Cell-type deconvolution in epigenome-wide association studies: a review and recommendations

A novel method for cell deconvolution using DNA methylation in PCA space

BACKGROUND

In this study, we present a novel method for reference-based cell deconvolution using data from DNA methylation arrays. Different from existing methods like IDOL-Ext, which operate on probe-level data, our approach represents features in the principal component analysis (PCA) space for cell type deconvolution.

RESULTS

Our method's accuracy in estimating cell compositions is validated across various public datasets, including blood samples from glioma patients. It demonstrates precision comparable to IDOL-Ext, with R2 values ranging from 0.73 to 0.99 for most cell types, while offering improved discrimination between similar cell types, particularly T cell subtypes in glioma patient samples (R2 0.42-0.75 vs. 0.36-0.66 for IDOL-Ext). However, both methods showed lower accuracy for certain cell types, such as memory CD8 T cells in glioma patients (R2 0.42 vs. 0.36 for IDOL-Ext), highlighting the challenges in distinguishing closely related cell populations. We have made this method available as an R package "BloodCellDecon" on GitHub.

CONCLUSIONS

Our study confirms the efficacy of cell type deconvolution in PCA space. The results indicate wide-ranging applicability and potential for adaptation to other forms of genomic data.

Computational single-cell methods for predicting cancer risk

Despite recent biotechnological breakthroughs, cancer risk prediction remains a formidable computational and experimental challenge. Addressing it is critical in order to improve prevention, early detection and survival rates. Here, I briefly summarize some key emerging theoretical and computational challenges as well as recent computational advances that promise to help realize the goals of cancer-risk prediction. The focus is on computational strategies based on single-cell data, in particular on bottom-up network modeling approaches that aim to estimate cancer stemness and dedifferentiation at single-cell resolution from a systems-biological perspective. I will describe two promising methods, a tissue and cell-lineage independent one based on the concept of diffusion network entropy, and a tissue and cell-lineage specific one that uses transcription factor regulons. Application of these tools to single-cell and single-nucleus RNA-seq data from stages prior to invasive cancer reveal that they can successfully delineate the heterogeneous inter-cellular cancer-risk landscape, identifying those cells that are more likely to turn cancerous. Bottom-up systems biological modeling of single-cell omic data is a novel computational analysis paradigm that promises to facilitate the development of preventive, early detection and cancer-risk prediction strategies.

Epigenetic age acceleration in follicular fluid and markers of ovarian response among women undergoing IVF

STUDY QUESTION

Are markers of epigenetic age acceleration in follicular fluid associated with outcomes of ovarian stimulation?

SUMMARY ANSWER

Increased epigenetic age acceleration of follicular fluid using the Horvath clock, but not other epigenetic clocks (GrimAge and Granulosa Cell), was associated with lower peak estradiol levels and decreased number of total and mature oocytes.

WHAT IS KNOWN ALREADY

In granulosa cells, there are inconsistent findings between epigenetic age acceleration and ovarian response outcomes.

STUDY DESIGN, SIZE, DURATION

Our study included 61 women undergoing IVF at an academic fertility clinic in the New England area who were part of the Environment and Reproductive Health Study (2006-2016).

PARTICIPANTS/MATERIALS, SETTING, METHODS

Participants provided a follicular fluid sample during oocyte retrieval. DNA methylation of follicular fluid was assessed using a genome-wide methylation screening tool. Three established epigenetic clocks (Horvath, GrimAge, and Granulosa Cell) were used to predict DNA-methylation-based epigenetic age. To calculate the age acceleration, we regressed epigenetic age on chronological age and extracted the residuals. The association between epigenetic age acceleration and ovarian response outcomes (peak estradiol levels, follicle stimulation hormone, number of total, and mature oocytes) was assessed using linear and Poisson regression adjusted for chronological age, three surrogate variables (to account for cellular heterogeneity), race, smoking status, initial infertility diagnosis, and stimulation protocol.

MAIN RESULTS AND ROLE OF CHANCE

Compared to the median chronological age of our participants (34 years), the Horvath clock predicted, on an average, a younger epigenetic age (median: 24.2 years) while the GrimAge (median: 38.6 years) and Granulosa Cell (median: 39.0 years) clocks predicted, on an average, an older epigenetic age. Age acceleration based on the Horvath clock was associated with lower peak estradiol levels (-819.4 unit decrease in peak estradiol levels per standard deviation increase; 95% CI: -1265.7, -373.1) and fewer total (% change in total oocytes retrieved per standard deviation increase: -21.8%; 95% CI: -37.1%, -2.8%) and mature oocytes retrieved (% change in mature oocytes retrieved per standard deviation increase: -23.8%; 95% CI: -39.9%, -3.4%). The age acceleration based on the two other epigenetic clocks was not associated with markers of ovarian response.

LIMITATIONS, REASONS FOR CAUTION

Our sample size was small and we did not specifically isolate granulosa cells from follicular fluid samples so our samples could have included mixed cell types.

WIDER IMPLICATIONS OF THE FINDINGS

Our results highlight that certain epigenetic clocks may be predictive of ovarian stimulation outcomes when applied to follicular fluid; however, the inconsistent findings for specific clocks across studies indicate a need for further research to better understand the clinical utility of epigenetic clocks to improve IVF treatment.

STUDY FUNDING/COMPETING INTEREST(S)

The study was supported by grants ES009718, ES022955, ES000002, and ES026648 from the National Institute of Environmental Health Sciences (NIEHS) and a pilot grant from the NIEHS-funded HERCULES Center at Emory University (P30 ES019776). RBH was supported by the Emory University NIH Training Grant (T32-ES012870).

TRIAL REGISTRATION NUMBER

N/A.

CelFiE-ISH: a probabilistic model for multi-cell type deconvolution from single-molecule DNA methylation haplotypes

Deconvolution methods infer quantitative cell type estimates from bulk measurement of mixed samples including blood and tissue. DNA methylation sequencing measures multiple CpGs per read, but few existing deconvolution methods leverage this within-read information. We develop CelFiE-ISH, which extends an existing method (CelFiE) to use within-read haplotype information. CelFiE-ISH outperforms CelFiE and other existing methods, achieving 30% better accuracy and more sensitive detection of rare cell types. We also demonstrate the importance of marker selection and of tailoring markers for haplotype-aware methods. While here we use gold-standard short-read sequencing data, haplotype-aware methods will be well-suited for long-read sequencing.

Brain cell-type shifts in Alzheimer's disease, autism, and schizophrenia interrogated using methylomics and genetics

Few neuropsychiatric disorders have replicable biomarkers, prompting high-resolution and large-scale molecular studies. However, we still lack consensus on a more foundational question: whether quantitative shifts in cell types-the functional unit of life-contribute to neuropsychiatric disorders. Leveraging advances in human brain single-cell methylomics, we deconvolve seven major cell types using bulk DNA methylation profiling across 1270 postmortem brains, including from individuals diagnosed with Alzheimer's disease, schizophrenia, and autism. We observe and replicate cell-type compositional shifts for Alzheimer's disease (endothelial cell loss), autism (increased microglia), and schizophrenia (decreased oligodendrocytes), and find age- and sex-related changes. Multiple layers of evidence indicate that endothelial cell loss contributes to Alzheimer's disease, with comparable effect size to APOE genotype among older people. Genome-wide association identified five genetic loci related to cell-type composition, involving plausible genes for the neurovascular unit (P2RX5 and TRPV3) and excitatory neurons (DPY30 and MEMO1). These results implicate specific cell-type shifts in the pathophysiology of neuropsychiatric disorders.

Benchmarking of methods for DNA methylome deconvolution

Defining the number and abundance of different cell types in tissues is important for understanding disease mechanisms as well as for diagnostic and prognostic purposes. Typically, this is achieved by immunohistological analyses, cell sorting, or single-cell RNA-sequencing. Alternatively, cell-specific DNA methylome information can be leveraged to deconvolve cell fractions from a bulk DNA mixture. However, comprehensive benchmarking of deconvolution methods and modalities was not yet performed. Here we evaluate 16 deconvolution algorithms, developed either specifically for DNA methylome data or more generically. We assess the performance of these algorithms, and the effect of normalization methods, while modeling variables that impact deconvolution performance, including cell abundance, cell type similarity, reference panel size, method for methylome profiling (array or sequencing), and technical variation. We observe differences in algorithm performance depending on each these variables, emphasizing the need for tailoring deconvolution analyses. The complexity of the reference, marker selection method, number of marker loci and, for sequencing-based assays, sequencing depth have a marked influence on performance. By developing handles to select the optimal analysis configuration, we provide a valuable source of information for studies aiming to deconvolve array- or sequencing-based methylation data.

