The BTLA-HVEM axis restricts CAR T cell efficacy in cancer.

The efficacy of T cell-based immunotherapies is limited by immunosuppressive pressures in the tumor microenvironment. Here we show a predominant role for the interaction between BTLA on effector T cells and HVEM (TNFRSF14) on immunosuppressive tumor microenvironment cells, namely regulatory T cells. High BTLA expression in chimeric antigen receptor (CAR) T cells correlated with poor clinical response to treatment. Therefore, we deleted BTLA in CAR T cells and show improved tumor control and persistence in models of lymphoma and solid malignancies. Mechanistically, BTLA inhibits CAR T cells via recruitment of tyrosine phosphatases SHP-1 and SHP-2, upon trans engagement with HVEM. BTLA knockout thus promotes CAR signaling and subsequently enhances effector function. Overall, these data indicate that the BTLA-HVEM axis is a crucial immune checkpoint in CAR T cell immunotherapy and warrants the use of strategies to overcome this barrier.

Genetic Inactivation of CD33 in Hematopoietic Stem Cells to Enable CAR T Cell Immunotherapy for Acute Myeloid Leukemia.

The absence of cancer-restricted surface markers is a major impediment to antigen-specific immunotherapy using chimeric antigen receptor (CAR) T cells. For example, targeting the canonical myeloid marker CD33 in acute myeloid leukemia (AML) results in toxicity from destruction of normal myeloid cells. We hypothesized that a leukemia-specific antigen could be created by deleting CD33 from normal hematopoietic stem and progenitor cells (HSPCs), thereby generating a hematopoietic system resistant to CD33-targeted therapy and enabling specific targeting of AML with CAR T cells. We generated CD33-deficient human HSPCs and demonstrated normal engraftment and differentiation in immunodeficient mice. Autologous CD33 KO HSPC transplantation in rhesus macaques demonstrated long-term multilineage engraftment of gene-edited cells with normal myeloid function. CD33-deficient cells were impervious to CD33-targeting CAR T cells, allowing for efficient elimination of leukemia without myelotoxicity. These studies illuminate a novel approach to antigen-specific immunotherapy by genetically engineering the host to avoid on-target, off-tumor toxicity.

Encyclopedia of bacterial gene circuits whose presence or absence correlate with pathogenicity--a large-scale system analysis of decoded bacterial genomes.

BACKGROUND: Bacterial infections comprise a global health challenge as the incidences of antibiotic resistance increase. Pathogenic potential of bacteria has been shown to be context dependent, varying in response to environment and even within the strains of the same genus. RESULTS: We used the KEGG repository and extensive literature searches to identify among the 2527 bacterial genomes in the literature those implicated as pathogenic to the host, including those which show pathogenicity in a context dependent manner. Using data on the gene contents of these genomes, we identified sets of genes highly abundant in pathogenic but relatively absent in commensal strains and vice versa. In addition, we carried out genome comparison within a genus for the seventeen largest genera in our genome collection. We projected the resultant lists of ortholog genes onto KEGG bacterial pathways to identify clusters and circuits, which can be linked to either pathogenicity or synergy. Gene circuits relatively abundant in nonpathogenic bacteria often mediated biosynthesis of antibiotics. Other synergy-linked circuits reduced drug-induced toxicity. Pathogen-abundant gene circuits included modules in one-carbon folate, two-component system, type-3 secretion system, and peptidoglycan biosynthesis. Antibiotics-resistant bacterial strains possessed genes modulating phagocytosis, vesicle trafficking, cytoskeletal reorganization, and regulation of the inflammatory response. Our study also identified bacterial genera containing a circuit, elements of which were previously linked to Alzheimer's disease. CONCLUSIONS: Present study produces for the first time, a signature, in the form of a robust list of gene circuitry whose presence or absence could potentially define the pathogenicity of a microbiome. Extensive literature search substantiated a bulk majority of the commensal and pathogenic circuitry in our predicted list. Scanning microbiome libraries for these circuitry motifs will provide further insights into the complex and context dependent pathogenicity of bacteria.

Deletion of CD38 enhances CD19 chimeric antigen receptor T cell function.

