Effective scaffolding of immunogens is crucial for generating conformationally selective antibodies through active immunization, particularly in the treatment of protein misfolding diseases such as Alzheimer's and Parkinson's disease. Previous computational work has revealed that a disorder-prone region of the tau protein, when in a stacked form, is predicted to structurally resemble a small, soluble protofibril, having conformational properties similar to those of experimental in vitro tau oligomers. Such an oligomeric structural mimic has the potential to serve as a vaccine immunogen design for Alzheimer's disease. In this study, we developed a cyclization scaffolding method in Rosetta, in which multiple cyclic peptides are stacked into a protofibril. Cyclization results in significant stabilization of protofibril-like structures by constraining the conformational space. Applying this method to the disorder-prone region of the tau fibril, we evaluated the metastability of the cyclized tau immunogen using molecular dynamics simulations, and we identified sequences of two cyclic constructs having high metastability in the protofibril. We then assessed their thermodynamic stability by computing the free energy required to separate a distal chain from the rest of the stacked structure. Our computational results, based on molecular dynamics simulations and free energy calculations, demonstrate that two cyclized constructs, cyclo-(VKSEKLDFKDRVQSKIFyN) and cyclo-(VKSEKLDFKDRVQSKIYvG) (lowercase letters indicate d-form amino acids), possess significantly increased thermodynamic stability in the protofibril over an uncyclized linear construct VKSEKLDFKDRVQSKI. The cyclization scaffolding approach proposed here holds promise as a means to effectively design immunogens for protein misfolding diseases, particularly those involving liposome-conjugated peptide constructs.

The COVID-19 pandemic has devastated the lives of millions and is still prevalent in our population today. Although several vaccines and therapeutics have been proposed, SARS-CoV-2 has evaded them by rapidly mutating. In this study, we propose the development of de novo designed antibodies called receptorbodies (Rbs), that are robust against SARS-CoV-2 mutations. 12 ACE2 decoys were designed in Rosetta, and tested for expression and function. Two of the 12 designed ACE2 decoys: 4H1 and 4H3, exhibited affinities of 160 nM and 137 nM, respectively, to the receptor binding domain (RBD) of SARS-CoV-2 omicron variant. The ACE2 decoys exhibited stable protein structures with predominant α-helix structure as predicted by CD spectroscopy, and showed moderate stability under urea titration. These positive leads were then grafted to the complementarity-determining regions of IgG heavy and light chains, to prolong the life of the designed Rbs in human blood and to elicit an immune response upon RBD binding. The ACE2 decoy-grafted Rbs were able to successfully engage HEK293T cells expressing BA.4/.5 spike protein in immunocytochemistry studies. Based on these results, we propose the Rbs as an active therapeutic that can target the spike protein of SARS-CoV-2, and exhibit resilience to viral RBD mutations. Additionally, the designed Rbs can be effective against other sarbecoviruses that utilize ACE2 for human host entry.

Tau pathology is associated with many neurodegenetive disorders, including Alzheimer’s disease (AD), where the spatio-temporal pattern of tau neurofibrillary tangles strongly correlates with disease progression, which motivates therapeutics selective for misfolded tau. Here, we introduce a new avidity-enhanced, multi-epitope approach for protein misfolding immunogen design, which is predicted to mimic the conformational state of an exposed epitope in toxic tau oligomers. A predicted oligomer-selective tau epitope 343 KLDFK 347 was scaffolded by designing a β -helix structure that incorporated multiple instances of the 16-residue tau fragment 339 VKSEKLDFKDRVQSKI 354 . Largescale conformational ensemble analyses involving Jensen-Shannon Divergence and the embedding depth 𝒟 showed that the multi-epitope scaffolding approach, employed in designing the β -helix scaffold, was predicted to better discriminate toxic tau oligomers than other “monovalent” strategies utilizing a single instance of an epitope for vaccine immunogen design. Using Rosetta, 10,000 sequences were designed and screened for the linker portions of the β -helix scaffold, along with a C-terminal stabilizing α -helix that interacts with the linkers, to optimize the folded structure and stability of the scaffold. Structures were ranked by energy, and the lowest 1% (82 unique sequences) were verified using AlphaFold. Several selection criteria involving AlphaFold are implemented to obtain a lead designed sequence. The structure was further predicted to have free energetic stability by using Hamiltonian replica exchange molecular dynamics (MD) simulations. The synthesized β -helix scaffold showed direct binding in surface plasmon resonance (SPR) experiments to several antibodies that were raised to the structured epitope using a designed cyclic peptide. Moreover the strength of binding of these antibodies to in vitro tau oligomers correlated with the strength of binding to the β -helix construct, suggesting that the construct presents an oligomer-like conformation and may thus constitute an effective oligomer-selective immunogen.

