Structure-based design of nanobodies that inhibit seeding of Alzheimer's patient-extracted tau fibrils.

Despite much effort, antibody therapies for Alzheimer's disease (AD) have shown limited efficacy. Challenges to the rational design of effective antibodies include the difficulty of achieving specific affinity to critical targets, poor expression, and antibody aggregation caused by buried charges and unstructured loops. To overcome these challenges, we grafted previously determined sequences of fibril-capping amyloid inhibitors onto a camel heavy chain antibody scaffold. These sequences were designed to cap fibrils of tau, known to form the neurofibrillary tangles of AD, thereby preventing fibril elongation. The nanobodies grafted with capping inhibitors blocked tau aggregation in biosensor cells seeded with postmortem brain extracts from AD and progressive supranuclear palsy (PSP) patients. The tau capping nanobody inhibitors also blocked seeding by recombinant tau oligomers. Another challenge to the design of effective antibodies is their poor blood-brain barrier (BBB) penetration. In this study, we also designed a bispecific nanobody composed of a nanobody that targets a receptor on the BBB and a tau capping nanobody inhibitor, conjoined by a flexible linker. We provide evidence that the bispecific nanobody improved BBB penetration over the tau capping inhibitor alone after intravenous administration in mice. Our results suggest that the design of synthetic antibodies that target sequences that drive protein aggregation may be a promising approach to inhibit the prion-like seeding of tau and other proteins involved in AD and related proteinopathies.

Sensitivity Analysis for Observational Studies with Recurrent Events.

We conduct an observational study of the effect of sickle cell trait Haemoglobin AS (HbAS) on the hazard rate of malaria fevers in children. Assuming no unmeasured confounding, there is strong evidence that HbAS reduces the rate of malarial fevers. Since this is an observational study, however, the no unmeasured confounding assumption is strong. A sensitivity analysis considers how robust a conclusion is to a potential unmeasured confounder. We propose a new sensitivity analysis method for recurrent event data and apply it to the malaria study. We find that for the causal conclusion that HbAS is protective against malarial fevers to be overturned, the hypothesized unmeasured confounder must be as influential as all but one of the measured confounders.

Simple and Low-Cost Electroactive Membranes for Ammonia Recovery.

Ammonia is considered a contaminant to be removed from wastewater. However, ammonia is a valuable commodity chemical used as the primary feedstock for fertilizer manufacturing. Here we describe a simple and low-cost ammonia gas stripping membrane capable of recovering ammonia from wastewater. The material is composed of an electrically conducting porous carbon cloth coupled to a porous hydrophobic polypropylene support, that together form an electrically conductive membrane (ECM). When a cathodic potential is applied to the ECM surface, hydroxide ions are produced at the water-ECM interface, which transforms ammonium ions into higher-volatility ammonia that is stripped across the hydrophobic membrane material using an acid-stripping solution. The simple structure, low cost, and easy fabrication process make the ECM an attractive material for ammonia recovery from dilute aqueous streams, such as wastewater. When paired with an anode and immersed into a reactor containing synthetic wastewater (with an acid-stripping solution providing the driving force for ammonia transport), the ECM achieved an ammonia flux of 141.3 ± 14.0 g.cm-2.day-1 at a current density of 6.25 mA.cm-2 (69.2 ± 5.3 kg(NH3-N)/kWh). It was found that the ammonia flux was sensitive to the current density and acid circulation rate.

Proximal Causal Inference without Uniqueness Assumptions.

We consider identification and inference about a counterfactual outcome mean when there is unmeasured confounding using tools from proximal causal inference. Proximal causal inference requires existence of solutions to at least one of two integral equations. We motivate the existence of solutions to the integral equations from proximal causal inference by demonstrating that, assuming the existence of a solution to one of the integral equations, n-estimability of a mean functional of that solution requires the existence of a solution to the other integral equation. Solutions to the integral equations may not be unique, which complicates estimation and inference. We construct a consistent estimator for the solution set for one of the integral equations and then adapt the theory of extremum estimators to find from the estimated set a consistent estimator for a uniquely defined solution. A debiased estimator is shown to be root-n consistent, regular, and semiparametrically locally efficient under additional regularity conditions.

