Poster Abstracts

Eva Agostino, Eric Porsch, Joe St. Geme III

Kingella kingae is a Gram-negative, bacterial pathogen that is a leading cause of osteoarticular infections and other invasive diseases in young children. To establish infection at distal sites, K. kingae must survive intravascularly by resisting complement-mediated killing. Our lab has identified a surface exposed lipoprotein (KK01_02920) that binds to human complement regulator factor H (FH) to protect K. kingae from complement-mediated killing. In this study, we examined the genomes of 102 epidemiologically diverse K. kingae clinical isolates and found that KK01_02920 is universal and highly conserved across the K. kingae population. Using AlphaFold structure predictions, we identified multiple residues in KK01_02920 that are potential binding sites for FH. We generated amino acid substitutions and found several residues to be important for FH binding. We have purified recombinant KK01_02920 that maintains its ability to bind FH. In ongoing studies, we are continuing assessment of other predicted FH binding residues, various biophysical properties of the KK01_02920-FH complex, and the impact of KK01_02920 and FH binding on K. kingae pathogenesis. This work will advance our understanding of K. kingae pathogenicity and may facilitate development of a vaccine against K. kingae.

Katrina Bazemore, Taha Iqbal, Jin Sha, Jacob Haut, Wan-Ping Lee, Yi Zhao, Otto Valladares, Li-San Wang, Gerald Schellenberg, Jin Jin, Adam Naj

Polygenic risk scores (PRS) are useful tools for measuring genetic liability to complex diseases but can overlook underlying heterogeneity across specific contributors to genetic liability. The pathway-specific PRS (pathway-PRS) model is more representative of heterogeneity in genetic liability and can inform inferences on biologic mechanisms. However, pathway analyses often do not consider the role of non-coding variants, which are key contributors to most complex traits including Alzheimer’s Disease (AD).

To examine pathway-PRS associations with AD and the potential of integrative annotation, we compared three variant-to-gene annotation strategies in the estimation of pathway-specific AD risk in the UK Biobank across 20 pathway-PRS. 

We show that predictive performance is meaningfully improved for 50% of AD pathway-PRS tested under our most integrative strategy, compared to a positional approach (35 kb upstream/10 kb downstream of gene boundaries). This inclusion of regulatory variants resulted in a stronger association with AD observed for protein and cellular localization pathways (OR=1.081, P=5.4×10-11) than for amyloid processing and formation pathways (OR=1.077, P=1.49×10-10).

Systematic incorporation of regulatory information into pathway-PRS construction yields substantial improvements in predictive performance. Our approach should be expanded upon to continue enhancing pathway-PRS based research in AD and other complex diseases.

Mark Bray, Prajna Mishra, Devin Kelly, Claire Woodward, Matthew Cruz, Jonah Park, Gregory Bowman

Ebola remains a severe global public health threat with limited therapeutic strategies. The single-stranded RNA virus is highly pathogenic, and encodes seven proteins: nucleoprotein (NP), viral protein 35 (VP35), VP40, glycoprotein, VP30, VP24, and an RNA-dependent RNA polymerase. Cell biological, biochemical, and structural evidence indicates that VP35, VP30, VP24, and nucleoprotein are sufficient and requisite components of the viral ribonucleocapsid complex, but characterization of this complex has remained elusive. Understanding this complex is imperative because clinical observations of Ebola patients suggest that the virus’ high fatality rate is correlated with high viral load due to uncontrolled viral replication. Critically, VP35 enables viral genome transcription by binding to nucleoprotein, and its C-terminal interferon inhibitory domain (VP35IID) binds dsRNA that forms as complementary viral RNA strands are synthesized. The Bowman lab previously identified a VP35IID cryptic pocket that is allosterically coupled to its dsRNA binding and identified mutations that control the cryptic pocket. When the cryptic pocket is open, dsRNA binding is reduced; and when the cryptic pocket is closed, dsRNA binding is enhanced. Here, we apply fluorescence polarization, HDX-MS, and NMR to characterize the nucleoprotein-VP35IID interaction. We additionally apply AI-based algorithms to design inhibitory peptides targeting the VP35IID cryptic pocket.

Taylor Brysgel, Nicolas Marzolinii, Eno-Obong Essien, Jichuan Wu, Manthan Patel, Jacob Myerson, Sachchidanand Tiwari, Vladimir Shuvaev, Elizabeth Hood, Oscar Marcos Contreras, Vladimir Muzykantov, Sahily Reyes-Esteves, Jacob Brenner

DNA-lipid nanoparticles (DNA-LNPs) loaded with inhibitors of the cGAS-STING pathway enable safe and effective delivery of DNA in vivo. Herein, we report the first instances of extrahepatic DNA-LNP targeting. DNA-LNPs conjugated to antibodies against PECAM-1 target the endothelium of the lungs. These LNPs drive robust transgene expression in their target organ, with greater magnitude and duration than untargeted LNPs. Lung specificity of PECAM-targeted transgene expression increases over two weeks, resulting in markedly higher lung-to-liver expression ratios than our previous PECAM-targeted mRNA-LNPs. Off-target liver DNA expression declines to undetectable levels but persists in the lungs, while mRNA expression uniformly decreases due to its short half-life. We further improve this expression specificity by replacing full-length antibodies with Fab fragments. Single-cell analysis reveals the mechanism underlying the improvements in organ-specificity: target organ expression is dominated by long-lived endothelial cells, while off-target liver expression is in non-endothelial cells with shorter half-lives. Collectively, these studies demonstrate that targeted DNA-LNPs achieve high levels of organ- and cell-type-specific transgene expression and thus provide a therapeutic platform for dozens of endothelial-centric diseases.

Marc Carceles-Cordon, Eliza Brody, Travis L. Unger, Michael D. Gallagher, Robert T Skrinak, Masen L. Boucher, Cooper K. Penner, Adama J. Berndt, Sromona Das, Rudolf Jaenisch, Vivianna Van Deerlin, Edward B. Lee, Kurt Brunden, Kelvin C. Luk, Alice Chen-Plotkin

Glycoprotein nonmetastatic melanoma B (GPNMB), encoded by the target gene (GPNMB) of a Parkinson’s disease (PD) risk locus (rs199347), acts as a secreted factor mediating inflammatory effects in the context of immunity and cancer. In a neurodegenerative disease context, GPNMB is critical to cellular uptake of pathological forms of alpha-synuclein (aSyn), the hallmark disease protein that misfolds and accumulates in PD. Here, we demonstrate that the non-membrane-anchored, extracellular domain of GPNMB, shed into conditioned medium or added as recombinant protein, is sufficient to enable uptake of aSyn fibrils in a non-cell-autonomous manner. In human postmortem brain, GPNMB is widely expressed in neurons and microglia, with increased microglial expression in the setting of neurodegenerative disease. In microglial cell lines and induced pluripotent stem cell-derived microglia (iMicroglia), GPNMB expression and secretion increases with exposure to neurodegeneration-related substrates. In the aSyn-fibril seeded model of PD, iMicroglia-derived GPNMB is sufficient for development of aSyn pathology in GPNMB knockout neurons. Finally, in 1675 human postmortem cases, GPNMB genotypes conferring higher GPNMB expression associate with more widespread aSyn pathology, without affecting beta-amyloid or tau pathology.

Nicholas A. Cerda, Brieyanna McWilliams, Awurama Akyianu, Manthan Patel, Elena Atochina-Vasserman, Jackie Meshanni, Qin Li, Amie Albertus, Tyler E. Papp, Nathan Ona, Faris Halilovic, Ishana Baboo, Michael Kegel, Drew Weissman, Hamideh Parhiz

Paramyxoviruses and pneumoviruses have been a leading cause of death worldwide. These virus families encompass many significant respiratory RNA viruses, including measles, mumps, parainfluenza 1-4, respiratory syncytial virus, and metapneumovirus. While the use of the MMR vaccine has reduced cases, over 66 million children under five are infected with these viruses every year. The recent licensure of the first vaccines against RSV for older adults has made a significant impact on pneumovirus infections. Still, children who are at the most considerable risk have no immunization available beyond the MMR vaccine. Here, we developed an mRNA vaccine encoding prefusion-stabilized antigens that self-assemble into virus-like particles (VLP). This VLP-forming mRNA vaccine showed higher efficacy in generating B cell responses, eliciting strong antibody titers, and increasing neutralizing antibodies. Additionally, we applied the same strategy for combination vaccination and observed strong humoral responses against all six viruses. These preclinical results establish a critical foundation for the development of the first vaccine with the potential to confer protection across entire viral families.

Xiaodi Chen, Haya Habib, Nima Naseri, Ophir Shalem, Roger Greenberg

The accumulation of DNA damage in neurons and the resulting disruption of cellular functions is central to aging and various neurological disorders. While the DNA damage response has been well characterized in dividing cells, how these pathways are adapted to meet the demands of terminally differentiated neurons remains less understood. Among the many sources of genome instability, transcription-coupled damage is particularly relevant in neurons. Neurons exhibit robust and rapid stimulus-driven gene transcription in response to synaptic input and other environmental signals. While this dynamic regulation is essential for neuronal plasticity, high level of transcription leads to endogenous DNA damage. Transcription-associated damage can arise from processes such as R-loop accumulation, topoisomerase trapping, and exposure of single-stranded DNA. To systematically identify regulators of this process, we performed a phenotypic CRISPR interference screen in human iPSC-derived neurons using gammaH2AX as a marker of DNA double-strand breaks. The screen recovered canonical DNA damage response factors, validating the approach, and revealed novel regulators, including candidates with apparent neuronal specificity and strong links to neurological disorders. These findings highlight both conserved and neuron-specific pathways that protect genome integrity and set the stage for mechanistic studies into how their disruption contributes to neuronal dysfunction and disease.

Diana Cruz, Julio Marques Ricarte Filho, Sarah Monteiro, Aime Franco, Guilherme Nader

Papillary thyroid cancer (PTC) is diagnosed by nuclear atypia enlarged/wrinkled nuclei. However, the causes and consequences of PTC nuclear atypia remain unknown. Our patient bulk RNAseq data revealed reduced expression of lamin B receptor (LBR) in PTC compared to normal tissue. To investigate, we employed PCCL3 HRAS, a rat thyroid follicular cell line with a dox-inducible HRASv12, a PTC driver. Upon HRASv12 induction, LBR levels decreased (by western blot) and these nuclei displayed increased size, wrinkles, and shape abnormalities consistent with PTC nuclear atypia (via super-resolution microscopy). Interestingly, immunofluorescence revealed a shift in LBR localization from the nuclear envelope to the ER, a pattern we also observed in patient tumors. LBR is a nuclear membrane protein and sterol reductase for lipid biosynthesis. Given the ER’s continuity with the nuclear membrane, we hypothesize that ER-enriched LBR enhances lipid synthesis at the ER expanding the nuclear membrane and causing wrinkles. This may buffer mechanical stress, a hallmark of solid tumors. To test this, we confined cells and measured nuclear envelope rupture using a cytosolic DNA sensor. We found PCCL3 cells expressing HRASv12-expressing cells displayed reduced rupture events under stress, supporting our model that ER-enriched LBR drives nuclear atypia and buffers mechanical stress.

Dominique Doyle, Mijeong Kim, Maria Basil

Chronic respiratory diseases (CRDs), including chronic obstructive respiratory disease (COPD), are among the top ten leading causes of death worldwide, due in part to a lack of new treatments. While valuable pulmonary research stems from using murine models, mice and humans have significant differences in the structural, cellular, and molecular components of their lungs. In humans, there are more distal airway generations termed respiratory bronchioles (RBs). RBs are “hybrid-like” regions that bridge the conducting airways and the gas-exchanging units of the alveoli, harboring a distinct epithelial airway cell population. These respiratory airway secretory cells (RASCs) are known to self-renew and act as facultative progenitor cells for alveolar type 2 (AT2) cells, but what regulates RASC fate is unknown. SOX4, a transcription factor with no known role in normal lung, shows regional specificity for RASCs. Our lab has demonstrated that SOX4 is essential for RASC maintenance, but the role of SOX4 in airway cell fate specification is unknown. However, in our lab’s human embryonic stem cell (hESC) organoid models, SOX4 expression is upregulated during airway cell differentiation from lung progenitors. Perturbating SOX4 expression may influence RASC fate and differentiation potential during lung progenitor development.

