Faculty Bibliography

The paucity of genetically informed, immune-competent tumor models impedes evaluation of conventional, targeted, and immune therapies. By engineering mouse fallopian tube epithelial organoids using lentiviral gene transduction and/or CRISPR/Cas9 mutagenesis, we generated multiple high grade serous tubo-ovarian carcinoma (HGSC) models exhibiting mutational combinations seen in HGSC patients. Detailed analysis of homologous recombination (HR)-proficient (Tp53-/-;Ccne1OE;Akt2OE ;KrasOE), HR-deficient (Tp53-/-;Brca1-/-;MycOE), and unclassified (Tp53-/-;Pten-/-;Nf1-/-) organoids revealed differences in in vitro properties (proliferation, differentiation, "secretome"), copy number aberrations, and tumorigenicity. Tumorigenic organoids had variable sensitivity to HGSC chemotherapeutics, evoked distinct immune microenvironments that could be modulated by neutralizing organoid-produced chemokines/cytokines. These findings enabled development of a chemotherapy/immunotherapy regimen that yielded durable, T-cell dependent responses in Tp53-/-;Ccne1OE;Akt2OE;Kras HGSC; by contrast, Tp53-/-;Pten-/-;Nf1-/- tumors failed to respond. Mouse and human HGSC models showed genotype-dependent similarities in chemosensitivity, secretome, and immune microenvironment. Genotype-informed, syngeneic organoid models could provide a platform for the rapid evaluation of tumor biology and therapeutics.

Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer death in the United States. Glycans, such as carbohydrate antigen 19-9, are biomarkers of PDAC and are emerging as important modulators of cancer phenotypes. Herein, we used a systems-based approach integrating glycomic analysis of the well-established KC mouse, which models early events in transformation, and analysis of samples from human pancreatic cancer patients to identify glycans with potential roles in cancer formation. We observed both common and distinct patterns of glycosylation in pancreatic cancer across species. Common alterations included increased levels of α-2,3-sialic acid and α-2,6-sialic acid, bisecting GlcNAc and poly-N-acetyllactosamine. However, core fucose, which was increased in human PDAC, was not seen in the mouse, indicating that not all human glycomic changes are observed in the KC mouse model. In silico analysis of bulk and single-cell sequencing data identified ST6 beta-galactoside alpha-2,6-sialyltransferase 1, which underlies α-2,6-sialic acid, as overexpressed in human PDAC, concordant with histological data showing higher levels of this enzyme at the earliest stages. To test whether ST6 beta-galactoside alpha-2,6-sialyltransferase 1 promotes pancreatic cancer, we created a novel mouse in which a pancreas-specific genetic deletion of this enzyme overlays the KC mouse model. The analysis of our new model showed delayed cancer formation and a significant reduction in fibrosis. Our results highlight the importance of a strategic systems approach to identifying glycans whose functions can be modeled in mouse, a crucial step in the development of therapeutics targeting glycosylation in pancreatic cancer.

Aims/UNASSIGNED:Histology, the microscopic study of normal tissues, is a crucial element of most medical curricula. Learning tools focused on histology are very important to learners who seek diagnostic competency within this important diagnostic arena. Recent developments in machine learning (ML) suggest that certain ML tools may be able to benefit this histology learning platform. Here, we aim to explore how one such tool based on a convolutional neural network, can be used to build a generalizable multi-classification model capable of classifying microscopic images of human tissue samples with the ultimate goal of providing a differential diagnosis (a list of look-alikes) for each entity. Methods/UNASSIGNED:We obtained three institutional training datasets and one generalizability test dataset, each containing images of histologic tissues in 38 categories. Models were trained on data from single institutions, low quantity combinations of multiple institutions, and high quantity combinations of multiple institutions. Models were tested against withheld validation data, external institutional data, and generalizability test images obtained from Google image search. Performance was measured with macro and micro accuracy, sensitivity, specificity, and f1-score. Results/UNASSIGNED:In this study, we were able to show that such a model's generalizability is dependent on both the training data source variety and the total number of training images used. Models which were trained on 760 images from only a single institution performed well on withheld internal data but poorly on external data (lower generalizability). Increasing data source diversity improved generalizability, even when decreasing data quantity: models trained on 684 images, but from three sources improved generalization accuracy between 4.05% and 18.59%. Maintaining this diversity and increasing the quantity of training images to 2280 further improved generalization accuracy between 16.51% and 32.79%. Conclusions/UNASSIGNED:This pilot study highlights the significance of data diversity within such studies. As expected, optimal models are those that incorporate both diversity and quantity into their platforms.s.