Slowing the loss of physical function in amyotrophic lateral sclerosis with edaravone: Post hoc analysis of ALSFRS-R item scores in pivotal study MCI186-19.

INTRODUCTION: Phase 3 study MCI186-19 demonstrated less loss of physical function with edaravone versus placebo, as measured by the revised Amyotrophic Lateral Sclerosis Functional Rating Scale (ALSFRS-R) total score. A 1-point drop in an individual ALSFRS-R item may be clinically meaningful. We assessed ALSFRS-R item score changes to identify clinical features protected by edaravone treatment. METHODS: Time-to-event analysis was used to assess the cumulative probabilities of reductions in ALSFRS-R item scores and Amyotrophic Lateral Sclerosis Assessment Questionnaire (ALSAQ-40) subdomain scores. RESULTS: Edaravone use was accompanied by: (1) delayed drop of ≥1 point in ALSFRS-R item score for four items: salivation, walking, climbing stairs, orthopnea (unadjusted), or for two items: walking, climbing stairs (after Bonferroni correction for multiple comparisons); (2) delayed score transition from 4 or 3 at baseline to ≤2 for five items: swallowing, eating motion, walking, climbing stairs, orthopnea (unadjusted), or for one item: climbing stairs (after Bonferroni correction for multiple comparisons); and (3) delayed worsening of ALSAQ-40 domain scores representing daily living/independence, eating and drinking (unadjusted). DISCUSSION: These post-hoc analyses identified the ALSFRS-R item scores and ALSAQ-40 domain scores that were associated with preserved gross motor function and health-related quality of life, respectively, after edaravone treatment. Limitations of post-hoc analyses should be considered when interpreting these results. We recommend that clinical trials employing the ALSFRS-R include this type of analysis as a pre-specified secondary outcome measure.

Auxora vs. placebo for the treatment of patients with severe COVID-19 pneumonia: a randomized-controlled clinical trial.

BACKGROUND: Calcium release-activated calcium (CRAC) channel inhibitors block proinflammatory cytokine release, preserve endothelial integrity and may effectively treat patients with severe COVID-19 pneumonia. METHODS: CARDEA was a phase 2, randomized, double-blind, placebo-controlled trial evaluating the addition of Auxora, a CRAC channel inhibitor, to corticosteroids and standard of care in adults with severe COVID-19 pneumonia. Eligible patients were adults with ≥ 1 symptom consistent with COVID-19 infection, a diagnosis of COVID-19 confirmed by laboratory testing using polymerase chain reaction or other assay, and pneumonia documented by chest imaging. Patients were also required to be receiving oxygen therapy using either a high flow or low flow nasal cannula at the time of enrolment and have at the time of enrollment a baseline imputed PaO2/FiO2 ratio > 75 and ≤ 300. The PaO2/FiO2 was imputed from a SpO2/FiO2 determine by pulse oximetry using a non-linear equation. Patients could not be receiving either non-invasive or invasive mechanical ventilation at the time of enrolment. The primary endpoint was time to recovery through Day 60, with secondary endpoints of all-cause mortality at Day 60 and Day 30. Due to declining rates of COVID-19 hospitalizations and utilization of standard of care medications prohibited by regulatory guidance, the trial was stopped early. RESULTS: The pre-specified efficacy set consisted of the 261 patients with a baseline imputed PaO2/FiO2≤ 200 with 130 and 131 in the Auxora and placebo groups, respectively. Time to recovery was 7 vs. 10 days (P = 0.0979) for patients who received Auxora vs. placebo, respectively. The all-cause mortality rate at Day 60 was 13.8% with Auxora vs. 20.6% with placebo (P = 0.1449); Day 30 all-cause mortality was 7.7% and 17.6%, respectively (P = 0.0165). Similar trends were noted in all randomized patients, patients on high flow nasal cannula at baseline or those with a baseline imputed PaO2/FiO2 ≤ 100. Serious adverse events (SAEs) were less frequent in patients treated with Auxora vs. placebo and occurred in 34 patients (24.1%) receiving Auxora and 49 (35.0%) receiving placebo (P = 0.0616). The most common SAEs were respiratory failure, acute respiratory distress syndrome, and pneumonia. CONCLUSIONS: Auxora was safe and well tolerated with strong signals in both time to recovery and all-cause mortality through Day 60 in patients with severe COVID-19 pneumonia. Further studies of Auxora in patients with severe COVID-19 pneumonia are warranted. Trial registration NCT04345614.