Jiayi Duan, Andrew D. Marques, Matthew Hogenauer, Young Hwang, Yanjia Zhang, Aaron T. Timperman, Stephanie Higgins, Aunie Fitts, Karissa Lim, Naomi Wilson, Ahmed Moustafa, Ronald G. Collman, Frederic D. Bushman 

The human virome is vast and diverse, with approximately 1013 virus particles per individual. In response to the NIH Human Virome Program’s initiative, we are optimizing methods to capture viral diversity across sample types by generating and using a synthetic viral community to assess several technical challenges important in virome analysis. Our mock community included phage T4, lambda, M13, MS2, phi6, an adenovirus-associated vector, murine hepatitis virus, and vaccinia virus. We spiked the mock community into different human sample types, including stool, saliva, oropharyngeal (OP) wash, and bronchoalveolar lavage (BAL), then passed the samples through different virus enrichment protocols for Illumina sequencing. 

Compared to direct metagenomic sequencing, VLP enrichment protocols greatly increased viral reads yields from stool and saliva, but not BAL. Phage MS2 was highly susceptible to nuclease treatment. Glucosylhydroxylmethyl modification on T4 genome could partially inhibit sequencing detection of bases. Amplification led to distorted representation of virus populations, with most methods greatly favoring small circular ssDNA viruses. Comparison of Illumina 1000-cycle kit versus 300-cycle kit showed that longer reads supported generation of longer virus genome assemblies. Overall, our study demonstrates that virus enrichment methods can significantly influence virome profiling outcomes, and spike-in studies can be used for optimization. 

Andrew F. Jarvis, Mohd Younis Bhat, Timothe Maujean, Islam Elsaid, Terence P. Gade, George M. Burslem, Donita C. Brady

Precision medicine has largely emphasized cancer genetics, overlooking decades of evidence that alterations in protein function drive cancer cell behavior. However, existing technologies have limited the ability to track oncogenic protein activity for therapeutic discovery. Our lab developed the probe-enabled activity reporting (PEAR) platform, a chemoproteomic approach that directly profiles the dynamic tumor proteome. PEAR integrates activity-based protein profiling (ABPP) and reverse-polarity ABPP (RP-ABPP), using probes targeting amino acids, post-translational modifications, and cofactors, to covalently label reactive protein sites. This allows high-resolution mapping of proteomic changes associated with cancer progression and treatment response. In totality, PEAR enables visualization of therapy-induced proteomic adaptations, links tumor biology to oncogenic protein activity, and guides development of functionalized therapeutics.

Although cancer-type agnostic, we applied PEAR to hepatocellular carcinoma (HCC), a malignancy with rising incidence, limited therapies, and no effective genomic targets. Using a model of transarterial chemoembolization (TACE), the HCC standard-of-care, PEAR revealed enriched cysteine accessibility in PARP1, a DNA repair regulator, under TACE-like stress. This suggests PARP1 as a therapeutic vulnerability in HCC. This project will define PARP1’s role in HCC adaptation to TACE and evaluate PARP inhibition as a therapeutic strategy, highlighting PEAR’s utility in uncovering functional protein liabilities for precision therapy.

Noa Erlitzki, Rahul Kohli

DNA methylation and demethylation dynamics are critical for normal development. DNA 
demethylation is mediated by TET enzymes, which generate products that feed into 
demethylation pathways by iteratively oxidizing 5mC to 5-hydroxymethylcytosine (5hmC), 
5-formylcytosine (5fC), and 5-carboxycytosine (5caC). 5hmC not only serves as a transient 
intermediate for demethylation, but also can be a stable modification that is defining of cell 
and tissue identity. Thus, it is important to know where in the genome 5hmC accumulates 
or gets depleted. DNA sequence context has been shown to strongly influence TET kinetics
in vitro, with potential implications for the determinants of 5hmC flux and demethylation 
dynamics in vivo. However, prior methods have been unable to parse the generation of 
5hmC from its decay to 5fC/5caC, the balance between which can promote or impede 
5hmC accumulation. Using a novel, high-throughput enzymology approach to observe the 
coupled generation and depletion of 5hmC in vitro, we have identified distinct sequence 
context preferences for 5hmC generation, depletion, and accumulation. Importantly, these
sequence context preferences correlate with sequence context-specific differences in 
5hmC levels in vivo. Our data suggest that efficient generation of 5hmC may be a more 
significant determinant of the 5hmC landscape in vivo than inefficient depletion of 5hmC, 
establishing a molecular basis for the biological roles of 5hmC as a stable epigenetic 
modification.

Carolann L Espy, Jichuan Wu, Serena Omo-Lamai, Liuqian Wang, Fengyi Dong, Aleksa Milosavljevic, Sarah O’Neill, Michael Zaleski, Yufei Wang, Emily Wolfe, Rhea Maheshwari, Manthan Patel, Jia Nong, Zhicheng Wang, Jacob S Brenner

Lipid nanoparticles (LNPs) are widely used for nucleic acid delivery, yet their clinical utility is limited by an acute inflammatory response post-administration. LNP-associated inflammation (LAI) manifests as de novo inflammation in select organs (e.g., in the lungs after inhalation) and severely worsens pre-existing inflammation.  NF-κB is a key transcription factor mediating this inflammation and represents a promising target for LAI mitigation strategies. However, systemic NF-κB inhibition carries risks of global immunosuppression, highlighting the need for localized inhibition. Here, we screen a panel of direct and indirect small molecule inhibitors of NF-κB for their ability to reduce LAI without disrupting LNPs’ expression. We identify inhibitors that load efficiently into LNPs, retain anti-inflammatory activity in vitro and in vivo, and preserve expression of cargo mRNA in target tissues. Our results demonstrate that co-formulation of NF-κB inhibitors with cargo nucleic acids enables localized immunomodulation in LNPs, offering a modular strategy to enhance the safety profile of LNPs for therapeutic use in inflammatory or high-risk patient populations.

Sarah M. Ferrigno, Nathan Zhang, Evan Iliakis, Saurabh Pandey, Jamie Galanaugh, Marc V. Fuccillo

Inhibitory control, or the ability to withhold action in certain situations, is behaviorally essential. Disrupted inhibitory control is linked to various neuropsychiatric symptoms, making it critical to understand the underlying neural basis. We examined how the tail of the striatum (TS), a major basal ganglia sensory hub, regulates actions to sensory stimuli. Mice performed an auditory Go/NoGo task where we recorded cell-specific activity of TS neurons. Both major striatal types were active during target sounds, but non-target sounds preferentially engaged indirect pathway neurons. Temporarily silencing this activity increased errors to non-target stimuli, indicating a role in suppressing inappropriate action. In mice deficient for the synaptic adhesion molecule Neurexin1alpha, a gene linked to autism spectrum disorder and ADHD, TS indirect pathway recruitment was reduced, and these mice demonstrated auditory-specific inhibitory control deficits. Altogether, these findings highlight a subcortical target to potentially improve attentional and behavioral regulation in neurodevelopmental disorders.

Carmen Gagliardi Flesher, Xi Lin, Xuechen Shi, Sarah Traynor, Natalie Moore, Kathleen Lee, Son Tran, Bat Tumenbayar, David Merrick

Obesity and its consequential metabolic diseases constitute a grave public health crisis. The introduction of GLP1-based medications has enabled millions of patients to achieve significant weight loss and amelioration of metabolic disease. However, multiple studies demonstrate a profound inclination for rapid weight regain following medication discontinuation. To understand gene expression changes in lean, obese, and weight loss states, we performed single cell RNA-sequencing in mice and humans. These data demonstrated obesity-induced changes in extracellular matrix (ECM)-modifying genes that failed to return to the lean expression state after weight loss. The adipose tissue ECM plays important mechanochemical roles in cell signaling and volumetric expansion to accommodate nutrient influx. However, almost nothing is known about the mechanochemical changes induced by weight gain and loss and the capacity for reversion following weight loss. To investigate the physical environment of adipose tissue in lean, obese, and weight loss states, I performed rheology to measure tissue stiffness using applied force that reproduces the physical tissue environment in vivo. These data showed that obese tissue is hyper-responsive to force, while weight loss results in maladaptive adipose softening. Together, these data suggest pathological ECM remodeling in obesity that fails to remodel properly following weight loss.

William Gao, Peng Hu, Qi Qiu, Xiangjin Kang, Kenneth C. Bedi, Kotaro Sasaki, Kenneth B. Margulies, Hao Wu

The first organ to develop in utero, the human heart undergoes significant changes during development and must sustain its function over a lifetime. To characterize molecular changes in human cardiac cell-types across sex, aging, developmental and disease, we analyzed single nucleus RNA sequencing (snRNA-seq) datasets from 299 donors, identifying many more differentially expressed genes (DEGs) across developmental and disease states than by sex and age. In cardiomyocytes and non-cardiomyocyte cell types, developmental and disease DEGs showed significant overlap. Cardiac development and disease had convergent changes in intercellular communication, including TGFβ signaling, but differences in cell-type proportions. By integrating snRNA-seq with 106 snATAC-seq datasets, we reveal potential transcriptional factors driving fetal reactivation in disease. Finally, using spatial transcriptomics data, we identify that fetal reactivation is highly localized in niches. This work offers the largest multimodal, cell-type resolved interrogation of the human heart, providing greater insights into convergence in development and disease.

Reyna Garcia Sillas, Igor Brodsky

Sepsis arises from a dysregulated immune response to infection. The breakdown in the equilibrium of the immune system is driven by uncontrolled cell death. Gram-negative pathogens like Yersinia pseudotuberculosis (Yp) are known to manipulate host response through the injection of effectors known as Yersinia outer proteins (YOPs). The infection model for Yp has been extensively studied in mouse macrophages. However, our recent data in human macrophages (hMDMs) suggests a distinct host response in humans that may involve alternative cell death or signaling pathways. To investigate this, we infected primary hMDMs with WT Yp, Yp delta-yopJ, or Yp delta-yop6 (lacking all YOPs). WT Yp induced moderate apoptosis with caspase-8 and -3 activation and minimal IL-1beta/IL-18 secretion. Yp lacking yopJ did not induce cell death but triggered strong cytokine release—indicating a possible hypersecretory phenotype. Yp lacking all yops caused robust cell death and cytokine release. NLRP3 inhibition with MCC950 had no effect on cytokine secretion, while caspase-1 inhibition (AC-YVAD) reduced cytokine secretion, suggesting caspase-1 involvement in this hypersecretory phenotype. By understanding how Yp manipulates cell death and cytokine release, this work will provide insight into the regulation of inflammation, immune homeostasis, and mechanisms that may underlie sepsis.

Alexis Michelle Garófalo, Perapa Chotiprasidhi, Josephine Princy Johnson, Karina Sato-Espinoza, Jun Ma, Hunter Miller, Daniel O'Brien, Lindsay Guare, Katie M. Cardone, Nicole Palmiero, Zachary Rodriguez, David E. Kaplan, Julie A. Lynch, Phil Tsao, Daniel J. Rader, Lewis R. Roberts, Jose Debes, Emuobor Odeghe, Edith Okeke, Jun Wang, Lifang Hou, Adwoa Agyei-Nkansah, Mary Yeboah, Yaw Asante Awuku, Albert Nyanga, Kyong-Mi Chang, Samuel O. Antwi, Marijana Vujkovic, Anurag Verma, Kirk Joseph Wangensteen

Liver cancer is the third leading cause of cancer death worldwide, with hepatocellular carcinoma (HCC) being the most common type. The prevalence of rare pathogenic/likely pathogenic (P/LP) germline variants in cancer predisposition genes is estimated at 3–17% in HCC patients, yet their contributions to HCC risk are poorly understood. We evaluated associations between rare P/LP variants in DNA-repair genes and HCC risk across ancestrally diverse cohorts.
We analyzed whole-exome and whole-genome sequencing data on 2,594 HCC cases and 290,547 cancer-free controls from four global biobanks: Penn Medicine BioBank, All of Us, Mayo Clinic, and Million Veteran Program. We focused on six DNA-repair genes routinely included on clinical cancer genetics panels: BRCA2, BRIP1, MSH6, PMS2, CHEK2, and FANCA. Gene-level burden analyses were performed within European and African populations, plus multi-population analysis.

In the European population, MSH6, a Lynch syndrome gene, showed strongest HCC association with 2.75-fold increased risk (OR=2.75[1.50,5.04], P=0.001, FDR q=0.02). PMS2 showed nominally significant association (OR=1.90[1.08,3.34], P=0.03, FDR q=0.09). Combined population analysis strengthened MSH6 findings (OR=2.53[1.43,4.49], P=0.001, FDR q=0.01) and revealed a significant BRCA2 association (OR=2.26[1.39,3.67], P=0.001, FDR q=0.01). These findings reveal a previously unrecognized role for DNA repair genes in HCC susceptibility and may inform genetic risk stratification strategies.

