Jenna C. Carlson’s research while affiliated with University of Pittsburgh and other places

Whole genome sequencing (WGS) studies have identified hundreds of millions of rare variants (RVs) and have enabled RV association tests (RVATs) of these variants with complex traits and diseases. Analysis of non-coding variants is challenged by the considerable variability in regulatory function which candidate Cis-Regulatory Elements (cCREs) exhibit across cell types. We propose cellSTAAR, which integrates WGS data with single-cell ATAC-seq data to capture variability in chromatin accessibility across cell types via the construction of cell-type-specific functional annotations and variant sets. To reflect the uncertainty in cCRE-gene linking, cellSTAAR also links cCREs to their target genes using an omnibus framework which aggregates results from a variety of popular linking approaches. We applied cellSTAAR on Freeze 8 (N = 60,000) of the NHLBI Trans-Omics for Precision Medicine (TOPMed) consortium data to four lipids phenotypes: LDL cholesterol, a binary variable corresponding to high LDL cholesterol, HDL cholesterol, and triglycerides. We also provide replication results for all four phenotypes using UK Biobank (N = 190,000). Evidence from simulation studies and our real data analysis demonstrates that cellSTAAR boosts power and improves interpretation of RVATs of cCREs.

Understanding psychosocial determinants of health is critical in understudied, high-risk populations, where such knowledge could have a major impact on health outcomes. In Samoa, where diabetes prevalence has rapidly increased over 30 years, associations between psychosocial factors and diabetes remain poorly understood. The purpose of our study was to identify psychosocial factors associated with HbA1c among Samoan adults while accounting for established correlates of HbA1c including age, body composition, and genetic and behavioral risk factors. Participants (n = 349 adult Samoans; 53% female) who were not receiving diabetes care were selectively sampled from the Soifua Manuia (2017 – 2019) study. Multiple linear regression with backwards elimination and a bootstrapping stability investigation was used to assess associations between HbA1c and participant demographics, genetics (rs373863828 A allele in CREBRF), biological, behavioral, and psychosocial variables (i.e., self-esteem, self-efficacy, social support, health related quality of life, perceived social conflict, food security). The rs373863828 genotype and age were statistically significantly associated with HbA1c—more copies of the rs373863828 A allele and younger age were associated with lower HbA1c. While not statistically significant based on confidence intervals, those with greater self-efficacy tended to have lower HbA1c, while those with higher social support tended to have higher HbA1c. Based on the frequent appearances of psychosocial factors in our models, we advocate for increased awareness and promotion of psychosocial health, and health promoting social support in Samoa. Integrating psychosocial and mental health awareness with standard biomedical protocols and health promotion materials provides an important and empowering means of diabetes prevention and management.

Orofacial clefts (OFCs) are the most common craniofacial birth defects, affecting 1 in 700 births, and have a strong genetic basis with a high recurrence risk within families. While many of the previous studies have associated common, noncoding genetic loci with OFCs, previous studies on de novo variants (DNVs) in OFC cases have focused on coding variants that could have a functional impact on protein structure, and the contribution of noncoding DNVs to the formation of OFCs has largely been ignored and is not well understood. Therefore, we reanalyzed an existing dataset of DNVs from 1,409 trios with OFCs that had undergone targeted sequencing of known OFC-associated loci. We then annotated these DNVs with information from datasets of predicted epigenetic function during human craniofacial development. Of the 66 DNVs in this cohort, 17 (25.7%) were within a predicted enhancer or promotor region. Two DNVs fell within the same enhancer region (hs1617), which is more than expected by chance (p = 0.0017). The sequence changes caused by these DNVs are predicted to create binding sites not seen in the reference sequence for transcription factors PAX6 and ZBTB7A and to disrupt binding sites for STAT1 and STAT3. This enhancer region is within the same topologically-associated domain as HHAT , SERTAD4 , and IRF6 , all of which are involved in craniofacial development. All three genes are highly expressed in human neural crest cells. Knockout mice for Hhat and Irf6 have abnormal embryonic development including a cleft palate, and variants in and around IRF6 are associated with nonsyndromic and syndromic forms of OFCs in humans. Taken together, this suggests that noncoding DNVs contribute to the genetic architecture of OFCs, with a burden of DNVs in OFC trios in enhancer regions near known OFC-associated genes. Overall, this adds to our understanding of the genetic mechanisms that underly OFC formation.

