While the addition of cyclin-dependent kinases 4 and 6 inhibitors (CDK4/6i) to endocrine therapy has improved disease control in advanced estrogen receptor-positive (ER+) breast cancer, acquired CDK4/6i resistance limits long-term efficacy. During CDK4/6i treatment, cancer cells can regain proliferative capacity by increasing responsiveness to growth factors, particularly ErbB signaling, to bypass G1/S checkpoint control. To investigate the molecular dynamics of cell cycle resistance, we performed temporal RNA-sequencing on ribociclib-sensitive and -resistant ER+ breast cancer cells during treatment with ribociclib monotherapy or in combination with the pan-ErbB, afatinib. Using generalized additive models, we characterized dynamic phenotypic responses to treatment unique to ribociclib-resistant cells, identifying cyclin E/CDK2 complex activation and E2F transactivation as key drivers of G1/S checkpoint bypass. This was accompanied by compensatory cell cycle reactivation and cell death pathway suppression in monotherapy-treated cells. In contrast, combination therapy of ribociclib plus afatinib synergistically induced early and sustained cell cycle arrest, followed by increased apoptosis in both ribociclib-sensitive and -resistant cells. Addition of afatinib restored G1/S checkpoint regulation by suppressing cyclin E/CDK2 complex, enforced G2/M checkpoint control by inhibiting cyclin B/CDK1 complex, and triggered caspase-dependent apoptosis. These findings identify the ErbB pathway as central to ribociclib resistance and support combined CDK4/6 and ErbB inhibition as a strategy to restore and maintain CDK4/6i sensitivity in ER+ breast cancer.
Other authors: Jason I. Griffiths, Eric F. Medina, Elena Farmaki, Aritro Nath, Andrea H. Bild
Our laboratory has previously discovered that metastasizing melanoma cells experience high levels of oxidative stress that limit metastasis. Here, we tested the hypothesis that human osteosarcoma metastasis is limited by oxidative stress and ferroptosis. We obtained human patient-derived osteosarcoma cell lines, HOS-MNNG, LM7, SJSA1, Hu09-H3 and MG63.2, tagged with stably expressed luciferase and orthotopically transplanted into immunocompromised NOD-SCID-Il2rg-/- (NSG) mice. The primary tumors that formed in the tibia spontaneously metastasized to distant sites such as the lung, liver, and kidneys. Using flow cytometry and mass spectrometry, we found that compared to primary tumors, circulating cancer cells and nascent metastatic nodules experienced higher levels of oxidative stress as indicated by elevated reactive oxygen species and depleted glutathione. Treatment of the mice with antioxidants, N-acetyl L-Cysteine (NAC) or dietary supplantation of vitamin E increased metastatic disease burden without affecting primary tumor growth. To determine whether ferroptosis limits the survival of osteosarcoma cells in the blood, we pretreated osteosarcoma cells with ferroptosis inhibitors, Liproxstatin-1 or NAC, and intravenously injected them into NSG mice. Pretreatment with Liproxstatin-1 or NAC, as compared to vehicle, significantly increased metastatic disease burden. Likewise, CRISPR-Cas9 mediated deletion of the Glutathione peroxidase 4 (GPX4) that protects cells from ferroptosis reduced spontaneous metastasis. These results suggest that osteosarcoma metastasis is limited by oxidative stress-induced ferroptosis. To gain insight into metabolic regulators of osteosarcoma metastasis, we compared primary tumors and nascent metastatic nodules by metabolomics. Notably, carnosine and its derivative anserine were approximately 200-fold depleted in nascent metastatic nodules as compared to primary tumors. Supplementation of carnosine in drinking water elevated carnosine levels in the nascent metastatic nodules and increased spontaneous metastasis. We are currently studying the mechanisms by which carnosine promotes metastasis.
Other authors: Derek A. Santiago-Ferrer, Daniel L. Cassidy, Amanda R. Reyes, Sean J. Morrison
Intratumoral heterogeneity is emerging as a key property of cancer cells. Recent advances in single-cell transcriptomics have led us to identify specific cellular programs being utilized by the malignant and non-malignant cells in the tumor. However, the origin and adaptive consequences of cancer cell states as well as the states of the tumor microenvironment (TME) component are not well understood. A model to explain two aspects of tumorigenesis: the tumor heterogeneity and the TME, is needed. By co-culturing cancer cells with macrophages or fibroblasts over 36 weeks, we observed gradual phenotypic changes in cancer cell states that induce changes in the states of the TME cell types towards a pro-tumoral environment. Single-cell RNA-sequencing revealed the induction of gene modules, including stress responses in macrophage-adapted cells and interferon responses in fibroblast-adapted cells, which led to larger tumors or increased drug resistance, respectively. Using spatial transcriptomics we test these interactions in patient-derived tumors. Our results suggest that the heterogeneity of cancer cell states is actively maintained through paracrine signaling with the local microenvironment cells, leading to the observed heterogeneous pattern. These findings support a model where cancer cell-TME interactions influence the fate of cancer cells, including metastasis and drug response.
Other authors: Itai Yanai
Alterations of the extracellular matrix (ECM), including both mechanical (such as stiffening of the ECM) and chemical (such as variation of adhesion proteins and deposition of hyaluronic acid (HA)) changes, in malignant tissues have been shown to mediate tumor progression. To survey how cells from different tissue types respond to various changes in ECM mechanics and composition, we measured physical characteristics (adherent area, shape, cell stiffness, and cell speed) of 25 cancer and 5 non-tumorigenic cell lines on 7 different substrate conditions. Our results indicate substantial heterogeneity in how cell mechanics changes within and across tissue types in response to mechanosensitive and chemosensitive changes in ECM. The analysis also underscores the role of HA in ECM with some cell lines showing changes in cell mechanics in response to presence of HA in soft substrate that are similar to those observed on stiff substrate. Lastly, using unsupervised machine learning, we identify phenotypic classes that characterize the physical plasticity, i.e. the distribution of physical feature values attainable, of a particular cell type in response to different ECM-based conditions.