Validating reference-based algorithms to determine cell-type heterogeneity in ovarian cancer DNA methylation studies

Information about cell composition in tissue samples is crucial for biomarker discovery and prognosis. Specifically, cancer tissue samples present challenges in deconvolution studies due to mutations and genetic rearrangements. Here, we optimized a robust, DNA methylation-based protocol, to be used for deconvolution of ovarian cancer samples. We compared several state-of-the-art methods (HEpiDISH, MethylCIBERSORT and ARIC) and validated the proposed protocol in an in-silico mixture and in an external dataset containing samples from ovarian cancer patients and controls. The deconvolution protocol we eventually implemented is based on MethylCIBERSORT. Comparing deconvolution methods, we paid close attention to the role of a reference panel. We postulate that a possibly high number of samples (in our case: 247) should be used when building a reference panel to ensure robustness and to compensate for biological and technical variation between samples. Subsequently, we tested the performance of the validated protocol in our own study cohort, consisting of 72 patients with malignant and benign ovarian disease as well as in five external cohorts. In conclusion, we refined and validated a reference-based algorithm to determine cell type composition of ovarian cancer tissue samples to be used in cancer biology studies in larger cohorts.

Placental and Immune Cell DNA Methylation Reference Panel for Bulk Tissue Cell Composition Estimation in Epidemiological Studies

To distinguish DNA methylation (DNAm) from cell proportion changes in whole placental tissue research, we developed a robust cell type-specific DNAm reference to estimate cell composition. We collated newly collected and existing cell type DNAm profiles quantified via Illumina EPIC or 450k microarrays. To estimate cell composition, we deconvoluted whole placental samples (n=36) with robust partial correlation based on the top 50 hyper- and hypomethylated sites per cell type. To test deconvolution performance, we evaluated RMSE in predicting principal component one of DNAm variation in 204 external placental samples. We analyzed DNAm profiles (n=368,435 sites) from 12 cell types: cytotrophoblasts (n=18), endothelial cells (n=19), Hofbauer cells (n=26), stromal cells (n=21), syncytiotrophoblasts (n=4), six lymphocyte types (n=36), and nucleated red blood cells (n=11). Median cell composition was consistent with placental biology: 60.4% syncytiotrophoblast, 17.1% stromal, 8.8% endothelial, 4.5% cytotrophoblast, 3.9% Hofbauer, 1.7% nucleated red blood cells, and 1.2% neutrophils. Our expanded reference outperformed an existing reference in predicting DNAm variation (15.4% variance explained, IQR=21.61) with cell composition estimates (RMSE:10.51 vs. 11.43, p-value<0.001). This cell type reference can robustly estimate cell composition from whole placental DNAm data to detect important cell types, reveal biological mechanisms, and improve casual inference.

Associations in cell type-specific hydroxymethylation and transcriptional alterations of pediatric central nervous system tumors

Although intratumoral heterogeneity has been established in pediatric central nervous system tumors, epigenomic alterations at the cell type level have largely remained unresolved. To identify cell type-specific alterations to cytosine modifications in pediatric central nervous system tumors, we utilize a multi-omic approach that integrated bulk DNA cytosine modification data (methylation and hydroxymethylation) with both bulk and single-cell RNA-sequencing data. We demonstrate a large reduction in the scope of significantly differentially modified cytosines in tumors when accounting for tumor cell type composition. In the progenitor-like cell types of tumors, we identify a preponderance differential Cytosine-phosphate-Guanine site hydroxymethylation rather than methylation. Genes with differential hydroxymethylation, like histone deacetylase 4 and insulin-like growth factor 1 receptor, are associated with cell type-specific changes in gene expression in tumors. Our results highlight the importance of epigenomic alterations in the progenitor-like cell types and its role in cell type-specific transcriptional regulation in pediatric central nervous system tumors.

Computational deconvolution of DNA methylation data from mixed DNA samples

In this review, we provide a comprehensive overview of the different computational tools that have been published for the deconvolution of bulk DNA methylation (DNAm) data. Here, deconvolution refers to the estimation of cell-type proportions that constitute a mixed sample. The paper reviews and compares 25 deconvolution methods (supervised, unsupervised or hybrid) developed between 2012 and 2023 and compares the strengths and limitations of each approach. Moreover, in this study, we describe the impact of the platform used for the generation of methylation data (including microarrays and sequencing), the applied data pre-processing steps and the used reference dataset on the deconvolution performance. Next to reference-based methods, we also examine methods that require only partial reference datasets or require no reference set at all. In this review, we provide guidelines for the use of specific methods dependent on the DNA methylation data type and data availability.

Prenatal cocaine exposure and its influence on pediatric epigenetic clocks and epigenetic scores in humans

The investigation of the effects of prenatal cocaine exposure (PCE) on offspring has been inconsistent, with few studies investigating biological outcomes in humans. We profiled genome-wide DNA methylation (DNAm) of umbilical cord blood (UCB) from newborns with (n = 35) and without (n = 47) PCE. We used DNAm data to (1) assess pediatric epigenetic clocks at birth and (2) to estimate epigenetic scores (ES) for lifetime disorders. We generated gestational epigenetic age estimates (DNAmGA) based on Knight and Bohlin epigenetic clocks. We also investigated the association between DNAmGA and UCB serum brain-derived neurotrophic factor (BDNF) levels. Considering the large-scale DNAm data availability and existing evidence regarding PCE as a risk for health problems later in life, we generated ES for tobacco smoking, psychosis, autism, diabetes, and obesity. A gene ontology (GO) analysis on the CpGs included in the ES with group differences was performed. PCE was associated with lower DNAmGA in newborns, and this effect remained significant when controlling for potential confounders, such as blood cell type composition predicted by DNAm and obstetric data. DNAmGA was negatively correlated with BDNF levels in the serum of UCB. Higher tobacco smoking, psychosis, and diabetes ES were found in the PCE group. The GO analysis revealed GABAergic synapses as a potential pathway altered by PCE. Our findings of decelerated DNAmGA and ES for adverse phenotypes associated with PCE, suggest that the effects of gestational cocaine exposure on the epigenetic landscape of human newborns are detectable at birth.

Immunological shifts during early-stage Parkinson's disease identified with DNA methylation data on longitudinally collected blood samples

Parkinson's disease (PD) is the second most common neurodegenerative disease in the United States. Decades before motor symptoms manifest, non-motor symptoms such as hyposmia and rapid eye movement (REM) sleep behavior disorder are highly predictive of PD. Previous immune profiling studies have identified alterations to the proportions of immune cells in the blood of clinically defined PD patients. However, it remains unclear if these phenotypes manifest before the clinical diagnosis of PD. We utilized longitudinal DNA methylation (DNAm) microarray data from the Parkinson's Progression Marker's Initiative (PPMI) to perform immune profiling in clinically defined PD and prodromal PD patients (Prod). We identified previously reported changes in neutrophil, monocyte, and T cell numbers in PD patients. Additionally, we noted previously unrecognized decreases in the naive B cell compartment in the defined PD and Prod patient group. Over time, we observed the proportion of innate immune cells in PD blood increased, but the proportion of adaptive immune cells decreased. We identified decreases in T and B cell subsets associated with REM sleep disturbances and early cognitive decline. Lastly, we identified increases in B memory cells associated with both genetic (LRRK2 genotype) and infectious (cytomegalovirus seropositivity) risk factors of PD. Our analysis shows that the peripheral immune system is dynamic as the disease progresses. The study provides a platform to understand how and when peripheral immune alterations occur in PD and whether intervention at particular stages may be therapeutically advantageous.

Matched analysis of detailed peripheral blood and tumor immune microenvironment profiles in bladder cancer

Background: Bladder cancer and therapy responses hinge on immune profiles in the tumor microenvironment (TME) and blood, yet studies linking tumor-infiltrating immune cells to peripheral immune profiles are limited. Methods: DNA methylation cytometry quantified TME and matched peripheral blood immune cell proportions. With tumor immune profile data as the input, subjects were grouped by immune infiltration status and consensus clustering. Results: Immune hot and cold groups had different immune compositions in the TME but not in circulating blood. Two clusters of patients identified with consensus clustering had different immune compositions not only in the TME but also in blood. Conclusion: Detailed immune profiling via methylation cytometry reveals the significance of understanding tumor and systemic immune relationships in cancer patients.

