Yanxiao Zhang

PePr: a peak-calling prioritization pipeline to identify consistent or differential peaks from replicated ChIP-Seq data

MOTIVATION ChIP-Seq is the standard method to identify genome-wide DNA-binding sites for transcription factors (TFs) and histone modifications. There is a growing need to analyze experiments with biological replicates, especially for epigenomic experiments where variation among biological samples can be substantial. However, tools that can perform group comparisons are currently lacking. RESULTS We present a peak-calling prioritization pipeline (PePr) for identifying consistent or differential binding sites in ChIP-Seq experiments with biological replicates. PePr models read counts across the genome among biological samples with a negative binomial distribution and uses a local variance estimation method, ranking consistent or differential binding sites more favorably than sites with greater variability. We compared PePr with commonly used and recently proposed approaches on eight TF datasets and show that PePr uniquely identifies consistent regions with enriched read counts, high motif occurrence rate and known characteristics of TF binding based on visual inspection. For histone modification data with broadly enriched regions, PePr identified differential regions that are consistent within groups and outperformed other methods in scaling False Discovery Rate (FDR) analysis. AVAILABILITY AND IMPLEMENTATION http://code.google.com/p/pepr-chip-seq/.

Subtypes of HPV-positive head and neck cancers are associated with HPV characteristics, copy number alterations, PIK3CA mutation, and pathway signatures

Purpose: There is substantial heterogeneity within human papillomavirus (HPV)-associated head and neck cancer (HNC) tumors that predispose them to different outcomes; however, the molecular heterogeneity in this subgroup is poorly characterized due to various historical reasons. Experimental Design: We performed unsupervised gene expression clustering on deeply annotated (transcriptome and genome) HPV+ HNC samples from two cohorts (84 total primary tumors), including 18 HPV− HNC samples, to discover subtypes and characterize the differences between subgroups in terms of their HPV characteristics, pathway activity, whole-genome somatic copy number alterations, and mutation frequencies. Results: We identified two distinct HPV+ subtypes (namely HPV-KRT and HPV-IMU). HPV-KRT is characterized by elevated expression of genes in keratinocyte differentiation and oxidation–reduction process, whereas HPV-IMU has strong immune response and mesenchymal differentiation. The differences in expression are likely connected to the differences in HPV characteristics and genomic changes. HPV-KRT has more genic viral integration, lower E2/E4/E5 expression levels, and higher ratio of spliced to full-length HPV oncogene E6 than HPV-IMU; the subgroups also show differences in copy number alterations and mutations, in particular the loss of chr16q in HPV-IMU and gain of chr3q and PIK3CA mutation in HPV-KRT. Conclusions: Our characterization of two subtypes of HPV+ HNC tumors provides valuable molecular level information that point to two main carcinogenic paths. Together, these results shed light on stratifications of the HPV+ HNCs and will help to guide personalized care for HPV+ HNC patients. Clin Cancer Res; 22(18); 4735–45. ©2016 AACR.

Genomic binding and regulation of gene expression by the thyroid carcinoma-associated PAX8-PPARG fusion protein

A chromosomal translocation results in production of an oncogenic PAX8-PPARG fusion protein (PPFP) in thyroid carcinomas. PAX8 is a thyroid transcription factor, and PPARG is a transcription factor that plays important roles in adipocytes and macrophages. PPFP retains the DNA binding domains of both proteins; however, the genomic binding sites of PPFP have not been identified, and only limited data exist to characterize gene expression in PPFP thyroid carcinomas. Therefore, the oncogenic function of PPFP is poorly understood. We expressed PPFP in PCCL3 rat thyroid cells and used ChIP-seq to identify PPFP genomic binding sites (PPFP peaks) and RNA-seq to characterize PPFP-dependent gene expression. PPFP peaks (~20,000) include known PAX8 and PPARG binding sites and are enriched with both motifs, indicating that both DNA binding domains are functional. PPFP binds to and regulates many genes involved in cancer-related processes. In PCCL3 thyroid cells, PPFP binds to adipocyte PPARG target genes in preference to macrophage PPARG target genes, consistent with the pro-adipogenic nature of PPFP and its ligand pioglitazone in thyroid cells. PPFP induces oxidative stress in thyroid cells, and pioglitazone increases susceptibility to further oxidative stress. Our data highlight the complexity of PPFP as a transcription factor and the numerous ways that it regulates thyroid oncogenesis.

