Cells sense and integrate extracellular cues through intracellular signaling networks that reshape transcription factor activity to dictate cellular responses. Signaling activity is difficult to decipher: it is non-linear, and it contains extensive feedback and crosstalk. Furthermore, the same perturbation can elicit markedly different responses depending on context (e.g., cell type, disease state, and tissue microenvironment) such that identical stimuli produce diverse responses in multicellular populations. Consequently, there is a vast combinatorial space of complex interactions and context-dependent responses that necessitate computational models. Computational models of single-cell perturbation responses are demonstrated to predict cellular responses, but are often limited in mechanistic insight. Prior knowledge networks offer a route to bridge predictive capability and interpretability. Here we present scLEMBAS, a context-aware, gray-box neural network that models signaling pathway activity at single-cell resolution while preserving mechanistic grounding. scLEMBAS encodes a prior-knowledge network of protein-protein interactions as a recurrent neural network whose learnable edge weights correspond to signaling interaction strengths. It also captures context and individual cell variance through compositional bias terms. An adversarial approach allows the model to answer a single-cell counterfactual - what a given cell's TF activity would be under a different perturbation or context - while involving mechanistic rather than simply relational information. Across two scRNA-seq datasets spanning single- and multi-perturbation settings, scLEMBAS accurately predicts out-of-distribution combinations of perturbation and context. Capturing population variance across individual cells enables the model to predict cell subtype specific perturbation responses, despite being agnostic to such labels. Beyond prediction, scLEMBAS' learned parameters are biologically interpretable: learned edge weights carry information beyond network topology and "self-prune" spurious interactions, while the categorical bias nominates proteins associated with cell-type-specific perturbation states. Overall, scLEMBAS enables quantitative dissection of how signaling pathway activity is reshaped by perturbation within specific cellular contexts.

For endemically circulating viruses, quantifying neutralization capacity of antibodies in a rapid and high-throughput manner is paramount given the ever-evolving targets of neutralization. Moreover, identifying features of an antibody response correlating with neutralization at the multivariate level can inform vaccine boosting strategies and next-generation monoclonal antibody therapies. In this study, we developed a systems-predicted neutralizing antibody platform that is capable of quantifying neutralization capacity at the multiplex level. We validated the platform using SARS-CoV-2 ancestral and highly diverged Omicron sublineage spikes simultaneously. Combining these results with a broader systems serology approach identified humoral signatures that predict broadly neutralizing responses that are divergent between vaccine or hybrid immunity. In both cases, though, predicted neutralization responses were enmeshed in antibody networks, showing that target-specific and broadly neutralizing antibody responses are multi-isotype- and subclass-influenced. We propose that neutralization against a diverse array of viral receptor-binding proteins can be quantified using a systems-based platform in a sample- and time-sparing approach, which can be further employed to predict protection against newly emerged viruses or variants.

Pregnancy is characterized by dynamic immunological adaptations which are essential for maintaining both maternal and fetal health. The first-time use of coronavirus disease 2019 (COVID-19) vaccines in pregnant individuals presented an opportunity to discover pregnancy-specific immunoproteomic signatures across gestation. In this study, we profiled abundance levels of 1,451 unique proteins at baseline and in response to de novo COVID-19 vaccination in 466 samples from 278 pregnant individuals. Self-organizing map analysis identified 11 clusters of proteins based on similar longitudinal trajectories, with each cluster associated with distinct biological processes. Further functional characterization of protein expression levels across gestation revealed inflection points at 18–20 and 30–32 weeks of gestation, providing insight into expression dynamics of proteins associated with regulation of immune tolerance. Generalized additive modeling inferred gestational age-specific responses to the first and second dose of the vaccine; these highlight enrichment in proteins associated with cellular motility and canonical immune signaling after first and third trimester vaccination, but relative suppression of proteins associated with immune and inflammatory signal transduction pathways after second trimester vaccination consistent with diminished antibody responses when pregnant individuals are vaccinated exclusively in the second trimester. These findings advance our understanding of trimester-specific immune responses to vaccines, providing opportunities to enhance maternal and neonatal health.

Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) are an uncommon and poorly understood malignancy with low mutational burden, lacking well-defined oncogenic drivers. GEP-NET mortality frequently results from extensive hepatic metastases. Accordingly, we interrogated phosphoproteomic data from GEP-NET liver metastases and patient-matched uninvolved liver to identify tumor-specific signaling and targetable tumor vulnerabilities using Kinase Motif Enrichment Analysis (KMEA), a new tool leveraging the recent Kinase Library compendium of the substrate motif specificity for nearly the entire human kinome. KMEA identified patient tumor-specific upregulation of mTOR or casein kinase 2 (CK2) activity that would be undiscoverable by standard personalized genomic and transcriptomic approaches. Striking concordance was observed between KMEA predictions for specific tumors, and their sensitivity to inhibitors of mTOR or CK2 using patient tumor-derived organoids. These findings reveal potential clinically-actionable protein kinases hyperactivated in GEP-NETs, and more broadly indicate a general method for personalized cancer treatment using phosphoproteomics and KMEA-derived kinase activity signatures.

Axl is a therapeutic target under active clinical investigation for a variety of cancer indications. While pre-clinical mouse studies have been promising, clinical trials targeting Axl in combination with standard-of-care therapies have yielded mixed results. We hypothesize that this clinical translation gap may in part be due to an incomplete understanding of the effects of Axl inhibition specifically on the myeloid compartment of the human tumor microenvironment, as Axl mediates immunoregulatory signaling in myeloid cells. Utilizing human in vitro melanoma tumor microenvironment model systems, herein we generated and analyzed single-cell RNA-sequencing data to understand the impacts of tumor microenvironment composition and tumor cell prior therapy on myeloid cell response to Axl inhibition. We used our previously established in vitro model system to generate single-cell RNA-sequencing samples of tumor cell-macrophage co-cultures as well as tumor cell-macrophage-dendritic cell tri-cultures from four healthy buffy coat donors. We first applied LIgand-receptor ANalysis frAmework (LIANA) to our transcriptional data to infer potential intercellular crosstalk, complemented by Tensor-cell2cell to identify prevalent patterns of potential communication across the various experimental conditions. We next probed the impact of Axl inhibition on intercellular communication with orthogonalized partial least squares discriminant modeling. Finally, we conducted gene set enrichment analysis to understand how each cell type’s state and signaling activity was impacted by each axis of experimental variation in our complex experimental design. Tensor-cell2cell yielded eight patterns of communication in the inferred intercellular interaction data, and these eight patterns were paired based on their communicating cell types as well as their shared top ligand-receptor interactions. These pairings demonstrated the impact of including dendritic cells in the model system and highlighted coordination among the myeloid cells. The specific effects of Axl inhibition on potential intercellular crosstalk were clarified with supervised modeling, demonstrating impaired tumor cell invasion and migration programs, as expected. Intriguingly, gene set enrichment analysis revealed that tumor cell prior injury with standard-of-care therapy impacted myeloid cell interferon signaling in response to Axl inhibition. In this work, we demonstrated that tumor microenvironment cellular composition as well as prior treatment with standard-of-care therapy modulate myeloid cell response to Axl inhibition in human in vitro melanoma model systems. While the inferred effects of Axl inhibition on intercellular crosstalk, cell state, and cell activity were largely consistent with prior literature, the impact of tumor cell prior injury on the myeloid interferon signaling response to Axl inhibition underscores the importance of intentionally incorporating treatment history as a design parameter when building model systems. Ultimately, complex human in vitro model systems that consider the nuanced effects of prior therapy on the tumor microenvironment can complement insights gained from pre-clinical mouse studies to better evaluate and iteratively design candidate therapies.

Protein tyrosine kinases activate signaling pathways by catalyzing the phosphorylation of tyrosine residues in their substrates. Mounting evidence suggests that, in addition to recognizing phosphorylated tyrosine (pTyr) residues through specific phosphobinding modules, many protein kinases selectively recognize pTyr directly adjacent to the tyrosine residue they phosphorylate and catalyze the formation of twin pTyr-pTyr sites. Here, we demonstrate the importance of this phosphopriming-driven twin pTyr signaling in promoting cell cycle progression through the cell cycle-inhibitory protein p27Kip1. We identify, structurally resolve, and tune two distinct molecular determinants driving the selective recognition of pTyr directly N- and C-terminal to the target phospho-acceptor tyrosine site. We further show structural and biochemical conservation in this recognition, and identify cancer-associated alterations to these determinants that are unable to recognize phosphoprimed substrates. Finally, using an in vivo mouse model of leukemia we show that Bcr-Abl mutants unable to recognize phosphoprimed substrates paradoxically result in enhanced tumor development and progression. These data indicate that Bcr-Abl, like other proto-oncogenes such as Ras or Myc, engages both pro- and anti-oncogenic programs - but in the case of Bcr-Abl, this is accomplished through a mechanism involving traditional and phosphoprimed substrate recognition.