Metastable epialleles in humans

First identified in isogenic mice, metastable epialleles (MEs) are loci where the extent of DNA methylation (DNAm) is variable between individuals but correlates across tissues derived from different germ layers within a given individual. This property, termed systemic interindividual variation (SIV), is attributed to stochastic methylation establishment before germ layer differentiation. Evidence suggests that some putative human MEs are sensitive to environmental exposures in early development. In this review we introduce key concepts pertaining to human MEs, describe methods used to identify MEs in humans, and review their genomic features. We also highlight studies linking DNAm at putative human MEs to early environmental exposures and postnatal (including disease) phenotypes.

DNA methylation modulated genetic variant effect on gene transcriptional regulation

BACKGROUND

Expression quantitative trait locus (eQTL) analysis has emerged as an important tool in elucidating the link between genetic variants and gene expression, thereby bridging the gap between risk SNPs and associated diseases. We recently identified and validated a specific case where the methylation of a CpG site influences the relationship between the genetic variant and gene expression.

RESULTS

Here, to systematically evaluate this regulatory mechanism, we develop an extended eQTL mapping method, termed DNA methylation modulated eQTL (memo-eQTL). Applying this memo-eQTL mapping method to 128 normal prostate samples enables identification of 1063 memo-eQTLs, the majority of which are not recognized as conventional eQTLs in the same cohort. We observe that the methylation of the memo-eQTL CpG sites can either enhance or insulate the interaction between SNP and gene expression by altering CTCF-based chromatin 3D structure.

CONCLUSIONS

This study demonstrates the prevalence of memo-eQTLs paving the way to identify novel causal genes for traits or diseases associated with genetic variations.

Uncertainty quantification of reference-based cellular deconvolution algorithms

The majority of epigenetic epidemiology studies to date have generated genome-wide profiles from bulk tissues (e.g., whole blood) however these are vulnerable to confounding from variation in cellular composition. Proxies for cellular composition can be mathematically derived from the bulk tissue profiles using a deconvolution algorithm; however, there is no method to assess the validity of these estimates for a dataset where the true cellular proportions are unknown. In this study, we describe, validate and characterize a sample level accuracy metric for derived cellular heterogeneity variables. The CETYGO score captures the deviation between a sample's DNA methylation profile and its expected profile given the estimated cellular proportions and cell type reference profiles. We demonstrate that the CETYGO score consistently distinguishes inaccurate and incomplete deconvolutions when applied to reconstructed whole blood profiles. By applying our novel metric to >6,300 empirical whole blood profiles, we find that estimating accurate cellular composition is influenced by both technical and biological variation. In particular, we show that when using a common reference panel for whole blood, less accurate estimates are generated for females, neonates, older individuals and smokers. Our results highlight the utility of a metric to assess the accuracy of cellular deconvolution, and describe how it can enhance studies of DNA methylation that are reliant on statistical proxies for cellular heterogeneity. To facilitate incorporating our methodology into existing pipelines, we have made it freely available as an R package (https://github.com/ds420/CETYGO).

Alterations to DNA methylation patterns induced by chemotherapy treatment are associated with negative impacts on the olfactory pathway

BACKGROUND

Exposure to cytotoxic chemotherapy treatment may alter DNA methylation (DNAm) in breast cancer patients.

METHODS

We performed DNAm analysis in 125 breast cancer patients with blood drawn before and after chemotherapy, using the Illumina MethylationEPIC array. DNAm changes of 588,798 individual CpGs (including 41,207 promoter regions) were evaluated using linear regression models adjusted for monocyte proportion. Gene set enrichment analyses (GSEA) were conducted to identify key Gene Ontology (GO) biological processes or Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with chemotherapy. Results were validated in a separate cohort of breast cancer patients who were treated (n = 1273) and not treated (n = 872) by chemotherapy (1808 blood, 337 saliva).

RESULTS

A total of 141 differentially methylated CpGs and 11 promoters were significantly associated with chemotherapy after multiple testing corrections in both the paired sample and single time point analyses. GSEA of promoter regions (pre-ranked by test statistics) identified six suppressed biological processes (p < 4.67e-8) related to sensory perception and detection of chemical stimuli, including smell perception (GO:0007606, GO:0007608, GO:0009593, GO:0050906, GO:0050907, and GO:0050911). The same six biological processes were significantly suppressed in the validation dataset (p < 9.02e-14). The KEGG pathway olfactory transduction (hsa04740) was also found to be significantly suppressed (ppaired-samples = 1.72e-9, psingle-timepoint-blood = 2.03e-15 and psingle-timepoint-saliva = 7.52e-56).

CONCLUSION

The enrichment of imprinted genes within biological processes and pathways suggests a biological mechanism by which chemotherapy could affect the perception of smell.

A crosstalk between epigenetic modulations and non-alcoholic fatty liver disease progression

Non-alcoholic fatty liver disease (NAFLD) has recently emerged as a major public health concern worldwide due to its rapidly rising prevalence and its potential to progress into end-stage liver disease. While the precise pathophysiology underlying NAFLD remains incompletely understood, it is strongly associated with various environmental triggers and other metabolic disorders. Epigenetics examines changes in gene expression that are not caused by alterations in the DNA sequence itself. There is accumulating evidence that epigenetics plays a key role in linking environmental cues to the onset and progression of NAFLD. Our understanding of how epigenetic mechanisms contribute to NAFLD pathophysiology has expanded considerably in recent years as research on the epigenetics of NAFLD has developed. This review summarizes recent insights into major epigenetic processes that have been implicated in NAFLD pathogenesis including DNA methylation, histone acetylation, and microRNAs that have emerged as promising targets for further investigation. Elucidating epigenetic mechanisms in NAFLD may uncover novel diagnostic biomarkers and therapeutic targets for this disease. However, many questions have remained unanswered regarding how epigenetics promotes NAFLD onset and progression. Additional studies are needed to further characterize the epigenetic landscape of NAFLD and validate the potential of epigenetic markers as clinical tools. Nevertheless, an enhanced understanding of the epigenetic underpinnings of NAFLD promises to provide key insights into disease mechanisms and pave the way for novel prognostic and therapeutic approaches.

Do not be scared of the genome's 5th base-Explaining phenotypic variability and evolutionary dynamics through DNA methylation analysis

Epigenetic processes have taken center stage for the investigation of many biological processes, and epigenetic modifications have shown to influence phenotype, morphology and behavioural traits such as stress resistance by affecting gene regulation and expression without altering the underlying genomic sequence. The multiple molecular layers of epigenetics synergistically construct the cell type-specific gene regulatory networks, characterized by a high degree of plasticity and redundancy to create cell-type-specific morphology and function. DNA methylation occurring on the 5' carbon of cytosines in different genomic sequence contexts is the most studied epigenetic modification. DNA methylation has been shown to provide a molecular record of the exposure to a large variety of environmental factors, which might be persistent through the entire lifetime of an organism and even be passed onto the offspring. Animals might display altered phenotypes mediated by epigenetic modifications depending on the developmental stage or the environmental conditions as well as during evolution. Therefore, the analysis of DNA methylation patterns might allow deciphering previous exposures, explaining ecologically relevant phenotypic diversity and predicting evolutionary trajectories enabling accelerated adaption to changing environmental conditions. Despite the explanatory potential of DNA methylation integrating genetic and environmental factors to shape phenotypic variation and contribute significantly to evolutionary dynamics, studies of DNA methylation are still scarce in the field of ecology. This might be at least partly due to the complexity of DNA methylation analysis and the interpretation of the acquired data. In the current issue of Molecular Ecology Resources, Laine and colleagues (Molecular Ecology Resources, 2022) provide a detailed summary of guidelines and valuable recommendations for researchers in the field of ecology to avoid common pitfalls and perform interpretable genome-wide DNA methylation analyses.

Gene-regulation modules in nonalcoholic fatty liver disease revealed by single-nucleus ATAC-seq

We investigated the progression of nonalcoholic fatty liver disease from fatty liver to steatohepatitis using single-nucleus and bulk ATAC-seq on the livers of rats fed a high-fat diet (HFD). Rats fed HFD for 4 wk developed fatty liver, and those fed HFD for 8 wk further progressed to steatohepatitis. We observed an increase in the proportion of inflammatory macrophages, consistent with the pathological progression. Utilizing machine learning, we divided global gene regulation into modules, wherein transcription factors within a module could regulate genes within the same module, reaffirming known regulatory relationships between transcription factors and biological processes. We identified core genes-central to co-expression and protein-protein interaction-for the biological processes discovered. Notably, a large part of the core genes overlapped with genes previously implicated in nonalcoholic fatty liver disease. Single-nucleus ATAC-seq, combined with data-driven statistical analysis, offers insight into in vivo global gene regulation as a combination of modules and assists in identifying core genes of relevant biological processes.

