., 2013 |
| 5H11C10 | Center of Neurol., Bonn | Mono | Mouse | p20-34 | pSer26 | 19–32 | Kumar et al., 2021 |
| pAb58-1 | Dept. of Psych., Göttingen | Poly | Rabbit | 4–9 | - | 4–9 | Bader et al., 2023 |
| Aβ-pE3 | Synaptic Systems (#218003) | Poly | Rabbit | pGlu3-7 | pE3 | 3–7 | Nussbaum et al., 2012 |
| 80C2 | Synaptic Systems (#218231) | Mono | Mouse | 1–5 | - | 1–5 | Zampar et al., 2020 |
| 2-48 | Synaptic Systems (#218011) | Momo | Mouse | pGlu3-7 | pE3 | 3–7 | Bayer and Wirths, 2014 |
| 101-1-1 | Dept. of Psych. Göttingen | Mono | Mouse | (−3)–5 | - | (−2)–3 | Wirths et al., 2024; Klafki et al., 2020 |
| IC16 | Dept. of Neuropath. Düsseldorf | Mono | Mouse | 1–16 | - | 1–8 | Richter et al., 2010; Dornieden et al., 2013; Jäger et al., 2009 |
| 6E10 | Covance (#SIG-39320) | Mono | Mouse | 1–16 | - | 3–8 | Baghallab et al., 2018; Klafki et al., 2020 |
| 1E8 | Nanotools (#bA4N-1E8) | Mono | Mouse | 1 x | - | 3–8 | Klafki et al., 2022a |
| Aducanumabbs | Proteogenix (#PX-TA1335) | Mono | Human | - | - | 1–7 | Arndt et al., 2018 |
| Donanemabbs | Proteogenix (#PX-TA1549) | Mono | Humanized | pGlu3-42 | pE3 | 3–10 | Sims et al., 2023; Demattos et al., 2012 |
| Lecanemabbs | Proteogenix (#PX-TA1746) and BioXCell (#BXC-SIM0032) | Mono | Humanized | 1–42 | - | 3–7 | van Dyck et al., 2023 |
-
*
names of the anti-amyloid beta antibodies.
-
†
Source of the antibodies used in this study, either commercial or academic.
-
‡
Clonality of the antibodies: "mono" indicates monoclonal, while "poly" stands for polyclonal antibodies.
-
§
Species of origin for the IgGs: mouse, rat, human, or rabbit.
-
¶
Immunogens, as reported in the literature, that were used to generate the corresponding antibodies.
-
**
If available, the anticipated PTM specificity is indicated: “p” represents phosphoserine (at positions Ser8 or Ser26), and “pE3” refers to an N-terminal pyroglutamate (at position 3).
-
††
epitope identified in this study on microarrays.
Additional files
-
MDAR checklist
- https://cdn.elifesciences.org/articles/106156/elife-106156-mdarchecklist1-v1.docx
-
Supplementary file 1
Probed Aβ peptide library.
- https://cdn.elifesciences.org/articles/106156/elife-106156-supp1-v1.xlsx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Aggregation-dependent epitope sequence and modification fingerprints of anti-Aβ antibodies
eLife 14:RP106156.
https://doi.org/10.7554/eLife.106156.3
