A comprehensive single-cell breast tumor atlas defines epithelial and immune heterogeneity and interactions predicting anti-PD-1 therapy response.

We present an integrated single-cell RNA sequencing atlas of the primary breast tumor microenvironment (TME) containing 236,363 cells from 119 biopsy samples across eight datasets. In this study, we leverage this resource for multiple analyses of immune and cancer epithelial cell heterogeneity. We define natural killer (NK) cell heterogeneity through six subsets in the breast TME. Because NK cell heterogeneity correlates with epithelial cell heterogeneity, we characterize epithelial cells at the level of single-gene expression, molecular subtype, and 10 categories reflecting intratumoral transcriptional heterogeneity. We develop InteractPrint, which considers how cancer epithelial cell heterogeneity influences cancer-immune interactions. We use T cell InteractPrint to predict response to immune checkpoint inhibition (ICI) in two breast cancer clinical trials testing neoadjuvant anti-PD-1 therapy. T cell InteractPrint was predictive of response in both trials versus PD-L1 (AUC = 0.82, 0.83 vs. 0.50, 0.72). This resource enables additional high-resolution investigations of the breast TME.

Histopathology and levels of proteins in plasma associate with survival after colorectal cancer diagnosis.

BACKGROUND: The TNM system is used to assess prognosis after colorectal cancer (CRC) diagnosis. Other prognostic factors reported include histopathological assessments of the tumour, tumour mutations and proteins in the blood. As some of these factors are strongly correlated, it is important to evaluate the independent effects they may have on survival. METHODS: Tumour samples from 2162 CRC patients were visually assessed for amount of tumour stroma, severity of lymphocytic infiltrate at the tumour margins and the presence of lymphoid follicles. Somatic mutations in the tumour were assessed for 2134 individuals. Pre-surgical levels of 4963 plasma proteins were measured in 128 individuals. The associations between these features and prognosis were inspected by a Cox Proportional Hazards Model (CPH). RESULTS: Levels of stroma, lymphocytic infiltration and presence of lymphoid follicles all associate with prognosis, along with high tumour mutation burden, high microsatellite instability and TP53 and BRAF mutations. The somatic mutations are correlated with the histopathology and none of the somatic mutations associate with survival in a multivariate analysis. Amount of stroma and lymphocytic infiltration associate with local invasion of tumours. Elevated levels of two plasma proteins, CA-125 and PPP1R1A, associate with a worse prognosis. CONCLUSIONS: Tumour stroma and lymphocytic infiltration variables are strongly associated with prognosis of CRC and capture the prognostic effects of tumour mutation status. CA-125 and PPP1R1A may be useful prognostic biomarkers in CRC.

Morphology-guided transcriptomic analysis of human pancreatic cancer organoids reveals microenvironmental signals that enhance invasion.

Pancreatic ductal adenocarcinoma (PDAC) frequently presents with metastasis, but the molecular programs in human PDAC cells that drive invasion are not well understood. Using an experimental pipeline enabling PDAC organoid isolation and collection based on invasive phenotype, we assessed the transcriptomic programs associated with invasion in our organoid model. We identified differentially expressed genes in invasive organoids compared with matched noninvasive organoids from the same patients, and we confirmed that the encoded proteins were enhanced in organoid invasive protrusions. We identified 3 distinct transcriptomic groups in invasive organoids, 2 of which correlated directly with the morphological invasion patterns and were characterized by distinct upregulated pathways. Leveraging publicly available single-cell RNA-sequencing data, we mapped our transcriptomic groups onto human PDAC tissue samples, highlighting differences in the tumor microenvironment between transcriptomic groups and suggesting that non-neoplastic cells in the tumor microenvironment can modulate tumor cell invasion. To further address this possibility, we performed computational ligand-receptor analysis and validated the impact of multiple ligands (TGF-ß1, IL-6, CXCL12, MMP9) on invasion and gene expression in an independent cohort of fresh human PDAC organoids. Our results identify molecular programs driving morphologically defined invasion patterns and highlight the tumor microenvironment as a potential modulator of these programs.

Triple negative breast tumors contain heterogeneous cancer cells expressing distinct KRAS-dependent collective and disseminative invasion programs.

