A shorter linker in the bispecific antibody RmAb158-scFv8D3 improves TfR-mediated blood-brain barrier transcytosis in vitro

In this study, we aimed to improve the BBB-passage of the symmetric antibody RmAb158-scFv8D3⁵ by varying the linker length between the therapeutic antibody RmAb158 and its TfR-targeting BBB-shuttle domain scFv8D3. The use of protein drugs for targets located inside the brain is highly restricted by the BBB, but it has been shown that binding to TfR can facilitate the transport of antibodies across the BBB. At elevated administered antibody concentrations, however, bivalent binding to TfR is thought to reduce the transcytosis efficiency through receptor crosslinking and therefore monovalent binding to TfR has been proposed to be preferred^{5,6,16,18,20,29}. Hence, our major focus was to investigate to which degree the binding strength of the RmAb158-scFv8D3 variants to TfR was affected by the different linker lengths and which effect this had on the antibodies’ transcytosis across BBB endothelial cells.

One major concern in designing the new RmAb158-scFv8D3 variants was that extremely short linkers could distort the secondary or tertiary structure of both scFv8D3 and the RmAb158 light chain³⁰. Long linkers, on the other hand, may be more prone to degradation and aggregation or, if applied in vivo cause immunogenicity³¹. However, we did not detect a significant degree of aggregation in any of the tested antibodies by SEC and the RmAb158-scFv8D3 variants with linker lengths between 1 aa and 80 aa had similar thermal stability. Only the − 2 aa linker variant, where the two C-terminal amino acids of scFv8D3 were removed, demonstrated an additional unfolding event at a lower temperature (59.6 °C), indicating that removing these two amino acids caused a destabilization of the protein under thermal stress. We conclude from these experiments, that varying the linker lengths between 1 aa and 80 aa in RmAb158-scFv8D3 did not affect the antibodies’ thermal stability. We furthermore confirmed by ELISA that the different linker lengths did not negatively affect the affinity of the RmAb158 arms to the therapeutic target Aβ protofibrils (Supplementary Fig. S3). In fact, the shorter-linker variants showed even enhanced binding strength to Aβ protofibrils compared to the parental antibody RmAb158.

To investigate the binding of the new RmAb158-scFv8D3 variants to mTfR_Pro119, we conducted ELISA and LigandTracer experiments and further evaluated the antibodies’ ability to cross the BBB in the In-Cell BBB-Trans assay.

With the indirect ELISA measuring mTfR binding and the more detailed kinetic analysis by LigandTracer, we observed that all tested RmAb158-scFv8D3 variants had similar apparent affinities to mTfR_Pro119, in contrast to the lower affinity of the monovalent control, indicating that all RmAb158-scFv8D3 variants can bind bivalently under the given conditions. Nevertheless, several differences seen between the indirect ELISA and LigandTracer results likely arise from differences between the two assay conditions such as end-point vs. real-time measurement, differences in coating concentration and presence of competing unlabeled antibody. In the ELISA, the highest apparent affinity was observed for 8D3, which was as expected because slight conformational changes in the scFv8D3 compared to the original binding sites of 8D3 may affect the epitope recognition negatively. This was supported by the LigandTracer data showing the fastest k_a for 8D3. The binding strength of RmAb158-scFv8D3 variants, despite being rather similar, showed a slight decrease with decreasing linker length in the ELISA. Thus, even though all RmAb158-scFv8D3 variants appear to bind bivalently as supported by the LigandTracer results, we speculate that shorter-linker variants may have less options to bind bivalently compared to those antibodies with longer linkers, which can span a greater distance between mTfR_Pro119 molecules (Fig. 4a). A higher number of longer-linker antibodies binding bivalently might result in a slightly higher apparent affinity of those antibodies compared to the shorter-linker antibodies. This apparent effect of antibodies with longer linkers being able to span between different TfR molecules likely resembles the antibodies’ ability to crosslink TfR molecules on the endothelial cell surface in vivo, suggesting that RmAb158-scFv8D3 with shorter linkers would crosslink TfR to a lesser extent in vivo. One caveat is that we do not know how well the density of mTfR_Pro119 at a coating concentration of 5 µg/ml on the ELISA plate relates to the TfR density on the endothelial cell surface. Furthermore, it remains unclear whether mTfR_Pro119 is monomeric or dimeric when applied as coating to the ELISA plate considering that the mass photometry analysis showed that mTfR_Pro119 is in a monomer-dimer equilibrium in solution (Fig. 2c). The TfR density and presence of monomeric vs. dimeric TfR may have a major impact on the binding behavior of the respective antibodies as these factors might determine the likelihood of an antibody to be able to span between different TfR molecules, i.e. to bind bivalently.

The more detailed kinetic analysis by LigandTracer with a competitive unlabeled binder added in the dissociation phase revealed a trend of decreasing k_d with decreasing linker length of RmAb158-scFv8D3 variants. This observation may seem counterintuitive, but could be ascribed to the ability of the unlabeled antibodies to compete for their epitope on the TfR. In that case, an antibody with a long linker has a larger radius of freedom when one arm is not bound than an antibody with a short linker. Thus, more competition with the unlabeled antibodies during unbinding and rebinding antibody arms is likely to happen with a long linker. Due to the very fast k_a of all antibodies tested in this study, any dissociation would likely be invisible in the absence of competing, unlabeled antibodies, because a dissociated antibody arm would quickly rebind. However, in the presence of unlabeled antibodies the competition for rebinding leads to a progressive displacement of target-bound, labeled antibodies by unlabeled antibodies and results in a faster drop in signal where more competition is taking place.

