Biology subjects

Jonsson, V. D.

Publications and source records attributed to Jonsson, V. D..

8 recordsLinked to original sources

bioRxiv2026

PRISM : Peptide-specificity annotation of T-cell receptors with uncertainty quantification

Mapping T-cell receptor (TCR) sequences to their cognate peptide-major histocompatibility complex (pMHC) ligands underlies both basic immunology and T-cell target discovery, yet current models aimed at predicting TCR specificity are limited by sparse labels, viral-biased training data, and an inability to recognize receptors outside their training distribution. We present PRISM, an uncertainty-aware metric-learning framework for TCR{beta} sequence representation. PRISM embeds receptors into a peptide-organized latent space, returns top-k peptides by nearest-neighbor retrieval, and abstains on out-of-distribution receptors by modeling an intrinsic uncertainty that tracks annotation correctness. To offset the viral bias of public databases, PRISM augments training data with structure-guided synthetic receptors that diversify TCR sequences while preserving the energetics of the TCR-pMHC interface. Across a held-out set of 923 peptides and the independent IMMREP23 benchmark, PRISM matches or exceeds sequence-based models, with largest gains on rare epitopes. Finally, PRISM learns attention weights on TCR residues that concentrate on the CDR3{beta} salt-bridge and hydrophobic contacts central to peptide recognition, linking PRISMs positional focus to the biochemical properties of TCR-pMHC structures.

immunology↗

bioRxiv2026

TRIOPS: A deep learning framework for prediction of T cell receptor-MHC binding specificity

T cell receptor (TCR) recognition is MHC-restricted, yet accurately predicting a TCRs restricting HLA allele remains an open problem. We present TRIOPS, a dual-branch convolutional model with soft cross-attention that predicts TCR-MHC restriction from amino acid sequence alone. TRIOPS uses cross-reactivity-aware negative sampling by HLA pseudosequence similarity to reduce allele-boundary label noise, extending prediction to alleles absent from training. TRIOPS reaches a held-out AUC of 0.97 for paired TCR{beta} and 0.92 for TCR{beta}-only inputs, generalizes to unseen receptors and HLA alleles, and after locus-specific calibration, assigns TCR clonotypes to their likeliest restricting allele across an individuals HLA genotype. In TCGA tumors, TCR repertoires preferentially engage the expression-lost allele at HLA-A and HLA-B and the retained allele at HLA-C, recapitulating from bulk tumor RNA-seq the allele specific HLA loss previously linked to immune escape.

immunology↗

bioRxiv2024

A rare HSC-derived megakaryocyte progenitor accumulates via enhanced survival and contributes to exacerbated thrombopoiesis upon aging

Distinct routes of cellular production from hematopoietic stem cells (HSCs) have defined our current view of hematopoiesis. Recently, we challenged classical views of platelet generation, demonstrating that megakaryocyte progenitors (MkPs), and ultimately platelets, can be specified via an alternate and additive route of HSC-direct specification specifically during aging. This "shortcut" pathway generates hyperactive platelets likely to contribute to age-related platelet-mediated morbidities. Here, we used single-cell RNA/CITEseq to demonstrate that these age-unique, non-canonical (nc)MkPs can be prospectively defined and experimentally isolated from wild type mice. Surprisingly, this revealed that a rare population of ncMkPs also exist in young mice. Young and aged ncMkPs are functionally distinct from their canonical (c)MkP counterparts, with aged ncMkPs paradoxically and uniquely exhibiting enhanced survival and platelet generation capacity. We further demonstrate that aged HSCs generate significantly more ncMkPs than their younger counterparts, yet this is accomplished without strict clonal restriction. Together, these findings reveal significant phenotypic, functional, and aging-dependent heterogeneity among the MkP pool and uncover unique features of megakaryopoiesis throughout life, potentially offering cellular and molecular targets for mitigation of age-related adverse thrombotic events.

cell biology↗

bioRxiv2024

Pregnancy Reduces Il33+ Hybrid Progenitor Accumulation in the Aged Mammary Gland

Aging increases breast cancer risk while an early first pregnancy reduces a womans life-long risk. Several studies have explored the effect of either aging or pregnancy on mammary epithelial cells (MECs), but the combined effect of both remains unclear. Here, we interrogate the functional and transcriptomic changes at single cell resolution in the mammary gland of aged nulliparous and parous mice to discover that pregnancy normalizes age-related imbalances in lineage composition, while also inducing a differentiated cell state. Importantly, we uncover a minority population of Il33-expressing hybrid MECs with high cellular potency that accumulate in aged nulliparous mice but is significantly reduced in aged parous mice. Functionally, IL33 treatment of basal, but not luminal, epithelial cells from young mice phenocopies aged nulliparous MECs and promotes formation of organoids with Trp53 knockdown. Collectively, our study demonstrates that pregnancy blocks the age-associated loss of lineage integrity in the basal layer through a decrease in Il33+ hybrid MECs, potentially contributing to pregnancy-induced breast cancer protection.