Aria Garrett, Elaine Mihelc, Bridget McVeigh, Vera Moiseenkova-Bell

The transient receptor potential mucolipin 1 (TRPML1) ion channel is implicated in diseases with altered lysosomal homeostasis, including Alzheimer’s and Parkinson’s disease, cancers, Duchenne muscular dystrophy, and lysosomal storage disorders. Lysosomes, although often depicted as spherical, can elongate into tubules or form tubular projections under conditions such as immune activation, nutrient deprivation, and aging. Increasing evidence suggests these tubular lysosomes are heterogeneous and functionally distinct, yet their subpopulations remain poorly defined. The objective of this study is to determine how TRPML1 activity influences the structural diversity of tubular lysosomes. To address this, I am using a combination of live-cell fluorescence imaging, focused ion beam scanning electron microscopy, and electron tomography to characterize lysosomal tubulation. Light microscopy confirmed that TRPML1 activity modulates lysosomal tubulation, while electron tomography is uncovering the subpopulations of tubular lysosomes in situ. Defining these subpopulations advances our understanding of lysosomal tubulation and clarifies TRPML1’s role in lysosomal biology.

James Gesualdi, Cagla Akay-Espinoza, Kelly Jordan-Sciutto

Despite the high efficacy of antiretroviral therapies, HIV persists in the central nervous system (CNS). This leads to sustained neuroinflammation and contributes to the development of neurocognitive impairment referred to as HIV-associated neurocognitive disorder (HAND) in people living with HIV. Microglia are the major population of HIV-susceptible cells in the CNS and astrocytes are also impacted by HIV infection via indirect activation by neighboring infected cells. The dynamics of HIV replication in microglia as well as the immune response of microglia and astrocytes to HIV infection remain poorly understood. Using human induced pluripotent stem cell-derived microglia (iMg) and astrocytes (iAst) we show that HIV replicates in iMg but not iAst. Surprisingly, coculture of iAst with iMg leads to robust increases in HIV replication. iAst exposed to infected iMg produce the pro-inflammatory cytokines TNF alpha and IL-6. These cytokines activate NF- kappa B signaling, which in turn drives transcription of the HIV genome. Inhibition of NF-kappa B signaling reduces HIV replication in iMg-iAst cocultures. Further, pro-inflammatory activation of iAst is associated with dysregulated lysosomal trafficking in iMg following HIV infection. This pro-inflammatory signaling axis may be a viable target of immunotherapies for HAND.

Lindsay Guare, Rachit Kumar, Jagyashila Das, Anurag Verma, Shefali Setia-Verma

Polygenic risk scores (PRSs) remain limited in clinical utility due to modest predictiveness and poor transferability. We focused on endometriosis (endo), a chronic and debilitating women’s health condition for which PRSs have thus far performed poorly. We hypothesize that PRSs are limited by collapsing biological information into one dimension. To address this, we present Geno2Vec, a deep heterogeneous graph learning method that embeds SNPs as vectors by integrating multi-omic data. We leveraged the learned vectors to test models for endo in ~250k female participants from the All of Us Research Program (AOU). The Geno2Vec-PRS had the highest pseudo-R-squared (0.011) as compared to traditional PRS (0.0099) and covariate-only null (0.0076) models, indicating that it explained the greatest proportion of variance. The training AUROC was highest for the Geno2Vec-PRS (0.5920). When applied to the validation subset, the traditional PRS model had the highest AUROC overall (0.5838) and for the AFR (0.5132) and AMR (0.6124) groups. Geno2Vec-PRS had the highest EUR test AUROC (0.5667). In the future, we will incorporate more –omics types and apply this methodology to a wide range of complex diseases. This approach represents innovation in genomic risk modeling, offering enhanced predictive capability through learned representations of complex multi-omic relationships.

Natalie Hagen, QianXuan She, Ahmed Moustafa, Joseph Zackular, Frederic D. Bushman

Clostridioides difficile infection (CDI) is the number one cause of nosocomial diarrhea worldwide and is an urgent antibiotic resistance threat. Antibiotics are the current treatment for CDI, however, this is insufficient due to a high rate of recurrent infection and development of antibiotic resistance. Therefore, development of new therapies for CDI is critical. Phage therapy is a promising option, but is challenging in part due to the lack of C. difficile phages which are strictly lytic. While the C. difficile genome is rich in prophages, administration of a temperate phage can lead to lysogeny and subsequent outgrowth of bacteria immune to re-infection.  Advancement of phage therapy requires better understanding of endogenous prophages. We have investigated the C. difficile “induce-ome” by testing the conditions such as mitomycin C, hydrogen peroxide, and bile acids on phage induction in over 20 clinical isolates. We purified the resulting virus-like particles and extracted DNA for next-generation sequencing. We have documented over 30 inducible prophages. Analysis is ongoing and includes annotation by nucleotide and protein structural homology and assessment for presence of antibiotic resistance genes and virulence factors. This work documents genetic diversity of C. difficile inducible elements, information useful for engineering phage therapies.

Avery Hawthorne, Kai Kelly, Mark Tigue, Elizabeth Desranleu, Yelda Erhan, Eric Waite, Klaus Kaestner

Objectives: Common gene variants in STEAP2 are strongly associated with Type 2 diabetes (T2D) risk. STEAP2 regulates iron uptake which plays a critical role in mitochondrial function necessary for pancreatic β-cell glucose-stimulated insulin secretion. Thus, we sought to identify the effect of β-cell specific knockout of STEAP2 with a STEAP2fl/fl;RipCreER mouse on glucose homeostasis in mice.

Methods: STEAP2 knockout was verified by RT-PCR of isolated islets. 16-hr fasted intraperitoneal glucose tolerance test (IP GTT), oral-GTT (OGTT); and unfasted insulin tolerance tests (ITT) were performed in 6-wk old mice. Blood glucose was measured by glucometer and plasma insulin was assessed by ELISA. Unfasted glucose and weight gain was monitored from 6-20 wks. 

Results: RT-PCR demonstrated a near complete reduction of STEAP2 mRNA. Blood glucose was increased in STEAP2fl/fl;RipCre+ mice during GTTs (p<0.01, n=24) but not ITT. Insulin release in early phases GTTs were significantly decreased in STEAP2fl/fl;RipCre+ mice (p<0.05, n=11). Unfasted glucose and weight were not significantly different between groups (n=7). 

Conclusions: We generated a STEAP2fl/fl mouse line and determined STEAP2 plays a critical role in β-cell insulin secretion in response to acute glucose challenge.

Katharine K. Hewlett, Amanda PeBenito, Aaron L. Hecht, Connor Tiffany, Ceylan Tanes, Rochelle C. Glover, Jibraan A. Fawad, Elliot S. Friedman, James C. Reynolds, Kyle Bittinger, James D. Lewis, Gary D. Wu, Nitin K. Ahuja, Joseph P. Zackular

Clostridioides difficile is the main cause of antibiotic-associated diarrhea in the United States. The main risk factor contributing to CDI is recent antibiotic use; however, some individuals remain susceptible to CDI for months post-antibiotic treatment for unknown reasons. Understanding the factors that contribute to CDI susceptibility is critical to controlling infection. Based on a clinical dietary intervention trial at our hospital, we hypothesized that dietary fiber may modulate susceptibility to C. difficile post-antibiotic treatment in patients. We investigated the metabolome and microbiota in human subjects on a low-fiber diet and found alterations to bacterial taxa and metabolites known to modulate colonization resistance against CDI. Next, we provided mice with fiber-rich or fiber-free diets and quantified CDI susceptibility after antibiotic treatment. We find that a low-fiber diet increases taurocholate, a bile acid known to promote C. difficile susceptibility, in both mouse and human stool. In both models, we further report long-lasting perturbation to the microbiota, highlighted by depletion of Clostridia and an enrichment of facultative anaerobes. In mice, we show that a low-fiber diet leads to increased CDI susceptibility and severity. Our work suggests that in the context of antibiotic treatment, diet is a critical, modifiable risk factor for CDI susceptibility.

Noah Hillman, Sarah M. Weinstein, Joëlle Bagautdinova, Kevin Y. Sun, Matthew Cieslak, Taylor Salo, Yong Fan, Arielle S. Keller, Aaron F. Alexander-Bloch, Simon N. Vandekar, Armin Raznahan, Theodore D. Satterthwaite, Haochang Shou, Russell T. Shinohara

Interpreting brain-behavior relationships through the lens of anatomical parcellations or functional networks is commonplace in human brain mapping. However, statistical approaches for testing whether brain–behavior associations are stronger (i.e., enriched) within a region of interest remain underdeveloped. Here, we propose a permutation-based approach for network enrichment testing using ordinal dominance curves (NETDOM). In simulation studies, we demonstrate that NETDOM properly controls the type I error rate -- unlike other prominent enrichment methods -- while exhibiting increased statistical power when enrichment occurs in a subset of in-network locations. Using data from two large-scale neurodevelopmental cohorts, we illustrate that NETDOM effectively detects enriched associations between structural and functional brain measures and neurocognitive performance.

Vivian Hoang, Krystal Ching, Matt Keller, Ken Cadwell

Autophagy proteins play critical roles beyond intracellular degradation, including the regulation of extracellular vesicle (EV) secretion. A specialized subset of EVs, termed defensosomes, act as protective decoys by displaying host receptors that neutralize microbial and viral toxins. For example, ADAM10+ defensosomes intercept Staphylococcus aureus α-toxin, while ACE2+ defensosomes block SARS-CoV-2 infection. Defensosome production can be induced through TLR9 activation with CpG-A in an ATG16L1-dependent manner, providing a tractable model to study EV biogenesis. Despite recent advances in EV isolation and characterization, the molecular mechanisms controlling defensosome formation remain poorly defined. Using improved EV detection platforms, we demonstrate that TMEM59 acts as a negative regulator of defensosome production and that their induction requires active transcription. These findings identify new molecular players in defensosome biogenesis and provide insight into how autophagy proteins regulate protective EV secretion.

Hannah H. Kim, Jose Angel Martinez-Sarmiento, Flavio R. Palma, Aayush Kant, Yujia (Ellen) Zhang, Zixian Guo, Robert L. Mauck, Su Chin Heo, Vivek Shenoy, Marcelo G.Bonini, Melike Lakadamyali

The eukaryotic genome is hierarchically organized across multiple spatial and genomic scales. Understanding how this organization relates to cell state is central to elucidating the molecular basis of development, cellular reprogramming, and disease. SMLM has been instrumental in revealing previously inaccessible aspects of chromatin organization and remodeling during physiological and pathological transitions.  While data generated by SMLM are complex and spatially rich, current analysis approaches fail to capitalize on this wealth of information.

We present O-SNAP (Objective Single-Molecule Nuclear Architecture Profiler), a comprehensive pipeline for automated extraction and classification of nuclear features from SMLM data. O-SNAP quantifies 144 biologically grounded features describing chromatin or histone mark organization at nanoscale resolution. It incorporates statistical tools inspired by RNA-Seq differential expression analysis, such as pairwise comparisons via volcano plots, feature set enrichment analysis, and pseudotime trajectory inference – enhancing biological interpretability and allowing for mechanistic insight into how specific chromatin features contribute to cell identity. Additionally, O-SNAP classifies cell states with high accuracy by training machine learning models on nuclear SMLM images.

We applied O-SNAP to study chromatin remodeling in broad biological contexts including human tenocyte cells from tendinosis patients, reprogramming human fibroblasts, and mammary epithelial cells with elevated levels of reactive oxygen species. In all cases, O-SNAP identified subtle changes to chromatin spatial organization and classified distinct cell states with >80% accuracy. 

O-SNAP offers a powerful framework to analyze physiological and pathological nuclear architecture transitions, providing mechanistic insights into the underlying chromatin dynamics and potentially enabling rational design of interventions to modulate cell state transitions.

Morgan Kindel, Ryan Post, Louise Lantier, Eusang Hwang, Jamie Carty, Nitsan Goldstein, Hallie Kern, Emily Lo, Kevin Willians, Erik Bloss, J. Nicholas Betley

Regular exercise training is one of the most robust and effective lifestyle interventions for improving health. It is known to induce a network of long-lasting peripheral changes to skeletomuscular, cardiovascular, and metabolic systems. The role of the brain in both responding to and driving these changes is not well understood. Here we show in mice that steroidogenic factor-1 (SF1) expressing neurons in the ventromedial hypothalamus (VMH) play a critical role in exercise-mediated changes. SF1 neurons are activated immediately following an exercise session, and this activity increases over time with more training. This increase in post-exercise activity is associated with increases in intrinsic excitability and synaptic plasticity onto SF1 neurons. Inhibiting this activity is sufficient to prevent the normal, training-induced improvements in endurance capacity. Conversely, optogenetically amplifying this post-exercise activity induces an additional boost in endurance capacity. Overall, these results demonstrate that exercise-induced hypothalamic SF1 neuron activity is essential for the coordination of physiological improvements from exercise training.