Cytoplasmically mislocalized aggregates of TDP‐43 have been implicated in the pathogenesis of amyotrophic lateral sclerosis (ALS), frontotemporal lobar degeneration (FTLD) and limbic‐predominant age‐related TDP‐43 encephalopathy (LATE) through direct toxicity, loss of function of normal TDP‐43, induction of misfolding of other neuronal proteins, and prion‐like, cell‐to‐cell propagation of disease. We generated monoclonal antibodies (mAbs) selective for the misfolded form of TDP‐43 and derived single chain intrabody constructs to allow for intracellular targeting of pathogenic TDP‐43 without interfering with physiological forms of TDP‐43 important for normal cell function.

Amyloid-β (Aβ) and tau proteins currently represent the two most promising targets to treat Alzheimer's disease. The most extensively developed method to treat the pathologic forms of these proteins is through the administration of exogenous antibodies, or passive immunotherapy. In this review, we discuss the molecular-level strategies that researchers are using to design an effective therapeutic antibody, given the challenges in treating this disease. These challenges include selectively targeting a protein that has misfolded or is pathological rather than the more abundant, healthy protein, designing strategic constructs for immunizing an animal to raise an antibody that has the appropriate conformational selectivity to achieve this end, and clearing the pathological protein species before prion-like cell-to-cell spread of misfolded protein has irreparably damaged neurons, without invoking damaging inflammatory responses in the brain that naturally arise when the innate immune system is clearing foreign agents. The various solutions to these problems in current clinical trials will be discussed.

Mislocalization, cleavage, and aggregation of the human protein TDP-43 is found in many neurodegenerative diseases. As is the case with many other proteins that are completely or partially structurally disordered, production of full-length recombinant TDP-43 in the quantities necessary for structural characterization has proved difficult. We show that the full-length TDP-43 protein and two truncated N-terminal constructs 1-270 and 1-263 can be heterologously expressed in E. coli. Full-length TDP-43 could be prevented from aggregation during purification using a detergent. Crystals grown from an N-terminal construct (1-270) revealed only the N-terminal domain (residues 1-80) with molecules arranged as parallel spirals with neighboring molecules arranged in head-to-tail fashion. To obtain detergent-free, full-length TDP-43 we mutated all six tryptophan residues to alanine. This provided sufficient soluble protein to collect small-angle X-ray scattering data. Refining relative positions of individual domains and intrinsically disordered regions against this data yielded a model of full-length TDP-43.