Genome-Wide DNA Methylation Profiles in Whole-Blood and Buccal Samples-Cross-Sectional, Longitudinal, and across Platforms

The field of DNA methylation research is rapidly evolving, focusing on disease and phenotype changes over time using methylation measurements from diverse tissue sources and multiple array platforms. Consequently, identifying the extent of longitudinal, inter-tissue, and inter-platform variation in DNA methylation is crucial for future advancement. DNA methylation was measured in 375 individuals, with 197 of those having 2 blood sample measurements ~10 years apart. Whole-blood samples were measured on Illumina Infinium 450K and EPIC methylation arrays, and buccal samples from a subset of 58 participants were measured on EPIC array. The data were analyzed with the aims to examine the correlation between methylation levels in longitudinal blood samples in 197 individuals, examine the correlation between methylation levels in the blood and buccal samples in 58 individuals, and examine the correlation between blood methylation profiles assessed on the EPIC and 450K arrays in 83 individuals. We identified 136,833, 7674, and 96,891 CpGs significantly and strongly correlated (>0.50) longitudinally, across blood and buccal samples as well as array platforms, respectively. A total of 3674 of these CpGs were shared across all three sets. Analysis of these shared CpGs identified previously found associations with aging, ancestry, and 7016 mQTLs as well.

DNA methylation and 28-year cardiovascular disease risk in type 1 diabetes: the Epidemiology of Diabetes Complications (EDC) cohort study

BACKGROUND

The potential for DNA methylation (DNAm) as an early marker for cardiovascular disease (CVD) and how such an association might differ by glycemic exposure has not been examined in type 1 diabetes, a population at increased CVD risk. We thus performed a prospective epigenome-wide association study of blood leukocyte DNAm (EPIC array) and time to CVD incidence over 28 years in a childhood-onset (< 17 years) type 1 diabetes cohort, the Pittsburgh Epidemiology of Diabetes Complications (EDC) study (n = 368 with DNA and no CVD at baseline), both overall and separately by glycemic exposure, as measured by HbA1c at baseline (split at the median: < 8.9% and ≥ 8.9%). We also assessed whether DNAm-CVD associations were independent of established cardiometabolic risk factors, including body mass index, estimated glucose disposal rate, cholesterol, triglycerides, blood pressure, pulse rate, albumin excretion rate, and estimated glomerular filtration rate.

RESULTS

CVD (first instance of CVD death, myocardial infarction, coronary revascularization, ischemic ECG, angina, or stroke) developed in 172 participants (46.7%) over 28 years. Overall, in Cox regression models for time to CVD, none of the 683,597 CpGs examined reached significance at a false discovery rate (FDR) ≤ 0.05. In participants with HbA1c < 8.9% (n = 180), again none reached FDR ≤ 0.05, but three were associated at the a priori nominal significance level FDR ≤ 0.10: cg07147033 in MIB2, cg12324048 (intergenic, chromosome 3), and cg15883830 (intergenic, chromosome 1). In participants with HbA1c ≥ 8.9% (n = 188), two CpGs in loci involved in calcium channel activity were significantly associated with CVD (FDR ≤ 0.05): cg21823999 in GPM6A and cg23621817 in CHRNA9; four additional CpGs were nominally associated (FDR ≤ 0.10). In participants with HbA1c ≥ 8.9%, DNAm-CVD associations were only modestly attenuated after cardiometabolic risk factor adjustment, while attenuation was greater in those with HbA1c < 8.9%. No pathways were enriched in those with HbA1c < 8.9%, while pathways for calcium channel activity and integral component of synaptic membrane were significantly enriched in those with HbA1c ≥ 8.9%.

CONCLUSIONS

These results provide novel evidence that DNAm at loci involved in calcium channel activity and development may contribute to long-term CVD risk beyond known risk factors in type 1 diabetes, particularly in individuals with greater glycemic exposure, warranting further study.

DNA methylation is differentially associated with glycemic outcomes by different types of weight-loss interventions: an epigenome-wide association study

BACKGROUND

Alterations in DNA methylation (DNAm) have been reported to be a mechanism by which bariatric surgeries resulted in considerable metabolic improvements. Previous studies have mostly focused on change in DNAm following weight-loss interventions, yet whether DNAm prior to intervention can explain the variability in glycemic outcomes has not been investigated. Here, we aim to examine whether baseline DNAm is differentially associated with glycemic outcomes induced by different types of weight-loss interventions.

METHODS

Participants were 75 adults with severe obesity who underwent non-surgical intensive medical intervention (IMI), adjustable gastric band (BAND) or Roux-en-Y gastric bypass (RYGB) (n = 25 each). Changes in fasting plasma glucose (FPG) and glycated hemoglobin (HbA1c) were measured at 1-year after intervention. DNAm was quantified by Illumina 450 K arrays in baseline peripheral blood DNA. Epigenome-wide association studies were performed to identify CpG probes that modify the effects of different weight-loss interventions on glycemic outcomes, i.e., changes in FPG and HbA1c, by including an interaction term between types of intervention and DNAm. Models were adjusted for weight loss and baseline clinical factors.

RESULTS

Baseline DNAm levels at 3216 and 117 CpGs were differentially associated with changes in FPG and HbA1c, respectively, when comparing RYGB versus IMI. Of these, 79 CpGs were significant for both FPG and HbA1c. The identified genes are enriched in adaptive thermogenesis, temperature homeostasis and regulation of cell population proliferation. Additionally, DNAm at 6 CpGs was differentially associated with changes in HbA1c when comparing RYGB versus BAND.

CONCLUSIONS

Baseline DNAm is differentially associated with glycemic outcomes in response to different types of weight-loss interventions, independent of weight loss and other clinical factors. Such findings provided initial evidence that baseline DNAm levels may serve as potential biomarkers predictive of differential glycemic outcomes in response to different types of weight-loss interventions.

Mixed model-based deconvolution of cell-state abundances (MeDuSA) along a one-dimensional trajectory

Deconvoluting cell-state abundances from bulk RNA-sequencing data can add considerable value to existing data, but achieving fine-resolution and high-accuracy deconvolution remains a challenge. Here we introduce MeDuSA, a mixed model-based method that leverages single-cell RNA-sequencing data as a reference to estimate cell-state abundances along a one-dimensional trajectory in bulk RNA-sequencing data. The advantage of MeDuSA lies primarily in estimating cell abundance in each state while fitting the remaining cells of the same type individually as random effects. Extensive simulations and real-data benchmark analyses demonstrate that MeDuSA greatly improves the estimation accuracy over existing methods for one-dimensional trajectories. Applying MeDuSA to cohort-level RNA-sequencing datasets reveals associations of cell-state abundances with disease or treatment conditions and cell-state-dependent genetic control of transcription. Our study provides a high-accuracy and fine-resolution method for cell-state deconvolution along a one-dimensional trajectory and demonstrates its utility in characterizing the dynamics of cell states in various biological processes.

Differences in white blood cell proportions between schizophrenia cases and controls are influenced by medication and variations in time of day

Cases with schizophrenia (SCZ) and healthy controls show differences in white blood cell (WBC) counts and blood inflammation markers. Here, we investigate whether time of blood draw and treatment with psychiatric medications are related to differences in estimated WBC proportions between SCZ cases and controls. DNA methylation data from whole blood was used to estimate proportions of six subtypes of WBCs in SCZ patients (n = 333) and healthy controls (n = 396). We tested the association of case-control status with estimated cell-type proportions and the neutrophil-to-lymphocyte ratio (NLR) in 4 models: with/without adjusting for time of blood draw, and then compared results from blood samples drawn during a 12-h (07:00-19:00) or 7-h (07:00-14:00) period. We also investigated WBC proportions in a subgroup of medication-free patients (n = 51). Neutrophil proportions were significantly higher in SCZ cases (mean=54.1%) vs. controls (mean=51.1%; p = <0.001), and CD8⁺T lymphocyte proportions were lower in SCZ cases (mean=12.1%) vs. controls (mean=13.2%; p = 0.001). The effect sizes in the 12-h sample (07:00-19:00) showed a significant difference between SCZ vs. controls for neutrophils, CD4⁺T, CD8⁺T, and B-cells, which remained significant after adjusting for time of blood draw. In the samples matched for time of blood draw during 07.00-14.00, we also observed an association with neutrophils, CD4⁺T, CD8⁺T, and B-cells that was unaffected by further adjustment for time of blood draw. In the medication-free patients, we observed differences that remained significant in neutrophils (p = 0.01) and CD4⁺T (p = 0.01) after adjusting for time of day. The association of SCZ with NLR was significant in all models (range: p < 0.001 to p = 0.03) in both medicated and unmedicated patients. In conclusion, controlling for pharmacological treatment and circadian cycling of WBC is necessary for unbiased estimates in case-control studies. Nevertheless, the association of WBC with SCZ remains, even after adjusting for the time of day.