Owen Koucky, Matthew Ho, Joe Fraietta

Chimeric antigen receptor (CAR) T cell therapy achieves high initial responses in multiple myeloma (MM) but is not curative and remains ineffective in solid tumors due to poor infiltration and persistence within the tumor microenvironment (TME), including the bone marrow (BM) in MM. The CXCR4-CXCL12 axis regulates immune cell trafficking and retention in BM and tumors; however, CXCR4 is more highly expressed on tumor-infiltrating lymphocytes than on peripheral blood lymphocytes, the source of CAR T cells (CARTs), and is downregulated upon CART activation, impairing residency. We hypothesize enhancing CXCR4 signaling will improve CART trafficking and efficacy. To test this, we engineered “self-driving CARs” by overexpressing gain-of-function CXCR4WHIM variants resistant to receptor internalization. The lead variant, L329Q fs*13, maintained 1.6-fold higher CXCR4 expression and 6.25-fold greater migration than WT, and will be evaluated for BM infiltration and MM control in vivo. Because CXCL12 also recruits immunosuppressive cells to tumors that inhibit CART function, we hypothesize that blocking CXCR4 with Motixafortide will favorably remodel TMEs but could impair CART infiltration. To overcome this, we created Motixafortide-resistant “self-driving CARs” with additional CXCR4D171N/D262N mutations that increase the drug IC50 9.5-fold. These approaches aim to enhance tumor residency and remodel the TME, broadening CART efficacy.

Michael Kuckyr, Omar Banda, Hooda Said, Mohamad-Gabriel Alameh, Beverly L. Davidson

Huntington’s Disease (HD) is a neurodegenerative disorder caused by an expanded CAG trinucleotide repeat in exon 1 of the Huntingtin gene (HTT). As individuals with HD age, they accrue repeats through somatic expansion caused by improper DNA repair of pathogenic DNA motifs. GWAS analysis of large HD patient cohorts have revealed protective CAA mutations within the repeat that delay disease onset, opening opportunities for disease modifying therapies via DNA editing. We hypothesize that this is due to a delay in somatic expansion caused by CAA interruptions. Utilizing cytosine base editing we inserted the protective mutations into pathogenic and non-pathogenic length HTT alleles.  In HEK293 cells we saw greater than 40% editing at any one nucleotide and most commonly inserting 4 repeats simultaneously.  LNP-delivered base editors and gRNAs reduced the rate of expansion in human cell lines that model accelerated rates of somatic expansion. Treated groups had no significant increase in repeat number over the course of 70 days while untreated cells gained 6 repeats within the pathogenic HTT locus. In Q140 mice we observed a reduction of somatic instability. Here we demonstrate the promise of base editing attenuating somatic expansion and, by inference from patient studies, slow HD onset.

Van Le, Krittin Trihemasava, Clemence Queriault, Kelly Rome, Will Bailis

CD8+ T cell functionality and persistence are central to protective immunity and are profoundly shaped by antigen exposure. Following antigen clearance, most CD8 T cells undergo contraction, leaving a long-lived memory population with enhanced metabolic fitness and rapid recall capacity. In contrast, persistent antigen induces CD8+ T cell exhaustion, characterized by mitochondrial dysfunction and elevated reactive oxygen species that limit survival and function. Yet, how CD8+ T cells persist under differing antigenic conditions remains poorly understood, and the metabolic requirements for persistence in the absence or presence of antigen may differ substantially. Here, we show that deletion of Slc25a51, a mitochondrial NAD+ transporter, in activated CD8+ T cells diminishes mitochondrial respiration and enhances glycolysis but minimally affects viability and proliferation at steady state, revealing metabolic vulnerabilities and proliferation defects only upon restimulation. Deletion in naive CD8+ T cells further impairs antigen-driven proliferation, particularly affecting effector expansion. Notably, adoptive transfer of Slc25a51+/- P14 CD8+ T cells into recipients infected with LCMV Clone 13 resulted in reduced persistence and a skew toward terminally exhausted T cells compared to controls. Together, these findings highlight the role of mitochondrial NAD+ in supporting CD8+ T cell proliferation and reveal distinct metabolic requirements for antigen-dependent persistence.

Alexis Leach, Colin Conine

In organisms ranging from C. elegans to mice, paternal environment has been demonstrated to non-genetically influence offspring phenotypes. While at least two classes of small RNAs, microRNAs and tRNA-fragments, have been confirmed to causally transmit environmentally regulated, epigenetically inherited phenotypes to progeny, the extent to which other RNAs (mRNAs, rRNAs, snoRNAs, snRNAs, and long noncoding RNAs) contribute to this transmission has yet to be explored. Due to technical challenges, comprehensive profiling of all RNAs from single sperm samples has been previously thwarted. Consequently, the diversity of sperm RNAs across mammalian species remains unknown.

To capture all RNAs from a single sperm sample and assess their diversity across organisms, we performed five RNA-sequencing techniques (ligation dependent and independent small RNA-seq, ribo-minus and poly-A purified mRNA-seq, and LIDAR-seq) on sperm from 10 mammalian species. We compared expression of known orthologs across a subset of these species, and findings suggest there exist widely conserved and highly expressed mammalian sperm RNAs. Due to their highly conserved nature, these sperm RNAs may be especially important in modulating preimplantation embryonic gene expression. Exploring the diversity of sperm RNAs and their impact on embryonic development will advance our understanding of the mechanisms and evolution of epigenetic inheritance.

Fenglin Li, Aurélia Balestra, Boris Striepen and Yi-Wei Chang

Cryptosporidium parvum is a major cause of diarrheal disease worldwide, particularly affecting children and immunocompromised individuals. Sexual reproduction is essential for parasite transmission, and disrupting fertilization has emerged as a promising strategy to block infection. Understanding the molecular mechanisms underlying this process could reveal new targets for therapeutic intervention.

Recent studies have revealed that male gametes of C. parvum express a complex and specialized fertilization machinery. While components such as the fusogen HAP2 have been identified (Tandel et al., 2019), the overall structure and function of this machinery remain poorly understood. To address this knowledge gap, we aim to use cryo-electron tomography (cryo-ET) to directly visualize the architecture of the male gamete and its fertilization apparatus at molecular resolution.

A critical bottleneck for structural studies is the ability to generate and isolate enough male gametes. To address this, we engineered a new C. parvum strain that integrates three key components: inducible maleness (previously validated in Walzer et al., 2024), a fluorescent male-specific reporter, and a high-production genetic background. Our results demonstrate that this system allows robust induction and enrichment of male gametes suitable for cryo-ET and downstream analyses.

This system paves the way for high-resolution structural studies and functional dissection of the fertilization machinery. Insights gained from this work will advance our understanding of parasite reproduction and may inform the development of transmission-blocking strategies against cryptosporidiosis.

Jialiu Liang, Jae Woo Jung, Kenneth Bedi, Kenneth Margulies, Zoltan Arany

Heart failure (HF) remains a leading cause of death worldwide despite therapeutic advances, underscoring the need for novel therapeutic targets. One promising target is polyamines – small, polycationic metabolites that regulate key cellular processes. Multiple lines of evidence suggest that spermidine, a major polyamine species, may be cardioprotective. However, how polyamines regulate cardiac physiology remains unclear.

In targeted metabolomics of human hearts explanted from patients with end-stage HF, compared to control discarded Gift-of-Life donor hearts, we found substantial elevations in polyamine levels. Bulk RNAseq revealed downregulation of spermidine/spermine N1-acetyltransferase 1 (SAT1), the rate-limiting enzyme for polyamine catabolism. These findings suggest that altered polyamine metabolism may play a previously unrecognized role in HF.

To further investigate this metabolic shift, we are using in vivo isotope tracing to map whole-body polyamine metabolism in conscious mice under both physiologic and HF conditions. In parallel, we are manipulating Sat1 expression in vivo to assess its impact on cardiac function. Given that SAT1 consumes acetyl-CoA, we hypothesize that increased polyamine catabolism exacerbates HF by depleting acetyl-CoA, disrupting cardiac energetics and epigenetic regulation. Together, these studies aim to quantify systemic polyamine metabolism dynamics and elucidate its role in HF, informing the development of novel HF therapeutic strategies.

Abby Lieberman, Jeff L. Gauthier, Timothy A. Machado

Movements emerge from the coordinated activity of distributed neural circuits across the brain. Many previous studies have led to insights about how cortical activity relates to behavior, but these results are difficult to interpret because cortex provides little to no monosynaptic input onto motor neurons in mice. Instead, motor neurons receive most of their inputs from premotor interneuron networks. Given the indirect influence of cortex over motor neurons, how do cortex and premotor circuits work together to drive behavior? Here, we take advantage of the fact that the entire orofacial motor system is in the brain, rather than the spinal cord, and that orofacial motor neurons receive input from premotor circuits in the medulla. This enables us to make simultaneous recordings from both cortical and premotor networks. We recorded from pyramidal tract (PT) neurons across the dorsal cortex using widefield Ca2+ imaging while mice performed a directed multi-spout licking task. Concurrently, we recorded from orofacial premotor networks in the medulla using Neuropixels probes. Preliminary results reveal that during our task, mice make licks that vary in direction, rate, and bout length. We found that cortical PT neuron engagement is stronger during cued pre-reward licks and that PT activity also ramps up with increases in lick rate. Lastly, we show that spike-triggered averaging reveals changes in functional coupling between the cortex and individual medulla units during different kinds of licks, and that different medulla units exhibit different patterns of functional input from the PT. These findings provide insight into how descending cortical commands interact with brainstem premotor circuits to drive orofacial behavior.

Anastasia Lucas, McKenna Reale, Yuri I. Wolf, Bryant Duong, Yichi Zhang, Jayamanna Wickramasinghe, Lindsey Behlman, Steven M. Jones, Stephanie Higgins, Ahmed M. Moustafa, Abdurrahman Elbasir, Ravi Amaravadi, Tara Mitchell, Alexander Huang, Noam Auslander

The gut microbiome has been causally implicated in the efficacy of immune-checkpoint inhibitor therapy (ICI), with an increasing number of clinical trials evaluating microbiome-targeting strategies. However, mechanisms by which gut microbiota elicit immune responses remain unknown and specific taxonomic biomarkers have not been validated in multiple cohorts. Here, we develop an approach to identify function-level determinants of ICI response and immune-related adverse events (irAEs) through quantification of whole-genome metagenomic shotgun sequencing into taxonomy-agnostic functional microbial protein clusters. Using these functional predictors, we identified microbial proteins that are reliably predictive of ICI outcomes across multiple cohorts in the United States. Most notably, we uncovered that above average relative abundance of a novel iron-sulfur functional cluster encoded by different bacteria is predictive of irAEs to ICI. We verified that this iron-sulfur protein cluster is strongly indicative of irAEs in a perspective cohort of melanoma patients treated with ICI at Penn Medicine, which were sequenced through this study. Overall, we show that interrogating metagenomes at the functional level using our taxa-agnostic framework can inform clinical risk stratification and identification of melanoma patients who should be more closely monitored for irAEs and non-response to ICI.

Pengfei Guo#*, Liran Mao#, Yufan Chen, Chin Nien Lee, Angelysia Cardilla, Mingyao Li, Marek Bartosovic*, Yanxiang Deng*

The phenotypic and functional states of a cell are modulated by a complex interactive molecular hierarchy of multiple omics layers, involving the genome, epigenome, transcriptome, proteome, and metabolome. Spatial omics approaches have enabled the capture of information from different molecular layers. However, current technologies are limited to mapping one to two modalities at the same time, which is inadequate to fully understand complex biological systems. Here, we present spatial-Mux-seq, a multi-modal spatial technology that allows simultaneous profiling of at most five different modalities, including genome-wide profiles of two histone modifications and open chromatin, whole transcriptome, and a panel of proteins at cellular level in a spatially resolved manner. We applied spatial-Mux-seq to generate multi-modal tissue maps in mouse embryos and brains, which discriminated refined cell types than unimodal data. We investigated the spatiotemporal relationship between histone modifications, chromatin accessibility, gene and protein expression in neuron differentiation, and were able to identify a radial glia spatial niche that revealed spatially changing gradient of epigenetic signals. Moreover, we discovered previously unappreciated involvement of repressive histone marks in the mouse hippocampus. Collectively, this spatial multi-omics approach heralds a new era for characterizing tissue and cellular heterogeneity, offering unprecedented potential for advanced biological discoveries.

