The association of cigarette smoking with DNA methylation and gene expression in human tissue samples.

Cigarette smoking adversely affects many aspects of human health, and epigenetic responses to smoking may reflect mechanisms that mediate or defend against these effects. Prior studies of smoking and DNA methylation (DNAm), typically measured in leukocytes, have identified numerous smoking-associated regions (e.g., AHRR). To identify smoking-associated DNAm features in typically inaccessible tissues, we generated array-based DNAm data for 916 tissue samples from the GTEx (Genotype-Tissue Expression) project representing 9 tissue types (lung, colon, ovary, prostate, blood, breast, testis, kidney, and muscle). We identified 6,350 smoking-associated CpGs in lung tissue (n = 212) and 2,735 in colon tissue (n = 210), most not reported previously. For all 7 other tissue types (sample sizes 38-153), no clear associations were observed (false discovery rate 0.05), but some tissues showed enrichment for smoking-associated CpGs reported previously. For 1,646 loci (in lung) and 22 (in colon), smoking was associated with both DNAm and local gene expression. For loci detected in both lung and colon (e.g., AHRR, CYP1B1, CYP1A1), top CpGs often differed between tissues, but similar clusters of hyper- or hypomethylated CpGs were observed, with hypomethylation at regulatory elements corresponding to increased expression. For lung tissue, 17 hallmark gene sets were enriched for smoking-associated CpGs, including xenobiotic- and cancer-related gene sets. At least four smoking-associated regions in lung were impacted by lung methylation quantitative trait loci (QTLs) that co-localize with genome-wide association study (GWAS) signals for lung function (FEV1/FVC), suggesting epigenetic alterations can mediate the effects of smoking on lung health. Our multi-tissue approach has identified smoking-associated regions in disease-relevant tissues, including effects that are shared across tissue types.

An integrative multi-context Mendelian randomization method for identifying risk genes across human tissues.

Mendelian randomization (MR) provides valuable assessments of the causal effect of exposure on outcome, yet the application of conventional MR methods for mapping risk genes encounters new challenges. One of the issues is the limited availability of expression quantitative trait loci (eQTLs) as instrumental variables (IVs), hampering the estimation of sparse causal effects. Additionally, the often context/tissue-specific eQTL effects challenge the MR assumption of consistent IV effects across eQTL and GWAS data. To address these challenges, we propose a multi-context multivariable integrative MR framework, mintMR, for mapping expression and molecular traits as joint exposures. It models the effects of molecular exposures across multiple tissues in each gene region, while simultaneously estimating across multiple gene regions. It uses eQTLs with consistent effects across more than one tissue type as IVs, improving IV consistency. A major innovation of mintMR involves employing multi-view learning methods to collectively model latent indicators of disease relevance across multiple tissues, molecular traits, and gene regions. The multi-view learning captures the major patterns of disease-relevance and uses these patterns to update the estimated tissue relevance probabilities. The proposed mintMR iterates between performing a multi-tissue MR for each gene region and joint learning the disease-relevant tissue probabilities across gene regions, improving the estimation of sparse effects across genes. We apply mintMR to evaluate the causal effects of gene expression and DNA methylation for 35 complex traits using multi-tissue QTLs as IVs. The proposed mintMR controls genome-wide inflation and offers new insights into disease mechanisms.

Integrative cross-omics and cross-context analysis elucidates molecular links underlying genetic effects on complex traits.

Genetic effects on functionally related 'omic' traits often co-occur in relevant cellular contexts, such as tissues. Motivated by the multi-tissue methylation quantitative trait loci (mQTLs) and expression QTLs (eQTLs) analysis, we propose X-ING (Cross-INtegrative Genomics) for cross-omics and cross-context integrative analysis. X-ING takes as input multiple matrices of association statistics, each obtained from different omics data types across multiple cellular contexts. It models the latent binary association status of each statistic, captures the major association patterns among omics data types and contexts, and outputs the posterior mean and probability for each input statistic. X-ING enables the integration of effects from different omics data with varying effect distributions. In the multi-tissue cis-association analysis, X-ING shows improved detection and replication of mQTLs by integrating eQTL maps. In the trans-association analysis, X-ING reveals an enrichment of trans-associations in many disease/trait-relevant tissues.

