Humanized monoclonal antibody armanezumab specific to N-terminus of pathological tau: characterization and therapeutic potency

Abstract

Background: The experience from clinical trials indicates that anti-Aβ immunotherapy could be effective in early/pre-clinical stages of AD, whereas at the late stages promoting the clearing of Aβ alone may be insufficient to halt the disease progression. At the same time, pathological tau correlates much better with the degree of dementia than Aβ deposition. Therefore, targeting pathological tau may provide a more promising approach for the treatment of advanced stages of AD. Recent data demonstrates that the N-terminal region of tau spanning aa 2-18 termed "phosphatase activation domain" that is normally hidden in the native protein in 'paperclip'-like conformation, becomes exposed in pathological tau and plays an essential role in the inhibition of fast axonal transport and in aggregation of tau. Hence, we hypothesized that anti-Tau_2-18 monoclonal antibodies (mAb) may recognize pathological, but not normal tau at very early stages of tauopathy and prevent or decrease the aggregation of this molecule.

Methods: Mouse mAbs were generated using standard hybridoma methodology. CDR grafting was used for humanization of mouse mAb. Humanized mAb (Armanezumab) was characterized and tested in vitro/ex vivo/in vivo using biochemical and immunological methods (HPLC, Biacore, ELISA, IHC, FRET, etc.). Stable DG44 cell line expressing Armanezumab was generated by clone selection with increased concentrations of methotrexate (MTX).

Results: A panel of mouse mAbs was generated, clone 1C9 was selected based on binding to pathological human tau with high affinity and humanized. Fine epitope mapping revealed conservation of the epitope of human tau recognized by the parent murine mAb and Armanezumab. Importantly, Armanezumab (i) bound to tau with high affinity as determined by Biacore; (ii) bound pathological tau in brains from AD, FTD and Pick's disease cases; (iii) inhibited seeding effect of aggregated tau from brain lysate of P301S Tg mice; (iv) inhibited cytotoxic effect of tau oligomers; (v) reduced total tau (HT7) and AT100, PHF1, AT8, AT180, p212, p214-positive tau species in brains of tau transgenic mice after intracranial injection. A stable CHO cell line producing >1.5 g/l humanized mAb, Armanezumab was generated.

Conclusion: These findings suggest that Armanezumab could be therapeutic in clinical studies for treatment of AD.

Keywords: Alzheimer’s disease; Humanization; Immunotherapy; Monoclonal antibody; Phosphatase activation domain; Tauopathy; Therapeutic efficacy.

Fig. 1

Characterization of mouse 1C9 anti-tau_2–18 monoclonal antibody. a Competition ELISA using peptides with Alanine substitution showed that 1C9 recognized epitope PRQEF comprising 4–8 amino acids of tau_2–18 peptide. The half maximal inhibitory concentration (IC₅₀) for each peptide is shown in Table. b 1C9 recognized full-length tau but not tau that lacks 2–18 domain in Western Blot. Lane 1- tauΔ2-18, lane 2-full-length tau. c 1C9 bound monomeric (spot 1), oligomeric (spot 2: cross-linked; spot 3: non-cross-linked) and fibrillar (spot 4) forms of recombinant tau protein in dot blot. d Anti-tau_2–18 mAb 1C9 bound to neuropil threads and neurofibrillary tangles in AD brains (Braak stage VI-C). No binding was observed with non-AD brain (Braak stage 0). Original magnification 40X, scale bar = 20 μm. e 1C9 bound different species of tau protein in brain homogenates from both AD cases and control subjects in denaturing conditions (lane 1: control 1; lane 2-control 2; lane 3-AD1; lane 4-AD2; lane 5-AD3) in Western Blot. f In non-denaturing conditions in Dot Blots 1C9 as well as commercial TNT-1 Ab specific to N-terminus of Tau selectively bound to soluble tau in AD brains but not in controls. Of note, HT7 and rabbit anti-tau-polyclonal Ab recognizing total tau, had bound tau in both control and AD brains

Fig. 2

a SDS-PAGE analyses of purified antibody, Lane 1: Armanezumab, reducing conditions, 2.00 μg; Lane 2: Armanezumab, non-reducing conditions, 2.00 μg. b Armanezumab purified from CHO cells supernatant had a 99% purity measured by HPLC. c Characterization of Armanezumab and parental mouse 1C9 mAb by Surface Plasmon Resonance (SPR). SPR sensorgrams showing the binding of human tau (0 N/4R isoform) with each of immobilized antibody. Tau protein was run with various concentrations (3, 9, 27, 81, 243 nM), curves and fitted curve are shown in the corresponding color. Table shows the association rate constant (K_a), dissociation rate constant (K_d), and binding constant (K_D) of antibodies with human tau. Biacore T200 evaluation software, version 1.0 was used to calculate K_a and K_d using 1:1 fitting model. Ms, millisecond; M, molar; s, second

