Faculty Bibliography

Abnormal JAK/STAT pathway activation is widespread in virtually all T-cell lymphoma (TCL) subtypes. However, activating mutations are insufficient to drive leukemic cell proliferation, which also requires enhanced upstream signaling. We have described that thyroid hormones (THs) contribute to the malignant phenotype of TCL by inducing intracellular transcriptional programs through integrin αvβ3 activation. Here, we evaluate the effect of THs on the JAK/STAT pathway and its implications on TCL therapy. We found that THs induce the activation of STAT1, 3, and 5, including the upregulation of target genes and metalloprotease activity. Furthermore, we observed that the integrin αvβ3 inhibitor, cilengitide, not only reverts these effects but also enhances the antilymphoma activity to a greater extent than the JAK1/2 inhibitor, ruxolitinib, when combined with bexarotene, a synthetic rexinoid clinically used for cutaneous TCL treatment. Furthermore, we explored the mechanisms of action of cilengitide and bexarotene combination using preclinical TCL in vivo models and proteomic analysis. We found that this combinatorial protocol significantly reduced tumor STATs phosphorylation, matrix metalloproteinase activity, and the number of metastatic foci by regulating proteins involved in cell proliferation, angiogenesis, metabolism, and immune response. In addition, we observed that high integrin αvβ3 messenger RNA levels are enriched in pathways associated with lymphoma progression and reduce overall survival in samples from patients with TCL. Our findings support the therapeutic potential of targeting THs signaling through integrin αvβ3 inhibition in combination with bexarotene as a less toxic therapeutic strategy to mitigate aberrant JAK/STAT activation and limit lymphoma dissemination.

Breast cancer has emerged as the leading cause of cancer death in women worldwide. The high recurrence and metastasis rates of malignant tumors impose significant limitations on existing mainstream treatments, including surgery, chemotherapy, and radiotherapy. Photodynamic therapy (PDT) is a clinically validated approach for cancer treatment. PDT requires three elements, photosensitizer, light, and oxygen, and mainly relies on the production of singlet oxygen (¹O₂) to elicit damage to the cancer tissue. In this study, we explored targeted photodynamic elimination of breast cancer cells overexpressing human epidermal growth factor receptor 2 (HER2). HER2 is enriched on the surface of certain cancer cells and targeted by commercially available monoclonal antibodies, including Trastuzumab, in the treatment of breast and stomach cancers. We engineered chimeric fusion proteins composed of Trastuzumab and genetically encoded photosensitizers, including SOPP3 and miniSOG. The production of ¹O₂ by these fusion proteins was directly measured by near-infrared spectroscopy centered at 1270 nm and further evaluated in the assay of targeted photodynamic neutralizations of SARS-CoV-2 pseudoviruses. To enhance the internalization of the antibody-photosensitizer fusion protein, cell-penetrating peptides (CPPs) were added to the fusion protein. HER2-positive (HER2⁺) cancer cells were incubated with the antibody-photosensitizer fusion protein and then exposed to light illumination. Cell viability assays revealed an over 50% reduction in cancer cell survival, with minimal impacts on the cells from the control group. In addition, we observed a long-lasting, over 24-h inhibition of the growth of the cancer cells after photodynamic treatment. Thus, based on these assays at the molecular and cellular levels, this study established a targeted photodynamic approach that can potentially be developed as an effective PDT for cancer treatment.

Peripheral T cell lymphomas (PTCLs) comprise heterogeneous malignancies with limited therapeutic options. To uncover targetable vulnerabilities, we generate a collection of PTCL patient-derived tumor xenografts (PDXs) retaining histomorphology and molecular donor-tumor features over serial xenografting. PDX demonstrates remarkable heterogeneity, complex intratumor architecture, and stepwise trajectories mimicking primary evolutions. Combining functional transcriptional stratification and multiparametric imaging, we identify four distinct PTCL microenvironment subtypes with prognostic value. Mechanistically, we discover a subset of PTCLs expressing Epstein-Barr virus-specific T cell receptors and uncover the capacity of cancer-associated fibroblasts of counteracting treatments. PDXs' pre-clinical testing captures individual vulnerabilities, mirrors donor patients' clinical responses, and defines effective patient-tailored treatments. Ultimately, we assess the efficacy of CD5KO- and CD30- Chimeric Antigen Receptor T Cells (CD5KO-CART and CD30_CART, respectively), demonstrating their therapeutic potential and the synergistic role of immune checkpoint inhibitors for PTCL treatment. This repository represents a resource for discovering and validating intrinsic and extrinsic factors and improving the selection of drugs/combinations and immune-based therapies.

