Whole-Genome Sequencing Identifies Novel Functional Loci Associated with Lung Function in Puerto Rican Youth

Abstract

Rationale: Puerto Ricans have the highest childhood asthma prevalence in the United States (23.6%); however, the etiology is uncertain.

Objectives: In this study, we sought to uncover the genetic architecture of lung function in Puerto Rican youth with and without asthma who were recruited from the island (n = 836).

Methods: We used admixture-mapping and whole-genome sequencing data to discover genomic regions associated with lung function. Functional roles of the prioritized candidate SNPs were examined with chromatin immunoprecipitation sequencing, RNA sequencing, and expression quantitative trait loci data.

Measurements and Main Results: We discovered a genomic region at 1q32 that was significantly associated with a 0.12-L decrease in the lung volume of exhaled air (95% confidence interval, −0.17 to −0.07; P = 6.62 × 10⁻⁸) with each allele of African ancestry. Within this region, two SNPs were expression quantitative trait loci of TMEM9 in nasal airway epithelial cells and MROH3P in esophagus mucosa. The minor alleles of these SNPs were associated with significantly decreased lung function and decreased TMEM9 gene expression. Another admixture-mapping peak was observed on chromosome 5q35.1, indicating that each Native American ancestry allele was associated with a 0.15-L increase in lung function (95% confidence interval, 0.08–0.21; P = 5.03 × 10⁻⁶). The region-based association tests identified four suggestive windows that harbored candidate rare variants associated with lung function.

Conclusions: We identified common and rare genetic variants that may play a critical role in lung function among Puerto Rican youth. We independently validated an inflammatory pathway that could potentially be used to develop more targeted treatments and interventions for patients with asthma.

At a Glance Commentary

Scientific Knowledge on the Subject

In the United States, asthma prevalence and severity vary significantly by race/ethnicity. Puerto Ricans have the highest childhood asthma prevalence (23.6%); however, it is not clear what causes Puerto Ricans to have such high occurrence. The structure of modern Hispanic/Latino populations is a mosaic of Native American, European, and African ancestral influences. A number of studies have reported the influence of genetic ancestry on the development of complex diseases. African genetic ancestry has been linked to decreased lung function, decreased drug response, and increased risk for the development of asthma in Latino populations.

What This Study Adds to the Field

We identified genomic regions that harbored candidate common and rare variants that could influence lung function among Puerto Rican youth with and without asthma using admixture-mapping, whole-genome sequencing, and functional genomics data. The findings of this study (i.e., the IL-6 and IL-1β inflammatory pathways mediated by TMEM9 gene expression) could potentially be used to treat Puerto Rican youth, a population that carries the majority of the asthma disease burden, and possibly benefit patients with asthma of other ethnicities, too. The findings of this manuscript therefore have important clinical and public health implications.

Asthma is an obstructive inflammatory lung disease that adversely influences the daily lives of 26 million Americans and costs $81 billion in annual health care (1). In the United States, asthma prevalence and severity vary significantly by race/ethnicity. Puerto Ricans have the highest childhood asthma prevalence (23.6%), followed by African Americans (18.1%), Mexican Americans (11.5%), and European Americans (9.5%) (2). These striking racial/ethnic disparities in asthma morbidity arise from a complex interplay of genetic and environmental risk factors (3–6). We previously demonstrated that the proportion of African genetic ancestry explained a substantial amount of the variance in baseline lung function in African Americans and Hispanics/Latinos, even after adjusting for environmental and social risk factors (7, 8).

Numerous studies have reported the influence of genetic ancestry on the development of complex diseases, including breast and prostate cancer, diabetes, cardiovascular disease, and asthma (9–15). Ancestry may reflect demographic history, unmeasured risk factors present in ancestral populations, and natural genetic variation, which can all be informative for understanding disease patterns within and across racial/ethnic groups. In addition, genetic risk alleles vary in frequency by race/ethnicity (16, 17). Therefore, among genetically admixed individuals, varying proportions of genetic ancestries may track with and therefore be a proxy of genetic risk alleles (7–15).