Large-scale whole-genome sequencing (WGS) studies have improved our understanding of the contributions of coding and noncoding rare variants to complex human traits. Leveraging association effect sizes across multiple traits in WGS rare variant association analysis can improve statistical power over single-trait analysis, and also detect pleiotropic genes and regions. Existing multi-trait methods have limited ability to perform rare variant analysis of large-scale WGS data. We propose MultiSTAAR, a statistical framework and computationally scalable analytical pipeline for functionally informed multi-trait rare variant analysis in large-scale WGS studies. MultiSTAAR accounts for relatedness, population structure and correlation among phenotypes by jointly analyzing multiple traits, and further empowers rare variant association analysis by incorporating multiple functional annotations. We applied MultiSTAAR to jointly analyze three lipid traits in 61,838 multi-ethnic samples from the Trans-Omics for Precision Medicine (TOPMed) Program. We discovered and replicated new associations with lipid traits missed by single-trait analysis.

Nonsyndromic orofacial clefts (OFCs) are common, heritable birth defects caused by both genetic and environmental risk factors. Despite the identification of many genetic loci harboring OFC-risk variants, there are many unknown genetic determinants of OFC. Furthermore, while the process of embryonic facial development is well characterized, the molecular mechanisms that underly it are not. This represents a major hurdle in understanding how disruptions in these biological processes result in OFC. Thus, we sought to identify novel OFC-risk loci through a genome-wide multi-ancestry study of five nested OFC phenotypes (isolated cleft lip [CLO], isolated cleft palate [CPO], cleft lip and palate [CLP], cleft lip with/without cleft palate [CL/P], and any cleft [ANY]) representing distinct cleft subtypes to identify subtype-specific signals and grouped types to maximize power to detect shared genetic effects. We performed genome-wide meta-analyses of these five OFC phenotypes from three cohorts totaling >14,000 individuals using METAL. In addition to replicating 13 known OFC-risk loci, we observed novel association in three regions: the 1p36.32 locus (lead variant rs584402, an intergenic variant, p CLO = 3.14e-8), the 7q33 locus (lead variant rs17168118, an intronic variant in CALD1 , p CLP = 9.17e-9), and the 16p13.3 locus (lead variant rs77075754, an intronic variant in RBFOX1 , p CL/P = 1.53e-9, p ANY = 1.93e-9). We also observed a novel association within the known risk locus 8q22.1 that was independent of the previously reported signal (lead variant rs4735314, an intronic variant in ESRP1 , p CLP = 1.07e-9, p CL/P = 3.88e-8). Next, we performed multi-tissue TWAS with s-MulTiXcan and identified four overlapping genes with significant genetically predicted transcription associated with OFC risk. These genes also overlapped the genome-wide significant association signals from the meta-analysis, including CALD1 and ESRP1 and known OFC-risk genes TANC2 and NTN1 . Each of the newly reported loci has potential regulatory effects, including evidence of craniofacial enhancer activity, that offer new clues as to the molecule mechanisms underlying embryonic facial development. Author Summary Orofacial clefts, including cleft lip and cleft palate, are common birth defects that can be caused by both genetic and environmental factors. While many of these factors are known, there are still significant gaps in our understanding of how and why clefts arise. To help address this deficiency, we measured association between variants across the genome and clefting in cases and controls from diverse genetic ancestries. We identified three new candidate genes ( CALD1 , ESRP1 , and RBFOX1 ) that may be involved in cleft risk and reaffirmed the role of 12 previously reported risk genes. We also found evidence that clefting was associated with predicted gene expression at CALD1 and ESRP1 and that the associated variants in these genes were located near regions known to be involved in the regulation of gene expression in craniofacial tissues during development.