Genomic binding of PAX8-PPARG fusion protein regulates cancer-related pathways and alters the immune landscape of thyroid cancer

PAX8-PPARG fusion protein (PPFP) results from a t(2;3)(q13;p25) chromosomal translocation, is found in 30% of follicular thyroid carcinomas, and demonstrates oncogenic capacity in transgenic mice. A PPARG ligand, pioglitazone, is highly therapeutic in mice with PPFP thyroid cancer. However, only limited data exist to characterize the binding sites and oncogenic function of PPFP, or to explain the observed therapeutic effect of pioglitazone. Here we used our previously characterized transgenic mouse model of PPFP follicular thyroid carcinoma to identify PPFP binding sites in vivo using ChIP-seq, and to distinguish genes and pathways regulated directly or indirectly by PPFP with and without pioglitazone treatment via integration with RNA-seq data. PPFP bound to DNA regions containing the PAX8 and/or the PPARG motif, near genes involved in lipid metabolism, the cell cycle, apoptosis, and cell motility; the binding site distribution was highly concordant with our previous study in a rat PCCL3 cell line. Most strikingly, pioglitazone induced an immune cell infiltration including macrophages and T cells only in the presence of PPFP, which may be central to its therapeutic effect.

HPV Integration in HNSCC Correlates with Survival Outcomes, Immune Response Signatures, and Candidate Drivers

The incidence of human papillomavirus (HPV)–related oropharynx cancer has steadily increased over the past two decades and now represents a majority of oropharyngeal cancer cases. Integration of the HPV genome into the host genome is a common event during carcinogenesis that has clinically relevant effects if the viral early genes are transcribed. Understanding the impact of HPV integration on clinical outcomes of head and neck squamous cell carcinoma (HNSCC) is critical for implementing deescalated treatment approaches for HPV+ HNSCC patients. RNA sequencing (RNA-seq) data from HNSCC tumors (n = 84) were used to identify and characterize expressed integration events, which were overrepresented near known head and neck, lung, and urogenital cancer genes. Five genes were recurrent, including CD274 (PD-L1). A significant number of genes detected to have integration events were found to interact with Tp63, ETS, and/or FOX1A. Patients with no detected integration had better survival than integration-positive and HPV− patients. Furthermore, integration-negative tumors were characterized by strongly heightened signatures for immune cells, including CD4+, CD3+, regulatory, CD8+ T cells, NK cells, and B cells, compared with integration-positive tumors. Finally, genes with elevated expression in integration-negative specimens were strongly enriched with immune-related gene ontology terms, while upregulated genes in integration-positive tumors were enriched for keratinization, RNA metabolism, and translation. Implications: These findings demonstrate the clinical relevancy of expressed HPV integration, which is characterized by a change in immune response and/or aberrant expression of the integration-harboring cancer-related genes, and suggest strong natural selection for tumor cells with expressed integration events in key carcinogenic genes. Mol Cancer Res; 16(1); 90–102. ©2017 AACR.

Expressed HNSCC variants by HPV-status in a well-characterized Michigan cohort

While whole-exome DNA sequencing is the most common technology to study genetic variants in tumors in known exonic regions, RNA-seq is cheaper, covers most of the same exonic regions, and is often more readily available. In this study, we show the utility of mRNA-seq-based variant analysis combined with targeted gene sequencing performed on both tumor and matched blood as an alternative when exome data is unavailable. We use the approach to study expressed variant profiles in the well-characterized University of Michigan (UM) head and neck squamous carcinoma (HNSCC) cohort (n = 36). We found that 441 out of 455 (~97%) identified cancer genes with an expressed variant in the UM cohort also harbor a somatic mutation in TCGA. Fourteen (39%) patients had a germline variant in a cancer-related Fanconi Anemia (FA) pathway gene. HPV-positive patients had more nonsynonymous, rare, and damaging (NRD) variants in those genes than HPV-negative patients. Moreover, the known mutational signatures for DNA mismatch repair and APOBEC activation were attributive to the UM expressed NRD variants, and the APOBEC signature contribution differed by HPV status. Our results provide additional support to certain TCGA findings and suggest an association of expressed variants in FA/DNA repair pathways with HPV-associated HNSCC tumorigenesis. These results will benefit future studies on this and other cohorts by providing the genetic variants of key cancer-related genes.