Other authors: Jonathan Nukpezah, Paul A. Janmey
Multiphoton microscopy is a powerful imaging technique widely used in biomedical research due to its deep tissue penetration, intrinsic optical sectioning, and broad excitation range. However, its conventional point-by-point scanning approach leads to slow frame rates, and the need for precise focusing with bulky optics limits its suitability for in vivo live imaging. We present a video-rate two-photon micro-endoscopy system that overcomes these limitations using a miniature, fiber-scanning distal probe. The probe integrates a custom-designed micro-objective and operates the fiber scanner at its second harmonic resonance mode to achieve high-speed imaging. With a diameter of just 2.5 mm, the probe delivers a lateral resolution of 1.2 μm and a 125 μm-diameter field of view. It enables real-time in vivo imaging at 20 frames per second, capturing both stained cells and label-free contrast from intrinsic fluorescence and second harmonic generation. To reduce cost and streamline fabrication, we developed 3D-printed mounts for the fiber scanner that maintain optical performance and mechanical stability. This compact, high-resolution system provides a practical platform for real-time cellular imaging in live subjects and paves the way for broader adoption of multiphoton techniques in translational and clinical research.
Other authors: Guillaume Ducourthial, Tarash Sharan, Arvind Mohan, Jiasen Hou, Sudip Timilsina, Rongguang Liang, Frederic Louradour, and Bryan Spring
Accurate quantification and spatial localization of small-molecule metabolites at the single-cell level remains a major challenge in cancer research. Standard techniques like LC/MS or isotope tracing offer only relative quantification or are too costly for exploratory imaging, particularly for resolving metabolites in complex biological samples. To address this, we present a novel, label-free imaging platform—Null-Deflection Infrared Scanning Probe Microscopy (NDIR). This method integrates infrared spectroscopy with atomic force microscopy, enabling high-resolution, sub-cellular topographic and chemical imaging. Unlike conventional infrared or mass spectrometry-based methods, NDIR allows direct, quantitative visualization of biomolecules in situ, with nanometer-scale precision and without the need for labeling or destructive sample prep. We hypothesize that coupling NDIR with optimized sample preparation protocols can enable accurate metabolic mapping. Thus far, we have established a robust, reproducible workflow and successfully localized 2-deoxyglucose-6-phosphate (2-DG-6-P)—a key metabolite of 2-deoxyglucose metabolism—in U87-MG glioblastoma cells, along with other cell and spheroid models. To support metabolite quantification, we developed a spectral calibration library using a custom 3D-printed microfluidic system for precise, reproducible deposition of metabolite standards. This library, combined with NDIR’s spatial precision, supports the development of machine learning–based tools for spectral deconvolution. Our goal is to build detailed chemical maps of cancer cells to reveal spatial distributions of key metabolites, opening new pathways for understanding tumor metabolism and heterogeneity.
Other authors: Seth Kenkel, Anirudha Rao, Brendan Harley, Rohit Bhargava
Background: Circulating tumor DNA (ctDNA) has emerged as a promising biomarker for real-time monitoring of cancer progression and treatment response in HPV-associated anal squamous cell carcinoma (ASCC). With 80-90% of ASCC cases linked to HPV infection, ctDNA demonstrates high sensitivity in detecting disease dynamics, often identifying progression earlier than conventional imaging while enabling frequent longitudinal assessment that correlates with overall tumor burden.
Methods: We analyzed longitudinal data from 32 HPV-associated ASCC patients receiving immunotherapy every 3 weeks for up to 2 years. We developed a patient-specific mathematical model integrating tumor volume dynamics with ctDNA kinetics to predict time to progression. The model was calibrated using correlations between tumor volume and ctDNA levels, then validated to identify parameters of response differences between patients achieving clinical benefit versus those who did not. Results: Relative changes in ctDNA positively correlated with tumor volume changes, with lower baseline ctDNA levels associated with superior clinical responses. In complete responders, ctDNA became undetectable before radiological confirmation, demonstrating both tumor volume reduction and ctDNA clearance. However, all patients eventually progressed. Parameter analysis revealed that treatment efficacy significantly impacts ctDNA shedding patterns, creating characteristic peaks that serve as early warning indicators for cancer progression, potentially enabling more timely therapeutic interventions.
Conclusions: Our patient-specific model accurately characterizes tumor and ctDNA dynamics in HPV-associated ASCC. Results suggest ctDNA-guided dosing strategies could optimize treatment regimens to improve responses and extend progression-free survival, establishing a foundation for integrating ctDNA surveillance into clinical monitoring protocols.
Other authors: Brandon Huffman MD, James Cleary MD, PhD, James Jabalee, and Renee Brady-Nicholls PhD
BACKGROUND: TIL therapy is an emerging immunotherapy where activated T cells are injected into the patient. This therapy can fail due to tumor-induced immunosuppression, for example via the PDL1/PD1 axis. PDL1 expression has been studied and can increase under IFNg, released by activated T cells, but PD-L1 relaxation is not understood. We hypothesize that there may be better strategies of therapy delivery that maximize tumor kill while minimizing immune suppression. We investigate these complex PD-L1 driven dynamics through a unique combination of in vitro studies and mathematical modeling.
METHODS: We have collected in vitro PD-L1 expression on NSCLC cells, which were either untreated or treated for 48 hours with high dose IFNg. To reinforce our in vitro findings, we developed a hybrid agent-based model (ABM). The agents in the ABM are tumor cells having variable PD-L1 expression, and immune cells that secrete IFNg and can kill tumor cells. We model TIL therapy as an immune cell pulse, given at regular or irregular intervals and with different quantities of cells.
RESULTS: Our ABM is calibrated to capture in vitro data dynamics. Under certain combinations of spatial configurations of the tumor and intermittent immune dosing schedules, we observe ‘sweet spots’ where tumor extinction is possible and immune exhaustion is avoided. This indicates that an appropriate pulsing of TIL therapy may lead to better overall immune efficacy than a bolus injection or continuous immunotherapy. Our novel findings can potentially benefit clinical cancer research by offering insights related to tumor extinction, equilibrium and escape.