Inter-patient and intra-tumoral heterogeneity complicate the identification of predictive biomarkers and effective treatments for basal triple negative breast cancer (b-TNBC). Invasion is the initiating event in metastasis and can occur by both collective and single-cell mechanisms. We cultured primary organoids from a b-TNBC genetically engineered mouse model in 3D collagen gels to characterize their invasive behavior. We observed that organoids from the same tumor presented different phenotypes that we classified as non-invasive, collective and disseminative. To identify molecular regulators driving these invasive phenotypes, we developed a workflow to isolate individual organoids from the collagen gels based on invasive morphology and perform RNA sequencing. We next tested the requirement of differentially regulated genes for invasion using shRNA knock-down. Strikingly, KRAS was required for both collective and disseminative phenotypes. We then performed a drug screen targeting signaling nodes upstream and downstream of KRAS. We found that inhibition of EGFR, MAPK/ERK, or PI3K/AKT signaling reduced invasion. Of these, ERK inhibition was striking for its ability to potently inhibit collective invasion and dissemination. We conclude that different cancer cells in the same b-TNBC tumor can express different metastatic molecular programs and identified KRAS and ERK as essential regulators of collective and single cell dissemination.

Hsa-miR-21-3p associates with breast cancer patient survival and targets genes in tumor suppressive pathways.

Breast cancer is the cancer most often diagnosed in women. MicroRNAs (MIRs) are short RNA molecules that bind mRNA resulting in their downregulation. MIR21 has been shown to be an oncomiR in most cancer types, including breast cancer. Most of the effects of miR-21 have been attributed to hsa-miR-21-5p that is transcribed from the leading strand of MIR21, but hsa-miR-21-3p (miR-21-3p), transcribed from the lagging strand, is much less studied. The aim of the study is to analyze whether expression of miR-21-3p is prognostic for breast cancer. MiR-21-3p association with survival, clinical and pathological characteristics was analyzed in a large breast cancer cohort and validated in three separate cohorts, including TCGA and METABRIC. Analytical tools were also used to infer miR-21-3p function and to identify potential target genes and functional pathways. The results showed that in the exploration cohort, high miR-21-3p levels associated with shorter survival and lymph node positivity. In the three validation cohorts, high miR-21-3p levels associated with pathological characteristics that predict worse prognosis. Specifically, in the largest validation cohort, METABRIC (n = 1174), high miR-21-3p levels associated with large tumors, a high grade, lymph node and HER2 positivity, and shorter breast-cancer-specific survival (HR = 1.38, CI 1.13-1.68). This association remained significant after adjusting for confounding factors. The genes with expression levels that correlated with miR-21-3p were enriched in particular pathways, including the epithelial-to-mesenchymal transition and proliferation. Among the most significantly downregulated targets were MAT2A and the tumor suppressive genes STARD13 and ZNF132. The results from this study emphasize that both 3p- and 5p-arms from a MIR warrant independent study. The data show that miR-21-3p overexpression in breast tumors is a marker of worse breast cancer progression and it affects genes in pathways that drive breast cancer by down-regulating tumor suppressor genes. The results suggest miR-21-3p as a potential biomarker.

Inhibition of phosphodiesterase type 9 reduces obesity and cardiometabolic syndrome in mice.

Central obesity with cardiometabolic syndrome (CMS) is a major global contributor to human disease, and effective therapies are needed. Here, we show that cyclic GMP-selective phosphodiesterase 9A inhibition (PDE9-I) in both male and ovariectomized female mice suppresses preestablished severe diet-induced obesity/CMS with or without superimposed mild cardiac pressure load. PDE9-I reduces total body, inguinal, hepatic, and myocardial fat; stimulates mitochondrial activity in brown and white fat; and improves CMS, without significantly altering activity or food intake. PDE9 localized at mitochondria, and its inhibition in vitro stimulated lipolysis in a PPARα-dependent manner and increased mitochondrial respiration in both adipocytes and myocytes. PPARα upregulation was required to achieve the lipolytic, antiobesity, and metabolic effects of PDE9-I. All these PDE9-I-induced changes were not observed in obese/CMS nonovariectomized females, indicating a strong sexual dimorphism. We found that PPARα chromatin binding was reoriented away from fat metabolism-regulating genes when stimulated in the presence of coactivated estrogen receptor-α, and this may underlie the dimorphism. These findings have translational relevance given that PDE9-I is already being studied in humans for indications including heart failure, and efficacy against obesity/CMS would enhance its therapeutic utility.

Myocardial Gene Expression Signatures in Human Heart Failure With Preserved Ejection Fraction.