Adipogenic Differentiation of Thyroid Cancer Cells Through the Pax8-PPARγ Fusion Protein Is Regulated by Thyroid Transcription Factor 1 (TTF-1)

A subset of thyroid carcinomas contains a t(2;3)(q13;p25) chromosomal translocation that fuses paired box gene 8 (PAX8) with the peroxisome proliferator-activated receptor γ gene (PPARG), resulting in expression of a PAX8-PPARγ fusion protein, PPFP. We previously generated a transgenic mouse model of PPFP thyroid carcinoma and showed that feeding the PPARγ agonist pioglitazone greatly decreased the size of the primary tumor and prevented metastatic disease in vivo. The antitumor effect correlates with the fact that pioglitazone turns PPFP into a strongly PPARγ-like molecule, resulting in trans-differentiation of the thyroid cancer cells into adipocyte-like cells that lose malignant character as they become more differentiated. To further study this process, we performed cell culture experiments with thyrocytes from the PPFP mouse thyroid cancers. Our data show that pioglitazone induced cellular lipid accumulation and the expression of adipocyte marker genes in the cultured cells, and shRNA knockdown of PPFP eliminated this pioglitazone effect. In addition, we found that PPFP and thyroid transcription factor 1 (TTF-1) physically interact, and that these transcription factors bind near each other on numerous target genes. TTF-1 knockdown and overexpression studies showed that TTF-1 inhibits PPFP target gene expression and impairs adipogenic trans-differentiation. Surprisingly, pioglitazone repressed TTF-1 expression in PPFP-expressing thyrocytes. Our data indicate that TTF-1 interacts with PPFP to inhibit the pro-adipogenic response to pioglitazone, and that the ability of pioglitazone to decrease TTF-1 expression contributes to its pro-adipogenic action.

Abstract 1511: Mutational signatures from RNA-seq data distinguish HPV(+) and HPV(-) HNSCC

IntroductionHead and neck squamous cell carcinoma (HNSCC) is the 6th most common cancer worldwide, and an increasing percentage of HNSCC cases are associated with human papillomavirus (HPV) which have improved prognosis compared to HPV(-) HNSCC. Identifying distinctions between HPV(+) and HPV(-) HNSCC may lead to improved prognosis or new treatments, such as determining mutagen(s) that are driving accumulation of genetic variants in each tumor. A recent approach uses the fact that different mutagens cause different types of mutations, referred to as a mutagen9s signature. By comparing the mutations in a tumor sample with the mutagen signatures, the mutagens responsible for most of the mutations in the tumor can be identified. We hypothesized that mutational signatures could distinguish HPV(+) from HPV(-) HNSCC tumors. Materials and Methods We performed RNA-sequencing on 18 HPV(+) and 18 HPV(-) tumor samples collected from HNSCC patients by the University of Michigan Head and Neck SPORE. Variant calling was performed on each sample using a GATK best practices for RNA-seq pipeline. These variants were then filtered to include only variants that were both rare ( Results In the studied HNSCC tumors, signatures related to aging and impaired DNA repair were responsible for the largest proportion of the mutations found. The signatures that were significantly different between HPV(+) and HPV(-) tumors were associated with APOBEC editing and defective DNA mismatch repair. Expression and mutational analysis validated that APOBEC is more active in HPV(+) than HPV(-) tumors. Interestingly, signatures associated with smoking were not among the most relevant. Conclusion The evaluation of mutational signatures in HNSCC tumors identified the main sources of mutations for HNSCC. The differences in mutational signatures between HPV(+) and HPV(-) identify distinct methods of carcinogenesis for these two subtypes of HNSCC. Citation Format: Pelle Hall, Shama Virani, Charles Warden, Yanxiao Zhang, Lada Koneva, Katie M. Rentschler, Alisha Virani, Laura S. Rozek, Maureen A. Sartor, The UM Head and Neck SPORE Investigators. Mutational signatures from RNA-seq data distinguish HPV(+) and HPV(-) HNSCC. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 1511.

Abstract 1515: Identification and characterization of HPV-host fusion transcripts in HNSCCs

