Biology subjects

Cuellar, C.

Publications and source records attributed to Cuellar, C..

6 recordsLinked to original sources

bioRxiv2025

Protein Language Models Capture Structural and Functional Epistasis in a Zero-Shot Setting

Protein language models (PLMs) learn from large collections of natural sequences and achieve striking success across prediction tasks, yet it remains unclear what biological principles underlie their representations. We use epistasis, the dependence of a mutations effect on its sequence context, as a lens to probe what PLMs capture about proteins. Comparing PLM-derived scores with deep mutational scanning data, we find that epistasis emerges naturally from pretrained models, without supervision on experimental fitness. Raw model scores align with residue-residue contacts, indicating that PLMs internalize structural proximity. Applying a nonlinear transformation to bring model outputs onto the experimental scale, however, shifts the signal toward functional couplings between distant sites. These findings show that PLMs capture both structural and functional dependencies from sequence data alone, and that epistasis provides a powerful window into the biological principles embedded in their representations.

bioinformatics↗

bioRxiv2024

Predicting purification process fit of monoclonal antibodies using machine learning

0In early stage development of therapeutic monoclonal antibodies, assessment of ease of purification process development typically requires extensive experimentation. However, upstream protein expression and downstream purification experiments are often in conflict with timeline pressures and material constraints, limiting the number of molecules and process conditions that can reasonably be assessed. Recently, high-throughput batch-binding screen data along with improved molecular descriptors have enabled development of robust quantitative structure-property relationship (QSPR) models which predict monoclonal antibody chromatographic binding behavior from the amino acid sequence. This work describes a QSPR strategy for in silico monoclonal antibody purification process fit assessment. Principal Component Analysis is applied to extract a one-dimensional basis for comparison of molecular chromatographic binding behavior from multi-dimensional high-throughput batch-binding screen data. Kernel Ridge Regression is employed to predict the first principal component for new molecular sequences. This workflow is demonstrated with a set of 97 monoclonal antibodies for five chromatography resins in two salt types across a range of pH and salt concentrations. Model development benchmarks four descriptor sets from biophysical structural models and protein language models. The investigation illustrates the value QSPR models can provide to purification process fit assessment, and selection of resins and operating conditions from sequence alone.

bioinformatics↗

bioRxiv2022

Canada goose fecal microbiota correlate with geography more than host-associated co-factors

The gut microbiota has many positive effects on the host, but how the microbiota is shaped and influenced can vary greatly. These factors affect the composition, diversity, and function of host-associated microbiota; however, these factors vary greatly from organism to organism and clade to clade. The avian microbiota often correlates more with the sampling locations rather than host-associated co-factors. These correlations between location and microbiota often only include a few sampling locations within the species range. To better understand the connection between geographic distance and the microbiota, were collected from non-migratory Canada geese across the United States. We expected host-associated factors to have minimal effect on the microbiota and geese microbiota will be strongly correlated to geography. We hypothesized more proximal geese will be exposed to more similar environmental microbes and will have more similar microbiota. Canada geese microbiota are largely similar across the entire sampling range. Several bacterial taxa were shared by more than half of the geese. Four phyla were found in the majority of the samples: Firmicutes, Proteobacteria, Bacteroidetes, and Actinobacteria. Three genera were also present in the majority of the samples: Helicobacter, Subdoligranulum, and Faecalibacterium. There were minimal differences in alpha diversity with respect to age, sex, and flyway. There were significant correlations between geography and beta diversity. Supervised machine learning models were able to predict the location of a fecal sample based on taxonomic data alone. Distance decay analysis show a positive relationship between geographic distance and beta diversity. Our work provides novel insights into the microbiota of the ubiquitous Canada goose and further supports the claim that the avian microbiota is largely dominated by the hosts environment. This work also suggests that there is a minimum distance that must be reached before significant differences in the microbiota between two individuals can be observed.

microbiology↗

bioRxiv2019

Changes in spinal cord hemodynamic reflect modulation of spinal network with different parameters of epidural stimulation