Modern Hispanic/Latino populations are a mosaic of Native American, European, and African ancestries. African genetic ancestry has been linked to decreased lung function, decreased drug response, and increased risk for the development of asthma in Latino populations (7, 8, 18). In contrast, increased Native American ancestry was associated with increased lung function in Latino children (19). It is possible that the large discrepancy in asthma prevalence and morbidity between Puerto Ricans and Mexican Americans may be partially explained by variation in genetic ancestry proportions. Previous studies have attempted to uncover genetic determinants of asthma in Puerto Ricans (20–24). However, it remains unclear why Puerto Ricans suffer such high asthma prevalence, morbidity, and mortality.

In this study, we used whole-genome sequencing (WGS) data to perform a genome-wide admixture-mapping analysis of lung function in asthma cases and controls from a Puerto Rican youth population from the island of Puerto Rico. Admixture mapping is a complementary method to traditional genome-wide allelic association tests. Typically, admixture linkage disequilibrium (LD) blocks are longer than haplotype blocks and therefore may have a higher chance of capturing disease-associated genes and reducing the multiple testing burden (25, 26). Admixture mapping identified multiple potential causal variants influencing lung function within two genomic regions. We used nasal epithelial cells from Puerto Rican youth and other available lung tissues to validate the functional roles of these variants and to explore possible biological pathways and mechanisms through which they could influence asthma prevalence and mortality. These results carry potentially important clinical implications for effective management of asthma in Puerto Rican youth.

Methods

Study Subjects

GALA II (Genes-Environment and Admixture in Latino Americans) is an ongoing case–control study of asthma in youth in the mainland United States and Puerto Rico (2006–present). Subjects were eligible to participate if they were 8–21 years of age and identified all four grandparents as Latino. In this study, only the subjects recruited from the island of Puerto Rico who had both lung function and genotype data were included in the main analysis (n = 836). Please see the online data supplement (Study Subjects) for study participants’ eligibility criteria.

Local institutional review boards approved the studies. All subjects and their legal guardians provided written informed assent/consent.

Pulmonary Function Testing

Pulmonary function testing was performed according to the American Thoracic Society standards with a KoKo PFT Spirometer (nSpire Health Inc.) to measure baseline lung function, particularly FEV₁. All spirometric measures were evaluated by a pulmonologist in Puerto Rico and at the University of California, San Francisco. Lung function measurements were log-transformed to make the distribution less skewed (Figure E1 in the online supplement). Analyses were performed on log-transformed FEV₁, which was treated as a continuous variable.

Local and Global Ancestry Estimation

More information about local and global ancestry estimations can be found in the supplement (Local and Global Ancestry Estimation).

Whole-Genome Sequencing

Genotypes were obtained from whole-genome sequences obtained as part of the TOPMed (Trans-Omics for Precision Medicine) sequencing program. Details regarding DNA extraction, WGS data generation, processing, and quality control steps can be found in the supplement (Whole Genome Sequencing Data Generation, Processing and Quality Control).

Admixture Mapping

Estimates of locus-specific ancestry were used to conduct admixture mapping. Each SNP was encoded with the number of alleles (0, 1, or 2 alleles) from a given ancestry. Additive linear regression models were used to test associations between local ancestry and lung function (measured FEV₁, liters) adjusting for age (years), sex (0 = male; 1 = female), asthma status (0 = control; 1 = case), Native American and African genome-wide genetic ancestry proportions (%), and height (cm). We adopted a similar approach as Shriner and colleagues and our previous studies to determine the effective number of tests (n = 595) by fitting an autoregression model to the summary statistics generated from the admixture-mapping analysis using the coda package in R (8, 12, 27). A Bonferroni corrected threshold of genome-wide significance at alpha level was then defined as α = 0.05/595 = 8.4 × 10⁻⁵ and applied to correct for multiple testing and reduce false positive. More information can be found in the supplement (Admixture Mapping).

Fine Mapping: Conditional Analysis

Details of the fine-mapping conditional analysis can be found in the supplement (Fine Mapping: Conditional Analysis).

Region-based Association Analysis

Please see the supplement (Region-based association analysis) regarding region-based association analyses and drop-one variant analyses using SKAT-O (28).