Other authors: Sandhya Prabhakaran, Mark Robertson-Tessi, Kimberly Luddy, Rafael Bravo, Taylor M. Bursell, Julian Pineiro, Jeffrey West1, Megan Johnson, Jhanelle E. Gray, Amer A. Beg, Scott Antonia, Robert A. Gatenby and Alexander R. A. Anderson
Topical application of Low-Frequency Ultrasound, also known as Low-Frequency Sonophoresis (LFS), has been of significant interest to non-invasively, safely, reversibly, and effectively deliver drugs through the skin barrier for the last few decades. Compared to the extensive study of transport mechanisms and transdermal permeation enhancement for small molecules, the immunological effects of the physical/chemical perturbation has been superficial in terms of quality and quantity. In this work, we highlight the shortcomings of published reports and through an interdisciplinary approach of fundamental immunological principles and engineering, address both the conflicting issues by providing a methodical framework of investigation for the physical perturbation impact represented through skin impedance on immunological T cell response. Additionally, gene expression profiling on LFS treated skin strongly indicates the adjuvancy effect of LFS and synergism with surfactant sodium lauryl sulfate at 1%w/v through activation of selective cell migration pathways and costimulation markers in the treated skin. The results of this study encourage use of LFS as a robust immunization platform for T cell dependent pathologies such as viral infections and cancer.
Other authors: Matthias Oberli, Gregory Szeto, Carmen Gerlach, Joshua Doloff, Siddharth Jhunjhunwala, Jeffery Wyckoff, Uli von Andrian, Robert Langer, Daniel Blankschtein
Fluorescence LiDAR (FLiDAR) is a technology utilized for depth estimation, however, its application in scattering media is limited by significant computational difficulties. The complex nature of the acquired signal in such environments makes isolating the photon time-of-flight (related to depth) from the intrinsic fluorescence lifetime challenging for existing analytical and deep learning methodologies, which often rely on simplifying assumptions or computationally expensive pre-processing steps. To overcome these limitations, we present an end-to-end Physics-Guided Mixture-of-Experts (MoE) framework for direct parameter estimation from time-resolved signals from the detector. Our approach, EvidenceMoE, assigns specialized roles to its expert models based on the underlying physics of photon transport; an 'Early Expert' focuses on initial photon arrivals indicative of depth, while a 'Late Expert' analyzes the decay characteristics associated with lifetime [1]. Central to this framework is the integration of Evidence-Based Dirichlet Critics (EDCs), a model that assesses the reliability of each expert's output by providing quantitative quality scores and corrective feedback. A Decider Network then leverages this reliability information to adaptively fuse the expert predictions into a single, robust final estimate. We validated our method using simulated FLiDAR using Monte Carlo simulation for non-invasive fluorescence inclusion. The framework demonstrates effective performance for both depth and lifetime estimations. References 1. L. Zhao, H. Yang, W. Cong, G. Wang, and X. Intes, “L p regularization for early gate fluorescence molecular tomography
Other authors: Ferhat Demirkiran, Karthik Swaminathan, Naigang Wang, Navid Ibtehaj Nizam, Stefan T. Radev, Kaoutar El Maghraoui, Xavier Intes, Vikas Pandey
Targeted cancer therapies exploit oncogenic dependencies, yet heterogeneity within tumors enables therapeutic resistance and recurrence. Mechanisms for how tumor heterogeneity affects drug responses have not been fully characterized, highlighting a critical need in order to treat patients more effectively. Studying tumor heterogeneity requires in-vitro systems that accurately recapitulate physiologically relevant contexts, such as collagen-based extracellular matrices. In a high density collagen I matrix an aggressive breast cancer cell model exhibits two phenotypic states: a migratory network or stationary spheroid state. Transcriptomic profiles of the network cells resemble a more aggressive, treatment-refractory subpopulation of triple-negative breast cancer patients compared to spheroid cells. As a potential therapeutic strategy, we sought to homogenize cell states by targeting mechanisms driving network cell formation to induce a shift to spheroid cell states. To this end, we systematically knocked out each of the twenty-six integrin genes, key adhesion heterodimers, and assessed their functional importance towards phenotypic presentation and survival in high density collagen matrices. We found that Integrin Alpha 2 and Beta 1 (A2/B1), a collagen-binding integrin heterodimer, are the most essential integrins required for survival. Additionally we found that the knock out A2/B1 consistently led to the formation of spheroid cells. These findings suggest that selective targeting of A2/B1 heterodimer may serve as a therapeutic strategy to homogenize tumor cell states which can improve treatment efficacy by reducing aggressive, drug-resistant subpopulations.
Other authors: Jeffery Turner, Erin Schiksnis, Katherine Licon, Stephanie Fraley, Trey Ideker
Hepatoblastoma (HB) is a unique pediatric cancer because it has the fewest genomic abnormalities and the strongest association with preterm birth across all cancers. Intermittent hypoxia (IH) is the most common complication in preterm infants, and its proinflammatory effect on the liver has been demonstrated in adults with apnea. Neonatal livers contain a distinct population of neutrophils that initiate inflammation through the recruitment of immune cells and play a critical role in tumorigenesis. Therefore, we hypothesize that IH-induced proinflammatory response from the distinct population of neutrophils in the neonatal liver accelerates HB pathogenesis. We showed that ABC-ER-Myc pups (P1) subjected to IH for five days from postnatal day 1 (P1) to P5 had a notable decrease in their median survival, increased tumor burden by the endpoint, and an increase in the number of β-catenin and Ki67+ cells. Mice treated with IH also showed increased expression of the hematopoietic marker CD45 and the neutrophil and inflammation marker MPO, compared to pups treated in room air. Moreover, neonatal liver from IH-treated pups showed aberrant expression of CXCR2 ligands, CXCL1, and CXCL2, suggesting IH induced strong proinflammatory responses in the liver and significantly accelerated HB progression. Our hypothesis was further confirmed in vitro by demonstrating that HB cells treated with recombinant antibodies CXCL1 and CXCL2 exhibited enhanced growth and survival compared to untreated cells cultured under IH conditions. In summary, we concluded that IH triggers CXCR2 signaling in neutrophils, thereby promoting HB pathogenesis in the neonatal liver.
Other authors: Tian Cheng, Ph.D., Xiaoli Liu, Jie Fang, Ph.D., Jun Yang, Ph.D., Liqin Zhu, Ph.D., Andrew J. Murphy, M.D.