In this study functional ultrasound (fUS) imaging has been implemented to explore the local hemodynamic response induced by electrical epidural stimulation and to study real-time in vivo functional changes of the spinal cord, taking advantage of the superior spatiotemporal resolution provided by fUS. By quantifying the hemodynamic and electromyographic response features, we tested the hypothesis that the transient hemodynamic response of the spinal cord to electrical epidural stimulation could reflect modulation of the spinal circuitry and accordingly respond to the changes in parameters of electrical stimulation. The results of this study for the first time demonstrate that the hemodynamic response to electrical stimulation could reflect functional organization of the spinal cord. Response in the dorsal areas to epidural stimulation was significantly higher and faster compared to the response in ventral spinal cord. Positive relation between the hemodynamic and the EMG responses was observed at the lower frequencies of epidural stimulation (20 and 40 Hz), which according to our previous findings can facilitate spinal circuitry after spinal cord injury, compared to higher frequencies (200 and 500 Hz). These findings suggest that different mechanisms could be involved in spinal cord hemodynamic changes during different parameters of electrical stimulation and for the first time provide the evidence that functional organization of the spinal cord circuitry could be related to specific organization of spinal cord vasculature and hemodynamic.\n\nSignificance StatementElectrical epidural stimulation (EES) has been successfully applied to control chronic refractory pain and was evolved to alleviate motor impairment after spinal cord injury, in Parkinsons disease, and other neurological conditions. The mechanisms underlying the EES remain unclear, and current methods for monitoring EES are limited in sensitivity and spatiotemporal resolutions to evaluate functional changes in response to EES. We tested the hypothesis that the transient hemodynamic response of the spinal cord to EES could reflect modulation of the spinal cord circuitry and accordingly respond to the changes in parameters of EES. The proposed methodology opens a new direction for quantitative evaluation of the spinal cord hemodynamic in understanding the mechanisms of spinal cord functional organization and effect of neuromodulation.

neuroscience↗

bioRxiv2019

A VP2/3-derived peptide exhibits potent antiviral activity against BK and JC polyomaviruses by targeting a novel VP1 binding site

In pursuit of effective therapeutics for human polyomaviruses, we identified a peptide derived from the BK polyomavirus (BKV) minor structural proteins VP2/3 that is a potent inhibitor of BKV infection with no observable cellular toxicity. The thirteen amino acid peptide binds to major structural protein VP1 in a new location within the pore with a low nanomolar KD. Alanine scanning of the peptide identified three key residues, substitution of each of which results in [~]1000-fold loss of affinity with a concomitant reduction in antiviral activity. NMR spectroscopy and an X-ray structurally-guided model demonstrate specific binding of the peptide to the pore of the VP1 pentamer that constitutes the BKV capsid. Cell-based assays with the peptide demonstrate nanomolar inhibition of BKV infection and suggest that the peptide likely blocks the viral entry pathway between endocytosis and escape from the host cell ER. The peptide motif is highly conserved among the polyomavirus clade, and homologous peptides exhibit similar binding properties for JC polyomavirus and inhibit infection with similar potency to BKV in a model cell line. Substitutions within VP1 or VP2/3 residues involved in VP1-peptide interaction negatively impact viral infectivity, potentially indicating the peptide-binding site within the VP1 pore is relevant for VP1-VP2/3 interactions. The inhibitory potential of the peptide-binding site first reported here may present a novel target for development of new anti-polyomavirus therapies. In summary, we present the first anti-polyomavirus inhibitor that acts via a novel mechanism of action by specifically targeting the pore of VP1.

microbiology↗

bioRxiv2019

Multifactorial motor behavior assessment for real-time evaluation of emerging therapeutics to treat neurologic impairments

Integrating multiple assessment parameters of motor behavior is critical for understanding neural activity dynamics during motor control in both intact and dysfunctional nervous systems. Here, we described a novel approach (termed Multifactorial Behavioral Assessment (MfBA)) to integrate, in real-time, electrophysiological and biomechanical properties of rodent spinal sensorimotor network activity with behavioral aspects of motor task performance. Specifically, the MfBA simultaneously records limb kinematics, multi-directional forces and electrophysiological metrics, such as high-fidelity chronic intramuscular electromyography synchronized in time to spinal stimulation in order to characterize spinal cord functional motor evoked potentials (fMEPs). Additionally, we designed the MfBA to incorporate a body weight support system to allow bipedal and quadrupedal stepping on a treadmill and in an open field environment to assess function in rodent models of neurologic disorders that impact motor activity This novel approach was validated using, a neurologically intact cohort, a cohort with unilateral Parkinsonian motor deficits due to midbrain lesioning, and a cohort with complete hind limb paralysis due to T8 spinal cord transection. In the SCI cohort, lumbosacral epidural electrical stimulation (EES) was applied, with and without administration of the serotonergic agonist Quipazine, to enable hind limb motor functions following paralysis. The results presented herein demonstrate the MfBA is capable of integrating multiple metrics of motor activity in order to characterize relationships between EES inputs that modulate mono- and polysynaptic outputs from spinal circuitry which in turn, can be used to elucidate underlying electrophysiologic mechanisms of motor behavior by synchronizing these datasets to metrics of movement and behavior. These results also demonstrate that proposed MfBA is an effective tool to integrate biomechanical and electrophysiology metrics, synchronized to therapeutic inputs such as EES or pharmacology, during body weight supported treadmill or open field motor activities, to target a high range of variations in motor behavior as a result of neurological deficit at the different levels of CNS.

neuroscience↗