Long-term edaravone efficacy in amyotrophic lateral sclerosis: Post-hoc analyses of Study 19 (MCI186-19).

BACKGROUND: In a Phase 3 study, amyotrophic lateral sclerosis (ALS) patients experienced significantly less physical functional decline with 24-week edaravone vs placebo, followed by open-label treatment for an additional 24 weeks. METHODS: Outcome (the change in ALS Functional Rating Scale-Revised, ALSFRS-R, from baseline) was projected for placebo patients through 48 weeks and compared with 48-week edaravone or 24-week edaravone after switching from placebo. RESULTS: A total of 123 patients received open-label treatment (65 edaravone-edaravone; 58 placebo-edaravone). The projected ALSFRS-R decline for placebo from baseline through week 48 was greater than for 48-week edaravone (P < .0001). For patients switching from placebo to edaravone, ALSFRS-R slope approached that of continued edaravone for 48 weeks. ALSFRS-R decline did not differ between actual and projected edaravone through week 48. CONCLUSIONS: Compared with placebo, these analyses suggest that edaravone is beneficial in ALS patients even after 6 mo of receiving placebo, and efficacy is maintained for up to 1 year.

Auxora versus standard of care for the treatment of severe or critical COVID-19 pneumonia: results from a randomized controlled trial.

BACKGROUND: Calcium release-activated calcium (CRAC) channel inhibitors stabilize the pulmonary endothelium and block proinflammatory cytokine release, potentially mitigating respiratory complications observed in patients with COVID-19. This study aimed to investigate the safety and efficacy of Auxora, a novel, intravenously administered CRAC channel inhibitor, in adults with severe or critical COVID-19 pneumonia. METHODS: A randomized, controlled, open-label study of Auxora was conducted in adults with severe or critical COVID-19 pneumonia. Patients were randomized 2:1 to receive three doses of once-daily Auxora versus standard of care (SOC) alone. The primary objective was to assess the safety and tolerability of Auxora. Following FDA guidance, study enrollment was halted early to allow for transition to a randomized, blinded, placebo-controlled study. RESULTS: In total, 17 patients with severe and three with critical COVID-19 pneumonia were randomized to Auxora and nine with severe and one with critical COVID-19 pneumonia to SOC. Similar proportions of patients receiving Auxora and SOC experienced ≥ 1 adverse event (75% versus 80%, respectively). Fewer patients receiving Auxora experienced serious adverse events versus SOC (30% versus 50%, respectively). Two patients (10%) receiving Auxora and two (20%) receiving SOC died during the 30 days after randomization. Among patients with severe COVID-19 pneumonia, the median time to recovery with Auxora was 5 days versus 12 days with SOC; the recovery rate ratio was 1.87 (95% CI, 0.72, 4.89). Invasive mechanical ventilation was needed in 18% of patients with severe COVID-19 pneumonia receiving Auxora versus 50% receiving SOC (absolute risk reduction = 32%; 95% CI, - 0.07, 0.71). Outcomes measured by an 8-point ordinal scale were significantly improved for patients receiving Auxora, especially for patients with a baseline PaO2/FiO2 = 101-200. CONCLUSIONS: Auxora demonstrated a favorable safety profile in patients with severe or critical COVID-19 pneumonia and improved outcomes in patients with severe COVID-19 pneumonia. These results, however, are limited by the open-label study design and small patient population resulting from the early cessation of enrollment in response to regulatory guidance. The impact of Auxora on respiratory complications in patients with severe COVID-19 pneumonia will be further assessed in a planned randomized, blinded, placebo-controlled study. TRIAL REGISTRATION: ClinicalTrials.gov, NCT04345614 . Submitted on 7 April 2020.

