A bispecific immunotweezer prevents soluble PrP oligomers and abolishes prion toxicity

Abstract

Antibodies to the prion protein, PrP, represent a promising therapeutic approach against prion diseases but the neurotoxicity of certain anti-PrP antibodies has caused concern. Here we describe scPOM-bi, a bispecific antibody designed to function as a molecular prion tweezer. scPOM-bi combines the complementarity-determining regions of the neurotoxic antibody POM1 and the neuroprotective POM2, which bind the globular domain (GD) and flexible tail (FT) respectively. We found that scPOM-bi confers protection to prion-infected organotypic cerebellar slices even when prion pathology is already conspicuous. Moreover, scPOM-bi prevents the formation of soluble oligomers that correlate with neurotoxic PrP species. Simultaneous targeting of both GD and FT was more effective than concomitant treatment with the individual molecules or targeting the tail alone, possibly by preventing the GD from entering a toxic-prone state. We conclude that simultaneous binding of the GD and flexible tail of PrP results in strong protection from prion neurotoxicity and may represent a promising strategy for anti-prion immunotherapy.

Fig 1. A bispecific immunotweezer formed by a combination of toxic (POM1) and non-toxic (POM2) antibodies.

(A) The single chain variable domains of POM1 and POM2 are joined by a flexible linker to yield the bispecific scPOM-bi, schematic representation. B) Computational molecular dynamics model of scPOM-bi in complex with mPrP; intra-molecular (left and S1 Movie) and inter-molecular (right and S2 Movie) binding modes are shown. They are both structurally accessible to scPOM-bi.

Fig 2. The bispecific scPOM-bi antibody protects against prion infection even when administered 21 days post infection (dpi).

(A) Chronic treatment with scPOM-bi for 21 days did not produce observable toxicity in COCS, contrary to POM1. Furthermore, simultaneous addition of the individuals POM1 and POM2 to COCS resulted in neurotoxicity, indicating that the bispecific has different properties than the simple sum of its parts Area staining of neuronal nuclei by NeuN is shown on the y axis (lower values correlate with toxicity). Column 3 (*) is from a different experiment with a related negative control on which the data was normalized to. (B) scPOM-bi prevents RML induced neurotoxicity even when added 21 dpi (top). Despite similar binding affinity for PrP, POM2-IgG does not achieve similar protection at 21 dpi even at 5 fold higher concentration (bottom). COCS inoculated with non-infectious brain homogenate (NBH) are used as control; the images show NeuN and DAPI staining of COCS, scale bar = 500 μm. ** p<0.01, *** p<0.001, n.s. = not significant, one-way ANOVA with Dunnett’s post-hoc test. Upper panel: n = 9 biological replicates (1 COCS = 1 biological replicate) for all treatment groups except for RML alone (n = 8). Lower panel: n = 9 biological replicates for all treatment groups. Images of all biological replicates depicted in S2 Fig. (C) Western blot shows the presence of PK resistant material in COCS inoculated with RML. Addition of scPOM-bi 21 days after prion inoculation of Tga20 COCS did not show conceivable reduction of PrP^Sc.

Fig 3. The bispecific antibody scPOM-bi binds simultaneously to GD and FT of PrP with high affinity.

SPR sensorgrams for binding of scPOM-bi to truncated PrP constructs lacking FT (A) or GD (B) indicate that both antigen binding sites of scPOM-bi correctly engage their target. The bispecific antibody (E) had a stronger affinity than its individual components (C-D) due to avidity resulting in a slower dissociation. The fitting of the experimental data used to calculate the binding constants is in grey. Values for the above plus the full IgG versions of POM1 and POM2 are summarized in (F).

Fig 4. scPOM-bi prevents the formation of soluble, PK resistant oligomers.

