Skip to main content
NCBI home page
Human Molecular Genetics logoLink to Human Molecular Genetics
. 2019 Jan 14;28(10):1682–1693. doi: 10.1093/hmg/ddz014

Sex differences in gene expression in response to ischemia in the human left ventricular myocardium

Gregory Stone 1, Ashley Choi 1, Oliva Meritxell 2, Joshua Gorham 3, Mahyar Heydarpour 1, Christine E Seidman 3, Jon G Seidman 3, Sary F Aranki 4, Simon C Body 1, Vincent J Carey 5, Benjamin A Raby 5,#, Barbara E Stranger 2,#, Jochen D Muehlschlegel 1,✉,#
PMCID: PMC6494791  PMID: 30649309

Abstract

Sex differences exist in the prevalence, presentation and outcomes of ischemic heart disease (IHD). Females have higher risk of heart failure post-myocardial infarction relative to males and are two to three times more likely to die after coronary artery bypass grafting surgery. We examined sex differences in human myocardial gene expression in response to ischemia. Left ventricular biopsies from 68 male/46 female patients undergoing aortic valve replacement surgery were obtained at baseline and after a median 74 min of cold cardioplegic arrest/ischemia. Transcriptomes were quantified by RNA-sequencing. Cell-type enrichment analysis was used to estimate the identity and relative proportions of different cell types in each sample. A sex-specific response to ischemia was observed for 271 genes. Notably, the expression FAM5C, PLA2G4E and CYP1A1 showed an increased expression in females compared to males due to ischemia and DIO3, MT1G and CMA1 showed a decreased expression in females compared to males due to ischemia. Functional annotation analysis revealed sex-specific modulation of the oxytocin signaling pathway and common pathway of fibrin clot formation. Expression quantitative trait locus (eQTL) analysis identified variant-by-sex interaction eQTLs, indicative of sex differences in the genotypic effects on gene expression. Cell-type enrichment analysis showed sex-bias in proportion of specific cell types. Common lymphoid progenitor cells and M2 macrophages were found to increase in female samples from pre- to post-ischemia, but no change was observed in male samples. These differences in response to myocardial ischemia provide insight into the sexual dimorphism of IHD and may aid in the development of sex-specific therapies that reduce myocardial injury.

Introduction

There is a growing recognition of sex differences in ischemic heart disease (IHD) outcomes. Women with non-obstructive coronary artery disease have a higher mortality and event rate than men (1) and are more likely to develop heart failure after myocardial infarction (2). The operative mortality of females undergoing coronary artery bypass grafting (CABG) is also two to three times higher than for men (3). Despite the systematic differences between men and women in the presentation and outcomes of IHD, women remain grossly underrepresented in research studies, and sex-specific outcomes are spuriously reported (4).

We have previously investigated changes in gene expression due to ischemia in the human left ventricle (LV) using cardioplegia and cardiopulmonary bypass (CPB) as a reproducible human ischemic model of myocardial infarction (5). While there has been some investigation into sex-specific transcription in the healthy myocardium (6), a dearth of knowledge remains regarding sex differences in the transcriptional response to myocardial ischemia. To date, most sex-specific investigations have focused on sex hormones and genes on the sex chromosomes (7). However, evidence exists to suggest that genes on autosomal chromosomes may show sex differences in the response to ischemia and that this variable response has implications for sex-specific outcomes (8).

We employed whole-genome RNA-sequencing (RNA-seq) to characterize sex differences in the human transcriptional response to myocardial ischemia and to identify sex-biased genomic loci that contribute to variation in expression levels of RNA after ischemia, namely expression quantitative trait loci (eQTLs). To our knowledge, this is the first study investigating sex differences in IHD at the transcriptional level.