Fully Automated Echocardiogram Interpretation in Clinical Practice.

BACKGROUND: Automated cardiac image interpretation has the potential to transform clinical practice in multiple ways, including enabling serial assessment of cardiac function by nonexperts in primary care and rural settings. We hypothesized that advances in computer vision could enable building a fully automated, scalable analysis pipeline for echocardiogram interpretation, including (1) view identification, (2) image segmentation, (3) quantification of structure and function, and (4) disease detection. METHODS: Using 14 035 echocardiograms spanning a 10-year period, we trained and evaluated convolutional neural network models for multiple tasks, including automated identification of 23 viewpoints and segmentation of cardiac chambers across 5 common views. The segmentation output was used to quantify chamber volumes and left ventricular mass, determine ejection fraction, and facilitate automated determination of longitudinal strain through speckle tracking. Results were evaluated through comparison to manual segmentation and measurements from 8666 echocardiograms obtained during the routine clinical workflow. Finally, we developed models to detect 3 diseases: hypertrophic cardiomyopathy, cardiac amyloid, and pulmonary arterial hypertension. RESULTS: Convolutional neural networks accurately identified views (eg, 96% for parasternal long axis), including flagging partially obscured cardiac chambers, and enabled the segmentation of individual cardiac chambers. The resulting cardiac structure measurements agreed with study report values (eg, median absolute deviations of 15% to 17% of observed values for left ventricular mass, left ventricular diastolic volume, and left atrial volume). In terms of function, we computed automated ejection fraction and longitudinal strain measurements (within 2 cohorts), which agreed with commercial software-derived values (for ejection fraction, median absolute deviation=9.7% of observed, N=6407 studies; for strain, median absolute deviation=7.5%, n=419, and 9.0%, n=110) and demonstrated applicability to serial monitoring of patients with breast cancer for trastuzumab cardiotoxicity. Overall, we found automated measurements to be comparable or superior to manual measurements across 11 internal consistency metrics (eg, the correlation of left atrial and ventricular volumes). Finally, we trained convolutional neural networks to detect hypertrophic cardiomyopathy, cardiac amyloidosis, and pulmonary arterial hypertension with C statistics of 0.93, 0.87, and 0.85, respectively. CONCLUSIONS: Our pipeline lays the groundwork for using automated interpretation to support serial patient tracking and scalable analysis of millions of echocardiograms archived within healthcare systems.

A multicenter case registry study on medication-related osteonecrosis of the jaw in patients with advanced cancer.

PURPOSE: This observational case registry study was designed to describe the natural history of cancer patients with medication-related osteonecrosis of the jaw (ONJ) and evaluate the ONJ resolution rate. METHODS: Adults with a diagnosis of cancer and with a new diagnosis of ONJ were enrolled and evaluated by a dental specialist at baseline and every 3 months for 2 years and then every 6 months for 3 years until death, consent withdrawal, or loss to follow-up. The primary endpoint was the rate and time course of ONJ resolution. Secondary endpoints included frequency of incident ONJ risk factors, ONJ treatment patterns, and treatment patterns of antiresorptive agents for subsequent ONJ. RESULTS: Overall, 327 patients were enrolled; 207 (63%) were continuing on study at data cutoff. Up to 69% of evaluable patients with ONJ had resolution or improvement during the study. ONJ resolution (AAOMS ONJ staging criteria) was observed in 114 patients (35%); median (interquartile range) time from ONJ onset to resolution was 7.3 (4.5-11.4) months. Most patients (97%) had received antiresorptive medication before ONJ development, 9 patients (3%) had not; 68% had received zoledronic acid, 38% had received denosumab, and 10% had received pamidronate (56% had received bisphosphonates only, 18% had received denosumab only, and 21% had exposure to both). CONCLUSIONS: These results are consistent with those observed in clinical trials evaluating skeletal-related events in patients with advanced malignancy involving bone. Longer follow-up will provide further information on ONJ recurrence and resolution rates between medically and surgically managed patients.