Lauren Elizabeth Lee, Ryo Kawakami MD PhD, Nicholas Forelli MD, Jiten Patel MS, Trace Thome PhD, Yijun Yang MD PhD, Zolt Arany MD PhD

Chronic kidney disease (CKD) affects 37 million US adults and is a leading cause of heart failure (HF). The failing kidney and heart engage in complex bidirectional crosstalk promoting the co-progression of both CKD and HF, underscoring the urgent need for the detailed evaluation of the cardiac-metabolic axis. We induced CKD in M and F 8-week-old C57BL/6N mice with a 0.2% adenine diet for 6–8 weeks. At 4 weeks, mice developed moderate CKD with mild renal fibrosis and elevated injury markers but no cardiac dysfunction. By 8 weeks, severe CKD was marked by pronounced myocardial fibrosis, cardiomyocyte hypertrophy, and diastolic dysfunction. Untargeted metabolomics revealed overlapping and organ-specific metabolic alterations featuring a glycolytic shift and impaired fatty acid oxidation. Preliminary in vivo steady-state ¹³C-labeled metabolite infusion data indicate similar nutrient contributions to TCA cycle intermediates in kidney and heart tissue. Importantly, isolated cardiac mitochondria from CKD mice exhibited ~20% reduction in oxidative phosphorylation and respiration capacity, underscoring mitochondrial dysfunction in cardiac pathology. These findings support the central role of CKD-associated cardiac metabolic remodeling and mitochondrial dysfunction and provide a robust platform for mechanistic studies and therapeutic development.

Joseph P. McGaunn, Haonian Zhou, Joseph A. Baur

Nicotinamide adenine dinucleotide (NAD) is a central coenzyme required for more than 500 enzymatic reactions, and its decline with age is linked to metabolic dysfunction and disease. Although NAD repletion extends healthspan in animal models, clinical trials of NAD-boosting compounds have shown modest benefits, underscoring a major translational barrier: the lack of reliable plasma biomarkers to confirm tissue delivery and functional impact. Evidence suggests that specific subcellular NAD pools may be disproportionately affected during aging, but the compartments most vulnerable—and the enzymes and fluxes most sensitive to their decline—remain unidentified. Existing “sledgehammer” approaches that severely deplete NAD implicate central carbon metabolism, including glycolysis and the pentose phosphate pathway, but they fail to capture the effects of physiologically relevant, partial declines observed with aging. To overcome this gap, we developed a culture model that enables stable, partial NAD depletion through precise nicotinamide titration within a narrow concentration window, without media changes. This system reveals cell-type–specific sensitivity to both moderate depletion and excess supplementation. By modeling age-consistent NAD decline, these studies provide a platform to identify molecular processes and subcellular compartments most impaired by moderate depletion, and to uncover candidate markers of compartment-specific NAD dysfunction as NAD declines.

Rushabh Mehta, Changya Chen, Fatemeh Alikarami, Xiaohuan Qin, Steven Foltz, Avi Loren, Elizabeth Li, Jason Xu, Marilyn M Li, Kathrin M. Bernt, Kai Tan

Survival outcomes for pediatric B-cell acute lymphoblastic leukemia (B-ALL) have dramatically improved, however patients who relapse continue to have poor outcomes. Risk of relapse is highly correlated with the level of minimal residual disease (MRD) after induction therapy, but the mechanisms driving chemoresistance are poorly understood. In this study, we used single-cell multiome sequencing to identify molecular features of treatment-resistant blasts and identify chemoresistance drivers in B-ALL.

We projected blasts onto a healthy bone marrow reference to identify patient-specific developmental states. Blasts displayed increased developmental heterogeneity after treatment towards progenitor-like and mature-like states. These shifts were largely patient-specific and unrelated to MRD level or genetic subtype.  

Our cell state analysis further revealed that nearly half of patients displayed increases in myeloid-like blasts following treatment. Myeloid-like blasts downregulate B cell genes while expressing markers with therapeutic relevance in acute myeloid leukemia. Increases in CD19-low blasts following treatment suggests that induction therapy may select for populations that drive CAR-T therapy failure in patients. 

Overall, our study presents the first comprehensive multiomics characterization of pediatric B-ALL minimal residual disease. Given the increased heterogeneity of blast arrest states following treatment, our findings underscore the need to identify lineage-agnostic targets for high-risk B-ALL frontline therapies.

Emily E Meyer, Lingqi Zhang, Nicolas P Cottaris, David H Brainard, Michael J Arcaro

A hallmark of primate vision is the ability to quickly and efficiently recognize objects despite considerable variations in object viewpoints, sizes, and other features. However, the evolutionary origins of this behavior remain poorly understood. Tree shrews (Tupaia belangeri), among primates’ closest relatives, possess an extensive visual cortex and use visually-guided behaviors which can illuminate the evolutionary origin of primate object recognition abilities.

We trained tree shrews on a match-to-sample task previously used to demonstrate object recognition in humans, macaques, and marmosets. Tree shrews successfully identified objects across variations in position, size, orientation, and scene backgrounds. To probe the neural representations driving their behavior, we compared tree shrew performance with predictions from visual processing models. We first used a front-end visual system model to address species’ acuity differences. We then tested models ranging from V1-like representations to convolutional neural networks and analyzed the correspondence of each model with tree shrew behavior patterns. The models most correlated with tree shrew behavior were deep convolutional layers and global shape representations, suggesting tree shrews may utilize complex, primate-like processing to discriminate objects.

These findings help establish tree shrews as a model for high-level processing and offer insights about the origin of primate object representations.

Natalie Moore, Cindy Lu, Sk Tousif Ahamed, Tejvir Khurana, David Merrick

The global obesity epidemic and recent surge in popularity of GLP-1 agonist drugs has shed light on the effects of whole-body weight loss on skeletal muscle. Fat mass loss is typically accompanied by significant lean mass loss, and many patients are encouraged to exercise to combat this problem, increasing the likelihood of exercise-induced muscle injury. Skeletal muscle is uniquely capable of regeneration after injury, during which muscle progenitor satellite cells (MuSCs) fuse into new myofibers. This process is dependent on fibroadipogenic progenitor cells (FAPs), a tissue-resident fibroblast-like cell that plays a supportive role in maintaining muscle mass even under homeostatic conditions. Preliminary work has identified a distinct transcriptional identity of “activated” FAPs in response to injury and has shown that the magnitude of activation differs depending on the body’s metabolic state. Interestingly, as injury recovery progresses, tissues from weight-loss animals display higher specific force compared to lean or obese counterparts. These data suggest that an imprinted state of catabolism regulates skeletal muscle regeneration after injury, perhaps favoring regeneration of highly efficient fibers and thus modulating resting energy expenditure after weight loss. Future studies will aim to delineate the mechanisms of FAP activation and their role in weight loss skeletal muscle.

Kira J Nightingale, Spandana Makeneni, Stephen Bonnet, Anjali Mahajan, Daniel Teixeira da Silva, Hannah Wall, Sarah Wood

Chlamydia trachomatis is the most common sexually transmitted infection (STI) in the US, with more than 60% of reported cases occurring in individuals aged 15-24. US guidelines recommend directly observed therapy (DOT) for the treatment of chlamydia in adolescents, but almost no research has been done to evaluate whether DOT results in improved outcomes. We conducted a retrospective cohort study to assess the relationship between receipt of DOT and chlamydia retesting within six months. Individuals were included if they were aged 13-25, had an initial chlamydia infection between 01/2013-01/2021 and were treated via DOT/pharmacy, and did not have syphilis/gonorrhea co-infection or pelvic inflammatory disease. Of the 1,970 patients included, 78.2% were female and 95.7% were Black/African American. 60.9% of the DOT group returned for STI retesting within six months, compared to 41.9% of individuals in the pharmacy group. After adjusting for age, sex, race, ethnicity, insurance status, symptomatic infection, and clinic type, individuals in the pharmacy group were 51% less likely than DOT patients to return for resting within six months (OR: 0.49 [0.37-0.65]). Directly observed therapy may be a viable option to increase retesting rates in this population and thus should be widely available at STI treatment sites.

Michael Noji, Christina Demetriadou, Madelyn Landis, Jennifer Pennise, Laura Pinheiro, Alison Jaccard, Adam Chatoff, Jack Drummond, Kevin Guo, Romie Azor, Maggie Robertson, Ryo Kawakami, Mariola Marcinkiewicz, John Tobias, Nathaniel Snyder, David Feldser, Zoltan Arany, Kathryn Wellen

Branched-chain amino acid (BCAA) metabolism is perturbed in patients with pancreatic cancer, but the contribution of systemic or pancreas-intrinsic BCAA catabolism to pancreatic carcinogenesis remains unclear. We show here that pancreas-specific loss of DBT, the E2 subunit of the branched-chain keto-acid dehydrogenase (BCKDH) complex required for BCAA oxidation, strikingly exacerbates premalignant pancreatic intraepithelial neoplasia (PanIN) lesions in KC (p48-Cre;KrasLSL-G12D/+) mice. However, deletion of upstream enzyme BCAT2 neither phenocopied nor rescued loss of DBT in KC mice, ruling out involvement of both upstream and downstream metabolites as mediators of PanIN promotion. Instead, we observed that DBT deficiency led to loss of the kinase BCKDK, a negative regulator of the BCKDH complex, and that, remarkably, pancreas-specific loss of BCKDK phenocopied DBT deficiency in accelerating PanIN formation. These data thus support a model in which pancreas BCKDK restrains tumorigenesis. In contrast, systemic treatment of KC mice with the BCKDK inhibitor BT2, which inhibits BCKDH phosphorylation across many tissues except the pancreas, reduced PanIN formation and preserved normal acinar area. Together the data reveal the promotion of BCAA catabolism systemically, but not within the pancreas, as a promising intervention strategy to suppress tumor initiation.

Corinna Oswell, Sophie Rogers, Justin James, Nora McCall, Alex Hsu, Malaika Mahmood, Gregory Salimando, Charu Ramakrishnan, Karl Deisseroth, Eric Yttri, Gregory Corder

Pain is an unpleasant emotional perception driven by the transformation of sensory neural signals into affective-cognitive information in cortical regions, including the anterior cingulate cortex (ACC). Opioid action in the ACC lessens aspects of the aversive quality of pain through mu opioid receptors (MORs) through both endogenous and exogenous pain relief. We hypothesized that MOR-expressing and functionally nociceptive ACC neurons represent a crucial neural circuit cell-type mediating pain affect. We used miniature microscopes in ACC to demonstrate that neural dynamics in acute and chronic pain are altered, and that exogenous opioids such as morphine selectively reverse such dynamics. To gain genetic access to ACC Oprm1 ensembles, we developed a suite of MOR-promotor driven viruses that induce selective expression in both mouse and macaque tissue. These viruses can be used in conjunction with circuit mapping tools and functional actuators with activity-, molecular-, and projection-dependence. Inhibition of ACC MOR+ neurons using MOR-promotor driven DREADDs mimicked opioid induced analgesia by decreasing affective but not reflexive nociceptive behaviors, as measured by a novel deep-learning system for unbiased pose-estimation of nocifensive behaviors. These approaches use state of the art techniques to improve our understanding of opioid-driven pain relief and offer a novel strategy for precision pain management targeting a key nociceptive cortical circuit.

Divyansi Pandey, Julia Chini PhD, Samuel J. McCright PhD, David A. Hill MD PhD

Obesity is a global epidemic associated with worse outcomes in several comorbidities. Adipose tissue (AT) stores excess energy but can undergo dysfunction leading to global insulin resistance and systemic inflammation. The liver is a critical regulator of AT function during obesity via the secretion of extracellular vesicles (EVs). EVs are nanometer-sized particles that deliver cargo. The mechanisms regulating this liver–AT axis in chronic inflammatory diseases remain poorly defined.

CD9 is a tetraspanin molecule with roles in EV biogenesis and uptake by target cells. We hypothesized that CD9 regulates an EV-mediated liver-AT axis during obesity. We generated liver-specific CD9 knock-out (CD9 LKO) mice and placed them on a high fat diet. Obese CD9 LKO mice exhibited marked AT inflammation, fibrosis, and systemic metabolic derangements. Primary hepatocytes isolated from CD9 LKO mice also had decreased EV production and altered contents compared to control mice suggesting that CD9-dependent EVs are contributing to the metabolic phenotype. 

These findings reveal a previously unrecognized role for liver CD9 in suppressing AT inflammation and metabolic dysfunction in obesity likely through modulation of EV signaling. Ongoing work aims to define liver-derived protective signals regulated by CD9 and to determine their broader impact on obesity-associated inflammatory diseases.