DNA methylation QTL mapping across diverse human tissues provides molecular links between genetic variation and complex traits.

Studies of DNA methylation (DNAm) in solid human tissues are relatively scarce; tissue-specific characterization of DNAm is needed to understand its role in gene regulation and its relevance to complex traits. We generated array-based DNAm profiles for 987 human samples from the Genotype-Tissue Expression (GTEx) project, representing 9 tissue types and 424 subjects. We characterized methylome and transcriptome correlations (eQTMs), genetic regulation in cis (mQTLs and eQTLs) across tissues and e/mQTLs links to complex traits. We identified mQTLs for 286,152 CpG sites, many of which (>5%) show tissue specificity, and mQTL colocalizations with 2,254 distinct GWAS hits across 83 traits. For 91% of these loci, a candidate gene link was identified by integration of functional maps, including eQTMs, and/or eQTL colocalization, but only 33% of loci involved an eQTL and mQTL present in the same tissue type. With this DNAm-focused integrative analysis, we contribute to the understanding of molecular regulatory mechanisms in human tissues and their impact on complex traits.

Mendelian randomization accounting for complex correlated horizontal pleiotropy while elucidating shared genetic etiology.

Mendelian randomization (MR) harnesses genetic variants as instrumental variables (IVs) to study the causal effect of exposure on outcome using summary statistics from genome-wide association studies. Classic MR assumptions are violated when IVs are associated with unmeasured confounders, i.e., when correlated horizontal pleiotropy (CHP) arises. Such confounders could be a shared gene or inter-connected pathways underlying exposure and outcome. We propose MR-CUE (MR with Correlated horizontal pleiotropy Unraveling shared Etiology and confounding), for estimating causal effect while identifying IVs with CHP and accounting for estimation uncertainty. For those IVs, we map their cis-associated genes and enriched pathways to inform shared genetic etiology underlying exposure and outcome. We apply MR-CUE to study the effects of interleukin 6 on multiple traits/diseases and identify several S100 genes involved in shared genetic etiology. We assess the effects of multiple exposures on type 2 diabetes across European and East Asian populations.

Severe COVID-19 infection is associated with aberrant cytokine production by infected lung epithelial cells rather than by systemic immune dysfunction.

The mechanisms explaining progression to severe COVID-19 remain poorly understood. It has been proposed that immune system dysregulation/over-stimulation may be implicated, but it is not clear how such processes would lead to respiratory failure. We performed comprehensive multiparameter immune monitoring in a tightly controlled cohort of 128 COVID-19 patients, and used the ratio of oxygen saturation to fraction of inspired oxygen (SpO2 / FiO2) as a physiologic measure of disease severity. Machine learning algorithms integrating 139 parameters identified IL-6 and CCL2 as two factors predictive of severe disease, consistent with the therapeutic benefit observed with anti-IL6-R antibody treatment. However, transcripts encoding these cytokines were not detected among circulating immune cells. Rather, in situ analysis of lung specimens using RNAscope and immunofluorescent staining revealed that elevated IL-6 and CCL2 were dominantly produced by infected lung type II pneumocytes. Severe disease was not associated with higher viral load, deficient antibody responses, or dysfunctional T cell responses. These results refine our understanding of severe COVID-19 pathophysiology, indicating that aberrant cytokine production by infected lung epithelial cells is a major driver of immunopathology. We propose that these factors cause local immune regulation towards the benefit of the virus.

A robust two-sample transcriptome-wide Mendelian randomization method integrating GWAS with multi-tissue eQTL summary statistics.

