907 Modulation of tumor immunogenicity by DNA methylation of immune synapse genes in cancers

BackgroundCancer immunotherapy represents a major paradigm shift in cancer care. Despite such breakthrough, majority of cancer patients remains refractory to existing immunotherapeutic modalities highlighting the inherent capacity of tumors to escape immunosurveillance mechanisms. Frequently, cancer cells utilize the epigenetic machinery to silence tumor suppressors or activate oncogenes for survival and proliferation. Likewise, tumor cells might employ the epigenetic reprogramming of immune-related pathways to evade the immune system. Methylation is one of the major epigenetic mechanisms modulating gene transcription. Thus, we investigated the methylation profile of both co-stimulatory and immune checkpoint genes in cancer.MethodsData from The Cancer Genome Atlas (TCGA) were used for methylation profiling and RNA-sequencing analysis. Twenty-six epithelial cancer cell lines with more than 3 mock and three 5-azacitidine–treated samples were selected for analysis from the GSE57342 dataset. t-distributed stochastic neighbor embedding (t-SNE) was calculated using 247 probes for the selected 20 genes across all TCGA samples. t-SNE analysis was performed on 8,186 solid tumors and 745 normal adjacent tissues for methylation levels for all probes. For principal component analysis, first and second principal components were used to represent the overall methylation status for 8,931 tumor and normal samples in the TCGA database. Survival analyses were retrieved from a prior publication.1ResultsWe found that methylation profile of immune synapse genes is distinct in tumor versus normal adjacent tissue. Interestingly, our results demonstrate hypermethylation of co-stimulatory genes such as CD40 and hypo-methylation of immune checkpoint genes such as HHLA2 and PDL1 across multiple tumor types in comparison with the normal adjacent tissue. In addition, an inverse correlation between methylation and gene expression was manifest among tumor and normal adjacent tissue, confirming the epigenetic mechanism of gene suppression by gene methylation. Furthermore, we observed a reversal of hypermethylation of the co-stimulatory genes including CD40 by the demethylating agent 5-azacytidine in the data set of 26 epithelial cancer cell lines. Finally, we found that that hypomethylation of co-stimulatory genes within the immune synapse correlates with functional T cell recruitment to the tumor microenvironment and is followed by a favorable clinical outcome in melanoma patients.ConclusionsOur finding unveils methylation of immune synapse genes as a crucial driver of the immune evasive phenotype of cancer cells. Notably, identification of actionable targets to restore tumor immunogenicity is an attractive strategy in combination with immune checkpoint blockade.AcknowledgementsThis work was supported by NIH grant K08 CA194273, the Immunology Innovation Fund, an NCI Cancer Center Support grant, (P30-CA076292), the Miriam and Sheldon G. Adelson Foundation, and the Moffitt Foundation.ReferenceLiu J, Lichtenberg T, Hoadley KA, Poisson LM, Lazar AJ, Cherniack AD, Kovatich AJ, Benz CC, Levine DA, Lee AV, Omberg L, Wolf DM, Shriver CD, Thorsson V, Cancer Genome Atlas Research N, Hu H. An integrated TCGA pan-cancer clinical data resource to drive high-quality survival outcome analytics. Cell 2018;173(2):400–16 e11.

Microbiome within Primary Tumor Tissue from Renal Cell Carcinoma May Be Associated with PD-L1 Expression of the Venous Tumor Thrombus

Objective. To perform a proof of concept microbiome evaluation and PD-L1 expression profiling in clear-cell renal cell carcinoma (cc-RCC) with associated tumor thrombus (TT). Methods. After IRB approval, six patients underwent radical nephrectomy (RN) with venous tumor thrombectomy (VTT). We collected fresh tissue specimens from normal adjacent, tumor, and thrombus tissues. We utilized RNA sequencing to obtain PD-L1 expression profiles and perform microbiome analysis. Statistical assessment was performed using Student’s t-test, chi-square, and spearman rank correlations using SPSS v25. Results. We noted the tumor thrombus to be mostly devoid of diverse microbiota. A large proportion of Staphylococcus epidermidus was detected and unknown if this is a surgical or postsurgical contaminant; however, it was noted more in the thrombus than other tissues. Microbiome diversity profiles were most abundant in the primary tumor compared to the thrombus or normal adjacent tissue. Differential expression of PD-L1 was examined in the tumor thrombus to the normal background tissue and noted three of the six subjects had a threshold above 2-fold. These three similar subjects had foreign microbiota that are typical residents of the oral microbiome. Conclusion. Renal tumors have more diverse microbiomes than normal adjacent tissue. Identification of resident oral microbiome profiles in clear-cell renal cancer with tumor thrombus provides a potential biomarker for thrombus response to PD-L1 inhibition.