We further hypothesize that the shortest linker variants with − 2 aa and 1 aa linkers are able to bind bivalently to the two subunits of one TfR dimer when the two scFv8D3 sit “on top” in between the two antibody arms (Fig. 4a), but are less likely to bind bivalently between two separate TfR subunits unless they are located very close to each other, due to shorter reach. This is in contrast to 8D3 that likely only binds bivalently to separate TfR subunits, but not to two subunits of the same TfR dimer (Fig. 4a). It has been reported that standard IgG antibodies have favored binding to epitopes at a distance of 10 nm, but a strongly reduced probability for binding to epitopes at distances below 8 nm³², which is the approximate distance of 8D3 epitopes on TfR (Supplementary Fig. S7). This higher stiffness of the 8D3 antibody at short epitope distances may also explain its rather low B_max value, as 8D3 likely “buries” unoccupied epitopes under a bivalently bound antibody and thus reaches target saturation faster (Fig. 4a).

Bivalent TfR binders are thought to cause significant crosslinking of TfR on the cell surface, which reduces the transcytosis efficiency at high, therapeutically relevant antibody doses^16,17,20. Therefore, monovalent TfR binders are considered to be of advantage for crossing the BBB when administered at high concentrations. At the high antibody concentration applied in our In-Cell BBB-Trans assay, TfR is considered to be saturated with antibodies which reflects the conditions found at a therapeutic antibody dose in in vivo experiments²⁷. In contrast, at the low antibody concentration in vitro, TfR is not saturated with antibodies²⁷ and bivalent TfR binders have been observed to perform better in crossing the BBB¹⁶. Despite the relatively similar apparent affinities of all RmAb158-scFv8D3 variants seen by ELISA and LigandTracer, the in vitro BBB assay showed higher transcytosis levels for the − 2 aa and 1 aa linker variants compared to all other bivalent antibodies tested at a high antibody concentration. The transcytosis levels of the − 2 aa and 1 aa linker variants were comparable to that of the monovalent control, which is surprising for antibodies that bind bivalently to TfR. These results suggest that the majority of -2 aa and 1 aa linker antibodies bind bivalently between subunits of one TfR dimer, but much less likely between separate TfR, resulting in a type of bivalent binding that does not affect the transcytosis efficiency negatively. The relative degree of trancytosis comparing different TfR-binders in the In-Cell BBB-Trans assay has previously been shown to translate very well to results obtained from in vivo experiments in mice^16,27. Hence, based on the improved in vitro transcytosis observed with the short-linker antibody variants here, we expect to also observe an enhanced delivery of these antibody variants to the brain in future in vivo experiments in mice models.

Despite reducing TfR-crosslinking to improve the antibody’s BBB transcytosis efficiency, increasing the antibody’s blood half-life is of strong interest for improved antibody delivery to the brain. Binding of a TfR-targeting antibody to TfR on reticulocytes is considered having a major impact on the antibody’s clearance from the blood³³. How the altered binding of the short-linker RmAb158-scFv8D3 variant to TfR affects the antibody’s blood half-life in vivo, remains to be investigated, but there is a chance that reduced TfR crosslinking on reticulocytes may also be beneficial for reducing its clearance from the blood circulation.

In conclusion, reducing the linker length in RmAb158-scFv8D3 does not change the apparent affinity of RmAb158-8D3 variants substantially, suggesting that a shorter linker does not force the antibodies to monovalent instead of bivalent binding. However, we hypothesize that the shorter linkers decrease the probability of forming larger crosslinked TfR networks on the endothelial cell surface, as indicated by the increased transcytosis of shorter-linker RmAb158-8D3 variants at high treatment concentrations in the In-Cell BBB-Trans assay. Hence, we suggest that the − 2 aa and 1 aa linker variants bind bivalently mostly to individual TfR dimers or dimers located very close to each other. This way of bivalent TfR binding likely does not affect the transcytosis efficiency negatively, but instead results in equally high transcytosis levels as with a purely monovalent TfR-binder administered at high concentrations. While monovalent TfR-binders are widely thought to be superior in crossing the BBB at therapeutic antibody concentrations, our results provide a new hypothesis about how a bivalent TfR-targeted BBB-shuttle can circumvent TfR-crosslinking and thus result in as good transcytosis levels as that shown for a monovalent BBB-shuttle administered at high concentrations. At low concentrations, on the other hand, the shortest linker RmAb158-scFv8D3 variants appeared to transcytose equally well as other bivalent TfR-binders and thus outperformed the purely monovalent binder at a low antibody dose. Hence, the shortest linker RmAb158-scFv8D3 variants may require lower doses than a purely monovalent TfR-binder to achieve the same brain concentrations, potentially decreasing the risk of side-effects while retaining the same therapeutic effect. Further, the simple, symmetric design of RmAb158-scFv8D3 also allows for easy production and a strong, bivalent engagement with multimeric therapeutic targets, which is highly relevant to achieve antibody-mediated clearance of toxic protein aggregates such as Aβ in Alzheimer’s disease. Thus, compared to Lecanemab, which is the FDA-approved humanized version of RmAb158³⁴, our short-linker RmA158-scFv8D3 variants present improved characteristics with the potential to achieve stronger therapeutic effects with lower antibody doses thanks to an active delivery to the brain.

Nevertheless, further in vitro and in vivo studies need to be performed to confirm our hypotheses about antibody-TfR binding stoichiometry and receptor crosslinking and that our conclusions also translate to improved in vivo brain uptake, increased blood half-life and higher therapeutic effects with the short-linker RmAb158-scFv8D3 variants.

• SEO Powered Content & PR Distribution. Get Amplified Today.
• PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
• PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
• PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
• PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
• Source: https://www.nature.com/articles/s41598-024-83627-6