H3K27ac ChIP-seq Assays in HBECs and Bronchial Smooth Muscle Cells

A detailed description of chromatin immunoprecipitation sequencing (ChIP-seq) analyses can be found in the supplement (Human bronchial epithelial cell culture and H3K27ac ChIP-seq assay).

Functional Annotation and Validation of Genetic Variants

Genomic annotation was conducted for the lead ancestry SNP, the SNPs identified from fine-mapping analyses, and all SNPs in linkage disequilibrium with them (r² > 0.8). More information can be found in the supplement (Functional Annotation and Validation of Genetic Variants).

RNA Sequencing and eQTL Analysis

Generation of gene expression and expression quantitative trait loci (eQTL) data and expression analyses are summarized in the supplement [RNA sequencing and Expression Quantitative Trait Loci (eQTL) analysis].

All statistical analyses were conducted in R Studio (version 1.2.5019) and PLINK 1.9 (29, 30).

Results

Study Participants

Baseline characteristics of study participants, by asthma status, are displayed in Table 1. In the main analyses, there were a total of 836 study participants (720 cases with asthma and 116 controls without asthma). On average, patients with asthma were slightly younger than the controls (12.4 ± 3.3 and 13.1 ± 1.7 yr, respectively; P = 0.01). The mean lung function between asthma cases (2.3 ± 0.8 L) and controls (3.0 ± 0.7 L) were significantly different (P < 2.2 × 10⁻¹⁶). Subjects without asthma were significantly taller than the participants with asthma (157.5 ± 10.4 and 150.1 ± 13.6 cm, respectively; P = 3.2 × 10⁻⁸).

Table 1.

Descriptive Statistics for Puerto Rican Study Participants Recruited from the Island (n = 836)

	Asthma Cases	Asthma Controls	P Value
No. of participants	720	116
Age, mean ± SD, yr	12.4 ± 3.3	13.1 ± 1.7	0.01*
Sex, M, n (%)	408 (56.7)	63 (54.3)	0.9†
FEV1, mean ± SD, L	2.3 ± 0.8	3.0 ± 0.7	2.2 × 10−16*
Height, mean ± SD, cm	150.1 ± 13.6	157.5 ± 10.4	3.2 × 10−8*
Ancestry, mean ± SD, %
African	22.9 ± 11.9	23.6 ± 8.1	0.5*
Native American	10.5 ± 3.0	10.7 ± 2.8	0.4*
European	66.5 ± 11.4	65.6 ± 8.0	0.4*

^*Unpaired two-sample t test was conducted for continuous variables.

^†Two-proportion z-test was performed for categorical variables.

Admixture Mapping

A genome-wide single ancestry admixture mapping was conducted for each of the three reference ancestries (YRI, African; CEU, European; NAM, Native American). We discovered two genome-wide significant peaks for African and Native American ancestries after adjusting for covariates (Figure 1). The association test between local genetic ancestry and FEV₁ identified a strong admixture signal on chromosomal region 1q32, indicating that each African ancestry allele was associated with a 0.12-L decrease in the volume of exhaled air (95% confidence interval, −0.17 to −0.07; P = 6.62 × 10⁻⁸) after transforming back to normal scale. Admixture mapping for Native American ancestry located a genomic region at 5q35.1 (Figure 1) that was significantly associated with a 0.15-L increase in lung function (95% confidence interval, 0.08–0.21; P = 5.03 × 10⁻⁶). The lead Native American ancestry SNP, rs12153426, is an intron variant within SLIT3. The lead African ancestry SNP (rs17696752) sits in a noncoding intergenic region (Figure 2).

Figure 1.

Admixture-mapping results. Genome-wide single ancestry admixture mapping for each reference ancestry (YRI, CEU, and NAM) with log-transformed FEV₁ adjusting for covariates including age, sex, height, asthma status, and African and Native American ancestries is shown.

Figure 2.

Admixture-mapping peaks were discovered at 1q32 and 5q35.1. Locus zoom displaying the lead locus-specific ancestry signals (rs17696752 and rs12144397) and nearby genes is shown. chr = chromosome.