The Microbiota of Breast Tissue and Its Association with Breast Cancer

ABSTRACTIn the United States, 1 in 8 women will be diagnosed with breast cancer in her lifetime. Along with genetics, the environment contributes to disease development, but what these exact environmental factors are remains unknown. We have previously shown that breast tissue is not sterile but contains a diverse population of bacteria. We thus believe that the host's local microbiome could be modulating the risk of breast cancer development. Using 16S rRNA amplicon sequencing, we show that bacterial profiles differ between normal adjacent tissue from women with breast cancer and tissue from healthy controls. Women with breast cancer had higher relative abundances ofBacillus,EnterobacteriaceaeandStaphylococcus.Escherichia coli(a member of theEnterobacteriaceaefamily) andStaphylococcus epidermidis, isolated from breast cancer patients, were shown to induce DNA double-stranded breaks in HeLa cells using the histone-2AX (H2AX) phosphorylation (γ-H2AX) assay. We also found that microbial profiles are similar between normal adjacent tissue and tissue sampled directly from the tumor. This study raises important questions as to what role the breast microbiome plays in disease development or progression and how we can manipulate this for possible therapeutics or prevention.IMPORTANCEThis study shows that different bacterial profiles in breast tissue exist between healthy women and those with breast cancer. Higher relative abundances of bacteria that had the ability to cause DNA damagein vitrowere detected in breast cancer patients, as was a decrease in some lactic acid bacteria, known for their beneficial health effects, including anticarcinogenic properties. This study raises important questions as to the role of the mammary microbiome in modulating the risk of breast cancer development.

Methylation-specific high-resolution melting analysis (MS-HRM)-based DNA promoter profiling in paired FFPE bladder cancer samples.

315 Background: DNA methylation and histone modification are widely studied epigenetic events that can decrease gene expression levels. Promoter hypermethylation has been proposed as a potential diagnostic or prognostic biomarker in various cancers. We studied DNA promoter methylation of 5 cancer-associated genes (MRE11, APX1, ERCC1, RASSF1A and RASSF2A ) in paired FFPE bladder cancer tissues and normal adjacent tissue. The MRE11/Rad50/NBS1 complex serves as a single-strand DNA nuclease which participates in the repair of DNA double-strand breaks and replication errors. APX1 plays a key role in regulating H2O2 levels and H2O2 signaling. DNA repair machinery, ERCC1 protein levels in tumor tissue have been shown to predict response to platinum-based chemotherapy. RASSF1 is thought to be a tumor suppressor gene, while the function of RASSF2 is less well understood. Methods: 16 bladder cancer cases with available paraffin-embedded tumor and matched normal adjacent tissue specimens (32 tissue samples) were analyzed. DNA was extracted by Ambion RecoverAll Total Nucleic Acid Isolation kit. FFPE DNA bisulfite modification was performed using Zymo EZ DNA Methylation Kit. Methylation Specific-High Resolution Melting (MS-HRM) analysis was used to assess methylation status. MS-HRM monitors the melting behavior of PCR amplicons by using a DNA intercalating fluorescent dye. Results: MRE11 and APX1 promoters were not methylated in either cancer or normal adjacent samples. The ERCC1 gene was heavily methylated in 94% (30/32) of all cancer and wild type samples. RASSF1A and RASSF2A promoter methylation was significantly different between cancer and normal adjacent tissue. 50% (8/16) RASSF1A and 25% (4/16) RASSF2A had promoter methylation in cancer tissues, while only 6% (1/16) RASSF1A and 0% (0/16) RASSF2A had promoter methylation in normal adjacent tissues. 69% (11/16) of bladder cancer tissues were positive for RASSF1A or RASSF2A promotor methylation while only 1/16 normal adjacent tissue samples was positive for either promotor methylation. Conclusions: The results show that cancer and non-cancer tissue have different RASSF1A and RASSF2A promoter methylation patterns.