Dynamics and consequences of nutrition-related microbial dysbiosis in early life: study protocol of the VITERBI GUT project

Introduction

Early life under- and overnutrition (jointly termed malnutrition) is increasingly recognized as an important risk factor for adult obesity and metabolic syndrome, a diet-related cluster of conditions including high blood sugar, fat and cholesterol. Nevertheless, the exact factors linking early life malnutrition with metabolic syndrome remain poorly characterized. We hypothesize that the microbiota plays a crucial role in this trajectory and that the pathophysiological mechanisms underlying under- and overnutrition are, to some extent, shared. We further hypothesize that a "dysbiotic seed microbiota" is transmitted to children during the birth process, altering the children's microbiota composition and metabolic health. The overall objective of this project is to understand the precise causes and biological mechanisms linking prenatal or early life under- or overnutrition with the predisposition to develop overnutrition and/or metabolic disease in later life, as well as to investigate the possibility of a dysbiotic seed microbiota inheritance in the context of maternal malnutrition.

Methods/design

VITERBI GUT is a prospective birth cohort allowing to study the link between early life malnutrition, the microbiota and metabolic health. VITERBI GUT will include 100 undernourished, 100 normally nourished and 100 overnourished pregnant women living in Vientiane, Lao People's Democratic Republic (PDR). Women will be recruited during their third trimester of pregnancy and followed with their child until its second birthday. Anthropometric, clinical, metabolic and nutritional data are collected from both the mother and the child. The microbiota composition of maternal and child's fecal and oral samples as well as maternal vaginal and breast milk samples will be determined using amplicon and shotgun metagenomic sequencing. Epigenetic modifications and lipid profiles will be assessed in the child's blood at 2 years of age. We will investigate for possible associations between metabolic health, epigenetics, and microbial changes.

Discussion

We expect the VITERBI GUT project to contribute to the emerging literature linking the early life microbiota, epigenetic changes and growth/metabolic health. We also expect this project to give new (molecular) insights into the mechanisms linking malnutrition-induced early life dysbiosis and metabolic health in later life, opening new avenues for microbiota-engineering using microbiota-targeted interventions.

Placental cell type deconvolution reveals that cell proportions drive preeclampsia gene expression differences

The placenta mediates adverse pregnancy outcomes, including preeclampsia, which is characterized by gestational hypertension and proteinuria. Placental cell type heterogeneity in preeclampsia is not well-understood and limits mechanistic interpretation of bulk gene expression measures. We generated single-cell RNA-sequencing samples for integration with existing data to create the largest deconvolution reference of 19 fetal and 8 maternal cell types from placental villous tissue (n = 9 biological replicates) at term (n = 40,494 cells). We deconvoluted eight published microarray case-control studies of preeclampsia (n = 173 controls, 157 cases). Preeclampsia was associated with excess extravillous trophoblasts and fewer mesenchymal and Hofbauer cells. Adjustment for cellular composition reduced preeclampsia-associated differentially expressed genes (log2 fold-change cutoff = 0.1, FDR < 0.05) from 1154 to 0, whereas downregulation of mitochondrial biogenesis, aerobic respiration, and ribosome biogenesis were robust to cell type adjustment, suggesting direct changes to these pathways. Cellular composition mediated a substantial proportion of the association between preeclampsia and FLT1 (37.8%, 95% CI [27.5%, 48.8%]), LEP (34.5%, 95% CI [26.0%, 44.9%]), and ENG (34.5%, 95% CI [25.0%, 45.3%]) overexpression. Our findings indicate substantial placental cellular heterogeneity in preeclampsia contributes to previously observed bulk gene expression differences. This deconvolution reference lays the groundwork for cellular heterogeneity-aware investigation into placental dysfunction and adverse birth outcomes.

CeDAR: incorporating cell type hierarchy improves cell type-specific differential analyses in bulk omics data

Bulk high-throughput omics data contain signals from a mixture of cell types. Recent developments of deconvolution methods facilitate cell type-specific inferences from bulk data. Our real data exploration suggests that differential expression or methylation status is often correlated among cell types. Based on this observation, we develop a novel statistical method named CeDAR to incorporate the cell type hierarchy in cell type-specific differential analyses of bulk data. Extensive simulation and real data analyses demonstrate that this approach significantly improves the accuracy and power in detecting cell type-specific differential signals compared with existing methods, especially in low-abundance cell types.

Nucleated red blood cells explain most of the association between DNA methylation and gestational age

Determining if specific cell type(s) are responsible for an association between DNA methylation (DNAm) and a given phenotype is important for understanding the biological mechanisms underlying the association. Our EWAS of gestational age (GA) in 953 newborns from the Norwegian MoBa study identified 13,660 CpGs significantly associated with GA (p_Bonferroni<0.05) after adjustment for cell type composition. When the CellDMC algorithm was applied to explore cell-type specific effects, 2,330 CpGs were significantly associated with GA, mostly in nucleated red blood cells [nRBCs; n = 2,030 (87%)]. Similar patterns were found in another dataset based on a different array and when applying an alternative algorithm to CellDMC called Tensor Composition Analysis (TCA). Our findings point to nRBCs as the main cell type driving the DNAm-GA association, implicating an epigenetic signature of erythropoiesis as a likely mechanism. They also explain the poor correlation observed between epigenetic age clocks for newborns and those for adults.

A new blood based epigenetic age predictor for adolescents and young adults

Children have special rights for protection compared to adults in our society. However, more than 1/4 of children globally have no documentation of their date of birth. Hence, there is a pressing need to develop biological methods for chronological age prediction, robust to differences in genetics, psychosocial events and physical living conditions. At present, DNA methylation is the most promising biological biomarker applied for age assessment. The human genome contains around 28 million DNA methylation sites, many of which change with age. Several epigenetic clocks accurately predict chronological age using methylation levels at age associated GpG-sites. However, variation in DNA methylation increases with age, and there is no epigenetic clock specifically designed for adolescents and young adults. Here we present a novel age Predictor for Adolescents and Young Adults (PAYA), using 267 CpG methylation sites to assess the chronological age of adolescents and young adults. We compared different preprocessing approaches and investigated the effect on prediction performance of the epigenetic clock. We evaluated performance using an independent validation data set consisting of 18-year-old individuals, where we obtained a median absolute deviation of just below 0.7 years. This tool may be helpful in age assessment of adolescents and young adults. However, there is a need to investigate the robustness of the age predictor across geographical and disease populations as well as environmental effects.

Data processing choices can affect findings in differential methylation analyses: an investigation using data from the LIMIT RCT

Objective

A wide array of methods exist for processing and analysing DNA methylation data. We aimed to perform a systematic comparison of the behaviour of these methods, using cord blood DNAm from the LIMIT RCT, in relation to detecting hypothesised effects of interest (intervention and pre-pregnancy maternal BMI) as well as effects known to be spurious, and known to be present.

Methods

DNAm data, from 645 cord blood samples analysed using Illumina 450K BeadChip arrays, were normalised using three different methods (with probe filtering undertaken pre- or post- normalisation). Batch effects were handled with a supervised algorithm, an unsupervised algorithm, or adjustment in the analysis model. Analysis was undertaken with and without adjustment for estimated cell type proportions. The effects estimated included intervention and BMI (effects of interest in the original study), infant sex and randomly assigned groups. Data processing and analysis methods were compared in relation to number and identity of differentially methylated probes, rankings of probes by p value and log-fold-change, and distributions of p values and log-fold-change estimates.

Results

There were differences corresponding to each of the processing and analysis choices. Importantly, some combinations of data processing choices resulted in a substantial number of spurious 'significant' findings. We recommend greater emphasis on replication and greater use of sensitivity analyses.