Fine Mapping: Conditional Analysis

Allelic association tests of lung function with WGS data were performed at the two genomic regions identified from admixture mapping. Conditional analyses on the association between lung function and locus-specific ancestry revealed that no single SNP was accountable for the observed admixture-mapping signals. Therefore, the joint effects of multiple SNPs on the locus-specific ancestry and lung function association were further investigated.

Stepwise linear regression models identified four SNPs mapped to African ancestry and seven SNPs mapped to Native American ancestry within the genomic regions harboring admixture signals (Table 2). The effect estimates shown in Table 2 represent the associations between the lead ancestry SNP and lung function, adjusting for additional SNPs and covariates. The individual allelic association tests between these SNPs and lung function are displayed in Table E1.

Table 2.

Fine Mapping of Genomic Regions Harboring Admixture Signal Using Sequenced Data

	β*	P Value
African ancestry
rs17696752	−0.02	6.62 × 10−8
rs17696752 + rs16847664	−0.02	5.29 × 10−6
rs17696752 + rs16847664 + rs6679485	−0.02	4.33 × 10−4
rs17696752 + rs16847664 + rs6679485 + rs116270262	−0.01	7.62 × 10−3
rs17696752 + rs16847664 + rs6679485 + rs116270262 + rs10920079	−0.01	0.09
Native American ancestry
rs12153426	0.03	5.04 × 10−6
rs12153426 + rs187841563	0.03	5.52 × 10−6
rs12153426 + rs187841563 + rs11134502	0.03	5.12 × 10−6
rs12153426 + rs187841563 + rs11134502 + rs73320187	0.02	2.96 × 10−3
rs12153426 + rs187841563 + rs11134502 + rs73320187 + rs4323254	0.02	0.03
rs12153426 + rs187841563 + rs11134502 + rs73320187 + rs4323254 + rs74677211	0.01	0.03
rs12153426 + rs187841563 + rs11134502 + rs73320187 + rs4323254 + rs74677211 + rs62384405	0.01	0.03
rs12153426 + rs187841563 + rs11134502 + rs73320187 + rs4323254 + rs74677211 + rs62384405 + rs12522512	0.01	0.21

Joint effects of multiple SNPs on lung function were evaluated using stepwise regression models. Conditional analyses on the locus-specific ancestry identified potential causal SNPs on chromosome 1 (1q32) and 5 (5q35.1).

^*β values represent the association between lead ancestry SNP (rs17696752 at 1q32 and rs12153426 at 5q35.1) and log-transformed FEV₁ adjusting for covariates and additional SNPs in forward stepwise regression analyses. All regression models were adjusted for age, sex, height, asthma status, and African and Native American ancestries.

Fine Mapping: Region-based Association Analysis

The main results of the region-based association tests are presented in Table 3. On chromosome 1, we did not find any significant window after correcting for multiple testing (P < 2.28 × 10⁻⁵). However, two windows were suggestively associated with lung function (P < 4.60 × 10⁻⁴). The P values of these two suggestive windows represent lung function associations driven solely by rare variants. The region-based analysis on chromosome 5 identified two windows that were suggestively associated with lung function (P < 5.30 × 10⁻⁴).

Table 3.

Region-based Association Tests within Each Fine-Mapping Region

Chr	Start Position	End Position	P Value	No. of Rare Markers	Nearest Gene(s)	Window ID
1	196762063	196763063	2.97 × 10−5*	1	CFH, CHFR3	1
1	199359563	199360536	6.76 × 10−5*	4	LINC02789, LOC400600	2
5	169911224	169912224	1.01 × 10−4†	6	DOCK2, INSYN2B	3
5	167726224	167727224	2.44 × 10−4†	6	TENM2	4

Definition of abbreviation: Chr = chromosome.

Genetic variants (minor allele frequency < 0.01) were aggregated in 1-kb sliding windows to examine the association with lung function adjusting for covariates using SKAT-O. Suggestive windows were reported within 1q32 and 5q35.1 regions.