By treating genetic variants as instrumental variables (IVs), two-sample Mendelian randomization (MR) methods detect genetically regulated risk exposures for complex diseases using only summary statistics. When considering gene expression as exposure in transcriptome-wide MR (TWMR) analyses, the eQTLs (expression-quantitative-trait-loci) may have pleiotropic effects or be correlated with variants that have effects on disease not via expression, and the presence of those invalid IVs would lead to biased inference. Moreover, the number of eQTLs as IVs for a gene is generally limited, making the detection of invalid IVs challenging. We propose a method, "MR-MtRobin," for accurate TWMR inference in the presence of invalid IVs. By leveraging multi-tissue eQTL data in a mixed model, the proposed method makes identifiable the IV-specific random effects due to pleiotropy from estimation errors of eQTL summary statistics, and can provide accurate inference on the dependence (fixed effects) between eQTL and GWAS (genome-wide association study) effects in the presence of invalid IVs. Moreover, our method can improve power and precision in inference by selecting cross-tissue eQTLs as IVs that have improved consistency of effects across eQTL and GWAS data. We applied MR-MtRobin to detect genes associated with schizophrenia risk by integrating summary-level data from the Psychiatric Genomics Consortium and the Genotype-Tissue Expression project (V8).

CCmed: cross-condition mediation analysis for identifying replicable trans-associations mediated by cis-gene expression.

MOTIVATION: Trans-acting expression quantitative trait loci (eQTLs) collectively explain a substantial proportion of expression variation, yet are challenging to detect and replicate since their effects are often individually weak. A large proportion of genetic effects on distal genes are mediated through cis-gene expression. Cis-association (between SNP and cis-gene) and gene-gene correlation conditional on SNP genotype could establish trans-association (between SNP and trans-gene). Both cis-association and gene-gene conditional correlation have effects shared across relevant tissues and conditions, and trans-associations mediated by cis-gene expression also have effects shared across relevant conditions. RESULTS: We proposed a Cross-Condition Mediation analysis method (CCmed) for detecting cis-mediated trans-associations with replicable effects in relevant conditions/studies. CCmed integrates cis-association and gene-gene conditional correlation statistics from multiple tissues/studies. Motivated by the bimodal effect-sharing patterns of eQTLs, we proposed two variations of CCmed, CCmedmost and CCmedspec for detecting cross-tissue and tissue-specific trans-associations, respectively. We analyzed data of 13 brain tissues from the Genotype-Tissue Expression (GTEx) project, and identified trios with cis-mediated trans-associations across brain tissues, many of which showed evidence of trans-association in two replication studies. We also identified trans-genes associated with schizophrenia loci in at least two brain tissues. AVAILABILITY AND IMPLEMENTATION: CCmed software is available at http://github.com/kjgleason/CCmed. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Determinants of telomere length across human tissues.

Telomere shortening is a hallmark of aging. Telomere length (TL) in blood cells has been studied extensively as a biomarker of human aging and disease; however, little is known regarding variability in TL in nonblood, disease-relevant tissue types. Here, we characterize variability in TLs from 6391 tissue samples, representing >20 tissue types and 952 individuals from the Genotype-Tissue Expression (GTEx) project. We describe differences across tissue types, positive correlation among tissue types, and associations with age and ancestry. We show that genetic variation affects TL in multiple tissue types and that TL may mediate the effect of age on gene expression. Our results provide the foundational knowledge regarding TL in healthy tissues that is needed to interpret epidemiological studies of TL and human health.

Primo: integration of multiple GWAS and omics QTL summary statistics for elucidation of molecular mechanisms of trait-associated SNPs and detection of pleiotropy in complex traits.

To provide a comprehensive mechanistic interpretation of how known trait-associated SNPs affect complex traits, we propose a method, Primo, for integrative analysis of GWAS summary statistics with multiple sets of omics QTL summary statistics from different cellular conditions or studies. Primo examines association patterns of SNPs to complex and omics traits. In gene regions harboring known susceptibility loci, Primo performs conditional association analysis to account for linkage disequilibrium. Primo allows for unknown study heterogeneity and sample correlations. We show two applications using Primo to examine the molecular mechanisms of known susceptibility loci and to detect and interpret pleiotropic effects.

IGREX for quantifying the impact of genetically regulated expression on phenotypes.