Results

We define up-regulated and down-regulated genes as those genes whose expression differs in the pre- and post-ischemic states (irrespectively of sex), where up-regulation refers to an increase in gene expression from the pre- to post-ischemia state and down-regulation to a decrease. We define sex-biased response genes as those that demonstrate significant sex differences in either the magnitude or the direction of the post-ischemia response. Notice that the sex-bias definition is exclusively restricted to the response to ischemia rather than to transcriptomic sex differences at baseline.

Sixty-six million reads per sample were available for alignment after adapter and quality trimming, and 84.5% of reads were uniquely aligned to the human genome (UCSC hg19) with Spliced Transcripts Alignment to a Reference (STAR). After removal of adapter sequences, quality trimming, alignment to the UCSC hg19 genome and removing genes with low counts in multiple samples, 15 265 genes were analyzed for differential expression in response to ischemia (Fig. 1). Of these genes, 8417 were differentially expressed in both sexes, 1263 were differentially expressed in females only and 2266 were differentially expressed in males only (Supplementary Material, Table S1). For 271 genes, we observed a differential response to ischemia by sex (as determined by formal interaction testing, Padj < 0.05) (Supplementary Material, Table S2) —70 genes changed in only males, 96 changed in only females and 105 changed in both sexes but to different degrees. Several of these genes, such as SCARA5, have previously been shown to respond to myocardial ischemia differently between male and female mice (9). The effect of sex on gene expression was visualized with a multidimensional scaling plot, which revealed stratification by sex in the 3rd dimension, showing a clear demarcation between male and female samples (Fig. 2). Furthermore, in Figure 3, we show the separation of genes with significant change that presents in only females, in only males and in females and males combined.

Figure 1.

Figure 1

Open in a new tab

Flow chart showing order of processing of RNA reads and SNPs. Processing was conducted from the top of the chart downwards. (A) Each box represents a step in the processing of the RNA reads prior to differential gene expression and eQTL analyses, culminating in 15 265 genes for differential expression analysis and 14 967 genes for baseline and 15 251 genes for ischemia cis-eQTL analysis. (B) Workflow of multiple SNPs filtering steps prior to eQTLs identification, giving a final number of 663 575 SNPs for cis-eQTL analyses.

Figure 2.

Figure 2

Open in a new tab

Multidimensional scaling plot showing separation of samples by sex. Each point represents a gene in either a male or female sample. The distance between any pair of samples is the LFC between the root-mean-square deviation of counts of the top 500 genes. A shorter distance indicates a greater similarity between a pair of samples. X- and Y-axis show distances in the 3rd and 4th dimensions, respectively. Male samples are coral and female samples are turquoise.

Figure 3.

Figure 3

Open in a new tab

Sex-specific changes in gene expression due to ischemia. Each point represents a gene with a significant LFC. A significant change in both sexes is red, change in only females is green and change in only males is blue. X- and Y-axis are LFCs in males and females, respectively.

Differential gene expression

Relative up-regulation in females

The gene with the largest difference between females and males in transcriptional response to ischemia was FAM5C, which was found to have a log2 fold change (LFC) of 1.91 in females and 0.56 in males (up-regulated in response to ischemia in both males and females) and an LFC of 1.36 for the ischemic response between females and males (relative up-regulation in females) (Fig. 4, Table 1). We also observed a sex difference in expression response to ischemia for PLA2G4E, which encodes the protein phospholipase A2 group IVE. In response to ischemia, PLA2G4E was found to have an LFC of 0.57 in females and −0.56 in males (Fig. 4, Table 1) and therefore was found to be relatively up-regulated in females at an LFC of 1.14. Another gene with a differential response to ischemia by sex was cytochrome P450-family 1-subfamily A-polypeptide 1 (CYP1A1), which was down-regulated in males (LFC = −0.63) and showed no significant change in females in response to ischemia (LFC = 0.33 with Padj = 0.08) (Fig. 4). Therefore, CYP1A1 was found to be relatively up-regulated in females at an LFC = 0.96 (Table 1).

Figure 4.