^*Chromosome 1 (after multiple test correction): adjusted P value = 2.28 × 10⁻⁵, adjusted suggestive P value = 4.6 × 10⁻⁴.

^†Chromosome 5 (after multiple test correction): adjusted P value = 2.64 × 10⁻⁵, adjusted suggestive P value = 5.3 × 10⁻⁴.

For each suggestive window, we performed drop-one variant analysis to investigate the effects of each rare variant on lung function. When the rare variant rs1000558853 (minor allele frequency = 0.0006) was removed from window 2, the P value increased by nearly four orders of magnitude (P = 0.41), suggesting that most of the association with lung function for window 2 was contributed by rs1000558853 (Figure E2 and Table E2). The rare variants rs943656663 and rs539192870 also resulted in larger P values when they were excluded from windows 3 and 4, respectively. rs943656663 is an intron variant of TENM2, whereas rs539192870 falls within an intronic region of the genes DOCK2 and INSYN2B. The minor allele of the rare variant rs539192870 was found exclusively in Puerto Ricans (Figure E3).

Bioinformatics Functional Annotation for SNP Prioritization

The candidate SNPs identified from fine mapping, along with other SNPs in high LD with them (r² > 0.80), were further investigated to understand their functional roles. Information from public data sources and H3K27ac ChIP-seq experiments indicated that four SNPs discovered by the African ancestry signal had regulatory functions in white blood cell and lung-related cell lines (Figure 3). We found that rs2400382 overlapped with a CEBPB (CCAAT/enhancer-binding protein β) transcription factor binding site in human lung fibroblast cells (IMR90) and H3K4me3 ChIP-seq and DNase I hypersensitivity signals (cCRE accession ID:EH38E1408465). The intron variants of MROH3P, rs10920079 and rs10920072, intersected with H3K27ac peaks in human bronchial epithelial cells (HBECs). The synonymous variant of CACNA1S, rs16847664, was also enriched with epigenetic markers in HBECs and may interact with TMEM9 based on high-throughput chromosome conformation capture data in IMR90 cells.

Figure 3.

Functional annotation of four variants identified from fine-mapping analysis at 1q32. Each color represents SNPs identified from fine-mapping conditional analysis and their linkage disequilibrium SNPs (R² > 0.8) with at least one or more than one regulatory element data. ^¶RegulomeDB (RegDB) score represents the likelihood that a variant will be located within a functional region. The RegDB variant classification scheme is defined as follows (50): Likely to affect binding: 2b, TF binding + matched TF motif + matched DNase footprint + DNase peak. Minimal binding evidence: 4, TF binding + DNase peak; 5, TF binding or DNase peak; 6, motif hit. ^‡Overlapping with H3K27ac chromatin immunoprecipitation sequencing peaks in human bronchial epithelial cells. ^†High signals of epigenetic markers in immune cells and lung-related tissues (Encyclopedia of DNA Elements data). BP = base pair position; CADD = Combined Annotation-Dependent Depletion; chr = chromosome; Hi-C = high-throughput chromosome conformation capture; TF = transcription factor.

Within the genomic region identified by admixture mapping with Native American ancestry, we identified 19 SNPs located in candidate regulatory regions (Figure 4). One of the candidate causal SNPs selected through fine-mapping analysis, rs6555875, was an intron variant of DOCK2 and fell within the H3K27ac peak area in bronchial smooth muscle cells. The high-throughput chromosome conformation capture data suggest that rs6555875 significantly interacts with FAM196B in IMR90 cells. The intron variant of RARS, rs11134516, overlapped with H3K4me3, H3K27ac, CTCF, and DNase peaks in white blood cells and lung-related tissues/cells according to Encyclopedia of DNA Elements data (cCRE accession ID:EH38E2429336). The ChIP-seq experiments in HBECs and bronchial smooth muscle cells further supported the regulatory potential of rs11134516. The RegulomeDB scores of rs185216867 and rs62384405 indicated that these SNPs were likely to affect transcription factor binding.

Figure 4.