BACKGROUND: Heart failure (HF) with preserved ejection fraction (HFpEF) constitutes half of all HF but lacks effective therapy. Understanding of its myocardial biology remains limited because of a paucity of heart tissue molecular analysis. METHODS: We performed RNA sequencing on right ventricular septal endomyocardial biopsies prospectively obtained from patients meeting consensus criteria for HFpEF (n=41) contrasted with right ventricular septal tissue from patients with HF with reduced ejection fraction (HFrEF, n=30) and donor controls (n=24). Principal component analysis and hierarchical clustering tested for transcriptomic distinctiveness between groups, effect of comorbidities, and differential gene expression with pathway enrichment contrasted HF groups and donor controls. Within HFpEF, non-negative matrix factorization and weighted gene coexpression analysis identified molecular subgroups, and the resulting clusters were correlated with hemodynamic and clinical data. RESULTS: Patients with HFpEF were more often women (59%), African American (68%), obese (median body mass index 41), and hypertensive (98%), with clinical HF characterized by 65% New York Heart Association Class III or IV, nearly all on a loop diuretic, and 70% with a HF hospitalization in the previous year. Principal component analysis separated HFpEF from HFrEF and donor controls with minimal overlap, and this persisted after adjusting for primary comorbidities: body mass index, sex, age, diabetes, and renal function. Hierarchical clustering confirmed group separation. Nearly half the significantly altered genes in HFpEF versus donor controls (1882 up, 2593 down) changed in the same direction in HFrEF; however, 5745 genes were uniquely altered between HF groups. Compared with controls, uniquely upregulated genes in HFpEF were enriched in mitochondrial adenosine triphosphate synthesis/electron transport, pathways downregulated in HFrEF. HFpEF-specific downregulated genes engaged endoplasmic reticulum stress, autophagy, and angiogenesis. Body mass index differences largely accounted for HFpEF upregulated genes, whereas neither this nor broader comorbidity adjustment altered pathways enriched in downregulated genes. Non-negative matrix factorization identified 3 HFpEF transcriptomic subgroups with distinctive pathways and clinical correlates, including a group closest to HFrEF with higher mortality, and a mostly female group with smaller hearts and proinflammatory signaling. These groupings remained after sex adjustment. Weighted gene coexpression analysis yielded analogous gene clusters and clinical groupings. CONCLUSIONS: HFpEF exhibits distinctive broad transcriptomic signatures and molecular subgroupings with particular clinical features and outcomes. The data reveal new signaling targets to consider for precision therapeutics.

Cancer cells educate natural killer cells to a metastasis-promoting cell state.

Natural killer (NK) cells have potent antitumor and antimetastatic activity. It is incompletely understood how cancer cells escape NK cell surveillance. Using ex vivo and in vivo models of metastasis, we establish that keratin-14+ breast cancer cells are vulnerable to NK cells. We then discovered that exposure to cancer cells causes NK cells to lose their cytotoxic ability and promote metastatic outgrowth. Gene expression comparisons revealed that healthy NK cells have an active NK cell molecular phenotype, whereas tumor-exposed (teNK) cells resemble resting NK cells. Receptor-ligand analysis between teNK cells and tumor cells revealed multiple potential targets. We next showed that treatment with antibodies targeting TIGIT, antibodies targeting KLRG1, or small-molecule inhibitors of DNA methyltransferases (DMNT) each reduced colony formation. Combinations of DNMT inhibitors with anti-TIGIT or anti-KLRG1 antibodies further reduced metastatic potential. We propose that NK-directed therapies targeting these pathways would be effective in the adjuvant setting to prevent metastatic recurrence.

Human primary liver cancer organoids reveal intratumor and interpatient drug response heterogeneity.

Liver cancer is the fourth leading cause of cancer-related mortality and is distinguished by a relative paucity of chemotherapy options. It has been hypothesized that intratumor genetic heterogeneity may contribute to the high failure rate of chemotherapy. Here, we evaluated functional heterogeneity in a cohort of primary human liver cancer organoid lines. Each primary human liver cancer surgical specimen was used to generate multiple cancer organoid lines, obtained from distinct regions of the tumor. A total of 27 liver cancer lines were established and tested with 129 cancer drugs, generating 3,483 cell survival data points. We found a rich intratumor, functional (drug response) heterogeneity in our liver cancer patients. Furthermore, we established that the majority of drugs were either ineffective, or effective only in select organoid lines. In contrast, we found that a subset of drugs appeared pan-effective, displaying at least moderate activity in the majority of these cancer organoid lines. These drugs, which are FDA approved for indications other than liver cancers, deserve further consideration as either systemic or local therapeutics. Of note, molecular profiles, obtained for a reduced sample set, did not correlate with the drug response heterogeneity of liver cancer organoid lines. Taken together, these findings lay the foundation for in-depth studies of pan-effective drugs, as well as for functional personalized oncology approaches. Lastly, these functional studies demonstrate the utility of cancer organoid drug testing as part of a drug discovery pipeline.