Grace Park, Marshall Lougee, Heejong Kim, Jiya Fowler, Zongtao Lin, Houssein Fazelinia, Lynn Spruce, Evan Yanagawa, Virginia M.-Y. Lee, Benjamin Garcia, Melike Lakadamyali, E. James Petersson

The formation of Lewy Bodies in the substantia nigra is a hallmark of Parkinson's Disease and is characterized by the aggregation of alpha-Synuclein. While a growing body of work has focused on finding neurodegenerative protein-protein interactions to understand the underlying mechanisms of disease, traditional proximity labeling techniques are limited in studying synucleinopathies due to the protein's small size and intrinsic disorder. This work focuses on the development of a next-generation photo-proximity labeling method (uMap) to tag even transient synuclein interactomes with high spatiotemporal control and resolution. Our studies have found uMap to be a robust toolkit that is uniquely suited for both region- and conformation-specific synuclein interactomes without traditional proximity labeling limitations.

Abigail Ridler, Catherine Osorio-Quintero, Paul Gadue

Pancreas agenesis (PA) is a disorder characterized by failure of pancreas development. 60% of all PA cases are due to heterozygous loss of the transcription factor (TF) GATA6 and further studies have confirmed a critical role for GATA6 in the generation of pancreatic progenitors. Individuals harboring GATA6 mutations show incomplete penetrance, presenting with agenesis, diabetes later in life or no phenotype. Murine studies have suggested GATA6 has roles beyond its known requirement in the generation of pancreas progenitors, but how exactly GATA6 loss is influencing these later stages has not been studied in a human model, due to the earlier developmental requirements of GATA6. To investigate the role of GATA6 in later development, we have employed an inducible degron technology in human pluripotent stem cells (hPSCs) to allow degradation of GATA6 at distinct stages of endocrine cell development. We have confirmed known early requirements for GATA6 through the generation of pancreas progenitors. We find GATA6 depletion after the generation of pancreas progenitors leads to a mild defect on specification of endocrine progenitors, but depletion in endocrine progenitors causes up regulation of the critical beta cell TF NKX6.1, which could influence both the specification and/or maturation of insulin producing beta cells.

Julia A. Rocereta, Ruth A. Pumroy, Anastasiia Sadybekov, Xi-Ping Huang, Kensuke Sakamoto, Andreas Leffler, Bryan L. Roth, Vsevolod Katritch, Vera Y. Moiseenkova-Bell

Transient receptor potential (TRP) channels are a diverse family of cation channels involved in numerous physiological processes. They respond to an extensive range of stimuli, including both physical and chemical signals. While many TRP channels have well-characterized, selective ligands that facilitate in-depth investigations of their functions, the pharmacological profile of transient receptor potential vanilloid 2 (TRPV2) remains relatively weak. This has hindered our understanding of TRPV2’s physiological role, despite its association with clinically relevant areas such as neuronal development, immune system regulation, cardiovascular maintenance, and cancer metastasis. We have previously identified several unique ligand binding pockets in rat TRPV2 (rTRPV2) using cryoEM for small molecules with suboptimal potency and selectivity. Here, we have screened these pockets in silico with a hierarchical synthon-based library, V-SYNTHES, to identify novel compounds with improved pharmacological properties at TRPV2. Top scaffolds were then screened in vitro to identify lead agonist candidates. We are using cryoEM of rTRPV2 with these small molecules to identify agonist binding and residue interactions that may be important for TRPV2 selectivity. Altogether, this work is essential for investigations into TRPV2’s physiological roles and therapeutic potentials.

Alexandra Schneider, Colin Conine

Sperm miRNAs are important to development and sufficient for the transmission of phenotypes; however, the molecular mechanism of how sperm miRNAs function postfertilization remains undetermined. Sperm miRNAs likely influence development by downregulating key mRNA targets in embryogenesis. Due to the difficulty of identifying targets in the embryo, the targets of sperm miRNAs are completely unidentified. The Fx-miRs, a cluster of sperm miRNAs, are highly expressed in sperm across mammals, but their postfertilization functions remain unknown. I hypothesize that the Fx-miRs target a subset of the genes critical to early mammalian development. Because of the lack of tools to identify targets in the embryo, I have established mouse embryonic stem cells (mESCs) as a model to study Fx-miR function. When six Fx-miRs are individually overexpressed in mESCs, hundreds of non-overlapping transcripts are dysregulated, including many early embryo transcripts. While my data is consistent with miRNA targeting prediction algorithms, I demonstrate both false negatives and positives, further highlighting the importance of determining targets experimentally. Furthermore, I discovered that overexpression of one Fx-miR causes profound alterations in translation, implying gross impacts on gene expression.  Overall, my work illuminates the molecular mechanisms underlying the regulatory functions of a prominent sperm miRNA cluster in development.

Taylor E. Senay, Xiaomei Li, Sneha G. Shirhattikar, Tiana T. Luo, and Jianxin You

Merkel Cell Polyomavirus (MCPyV) is an oncogenic human polyomavirus latently infecting most adults. Although the causative link between MCPyV and Merkel Cell Carcinoma is well established, the molecular mechanisms governing viral latency and preventing oncogenic progression remain poorly understood. We previously reported that the MCPyV protein ALTO modulates STING-TBK1 signaling, enabling the virus to co-opt host innate immune pathways to suppress viral replication and promote latency over transformation. In this study, we expand this model by identifying a short domain within ALTO, termed LIT (Lost in Tau), which is necessary for ALTO–TBK1 interaction and activation but dispensable for ALTO trafficking and interactions with STING or Src. The LIT domain functions as a dominant negative inhibitor of wild-type ALTO, competitively blocking TBK1 activation through a novel interaction domain. Deletion of the LIT domain abolishes ALTO-TBK1 interaction, downstream phosphorylation, and TBK1-mediated suppression of MCPyV replication during early infection of human dermal fibroblasts (HDFs). These findings highlight the functional importance of intrinsically disordered regions in modulating host–virus interactions and suggest that MCPyV latency is actively maintained through a balance of pro- and anti-viral signaling. Targeting domains such as LIT may offer new strategies for regulating TBK1 activity or disrupting viral persistence.

Junyoung Shin, Anushka Singhal, Marisa A. Jeffries, Caela C. Long, Lindsay K. Festa, Kelly L. Jordan-Sciutto, Judith B. Grinspan

Approximately 30-50% of people with HIV (PWH) are affected by HIV-associated neurocognitive disorders (HAND) that are strongly linked to white matter abnormalities. In this study, primary mixed glial cultures from neonatal HIV-1 transgenic (Tg) rats showed reduced oligodendrocyte precursor cell (OPC) numbers and potentially impaired differentiation into mature OLs compared to wild-type (WT) cultures. While MBP expression remained unchanged, CNP—an essential myelin enzyme—was significantly decreased. We also observed reduced expression of low-density lipoprotein receptor (LDLR), a key mediator of lipid uptake for myelination. Similarly, WT OPCs treated with an antiretroviral therapy (ART) drug combination (tenofovir disoproxil fumarate and emtricitabine) showed decreased LDLR fluorescence intensity. These findings suggest that HIV and ART disrupt OL maturation and myelin lipid homeostasis, potentially contributing to white matter deficits in HAND. Ongoing studies are examining how HIV and ART impair astrocyte-to-OL lipid transport, a critical pathway for myelin synthesis. Understanding these mechanisms could inform therapeutic strategies to improve cognitive outcomes in PWH.

Erin M. Smith, Natali L. Chanaday, Sandra Maday

Lysosomal damage impairs proteostasis and contributes to neurodegenerative diseases, yet cell-type-specific differences in lysosomal repair remain unclear. Here, we use a neuron-astrocyte coculture to compare the response of neurons versus astrocytes to damage induced by LLOMe, a lysosomotropic methyl ester, with a focus on three key pathways: ESCRT-mediated membrane repair, TBC1D15-mediated membrane reformation, and PITT pathway-mediated lipid shuttling. Both neurons and astrocytes showed lysosomal damage, marked by galectin-3 recruitment to lumenal lysosomal beta-galactosides, elevated lysosomal pH, and engagement of lysophagy receptors. Despite lysosomal damage occurring in both cell types, astrocytes showed a preferential recruitment of ESCRT repair machinery (e.g., CHMP2B, CHMP2A, ALIX and IST1) to damaged lysosomes, as compared to neurons. Moreover, enhanced ESCRT recruitment was conserved in astrocytes isolated from different brain regions (e.g., the hippocampus or cortex). Additionally, the lysosomal membrane reformation pathway regulated by TBC1D15, was more robustly activated in astrocytes. By contrast, PITT pathway components (e.g., PI4K2A and ORP9) were activated in both cell types. Our data reveal a divergence in how neurons and astrocytes mobilize repair pathways to manage lysosomal damage. These data may reflect differences in lysosomal resilience between astrocytes and neurons and inform therapeutic strategies to correct lysosomal dysfunction in neurodegenerative diseases.

Sidney Smith, Hannah Loo, PhD, Ayman Rezk, PhD, Gautier Breville, MD, Amit Bar-Or, MD, Jennifer Orthmann-Murphy, MD, PhD

Adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP) is a rare, autosomal-dominant neurodegenerative disorder characterized by progressive psychiatric, motor, and cognitive impairment. Brain pathology includes demyelination, loss of microglia, and axonal spheroids. ALSP is caused by pathogenic variants in CSF1R, which is necessary for myeloid-lineage cell survival. It is not well understood how mutant monocytes and microglia cause CNS damage. Due to heterogeneous symptoms and age of onset, ALSP is often misdiagnosed, can be progressively fatal, and there is no FDA-approved disease modifying therapy. There is an unmet need for early diagnosis and development of treatments. 

To define a distinct immune signature in ALSP monocytes, we use flow cytometry to quantify monocyte phenotype, CSF1R cell surface expression and response to ligands, and compare to healthy control monocytes. We have defined a unique signature in a patient with biallelic CSF1R variants, which I will compare to ALSP samples. To determine whether reduced Csf1r signaling in microglia alters recovery from demyelination, I treated Csf1r-haploinsufficient mice with cuprizone to induce demyelination followed by recovery. Based on this preliminary data, Csf1r-haploinsufficient microglia alter oligodendrocyte recovery from demyelination. Currently, I am defining the reactive state of Csf1r-haploinsufficient mice exposed to cuprizone.

Sarina Smith, Katherine Whiteman, Josh Fuller, Ellie Carrell, Beverly Davidson, Defne Amado

Amyotrophic lateral sclerosis (ALS) is a fatal disease characterized by motor neuron loss, paralysis, and respiratory failure. Nearly all patients exhibit nuclear clearance and cytoplasmic aggregation of TDP-43, which associates with stress granules. Ataxin-2 regulates stress granule dynamics. Our lab previously demonstrated that AAV-mediated delivery of a miRNA targeting Atxn2 (miAtxn2) significantly improves survival and motor function in TDP-43 overexpression mouse models. However, our collaborators in the Davidson lab showed toxicity of some of the construct elements in nonhuman primates, prompting development of another version with a different promoter and backbone. Here, we evaluated the efficacy and safety of miAtxn2 delivered by this new construct. Wildtype (C57BL/6) mice received ICV injections of vehicle, original or new miAtxn2-containing constructs, or our nontargeting control at P1. qPCR revealed ~40% Atxn2 knockdown in the frontal cortex (p=0.0005) and lumbar spine (p=0.0047) with our original construct but only a trend with our new construct. Aif1 and Gfap levels indicated no major signs of neuroinflammation, muscle masses were unchanged, and 13-week rotarod performance did not differ. Future analyses include assessing protein knockdown, later-timepoint behavior, and histology. Taken together, this data supports the safety of these constructs yet indicates limited Atxn2 knockdown with the new construct.

Ebony Smith, Kisha Patel, Donna Gonzalez, Regina Young, Carl June

CAR T cell therapy has had great success in treating B cell leukemia and lymphoma, however, remains ineffective in treating solid tumors. To overcome the limitations of CAR T cell therapy we are evaluating MAITs as an alternative platform to improve cancer immunotherapies targeting solid tumors. MAITs are alpha-beta T cells with a very limited TCR repertoire that is only specific for microbial peptides derived from riboflavin. Thus, MAITs do not mediate alloreactivity warranting their potential use as an allogeneic CAR product. They have a distinct cytokine profile and express various cytotoxic receptors, reducing the risk of toxicity and antigen escape. Moreover, they are more likely to infiltrate solid tumors in the liver and lung due to their natural homing to these tissues. Recently, MAITs have been expanded 200-fold from PBMCs, transduced with the anti-CD19 CAR, and demonstrated to kill CD19 expressing leukemia cells in vitro. However, the anti-tumor efficacy of CAR MAITs targeting solid tumors in vivo is unknown. In this study, we demonstrate that MAITs expressing a CAR specific for mesothelin (M5 CAR) can kill a mesothelin expressing pancreatic tumor cell line, AsPC-1, in vitro and demonstrate limited anti-tumor efficacy in an in vivo AsPC-1 mice model.

