Faculty Bibliography

Breast and prostate cancers are both hormone-driven adenocarcinomas that undergo analogous invasion programs. Using lightsheet microscopy on intact tumors, we identified transitional junctions between precancerous and invasive regions. We then developed a multimodal serial-section workflow integrating volumetric reconstruction with spatial transcriptomics. Analysis of 319 spatial assays from 51 cases revealed gene-expression features and novel structural insights defining the shift from precancer to invasive disease. In breast cancer, loss of MGP and PLAT was associated with invasive transition and promoted tumorigenesis in functional assays. In prostate cancer, GDF15, ALDH1A3, ANPEP, and FASN were upregulated along invasive progression, and their knockdown in PC-3 cells suppressed proliferation and migration. Enrichment of tumor-associated macrophages (SPP1⁺, MS4A6A⁺) along non-TNBC breast cancer transitions highlights immune involvement as a potential driver of invasiveness.

Glaucoma is a chronic neurodegenerative disease of the visual system, and treatment is targeted toward lowering intraocular pressure. However, some patients fail to respond to treatment and their intraocular pressure levels remain high, risking continuous vision loss. Explainable machine learning provides a mechanism for both individual prognostication and the identification of factors associated with treatment outcome. Here, we used explainable machine learning to predict intraocular pressure for glaucoma patients receiving medication treatment. We accessed the UK Biobank to obtain information on 290 eyes from 161 participants who reported a diagnosis of glaucoma and were receiving treatment. Features were divided into three distinct datasets containing demographic data only, physiometabolic parameters and medication prescription data, and all data combined. We evaluated five machine learning techniques for each feature set in terms of their ability to predict intraocular pressure at a follow-up visit in a classification task. We then calculated SHapley Additive exPlanation (SHAP) values for the best performing model to determine feature importance, stability, and interactions. We found that eXtreme Gradient Boosting (XGBoost) outperformed all other models when trained and tested on the combined feature set with an area under receiver operating characteristic curve (AUC) of 0.708. Insulin-like growth factor 1 (IGF-1), low-density lipoprotein (LDL), and lymphocyte count ranked as the three most important features for this model. LDL and IGF-1 exhibited a low degree of global variability in contribution to the model output across all cross-validation repeats. SHAP values demonstrated the strongest interactions being between LDL and IGF-1. In summary, our studies indicated the importance of blood LDL and IGF-1 in contributing to the outcomes of intraocular pressure lowering treatment and demonstrated the ability of XGBoost to predict these outcomes.

Glaucoma is a chronic neurodegenerative disease of the visual system, and treatment is targeted toward lowering intraocular pressure. However, some patients fail to respond to treatment and their intraocular pressure levels remain high, risking continuous vision loss. Explainable machine learning provides a mechanism for both individual prognostication and the identification of factors associated with treatment outcome. Here, we used explainable machine learning to predict intraocular pressure for glaucoma patients receiving medication treatment. We accessed the UK Biobank to obtain information on 290 eyes from 161 participants who reported a diagnosis of glaucoma and were receiving treatment. Features were divided into three distinct datasets containing demographic data only, physiometabolic parameters and medication prescription data, and all data combined. We evaluated five machine learning techniques for each feature set in terms of their ability to predict intraocular pressure at a follow-up visit in a classification task. We then calculated SHapley Additive exPlanation (SHAP) values for the best performing model to determine feature importance, stability, and interactions. We found that eXtreme Gradient Boosting (XGBoost) outperformed all other models when trained and tested on the combined feature set with an area under receiver operating characteristic curve (AUC) of 0.708. Insulin-like growth factor 1 (IGF-1), low-density lipoprotein (LDL), and lymphocyte count ranked as the three most important features for this model. LDL and IGF-1 exhibited a low degree of global variability in contribution to the model output across all cross-validation repeats. SHAP values demonstrated the strongest interactions being between LDL and IGF-1. In summary, our studies indicated the importance of blood LDL and IGF-1 in contributing to the outcomes of intraocular pressure lowering treatment and demonstrated the ability of XGBoost to predict these outcomes.