By leveraging existing GWAS and eQTL resources, transcriptome-wide association studies (TWAS) have achieved many successes in identifying trait-associations of genetically regulated expression (GREX) levels. TWAS analysis relies on the shared GREX variation across GWAS and the reference eQTL data, which depends on the cellular conditions of the eQTL data. Considering the increasing availability of eQTL data from different conditions and the often unknown trait-relevant cell/tissue-types, we propose a method and tool, IGREX, for precisely quantifying the proportion of phenotypic variation attributed to the GREX component. IGREX takes as input a reference eQTL panel and individual-level or summary-level GWAS data. Using eQTL data of 48 tissue types from the GTEx project as a reference panel, we evaluated the tissue-specific IGREX impact on a wide spectrum of phenotypes. We observed strong GREX effects on immune-related protein biomarkers. By incorporating trans-eQTLs and analyzing genetically regulated alternative splicing events, we evaluated new potential directions for TWAS analysis.

Integrated Proteogenomic Characterization of Clear Cell Renal Cell Carcinoma.

To elucidate the deregulated functional modules that drive clear cell renal cell carcinoma (ccRCC), we performed comprehensive genomic, epigenomic, transcriptomic, proteomic, and phosphoproteomic characterization of treatment-naive ccRCC and paired normal adjacent tissue samples. Genomic analyses identified a distinct molecular subgroup associated with genomic instability. Integration of proteogenomic measurements uniquely identified protein dysregulation of cellular mechanisms impacted by genomic alterations, including oxidative phosphorylation-related metabolism, protein translation processes, and phospho-signaling modules. To assess the degree of immune infiltration in individual tumors, we identified microenvironment cell signatures that delineated four immune-based ccRCC subtypes characterized by distinct cellular pathways. This study reports a large-scale proteogenomic analysis of ccRCC to discern the functional impact of genomic alterations and provides evidence for rational treatment selection stemming from ccRCC pathobiology.

Integrative Proteo-genomic Analysis to Construct CNA-protein Regulatory Map in Breast and Ovarian Tumors.

Recent development in high throughput proteomics and genomics profiling enable one to study regulations of genome alterations on protein activities in a systematic manner. In this article, we propose a new statistical method, ProMAP, to systematically characterize the regulatory relationships between proteins and DNA copy number alterations (CNA) in breast and ovarian tumors based on proteogenomic data from the CPTAC-TCGA studies. Because of the dynamic nature of mass spectrometry instruments, proteomics data from labeled mass spectrometry experiments usually have non-ignorable batch effects. Moreover, mass spectrometry based proteomic data often possesses high percentages of missing values and non-ignorable missing-data patterns. Thus, we use a linear mixed effects model to account for the batch structure and explicitly incorporate the abundance-dependent-missing-data mechanism of proteomic data in ProMAP. In addition, we employ a multivariate regression framework to characterize the multiple-to-multiple regulatory relationships between CNA and proteins. Further, we use proper statistical regularization to facilitate the detection of master genetic regulators, which affect the activities of many proteins and often play important roles in genetic regulatory networks. Improved performance of ProMAP over existing methods were illustrated through extensive simulation studies and real data examples. Applying ProMAP to the CPTAC-TCGA breast and ovarian cancer data sets, we identified many genome regions, including a few novel ones, whose CNA were associated with protein and or phosphoprotein abundances. For example, in breast tumors, a small region in 8p11.21 was recognized as the second biggest hub in the CNA-phosphoprotein regulatory map, and further investigation of the regulatory targets suggests the potential role of 8p11.21 CNA in perturbing oxygen binding and transport activities in tumor cells. This and other findings from our analyses help to characterize the impacts of CNAs on protein activity landscapes and cast light on the genetic regulation mechanisms underlying these tumors.

Insights into Impact of DNA Copy Number Alteration and Methylation on the Proteogenomic Landscape of Human Ovarian Cancer via a Multi-omics Integrative Analysis.