Functional annotation of 19 variants identified from fine-mapping analysis at 5q35.1. Each color represents SNPs identified from the fine-mapping conditional analysis and their linkage disequilibrium SNPs (R² > 0.8) with at least one or more than one regulatory element data. ^¶RegulomeDB (RegDB) score represents the likelihood that a variant will be located within a functional region. The RegDB variant classification scheme is defined as follows (50): Likely to affect binding: 2b, TF binding + matched TF motif + matched DNase footprint + DNase peak. Minimal binding evidence: 4, TF binding + DNase peak; 5, TF binding or DNase peak; 6, motif hit. ^§Overlapping with H3K27ac chromatin immunoprecipitation sequencing peaks in bronchial smooth muscle cells. ^†High signals of epigenetic markers in immune cells and lung (Encyclopedia of DNA Elements data). ^‡Overlapping with H3K27ac chromatin immunoprecipitation sequencing peaks in human bronchial epithelial cells. For definition of abbreviations, see Figure 3.

Expression Analyses

We found two of the African ancestry–associated SNPs prioritized by admixture mapping (rs10920072 and rs10920079), which were in high LD with one another and were also cis-eQTL variants of the TMEM9 gene in nasal epithelial cells and MROH3P in esophagus mucosa (Figure E4 and Table 4). The minor alleles of these SNPs were found more frequently among the African population and associated with decreased expression of TMEM9 (P_rs10920072 = 2.00 × 10⁻⁹ and P_rs10920079 = 1.31 × 10⁻⁹) and decreased lung function after adjusting for covariates (P_rs10920072 = 1.18 × 10⁻⁵ and P_rs10920079 = 4.23 × 10⁻⁶). Using nasal epithelial gene expression data, we showed decreasing trends of TMEM9 expression and lung function with respect to one or two copies of the minor allele of rs10920072 and rs10920079 (Figure 5). TMEM9 expression was correlated with lung function (R² = 0.18; P = 0.93 × 10⁻⁵) (Figure E5).

Table 4.

Expression Analysis Results: African Ancestry

SNP ID	Chr	Position (hg38)	Minor Allele	Minor Allele Frequency			Nasal Epithelial Cells			Esophagus Mucosa (GTEx)
				YRI	CEU	PUR*	β	P Value	Gene	β	P Value	Gene
rs10920079	1	200950242	C	0.47	0.05	0.33	−0.19	1.31 × 10−9	TMEM9	−0.36	1.60 × 10−5	MROH3P
rs10920072	1	200942726	A	0.66	0.06	0.42	−0.21	2.00 × 10−9	TMEM9	−0.36	1.40 × 10−5	MROH3P

Definition of abbreviations: CEU = Utah Residents with Northern and Western European ancestry; Chr = chromosome; GTEx = Genotype-Tissue Expression Project; PUR = Puerto Ricans from Puerto Rico; YRI = Yoruba in Ibadan, Nigeria.

The expression quantitative trait locus mapping in nasal epithelial cells identified two significant cis-acting expression quantitative trait loci of TMEM9. These SNPs were also significantly associated with MROH3P expression in esophagus mucosa according to GTEx data. The minor allele of these SNPs tracked with African population, and these alleles were also found more frequently among our study participants than among people with European ancestry.

^*Minor allele frequency computed from the Puerto Rican study participants (n = 836).

Figure 5.

Expression analysis results. Expression of TMEM9 gene and lung function by minor allele of the candidate SNPs (rs10920072 and rs10920079) is shown.

Discussion

Findings from our previous studies have suggested that varying proportions of genetic ancestry may play a critical part in asthma prevalence and severity in African American and Latino populations (7, 8, 12, 19). Thus, we reasoned that the admixed population structure of Puerto Ricans could be leveraged to identify genetic variants associated with lung function. To our knowledge, our study is the first study to have delineated and validated potential biological pathways associated with lung function in Puerto Rican youth using admixture-mapping, whole-genome sequence, and functional genomics data. We discovered two genomic regions associated with lung function and candidate SNPs with important functional elements that may influence inflammatory pathways in lung-related tissues and peripheral blood cells.