Using several melanoma proteomics data sets we created a single analysis platform that enables the discovery, knowledge build, and validation of diagnostic, predictive, and prognostic biomarkers at the protein level. Quantitative mass-spectrometry-based proteomic data was obtained from five independent cohorts, including 489 tissue samples from 394 patients with accompanying clinical metadata. We established an interactive R-based web platform that enables the comparison of protein levels across diverse cohorts, and supports correlation analysis between proteins and clinical metadata including survival outcomes. By comparing differential protein levels between metastatic, primary tumor, and nonmalignant samples in two of the cohorts, we identified 274 proteins showing significant differences among the sample types. Further analysis of these 274 proteins in lymph node metastatic samples from a third cohort revealed that 45 proteins exhibited a significant effect on patient survival. The three most significant proteins were HP (HR = 4.67, p = 2.8e-06), LGALS7 (HR = 3.83, p = 2.9e-05), and UBQLN1 (HR = 3.2, p = 4.8e-05). The user-friendly interactive web platform, accessible at https://www.tnmplot.com/melanoma, provides an interactive interface for the analysis of proteomic and clinical data. The MEL-PLOT platform, through its interactive capabilities, streamlines the creation of a comprehensive knowledge base, empowering hypothesis formulation and diligent monitoring of the most recent advancements in the domains of biomedical research and drug development.

Human endogenous retroviruses (HERVs) occupy a large portion of the human genome. Most HERVs are transcriptionally silent, but they can be reactivated during pathological states such as viral infection and certain cancers. The HERV-K HML-2 clade includes elements that recently integrated have in the human germ line and often contain intact open reading frames that possibly support peptide and protein expression. Understanding HERV-K-host interactions and their potential as biomarkers is problematic due to the high similarity among different elements. Previously, we described a long-read single molecule real-time sequencing (PacBio) strategy to analyze HERV-K RNA expression profiles in different cell types. However, identifying HERV-K HML-2 proteins accurately is difficult without robust and reliable methods and reagents. Here we present a new approach to characterize the HML-2 elements that (a) are being translated and (b) produce enough protein to be detected and identified by mass spectrometry. Our data reveal that RNA expression profiling alone cannot accurately predict which HML-2 elements are responsible for protein production, as we observe several differences between the highest expressed RNAs and the elements that are the predominant source of HERV-K HML-2 protein synthesis. These studies represent an important advance toward untangling the complexity of HERV-K-host interactions.

[This corrects the article DOI: 10.3389/fimmu.2025.1590165.].

INTRODUCTION/UNASSIGNED:Recent investigations have identified patients of African ancestry (AA) with Multiple Sclerosis (MS), who display more rapid B-cell repopulation after standard semi-annual infusions with an anti-CD20 monoclonal antibody for B cell depletion. In this study, we explored the immunologic and genetic factors, with, serum drug monitoring that may contribute to a faster rate of B-cell repletion that follows during recovery from treatment with anti-CD20 antibody. METHODS/UNASSIGNED:In AA MS patients treated with an anti-CD20 antibody that had early repopulation of peripheral blood B cells, we assessed for extrinsic factors, including the presence of anti-drug antibodies against ocrelizumab, which may contribute to early repletion. We also documented the associated serum drug levels. In addition, we examined for inheritance of intrinsic gene polymorphisms associated with B cell survival and immune function. RESULTS/UNASSIGNED:Our findings identified a subset of AA patients with early B cell repletion after anti-CD20 treatment associated with anti-drug antibodies and an absence of detectable drug. Furthermore, a separate set of AA patients with the early B cell repletion phenotype without anti-drug antibodies had significant over-representation of genetic polymorphisms that map to genes for the B cell survival factor, BAFF, to antibody-dependent cytotoxicity, and to pathways involved in inflammation, leukocyte activation and B cell differentiation. DISCUSSION/UNASSIGNED:In AA patients with MS, after anti-CD20 antibody treatment we found an unexpected high occurrence of early B cell replenishment. This was associated with the presence of anti-drug antibodies and/or specific genetic polymorphisms. Larger studies are now needed to determine whether these factors may lead to impaired therapeutic benefits of B cell targeted therapy and clinical progression, and these findings may be useful to guide future optimized personalized therapeutic strategies.