The antibody-drug conjugate (ADC) Mirvetuximab soravtansine (Mirv) has emerged as a powerful tool for suppressing and delaying the progression of epithelial ovarian cancer, including high-grade serous ovarian cancer (HGSOC). This ADC targets ovarian cancer cells based on their elevated expression of the folate receptor alpha (FRα) and, upon internalization, delivers a potent microtubule (MT) polymerization inhibitor (DM4). Mirv, however, is approved only for patients with high FRα expression and the durability of the response is finite as a consequence of drug resistance. We discovered that inhibitors of Polo-like kinase 1 (Plk1), a critical mitotic kinase, synergistically kill a variety of tumor cell types, including many HGSOC cell lines, when combined with microtubule polymerization inhibitors, including DM4 the payload of Mirv. This synergistic tumor cell death results from a cancer cell-specific mitotic vulnerability, leading to drug combination-specific mitotic spindle defects that cause mitotic arrest and subsequent spindle assembly checkpoint-dependent cell death. The efficacy of this synergistic drug combination was further validated in human patient-derived HGSOC organoids, and in an in vivo human HGSOC murine xenograft model. To elucidate the molecular mechanism underlying this Plk1i/anti-microtubule drug synergy, we performed a genome-wide CRISPR-A screen, implicating specific molecular components of the kinetochore (KT) as the targets responsible for synergistic killing. To further validate this molecular mechanism, we report here that this drug combination potently disrupts KT-MT attachments due to the loss of Plk1-dependent kinetochore protein phosphorylation required for optimal MT capture. These data are consistent with a model in which inhibition of Plk1 activity compromises KT function, rendering cancer cells hyper-reliant on MT dynamics to achieve the proper KT-MT attachments that are essential for accurate chromosome segregation. Our work indicates that targeting Plk1 could enhance and extend the durability of clinical response in HGSOC patients treated with Mirv and possibly expand the utility of Mirv to patients with lower FRα expression. To translate these findings into the clinic, we have acquired funding to support an investigator-sponsored phase 1 clinical trial testing this combination in ovarian cancer patients through the MIT — Dana-Farber/Harvard Cancer Center Bridge Program. This trial utilizes the ADC Mirv, currently FDA approved for ovarian cancer treatment, in combination with the Plk1 inhibitor Onvansertib, which has been used in multiple prior and ongoing clinical trials with an established safety profile. The trial will recruit patients planning to initiate Mirv with the addition of Onvansertib, using a BOIN dose escalation strategy. The primary endpoint is safety with secondary endpoints assessing radiographic responses and molecular correlations of any clinical responses. Jesse C. Patterson, Nabihah Tayob, Robin Lu, Yi Wen Kong, Brian A. Joughin, Eugene Kim, Bo Rueda, Joyce F. Liu, Ursula A. Matulonis, Meghan Shea, Michael B. Yaffe. Targeting a cancer-specific mitotic vulnerability in ovarian cancer by combining Mirvetuximab soravtansine with inhibitors of Polo-like kinase 1 [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Ovarian Cancer Research; 2025 Sep 19-21; Denver, CO. Philadelphia (PA): AACR; Cancer Res 2025;85(18_Suppl):Abstract nr A042.