We previously demonstrated that African genetic ancestry was associated with lower lung function in participants with and without asthma. Therefore, we performed an African-specific ancestry admixture-mapping analysis and discovered a genomic region at 1q32 that was associated with a risk of decreased lung function among the 836 Puerto Ricans with and without asthma. The lead locus-specific ancestry SNP, rs17696752, resides in a potential regulatory region in the intergenic region of the genes LINC02789 and NR5A2 based on overlap with a GATA3 ChIP-seq cluster in the human neuroblast cell line (SH-SY5Y); GATA3 is a master transcription factor involved with T-cell development and expression of T-helper cell type 2 cytokine genes (31). This SNP was also associated with decreased FEV₁ in the U.K. Biobank data set (β = −0.01; P = 1.9 × 10⁻⁴) (32). The conditional analysis with sequenced data identified four candidate SNPs that could potentially explain the ancestry signal at 1q32. One of the SNPs was a synonymous variant of the CACNA1S gene, which plays a key role in skeletal muscle movement (contraction and relaxation) by generating calcium channels in muscle cells (33). It is not clear how CACNA1S is involved with lung function. However, rs3850625, a functionally deleterious missense variant of CACNA1S, was associated with FVC in the U.K. Biobank data set, further validating our study results (34). Previous genome-wide association studies identified genetic variants associated with asthma, lung function, and bronchodilator drug response at 1q32 (35–37). In addition, we previously identified this locus in a genome-wide association study of Puerto Rican asthma trios (father, mother, and affected child), independent of the current study, and found that the genomic region 1q32.2 was associated with asthma (21).

The potential biologic plausibility of the SNPs selected from the fine mapping above, as well as SNPs in high LD with them (r² > 0.8), were further investigated by performing a bioinformatics search and functional experiments. We found four SNPs located in potential regulatory regions in lung-related tissues and peripheral blood cells. Two of our lung function–associated SNPs (rs10920079 and rs10920072) were eQTLs of TMEM9, and their minor alleles were significantly associated with decreased expression of TMEM9 in nasal epithelial cells from Puerto Rican youth. These two SNPs were also significantly associated with decreased MROH3P expression in esophagus mucosa according to GTEx data. The minor alleles of these SNPs were more prevalent in both African populations and our study participants compared with European populations. Another study demonstrated that expression of the TMEM9 gene was significantly associated with lung function (38). Although the direction of the association between TMEM9 and lung function depends on tissue and cell types, their findings indicate that TMEM9 gene expression could be used as an important biologic target for pulmonary diseases.

A recent study suggested that expression of TMEM9 regulates secretion of IL-6 and IL-1β, which are important cytokines involved with airway inflammation and asthma (39, 40). Our previous study has demonstrated that the interaction between genetic ancestry and variants in IL-6 and IL-6R modified the effects of bronchodilator drug response in patients with asthma (41). We further tested this hypothesis with our RNA-sequencing data and found that increased expression of TMEM9 was inversely correlated with both IL-6 and IL-1β expressions in nasal epithelial cells (Figures E6 and E7, respectively). Among our study subjects, having one or two copies of the minor allele of rs10920072 and rs10920079 appeared to reduce expression of TMEM9 and yield lower FEV₁ values. Therefore, we speculate that the poor lung function of Puerto Rican youth could be mediated through the IL-6 and IL-1β inflammatory pathways and by TMEM9 gene expression.

We additionally discovered an admixture-mapping peak on chromosome 5q35.1, indicating that with each allele of Native American ancestral origin, lung function increased after adjusting for covariates. This is consistent with previous work demonstrating that Native American genetic ancestry was associated with improved lung function (19, 41). The Native American ancestry SNP with the smallest P value, rs12153426, fell within a potential regulatory region in intron 4 of SLIT3 based on overlap with a GATA3 ChIP-seq cluster in the human neuroblast cells (SH-SY5Y). SLIT3 is a protein-coding gene highly expressed in the lung (42). Previous studies demonstrated that expression of SLIT3 was involved with tumor suppression in lung cancer and possibly involved with organ developments including the diaphragm and kidney in mice (42–44). Fine-mapping analysis with sequenced data identified seven candidate SNPs to further explain the association between Native American genetic ancestry signal and lung function. Two SNPs (rs187841563 and rs73320187) were located within a potential regulatory region of intron 25 in DOCK2 based on overlap with ELF and P300 ChIP-seq clusters in B-lymphocyte cells. Expression of DOCK2 influences spatial rearrangement of the cytoskeleton, a necessary step prior to lymphocyte migration triggered by chemokine signaling upon antigen challenge (45).