Discovery of Pyrophosphate Diesters as Tunable, Soluble, and Bioorthogonal Linkers for Site-Specific Antibody-Drug Conjugates.

As part of an effort to examine the utility of antibody-drug conjugates (ADCs) beyond oncology indications, a novel pyrophosphate ester linker was discovered to enable the targeted delivery of glucocorticoids. As small molecules, these highly soluble phosphate ester drug linkers were found to have ideal orthogonal properties: robust plasma stability coupled with rapid release of payload in a lysosomal environment. Building upon these findings, site-specific ADCs were made between this drug linker combination and an antibody against human CD70, a receptor specifically expressed in immune cells but also found aberrantly expressed in multiple human carcinomas. Full characterization of these ADCs enabled procession to in vitro proof of concept, wherein ADCs 1-22 and 1-37 were demonstrated to afford potent, targeted delivery of glucocorticoids to a representative cell line, as measured by changes in glucocorticoid receptor-mediated gene mRNA levels. These activities were found to be antibody-, linker-, and payload-dependent. Preliminary mechanistic studies support the notion that lysosomal trafficking and enzymatic linker cleavage are required for activity and that the utility for the pyrophosphate linker may be general for internalizing ADCs as well as other targeted delivery platforms.

A Study of Biomedical Relation Extraction Using GPT Models.

Relation Extraction (RE) is a natural language processing (NLP) task for extracting semantic relations between biomedical entities. Recent developments in pre-trained large language models (LLM) motivated NLP researchers to use them for various NLP tasks. We investigated GPT-3.5-turbo and GPT-4 on extracting the relations from three standard datasets, EU-ADR, Gene Associations Database (GAD), and ChemProt. Unlike the existing approaches using datasets with masked entities, we used three versions for each dataset for our experiment: a version with masked entities, a second version with the original entities (unmasked), and a third version with abbreviations replaced with the original terms. We developed the prompts for various versions and used the chat completion model from GPT API. Our approach achieved a F1-score of 0.498 to 0.809 for GPT-3.5-turbo, and a highest F1-score of 0.84 for GPT-4. For certain experiments, the performance of GPT, BioBERT, and PubMedBERT are almost the same.

A phase 2, open-label study of anti-inflammatory NE3107 in patients with dementias.

BACKGROUND: Alzheimer's disease (AD) is a progressive, multifactorial, neurodegenerative disorder affecting >6 million Americans. Chronic, low-grade neuroinflammation, and insulin resistance may drive AD pathogenesis. We explored the neurophysiological and neuropsychological effects of NE3107, an oral, anti-inflammatory, insulin-sensitizing molecule, in AD. METHODS: In this phase 2, open-label study, 23 patients with mild cognitive impairment or mild dementia received 20-mg oral NE3107 twice daily for 3 months. Primary endpoints assessed changes from baseline in neurophysiological health and oxidative stress (glutathione level) using advanced neuroimaging analyses. Secondary endpoints evaluated changes from baseline in neuropsychological health using cognitive assessments, including the 11-item Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog11), Mini-Mental State Examination (MMSE), Montreal Cognitive Assessment, Clinical Dementia Rating, Quick Dementia Rating Scale, Alzheimer's Disease Composite Score, and Global Rating of Change (GRC). Exploratory endpoints assessed changes from baseline in neuroinflammation biomarkers (tumor necrosis factor alpha, TNF-α) and AD (amyloid beta and phosphorylated tau [P-tau]). RESULTS: NE3107 was associated with clinician-rated improvements in cerebral blood flow and functional connectivity within the brain. In patients with MMSE ≥ 20 (mild cognitive impairment to mild AD; n = 17), NE3107 was associated with directional, but statistically nonsignificant, changes in brain glutathione levels, along with statistically significant improvements in ADAS-Cog11 (P = .017), Clinical Dementia Rating (P = .042), Quick Dementia Rating Scale (P = .002), Alzheimer's Disease Composite Score (P = .0094), and clinician-rated GRC (P < .001), as well as in cerebrospinal fluid P-tau levels (P = .034) and P-tau:amyloid beta 42 ratio (P = .04). Biomarker analyses also demonstrated directional, but statistically non-significant, changes in plasma TNF-α, consistent with the expected mechanism of NE3107. Importantly, we observed a statistically significant correlation (r = 0.59) between improvements in TNF-α levels and ADAS-Cog11 scores (P = .026) in patients with baseline MMSE ≥ 20. CONCLUSION: Our results indicate that in this study NE3107 was associated with what appear to be positive neurophysiological and neuropsychological findings, as well as evidence of improvement in biomarkers associated with neuroinflammation and AD in patients diagnosed with dementia. Our findings are consistent with previous preclinical and clinical observations and highlight a central role of neuroinflammation in AD pathogenesis.