Maternal exposures are known to influence the risk of isolated cleft lip with or without cleft palate (CL/P) – a common and highly heritable birth defect with a multifactorial etiology. To identify new CL/P risk loci, we conducted a genome-wide gene-environment interaction (GEI) analysis of CL/P on a sample of 540 cases and 260 controls recruited from the Philippines, incorporating the interaction effects of genetic variants with maternal smoking and vitamin use. As GEI analyses are typically low in power and the results can be difficult to interpret, we used multiple testing frameworks to evaluate potential GEI effects: 1 degree-of-freedom (1df) GxE test, the 3df joint test, and the two-step EDGE approach. While we did not detect any genome-wide significant interactions, we detected 12 suggestive GEI with smoking and 25 suggestive GEI with vitamin use between all testing frameworks. Several of these loci showed biological plausibility. Notable interactions with smoking include loci near FEZF1 , TWIST2, and NET1. While FEZF1 is involved in early neuronal development, TWIST2 and NET1 regulate epithelial-mesenchymal transition which is required for proper lip and palate fusion. Interactions with vitamins encompass CECR2 — a chromatin remodeling protein required for neural tube closure—and FURIN, a critical protease during early embryogenesis that activates various growth factor and extracellular-matrix protein. The activity of both proteins is influenced by folic acid. Our findings highlight the critical role of maternal exposures in identifying genes associated with structural birth defects such as CL/P and provide new paths to explore for CL/P genetics.

Study question: Can a genome–wide association study (GWAS) and transcriptome–wide association study (TWAS) help identify genetic variation or genes associated with circulating anti–Müllerian hormone (AMH) levels in Samoan women? Summary answer: We identified eleven genome–wide suggestive loci (strongest association signal in ARID3A 19–946163–G–C [p = 2.32 × 10⁻⁷]) and seven transcriptome–wide significant genes (GINS2, SENP3, USP7, TUSC3, MAFA, METTL4, NDFIP1 [all with a p < 2.50 × 10⁻⁶]) associated with circulating AMH levels in Samoan women. What is known already: Three prior GWASs of AMH levels identified eight loci in premenopausal women of European ancestry (AMH, MCM8, TEX41, CHECK2, CDCA7, EIF4EBP1, BMP4 and an uncharacterized non–coding RNA gene CTB–99A3.1), among which the MCM8 locus was shared among all three studies. Study design, size, duration: We included a sample of 1,185 women from two independently recruited samples: a family study (n = 212; age: 18 to 40 years) recruited in 2002–03 from Samoa and American Samoa; and the Soifua Manuia Study (n = 973; age: 25 to 51 years), a cross–sectional population–based study recruited in 2010 from Samoa. Participants/materials, setting, methods: Serum AMH levels were measured using enzyme linked immunosorbent assays (ELISA). We performed GWASs in the two participant samples using a Cox mixed–effects model to account for AMH levels below detectable limits and adjusted for centered age, centered age2, polity, and kinship via kinship matrix. The summary statistics were then meta–analyzed using a fixed–effect model. We annotated the variants with p < 1 × 10⁻⁵ and calculated posterior probability of causality for prioritization. We further annotated variants using FUMA and performed colocalization and transcriptome–wide association analysis. We also assessed whether any previously reported loci were replicated in our GWAS. Main results and the role of chance: We identified eleven novel genome–wide suggestive loci (p < 1 × 10⁻⁵) associated with AMH levels and replicated EIF4EBP1, a previously reported AMH locus, in the GWAS. The lead variant in ARID3A, 19–946163–G–C is in high linkage disequilibrium (r2 = 0.79) with the known age–at–menopause variant 19–950694–G–A. Nearby KISS1R is a bio-logically plausibility causal gene in the region; kisspeptin regulates ovarian follicle development and has been linked to AMH levels. Further investigation of the ARID3A locus is warranted. Limitations, reasons for caution: The main limitations of our study are the small sample size for a GWAS and the use of the transcription model trained on mostly European samples from the Genotype Tissue Expression (GTEx) project, which may have led to reduced power to detect genotype–expression associations. Our findings need to be validated in larger Polynesian cohorts. Wider implications of the findings: In addition to replicating one of the eight previously discovered AMH loci, we identified new suggestive associations. It is known that the inclusion of founder populations aids in the discovery of novel loci. These findings could enhance our understanding of AMH and AMH–related reproductive phenotypes (ovarian reserve, age at menopause, premature ovarian failure, and polycystic ovary syndrome) and help build a screening approach for women at risk for these phenotypes using genetically predicted AMH levels. Study funding/competing interest(s): This work was funded by NIH grants R01–HL093093 (PI: S.T.M.), R01–HL133040 (PI: R.L.M.), and T90–DE030853 (PI: C.S. Sfeir). Molecular data for the Trans–Omics in Precision Medicine (TOPMed) Program was supported by the National Heart, Lung and Blood Institute (NHLBI). The content is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health.