AbstractBackgroundMisfolded, aggregated TDP‐43 has been implicated in the pathogenesis of amyotrophic lateral sclerosis (ALS), frontotemporal lobar dementia (FTLD) and limbic‐predominant age‐related TDP‐43 encephalopathy (LATE) through direct toxicity, loss of function of normal TDP‐43, induction of misfolding of other neuronal proteins, and prion‐like, cell‐to‐cell propagation of disease. We sought to generate antibodies selectively targeting the misfolded, pathogenic form of TDP‐43 while sparing physiological forms of TDP‐43 important for normal cell function.MethodMice and rabbits were immunized with an unfolded N‐terminal domain (NTD) linear epitope predicted to become exposed in cytosolically mislocalized, aggregated TDP‐43 but otherwise buried in natively folded TDP‐43. Selectivity of polyclonal antibody (pAb) for misfolded NTD was confirmed by studies with denaturing and native gel electrophoresis followed by immunoblotting. Monoclonal antibody (mAb) affinity for the immunizing peptide was measured by surface plasmon resonance (SPR). mAb selectivity for pathogenic, aggregated TDP43 was assessed by immunocytochemistry (ICC) of HEK293FT cells transfected with mutant TDP‐43 lacking a functional nuclear localization signal (DNLS) or upon arsenite stress, and by immunohistochemistry (IHC) on patient samples. The ability of mAbs to inhibit cell‐to‐cell transmission of DNLS TDP‐43 was evaluated in cell culture.ResultAffinity‐purified pAb from immunized animals displayed reactivity with recombinant NTD under denaturing but not native conditions, indicating selectivity for unfolded NTD. mAb clones displayed pM affinity for the NTD epitope by SPR. ICC showed mAb reactivity with cytoplasmic aggregates of DNLS‐TDP‐43 but not wild‐type nuclear TDP‐43. Antibodies also did not recognize TDP‐43 in physiological stress granules in HEK‐293FT cells. IHC of ALS and FTLD CNS sections, but not normal control, confirmed selective immunoreactivity of mAbs with pathogenic TDP‐43. In cell culture, mAbs inhibited transmission of misfolding TDP‐43 from the conditioned medium of donor HEK293FT cells transfected with DNLS‐TDP‐43 to naïve recipient cells.ConclusionImmunization with an NTD epitope of misfolded TDP‐43 gave rise to mAbs selective for pathogenic vs physiologically important forms of TDP‐43. The mAbs were capable of inhibiting cell‐to‐cell propagation of misfolding TDP43 in vitro. Such antibodies may have value against extracellular transmission of misfolding TDP‐43 or as pathogenic TDP43‐specific intrabodies expressed intracellularly.

Previous studies of Alzheimer’s disease (AD) pathology point to cytotoxic tau as a cause of neuronal cell death, which is induced or exacerbated by soluble misfolded Aβ oligomers. Soluble misfolded species of both tau and Aβ are both observed to propagate cell‐to‐cell. A method for identifying antibodies to tau and Aβ that are conformationally‐selective to propagative misfolded oligomeric forms, and which also have low affinity to isolated monomers or, particularly for Aβ, low affinity to fibrils, is thus a highly desired goal that holds significant promise for AD therapy.

Natively folded TAR-DNA binding protein-43 (TDP-43) possesses a large physiological protein interactome, but pathological TDP-43 can co-aggregate with additional proteins, including karyopherins (Chou et al, Nat Neurosci 2018) and disrupted in schizophrenia-1 (DISC1, Endo et al, Biological Psych 2018). We have reported that cytosolically mislocalized TDP-43 (dNLS-TDP-43) induces misfolding of human wild-type SOD1 (HuWtSOD1) in cultured cells, including primary neurons (Pokrishevsky et al, PLoS One 2012), and that such misfolded SOD1 acquires the prion-like property of transmitting its misfold to recipient cells (Pokrishevsky et al, Sci Rep 2016). We have also reported that the sole tryptophan (Trp) residue at codon 32 restricts the self-seeding of mutant and wild-type SOD1 in cell culture (Grad et al, PNAS 2011; Pokrishevsky et al, Sci Rep 2018). We generated Trp-less versions of dNLS-TDP-43 containing various Trps/serine substitutions, and assessed their ability to induce SOD1 misfolding and aggregation in HEK293 cells and in zebrafish. Immunocytochemical studies revealed that dNLS-TDP-43, but not its Trp-less version lacking all Trps, induces significant misfolding of endogenous HuWtSOD1, as well as aggregation of a co-transfected fluorescent reporter protein, SOD1-G85R-GFP. We determined that three Trps in the C-terminal low complexity domain of TDP-43 subserve TDP-43 aggregation, consistent with recent publications (Hughes et al, Science 2018; Li et al, JBC 2018), and that N-terminal Trps are essential for the cross-seeding and induced misfolding of HuWtSOD1. Cytosolic inclusions of TDP-43 were reactive with the novel RRM1 misfolding-selective monoclonal antibody 9C5 (ProMIS Neurosciences), confirming disruption of the native structure of this RNA binding domain, which possesses two of the N-terminal Trps. While co-expression of HuWtSOD1 with wtTDP-43 in zebrafish triggers axonopathy and swimming defects, Trp-less versions of either protein significantly alleviate this pathology. Pathological TDP-43 can trigger misfolding and aggregation of HuWtSOD1, a toxic collaboration consistent with homophillic Trp-Trp interactions involving misfolding-exposed Trps in the N-terminal structured domains of TDP-43 and Trp32 of SOD1. Further studies will elucidate the consequences of this and other pathological interactions for TDP-43 in frontal temporal dementia, amyotrophic lateral sclerosis, and other diseases characterized by cytosolic mislocalization and aggregation of TDP-43.