Figure 4

Open in a new tab

Relative change in gene expression due to ischemia. Orange and blue bars represent the mean LFC in gene expression in female and male samples, respectively, due to ischemia. The black brackets denote examples of the changes in gene expression in females relative to males due to ischemia.

Table 1.

Top eight up- and down-regulated genes in females that are significantly differently regulated relative to males

Gene name Gene symbol LFC (Benjamini-Hochberg-adjusted P-value)
Female ischemic response Male ischemic response Female–male difference in ischemic response
Genes associated with a more positive ischemic response in females than in males
Family with sequence similarity 5, member C FAM5C 1.91 (1.86 × 10−8) 0.56 (0.04) 1.36 (0.04)
Phospholipase A2 group IVE PLA2G4E 0.57 (0.01) −0.56 (0.003) 1.14 (0.01)
LOC400794 LOC400794 1.57 (4.45 × 10−12) 0.52 (0.003) 1.05 (0.01)
Spindle and kinetochore-associated complex subunit 1 SKA1 0.84 (5.02 × 10−8) −0.17 (0.17) 1.01 (0.0003)
PDZ binding kinase PBK 1.04 (2.14 × 10−8) 0.07 (0.65) 0.97 (0.005)
Cytochrome P450, family 1, subfamily A, polypeptide 1 CYP1A1 0.33 (0.08) −0.63 (1.03 × 10−5) 0.96 (0.005)
Relaxin/insulin-like family peptide receptor 1 RXFP1 1.46 (1.96 × 10−12) 0.5 (0.001) 0.96 (0.01)
LOC440896 LOC440896 0.99 (1.66 × 10−5) 0.08 (0.7) 0.91 (0.05)
Genes associated with a more negative ischemic response in females than in males
Iodothyronine deiodinase 3 DIO3 −1.72 (2.20 × 10−10) −0.38 (0.07) −1.34 (0.007)
Metallothionein-1G MT1G −0.28 (0.30) 0.89 (2.10 × 10−5) −1.17 (0.03)
Chymase 1 CMA1 −1.16 (8.63 × 10−9) 0.00 (0.98) −1.16 (0.001)
Ras protein-specific guanine nucleotide-releasing factor 1 RASGRF1 −0.59 (0.005) 0.48 (0.004) −1.08 (0.007)
Chitinase-3-like protein 2 CHI3L2 −1.53 (2.56 × 10−12) −0.48 (0.003) −1.04 (0.009)
Zinc finger protein 385B ZNF385B −1.67 (1.02 × 10−14) −0.72 (1.54 × 10−5) −0.96 (0.02)
Potassium voltage-gated channel subfamily D member 2 KCND2 −1.23 (2.21 × 10−7) −0.28 (0.14) −0.95 (0.05)
MyoD family inhibitor MDFI −1.15 (2.58 × 10−7) −0.25 (0.16) −0.9 (0.04)
Open in a new tab

Relative down-regulation in females

The gene with the largest sex difference in response to ischemia was DIO3, which encodes the protein iodothyronine deiodinase 3 (D3) (Table 1). In response to ischemia, DIO3 was found to have an LFC of −1.72 in females and −0.38 in males (down-regulated in both sexes) and an LFC of −1.34 between females and males (relative down-regulation in females) (Fig. 4). Additionally, MT1G, which encodes the protein Metallothionein-1G, was found to have an LFC of −0.28 with Padj = 0.30 in females in response to ischemia and an LFC of 0.89 in males, resulting in an LFC = −1.17 between the sexes (Fig. 4). Also, CMA1, which encodes the protein chymase, was significantly down-regulated in females (LFC = −1.16) and showed no change in males in response to ischemia (LFC = 0.00 with Padj = 0.98) (Fig. 4).