Sleep apnea is a global public health concern, but little research has examined this issue in low- and middle-income countries, including Samoa. The purpose of this study was to examine the sample prevalence and characteristics of sleep apnea using a validated home sleep apnea device (WatchPAT, Itamar) and explore factors that may influence sleep health in the Samoan setting. This study used data collected through the Soifua Manuia (“Good Health”) study, which investigated the impact of the body mass index (BMI)-associated genetic variant rs373863828 in CREBRF on metabolic traits in Samoan adults (sampled to overrepresent the obesity-risk allele of interest). A total of 330 participants had sleep data available. Participants (53.3 % female) had a mean (SD) age of 52.0 (9.9) years and BMI of 35.5 (7.5) kg/m², and 36.3 % of the sample had type 2 diabetes. Based on the 3 % and 4 % apnea hypopnea indices (AHI) and the 4 % oxygen desaturation index (ODI), descriptive analyses revealed moderate to severe sleep apnea (defined as ≥15 events/hr) in 54.9 %, 31.5 %, and 34.5 % of the sample, respectively. Sleep apnea was more severe in men (e.g., AHI 3 % ≥15 in 61.7 % of men and 48.9 % of women). Correction for non-representational sampling related to the CREBRF obesity-risk allele resulted in only slightly lower estimates. Multiple linear regression linked a higher number of events/hr to higher age, male sex, higher BMI, higher abdominal-hip circumference ratio, and geographic region of residence. Further research and an increased focus on equitable and affordable diagnosis and access to treatment are crucial to addressing sleep apnea in Samoa and globally.