TXNIP DNA methylation is associated with glycemic control over 28 years in type 1 diabetes: findings from the Pittsburgh Epidemiology of Diabetes Complications (EDC) study

INTRODUCTION

DNA methylation (DNAme) has been cross-sectionally associated with type 2 diabetes and hemoglobin A1c (HbA1c) in the general population. However, longitudinal data and data in type 1 diabetes are currently very limited. Thus, we performed an epigenome-wide association study (EWAS) in an observational type 1 diabetes cohort to identify loci with DNAme associated with concurrent and future HbA1cs, as well as other clinical risk factors, over 28 years.

RESEARCH DESIGN AND METHODS

Whole blood DNAme in 683 597 CpGs was analyzed in the Pittsburgh Epidemiology of Diabetes Complications study of childhood onset (<17 years) type 1 diabetes (n=411). An EWAS of DNAme beta values and concurrent HbA1c was performed using linear models adjusted for diabetes duration, sex, pack years of smoking, estimated cell type composition variables, and technical/batch covariates. A longitudinal EWAS of subsequent repeated HbA1c measures was performed using mixed models. We further identified methylation quantitative trait loci (meQTLs) for significant CpGs and conducted a Mendelian randomization.

RESULTS

DNAme at cg19693031 (Chr 1, Thioredoxin-Interacting Protein (TXNIP)) and cg21534330 (Chr 17, Casein Kinase 1 Isoform Delta) was significantly inversely associated with concurrent HbA1c. In longitudinal analyses, hypomethylation of cg19693031 was associated with consistently higher HbA1c over 28 years, and with higher triglycerides, pulse rate, and albumin:creatinine ratio (ACR) independently of HbA1c. We further identified 34 meQTLs in SLC2A1/SLC2A1-AS1 significantly associated with cg19693031 DNAme.

CONCLUSIONS

Our results extend prior findings that TXNIP hypomethylation relates to worse glycemic control in type 1 diabetes by demonstrating the association persists over the long term. Additionally, the associations with triglycerides, pulse rate, and ACR suggest TXNIP DNAme could play a role in vascular damage independent of HbA1c. These findings strengthen potential for interventions targeting TXNIP to improve glycemic control in type 1 diabetes through its role in SLC2A1/glucose transporter 1-mediated glucose regulation.

A systematic assessment of cell type deconvolution algorithms for DNA methylation data

We performed systematic assessment of computational deconvolution methods that play an important role in the estimation of cell type proportions from bulk methylation data. The proposed framework methylDeConv (available as an R package) integrates several deconvolution methods for methylation profiles (Illumina HumanMethylation450 and MethylationEPIC arrays) and offers different cell-type-specific CpG selection to construct the extended reference library which incorporates the main immune cell subsets, epithelial cells and cell-free DNAs. We compared the performance of different deconvolution algorithms via simulations and benchmark datasets and further investigated the associations of the estimated cell type proportions to cancer therapy in breast cancer and subtypes in melanoma methylation case studies. Our results indicated that the deconvolution based on the extended reference library is critical to obtain accurate estimates of cell proportions in non-blood tissues.

Designing Studies for Epigenetic Biomarker Development in Autoimmune Rheumatic Diseases

In just a few years, the number of epigenetic studies in autoimmune rheumatic and inflammatory diseases has greatly increased. This is in part due to the need of identifying additional determinants to genetics to explain the pathogenesis and development of these disorders. In this regard, epigenetics provides potential mechanisms that determine gene function, are linked to environmental factors, and could explain a wide range of phenotypic variability among patients with these diseases. Despite the high interest and number of studies describing epigenetic alterations under these conditions and exploring their relationship to various clinical aspects, few of the proposed biomarkers have yet reached clinical practice. The potential of epigenetic markers is high, as these alterations link measurable features with a number of biological traits. In the present article, we present published studies in the field, discuss some frequent limitations in the existing research, and propose a number of considerations that should be taken into account by those starting new projects in the field, with an aim to generate biomarkers that could make it into the clinics.

Maternal obesity alters methylation level of cytosine in CpG island for epigenetic inheritance in fetal umbilical cord blood

BACKGROUND

Over the past few decades, global maternal obesity prevalence has rapidly increased. This condition may induce long-lasting pathophysiological effects on either fetal or infant health that could be attributable to unknown unique changes in the umbilical blood composition.

METHODS

A total of 34 overweight/obese and 32 normal-weight pregnant women were recruited. Fifteen umbilical blood samples including 8 overweight/obese subjects and 7 normal weight women were sequenced using Targeted Bisulfite Sequencing technology to detect the average methylation level of cytosine and identify the differentially methylated region (DMR). GO and KEGG analyses were then employed to perform pathway enrichment analysis of DMR-related genes and promoters. Moreover, the mRNA levels of methylation-related genes histone deacetylases (HDACs) and DNA methyltransferases (DNMTs) were characterized in the samples obtained from these two groups.

RESULTS

Average methylated cytosine levels in both the CpG islands (CGI) and promoter significantly decreased in overweight/obese groups. A total of 1669 DMRs exhibited differences in their DNA methylation status between the overweight/obese and control groups. GO and KEGG analyses revealed that DMR-related genes and promoters were enriched in the metabolism, cancer and cardiomyopathy signaling pathways. Furthermore, the HDACs and DNMTs mRNA levels trended to decline in overweight/obese groups.

CONCLUSIONS

Decreased methylated cytosine levels in overweight/obese women induce the gene expression activity at a higher level than in the control group. DMRs between these two groups in the fetal blood may contribute to the changes in gene transcription that underlie the increased risk of metabolic disorders, cancers and cardiomyopathy in their offspring.

Pre-surgery immune profiles of adult glioma patients

INTRODUCTION

Although immunosuppression is a known characteristic of glioma, no previous large studies have reported peripheral blood immune cell profiles prior to patient surgery and chemoradiation. This report describes blood immune cell characteristics and associated variables prior to surgery among typical glioma patients seen at a large University practice.

METHODS

We analyzed pre-surgery blood samples from 139 glioma patients diagnosed with a new or recurrent grade II/III glioma (LrGG, n = 64) or new glioblastoma (GBM, n = 75) and 454 control participants without glioma. Relative cell fractions of CD4, CD8, B-cells, Natural Killer cells, monocytes, and neutrophils, were estimated via a validated deconvolution algorithm from blood DNA methylation measures from Illumina EPIC arrays.

RESULTS

Dexamethasone use at time of blood draw varied by glioma type being highest among patients with IDH wild-type (wt) GBM (75%) and lowest for those with oligodendroglioma (14%). Compared to controls, glioma patients showed statistically significant lower cell fractions for all immune cell subsets except for neutrophils which were higher (all p-values < 0.001), in part because of the higher prevalence of dexamethasone use at time of blood draw for IDHwt GBM. Patients who were taking dexamethasone were more likely to have a low CD4 count (< 200, < 500), increased neutrophils, low absolute lymphocyte counts, higher total cell count and higher NLR.

CONCLUSION

We show that pre-surgery blood immune profiles vary by glioma subtype, age, and more critically, by use of dexamethasone. Our results highlight the importance of considering dexamethasone exposures in all studies of immune profiles and of obtaining immune measures prior to use of dexamethasone, if possible.

Metabolomic predictors of phenotypic traits can replace and complement measured clinical variables in population-scale expression profiling studies

Population-scale expression profiling studies can provide valuable insights into biological and disease-underlying mechanisms. The availability of phenotypic traits is essential for studying clinical effects. Therefore, missing, incomplete, or inaccurate phenotypic information can make analyses challenging and prevent RNA-seq or other omics data to be reused. A possible solution are predictors that infer clinical or behavioral phenotypic traits from molecular data. While such predictors have been developed based on different omics data types and are being applied in various studies, metabolomics-based surrogates are less commonly used than predictors based on DNA methylation profiles.In this study, we inferred 17 traits, including diabetes status and exposure to lipid medication, using previously trained metabolomic predictors. We evaluated whether these metabolomic surrogates can be used as an alternative to reported information for studying the respective phenotypes using expression profiling data of four population cohorts. For the majority of the 17 traits, the metabolomic surrogates performed similarly to the reported phenotypes in terms of effect sizes, number of significant associations, replication rates, and significantly enriched pathways.The application of metabolomics-derived surrogate outcomes opens new possibilities for reuse of multi-omics data sets. In studies where availability of clinical metadata is limited, missing or incomplete information can be complemented by these surrogates, thereby increasing the size of available data sets. Additionally, the availability of such surrogates could be used to correct for potential biological confounding. In the future, it would be interesting to further investigate the use of molecular predictors across different omics types and cohorts.