Assessment of Insertional Mutagenesis Risk Following AAV Vector-Mediated Factor IX Gene Transfer in Mice.

Abstract 2465 Poster Board II-442 Gene therapy for Hemophilia B is a promising alternative to recombinant protein therapy. Long-term expression of coagulation factor IX via stable gene transfer could reduce costs and risks associated with intravenous recombinant factor IX infusions. Liver-directed, adeno-associated virus (AAV) vector-mediated gene transfer of factor IX (FIX) is a gene therapy strategy currently used in two human clinical trials. Our group is performing one of these trials using an AAV vector (AAV-hFIX16) expressing human FIX from a liver-specific promoter. Hepatic artery delivery of AAV-hFIX16 has shown short-term efficacy and disease correction. A fundamental issue facing clinical gene transfer has to do with risk related to vector integration. Though AAV vector DNA is predominantly episomal in transduced cells, chromosomal integration can occur, and consequences of AAV genotoxicity at the molecular level in in vivo systems require further analysis. Here we present the results of a large-scale longitudinal study in which we followed 120 male, wild-type C57BL/6 mice for 18 months after portal vein injection of either 5e12 vg/kg of AAV-hFIX16 (60 mice), 1e14 vg/kg of AAV-hFIX16 (20 mice), 1e14 vc/kg of empty AAV capsid (20 mice), or saline (20 mice). At 18 months we found circulating hFIX levels of 12 ug/mL in mice receiving the lower AAV dose, and 25 ug/mL in mice receiving the higher AAV dose, with no detectable hFIX in mice receiving empty capsid or saline. To assess the risk of insertional mutagenesis in cells transduced by AAV vector, we analyzed the incidence of hepatocellular carcinoma (HCC). We found an HCC incidence rate of 9.1% in untreated mice, 0% in mice receiving saline, 0% in mice receiving empty capsid, 3.8% in mice receiving the lower AAV dose, and 12.5% in mice receiving the higher AAV dose, resulting in a p-value of 0.64 when comparing both AAV groups to controls. We quantified vector genome copy number by qPCR in the HCC's occurring in mice treated with AAV and compared them to vector genome copy number in normal tissue adjacent to the tumors and found no statistically significant difference (p=0.3). We then used LM-PCR to recover vector-chromosome integration junctions from tumor tissue and normal adjacent tissue. We were able to clone 249 unique integrants from tumor tissue and 862 unique integrants from normal adjacent tissue. This yields an average of 0.00018 unique integrants per diploid genome in tumor tissue and 0.00064 unique integrants per diploid genome in normal adjacent tissue. When determining integration profiles within the mouse genome, we found that integrants in both tumor tissue and normal adjacent tissue had a preference for integrating within genes, particularly within exons, as well as within CpG-rich regions. In addition, we found that integrants in both tumor tissue and normal adjacent tissue were more likely than not to be located within 50kb of the 5′ end of an oncogene. While this preference was stronger for integrants in normal adjacent tissue (p=6.3e-5) than integrants in tumor tissue (p=7.7e-2), it was even more likely for the HIV virus to integrate within 50kb of the 5′ end of an oncogene (p=4.1e-11). Functional consequences of these sites of AAV vector integration are not yet known, but are currently being investigated further. In conclusion, our large-scale prospective study of mice receiving AAV-hFIX16 did not demonstrate a statistically significant difference in HCC incidence between AAV-treated groups and control groups, although there was a trend of increasing HCC incidence with very high AAV doses. When analyzing tumors from mice treated with AAV, we found no relationship between vector genome copy number and tumor formation. When analyzing integrated vector in tumor tissue and normal adjacent tissue, we found that integrants in both normal adjacent tissue and tumor tissue had less likelihood for localizing within 50kb of the 5' end of oncogenes than HIV. Disclosures: Wright: Genzyme: Consultancy; Tacere: Consultancy.