Joseph Stucynski, Marc Lubman, Alan de Araujo, Sydney Liu, Guillaume de Lartigue, Franz Weber, Shinjae Chung

Increased sleep in animals during a peripheral immune challenge such as bacterial or viral infection comprises one component of a constellation of behaviors collectively termed ‘sickness’. Body-to-brain communication along neural pathways remains understudied, and in particular the mechanisms through which the immune system interacts with the brain to influence sleep regulation during sickness remain poorly understood. The caudal nucleus of the solitary tract (cNTS) in the brainstem modulates systemic levels of pro-inflammatory cytokines including interleukins such as IL-1B during sickness. The cNTS receives neural inputs from the vagus nerve, which innervates various organ systems of the body and has also been shown to regulate levels of cytokines during sickness. Shown here in mice, IL-1beta increases NREM sleep, fragments it, and suppresses REM sleep, while reducing slow wave delta power. Chemogenetic manipulations of NTS IL1B-activated neurons, as well as DBH-expressing neurons increase sleep and microarousals. Likewise, chemogenetic manipulations of vagal IL1B-activated neurons increase sleep, and vagal ablation reduces the sleep effects of IL1B.

Nehalee Surve, Sunetra Sase, Anjali Bhagavatula, Prabhat Napit, Adeline Vanderver

Microtubules (MTs), composed of α-β tubulin heterodimers, are essential for neuronal function, intracellular trafficking, and myelination. Mutations in tubulin genes cause a range of neurological disorders, including leukodystrophies—rare genetic diseases involving abnormalities in brain white matter. Heterozygous missense mutations in the TUBB4A gene result in TUBB4A-related leukodystrophy (TUBB4A-LD), with a broad clinical spectrum. Our team has defined three novel mouse models across the disease spectrum, including recurring mutations corresponding to three human phenotypes: early infantile encephalopathy (p.Arg262His), the severe Hypomyelination with Atrophy of the Basal Ganglia and Cerebellum (H-ABC, p.Asp249Asn) and a late infantile hypomyelination (p.Arg391His). TUBB4A-LD may impact oligodendrocytes, striatal neurons, and cerebellar granule neurons (CGNs). While CGN loss is a classical hallmark feature of H-ABC, this is less well understood in other variants. Additionally, the impact of the Tubb4aD249N variant on CGN death remains unclear. We will assess CGN loss in the early and late infantile models and use high-resolution imaging and biochemical tools to examine MT dynamics and trafficking across genotypes. Although an existing antisense oligonucleotide therapy mitigates some pathological features, it fails to prevent CGN loss—highlighting the need for more targeted strategies. This work aims to uncover cell-type–specific mechanisms to guide future therapies.

Seth Talyansky, Nadia Dehghani, Edward B. Lee, David J. Irwin, David A. Wolk, Vivianna Van Deerlin, and Corey T. McMillan

Background: Transactive response DNA binding protein of 43 kDa (TDP-43) aggregates are observed in cognitive disorders of aging including frontotemporal lobar degeneration (FTLD-TDP) and limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC). Divergence in the genetic architecture of these disorders remains unclear.

Methods: We identified 391 individuals in the Penn Integrated Neurodegenerative Disease Database (INDD) with a neuropathological diagnosis of FTLD-TDP or LATE-NC, an ABC score of Alzheimer’s disease neuropathologic change (ADNC), and no pathogenic mutation. We compared effect allele frequency of 13 candidate TDP-43-related SNPs between the FTLD-TDP and LATE-NC groups. We validated significant associations in a distinct cohort of 450 individuals from the National Alzheimer’s Coordinating Center (NACC). We tested the association of validated SNPs with average regional TDP-43 severity in neocortex and medial temporal lobe in INDD.

Results: Most LATE-NC cases (78%) showed intermediate/high ADNC. Adjusting for sex and ADNC level, UNC13A-rs12973192-G was more frequent in FTLD-TDP than LATE-NC in INDD and replicated in NACC. UNC13A-rs12973192-G associated with greater neocortical TDP-43 severity.

Conclusions: A variant in UNC13A appears to increase risk of FTLD-TDP relative to LATE-NC. UNC13A may warrant the focus of translational research in all three major TDP-43-opathies: FTLD-TDP, ALS, as well as LATE-NC.

Amelia Taylor, Tanay Parnaik, Stephanie Jones, Vineeth Vengayil, Ph.D., Helen Jordan, Ph.D., Nathaniel Snyder, Ph.D., Caroline Bartman, Ph.D.

Previous research shows that multiple solid tumors, including pancreatic ductal adenocarcinoma (PDAC), exhibit reduced tricarboxylic acid (TCA) cycle flux and ATP production flux compared to healthy tissues. This raises the question of whether a limiting metabolite or cofactor is responsible for this impaired energy metabolism. One candidate is coenzyme A (CoA), a key TCA cycle cofactor derived from the vitamin B5 (pantothenate). Using mass spectrometry-based metabolomics, we demonstrate that PDAC tumors have a >3-fold decrease in the level of total CoA than healthy pancreas despite similar pantothenate levels, suggesting that tumors may have altered CoA metabolism leading to reduced energy production. Therefore, I hypothesize that pancreatic tumor energy metabolism and growth can be suppressed by targeting CoA production through the pantothenate pathway. The work presented herein (1) uncovers the mechanisms underlying reduced CoA levels in PDAC utilizing stable isotopically labelled carbon tracing experiments, and (2) elucidates the dependence of pancreatic cancer on pantothenate for proliferation both in vitro and in vivo through genetic and environmental manipulation.

Haedi Thelen, Wei Yang, Sean Hennessy, Jordana Cohen, Wensheng Guo, Todd Miano

Background: Estimating treatment effects across baseline disease risk scores (DRS) is a common approach for assessing treatment effect heterogeneity, but application to real-world data is challenged by bias from confounding and model overfitting. We propose an inverse probability–weighted split-sample (SS) method for fitting DRS models.

Objective: Compare bias in treatment effect estimates using SS versus propensity score matched-sample (MS) methods.

Methods: We simulated randomized samples (n=3,600) from a population with 12 covariates and confounded treatment–outcome relationships under varying treatment effects (odds ratios [OR] 1, 0.8, 0.5) and treatment–covariate interactions. DRS models were fit using (1) the SS method, with half of comparators withheld after model training, or (2) the MS method, with 1:1 matched treated and comparator subjects. Treatment effects were estimated within DRS strata and percent bias was summarized across scenarios.

Results: Bias increased in the MS method with stronger treatment effects and interactions (overall percent bias [OPB] 17.5% for OR=0.5 with interactions), whereas SS bias remained small (OPB 7.7%). Bias declined more with SS than MS with larger sample sizes, higher outcome prevalence, and greater comparator-to-treated ratios. 

Conclusions: Weighted, split-sample DRS fitting reduces bias and offers advantages for studying treatment effect heterogeneity in real-world data.

Van Truong, Shu Yang, Li Shen, John Wherry, Marylyn Ritchie

Over the past quarter-century, bioinformatics has been repeatedly reshaped by advances in computation, from bespoke command-line tools to scalable cloud platforms. Today, large language models (LLMs) mark the next transformation, offering natural language interfaces, multi-modal data integration, and lightweight reasoning agents. Yet their adoption raises pressing questions of accuracy, transparency, and reproducibility, making it vital to design systems that are both powerful and trustworthy.

My dissertation addresses this challenge through two complementary projects. The first, ToolsyBio, is a retrieval-augmented generation (RAG) system that connects researchers to the rapidly expanding landscape of free and open-source bioinformatics software. Our approach is built on structured registries (bio.tools and EDAM) and a locally served Ollama language model to improve tool discoverability through conversational search. We showcase the merits of fuzzy matching over traditional keyword-based searches.

The second project evaluates the tool-calling capabilities of LLM agents for genomics annotation tasks such as variant-to-position and variant-to-gene mapping. We benchmark over a hundred LLMs on OpenRouter’s meta-gateway. Our results show that tool-boosted LLMs outperform their baseline models when provided with the ability to dynamically retrieve grounded information from well-curated biomedical databases.

Together, these studies demonstrate how LLMs can enhance usability and reliability in biomedical informatics, while providing a framework for developing the next generation of trustworthy AI systems in science.

Eliana von Krusenstiern, Lindsay Festa, Lindsay Roth, Kelly Jordan-Sciutto, Judith Grinspan

Approximately half of people with HIV (PWH) experience HIV-Associated Neurocognitive Disorder (HAND). A persistent pathologic feature of HAND in the post-antiretroviral therapy (ART) era is white matter abnormalities. The duration of ART treatment in patients correlates with thinning of the corpus callosum, suggesting that ART drugs contribute to this pathology. We have found that the ART drug elvitegravir (EVG) prevents the maturation of oligodendrocytes (OLs) and remyelination via activation of the Integrated Stress Response (ISR). During ISR activation, stress granules (SGs) often form, sequestering proteins and mRNAs. Here we show that treating differentiating OLs with the ART drugs bictegravir (BIC) and EVG leads to formation of cytoplasmic SGs. Co-treatment of these ART drugs with an ISR inhibitor and PERK inhibitor prevents SG formation, indicating the SGs form via the PERK activated ISR. Analysis of mice treated with BIC demonstrates SGs within OLs of the corpus callosum; In post-mortem cortical white matter of PWH with HAND, we observed increased SG formation in OLs when compared with neurocognitively normal individuals. These findings suggest that SGs in OLs may contribute to white matter pathology in PWH with HAND, and implicate SGs as a potential therapeutic target for improving outcomes for PWH on ART.

Anna Voss, Omar Bayoumy, Emily Feierman, Roxanne Perez Tremble, Eric Campos, Erica Korb

Histone variants replace canonical histones and are encoded by unique genes with distinct sequences from their canonical counterparts. Recently, the importance of H2A and H3 histone variants in neurodevelopment has become increasingly appreciated. However, the roles of H2B variants in neurodevelopment are poorly understood. H2BE, encoded by H2BC21, is the sole variant of H2B known to be expressed outside of germline tissues. Our lab discovered that H2BE is broadly expressed throughout the brain and is highly localized to promoters. Further, we found that loss of H2BE leads to decreases in chromatin accessibility throughout the genome, changes in neuronal gene expression, and deficits in long-term memory. In collaboration with clinicians, we identified multiple de novo H2BC21 putative germline variants (PGVs) in patients presenting with developmental delays, intellectual disability, short stature, and microcephaly. Utilizing biochemical approaches, CUT&Tag, and ATAC-sequencing, we aim to elucidate how H2BC21 PGVs affect H2BE incorporation, localization within the genome, and chromatin accessibility in human embryonic stem cell-derived neural progenitor cells and neurons. This study is the first to directly implicate a H2B variant histone in neurodevelopmental disorders and is critical to our understanding of how H2BE, and histone variants broadly, alter chromatin regulation and function in the brain.

Patrick J. Walsh*, Elizabeth B. Kraeutler*, Ricardo Linares-Saldana, May Wai, Son C. Nguyen, Shuo Zhang, Parisha P. Shah, Daniel S. Park, Haris A. Muzaffar, Rajan Jain, and Eric F. Joyce§

The nuclear periphery is a universal site of heterochromatin organization, where lamina-associated domains (LADs) play a crucial role in transcriptional repression and genome stability. Yet the molecular mechanisms that tether LADs in human cells have remained elusive. Here, we overcome this barrier by performing the first genome-wide, imaging-based siRNA screen of LAD positioning, identifying more than 100 genes required for perinuclear localization. We isolate a striking enrichment of RNA-binding proteins, revealing an unanticipated layer of regulation. Among these, hnRNPK emerged as a central factor, required for the positioning of nearly two-thirds of LADs in the human genome. Its loss drives large-scale LAD detachment from the nuclear periphery without altering heterochromatin states, but induces transcriptional misregulation within these domains. Strikingly, hnRNPK-sensitive LADs are uniquely enriched for H3K27me3, distinguishing them from hnRNPK-insensitive LADs enriched in H3K9me3. Together, these findings uncover a pivotal role for hnRNPK in spatial genome organization, highlight the broader diversity of LAD localization mechanisms, and establish a framework for systematically mapping genetic control of nuclear architecture.

Alice Wang, Luca Paruzzo, Martina Bonomi, Chia-hui Chen, Ivan Cohen, John DiMaio, Kyung Jin Ahn, Peter Michener,  Jaryse Harris, Arati Inamdar, Pooja Devi, Emeline Chong,  Siddarth Bhattacharyya, Vrutti Patel, Patrizia Porazzi, Joseph Fraietta, Stephen Schuster, Fabio Luciani, Marco Ruella, Kai Tan

CD19-directed CART (CART19) cell therapy is standard of care for relapsed/refractory large B-cell lymphomas (r/r LBCL), but only 33% of patients maintain remission beyond 5 years. Resistance to CART19 can arise from antigen-negative escape or CART cell dysfunction, while emerging data shows myeloid cells can shape CART functionality. To further understand longitudinal interactions between myeloid and CART cells, we performed single-cell CITE and TCR sequencing on the peripheral blood of 8 complete responders (CR) and 7 patients with progressive disease (PD) after CART19. In total, 65,618 CART and 298,855 non-CAR cells were analyzed.