developmental biology↗

bioRxiv2023

A data-efficient deep learning tool for scRNA-Seq label transfer in neuroscience

Large single-cell RNA datasets have contributed to unprecedented biological insight. Often, these take the form of cell atlases and serve as a reference for automating cell labeling of newly sequenced samples. Yet, classification algorithms have lacked the capacity to accurately annotate cells, particularly in complex datasets. Here we present SIMS (Scalable, Interpretable Ma-chine Learning for Single-Cell), an end-to-end data-efficient machine learning pipeline for discrete classification of single-cell data that can be applied to new datasets with minimal coding. We benchmarked SIMS against common single-cell label transfer tools and demonstrated that it performs as well or better than state of the art algorithms. We then use SIMS to classify cells in one of the most complex tissues: the brain. We show that SIMS classifies cells of the adult cerebral cortex and hippocampus at a remarkably high accuracy. This accuracy is maintained in trans-sample label transfers of the adult hu-man cerebral cortex. We then apply SIMS to classify cells in the developing brain and demonstrate a high level of accuracy at predicting neuronal sub-types, even in periods of fate refinement, shedding light on genetic changes affecting specific cell types across development. Finally, we apply SIMS to single cell datasets of cortical organoids to predict cell identities and unveil genetic variations between cell lines. SIMS identifies cell-line differences and misannotated cell lineages in human cortical organoids derived from different pluripotent stem cell lines. When cell types are obscured by stress signals, label transfer from primary tissue improves the accuracy of cortical organoid annotations, serving as a reliable ground truth. Altogether, we show that SIMS is a versatile and robust tool for cell-type classification from single-cell datasets.

bioinformatics↗

bioRxiv2022

PD-L1 blockade restores CAR T cell activity through IFNγ-regulation of CD163+ macrophages

BackgroundThe immune suppressive tumor microenvironment (TME) that inhibits T cell infiltration, survival, and anti-tumor activity has posed a major challenge for developing effective immunotherapies for solid tumors. Chimeric antigen receptor (CAR)-engineered T cell therapy has shown unprecedented clinical response in treating patients with hematological malignancies, and intense investigation is underway to achieve similar responses with solid tumors. Immunologically cold tumors, including prostate cancers, are often infiltrated with abundant tumor-associated macrophages (TAMs), and infiltration of CD163+ M2 macrophages correlates with tumor progression and poor responses to immunotherapy. However, the impact of TAMs on CAR T cell activity alone and in combination with TME immunomodulators is unclear. MethodsTo model this in vitro, we utilized a novel co-culture system with tumor cells, CAR T cells, and polarized M1 or M2 macrophages from CD14+ PBMCs collected from healthy human donors. Tumor cell killing, T cell activation and proliferation, and macrophage phenotypes were evaluated by flow cytometry, cytokine production, RNA sequencing, and functional blockade of signaling pathways using antibodies and small molecule inhibitors. We also evaluated the TME in humanized mice following CAR T cell therapy for validation of our in vitro findings. ResultsWe observed inhibition of CAR T cell activity with the presence of M2 macrophages, but not M1 macrophages, coinciding with a robust induction of PD-L1 in M2 macrophages. We observed similar PD-L1 expression in TAMs following CAR T cell therapy in the TME of humanized mice. PD-L1, but not PD-1, blockade in combination with CAR T cell therapy altered phenotypes to more M1-like subsets and led to loss of CD163+ M2 macrophages via IFN{gamma} signaling, resulting in improved anti-tumor activity of CAR T cells. ConclusionThis study reveals an alternative mechanism by which the combination of CAR T cells and immune checkpoint blockade modulates the immune landscape of solid tumors to enhance therapeutic efficacy of CAR T cells.

immunology↗

bioRxiv2021

CAR T cell therapy drives endogenous locoregional T cell dynamics in a responding patient with glioblastoma

CAR T cell therapy has transformed clinical care and management of patients with certain hematological cancers. However, it remains unclear whether the success of CAR T cell therapy relies solely on CAR T cell engagement with tumor antigen, or if it also requires the stimulation of an individual patients endogenous T cell response. Here, we performed combined analysis of longitudinal, single cell RNA and T cell receptor sequencing on glioblastoma tumors, peripheral blood (PB), and cerebrospinal fluid (CSF) from a patient with recurrent multifocal glioblastoma that underwent a remarkable response followed by recurrence on IL13RA2-targeted CAR T cell therapy (Brown et al. 2016). Single cell analysis of a tumor resected prior to CAR T cell therapy revealed the existence of an inflamed tumor microenvironment including a CD8+ cytotoxic, clonally expanded and antigen specific T cell population that disappeared in the recurrent setting. Longitudinal tracking of T cell receptors uncovered distinct T cell dynamics classes in the CSF during CAR T cell therapy. These included T cell clones with transient dynamics, representing intraventricular CAR T cell delivery and endogenous T cell recruitment from the PB into the CSF; and a group of T cells in the cerebrospinal fluid, that tracked with clonally expanded tumor resident T cells and whose dynamics contracted concomitantly with tumor volume. Our results suggest the existence of an endogenous T cell population that was invigorated by intraventricular CAR T cell infusions, and combined with the therapy to produce a complete response.

immunology↗

bioRxiv2016

Novel computational method for predicting polytherapy switching strategies to overcome tumor heterogeneity and evolution

1The success of targeted cancer therapy is limited by drug resistance that can result from tumor genetic heterogeneity. The current approach to address resistance typically involves initiating a new treatment after clinical/radiographic disease progression, ultimately resulting in futility in most patients. Towards a potential alternative solution, we developed a novel computational framework that uses human cancer profiling data to systematically identify dynamic, pre-emptive, and sometimes non-intuitive treatment strategies that can better control tumors in real-time. By studying lung adenocarcinoma clinical specimens and preclinical models, our computational analyses revealed that the best anti-cancer strategies addressed existing resistant subpopulations as they emerged dynamically during treatment. In some cases, the best computed treatment strategy used unconventional therapy switching while the bulk tumor was responding, a prediction we confirmed in vitro. The new framework presented here could guide the principled implementation of dynamic molecular monitoring and treatment strategies to improve cancer control.

cancer biology↗