Biology subjects

Kelkar, A.

Publications and source records attributed to Kelkar, A..

12 recordsLinked to original sources

bioRxiv2026

Predicting Post-Stroke Aphasia Speech Performance from Multimodal Data with Explainable Machine Learning

Aphasia, an acquired language deficit, is the most common post-stroke focal cognitive impairment, and roughly 60% cases become chronic (duration >6 months). Aphasia therapies could be optimized if clinicians could make personalized predictions of how individual persons with aphasia (PWA) would be likely to perform on particular language tasks. However, current approaches relying on imaging, lesion volume, patient demographics, and clinical scores achieve less than 50% accuracy in predicting performance in PWA. Research algorithms using complex imaging and fMRI can make binary predictions about the presence or absence of aphasia but do not give more clinically relevant information. We aim to predict word-by-word speech accuracy in PWA to better enable personalized speech therapies. To be clinically informative, machine learning models developed for this purpose should use clinically available inputs, explain key features behind a prediction, and generalize to new PWA and previously unseen words. This study combines multimodal input features from clinical testing scores and structural MRI neuroimaging with a novel data source: word-by-word linguistic difficulty. We computed metrics of cognitive burden, such as semantic selection and recall demands, and articulatory burden, such as word length in phonemes and syllables, using naturalistic corpora containing over a billion words of English text. Retrospective training, ten-fold cross validation and 500-run bootstrapping of different machine learning models with various combinations of input features was conducted using 4620 trials. A simplified version of the best model using widely available inputs was deployed clinically through a web app, and prospective generalization was tested on 570 trials with unseen words and different naming tasks in new PWA. We found the best performances with random forest classifiers using linguistic difficulty combined with either clinical information (AUROC {+/-} SEM = 0.87 {+/-} 0.07), or all together with structural imaging connectivity (0.90 {+/-} 0.04). Classifiers using multimodal inputs significantly outperformed others employing single inputs (range 0.66-0.85, p<0.05). Extracting feature importances from the best model showed that Western Aphasia Battery scores, semantic demands, number of phonemes, and syllables were predictive of PWA speech accuracy. Structural integrity in peri-lesional brain regions predicted better language performance whereas higher connectivity of select contralateral homotopes contributed to prediction of worse speech. Without the inclusion of MRI data, lesion volume was a key predictor of PWA speech as well. A simplified, clinically ready, explainable model (publicly available as AphasiaLENS web application) predicted PWA accuracy for any user-entered word, not restricted to a standardized battery. Its prospective generalization performance was not significantly different from the best model using full inputs (AUROC ranges 0.81-0.89, p>0.05). Thus, our research can help inform individualized treatment planning for PWA, while also suggesting research targets through better understanding of brain-behavior relationships.

bioengineering↗

bioRxiv2025

Impairment of neuronal activity occurs at the early stages of the aggregation cascade of Ab1-42 and mutant Tau

Alzheimers disease (AD) is a progressive neurodegenerative disease that is characterized by the accumulation of amyloid-{beta} (A{beta}) plaques and neurofibrillary Tau tangles, ultimately leading to brain atrophy and death. To elucidate the relationship between the aberrant folding and aggregation of A{beta} and mutant Tau and neuronal function, we monitored neuronal activity in C. elegans AD models across age. Our findings reveal that expression of both A{beta} and Tau lead to significant reductions in neuronal activity and function in young adult animals preceding the accumulation of amyloid aggregates. Notably, A{beta} expression and aggregation in muscle tissue produced comparable detrimental effects on neuronal activity as its expression in neurons, suggesting that proteotoxic stress in muscle can influence neuronal function. This may occur through the propagation of A{beta} from muscle to neurons or through retrograde signaling pathways. Further, our new sub-stoichiometrically labeled Tau strains highlight that TauP301L,V337M has a significant impact on neuronal activity throughout aging. These results enhance our understanding of the early functional effects of amyloid aggregation in Alzheimers disease.

neuroscience↗

bioRxiv2024

Pelvic Ring Fractures: Does Spinopelvic Alignment Affect Sacral and Lumbopelvic Fixation Stability?