Oligomers of amyloid-β (AβO) are deemed key in synaptotoxicity and amyloid seeding of Alzheimer's disease (AD). However, the heterogeneous and dynamic nature of AβO and inadequate markers for AβO subtypes have stymied effective AβO identification and therapeutic targeting in vivo. We identified an AβO-subclass epitope defined by differential solvent orientation of the lysine 28 side chain in a constrained loop of serine-asparagine-lysine (cSNK), rarely displayed in molecular dynamics simulations of monomer and fibril ensembles. A mouse monoclonal antibody targeting AβOcSNK recognizes ∼50-60 kDa SDS-resistant soluble Aβ assemblages in AD brain and prolongs the lag phase of Aβ aggregation in vitro. Acute peripheral infusion of a murine IgG1 anti-AβOcSNK in two AD mouse models reduced soluble brain Aβ aggregates by 20-30%. Chronic cSNK peptide immunization of APP/PS1 mice engendered an anti-AβOcSNK IgG1 response without epitope spreading to Aβ monomers or fibrils and was accompanied by preservation of global PSD95 expression and improved cued fear memory. Our data indicate that the oligomer subtype AβOcSNK participates in synaptotoxicity and propagation of Aβ aggregation in vitro and in vivo.

P2-048 HUMANIZED PMN310 SHOWS ENHANCED THERAPEUTIC POTENTIAL BY BINDING TOXIC LOW MOLECULARWEIGHTAb OLIGOMERS WHILE AVOIDING ARIARELATED BINDING TO Ab DEPOSITS IN AD PATIENT BRAINS Johanne Kaplan, Ebrima Gibbs, Judith M. Silverman, Jing Wang, Xubiao Peng, Steven S. Plotkin, Neil R. Cashman, ProMIS Neurosciences, Toronto, ON, Canada; University of British Columbia, Vancouver, BC, Canada. Contact e-mail: johanne.kaplan@ promisneurosciences.com

Mutations in the SOD1 gene are associated with some forms of familial amyotrophic lateral sclerosis (fALS). There are more than 150 different mutations in the SOD1 gene that have various effects on the copper-zinc superoxide dismutase (SOD1) enzyme structure, including the loss of metal binding and a decrease in dimer affinity. The copper-based therapeutic CuATSM has been proven to be effective at rescuing neuronal cells from SOD1 mutant toxicity and has also increased the life expectancy of mice expressing the human transgenes SOD1G93A and SOD1G37R. Furthermore, CuATSM is currently the subject of a phase I/II clinical trial in Australia as a treatment for ALS. To determine if CuATSM protects against a broad variety of SOD1 mutations, we used a well-established cell culture model of SOD1-fALS. NSC-34 cells expressing SOD1-EGFP constructs were treated with CuATSM and examined by time-lapse microscopy. Our results show a concentration-dependent protection of cells expressing mutant SOD1A4V over the experimental time period. We tested the efficacy of CuATSM on 10 SOD1-fALS mutants and found that while protection was observed in cells expressing pathogenic wild-type-like mutants, cells expressing a truncation mutant or metal binding region mutants were not. We also show that CuATSM rescue is associated with an increase in human SOD1 activity and a decrease in the level of SOD1 aggregation in vitro. In conclusion, CuATSM has shown to be a promising therapeutic for SOD1-associated ALS; however, our in vitro results suggest that the protection afforded varies depending on the SOD1 variant, including negligible protection to mutants with deficient copper binding.