After infusion, CR patients saw significant expansion of CD8+ effector memory (EM) CART cells, which displayed increased tumor necrosis factor signaling. Conversely, CD8+ EM CART cells in PD patients exhibited increased interferon (IFN) signaling. Cell fate analysis also revealed that IFN genes drive the evolution of CD8+ EM CART cells in PD patients. Interestingly, a similar IFN signature was also detected in myeloid cells from PD patients. Finally, cell-cell communication analysis revealed IFNg signaling increased downstream IFNg driven interactions (HLA-DR/LAG3) between monocytes and CD8+ EM CART cells. These findings suggest a dysfunctional IFN-mediated cross-talk between myeloid and CAR T cells, correlating myeloid-driven IFN signaling to CART19 resistance.

Morgan Watson BSPH, Kechna Cadet PhD, Gabrielle Zuckerman BSPH, Elizabeth Nesoff PhD MPH

Housing insecurity, including homelessness and eviction, is linked to higher fatal overdose (OD) risk. However, the relationship between rent burden, spending 30% or more of income on rent, and fatal opioid-involved OD is unclear. This cross-sectional ecological study analyzed Chicago census tracts from 2015-2023. Fatal opioid-involved ODs were identified using toxicology reports from the Cook County Medical Examiner. Rent burden was derived from the American Community Survey. Across study years, there were between 24-27% rent-burdened households and 7,926 total fatal opioid-involved ODs. Mapping and Moran’s I plots assessed spatial patterns. Multivariable negative binomial models estimated associations between rent burden and fatal opioid-involved ODs for each year, controlling for neighborhood variables (e.g., poverty, segregation, population density). Moran’s I plots indicated positive spatial autocorrelation with clustered values of high rent burden and OD. For all years except 2022, each 10% increase in neighborhood rent burden was significantly associated with increased fatal OD rates. For example, in 2023, each 10% increase in rent burden was associated with a 22% increase in OD rate (aIRR = 1.22, 95% CI [1.09, 1.36]). Neighborhoods with high proportions of rent-burdened households experience higher rates of OD. Upstream housing interventions, such as rental assistance, may mitigate fatal OD.

Emma Welter, Natalie Toothacre, Claudia Lovell, Marisa Bartolomei, Montserrat Anguera

Most patients with autoimmune disease are female, with hormones and genetics contributing to this sex bias. Autoimmune disease patients aberrantly express X-linked genes, and B cells from patients with the female-biased autoimmune disease lupus exhibit alterations with X-Chromosome Inactivation (XCI). XCI silences most of one X in XX females via enrichment of repressive epigenetic marks including Xist RNA, heterochromatic histone marks, and DNA methylation (DNAm). My project investigates the role of DNAm in XCI maintenance in B cells, wherein epigenetic modifications dynamically localize to the inactive X (Xi) following cellular activation, and how lupus-like disease alters the epigenome. DNAm is highly enriched on the Xi in both naïve and stimulated B cells and dynamically changes at select Xi regions preceding the first cell division. Lupus-like disease in mice alters DNAm on and gene expression from the Xi in B cells. Using ex vivo deletion, I am investigating the role of DNA methyltransferases (DNMT1 and DNMT3a/3b) for regulating DNAm levels on the Xi and how gene expression is affected. Understanding the regulatory mechanisms for DNAm enrichment across the Xi and how lupus disease alters DNAm levels and Xi gene expression will reveal sex-specific pathways underlying female biased autoimmune disease.

Jakob Woerner*, Thomas Westbrook*, Jaehyun Joo, Manu Shivakumar, Rasika Venkatesh, Tess Cherlin,  Damian Maseda,  Michelle McKeague, Shwetank, Matei Ionita, Joost Wagenaar, Sarah A. Abramowitz, Anurag Verma, Bingxin Zhao, Seunggeun Lee,  Scott Damrauer, Michael G. Levin, Thomas P. Cappola, Daniel J. Rader, Sharlene Day, Rajat Deo, Joel M. Gelfand, Ravi Ramessur, Marie Guerraty, Shefali Setia-Verma, Bogdan Pasaniuc, Marylyn D. Ritchie, Sokratis A. Apostolidis, Allison R. Greenplate, E. John Wherry, Penn Medicine Biobank, Yonghyun Nam, Dokyoon Kim

Polygenic risk scores (PRS) and proteomic risk scores (ProRS) have independently demonstrated predictive utility for complex diseases, but their combined value across many diseases remains underexplored. Integrating genetic and 2,920 plasma protein measurements in 39,843 UK Biobank participants, we computed PRS and ProRS for 301 diverse diseases. Models trained on prevalent cases and controls were evaluated for their ability to predict incident disease. 
ProRS improved prediction of future disease onset over PRS in 95% of traits, especially for conditions with lower heritability. Combined modalities (PRS+ProRS) achieved the highest accuracy and improved risk stratification, especially in those at the highest risk. In proportional hazards models, PRS explained >20% of the relative variation in eight autoimmune diseases. While ProRS provided limited added value for most cancer outcomes, genetics were significant in predicting incidence. Although ProRS performance wanes with time since blood draw, it can still identify at-risk individuals a decade before disease onset.

Our findings support leveraging both static genetic and dynamic proteomic markers for the care and screening of complex diseases. While previous studies have questioned whether PRS provides any additional information beyond protein risk scores, we showed that PRS further stratifies risk even among individuals in the highest ProRS strata.

David Wei Wu, Feng Wang, Quan Sun, Xinjun Ji, Stacy Woyciechowski, Robert Wang, Joseph Park, Ryan Park, William Gaynor, Yi Xing

Splicing variants represent 15-60% of disease variants, but they are poorly cataloged. A key driver of this gap is the use of short-read RNA sequencing (srRNA-seq), which incompletely characterizes splicing. Long-read RNA sequencing (lrRNA-seq) overcomes the limitations of srRNA-seq, but current lrRNA-seq studies are limited in scale (n ≤ 67, reads < 13 million/sample) and profile healthy cohorts.

To address this limitation, we generated whole genome sequencing, clinical phenotyping, and whole-transcriptome Nanopore lrRNA-seq data on 91 patients from the Children’s Hospital of Philadelphia Birth Defects Biorepository. With a median of 27.8 million lrRNA-seq reads/sample, our cohort is one of the largest, most deeply sequenced lrRNA-seq studies to date and the only one to profile birth defects. We use this dataset to discover splicing variants that explain complex disease risk and cause Mendelian conditions, for which our cohort is enriched.

To uncover common splicing variants, we performed splicing QTL mapping, discovering 336,800 sQTLs across 5,930 genes at 0.05 FDR. 43,655 (13.0%) of sQTLs were not significant in any GTEx tissue, despite GTEx having ~10X more samples. Colocalization with GWAS Catalog and FinnGen data revealed 1,728 associations (H4 PP ≥ 0.8) across 207 complex diseases. Importantly, we find a novel CD2AP sQTL, rs10676828 (FDR = 4.2×10-11), that activates an unannotated transposon-derived poison exon and increases the risk for Alzheimer's disease (H4 PP = 0.91). We also find two Neanderthal-introgressed variants, rs5743596 and rs5743566, that together cause skipping of TLR1 exons 2 and 3 (FDR = 1.2×10-16) and protect against spirochete infections (H4 PP = 0.9). 

Lastly, we developed a pipeline to discover rare splicing variants that cause Mendelian diseases. Our method finds patient splicing outliers, links them with causal mutations, and uses machine learning to prioritize mutations that explain the patient’s phenotypes. Using this approach, we find Mendelian diagnoses for two undiagnosed patients. In one patient, we made a new diagnosis of TARP Syndrome by identifying and confirming with a minigene assay that a synonymous RBM10 VUS causes pathogenic intron retention.

In summary, we establish a comprehensive lrRNA-seq resource and use it to reveal novel splicing variants that contribute to complex and Mendelian diseases.

Jiazheng Xing, Qing Tang, Siewert Hugelier, Melike Lakadamyali

To meet the local energy demand at different locations within a cell, mitochondria are transported by microtubule motors kinesin and dynein-dynactin. Previous studies have confirmed mitochondrial outer membrane protein Miro and microtubule motor adaptor protein TRAK to play critical role in the transportation of mitochondria, though the spatiotemporal organization of these proteins as an adaptor complex remains to be investigated. Using live cell single-particle tracking, we visualized the heterogeneous dynamics of Miro and TRAK on mitochondria, in either a lateral diffusing or a confined state. Characterization and analysis of these motility states could provide information on the Miro-TRAK motor adaptor complex, and how it regulates the transportation of mitochondria.

Lingke Zhong, Sukanya Madhwal, Amita Sehgal

Circadian rhythms are internal timekeepers that regulate various physiological and behavioral processes. With aging, circadian rhythms are dampened and advanced, leading to fragmented sleep and earlier activity schedules. These changes not only reduce the quality of life but also increase susceptibility to cognitive decline, chronic diseases, and weakened immune systems. At the molecular level, circadian rhythms are maintained by a transcriptional-translational feedback loop of four core clock genes: CLOCK, BMAL, CRY, and PER, which also regulate thousands of clock-controlled genes across peripheral tissues. Circadian gene expression in peripheral tissues dampens or phase-shifts with age, paralleling the phenotypes observed in the elderly. Our previous work showed that human age-dependent blood serum alters rhythmicity of the clock-controlled gene expression in cultured fibroblasts. Nevertheless, the responsible serum factors are yet to be identified. 

We hypothesize that age-dependent serum factors affect circadian rhythms in peripheral cells, driving aging-related impairment. Here, we established a mouse model that recapitulates age-related circadian decline observed in humans. Using this system, I identified young blood-enriched serum factors that restore rhythmicity of clock-controlled genes in periphery. These findings lay the foundation for developing therapeutic strategies to mitigate age-related circadian decline.

Chaoting Zhou, Jing Yu Carolina Cen Feng, Shan Liu, Katherine S. Ventre5, Jordan E. Axelrad, Kyung Ku Jang, Ken Cadwell

Intestinal stem cells (ISCs) mediate the continuous renewal of the epithelium during homeostasis and recovery from injury. To investigate the molecular impact of inflammation on human ISCs, we established an organoid model derived from individuals with pouchitis, a form of inflammatory bowel disease that develops in an organ generated from ileal pouch-anal anastomosis (IPAA) surgery. Compared with non-inflamed specimens, pouchitis organoids exhibited increased apoptosis and secretory lineage differentiation that were stable after multiple passages, mirroring findings in primary pouch tissue. Chromatin accessibility and histone modification profiling of ISCs revealed inflammation-driven epigenetic remodeling, particularly involving AP-1 transcription factors like c-Jun and excessive STAT1 activity. Loss of c-Jun disrupted ISC viability and enhanced secretory cell differentiation in non-inflamed organoids, while therapies targeting JAK/STAT reversed these ISC defects in pouchitis organoids. These results associate inflammation with epigenetic changes in human ISCs, suggesting that immune-mediated injury has lasting effects on the pouch epithelium that can potentially be reversed through available therapies.

Andrew Zolensky, Kuk Jang, Janice Sabin, Andrea Hartzler, Basam Alasaly, Sriharsha Mopidevi, Mark Liberman, Kevin Johnson

Patient-clinician communication research is crucial for understanding interaction dynamics and for predicting outcomes that are associated with clinical discourse. Traditionally, interaction analysis is conducted manually because of challenges such as Speaker Role Identification (SRI), which must reliably differentiate between doctors, medical assistants, patients, and other caregivers in the same room. Although automatic speech recognition with diarization can efficiently create a transcript with separate labels for each speaker, these systems are not able to assign roles to each person in the interaction. Previous SRI studies in task-oriented scenarios have directly predicted roles using linguistic features, bypassing diarization. However, to our knowledge nobody has investigated SRI in clinical settings. We explored whether Large Language Models (LLMs) such as BERT could accurately identify speaker roles in clinical transcripts, with and without diarization. We used veridical turn segmentation and diarization identifiers, fine-tuning each model at varying levels of identifier corruption to assess impact on performance. Our results demonstrate that BERT achieves high performance with linguistic signals alone (82% accuracy/82% F1-score), while incorporating accurate diarization identifiers further enhances accuracy (95%/95%). We conclude that fine-tuned LLMs are effective tools for SRI in clinical settings.