Functional analysis

Functional annotation analysis of the 271 genes that showed a sex-bias in response to ischemia revealed significant enrichment of several GO, KEGG, REACTOME, Human Phenotype Ontology (HP), CORUM, miRBase and TRANSFAC terms (Table 2, Supplementary Material, Table S3). For example, in the 271 genes showing a sex-bias in response to ischemia and enrichment of genes involved in the oxytocin, signaling pathway was identified by analysis with g:Profiler. The genes found to be associated with the oxytocin signaling pathway, PLA2G4E, KCNJ3, PLCB3, SRC, CD38, RYR1 and TRPM2, were either down-regulated in response to ischemia in females and up-regulated in males, or down-regulated in both sexes, but to a greater degree in females, meaning that the expression of the aforementioned genes is reduced in females due to ischemia relative to males (Table 2; see Supplementary Material, Table S2 for specific LFC and P-values for the listed genes). PLCB3 was the only gene to be up-regulated in both sexes but to a lesser extent in females. Additionally, though transcription of the gene encoding oxytocin receptor (OXTR) is down-regulated in the rat infarcted LV (10), OXTR was down-regulated in females while up-regulated in males. Previous studies have shown oxytocin to be cardioprotective post-infarction (10) and after ischemia/reperfusion injury (11). Here, the female ischemic myocardium suppresses the oxytocin signaling pathway more so than in males, thereby negating its protective effects and possibly inducing vulnerability in females due to ischemia.

Table 2.

Functional analysis. GO and KEGG associations elucidated genes involved in biological pathways

Functional term Proportion of category genes expressed in a sex-specific manner in response to ischemia Benjamini–Hochberg-adjusted P-value Number of genes
GO
Multi-multicellular organism process 0.055 0.02 12
Monocyte differentiation 0.192 0.05 5
KEGG Pathway
Glutamatergic synapse 0.053 0.050 6
Oxytocin signaling 0.044 0.03 7
Estrogen signaling 0.05 0.02 5
Ovarian steroidogenesis 0.039 0.005 2
REACTOME
Signaling by Platelet-derived growth factor 0.062 0.05 4
The ligand trap binds the ligand Bone morphogenetic protein (BMP) 2, blocking BMP signaling 0.333 0.04 2
Gab1 signalosome 0.182 0.03 2
Sustained activation of SRC kinase by SHP2 0.25 0.02 2
Hemostasis 0.038 0.02 9
PECAM1 interactions 0.167 0.04 2
Phosphorylation of PECAM-1 by Fyn or Lyn or c-Src 0.4 0.006 2
Common pathway of fibrin formation 0.182 0.05 2
Open in a new tab

REACTOME enrichments identified genes involved specific biological processes and reactions. The results presented in this table were obtained by functional analysis of the 271 genes with sex-specific differential expression due to ischemia using g:Profiler. Proportion of category genes expressed in a sex-specific manner in response to ischemia = the proportion of the 271 genes with sex-specific differential expression due to ischemia that were found to be relevant to the functional term. Number of genes = the number of genes, out of the 271 queried, found to be relevant to the functional term.

Sex-biased genes were also enriched in the common pathway in fibrin formation and hemostasis. The gene for coagulation factor X (F10) was up-regulated to a greater degree in males than females, while the gene encoding thrombomodulin (THBD) was down-regulated to a greater degree in females than males in response to ischemia. F10 and THBD therefore both show a relative down-regulation in females, and the expression of both genes has been shown to be cardioprotective in animal models (12,13). Additionally, SERPINA5, which encodes Protein C inhibitor that inhibits thrombin–thrombomodulin complex formation (14), was down-regulated in both sexes in response to ischemia but to a greater magnitude in females. Sex-specific changes in hemostasis due to myocardial ischemia, and the mediation of hemostasis by the LV, require further investigation.