Me-LLaMA: Foundation Large Language Models for Medical Applications.

Recent advancements in large language models (LLMs) such as ChatGPT and LLaMA have hinted at their potential to revolutionize medical applications, yet their application in clinical settings often reveals limitations due to a lack of specialized training on medical-specific data. In response to this challenge, this study introduces Me-LLaMA, a novel medical LLM family that includes foundation models - Me-LLaMA 13/70B, along with their chat-enhanced versions - Me-LLaMA 13/70B-chat, developed through continual pre-training and instruction tuning of LLaMA2 using large medical datasets. Our methodology leverages a comprehensive domain-specific data suite, including a large-scale, continual pre-training dataset with 129B tokens, an instruction tuning dataset with 214k samples, and a new medical evaluation benchmark (MIBE) across six critical medical tasks with 12 datasets. Our extensive evaluation using the MIBE shows that Me-LLaMA models achieve overall better performance than existing open-source medical LLMs in zero-shot, few-shot and supervised learning abilities. With task-specific instruction tuning, Me-LLaMA models outperform ChatGPT on 7 out of 8 datasets and GPT-4 on 5 out of 8 datasets. In addition, we investigated the catastrophic forgetting problem, and our results show that Me-LLaMA models outperform other open-source medical LLMs in mitigating this issue. Me-LLaMA is one of the largest open-source medical foundation LLMs that use both biomedical and clinical data. It exhibits superior performance across both general and medical tasks compared to other open-source medical LLMs, rendering it an attractive choice for medical AI applications. We release our models, datasets, and evaluation scripts at: https://github.com/BIDS-Xu-Lab/Me-LLaMA.

MRI-Based Brain Volume Scoring in Cerebral Malaria Is Externally Valid and Applicable to Lower-Resolution Images.

BACKGROUND AND PURPOSE: Children with cerebral malaria have an elevated risk of mortality and neurologic morbidity. Both mortality and morbidity are associated with initially increased brain volume on MR imaging, as graded by the Brain Volume Score, a subjective ordinal rating scale created specifically for brain MRIs in children with cerebral malaria. For the Brain Volume Score to be more widely clinically useful, we aimed to determine its independent reproducibility and whether it can be applicable to lower-resolution MRIs. MATERIALS AND METHODS: To assess the independent reproducibility of the Brain Volume Score, radiologists not associated with the initial study were trained to score MRIs from a new cohort of patients with cerebral malaria. These scores were then compared with survival and neurologic outcomes. To assess the applicability to lower-resolution MRI, we assigned Brain Volume Scores to brain MRIs degraded to simulate a very-low-field (64 mT) portable scanner and compared these with the original scores assigned to the original nondegraded MRIs. RESULTS: Brain Volume Scores on the new cohort of patients with cerebral malaria were highly associated with outcomes (OR for mortality = 16, P < .001). Scoring of the simulated degraded images remained consistent with the Brain Volume Scores assigned to the original higher-quality (0.35 T) images (intraclass coefficients > 0.86). CONCLUSIONS: Our findings demonstrate that the Brain Volume Score is externally valid in reproducibly predicting outcomes and can be reliably assigned to lower-resolution images.