Introduction Head and neck squamous cell carcinoma (HNSCC) is the sixth most incident cancer worldwide. Human papillomavirus (HPV) is implicated in at least 70% of oropharyngeal squamous cell carcinomas in the US. HPV genomic integration events are a likely critical step in progression to cancer and occur as a consequence of HPV oncogene-induced chromosomal instability. The majority of viral integration events in the human genome appear to occur within or near genic regions. Identified HPV integration events in HNSCC were associated with alterations in DNA copy number, mRNA transcript abundance and splicing, and inter-/intrachromosomal rearrangements. Materials and methods We analyzed 84 HPV-positive HNSCC samples collected at the University of Michigan (18 tumors) and from The Cancer Genome Atlas (66 tumors). We used VirusSeq software for detection of integration events. Integrations were evidenced by host-virus fusion transcripts in RNA-seq, considering only breakpoints supported by at least four discordant read pairs and at least one junction spanning read. Results We identified 50 integration-positive tumors, which consisted of 41 with HPV16, one tumor with HPV18, 5 tumors with HPV33 and 3 with HPV35. We found an overall 271 breakpoints of integration within or near 83 human genes. Fusion virus-host transcripts with breakpoints within E6 and E7 HPV oncogenes were more common (59.3%) compared with breakpoints into other viral genes: E1 and E2 (19%), E4 and E5 (16.8%), L1 and L2 (4.7%). The detected viral integration events were widespread across the human genome with a few hotspots of recurrent integrations in genic regions at 1p36, 9p24, and 13p22. After accounting for regions covered by the data, we did not find an association of integration sites with aphidicolin-induced common fragile sites (CFSs), as observed for HPV integration sites in cervical cancer. Analysis of the protein interaction network constructed from the 83 host genes harboring viral integrants showed that 56 of them were linked into a highly connected sub-network with direct interactions. ETS2, TP63, FOXA1, CTGF and KLF5 were hubs in this network. The genes were statistically enriched for cancer genes known as important in “Head and Neck Neoplasms”(p = 1.74E-11). Conclusion HPV integration events in HNSCC are largely different from those common to cervical cancers. While TP63 appears to be a recurrent target in both cervical cancer and HNSCC, HNSCCs have a surprisingly high percent of integration events in known HNSCC-related genes. Overall, our results suggest that there is strong selective pressure for integration events that occur in HNSCC relevant genes. Citation Format: Lada A. Koneva, Yanxiao Zhang, Pelle Hall, Shama Virani, Alisha Virani, Thomas E. Carey, Laura S. Rozek, Maureen A. Sartor. Identification and characterization of HPV-host fusion transcripts in HNSCCs. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 1515.

Abstract 4808: Identification of genes with variants in HPV-associated head and neck squamous cell carcinoma

Proceedings: AACR 106th Annual Meeting 2015; April 18-22, 2015; Philadelphia, PA Introduction: Human papilloma virus (HPV)-associated head and neck squamous cell carcinomas have a distinct etiologic mechanism of carcinogenesis compared to non-HPV associated HNSCCs. These tumors are also generally thought to harbor less mutations compared to non-HPV associated HNSCCs. However, recent research suggests that not only do HPV-associated HNSCCs contain mutations, they are distinct from mutations found in non-HPV associated HNSCCs. Identification of these genetic variants is necessary to determine potential prognostic markers or novel insight into the mechanism of HPV-induced carcinogenesis. In this study, we identified genes with variants that are more common in HPV-associated HNSCCs compared to non-HPV associated HNSCCs. Materials and methods: Thirty-six primary head and neck SCC (HNSCC) tumor samples were collected between 2011-2013 at University of Michigan hospital, of which 18 were determined to be HPV+ (type16, 18 or 35). Paired-end RNA-Seq libraries were generated using Illumina HiSeq for all samples and aligned to both the human and HPV genome. Variant calling was employed using an established pipeline to identify single nucleotide polymorphisms (SNPs) from RNA-Seq data. Briefly, data was aligned using STAR, followed by indel realignment and base recalibration. Variants were called using HaplotypeCaller and filtered by Fisher Strand values (FS > 30.0) and Qual By Depth values (QD < 2.0). Variants were annotated and filtered by SnpEff and SnpSift to determine genes with nonsynonymous variants by HPV status. Results: Out of all samples, nine genes showed nonsynonymous variants in HPV-associated tumors greater than 60% of the time compared to non-HPV associated tumors. These are PTCH1 (associated with eosophageal squamous cell carcinoma), ATRX (involved in chromatin binding), ERCC2 (associated with skin cancers), FOXA1 (transcription factor), CCDC6 (associated with thyroid papillary carcinoma), FNBP1 (involved in lipid binding), and POU5F1 (transcription factor). Seventy-five percent of the samples with nonsynonymous variants in TMPRSS2, a gene shown to be associated with prostate cancer, were HPV-associated tumors. Eighty-two percent of all samples with nonsynonymous variants in FANCD2, a gene involved in maintenance of chromosomal stability, were HPV-associated tumors. Conclusion: Although HPV-associated HNSCCs generally harbor fewer mutations than non-HPV associated HNSCCs, our data suggests they do contain variants in cancer-related genes. This is important as these genes may play a role in the mechanism of HPV-induced tumorigenesis subsequent to viral initiation. Citation Format: Shama Virani, Pelle B. Hall, Yanxiao Zhang, Lada A. Koneva, Alisha Virani, Katie M. Rentschler, Thomas E. Carey, Laura S. Rozek, Maureen A. Sartor. Identification of genes with variants in HPV-associated head and neck squamous cell carcinoma. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4808. doi:10.1158/1538-7445.AM2015-4808