Polarization and migration in the zebrafish posterior lateral line system.

Collective cell migration plays an important role in development. Here, we study the posterior lateral line primordium (PLLP) a group of about 100 cells, destined to form sensory structures, that migrates from head to tail in the zebrafish embryo. We model mutually inhibitory FGF-Wnt signalling network in the PLLP and link tissue subdivision (Wnt receptor and FGF receptor activity domains) to receptor-ligand parameters. We then use a 3D cell-based simulation with realistic cell-cell adhesion, interaction forces, and chemotaxis. Our model is able to reproduce experimentally observed motility with leading cells migrating up a gradient of CXCL12a, and trailing (FGF receptor active) cells moving actively by chemotaxis towards FGF ligand secreted by the leading cells. The 3D simulation framework, combined with experiments, allows an investigation of the role of cell division, chemotaxis, adhesion, and other parameters on the shape and speed of the PLLP. The 3D model demonstrates reasonable behaviour of control as well as mutant phenotypes.

3-D individual cell based computational modeling of tumor cell-macrophage paracrine signaling mediated by EGF and CSF-1 gradients.

High density of macrophages in mammary tumors has been associated with a higher risk of metastasis and thus increased mortality in women. The EGF/CSF-1 paracrine signaling increases the number of invasive tumor cells by both recruiting tumor cells further away and manipulating the macrophages' innate ability to open up a passage into blood vessels thus promoting intravasation and finally metastasis. A 3-D individual-cell-based model is introduced, to better understand the tumor cell-macrophage interactions, and to explore how changing parameters of the paracrine signaling system affects the number of invasive tumor cells. The simulation data and videos of the cell movements correlated well with findings from both in vitro and in vivo experimental results. The model demonstrated how paracrine signaling is necessary to achieve co-migration of tumor cells and macrophages towards a specific signaling source. We showed how the paracrine signaling enhances the number of both invasive tumor cells and macrophages. The simulations revealed that for the in vitro experiments the imposed no-flux boundary condition might be affecting the results, and that changing the setup might lead to different experimental findings. In our simulations, the 3 : 1 tumor cell/macrophage ratio, observed in vivo, was robust for many parameters but sensitive to EGF signal strength and fraction of macrophages in the tumor. The model can be used to identify new agents for targeted therapy and we suggest that a successful strategy to prevent or limit invasion of tumor cells would be to block the tumor cell-macrophage paracrine signaling. This can be achieved by either blocking the EGF or CSF-1 receptors or supressing the EGF or CSF-1 signal.

Mathematical model of macrophage-facilitated breast cancer cells invasion.

Mortality from breast cancer stems from its tendency to invade into surrounding tissues and organs. Experiments have shown that this metastatic process is facilitated by macrophages in a short-ranged chemical signalling loop. Macrophages secrete epidermal growth factor, EGF, and respond to the colony stimulating factor 1, CSF-1. Tumor cells secrete CSF-1 and respond to EGF. In this way, the cells coordinate aggregation and cooperative migration. Here we investigate this process in a model for in vitro interactions using two distinct but related mathematical approaches. In the first, we analyze and simulate a set of partial differential equations to determine conditions for aggregation. In the second, we use a cell-based discrete 3D simulation to follow the fates and motion of individual cells during aggregation. Linear stability analysis of the PDE model reveals that decreasing the chemical secretion, chemotaxis coefficients or density of cells or increasing the chemical degradation in the model could eliminate the spontaneous aggregation of cells. Simulations with the discrete model show that the ratio between tumor cells and macrophages in aggregates increases when the EGF secretion parameter is increased. The results also show how CSF-1/CSF-1R autocrine signalling in tumor cells affects the ratio between the two cell types. Comparing the continuum results with simulations of a discrete cell-based model, we find good qualitative agreement.