How short peptides can disassemble ultra-stable tau fibrils extracted from Alzheimer's disease brain by a strain-relief mechanism.

Reducing fibrous aggregates of protein tau is a possible strategy for halting progression of Alzheimer's disease (AD). Previously we found that in vitro the D-peptide D-TLKIVWC disassembles tau fibrils from AD brains (AD-tau) into benign segments with no energy source present beyond ambient thermal agitation. This disassembly by a short peptide was unexpected, given that AD-tau is sufficiently stable to withstand disassembly in boiling SDS detergent. To consider D peptide-mediated disassembly as a potential therapeutic for AD, it is essential to understand the mechanism and energy source of the disassembly action. We find assembly of D-peptides into amyloid-like fibrils is essential for tau fibril disassembly. Cryo-EM and atomic force microscopy reveal that these D-peptide fibrils have a right-handed twist and embrace tau fibrils which have a left-handed twist. In binding to the AD-tau fibril, the oppositely twisted D-peptide fibril produces a strain, which is relieved by disassembly of both fibrils. This strain-relief mechanism appears to operate in other examples of amyloid fibril disassembly and provides a new direction for the development of first-in-class therapeutics for amyloid diseases.

How short peptides can disassemble ultra-stable tau fibrils extracted from Alzheimer's disease brain by a strain-relief mechanism.

Reducing fibrous aggregates of protein tau is a possible strategy for halting progression of Alzheimer's dis-ease (AD). Previously we found that in vitro the D-peptide D-TLKIVWC disassembles tau fibrils from AD brains (AD-tau) into benign segments with no energy source present beyond ambient thermal agitation. This disassembly by a short peptide was unexpected, given that AD-tau is sufficiently stable to withstand disas-sembly in boiling SDS detergent. To consider D peptide-mediated disassembly as a potential therapeutic for AD, it is essential to understand the mechanism and energy source of the disassembly action. We find as-sembly of D-peptides into amyloid-like fibrils is essential for tau fibril disassembly. Cryo-EM and atomic force microscopy reveal that these D-peptide fibrils have a right-handed twist and embrace tau fibrils which have a left-handed twist. In binding to the AD-tau fibril, the oppositely twisted D-peptide fibril produces a strain, which is relieved by the disassembly of both fibrils. This strain-relief mechanism appears to operate in other examples of amyloid fibril disassembly and provides a new direction for the development of first-in-class therapeutics for amyloid diseases.

Stability Oracle: a structure-based graph-transformer framework for identifying stabilizing mutations.

Engineering stabilized proteins is a fundamental challenge in the development of industrial and pharmaceutical biotechnologies. We present Stability Oracle: a structure-based graph-transformer framework that achieves SOTA performance on accurately identifying thermodynamically stabilizing mutations. Our framework introduces several innovations to overcome well-known challenges in data scarcity and bias, generalization, and computation time, such as: Thermodynamic Permutations for data augmentation, structural amino acid embeddings to model a mutation with a single structure, a protein structure-specific attention-bias mechanism that makes transformers a viable alternative to graph neural networks. We provide training/test splits that mitigate data leakage and ensure proper model evaluation. Furthermore, to examine our data engineering contributions, we fine-tune ESM2 representations (Prostata-IFML) and achieve SOTA for sequence-based models. Notably, Stability Oracle outperforms Prostata-IFML even though it was pretrained on 2000X less proteins and has 548X less parameters. Our framework establishes a path for fine-tuning structure-based transformers to virtually any phenotype, a necessary task for accelerating the development of protein-based biotechnologies.