Tracing the Origin of Cell-Free DNA Molecules through Tissue-Specific Epigenetic Signatures

All cell and tissue types constantly release DNA fragments into human body fluids by various mechanisms including programmed cell death, accidental cell degradation and active extrusion. Particularly, cell-free DNA (cfDNA) in plasma or serum has been utilized for minimally invasive molecular diagnostics. Disease onset or pathological conditions that lead to increased cell death alter the contribution of different tissues to the total pool of cfDNA. Because cfDNA molecules retain cell-type specific epigenetic features, it is possible to infer tissue-of-origin from epigenetic characteristics. Recent research efforts demonstrated that analysis of, e.g., methylation patterns, nucleosome occupancy, and fragmentomics determined the cell- or tissue-of-origin of individual cfDNA molecules. This novel tissue-of origin-analysis enables to estimate the contributions of different tissues to the total cfDNA pool in body fluids and find tissues with increased cell death (pathologic condition), expanding the portfolio of liquid biopsies towards a wide range of pathologies and early diagnosis. In this review, we summarize the currently available tissue-of-origin approaches and point out the next steps towards clinical implementation.

Differential placental CpG methylation is associated with chronic lung disease of prematurity

BACKGROUND

Chronic lung disease (CLD) is the most common pulmonary morbidity in extremely preterm infants. It is unclear to what extent prenatal exposures influence the risk of CLD. Epigenetic variation in placenta DNA methylation may be associated with differential risk of CLD, and these associations may be dependent upon sex.

METHODS

Data were obtained from a multi-center cohort of infants born extremely preterm (<28 weeks' gestation) and an epigenome-wide approach was used to identify associations between placental DNA methylation and CLD (n = 423). Associations were evaluated using robust linear regression adjusting for covariates, with a false discovery rate of 0.05. Analyses stratified by sex were used to assess differences in methylation-CLD associations.

RESULTS

CLD was associated with differential methylation at 49 CpG sites representing 46 genes in the placenta. CLD was associated with differential methylation of probes within genes related to pathways involved in fetal lung development, such as p53 signaling and myo-inositol biosynthesis. Associations between CpG methylation and CLD differed by sex.

CONCLUSIONS

Differential placental methylation within genes with key roles in fetal lung development may reflect complex cell signaling between the placenta and fetus which mediate CLD risk. These pathways appear to be distinct based on fetal sex.

IMPACT

In extremely preterm infants, differential methylation of CpG sites within placental genes involved in pathways related to cell signaling, oxidative stress, and trophoblast invasion is associated with chronic lung disease of prematurity. DNA methylation patterns associated with chronic lung disease were distinctly based on fetal sex, suggesting a potential mechanism underlying dimorphic phenotypes. Mechanisms related to fetal hypoxia and placental myo-inositol signaling may play a role in fetal lung programming and the developmental origins of chronic lung disease. Continued research of the relationship between the placental epigenome and chronic lung disease could inform efforts to ameliorate or prevent this condition.

A pan-tissue DNA methylation atlas enables in silico decomposition of human tissue methylomes at cell-type resolution

Bulk-tissue DNA methylomes represent an average over many different cell types, hampering our understanding of cell-type-specific contributions to disease development. As single-cell methylomics is not scalable to large cohorts of individuals, cost-effective computational solutions are needed, yet current methods are limited to tissues such as blood. Here we leverage the high-resolution nature of tissue-specific single-cell RNA-sequencing datasets to construct a DNA methylation atlas defined for 13 solid tissue types and 40 cell types. We comprehensively validate this atlas in independent bulk and single-nucleus DNA methylation datasets. We demonstrate that it correctly predicts the cell of origin of diverse cancer types and discovers new prognostic associations in olfactory neuroblastoma and stage 2 melanoma. In brain, the atlas predicts a neuronal origin for schizophrenia, with neuron-specific differential DNA methylation enriched for corresponding genome-wide association study risk loci. In summary, the DNA methylation atlas enables the decomposition of 13 different human tissue types at a high cellular resolution, paving the way for an improved interpretation of epigenetic data.

This resource presents an in silico generated DNA methylation atlas that can be used for cell-type deconvolution of human tissues.

Cell-type heterogeneity: Why we should adjust for it in epigenome and biomarker studies

Most studies aiming to identify epigenetic biomarkers do so from complex tissues that are composed of many different cell-types. By definition, these cell-types vary substantially in terms of their epigenetic profiles. This cell-type specific variation among healthy cells is completely independent of the variation associated with disease, yet it dominates the epigenetic variability landscape. While cell-type composition of tissues can change in disease and this may provide accurate and reproducible biomarkers, not adjusting for the underlying cell-type heterogeneity may seriously limit the sensitivity and precision to detect disease-relevant biomarkers or hamper our understanding of such biomarkers. Given that computational and experimental tools for tackling cell-type heterogeneity are available, we here stress that future epigenetic biomarker studies should aim to provide estimates of underlying cell-type fractions for all samples in the study, and to identify biomarkers before and after adjustment for cell-type heterogeneity, in order to obtain a more complete and unbiased picture of the biomarker-landscape. This is critical, not only to improve reproducibility and for the eventual clinical application of such biomarkers, but importantly, to also improve our molecular understanding of disease itself.

Enhanced cell deconvolution of peripheral blood using DNA methylation for high-resolution immune profiling

DNA methylation microarrays can be employed to interrogate cell-type composition in complex tissues. Here, we expand reference-based deconvolution of blood DNA methylation to include 12 leukocyte subtypes (neutrophils, eosinophils, basophils, monocytes, naïve and memory B cells, naïve and memory CD4 + and CD8 + T cells, natural killer, and T regulatory cells). Including derived variables, our method provides 56 immune profile variables. The IDOL (IDentifying Optimal Libraries) algorithm was used to identify libraries for deconvolution of DNA methylation data for current and previous platforms. The accuracy of deconvolution estimates obtained using our enhanced libraries was validated using artificial mixtures and whole-blood DNA methylation with known cellular composition from flow cytometry. We applied our libraries to deconvolve cancer, aging, and autoimmune disease datasets. In conclusion, these libraries enable a detailed representation of immune-cell profiles in blood using only DNA and facilitate a standardized, thorough investigation of immune profiles in human health and disease.

Reliability of a novel approach for reference-based cell type estimation in human placental DNA methylation studies

The placenta is a central organ during early development, influencing trajectories of health and disease. DNA methylation (DNAm) studies of human placenta improve our understanding of how its function relates to disease risk. However, DNAm studies can be biased by cell type heterogeneity, so it is essential to control for this in order to reduce confounding and increase precision. Computational cell type deconvolution approaches have proven to be very useful for this purpose. For human placenta, however, an assessment of the performance of these estimation methods is still lacking. Here, we examine the performance of a newly available reference-based cell type estimation approach and compare it to an often-used reference-free cell type estimation approach, namely RefFreeEWAS, in placental genome-wide DNAm samples taken at birth and from chorionic villus biopsies early in pregnancy using three independent studies comprising over 1000 samples. We found both reference-free and reference-based estimated cell type proportions to have predictive value for DNAm, however, reference-based cell type estimation outperformed reference-free estimation for the majority of data sets. Reference-based cell type estimations mirror previous histological knowledge on changes in cell type proportions through gestation. Further, CpGs whose variation in DNAm was largely explained by reference-based estimated cell type proportions were in the proximity of genes that are highly tissue-specific for placenta. This was not the case for reference-free estimated cell type proportions. We provide a list of these CpGs as a resource to help researchers to interpret results of existing studies and improve future DNAm studies of human placenta.

Immune profiles and DNA methylation alterations related with non-muscle-invasive bladder cancer outcomes

BACKGROUND

Non-muscle-invasive bladder cancer (NMIBC) patients receive frequent monitoring because ≥ 70% will have recurrent disease. However, screening is invasive, expensive, and associated with significant morbidity making bladder cancer the most expensive cancer to treat per capita. There is an urgent need to expand the understanding of markers related to recurrence and survival outcomes of NMIBC.