In-situ hybridization (ISH) analysis is a highly desirable, versatile approach for assessing biomarker expression status in a spatial context. Most researchers rely on immunostaining (protein targets) or qPCR (mRNA). However, not all proteins can be immunolabeled due to a lack of well-validated antibodies. The qPCR approach, although highly specific, cannot provide spatial information. RNAscope employs a unique double Z probe that has to bind to the target RNA in tandem in order to be recognized by the preamplifiers and amplifiers. A fluorescent/chromogenic labeled probe then binds to the multiple binding sites of the amplifiers, which improves detection of low expressing RNA and reduces non-specific binding. RNAscope replaces cumbersome radioactive and chromogenic ISH with more hassle-free chromogen and fluorescence-labelled probes. At the NYULMC Experimental Pathology Core we have integrated RNAscope with Polaris multispectral imaging and quantitative analysis using different software platforms. About 21 laboratories have used this workflow to address their specific questions. We have also established and validated the newer BaseScopeTM assay. In contrast to RNAscope, which targets lncRNA and mRNA sequences greater than 300nt, BaseScopeTM enables detection of short RNA target sequences between 50-300nt. It can be used to detect exon junctions/splice variants, circular RNA, pre-miRNA, and point mutations. We adapted BaseScopeTM to co-detect circular RNA and its linear counterpart in a differentiating cell population, which could not be established on glass chamber slides and had to be stained on a plastic petri dish. In conclusion, RNAscope and BaseScopeTM RNA-ISH are powerful alternative strategies for assessing the spatial distribution of critical biomarkers within intact tissues and cells. This approach coupled with sophisticated imaging modalities and downstream analysis support provides new collaborative opportunities for Core aboratories.
Copyright

Despite remarkable responses to cancer immunotherapy in a subset of patients, many patients remain resistant to therapies. It is now clear that elevated levels of tumor-infiltrating T cells as well as a systemic anti-tumor immune response are requirements for successful immunotherapies. However, the tumor microenvironment imposes an additional resistance mechanism to immunotherapy. We have developed a practical and improved strategy for cancer immunotherapy using an oncolytic virus and anti-OX40. This strategy takes advantage of a preexisting T cell immune repertoire in vivo, removing the need to know about present tumor antigens. We have shown in this study that the replication-deficient oncolytic Sindbis virus vector expressing interleukin-12 (IL-12) (SV.IL12) activates immune-mediated tumor killing by inducing OX40 expression on CD4 T cells, allowing the full potential of the agonistic anti-OX40 antibody. The combination of SV.IL12 with anti-OX40 markedly changes the transcriptome signature and metabolic program of T cells, driving the development of highly activated terminally differentiated effector T cells. These metabolically reprogrammed T cells demonstrate enhanced tumor infiltration capacity as well as anti-tumor activity capable of overcoming the repressive tumor microenvironment. Our findings identify SV.IL12 in combination with anti-OX40 to be a novel and potent therapeutic strategy that can cure multiple types of low-immunogenic solid tumors.

A subset of B cell acute lymphoblastic leukemia (B-ALL) patients will relapse and succumb to therapy-resistant disease. The bone marrow microenvironment may support B-ALL progression and treatment evasion. Utilizing single-cell approaches, we demonstrate B-ALL bone marrow immune microenvironment remodeling upon disease initiation and subsequent re-emergence during conventional chemotherapy. We uncover a role for non-classical monocytes in B-ALL survival, and demonstrate monocyte abundance at B-ALL diagnosis is predictive of pediatric and adult B-ALL patient survival. We show that human B-ALL blasts alter a vascularized microenvironment promoting monocytic differentiation, while depleting leukemia-associated monocytes in B-ALL animal models prolongs disease remission in vivo. Our profiling of the B-ALL immune microenvironment identifies extrinsic regulators of B-ALL survival supporting new immune-based therapeutic approaches for high-risk B-ALL treatment.