Background ContextPelvic ring is a complex boney and ligamentous complex and its structural integrity is paramount to be able to withstand the high forces. Pelvic ring fractures, which can result from low-energy impacts or high-velocity injuries and falls from great heights, can severely impair the structural integrity of the pelvic ring. These injuries can be fatal, especially in geriatric patients, with mortality rates ranging between 10% and 16%. Pelvic ring injuries require immediate stabilization, typically performed through internal fixation, to maintain circulatory balance and achieve anatomical restoration of the pelvis. MethodsA previously validated FE model of the spine, pelvis, and femur model was used to evaluate how different fixation techniques for pelvic ring injury affect spinopelvic biomechanics. The SS, PI, PT and LL of the normal, intact model was modified to generate two more SS models for this study. One modified model with an SS of 20{o} was considered the low SS model (SSD), while the second modified model with an SS of 32{o} was termed the high SS model (SSI). A unilateral pelvic ring fracture was simulated by resecting the left side of the sacrum and pelvis. The fractures were stabilized using two different posterior stabilization techniques 1) The fractures were stabilized using two different posterior stabilization techniques, and 2) L5-Ilium posterior screw fixation without cross connector (L5_PF_WO_CC). ResultsSSI demonstrated the higher range of motion, followed by Normal model and least by SSD model at right Sacroiliac Joint. No differences were seen between with and without cross connector model for range of motion at right SIJ. For L5-S1 motion models treated with cross connectors demonstrated the least motion compared to those treated without cross connectors for all configurations and loading conditions. For the sacrum fracture, models with high sacral slopes recorded the least horizontal displacement, followed by decreased sacral slope models for both configurations and loading conditions.

bioengineering↗

bioRxiv2024

Conserved role of spike S2 domain N-glycosylation across beta-coronavirus family

Besides acting as an immunological shield, the N-glycans of SARS-CoV-2 are also critical for viral life cycle. As the S2 subunit of spike is highly conserved across beta-coronaviruses, we determined the functional significance of the five stem N-glycans located in S2 between N1098-N1194. Studies were performed with 31 Asn-to-Gln mutants, beta-coronavirus virus-like particles and single-cycle viral replicons. Deletions of stem N-glycans enhanced S1 shedding from trimeric spike, reduced ACE2 binding and abolished syncytia formation. When three or more N-glycans were deleted, spike expression on cell surface and incorporation into virions was both reduced. Viral entry function was progressively lost upon deleting the N1098 glycan in combination with additional glycosite modifications. In addition to SARS-CoV-2, deleting stem N-glycans in SARS-CoV and MERS-CoV spike also prevented viral entry into target cells. These data suggest multiple functional roles for the stem N-glycans, and evolutionarily conserved properties for these complex carbohydrates across human beta-coronaviruses. Author SummaryPrevious work shows that the N-linked glycans of SARS-CoV-2 are essential for viral life cycle. Few natural mutations have been observed in the S2-subunit of the viral spike glycoprotein in GISAID data, and mutations are absent in the five stem N-glycans located between N1098-N1194. In the post-fusion spike structure these glycans lie equidistant, ~4 nm apart, suggesting functional significance. Upon testing the hypothesis that these glycans are critical for SARS-CoV-2 function, we noted multiple roles for the complex carbohydrates including regulation of S1-subunit shedding, spike expression on cells and virions, syncytial formation/cell-cell fusion and viral entry. Besides SARS-CoV-2, these glycans were also critical for other human beta-coronaviruses. Thus, these carbohydrates represent targets for the development of countermeasures against future outbreaks.

microbiology↗

bioRxiv2023

Structure prediction of protein-ligand complexes from sequence information with Umol

Protein-ligand docking is an established tool in drug discovery and development to narrow down potential therapeutics for experimental testing. However, a high-quality protein structure is required and often the protein is treated as fully or partially rigid. Here we develop an AI system that can predict the fully flexible all-atom structure of protein-ligand complexes directly, given a multiple sequence alignment representation of the protein and a SMILES string representing the ligand. At a high accuracy threshold, unseen protein-ligand complexes can be predicted more accurately than for RoseTTAFold-AA, and at medium accuracy even classical docking methods that use known protein structures as input are surpassed. The high accuracy presented here suggests that the goal of AI-based drug discovery is one step closer, but there is still a way to go to fully grasp the complexity of protein-ligand interactions. Umol is available at: https://github.com/patrickbryant1/Umol