The subsequent functional analysis of the seven candidate SNPs and their LD SNPs (r² > 0.8) revealed that 19 SNPs had an evidence of regulatory function in lung-related tissues and peripheral blood cells. Based on relevant functional annotation, two SNPs (rs11134516 and rs12516801) appeared to have a strong enrichment of H3K4me3, H3K27ac, and DNase I hypersensitivity ChIP-seq signals, indicating a potential role in gene expression regulation. These two SNPs fell within an intronic region of RARS (arginyl-tRNA synthetase) and may interact with the FBLL1 gene. In addition, based on the RegulomeDB categories and Encyclopedia of DNA Elements epigenetics data, rs62384405 and rs185216867 may have regulatory functions, too. We do not know how rs62384405, an exon variant of a long noncoding RNA transcript overlapping with TENM2, influences lung function. However, a different variant residing within the intronic region of TENM2, rs73803919, was associated with a positive response to bronchodilator drug response (46). The exact mechanism(s) of how gene expression of RARS and FBLL1 influences lung function among our participants is an important direction for future study.

The region-based association tests did not find any window that was significantly associated with lung function after multiple testing correction. However, we identified four windows within the fine-mapping regions of chromosome 1 and 5 that passed the suggestive threshold levels, indicating that both common and rare variants in our admixture-mapping peaks can influence lung function.

Certain genetic variants may play important roles in development of childhood-onset asthma (47, 48). The mean (±SD) age for the asthma onset among our participants was 2.1 (±2.6) years. We found that the age of asthma onset significantly interacted with the lead Native American ancestry SNP (rs12153426) on lung function, suggesting that those who developed asthma later in life and had one or more than one copy of a Native American ancestry allele at the given locus resulted in positive lung function (βeta = 0.008; P = 0.001) (Table E3). Because our participants were composed of broad age groups (8–21 yr), we also examined the effects of age. After adjusting for covariates, age was significantly associated with lung function. However, the interactions between the lead ancestry SNPs and age were not statistically significant.

This study is among the few asthma genetic studies focusing exclusively on Puerto Rican youth recruited from the island of Puerto Rico. Applying the admixture-mapping technique over traditional genome-wide allelic association tests with genetically admixed Puerto Ricans enabled discovery of the disease-associated genetic variant(s) attributable to different ancestry haplotype blocks. In addition, our uniquely rich data set includes WGS, RNA sequencing, and eQTL in ciliated nasal epithelial cells from Puerto Rican youth and functional assays in lung-related tissues. We previously demonstrated that gene expression patterns between ciliated nasal epithelial cells and bronchial tissue biopsies overlap by 93% (49). Our WGS facilitated both common and rare variant analyses, thereby considering all potential causal genetic variants. Furthermore, we validated important biological mechanisms and pathways that could potentially be used to treat patients with asthma. Finally, the association between the 1q32.2 genomic region and lung function/asthma was also reported in two independent data sets, the U.K. Biobank and Puerto Rican families from the GALA I study (21, 32).

Although our study significantly contributes to improving our understanding of the genetic components of lung function among Puerto Rican youth, our findings should be interpreted with caution. First, our sample size was limited because we were studying a population of minority youth from a specific geographic location. Although our identified genomic regions reached genome-wide significance after correcting multiple testing, adequate power in future studies will invariably require larger sample sizes. Second, the findings from this study may not be generalizable or applicable to other populations. Although it is likely that the Native American ancestry SNPs that we identified in this study are population-specific, it is also possible that our candidate SNPs are relevant to other admixed Latino populations.

In summary, admixture mapping identified genomic regions that harbored candidate genetic variants that could influence lung function among the Puerto Rican youth with and without asthma. We validated multiple inflammatory pathways with our biological data, which has important clinical and public health implications.