(A) DLS showed the presence of soluble oligomers (red shades in histograms, reported as percentage) upon addition of the POM1 toxic antibody to recombinant mPrP in vitro. Subsequent addition of POM2 did not remove the oligomers or inhibit toxicity. Smaller species comparable to monomeric forms (blue) were detected in solution when POM1 was in complex with ΔmPrP^C_90-230, lacking the flexible tail, and when POM2 was added to mPrP prior to POM1 addition. Similarly small species were found when the neuroprotective scPOM-bi was added to mPrP; the bispeficic was also capable of removing the soluble oligomers generated by POM1. DLS data is shown for 3 time points after complex formation. (n = 5 for scPOM1:mPrP, scPOM2:mPrP and scPOM-bi:mPrP; n = 3 for scPOM1:mPrP then scPOM2, scPOM2:mPrP then scPOM1 and scPOM1:mPrP then scPOM-bi) (B) DLS can only detect soluble material. To investigate the presence of insoluble aggregates we formed the mPrP:Ab complexes in vitro, centrifuged them and analyzed the resulting supernatant with PAGE/Western blot. Soluble material was only detected in toxic combinations (POM1:mPrP or POM1:mPrP followed by POM2, red and orange). The percentage of mPrP and antibody in solution (normalized against isolated PrP or antibody) is shown; data from quantification of band intensity on SDS-PAGE (images in S5 Fig—n = 7 for all samples tested). (C) In order to characterize both soluble oligomers and insoluble aggregates we formed the mPrP:Ab complexes in vitro and deposited the resulting material on microscopy slides. Confocal microscopy indicates that toxic antibody combinations (e.g. POM1:mPrP or POM1:mPrP followed by POM2, red and orange) generate species with smaller average size than protective antibody combinations. The surface area of the detected species is reported on the y axis, the horizontal line represents the average. Differences can also be appreciated by visual inspection of the confocal microscopy images (S4 Fig—scPOM1:mPrP n = 166, scPOM2:mPrP n = 1136, scPOM-bi:mPrP n = 204, scPOM1:mPrP then scPOM2 n = 444, scPOM2:mPrP then scPOM1 n = 74 and scPOM1:mPrP then scPOM-bi n = 1767). D) The soluble oligomers generated by POM1 showed increased resistance to in vitro degradation by proteinase K at 2μg/ml (red). Such resistance was abolished when POM1 bound a mPrP construct lacking the FT (light red) or in non-toxic antibodies (shades of blue). Data from quantification of PK resistant bands on western blot, normalized against isolated PrP (images in S6 Fig—scPOM1:mPrP n = 5, scPOM1:ΔPrP n = 3, scPOM2:mPrP n = 4, scPOM-bi:mPrP n = 4).

Fig 5. scPOM-bi does not induce toxicity on CAD5 expressing PrPC in comparison to scPOM1.

The percentage of PI positive cells for different mPrP:antibodies complexes (A) or antibodies alone (B) on CAD5 PrP^C (left) and on CAD5 Prnp^-/- (right) are shown; each sample was added to cells after 10’ or 60’ of incubation at RT. The scPOM1:mPrP soluble oligomers caused toxicity in PrP^C-expressing CAD5 cells but not in PrP^C knock-out (PrP^-/-) CAD5. No toxicity was detected when scPOM1 was in complex with the truncated ΔmPrP_90-230 lacking the FT. Addition of the scPOM2:mPrP complex resulted in no toxicity, as well. Addition of the scPOM-bi:mPrP material to cells 10 minutes after complex formation resulted in toxicity, albeit lower than with scPOM1. However, toxicity was not significant if the material was added 60 minutes after complex formation. (n = 4 for all samples tested).

Fig 6. The bispeficic antibody scPOM-bi prevents the formation of soluble PrP oligomers and protects from prion neurotoxicity even when administered late after infection.

Addition of POM1 antibody (top), but not POM2 or scPOM-bi, to PrP^C generates soluble, pK resistant PrP oligomers (red) whose presence correlates with toxicity (top); subsequent addition of the neuroprotective scPOM-bi, but not POM2, eliminates them in favor of larger, non-toxic aggregates (blue). Small soluble oligomers might also be responsible for prion induced toxicity (bottom) similarly to other amyloidosis. scPOM-bi might be able to eliminate them just as it does with the POM1-induced oligomers whereas POM2 might not, which would explain why only the bispecific is neuroprotective even at late administration (21 days post infection, dpi).