Fig. 3

a “Alanine scanning” showed that CDR grafting did not affect the epitope specificity. Armanezumab recognized epitope PRQEF comprising 4–8 amino acids of tau_2–18 peptide. Inhibition of binding of Armanezumab to Tau_2–18 by peptides with alanine substitution in competition ELISA. The half maximal inhibitory concentration (IC₅₀) for each peptide is shown in Table. b Armanezumab recognized (i) full-length recombinant tau protein, but not tau that lacks aa 2–18 (ΔTau2-18), lanes 1–2 and (ii) aggregated forms of tau in brain homogenates from AD cases (Braak stage VI), lanes 3–7. Lane 1: ΔTau2-18; lane 2: Full-length tau; lane 3: control brain 1; lane 4: control brain 2; lane 5: AD brain 1; lane 6: AD brain 2; lane 7: AD brain 3. HT7 recognizing total tau and TNT-1 specific to N-terminus of tau were used as positive controls

Fig. 4

Armanezumab bound to pathological tau in brain tissues from inferior parietal gyrus of AD (a), midfrontal cortices of both Pick’s Disease (b) and Frontotemporal Dementia (c), while no binding was observed in the inferior parietal gyrus of non-AD brain (d). Adjacent brain sections stained with other antibodies against pathological tau such as PHF1 (e-g), AT8 (i-k), AT100 (m-o), N-terminal tau TNT1 (q-s), as well as HT7 anti-total tau (u-w), showed similar patterns of pathological profiles in perikarya and neuritic processes, while no binding was observed in the adjacent sections from the control brain (d, h, l, p, t, x). Original magnification 60X, scale bar =20 um

Fig. 5

Armanezumab inhibited the seeding activity of pathological tau. a Co-incubation with Armanezumab blocked seeding activity of brain lysate of tau (P301S)/Tg mice and significantly decreased the ability to induce the aggregation of RDΔK280-CFP/RD-YFP in transiently transfected HEK293 cells

Fig. 6

a Armanezumab/protein A/G complex bound and removed pathological tau from brain lysate of tau(P301S)/Tg mice significantly decreasing the ability of lysate to induce the aggregation of RD-CFP/RD-YFP in HEK293 cell line constitutively expressing RD-CFP/RD-YFP. FRET positive cells were analyzed by flow cytometry and integrated FRET density was calculated. Representative plots of flow cytometric analyses for each sample are shown. % of FRET positive cells are indicated in plots. b Brain lysate of tau(P301S)/Tg mouse immunodepleted with Armanezumab/protein A/G complex or control IgG/protein A/G complex were analyzed by western blot. Bands were visualized using rabbit anti-tau polyclonal antibody. Lane 1-brain lysate; lane 2-brain lysate immunodepleted with Armanezumab; lane 3- brain lysate immunodepleted with control human IgG

Fig. 7

Armanezumab inhibited cytotoxicity of oligomeric recombinant tau protein. a SH-SY5Y human neuroblastoma cell line and mouse primary neurons b were incubated with tau oligomers in the presence or absence of Armanezumab or control IgG. Control cells were treated with vehicle, and cell viability was assayed in all cultures using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Data were collected (four replicates) and expressed as percentages of control ± SD

Fig. 8

Armanezumab reduced total (HT7) and pThr212, pSer214, PHF1, AT100, AT180, AT8-positive phosphorylated tau in the brains of 6-month-old Thy22-Tau Tg mice after intracranial injection. Quantitative analysis using ImageJ software showed reduced % of stained total area for each ipsilateral region injected with Armanezumab compared to contralateral region injected with control IgG. Bars represent mean ± SD from n = 7 mice. Corresponding representative images of injected regions where antibodies diffused after injection, stained with various tau-specific antibodies are shown in the boxed areas for each hemisphere. Original magnifications 10X, scale bar = 100 um

Fig. 9

Mean serum concentrations and pharmacokinetics parameters of Armanezumab following a single IV dose to PS19 tau/Tg and C57BL6 mice. Error bars represent average ± SD (n = 3 for C57BL6 and n = 9 for PS19 tau/Tg groups). Pharmacokinetic analyses were performed as described in Materials and Methods. Kel, elimination constant; T-1/2, elimination half-life; Tmax, the time to reach max concentration; Cmax, the maximum concentration; AUC 0-t, area under the concentration-time curve; AUC 0-inf, the area to infinity; AUC 0-inf %extrap, % of extrapolated AUC to infinity; Vd, the apparent volume of distribution