Using Molecular Dynamic simulations, we predicted the epitopes on Aβ oligomers, which would be distinct from those in either Aβ monomer or Aβ fibril. Starting from the predicted epitopes, 66 antibodies are raised by passive immunization in mice, and their binding to Aβ monomer, oligomer and plaque, along with preferential binding to Aβ protein in the cerebrospinal fluid and brain homogenate from deceased Alzheimer Disease (AD) patients vs healthy controls are measured experimentally. These experimental measurements provide up to eight different screening criteria to select lead candidates for clinical stages of drug development. The screening criteria do not perfectly correlate with each other and there is uncertainty in the criteria values as well as their importance/weights, making the selection of lead candidates a nontrivial maximum-likelihood ranking problem. In other fields, this problem is referred to generally as multiple criteria decision making or MCDM, and has been applied in various fields including finance and economics, energy policy, and environmental science, but not—at least to our knowledge—to the problem of screening candidate drug therapeutics. We employ a stochastic MCDM method named "SMAA-TOPSIS" to screen and rank the measured large cohort of the antibodies, in order to find lead candidate therapeutics. We obtain distributions of candidate rankings due to uncertainty in screening measurements, as well as the user-defined weight of importance attributed to each screening criterion. In choosing lead candidates, we propose a quantity "topness" as the most robust measure of ranking. Our hope is that this method may enable more systematic screening of candidate therapeutics when it becomes difficult intuitively to process multi-variate screening data that distinguishes candidates, so that additional candidates will be exposed as potential leads, increasing the likelihood of success in downstream clinical trials.

P2-048 HUMANIZED PMN310 SHOWS ENHANCED THERAPEUTIC POTENTIAL BY BINDING TOXIC LOW MOLECULARWEIGHTAb OLIGOMERS WHILE AVOIDING ARIARELATED BINDING TO Ab DEPOSITS IN AD PATIENT BRAINS Johanne Kaplan, Ebrima Gibbs, Judith M. Silverman, Jing Wang, Xubiao Peng, Steven S. Plotkin, Neil R. Cashman, ProMIS Neurosciences, Toronto, ON, Canada; University of British Columbia, Vancouver, BC, Canada. Contact e-mail: johanne.kaplan@ promisneurosciences.com

We introduce a global, collective coordinate bias into molecular dynamics simulations that partially unfolds a protein, in order to predict misfolding-specific epitopes based on the regions that locally unfold. Several metrics are used to measure local disorder, including solvent exposed surface area (SASA), native contacts (Q), and root mean squared fluctuations (RMSF). The method is applied to Cu, Zn superoxide dismutase (SOD1). For this protein, the processes of monomerization, metal loss, and conformational unfolding due to microenvironmental stresses are all separately taken into account. Several misfolding-specific epitopes are predicted, and consensus epitopes are calculated. These predicted epitopes are consistent with the "lower-resolution" peptide sequences used to raise disease-specific antibodies, but the epitopes derived from collective coordinates contain shorter, more refined sequences for the key residues constituting the epitope.