The shape and structure of mitochondria is extensively to its functional state, with mitochondria moving between states of fusion and fission to regulate metabolic activity. This phenomenon is particularly important in understanding the effects of glucose damage on retinal cells as recenter evidence has shown that retinal vascular cells develop a resilience to high glucose induced damage. Though this phenomenon is known, no quantitative methods exist to quantify mitochondrial alterations at nanoscale resolutions. Three-dimensional single-molecule localization microscopy is a super-resolution technique that can visualize such nanoscale alterations. In this talk, we will demonstrate how we used SMLM to quantitatively assess mitochondrial shape changes as human retinal endothelial cells are exposed to high glucose environments. In doing so, we found that though mitochondria undergo significant fragmentation in their initial exposure to high glucose, cells recovered after 10 days, successfully showing that mitochondria develop high glucose induced adaptations to protect from damage. In addition human retinal endothelial cells, we will also discuss how we quantified mitochondrial morphological changes associated with cisplatin resistance and brequinar treatment in OVCAR5 ovarian cancer cells.
Other authors: Manav Gandhi, Andrius Kazlauskas, Hao F. Zhang
Improving T cell infiltration into solid tumors remains a critical goal in immunotherapy. In this project, we analyzed the spatial distribution of T cells relative to the edge of 3D Matrigel domes containing OVCAR5 ovarian tumor cells, following treatment with PINC-enhanced photoimmunotherapy. Using fluorescence microscopy time-lapse data, I developed a custom image analysis pipeline to segment T cells, filter by size, and quantify their distance from the dome boundary across multiple time points. This approach enables assessment of infiltration depth and density over time, allowing for comparison between treatment and control conditions. Preliminary results suggest that PINC/photoimmunotherapy enhances T cell penetration into the tumor matrix. These findings support the potential of photoimmunotherapy to modulate immune-tumor interactions in 3D environments and provide a framework for spatial quantification in similar experimental systems.
Other authors: Rebecca Harman, Ivonne Lozano-Pope, Bryan Spring
Solid tumor cancers develop within protective microenvironments that pose significant challenges to immunotherapeutic strategies due to their highly immunosuppressive nature. Tumor-derived extracellular vesicles (TEVs) potently mediate such immunosuppression through the bioactive cargo they carry. The immune checkpoint molecule PD-L1 is one such cargo that potently suppresses recipient cytotoxic T-cells. However, given the nanoscale size and heterogeneity of TEVs, the mechanism driving their potent immunosuppression remains unclear. Previously, we found that PD-L1+ TEVs can alter the localization of surface PD-1 molecules to significantly inhibit T-cell signaling in silico. Thus, we hypothesize that TEVs induce PD-1 “nanoclusters” on the T-cell membrane and that these nanoclusters are productive, inhibitory signaling complexes that drive T-cell suppression. To test this, we characterized TEVs derived from the triple-negative breast cancer cell line MDA-MB-231 and then assessed their interactions with T-cells using dSTORM superresolution microscopy. We find that only a small fraction of the TEV population is enriched for PD-L1, but that these TEVs are capable of altering PD-1 localization on recipient T-cells. These results novelly visualize EV/cell interactions and provide evidence suggesting PD-1 nanoclusters as a mechanism potentiating TEV suppression of cytotoxic T-cells.
Other authors: Xiaowei Xu, Wei Guo, Su Chin Heo, Jina Ko, Ravi Radhakrishnan
Understanding how the immune system “recognizes” foreign bodies, such as nanocarriers shuttling anti-cancer drugs, engineered CAR-T cells, and cancer cells themselves, is essential to designing effective cancer treatments. The innate immune system depends on the complement cascade, a 500-million-year-old protein network which deposits C3b on protein-binding sites on foreign body surfaces, labeling them for phagocytic clearance. Unwanted immunogenicity prevents therapies from reaching their targets and can lead to adverse patient reactions. However, the principles governing how the system “decides” to coat an NC and how this alters immune uptake have eluded researchers.
We utilize a multi-scale systems approach to investigate this question. We integrate ordinary differential equations and an agent-based model, validated by experiment, to reveal that complement activation follows a percolation phase transition governed by NC surface protein density. We reveal that a critical percolation threshold separates uncoated and C3b-coated NC regimes; NCs above the threshold exhibit multistability within a bimodal distribution, occupying states of low, high, or no C3b coverage. Using our Dynamically Triangulated Monte Carlo membrane simulation model, we show that NC–membrane interactions vary across these regimes. High-C3b NCs form numerous ligand–receptor bonds dominated by C3b-mediated adhesion and induce local curvature—suggesting greater internalization propensity. Low-C3b NCs require higher site density to achieve similar effects, mediated by both C3b and other NC surface proteins. NCs lacking C3b evade immune recognition.
This biophysics-based understanding of nanocarrier immunogenicity can help inform the design of safer and more effective treatments. * We acknowledge support in part from NCI
Other authors: Sahil Kulkarni, Ravi Radhakrishnan
Cancer exhibits substantial heterogeneity, manifesting as distinct morphological and molecular variations across tumors. Intrinsic genetic drivers and tumor microenvironment (TME) driven physical heterogeneity together affect tumor progression, therapy efficacy, and consequently, patient prognosis. This heterogeneity necessitates the development of precision therapies utilizing multi-physics, multi-scale modeling. In this work, we present our multi-scale Wilms Tumor model, which makes genotype-specific cell fate predictions under combination chemotherapy and uses them to inform a tissue-level agent-based model to capture spatiotemporal dynamics driven by intrinsic as well as TME-driven heterogeneity. This work intends to accelerate our efforts towards building patient-specific cancer digital twins. The aggressive blastemal subtype of Wilms tumor is known to be driven by the cells' MYCN status. Clinically, low cell-matrix adhesion is a signature of Wilms tumor tissue as compared to normal kidney. This morphological phenotype is potentially mediated by MYCN-amplification in blastemal cells, and can potentially affect chemotherapy efficacy. We thus simulated combination chemotherapy in our multiscale framework, and tailored the model to different clinical patient cohort-specific genetic information and chemotherapy treatment regimes. We performed sensitivity analyses on key parameters, including MYCN status, tissue composition, and adhesion-motility phenotypes, to explore their impact on emergent tumor behavior. Our results reveal that chemotherapy efficacy varies significantly with tumor subtype composition. Blastemal cells respond poorly to chemotherapies. This phenomenon is further exacerbated at lower adhesion strengths. Beyond adhesion-motility traits, blastemal Wilms tumors often exhibit immunosuppressive TMEs. As a future direction, we aim to test whether exosome-mediated signaling drives this effect by incorporating tumor-immune communication and immune infiltration into our model. We acknowledge funding from the National Cancer Institute through the Physical Sciences Oncology Network.