Uncontrolled neuroinflammation mediates traumatic brain injury (TBI) pathology and impairs recovery. Interleukin-6 (IL-6), a pleiotropic inflammatory regulator, is associated with poor clinical TBI outcomes. IL-6 operates via classical-signaling through membrane-bound IL-6 receptor (IL-6R) and trans-signaling through soluble IL-6 receptor (s)IL-6R. IL-6 trans-signaling specifically contributes to neuropathology, making it a potential precision therapeutic TBI target. Soluble glycoprotein 130 (sgp130) prevents IL-6 trans-signaling, sparing classical signaling, thus is a possible treatment. Mice received either controlled cortical impact (CCI) (6.0 ± 0.2 m/s; 2 mm; 50-60ms) or sham procedures. Vehicle (VEH) or sgp130-Fc was subcutaneously administered to sham (VEH or 1 µg) and CCI (VEH, 0.25 µg or 1 µg) mice on days 1, 4, 7, 10 and 13 post-surgery to assess effects on cognition [Morris Water Maze (MWM)] and ipsilateral hemisphere IL-6 related biomarkers (day 21 post-surgery). CCI + sgp130-Fc groups (0.25 µg and 1 µg) were combined for analysis given similar behavior/biomarker outcomes. CCI + VEH mice had longer latencies and path lengths to the platform and increased peripheral zone time versus Sham + VEH and Sham + sgp130-Fc mice, suggesting injury-induced impairments in learning and anxiety. CCI + sgp130-Fc mice had shorter platform latencies and path lengths and had decreased peripheral zone time, indicating a therapeutic benefit of sgp130-Fc after injury on learning and anxiety. Interestingly, Sham + sgp130-Fc mice had shorter platform latencies, path lengths and peripheral zone times than Sham + VEH mice, suggesting a beneficial effect of sgp130-Fc, independent of injury. CCI + VEH mice had increased brain IL-6 and decreased sgp130 levels versus Sham + VEH and Sham + sgp130-Fc mice. There was no treatment effect on IL-6, sIL6-R or sgp130 in Sham + VEH versus Sham + sgp130-Fc mice. There was also no treatment effect on IL-6 in CCI + VEH versus CCI + sgp130-Fc mice. However, CCI + sgp130-Fc mice had increased sIL-6R and sgp130 versus CCI + VEH mice, demonstrating sgp130-Fc treatment effects on brain biomarkers. Inflammatory chemokines (MIP-1β, IP-10, MIG) were increased in CCI + VEH mice versus Sham + VEH and Sham + sgp130-Fc mice. However, CCI + sgp130-Fc mice had decreased chemokine levels versus CCI + VEH mice. IL-6 positively correlated, while sgp130 negatively correlated, with chemokine levels. Overall, we found that systemic sgp130-Fc treatment after CCI improved learning, decreased anxiety and reduced CCI-induced brain chemokines. Future studies will explore sex-specific dosing and treatment mechanisms for sgp130-Fc therapy.

Background The A allele of rs373863828 in CREB3 regulatory factor is associated with high Body Mass Index, but lower odds of type 2 diabetes. These associations have been replicated elsewhere, but to date all studies have been cross-sectional. Our aims were (1) to describe the development of type 2 diabetes and change in fasting glucose between 2010 and 2018 among a longitudinal cohort of adult Samoans without type 2 diabetes or who were not using diabetes medications at baseline, and (2) to examine associations between fasting glucose rate-of-change (mmol/L per year) and the A allele of rs373863828. Methods We describe and test differences in fasting glucose, the development of type 2 diabetes, body mass index, age, smoking status, physical activity, urbanicity of residence, and household asset scores between 2010 and 2018 among a cohort of n = 401 adult Samoans, selected to have a ~2:2:1 ratio of GG:AG: AA rs373863828 genotypes. Multivariate linear regression was used to test whether fasting glucose rate-of-change was associated with rs373863828 genotype, and other baseline variables. Results By 2018, fasting glucose and BMI significantly increased among all genotype groups, and a substantial portion of the sample developed type 2 diabetes mellitus. The A allele was associated with a lower fasting glucose rate-of-change (β = −0.05 mmol/L/year per allele, p = 0.058 among women; β = −0.004 mmol/L/year per allele, p = 0.863 among men), after accounting for baseline variables. Mean fasting glucose and mean BMI increased over an eight-year period and a substantial number of individuals developed type 2 diabetes by 2018. However, fasting glucose rate-of-change, and type 2 diabetes development was lower among females with AG and AA genotypes. Conclusions Further research is needed to understand the effect of the A allele on fasting glucose and type 2 diabetes development. Based on our observations that other risk factors increased over time, we advocate for the continued promotion for diabetes prevention and treatment programming, and the reduction of modifiable risk factors, in this setting.