In this work, we propose iProFun, an integrative analysis tool to screen for proteogenomic functional traits perturbed by DNA copy number alterations (CNAs) and DNA methylations. The goal is to characterize functional consequences of DNA copy number and methylation alterations in tumors and to facilitate screening for cancer drivers contributing to tumor initiation and progression. Specifically, we consider three functional molecular quantitative traits: mRNA expression levels, global protein abundances, and phosphoprotein abundances. We aim to identify those genes whose CNAs and/or DNA methylations have cis-associations with either some or all three types of molecular traits. Compared with analyzing each molecular trait separately, the joint modeling of multi-omics data enjoys several benefits: iProFun experienced enhanced power for detecting significant cis-associations shared across different omics data types, and it also achieved better accuracy in inferring cis-associations unique to certain type(s) of molecular trait(s). For example, unique associations of CNAs/methylations to global/phospho protein abundances may imply posttranslational regulations.We applied iProFun to ovarian high-grade serous carcinoma tumor data from The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium and identified CNAs and methylations of 500 and 121 genes, respectively, affecting the cis-functional molecular quantitative traits of the corresponding genes. We observed substantial power gain via the joint analysis of iProFun. For example, iProFun identified 117 genes whose CNAs were associated with phosphoprotein abundances by leveraging mRNA expression levels and global protein abundances. By comparison, analyses based on phosphoprotein data alone identified none. A network analysis of these 117 genes revealed the known oncogene AKT1 as a key hub node interacting with many of the rest. In addition, iProFun identified one gene, BIN2, whose DNA methylation has cis-associations with its mRNA expression, global protein, and phosphoprotein abundances. These and other genes identified by iProFun could serve as potential drug targets for ovarian cancer.

Author Correction: Characterization of Nigerian breast cancer reveals prevalent homologous recombination deficiency and aggressive molecular features.

The original version of this Article contained an error in the author affiliations. The affiliation of Kevin P. White with Tempus Labs, Inc. Chicago, IL, USA was inadvertently omitted. This has now been corrected in both the PDF and HTML versions of the Article.

Using multivariate mixed-effects selection models for analyzing batch-processed proteomics data with non-ignorable missingness.

In quantitative proteomics, mass tag labeling techniques have been widely adopted in mass spectrometry experiments. These techniques allow peptides (short amino acid sequences) and proteins from multiple samples of a batch being detected and quantified in a single experiment, and as such greatly improve the efficiency of protein profiling. However, the batch-processing of samples also results in severe batch effects and non-ignorable missing data occurring at the batch level. Motivated by the breast cancer proteomic data from the Clinical Proteomic Tumor Analysis Consortium, in this work, we developed two tailored multivariate MIxed-effects SElection models (mvMISE) to jointly analyze multiple correlated peptides/proteins in labeled proteomics data, considering the batch effects and the non-ignorable missingness. By taking a multivariate approach, we can borrow information across multiple peptides of the same protein or multiple proteins from the same biological pathway, and thus achieve better statistical efficiency and biological interpretation. These two different models account for different correlation structures among a group of peptides or proteins. Specifically, to model multiple peptides from the same protein, we employed a factor-analytic random effects structure to characterize the high and similar correlations among peptides. To model biological dependence among multiple proteins in a functional pathway, we introduced a graphical lasso penalty on the error precision matrix, and implemented an efficient algorithm based on the alternating direction method of multipliers. Simulations demonstrated the advantages of the proposed models. Applying the proposed methods to the motivating data set, we identified phosphoproteins and biological pathways that showed different activity patterns in triple negative breast tumors versus other breast tumors. The proposed methods can also be applied to other high-dimensional multivariate analyses based on clustered data with or without non-ignorable missingness.

The contribution of parent-to-offspring transmission of telomeres to the heritability of telomere length in humans.