Other authors: Dr. Alokendra Ghosh, Dr. Lindsey Fernandez, Kewei Ni, Dr. Ravi Radhakrishnan
Trastuzumab (TZM), a monoclonal antibody targeting HER2+ breast cancer, often faces delivery and tumor penetration limitations, leading to variable therapeutic response and resistance. Near infrared Fluorescence Lifetime Förster Resonance Energy Transfer (NIR FLI-FRET) imaging offers a powerful approach to spatially and quantitatively assess TZM-HER2 engagement in vivo, capturing intratumoral heterogeneity influenced by collagen content and vascularity. However, a major limitation in FLI-FRET imaging lies in its dependency on amine-based N hydroxysuccinimide (NHS) ester stochastic labeling for fluorescent dye conjugation, creating heterogeneous antibody-dye conjugates, leading to inconsistent binding and FLI-FRET measurements. We developed an engineered meditope-enabled TZM (TZM-MDT) antibody, site specifically labeled with two NIR dyes per IgG via cyclic peptides on each Fab arm. This enables a consistent labeling scheme critical for high-resolution FLI-FRET imaging of antibody-HER2 interactions. We propose that site-specific TZM-MDT labeling enhances tumor binding, penetration, and FLI-FRET imaging accuracy compared to TZM-NHS labeling. Using in vitro microscopy and ex vivo time-domain mesoscopic fluorescence molecular tomography (TD MFMT), we compared NIR-labeled TZM-MDT and TZM-NHS in HER2+ AU565 cells and xenografts. TZM-MDT exhibited superior HER2 engagement, evidenced by consistent fluorescence lifetimes, elevated donor FRET fractions, and deeper tumor penetration. Pixel-wise FRET mapping revealed widespread intratumoral engagement by TZM-MDT in contrast to peripheral binding by TZM-NHS. These findings were validated by immunohistochemistry (IHC), with spatial correlation confirmed via ImageJ and GIMP-based image analysis. To investigate resistance mechanisms, we assessed HER2 engagement in Estrogen receptor (ER+) and HER2+ BT474 and ER- HER2+ AU565 tumors. FRET imaging revealed significantly reduced HER2 engagement in BT474 tumors, consistent with ER-mediated resistance and reduced TZM efficacy. Notably, regions with high FRET signal corresponded to areas of decreased tumor cellularity on H&E staining, indicating therapeutic impact. Thus, FLI-FRET imaging using MDT-TZM provides a non-invasive, quantitative, and spatially resolved assessment of HER2 engagement supporting its application in personalized cancer therapy.
Other authors: NA
Genomic alterations in cancer arise from selective pressures acting on genes and hallmark molecular systems, layered over a background of random mutagenic events. Here we introduce CanSRMaPP, an approach to deconvolve the influence of each of these distinct factors within a given tumor type. When applied to lung adenocarcinoma genomes, CanSRMaPP identifies positive selection on a set of 51 molecular systems, yielding a model which is highly predictive of mutation patterns in new cohorts. These systems aggregate rare and common mutations across 566 genes, substantially expanding the collection of genes implicated in this cancer type. CanSRMaPP also improves the translation of genetic alterations to phenotypes, which we demonstrate by using tumor genomes to predict levels of prominent protein biomarkers. This study presents a general strategy for pathway mapping and phenotypic prediction in cancer, with implications for understanding other cancer types and polygenic diseases.
Other authors: Burçak Otlu, Roded Sharan, Trey Ideker
For in vitro cell culture, providing appropriate growth supplements is essential. Fetal bovine serum (FBS) is the most commonly used supplement added to culture media. FBS is derived from bovine fetuses obtained from slaughtered pregnant livestock. While widely used, FBS raises several concerns, most notably ethical issues, as well as high cost, batch-to-batch variability, and supply chain challenges. These limitations have led researchers to explore alternatives, such as human platelet lysate (hPL). Although hPL performs well in many applications, it still suffers from variability due to donor characteristics, limited availability, and risk of pathogen contamination [1]. Recently, a new supplement called bovine platelet lysate (bPL), marketed as Plenty™, has been introduced. bPL is significantly more affordable than FBS and is derived from live cattle in a minimally invasive manner, reducing ethical concerns. While bPL has shown promise in supporting breast cancer cell growth, its performance in prostate cancer (PC) models, particularly under physiologically relevant oxygen conditions, remains largely unexplored. This study investigates the viability and proliferative behavior of PC3 prostate cancer cells cultured in RPMI medium supplemented with bPL under normoxic (21% O₂) and hypoxic (1% O₂) environments. The present work also evaluates the viability of LNCaP prostate cancer cells in bPL-supplemented media under the same oxygen conditions, and further assessed the ability of bPL to support 3D spheroid formation of LNCaP cells in both normoxia and hypoxia.
Other authors: Michael R. King
Lipid metabolic programming plays an essential role in immune cell activation, differentiation, function, and in shaping the antitumor immune response. Expanding upon our previous pan-cancer analyses in The Cancer Genome Atlas (TCGA) that identified germline variants shaping the immune tumor microenvironment (TME), we assessed the heritability of 22 KEGG lipid metabolic pathway expression signatures across >9,000 individuals encompassing 30 cancer types. Our analysis uncovered significant heritability (FDR p<0.05) for 73% of lipid metabolic signatures, including fatty acid, glycosphingolipid, glycerophospholipid, and steroid hormone pathways. We identified 53 genome-wide significant (p<5×10⁻⁸) and 260 suggestive (p<1×10⁻⁶) loci, and >1,300 germline variants associated with these heritable lipid metabolic traits. Lipid metabolic signatures robustly predicted pan-cancer immune subtypes (AUC≥0.8). Notably, metabolic traits linked to fatty acid, glycosphingolipid, and glycerophospholipid pathways strongly correlated with immune signatures associated with enrichment of cytotoxic cells, CD8 T cells, and T helper cells, as well as NK cells, neutrophils, and eosinophils. These signatures were also significantly associated with improved progression-free survival pan-cancer and within a majority of cancer types (p<0.05). We observed considerable pleiotropy at 28 genetic loci affecting both lipid metabolic and immune traits. Additionally, the expression of heritable immune-lipid signatures and frequency of associated germline variants significantly varied across genetic ancestry groups (FDR p<0.05). Our findings illustrate that germline genetics partially shapes the immune-lipid metabolic TME. The differential prevalence of these heritable TME features across genetic ancestries further suggests that inherited genetic factors, along with social and structural determinants of health, may contribute to disparities in cancer outcomes.