Cell surface molecules transiently upregulated on activated T cells can play a counter-regulatory role by inhibiting T cell function. Deletion or blockade of such immune checkpoint receptors has been investigated to improve the function of engineered immune effector cells. CD38 is upregulated on activated T cells, and although there have been studies showing that CD38 can play an inhibitory role in T cells, how it does so has not fully been elucidated. In comparison with molecules such as PD1, CTLA4, LAG3, and TIM3, we found that CD38 displays more sustained and intense expression following acute activation. After deleting CD38 from human chimeric antigen receptor (CAR) T cells, we showed relative resistance to exhaustion in vitro and improved anti-tumor function in vivo. CD38 is a multifunctional ectoenzyme with hydrolase and cyclase activities. Reintroduction of CD38 mutants into T cells lacking CD38 provided further evidence supporting the understanding that CD38 plays a crucial role in producing the immunosuppressive metabolite adenosine and utilizing nicotinamide adenine dinucleotide (NAD) in human T cells. Taken together, these results highlight a role for CD38 as an immunometabolic checkpoint in T cells and lead us to propose CD38 deletion as an additional avenue for boosting CAR T cell function.

Anti-CD19 CAR T cells in combination with ibrutinib for the treatment of chronic lymphocytic leukemia.

In chronic lymphocytic leukemia (CLL) patients who achieve a complete remission (CR) to anti-CD19 chimeric antigen receptor T cells (CART-19), remissions are remarkably durable. Preclinical data suggesting synergy between CART-19 and the Bruton's tyrosine kinase (BTK) inhibitor ibrutinib prompted us to conduct a prospective single-center phase 2 trial in which we added autologous anti-CD19 humanized binding domain T cells (huCART-19) to ibrutinib in patients with CLL not in CR despite ≥6 months of ibrutinib. The primary endpoints were safety, feasibility, and achievement of a CR within 3 months. Of 20 enrolled patients, 19 received huCART-19. The median follow-up for all infused patients was 41 months (range, 0.25-58 months). Eighteen patients developed cytokine release syndrome (CRS; grade 1-2 in 15 of 18 subjects), and 5 developed neurotoxicity (grade 1-2 in 4 patients, grade 4 in 1 patient). While the 3-month CR rate among International Working Group on CLL (iwCLL)-evaluable patients was 44% (90% confidence interval [CI], 23-67%), at 12 months, 72% of patients tested had no measurable residual disease (MRD). The estimated overall and progression-free survival at 48 months were 84% and 70%, respectively. Of 15 patients with undetectable MRD at 3 or 6 months, 13 remain in ongoing CR at the last follow-up. In patients with CLL not achieving a CR despite ≥6 months of ibrutinib, adding huCART-19 mediated a high rate of deep and durable remissions. ClinicalTrials.gov number, NCT02640209.

Human chimeric antigen receptor macrophages for cancer immunotherapy.

Chimeric antigen receptor (CAR) T cell therapy has shown promise in hematologic malignancies, but its application to solid tumors has been challenging1-4. Given the unique effector functions of macrophages and their capacity to penetrate tumors5, we genetically engineered human macrophages with CARs to direct their phagocytic activity against tumors. We found that a chimeric adenoviral vector overcame the inherent resistance of primary human macrophages to genetic manipulation and imparted a sustained pro-inflammatory (M1) phenotype. CAR macrophages (CAR-Ms) demonstrated antigen-specific phagocytosis and tumor clearance in vitro. In two solid tumor xenograft mouse models, a single infusion of human CAR-Ms decreased tumor burden and prolonged overall survival. Characterization of CAR-M activity showed that CAR-Ms expressed pro-inflammatory cytokines and chemokines, converted bystander M2 macrophages to M1, upregulated antigen presentation machinery, recruited and presented antigen to T cells and resisted the effects of immunosuppressive cytokines. In humanized mouse models, CAR-Ms were further shown to induce a pro-inflammatory tumor microenvironment and boost anti-tumor T cell activity.

A heuristic method for simulating open-data of arbitrary complexity that can be used to compare and evaluate machine learning methods.

A central challenge of developing and evaluating artificial intelligence and machine learning methods for regression and classification is access to data that illuminates the strengths and weaknesses of different methods. Open data plays an important role in this process by making it easy for computational researchers to easily access real data for this purpose. Genomics has in some examples taken a leading role in the open data effort starting with DNA microarrays. While real data from experimental and observational studies is necessary for developing computational methods it is not sufficient. This is because it is not possible to know what the ground truth is in real data. This must be accompanied by simulated data where that balance between signal and noise is known and can be directly evaluated. Unfortunately, there is a lack of methods and software for simulating data with the kind of complexity found in real biological and biomedical systems. We present here the Heuristic Identification of Biological Architectures for simulating Complex Hierarchical Interactions (HIBACHI) method and prototype software for simulating complex biological and biomedical data. Further, we introduce new methods for developing simulation models that generate data that specifically allows discrimination between different machine learning methods.