Leukocyte telomere length (LTL) is a heritable trait with two potential sources of heritability (h2): inherited variation in non-telomeric regions (e.g., SNPs that influence telomere maintenance) and variability in the lengths of telomeres in gametes that produce offspring zygotes (i.e., "direct" inheritance). Prior studies of LTL h2 have not attempted to disentangle these two sources. Here, we use a novel approach for detecting the direct inheritance of telomeres by studying the association between identity-by-descent (IBD) sharing at chromosome ends and phenotypic similarity in LTL. We measured genome-wide SNPs and LTL for a sample of 5069 Bangladeshi adults with substantial relatedness. For each of the 6318 relative pairs identified, we used SNPs near the telomeres to estimate the number of chromosome ends shared IBD, a proxy for the number of telomeres shared IBD (Tshared). We then estimated the association between Tshared and the squared pairwise difference in LTL ((ΔLTL)2) within various classes of relatives (siblings, avuncular, cousins, and distant), adjusting for overall genetic relatedness (ϕ). The association between Tshared and (ΔLTL)2 was inverse among all relative pair types. In a meta-analysis including all relative pairs (ϕ > 0.05), the association between Tshared and (ΔLTL)2 (P = 0.01) was stronger than the association between ϕ and (ΔLTL)2 (P = 0.43). Our results provide strong evidence that telomere length (TL) in parental germ cells impacts TL in offspring cells and contributes to LTL h2 despite telomere "reprogramming" during embryonic development. Applying our method to larger studies will enable robust estimation of LTL h2 attributable to direct transmission of telomeres.

Characterization of Nigerian breast cancer reveals prevalent homologous recombination deficiency and aggressive molecular features.

Racial/ethnic disparities in breast cancer mortality continue to widen but genomic studies rarely interrogate breast cancer in diverse populations. Through genome, exome, and RNA sequencing, we examined the molecular features of breast cancers using 194 patients from Nigeria and 1037 patients from The Cancer Genome Atlas (TCGA). Relative to Black and White cohorts in TCGA, Nigerian HR + /HER2 - tumors are characterized by increased homologous recombination deficiency signature, pervasive TP53 mutations, and greater structural variation-indicating aggressive biology. GATA3 mutations are also more frequent in Nigerians regardless of subtype. Higher proportions of APOBEC-mediated substitutions strongly associate with PIK3CA and CDH1 mutations, which are underrepresented in Nigerians and Blacks. PLK2, KDM6A, and B2M are also identified as previously unreported significantly mutated genes in breast cancer. This dataset provides novel insights into potential molecular mechanisms underlying outcome disparities and lay a foundation for deployment of precision therapeutics in underserved populations.

Germline Variation and Breast Cancer Incidence: A Gene-Based Association Study and Whole-Genome Prediction of Early-Onset Breast Cancer.

Background: Although germline genetics influences breast cancer incidence, published research only explains approximately half of the expected association. Moreover, the accuracy of prediction models remains low. For women who develop breast cancer early, the genetic architecture is less established.Methods: To identify loci associated with early-onset breast cancer, gene-based tests were carried out using exome array data from 3,479 women with breast cancer diagnosed before age 50 and 973 age-matched controls. Replication was undertaken in a population that developed breast cancer at all ages of onset.Results: Three gene regions were associated with breast cancer incidence: FGFR2 (P = 1.23 × 10-5; replication P < 1.00 × 10-6), NEK10 (P = 3.57 × 10-4; replication P < 1.00 × 10-6), and SIVA1 (P = 5.49 × 10-4; replication P < 1.00 × 10-6). Of the 151 gene regions reported in previous literature, 19 (12.5%) showed evidence of association (P < 0.05) with the risk of early-onset breast cancer in the early-onset population. To predict incidence, whole-genome prediction was implemented on a subset of 3,076 participants who were additionally genotyped on a genome wide array. The whole-genome prediction outperformed a polygenic risk score [AUC, 0.636; 95% confidence interval (CI), 0.614-0.659 compared with 0.601; 95% CI, 0.578-0.623], and when combined with known epidemiologic risk factors, the AUC rose to 0.662 (95% CI, 0.640-0.684).Conclusions: This research supports a role for variation within FGFR2 and NEK10 in breast cancer incidence, and suggests SIVA1 as a novel risk locus.Impact: This analysis supports a shared genetic etiology between women with early- and late-onset breast cancer, and suggests whole-genome data can improve risk assessment. Cancer Epidemiol Biomarkers Prev; 27(9); 1057-64. ©2018 AACR.