Other authors: Zheyun Xu, Denise M. Wolf, Vésteinn Thorsson, Mohamad Saad, Donglei Hu, Scott Huntsman, Christina Yau, Davide Bedognetti, Jennifer Rosenbluth, Michael J. Campbell, Laura J. van ‘t Veer, Elad Ziv
Our lab previously identified a hybrid epithelial-mesenchymal (E/M) Cdh1+/Vim+ cell state found at the leading edge of invasion in triple-negative breast cancer (TNBC) tumor using longitudinal single-cell RNA sequencing and dimensionality reduction via non-negative matrix factorization. Further, hybrid epithelial-mesenchymal cells states have previously been implicated throughout mammary gland development, particularly as the terminal end buds invade into the mammary fat pad during puberty. Such gene expression similarities suggest common cellular programs between regulated invasion during development and aberrant invasion in TNBC. To further characterize our hybrid E/M gene signature, I generated preliminary data using the projectR package developed by my lab. ProjectR is a Bioconductor R package that employs a transfer learning framework to transcriptomic or genomic data and ultimately exploits shared features (in this case genes) between independent datasets to evaluate presence of latent spaces (i.e. PCs, CoGAPS patterns, trajectories) in a new context. This powerful tool allows for more robust in silico interpretation of previously learned latent spaces by evaluating their use across different technological modalities, tissue types, and even species. I sought to establish whether our hybrid E/M gene signature is even implicated during developmental timepoints. To do so, I projected three publicly available scRNAseq datasets of the developing mouse mammary epithelium into all three EMT-associated patterns from Grasset et al, 2022. Intriguingly, when correlating one-hot-encoded metadata (in this case, developmental timepoints) and projection weights, the Cdh1+/Vim+ hybrid E/M signature showed the strongest correlation within the window of puberty, ranging from 5 to 10 weeks of age. This highlights an interesting developmental period to follow up on, since this timeframe coincides with when most ductal branching occurs. Further, preliminary immunofluorescence experiments demonstrate the presence of these Cdh1+/Vim+ hybrid E/M cells present in epithelial organoids derived from healthy mouse mammary glands subjected to an ex vivo assay (via embedding in a 1:1 mixture of collagen I and Cultrex) that recapitulates healthy branching morphogenesis. My preliminary work using in silico transfer learning and ex vivo organoid culture establishes that the Cdh1+/Vim+ hybrid E/M cell state previously found in triple-negative breast cancer invasion is not unique to this disease state; rather, it is aberrantly co-opted from its rightful timeframe during pubertal development.
Other authors: Andrew Ewald, Genevieve Stein-O’Brien
Protein–protein interactions mediated by short linear motifs (SLiMs) are essential for cellular signaling and regulation yet remain underrepresented in current interactome maps due to their transient and low-affinity nature. Here, we present PrePPI-SLiM, a proteome-scale computational pipeline that integrates domain–motif interaction data from the ELM database with structural domain annotations and sequence-derived features such as disorder and conservation. Utilizing a naïve Bayes classifier and a likelihood ratio-based scoring framework, PrePPI-SLiM systematically evaluates all possible protein pairs within a proteome to identify interactions plausibly mediated by domain–motif recognition. Applied to the human and yeast proteomes, PrePPI-SLiM predicted over 21 million PPepIs in humans and 1.7 million in yeast, including nearly 4 million and 430,000 interactions, respectively, not found in the current PrePPI database containing domain-domain interactions. At a stringent false positive rate (FPR ≤ 0.005), the pipeline identified 114,935 high-confidence intreractions in humans spanning 59 Pfam domains, and 8,580 PPepIs in yeast spanning 32 Pfam domains, demonstrating broad coverage of peptide recognition domains across species. To assess structural plausibility, we modeled high-confidence interactions using AF3Complex and validated them against experimentally resolved receptor–peptide complexes from the Propedia database. The strong agreement between predicted and known structures highlights the biological relevance of these interactions. Furthermore, clustering of the high-confidence PrePPI-SLiM interactome yields functionally coherent networks that reveal mechanistic insights into cellular processes. Altogether, PrePPI-SLiM provides a scalable and statistically robust framework for uncovering peptide-mediated interactions that are often missed by existing experimental and computational methods.
Other authors: NA
Glioblastoma (GBM) is a lethal brain cancer with a median survival time of less than 15 months. GBM tumors have heterogeneous matrix stiffness and oxygen levels that create distinct biophysical niches within the tumor and promote growth and survival of GBM cells. However, few studies have investigated the overlapping effects of these features on GBM aggression. Thus, we investigated metabolic and phenotypic changes in GBM cells cultured in matrices mimicking the soft, brain-like (G’=100 Pa) or the stiff, tumor-like (G’=1000 Pa) environment at 1% or 21% O2. We found that polyamine metabolism is altered by matrix stiffness in two patient lines. We observed higher levels of polyamines in soft matrices than stiff in both 21% and 1% O2. This may be due to increased polyamine catabolism through SAT1 and/or use of spermidine downstream to activate EIF5A, both of which were observed in stiff matrices. Both cell lines also have increased proliferation in stiff matrices, particularly in 1% O2 conditions, which may be driven by higher EIF5AH. We also identified changes in lipid metabolism between conditions. In the HK177 line, 1% O2 increased triglycerides and polyunsaturated fatty acids, with stiffness also contributing but to a lesser extent. In the HK408 line, matrix stiffness primarily drove differences in lipid metabolism, with stiffer matrices decreasing glycerophospholipids and increasing polyunsaturated fatty acids. These findings reveal the important interplay between matrix stiffness and oxygen levels in the tumor microenvironment and their patient-specific effects on cancer metabolism.