eQTL identification and cell-type deconvolution

We performed eQTL analysis to assess whether genetic variation influences sex differences in ischemic response. Of the 114 patients in this study, complete genotype information was obtained for 110 (66 males, 44 females). After applying filtering criteria, 663 575 Single nucleotide polymorphisms (SNPs) were tested for eQTLs, with 14 967 genes at baseline and 15 251 genes at post-ischemia. cis-eQTL analyses at baseline and post-ischemia were performed separately: 1) females only, 2) males only and 3) females and males combined. cis-eQTLs at baseline for 95 genes were identified exclusively in females, 359 genes exclusively in males and 932 genes in females and males combined (Padj < = 0.05). Fewer post-ischemia cis-eQTLs were identified: 46 genes in females only, 317 genes in males only and 804 genes in females and males combined (Padj < 0.05). We identified SNP-by-sex interaction cis-eQTLs at baseline (LETM2, rs11985898, Padj = 0.02) and at post-ischemia (SLC30A7, kgp2665681, Padj = 0.04 and VGLL4, rs11707097, Padj = 0.05), indicative of sex differences in the genotypic effects on gene expression (Fig. 5, Supplementary Material, Table S4). Out of the 271 genes that had gender-specific ischemic response, we have identified 7 genes that are significant eQTL at post-ischemia for females and males combined: ATP6AP1L, C1QTNF9, GSTM5, SPATA7, THEM178, WBSCR27 and ZNF385B.

Figure 5.

Figure 5

Open in a new tab

DNA sequence variant rs11985898 has a sex-specific effect on the expression of LETM2 at baseline only. The boxplots in pink and blue show female and male LETM2 expression, respectively. The Y-axis denotes expression levels, whereas the X-axis shows, from left to right, homozygous for the reference allele (0), heterozygous (1) and homozygous for the alternate allele (2) genotypes. At baseline, males with the homozygous reference genotype have significantly higher expression of LETM2 than females, while males with the homozygous alternate genotype have significantly lower expression. At post-ischemia, there is no difference between sexes.

Clustering of the cell-type enrichment profiles of 220 samples revealed clustering by individual, and not by ischemia (Supplementary Material, Fig. S2), demonstrating that intra-individual samples exhibit similar cell-type composition profiles that are not altered by ischemia. Furthermore, when clustered with the multi-tissue genotype-tissue expression (GTEx) cell-type enrichment profiles, the biopsy samples cluster with the GTEx LV tissues, showing that the tissue collected is comparable between studies (Supplementary Material, Fig. S3).

At both baseline and ischemia conditions, we observe sex differences in cell-type enrichment scores (Supplementary Material, Fig. S4). Baseline samples exhibited significant sex differences in enrichment of several cell-types: hematopoietic stem cells, chondrocytes, macrophages M1 and immature dendritic cells are significantly more abundant in females than males (1.1, 1.22, 1.39, 1.47 female/male ratio, respectively; t-test, P < 0.01). In female samples, we observed an increase in the abundance of common lymphoid progenitor cells post-ischemia relative to baseline (fold change, 1.13) and a decrease in macrophages M2 abundance post-ischemia (fold change, 0.89). These effects were not observed in males.

Discussion

The female sex is an independent risk factor for hospital and operative mortality after cardiac surgery (3). Furthermore, females have higher risk of heart failure post-myocardial infarction relative to males (2). While the transcriptional response to ischemia in the LV myocardium has previously been characterized (5), the sexual dimorphism of IHD, and a paucity of sex-specific investigation to date (4), suggests sex-specific transcriptional responses might explain these mechanisms. In this investigation, we conducted a whole-genome RNA-seq analysis to identify sex-specific changes in gene expression in the human LV myocardium using CPB and cold blood cardioplegia as a model for ischemia. Functional analysis of identified genes revealed enrichment of several biological processes. An eQTL analysis identified sex differences in the genetic regulation of gene expression post-ischemia, and cell-type deconvolution analysis showed sex-bias in proportion of specific cell types.