METHODS AND RESULTS

We used the Illumina HumanMethylationEPIC array to measure peripheral blood DNA methylation profiles of NMIBC patients (N = 603) enrolled in a population-based cohort study in New Hampshire and applied cell type deconvolution to estimate immune cell-type proportions. Using Cox proportional hazard models, we identified that increasing CD4T and CD8T cell proportions were associated with a statistically significant decreased hazard of tumor recurrence or death (CD4T: HR = 0.98, 95% CI = 0.97-1.00; CD8T: HR = 0.97, 95% CI = 0.95-1.00), whereas increasing monocyte proportion and methylation-derived neutrophil-to-lymphocyte ratio (mdNLR) were associated with the increased hazard of tumor recurrence or death (monocyte: HR = 1.04, 95% CI = 1.00-1.07; mdNLR: HR = 1.12, 95% CI = 1.04-1.20). Then, using an epigenome-wide association study (EWAS) approach adjusting for age, sex, smoking status, BCG treatment status, and immune cell profiles, we identified 2528 CpGs associated with the hazard of tumor recurrence or death (P < 0.005). Among these CpGs, the 1572 were associated with an increased hazard and were significantly enriched in open sea regions; the 956 remaining CpGs were associated with a decreased hazard and were significantly enriched in enhancer regions and DNase hypersensitive sites.

CONCLUSIONS

Our results expand on the knowledge of immune profiles and methylation alteration associated with NMIBC outcomes and represent a first step toward the development of DNA methylation-based biomarkers of tumor recurrence.

DNA methylation-derived systemic inflammation indices and their association with oropharyngeal cancer risk and survival

BACKGROUND

Head and neck squamous cell carcinomas are associated with systemic inflammation (SI). We evaluated whether DNA methylation-derived SI (mdSI) indices are associated with oropharyngeal cancer risk and survival.

METHODS

Ninety-four oropharyngeal squamous cell carcinoma (OPSCC) cases and 57 controls with DNA methylation data were included. Logistic regression analysis and survival analysis were performed to test the association of mdSI indices with OPSCC risk and survival.

RESULTS

Higher methylation-derived neutrophil-to-lymphocyte ratio (mdNLR) was associated with increased risk of OPSCC (OR = 1.21, 95%CI: 1.11-1.40) while no association was found with methylation-derived lymphocyte-to-monocyte ratio (mdLMR). For 5-year overall survival, higher mdLMR was significantly associated with decreased risk of death (HR = 0.25, 95%CI: 0.10-0.64) while the converse was observed for mdNLR (HR = 2.48, 95%CI: 1.04-5.92).

CONCLUSION

We observed an association between mdSI indices and OPSCC risk and 5-year overall survival. It is possible to use mdLMR as an independent prognostic factor for OPSCC.

Experimental and Bioinformatic Approaches to Studying DNA Methylation in Cancer

DNA methylation is an essential epigenetic mark. Alterations of normal DNA methylation are a defining feature of cancer. Here, we review experimental and bioinformatic approaches to showcase the breadth and depth of information that this epigenetic mark provides for cancer research. First, we describe classical approaches for interrogating bulk DNA from cell populations as well as more recently developed approaches for single cells and multi-Omics. Second, we focus on the computational analysis from primary data processing to the identification of unique methylation signatures. Additionally, we discuss challenges such as sparse data and cellular heterogeneity.

Using R for Cell-Type Composition Imputation in Epigenome-Wide Association Studies

Adjusting cell type composition is challenging but critical in epigenome-wide association studies (EWAS). In this chapter, we describe how to apply reference-based and reference-free methods in R to impute cell type composition in whole blood samples.

DNA Methylation Profiles of Purified Cell Types in Bronchoalveolar Lavage: Applications for Mixed Cell Paediatric Pulmonary Studies

In epigenome-wide association studies analysing DNA methylation from samples containing multiple cell types, it is essential to adjust the analysis for cell type composition. One well established strategy for achieving this is reference-based cell type deconvolution, which relies on knowledge of the DNA methylation profiles of purified constituent cell types. These are then used to estimate the cell type proportions of each sample, which can then be incorporated to adjust the association analysis. Bronchoalveolar lavage is commonly used to sample the lung in clinical practice and contains a mixture of different cell types that can vary in proportion across samples, affecting the overall methylation profile. A current barrier to the use of bronchoalveolar lavage in DNA methylation-based research is the lack of reference DNA methylation profiles for each of the constituent cell types, thus making reference-based cell composition estimation difficult. Herein, we use bronchoalveolar lavage samples collected from children with cystic fibrosis to define DNA methylation profiles for the four most common and clinically relevant cell types: alveolar macrophages, granulocytes, lymphocytes and alveolar epithelial cells. We then demonstrate the use of these methylation profiles in conjunction with an established reference-based methylation deconvolution method to estimate the cell type composition of two different tissue types; a publicly available dataset derived from artificial blood-based cell mixtures and further bronchoalveolar lavage samples. The reference DNA methylation profiles developed in this work can be used for future reference-based cell type composition estimation of bronchoalveolar lavage. This will facilitate the use of this tissue in studies examining the role of DNA methylation in lung health and disease.

Computational challenges in detection of cancer using cell-free DNA methylation

Cell-free DNA(cfDNA) methylation profiling is considered promising and potentially reliable for liquid biopsy to study progress of diseases and develop reliable and consistent diagnostic and prognostic biomarkers. There are several different mechanisms responsible for the release of cfDNA in blood plasma, and henceforth it can provide information regarding dynamic changes in the human body. Due to the fragmented nature, low concentration of cfDNA, and high background noise, there are several challenges in its analysis for regular use in diagnosis of cancer. Such challenges in the analysis of the methylation profile of cfDNA are further aggravated due to heterogeneity, biomarker sensitivity, platform biases, and batch effects. This review delineates the origin of cfDNA methylation, its profiling, and associated computational problems in analysis for diagnosis. Here we also contemplate upon the multi-marker approach to handle the scenario of cancer heterogeneity and explore the utility of markers for 5hmC based cfDNA methylation pattern. Further, we provide a critical overview of deconvolution and machine learning methods for cfDNA methylation analysis. Our review of current methods reveals the potential for further improvement in analysis strategies for detecting early cancer using cfDNA methylation.

DNA Methylation Patterns in CD8+ T Cells Discern Psoriasis From Psoriatic Arthritis and Correlate With Cutaneous Disease Activity

Background: Psoriasis is a T cell-mediated chronic autoimmune/inflammatory disease. While some patients experience disease limited to the skin (skin psoriasis), others develop joint involvement (psoriatic arthritis; PsA). In the absence of disease- and/or outcome-specific biomarkers, and as arthritis can precede skin manifestations, diagnostic and therapeutic delays are common and contribute to disease burden and damage accrual.

Objective: Altered epigenetic marks, including DNA methylation, contribute to effector T cell phenotypes and altered cytokine expression in autoimmune/inflammatory diseases. This project aimed at the identification of disease-/outcome-specific DNA methylation signatures in CD8⁺ T cells from patients with psoriasis and PsA as compared to healthy controls.

Method: Peripheral blood CD8⁺ T cells from nine healthy controls, 10 psoriasis, and seven PsA patients were collected to analyze DNA methylation marks using Illumina Human Methylation EPIC BeadChips (>850,000 CpGs per sample). Bioinformatic analysis was performed using R (minfi, limma, ChAMP, and DMRcate packages).

Results: DNA methylation profiles in CD8⁺ T cells differentiate healthy controls from psoriasis patients [397 Differentially Methylated Positions (DMPs); 9 Differentially Methylated Regions (DMRs) when ≥CpGs per DMR were considered; 2 DMRs for ≥10 CpGs]. Furthermore, patients with skin psoriasis can be discriminated from PsA patients [1,861 DMPs, 20 DMRs (≥5 CpGs per region), 4 DMRs (≥10 CpGs per region)]. Gene ontology (GO) analyses considering genes with ≥1 DMP in their promoter delivered methylation defects in skin psoriasis and PsA primarily affecting the BMP signaling pathway and endopeptidase regulator activity, respectively. GO analysis of genes associated with DMRs between skin psoriasis and PsA demonstrated an enrichment of GABAergic neuron and cortex neuron development pathways. Treatment with cytokine blockers associated with DNA methylation changes [2,372 DMPs; 1,907 DMPs within promoters, 7 DMRs (≥5 CpG per regions)] affecting transforming growth factor beta receptor and transmembrane receptor protein serine/threonine kinase signaling pathways. Lastly, a methylation score including TNF and IL-17 pathway associated DMPs inverse correlates with skin disease activity scores (PASI).

Conclusion: Patients with skin psoriasis exhibit DNA methylation patterns in CD8⁺ T cells that allow differentiation from PsA patients and healthy individuals, and reflect clinical activity of skin disease. Thus, DNA methylation profiling promises potential as diagnostic and prognostic tool to be used for molecular patient stratification toward individualized treatment.