Other authors: Shuxin Dong, Alireza Sohrabi, Nadia Toh, Alessia Lodi, Stephanie Seidlits
Introduction:
Our gut is home to trillions of bacteria forming a dynamic community that plays a vital role in our health. Understanding how gut bacteria colonize is key to unlocking their impact on health and disease. These microbial communities play vital roles in regulating important physiological processes such as bile acid metabolism. Gut bacteria facilitate the synthesis, conjugation, uptake and recirculation of the bile acids to the liver through modulation of transcription factor Farnesoid X receptor (FXR), and G-protein coupled receptor (TGR5). [Wahlström A, et. al. doi: 10.1016/j.cmet.2016.05.005] Elevated levels of bile acids are associated with increased risk of colorectal cancer. [Chao A, et. al. doi: 10.1001/jama.293.2.172] Given these links, it is essential to understand the bile acid–microbiota axis in colorectal carcinogenesis. However, bile salts are not typically incorporated in organoid cultures, particularly in cases where host-microbial interactions are studied. This study aims to study colonization of organoids by gut bacteria in the presence of bile acids. We further study how bile acid salts affect the growth of bacteria as well as the impact on tumor-associated genes in colorectal organoids in a microfluidic colon organoid-on-a-chip.
Material and methods:
We isolated colon organoids isolated from healthy and tumor tissues from Apcfl CDX2 Cre mice. We used two bacterial strains - EcAZ1, a native strain isolated from these mice, and EcAZ2, where EcAZ1 is encoded with the Bile Salt Hydrolase (BSH) gene. This gene upregulates the production of bile salt hydrolase, which deconjugates primary bile acids. The bile salts Tauro β muricholic acid (TβmCA) and Chenodeoxycholic acid (CDCA) were added to the culture media. The organoids were cultured in Matrigel. Bacteria were added at a MOI of 1:10, and the colonization was studied for 72 hr with a wash every 24 hr. Both bacterial strains were encoded with green fluorescent protein to help to track them during the experiments. The cells and bacteria were visualized using confocal microscopy and gene expression analysis via qPCR was conducted to assess the regulation of key colorectal genes, including stem cell markers like LGR5, differentiation genes CK20, and markers specific for bile acid receptors including FGF15 and TGR5. Experiments were also carried out in two-chamber microfluidic devices, which were fabricated as previously described by our group [Wang C, et. al. doi: 10.1096/fj.202001600RR].
Results, Conclusions, and Discussions:
We observed bacterial penetration of the Matrigel and colonization of organoids, unlike the control with only Matrigel. The two bacterial strains exhibited distinct colonization patterns, influenced by bile salts. While bile salts generally inhibited bacterial growth, and a similar trend was seen in organoid colonization, qPCR quantification revealed specific effects. Compared to conditions without bile salts, EcAZ1 colonization of tumor organoids was approximately 2-fold higher, whereas EcAZ2 colonization was suppressed by 2.5-fold. Conversely, bile salt addition significantly downregulated the colonization of healthy organoids by both strains. Notably, EcAZ1 showed the highest colonization of tumor organoids in the presence of bile salts. Our assessment of cellular markers via qPCR, with normalization to bacterial counts, yielded interesting contrasts. Both engineered bacteria upregulated FGF15 and TGR5 in both healthy and tumor organoids, highlighting their potential influence on intestinal bile acid homeostasis [3]. Our data shows EcAZ2's strong ability to upregulate these factors, implying enhanced bile acid deconjugation and potential CRC suppression through FXR activation. In contrast, the effects on cell lineage markers (LGR5, LYZ, CK20) differed significantly. EcAZ1 strongly upregulated these markers in healthy organoids but downregulated them in tumor organoids, while EcAZ2 showed a weaker upregulation in healthy organoids. These findings suggest a potential role for engineered bacteria in modulating cell lineage homeostasis in tumors. The specific upregulation of LYZ points to a possible increase in antimicrobial enzyme production by Paneth cells, critical players in intestinal homeostasis and inflammatory processes [Yu S, et. al. doi: 10.1016/j.immuni.2020.07.010]. This research explores the impact of gut bacteria on the microenvironment of tumor versus healthy organoids in the presence of bile acids. Utilizing in vitro models in plates and organ-on-chip devices, we examined the colonization of organoids, enabling analysis of bacterial behavior and host responses. Normalizing the data by the colonizing bacteria presented a more accurate comparison of cellular effects; the inclusion of bile salts is innovative and enhanced the physiological relevance.
Other authors: Arianna Brevi, Durga Prasad Rangineni, Yixiao Ma, Amir Zarrinpar
Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy characterized by a complex, spatially organized malignant and stromal microenvironment. This comprises a heterogeneous collection of coexisting normal and malignant transcriptional states that are highly conserved across patients. Understanding how malignant cells communicate with the stroma via receptor/ligand-mediated interactions and how these affect tumor progression remains a fundamental yet highly elusive challenge. In this project, we investigate how cascades of ligand/receptor-mediated interactions drive emergent PDAC tumor phenotypes with different functional characteristics. By leveraging spatial transcriptomics to assess cell-cell distances, we integrate transcriptional, post-transcriptional, and protein-protein interactions in complexes—as inferred by network-based algorithms such as ARACNE, MINDy, and PrePPI—to construct a novel framework to decode interactions between tumor and tumor microenvironment cells. Our approach identifies key signals supporting the immunosuppressive nature of PDAC’s tumor microenvironment, as mediated by fibroblast polarization and angiogenic quiescence—critical hallmarks of PDAC’s desmoplastic microenvironment and therapy resistance. Moving forward, we aim to functionally validate these interactions in engineered co-culture systems to test the top ligand-receptor interactions. Ultimately, this work lays the groundwork for designing combination therapies that cooperate with tumor ecosystems—transforming immune-excluded tumors into immune-responsive ones and opening new therapeutic frontiers for a historically intractable disease.