bioinformatics↗

bioRxiv2022

Individual-level Functional ConnectivityPredicts Cognitive Control Efficiency

Cognitive control (CC) is a vital component of cognition associated with problem-solving in everyday life. Many neurological and neuropsychiatric conditions have deficits associated with CC. CC is composed of multiple behaviors including switching, inhibiting, and updating. The fronto-parietal control network B (FPCN-B), the dorsal attention network (DAN), the cingulo-opercular network (CON) and the dorsal default-mode network (dorsal-DMN) have been associated with switching and inhibiting behaviors. However, our understanding of how these brain regions interact to bring about CC behaviors is still unclear. In the current study, participants performed two in-scanner tasks that required switching and inhibiting. We then used a series of support vector regression (SVR) models containing individually-estimated functional connectivity between the networks of interest derived during tasks and at rest to predict inhibition and switching behaviors in individual subjects. We observed that the combination of between-network connectivity from these individually estimated functional networks predicted accurate and timely inhibition and switching behaviors in individuals. We also observed that the relationships between canonical task-positive and task-negative networks predicted inhibiting and switching behaviors. Finally, we observed a functional dissociation between the FPCN-A and FPCNB during rest, and task performance predicted inhibiting and switching behaviors. These results suggest that individually estimated networks can predict individual CC behaviors, that between-network functional connectivity estimated within individuals is vital to understanding how CC arises, and that the fractionation of the FPCN and the DMN may be associated with different behaviors than their canonically accepted behaviors.

neuroscience↗

bioRxiv2022

Autophagy-cell death balance is maintained by Polycomb-mediated regulation during stem cell differentiation

Autophagy is a conserved cytoprotective process, aberrations in which, lead to numerous degenerative disorders. While the cytoplasmic components of autophagy have been extensively studied, the epigenetic regulation of autophagy genes, especially in stem cells, is less understood. Deciphering the epigenetic regulation of autophagy genes becomes increasingly relevant given the therapeutic benefits of small-molecule epigenetic inhibitors in novel treatment modalities. We observe that, during retinoic acid-mediated differentiation of mouse embryonic stem cells (mESCs), autophagy is induced, and identify the Polycomb enzyme EZH2 as a regulator of this process. In mESCs, EZH2 represses several autophagy genes including the autophagy regulator Dram1. EZH2 facilitates the formation of a bivalent chromatin domain at the Dram1 promoter, which allows the expression of the gene and induction of autophagy during differentiation, while still retaining the repressive H3K27me3 mark. EZH2 inhibition leads to loss of the bivalent domain, and a consequential "hyper- expression" of Dram1, with extensive cell death. This study shows that Polycomb group proteins help maintain an autophagy-cell death balance during stem cell differentiation, in part, by regulating the expression of the Dram1 gene.

cell biology↗

bioRxiv2021

Simulated attack reveals how lesions affect network properties in post-stroke aphasia

Aphasia is one of the most prevalent cognitive syndromes caused by stroke. The rarity of premorbid imaging and heterogeneity of lesion size and extent obfuscates the links between the local effects of the lesion, global anatomical network organization, and aphasia symptoms. We applied a simulated attack approach to examine the effects of 39 stroke lesions on network topology by simulating their effects in a control sample of 36 healthy brain networks. We focused on measures of global network organization thought to support overall brain function and resilience in the whole brain and within the left hemisphere. After removing lesion volume from the network topology measures and behavioral scores (the Western Aphasia Battery Aphasia Quotient; WAB-AQ), four behavioral factor scores obtained from a neuropsychological battery, and a factor sum), we compared the behavioral variance accounted for by simulated post-stroke connectomes to that observed in the randomly permuted data. Overall, global measures of network topology in the whole brain and left hemisphere accounted for 10% variance or more of the WAB-AQ and the lexical factor score beyond lesion volume and null permutations. Streamline networks provided more reliable point estimates than FA networks. Edge weights and network efficiency were weighted most highly in predicting the WAB-AQ for FA networks. Overall, our results suggest that global network measures can provide modest statistical value predicting overall aphasia severity, but less value in predicting specific behaviors. Variability in estimates could be induced by premorbid ability, deafferentation and diaschisis, and neuroplasticity following stroke.

neuroscience↗

bioRxiv2021

Structural Disconnection of the Posterior Medial Frontal Cortex Reduces Speech Error Monitoring