Large language models (LLMs) are routinely used by physicians and patients for medical advice, yet their clinical safety profiles remain poorly characterized. We present NOHARM (Numerous Options Harm Assessment for Risk in Medicine), a benchmark using 100 real primary care-to-specialist consultation cases to measure frequency and severity of harm from LLM-generated medical recommendations. NOHARM covers 10 specialties, with 12,747 expert annotations for 4,249 clinical management options. Across 31 LLMs, potential for severe harm from LLM recommendations occurs in up to 22.2% (95% CI 21.6-22.8%) of cases, with harm of omission accounting for 76.6% (95% CI 76.4-76.8%) of errors. Safety performance is only moderately correlated (r = 0.61-0.64) with existing AI and medical knowledge benchmarks. The best models outperform generalist physicians on safety (mean difference 9.7%, 95% CI 7.0-12.5%), and a diverse multi-agent approach improves safety compared to solo models (mean difference 8.0%, 95% CI 4.0-12.1%). Therefore, despite strong performance on existing evaluations, widely used AI models can produce severely harmful medical advice at nontrivial rates, underscoring clinical safety as a distinct performance dimension necessitating explicit measurement.

The mechanisms utilized by differentiating B cells to withstand highly damaging conditions generated during severe infections, like the massive hemolysis that accompanies malaria, are poorly understood. Here, we demonstrate that ROCK1 regulates B cell differentiation in hostile environments replete with pathogen-associated molecular patterns (PAMPs) and high levels of heme by controlling 2 key heme-regulated molecules, BACH2 and heme-regulated eIF2α kinase (HRI). ROCK1 phosphorylates BACH2 and protects it from heme-driven degradation. As B cells differentiate, furthermore, ROCK1 restrains their pro-inflammatory potential and helps them handle the heightened stress imparted by the presence of PAMPs and heme by controlling HRI, a key regulator of the integrated stress response and cytosolic proteotoxicity. ROCK1 controls the interplay of HRI with HSP90 and limits the recruitment of HRI and HSP90 to unique p62/SQSTM1 complexes that also contain critical kinases like mTOR complex 1 and TBK1, and proteins involved in RNA metabolism, oxidative damage, and proteostasis like TDP-43. Thus, ROCK1 helps B cells cope with intense pathogen-driven destruction by coordinating the activity of key controllers of B cell differentiation and stress responses. These ROCK1-dependent mechanisms may be widely employed by cells to handle severe environmental stresses, and these findings may be relevant for immune-mediated and age-related neurodegenerative disorders.

Scale invariance is a ubiquitous observation in the dynamics of large distributed complex systems. The computation of its scaling exponents, which provide clues on its origin, is often hampered by the limited available sampling data, making an appropriate mathematical description a challenge. This work investigates the behavior of correlation functions in fractal systems under conditions of severe subsampling. Analytical and numerical results reveal a striking robustness: the correlation functions continue to capture the expected scaling exponents despite substantial data reduction. This behavior is demonstrated numerically for the random 2D Cantor set and the Sierpinski gasket, both consistent with exact analytical predictions. Similar robustness is observed in 1D time series both synthetic and experimental, as well as in high resolution images of a neuronal structure. Overall, these findings are broadly relevant for the structural characterization of biological systems under realistic sampling constraints.

Traumatic injury is one of the leading causes of emergent hospital evaluations. Specifically, blunt bowel and mesenteric injury (BBMI) account for 1% to 5% of abdominal traumas with a high morbidity and mortality, as clinical signs and nonspecific imaging findings make the initial diagnosis challenging. Understanding key imaging findings and the clinical symptoms can increase the radiologist's suspicion for BBMI and ultimately improve patient outcomes.