Phosphorylation of proteins on tyrosine (Tyr) residues evolved in metazoan organisms as a mechanism of coordinating tissue growth1. Multicellular eukaryotes typically have more than 50 distinct protein Tyr kinases that catalyse the phosphorylation of thousands of Tyr residues throughout the proteome1-3. How a given Tyr kinase can phosphorylate a specific subset of proteins at unique Tyr sites is only partially understood4-7. Here we used combinatorial peptide arrays to profile the substrate sequence specificity of all human Tyr kinases. Globally, the Tyr kinases demonstrate considerable diversity in optimal patterns of residues surrounding the site of phosphorylation, revealing the functional organization of the human Tyr kinome by substrate motif preference. Using this information, Tyr kinases that are most compatible with phosphorylating any Tyr site can be identified. Analysis of mass spectrometry phosphoproteomic datasets using this compendium of kinase specificities accurately identifies specific Tyr kinases that are dysregulated in cells after stimulation with growth factors, treatment with anti-cancer drugs or expression of oncogenic variants. Furthermore, the topology of known Tyr signalling networks naturally emerged from a comparison of the sequence specificities of the Tyr kinases and the SH2 phosphotyrosine (pTyr)-binding domains. Finally we show that the intrinsic substrate specificity of Tyr kinases has remained fundamentally unchanged from worms to humans, suggesting that the fidelity between Tyr kinases and their protein substrate sequences has been maintained across hundreds of millions of years of evolution.

Homologous Recombination (HR) is a high-fidelity repair mechanism of DNA Double-Strand Breaks (DSBs), which are induced by irradiation, genotoxic chemicals or physiological DNA damaging processes. DSBs are also generated as intermediates during the repair of interstrand crosslinks (ICLs). In this context, the Fanconi anemia (FA) core complex, which is effectively recruited to ICLs, promotes HR-mediated DSB-repair. However, whether the FA core complex also promotes HR at ICL-independent DSBs remains controversial. Here, we identified the FA core complex members FANCL and Ube2T as HR-promoting factors in a CRISPR/Cas9-based screen with cells carrying the DSB-repair reporter DSB-Spectrum. Using isogenic cell-line models, we validated the HR-function of FANCL and Ube2T, and demonstrated a similar function for their ubiquitination-substrate FANCD2. We further show that FANCL and Ube2T are directly recruited to DSBs and are required for the accumulation of FANCD2 at these break sites. Mechanistically, we demonstrate that FANCL ubiquitin ligase activity is required for the accumulation of the nuclease CtIP at DSBs, and consequently for optimal end-resection and Rad51 loading. CtIP overexpression rescues HR in FANCL-deficient cells, validating that FANCL primarily regulates HR by promoting CtIP recruitment. Together, these data demonstrate that the FA core complex and FANCD2 have a dual genome maintenance function by promoting repair of DSBs as well as the repair of ICLs.

GSEA plots of carcinogenesis gene sets in SRC compared with NEP.

Data collection and refinement statistics

GSEA comparing SRC to NEP using gene sets for parenchymal gastric cells.

Representative slides at 20x magnification from every IHC specimen used for quantification.

Figure S1 shows that pHLIP localization to the acidic tissues in vivo is pH-dependent and pHLIP- CA9 colocalization in lung mets and enrichment of pHLIP and CA9 at tumor stroma interface in mouse and human tumors

Abstract KRAS is the most frequently mutated oncogene. The incidence of specific KRAS alleles varies between cancers from different sites, but it is unclear whether allelic selection results from biological selection for specific mutant KRAS proteins. We used a cross-disciplinary approach to compare KRASG12D, a common mutant form, and KRASA146T, a mutant that occurs only in selected cancers. Biochemical and structural studies demonstrated that KRASA146T exhibits a marked extension of switch 1 away from the protein body and nucleotide binding site, which activates KRAS by promoting a high rate of intrinsic and guanine nucleotide exchange factor–induced nucleotide exchange. Using mice genetically engineered to express either allele, we found that KRASG12D and KRASA146T exhibit distinct tissue-specific effects on homeostasis that mirror mutational frequencies in human cancers. These tissue-specific phenotypes result from allele-specific signaling properties, demonstrating that context-dependent variations in signaling downstream of different KRAS mutants drive the KRAS mutational pattern seen in cancer. Significance: Although epidemiologic and clinical studies have suggested allele-specific behaviors for KRAS, experimental evidence for allele-specific biological properties is limited. We combined structural biology, mass spectrometry, and mouse modeling to demonstrate that the selection for specific KRAS mutants in human cancers from different tissues is due to their distinct signaling properties. See related commentary by Hobbs and Der, p. 696. This article is highlighted in the In This Issue feature, p. 681