Optimal performance in any task relies on the ability to detect and repair errors. The anterior cingulate cortex and the broader posterior medial frontal cortex (pMFC) are active during error processing. However, it is unclear whether damage to the pMFC impairs error monitoring. We hypothesized that successful error monitoring critically relies on connections between the pMFC and broader cortical networks involved in executive functions and the task being monitored. We tested this hypothesis in the context of speech error monitoring in people with post-stroke aphasia. Diffusion weighted images were collected in 51 adults with chronic left-hemisphere stroke and 37 age-matched control participants. Whole-brain connectomes were derived using constrained spherical deconvolution and anatomically-constrained probabilistic tractography. Support vector regressions identified white matter connections in which lost integrity in stroke survivors related to reduced error detection during confrontation naming. Lesioned connections to the bilateral pMFC were related to reduce error monitoring, including many connections to regions associated with speech production and executive function. We conclude that connections to the pMFC support error monitoring. Error monitoring in speech production is supported by the structural connectivity between the pMFC and regions involved in speech production and executive function. Interactions between pMFC and other taskrelevant processors may similarly be critical for error monitoring in other task contexts.

neuroscience↗

bioRxiv2021

Glass Half Full: Preserved Anatomical Bypasses Predict Variance in Language Functions After Stroke

The severity of post-stroke aphasia is related to damage to white matter connections. However, neural signaling can route not only through direct connections, but also along multi-step network paths. When brain networks are damaged by stroke, paths can bypass around the damage to restore communication. The shortest network paths between regions could be the most efficient routes for mediating bypasses. We examined how shortest-path bypasses after left hemisphere strokes were related to language performance. Regions within and outside of the canonical language network could be important in aphasia recovery. Therefore, we innovated methods to measure the influence of bypasses in the whole brain. Distinguishing bypasses from all residual shortest paths is difficult without pre-stroke imaging. We identified bypasses by finding shortest paths in subjects with stroke that were longer than those observed in the average network of the most reliably observed connections in age-matched controls. We tested whether features of those bypasses predicted scores in four orthogonal dimensions of language performance derived from a factor analysis of a battery of language tasks. The features were the length of each bypass in steps, and how many bypasses overlapped on each individual direct connection. We related these bypass features to language factors using grid-search cross-validated Support Vector Regression, a technique that extracts robust relationships in high-dimensional data analysis. We discovered that the length of bypasses reliably predicted variance in lexical production (R2 = .576) and auditory comprehension scores (R2 = .164). Bypass overlaps reliably predicted variance in Lexical Production scores (R2 = .247). The predictive elongation features revealed that bypass efficiency along the dorsal stream and ventral stream were most related to Lexical Production and Auditory Comprehension, respectively. Among the predictive bypass overlaps, increased bypass routing through the right hemisphere putamen was negatively related to lexical production ability.

neuroscience↗

bioRxiv2020

Inhibition of SARS-CoV-2 viral entry in vitro upon blocking N- and O-glycan elaboration

The Spike protein of SARS-CoV-2, its receptor binding domain (RBD), and its primary receptor ACE2 are extensively glycosylated. The impact of this post-translational modification on viral entry is yet unestablished. We expressed different glycoforms of the Spike-protein and ACE2 in CRISPR-Cas9 glycoengineered cells, and developed corresponding SARS-CoV-2 pseudovirus. We observed that N- and O-glycans had only minor contribution to Spike-ACE2 binding. However, these carbohydrates played a major role in regulating viral entry. Blocking N-glycan biosynthesis at the oligomannose stage using both genetic approaches and the small molecule kifunensine dramatically reduced viral entry into ACE2 expressing HEK293T cells. Blocking O-glycan elaboration also partially blocked viral entry. Mechanistic studies suggest multiple roles for glycans during viral entry. Among them, inhibition of N-glycan biosynthesis enhanced Spike-protein proteolysis. This could reduce RBD presentation on virus, lowering binding to host ACE2 and decreasing viral entry. Overall, chemical inhibitors of glycosylation may be evaluated for COVID-19.

biochemistry↗

bioRxiv2019

A novel cis regulatory element regulates human XIST in CTCF-dependent manner

The long non-coding RNA XIST is the master regulator for the process of X chromosome inactivation (XCI) in mammalian females. Here we report the existence of a hitherto uncharacterized cis regulatory element (cRE) within the first exon of human XIST, which determines the transcriptional status of XIST during the initiation and maintenance phases of XCI. In the initiation phase, pluripotency factors bind to this cRE and keep XIST repressed. In the maintenance phase of XCI, the cRE is enriched for CTCF which activates XIST transcription. By employing a CRISPR-dCas9-KRAB based interference strategy, we demonstrate that binding of CTCF to the newly identified cRE is critical for regulating XIST in a YY1-dependent manner. Collectively, our study uncovers the combinatorial effect of multiple transcriptional regulators influencing XIST expression during the initiation and maintenance phases of XCI.

molecular biology↗