GPR133 (ADGRD1) is an adhesion G-protein-coupled receptor that signals through Gαs/cyclic AMP (cAMP) and is required for the growth of glioblastoma (GBM), an aggressive brain malignancy. The regulation of GPR133 signaling is incompletely understood. Here, we use proximity biotinylation proteomics to identify ESYT1, a Ca²⁺-dependent mediator of endoplasmic reticulum-plasma membrane bridge formation, as an intracellular interactor of GPR133. ESYT1 knockdown or knockout increases GPR133 signaling, while its overexpression has the opposite effect, without altering GPR133 levels in the plasma membrane. The GPR133-ESYT1 interaction requires the Ca²⁺-sensing C2C domain of ESYT1. Thapsigargin-mediated increases in cytosolic Ca²⁺ relieve signaling-suppressive effects of ESYT1 by promoting ESYT1-GPR133 dissociation. ESYT1 knockdown or knockout in GBM slows tumor growth, suggesting tumorigenic functions of ESYT1. Our findings demonstrate a mechanism for the modulation of GPR133 signaling by increased cytosolic Ca²⁺, which reduces the signaling-suppressive interaction between GPR133 and ESYT1 to raise cAMP levels.

INTRODUCTION/UNASSIGNED:Biomarker discovery is increasingly moving from single omics to multiomics, as well as from multi-cell omics to single-cell omics. These transitions have increasingly adopted digital transformation technologies to accelerate the progression from data to insight. Here, we will discuss the concept of 'digitalomics' and how digital transformation directly impacts biomarker discovery. This will ultimately assist clinicians in personalized therapy and precision-medicine treatment decisions. AREAS COVERED/UNASSIGNED:Genotype-to-phenotype-based insight generation involves integrating large amounts of complex multiomic data. This data integration and analysis is aided through digital transformation, leading to better clinical outcomes. We also highlight the challenges and opportunities of Digitalomics, and provide examples of the application of Artificial Intelligence, cloud- and high-performance computing, and use of tensors for multiomic analysis workflows. EXPERT OPINION/UNASSIGNED:Biomarker discovery, aided by digital transformation, is having a significant impact on cancer, cardiovascular, infectious, immunological, and neurological diseases, among others. Data insights garnered from multiomic analyses, combined with patient meta data, aids patient stratification and targeted treatment across a broad spectrum of diseases. Digital transformation offers time and cost savings while leading to improved patent healthcare. Here, we highlight the impact of digital transformation on multiomics- based biomarker discovery with specific applications related to oncology.

OBJECTIVE:Whereas genetic susceptibility for systemic lupus erythematosus (SLE) has been well explored, the triggers for clinical disease flares remain elusive. To investigate relationships between microbiota community resilience and disease activity, we performed the first longitudinal analyses of lupus gut-microbiota communities. METHODS:In an observational study, taxononomic analyses, including multivariate analysis of ß-diversity, assessed time-dependent alterations in faecal communities from patients and healthy controls. From gut blooms, strains were isolated, with genomes and associated glycans analysed. RESULTS:(RG) occurred at times of high-disease activity, and were detected in almost half of patients during lupus nephritis (LN) disease flares. Whole genome sequence analysis of RG strains isolated during these flares documented 34 genes postulated to aid adaptation and expansion within a host with an inflammatory condition. Yet, the most specific feature of strains found during lupus flares was the common expression of a novel type of cell membrane-associated lipoglycan. These lipoglycans share conserved structural features documented by mass spectroscopy, and highly immunogenic repetitive antigenic-determinants, recognised by high-level serum IgG2 antibodies, that spontaneously arose, concurrent with RG blooms and lupus flares. CONCLUSIONS:Our findings rationalise how blooms of the RG pathobiont may be common drivers of clinical flares of often remitting-relapsing lupus disease, and highlight the potential pathogenic properties of specific strains isolated from active LN patients.