Other authors: Bianca Xue, Andrea Califano
High-grade serous ovarian carcinoma (HGSOC) is one of the most aggressive gynecological diseases, with a 5-year survival rate of < 50%. Nearly 80% of HGSOC cases metastasize to the omentum, which undergoes significant remodeling with disease progression, characterized by a loss of adipocytes and increased stromal content. This transition from an adipose- to stroma-rich tissue alters biophysical properties, potentially influencing tumor invasion; however, these factors remain poorly understood. We hypothesized that changes in adipocyte packing and collagen density impact tumor cell invasion. Since existing omental models often lack adipocytes or fail to isolate their biophysical role, we engineered a hybrid hydrogel system with alginate microspheres mimicking adipocytes within an interpenetrating network of gelatin methacrylate and collagen. Histological analysis of healthy vs. diseased omenta revealed a reduction in adipocyte diameter (41µm to 30µm) and packing fraction (72% to 63%), coupled with increased collagen deposition and intercellular gap. Our hybrid model successfully mimicked these key structural changes. Invasion assays with RFP-tagged OVCAR8 spheroids revealed significantly greater invasion in hybrid gels, especially at the stage III-IV packing (63%), than in the benign packing (72%). Enhanced invasion at high collagen density was only observed under the 63% packing condition. To elucidate the mechanism, we inhibited Yes-Associated Protein (YAP), a key effector of mechanotransduction. YAP inhibition ablated the enhanced invasion observed with reduced packing, implicating the role of YAP signaling in spatial heterogeneity-driven invasion. Overall, our findings suggest that the biophysical impact of adipocyte loss promotes HGSOC invasion via YAP-mediated mechanotransduction.
Other authors: Kaitlyn Menegoni, Ning Yang, Pamela K. Kreeger
Breast cancer recurrence can arise years after initial therapy due to the survival of dormant cancer cells that evade immune surveillance. To investigate the mechanisms enabling dormancy escape and recurrence, we used the D2.0R syngeneic mouse model, in which orthotopically transplanted breast cancer cells are initially eliminated by adaptive immunity but relapse after prolonged latency. We derived seven “late-escaper” cell lines from these spontaneous recurrences and found that all could reinitiate tumor growth in immunocompetent mice, though with distinct proliferation rates, immune landscapes, and metastatic capacities. Transcriptomic profiling revealed that late-escaper cells resolved endoplasmic reticulum (ER) stress and downregulated unfolded protein response (UPR) pathways—a change associated with dormancy exit. While some lines acquired shared adaptations, such as immune evasion and EMT, others exhibited unique gene expression and mutational profiles, including Myc amplification and MHC-I loss. Spatial proteomic and in vivo analyses revealed that the composition of the tumor microenvironment, particularly T cell infiltration, correlated inversely with metastatic potential. Notably, neutrophil extracellular traps (NETs) promoted dormancy awakening and lung metastasis in select lines, linking inflammation to recurrence. Together, our results define a set of molecular and immune mechanisms that enable dormant cancer cells to escape immune control and drive relapse, offering new avenues for therapeutic intervention.
Other authors: Emilis Bružas, Xue-Yan He, Margaret Shevik, Longling S. Shui, J. Erby Wilkinson, Scott W. Lowe, Camila dos Santos
Pancreatic ductal adenocarcinoma (PDAC) is an aggressive cancer with a dismal five-year survival rate of 13%. Most patients present with metastatic disease, and 80% undergoing surgery relapse within two years. Research on PDAC has traditionally focused on primary tumors, thus most treatment derives from primary tumor biology. Recent evidence unveiled significant differences in the tumor immune microenvironment (TIME) between primary and metastatic lesions. Understanding these differences is crucial, as immunotherapies are predominantly administered to patients with refractory metastatic PDAC, yet traditional treatment fails to account for organ specificity and patient TIME signatures. In this study, we wanted to know why the same cancer cell type in different organs would create different microenvironments in the pancreas, liver, and lungs. To do this, we investigated the TIME in early and late-stage metastatic cohorts by doing multiplex immunohistochemistry (mIHC), transcriptomic assays and spatial analysis.
Other authors: N/A
Metastatic triple-negative breast cancer (TNBC) is a major cause of mortality in women due to its aggressiveness and drug resistance. The divalent metal transporter 1 (DMT1) regulates TNBC invasion by modulating iron metabolism and extracellular matrix (ECM) remodeling, particularly collagen integrity, via endoplasmic reticulum (ER) stress. Using CRISPR, we generated DMT1 knockout (KO) and overexpression (OE) MDA-MB-231 cells, cultured in 2D and 3D models to assess phenotype and invasion. DMT1 KO spheroids exhibited reduced sphericity, increased volume, and lower cell density, with enhanced invasion in 3D Collagen1A and Matrigel matrices compared to compact OE spheroids and wild-type controls, corroborated by lung metastasis data. Ferro-Orange assays and transcriptomic analysis revealed that DMT1 KO increases the labile iron pool (LIP), inducing ER stress and activating the unfolded protein response (UPR). This disrupts oxidative folding and disulfide bond formation, impairing Collagen 1A production and reducing ECM integrity, thus promoting epithelial-to-mesenchymal transition (EMT). Conversely, DMT1 OE cells maintained low LIP, reduced ER stress, and preserved collagen expression, slowing EMT. Transcriptomic data confirmed reduced collagen in DMT1 KO cells, linked to mesenchymal traits. Notably, 2D wound-healing assays showed faster migration in OE cells, highlighting 3D models’ relevance to in vivo conditions. TGF-β inhibition analysis further connected DMT1 KO to collagen loss and increased invasion. These findings demonstrate that DMT1 regulates TNBC invasion through ER stress-mediated collagen disruption, highlighting collagen as a key ECM component and potential therapeutic target.
Other authors: N/A
Platelets play a critical role in the development of the pre-metastatic niche, facilitating tumor cell extravasation, and immune evasion. The mechanism of how this is propagated is still being actively studied but there are several biophysical factors are attributed to the development of metastasis. By the promotion of endothelial permeability and modulating shear forces, activated platelets can induce an environment conducive to metastasis. Using intravital microscopy (IVM) of GP1b-labeled platelets in the liver vasculature, we quantified platelet flow dynamics in a 4T1 Balb/C model of triple negative breast cancer (TNBC). We evaluated 3 drugs targeting different mechanisms of platelet activation, and found that Eptifibatide, an integrin αIIbβIII antagonist, most effectively restored platelet fluidity and reduced stationary platelet accumulation.
Other authors: Ning Shao, Milos Kojic, Arturas Ziemys, Andrew Schafer, Thomas Wong, Yan Ting Liu, Thao Nguyen, Kyuson Yun