The underlying physical causes of SOD1-related ALS are still not well-understood. We address this problem here by computationally designing two de novo mutants, A89R and K128N, which were predicted theoretically to be either significantly destabilizing or stabilizing respectively. We subjected these in silico designed mutants to a series of experimental tests, including in vitro measures of thermodynamic stability, cell-based aggregation and toxicity assays, and an in vivo developmental model of zebrafish motor neuron axonopathy. The experimental tests validated the theoretical predictions: A89R is an unstable, highly-deleterious mutant, and K128N is a stable, non-toxic mutant. Moreover, K128N is predicted computationally to form an unusually stable heterodimer with the familial ALS mutant A4V. Consistent with this prediction, co-injection of K128N and A4V into zebrafish shows profound rescue of motor neuron pathology. The demonstrated success of these first principles calculations to predict the physical properties of SOD1 mutants holds promise for rationally designed therapies to counter the progression of ALS.Significance Mutations in the protein superoxide dismutase cause ALS, and many of these mutants have decreased folding stability. We sought to pursue this thread using a synthetic biology approach, where we designed two de novo mutations, one stabilizing and one destabilizing, as predicted using computational molecular dynamics simulations. We then tested these mutants using in vitro , cell-based, and in vivo zebrafish models. We found that the unstable mutant was toxic, and induced a severe ALS phenotype in zebrafish; the predicted stable mutant, on the other hand, behaved even better than WT. In fact, it was able to rescue the ALS phenotype caused by mutant SOD1. We propose a mechanism for this rescue, which may provide an avenue for therapeutic intervention.

Mechanical unfolding of mutated apo, disulfide-reduced, monomeric superoxide dismutase 1 protein (SOD1) has been simulated via force spectroscopy techniques, using both an all-atom (AA), explicit solvent model and a coarse-grained heavy-atom Gō (HA-Gō) model. The HA-Gō model was implemented at two different pulling speeds for comparison. The most-common sequence of unfolding in the AA model agrees well with the most-common unfolding sequence of the HA-Gō model, when the same normalized pulling rate was used. Clustering of partially-native structures as the protein unfolds shows that the AA and HA-Gō models both exhibit a dominant pathway for early unfolding, which eventually bifurcates repeatedly to multiple branches after the protein is about half-unfolded. The force-extension curve exhibits multiple force drops, which are concomitant with jumps in the local interaction potential energy between specific β-strands in the protein. These sudden jumps in the potential energy coincide with the dissociation of specific pairs of β-strands, and thus intermediate unfolding events. The most common sequence of β-strand dissociation in the unfolding pathway of the AA model is β-strands 5, 4, 8, 7, 1, 2, then finally β-strands 3 and 6. The observation that β-strand 5 is among the first to unfold here, but the last to unfold in simulations of loop-truncated SOD1, could imply the existence of an evolutionary compensation mechanism, which would stabilize β-strands flanking long loops against their entropic penalty by strengthening intramolecular interactions. This article is part of a Special Issue entitled: Biophysics in Canada, edited by Lewis Kay, John Baenziger, Albert Berghuis and Peter Tieleman.

Mechanical unfolding of a single domain of loop-truncated superoxide dismutase protein has been simulated via force spectroscopy techniques with both all-atom (AA) models and several coarse-grained models having different levels of resolution: A Gō model containing all heavy atoms in the protein (HA-Gō), the associative memory, water mediated, structure and energy model (AWSEM) which has 3 interaction sites per amino acid, and a Gō model containing only one interaction site per amino acid at the Cα position (Cα-Gō). To systematically compare results across models, the scales of time, energy, and force had to be suitably renormalized in each model. Surprisingly, the HA-Gō model gives the softest protein, exhibiting much smaller force peaks than all other models after the above renormalization. Clustering to render a structural taxonomy as the protein unfolds showed that the AA, HA-Gō, and Cα-Gō models exhibit a single pathway for early unfolding, which eventually bifurcates repeatedly to multiple branches only after the protein is about half-unfolded. The AWSEM model shows a single dominant unfolding pathway over the whole range of unfolding, in contrast to all other models. TM alignment, clustering analysis, and native contact maps show that the AWSEM pathway has however the most structural similarity to the AA model at high nativeness, but the least structural similarity to the AA model at low nativeness. In comparison to the AA model, the sequence of native contact breakage is best predicted by the HA-Gō model. All models consistently predict a similar unfolding mechanism for early force-induced unfolding events, but diverge in their predictions for late stage unfolding events when the protein is more significantly disordered.