Programmed cell death protein 1 (PD-1) ligation delimits immunogenic responses in T cells. However, the consequences of programmed cell death 1 ligand 1 (PD-L1) ligation in T cells are uncertain. We found that T cell expression of PD-L1 in cancer was regulated by tumor antigen and sterile inflammatory cues. PD-L1⁺ T cells exerted tumor-promoting tolerance via three distinct mechanisms: (1) binding of PD-L1 induced STAT3-dependent 'back-signaling' in CD4⁺ T cells, which prevented activation, reduced T_H1-polarization and directed T_H17-differentiation. PD-L1 signaling also induced an anergic T-bet^-IFN-γ^- phenotype in CD8⁺ T cells and was equally suppressive compared to PD-1 signaling; (2) PD-L1⁺ T cells restrained effector T cells via the canonical PD-L1-PD-1 axis and were sufficient to accelerate tumorigenesis, even in the absence of endogenous PD-L1; (3) PD-L1⁺ T cells engaged PD-1⁺ macrophages, inducing an alternative M2-like program, which had crippling effects on adaptive antitumor immunity. Collectively, we demonstrate that PD-L1⁺ T cells have diverse tolerogenic effects on tumor immunity.

An increasing number of technologies combine spatial discrimination with molecule identification, as has been done for decades through immunohistochemistry and in situ hybridization. The difference is that now a sleuth of commercially available platforms promises to significantly increase the number of targets that can be identified on a single sample. They also provide higher throughput in terms of number of samples that can be interrogated, and some offer semiautomation with straightforward sample preparation. All these characteristics set up the new spatial profiling technologies as attractive candidates for biomedical research facilities. However, to be able to discriminate between them it is necessary to share real world experiences using them in our cores. This panel session brings together several researchers with hands on experience using two or more spatial profiling technologies. These will include antibody based (Opal/Vectra Polaris, Ultivue, MIBI, Hyperion) and nucleic acid (Visium, GeoMx) based platforms. Each panelist will give a quick overview of the technology, its biochemical principles, costs, advantages and disadvantages as they have identified them and compared them to each other or a gold standard, if available. At the end, we will have an open discussion where the audience is encouraged to ask questions and provide their experiences

Several modalities of multiplex immunofluorescence histology currently available require significant time and resources to implement. Many research laboratories develop questions benefiting from multiplex staining and analysis but do not have the human resources and/or equipment to perform the assay. Furthermore, pilot studies use similar metrics to evaluate multiplex histology data making them ideal for a core laboratory setup. The objective of this study was to establish a semi-automated workflow for multiplex immunofluorescence staining and initial quantification of cell populations in whole slide microscopy scans. The requirements for the workflow included: A. minimal transfer of decision making from the researcher to core personnel (Semi-Automation), B. modifiable in terms of antigen targets and tissue types with minimal disruption to the process and C. reproducible across samples submitted at different time periods (eg patient samples).

Identification of biomarkers is a major goal of personalized medicine. Large transcriptome screens have identified new targets and molecular signatures for disease sub-types. However, tissue spatial information, which fundamentally alters in vivo cell behavior and gene expression, is lost. To understand spatial context and validate bulk tissue screens, most researchers rely exclusively on antibodies and immunostaining assays. Unfortunately, this is either not the appropriate choice for some targets, such as long non-coding RNAs, or it is not feasible because no reliable antibodies exist. To address these issues, we have established an alternative work-flow incorporating RNA in situ hybridization (CISH and FISH), whole slide and/or multispectral scanning and image analysis. Methods: RNAscope technology; Leica SCN scanner or Vectra3 multispectral imaging system for image acquisition; ImageJ2/FIJI, R, InForm and Visiopharm software platforms for quantitative analysis. Results: Several laboratories have used this workflow to address their specific questions. For example, we established and validated RNAscope assays for signaling factor transcripts, which are now integrated into an ongoing clinical trial. In this case, all tested commercial antibodies failed the validation assay. We also assessed expression of LNC RNAs in prostate cancer and put in place protocols for normalizing probe quantification across samples. The analysis revealed that storage and/or sample preparation affected the detection of certain LNC RNAs more than others identifying important factors regarding banking specimens. Spatial heat map visualization of RNAscope probes revealed an unexpected distribution of inflammatory cytokine targets in the kidney which are now being further investigated. In conclusion, RNA ISH is a powerful alternative strategy for assessing the spatial distribution of specific cell populations and critical biomarkers within intact tissues. This approach coupled with sophisticated imaging modalities and downstream analysis support provides new collaborative opportunities for Core laboratories.