Differential gene expression

FAM5C is expressed in coronary artery smooth muscle and endothelium where over-expression increases monocyte adhesion (15). The greater up-regulation of FAM5C in women in response to ischemia in our study indicates that women may experience increased monocyte adhesion, and therefore a larger inflammatory infiltrate, in response to ischemia than do men. FAM5C may therefore serve as a novel target to modulate cardiac inflammation following ischemia.

The superfamily of phospholipase A2 enzymes is composed of five subgroups, several of which are biomarkers for ischemic events (16). Phospholipase A2 group IVE, encoded by PLA2G4E, has recently been implicated in endocytosis (17) and in the biosynthesis of N-acyl phosphatidylethanolamines (NAPEs) (18). Cytosolic PLA2E has been observed to share an endosomal binding site with PLD2 (17), which localizes to the sarcolemmal membrane and regulates endocytosis of angiotensin-II type I receptor, which is injurious to the myocardium after ischemia/reperfusion injury (19,20). In our study, PLD2 was significantly up-regulated in both sexes but to a lesser degree in women. Changes in PLA2G4E expression, and its relationship to PLD2, may therefore modulate cardiac angiotensin action. Additionally, PLA2G4E-transfected cells accumulated downstream NAPE metabolites, which may come from a specific source of NAPEs (18) and would reveal an injurious modulation of a specific, potentially harmful reservoir of biological molecules in only females during myocardial ischemia.

CYP1A1, which was down-regulated in males and showed no significant change in females in response to ischemia, metabolizes estrogens in the myocardium to form a 2-hydroxy metabolite (2H) (21), which can then be metabolized to a 2-methoxy product (2ME) (22). 2ME has been demonstrated to be anti-angiogenic and augment apoptosis in endothelial cells (23), and, in the myocardium, both metabolites increase apoptosis in ischemia–reperfusion injury (24,25). Furthermore, ESR1, which encodes estrogen receptor alpha (ERα), was expressed similarly between the sexes at baseline but showed a differential response to ischemia with up-regulation in males and down-regulation in females in response to ischemia. ERα has been shown to protect against ischemic injury (26). Estrogen is an established cardioprotective agent in pre-menopausal females, and hormone replacement therapy (HRT) has been investigated in the reduction of cardiovascular events. However, HRT has been found to increase cardiovascular disease in post-menopausal females within the first several years of treatment (27), though no patients in this study were undergoing HRT. Our results suggest that the changes seen in post-menopausal females, the expression of genes that increase estrogen metabolism and reduce estrogen binding, may be injurious in IHD, and/or that the increased action and reduced metabolism of estrogen in males following ischemia may be a protective mechanism not shared by post-menopausal females.

D3, down-regulated to a significantly greater degree in females, is one of three iodothyronine deiodinases and is responsible for deactivating thyroid hormone (28). Myocardial DIO3 induction has been observed in infarcted rat models (29). While in both sexes there were changes in iodothyronine deiodinase 2 and 3 gene expression, thyroid hormone levels were not measured. Despite a cardioprotective indication for thyroid hormone post-ischemia (30), it may not be metabolized in the hearts of patients undergoing cardiac surgery (31). Epigenetic down-regulation of DIO3, which has been observed in atherosclerotic arteries (32), may therefore be responsible for the detected change in females and may reflect a sex-specific injurious vascular response in the myocardium to ischemic insult.

Chymase inhibition has been shown to be cardioprotective following ischemia/reperfusion injury and myocardial infarction (33). However, in this experiment, CMA1 was down-regulated in females and showed no change in males due to ischemia, suggesting that chymase is injurious to the myocardium. While the precise role of chymase in the myocardium remains unclear (34), this investigation complicates the role of chymase in cardiac remodeling and warrants further investigation into the role of chymase gene expression in the myocardium following ischemic injury.

cis-eQTL

The effect on gene expression of the SNP-by-sex interaction response QTLs identified in this investigation has not been previously described. We demonstrated that not only could cis-eQTLs have sex-dependent effects on gene expression at baseline but also in response to ischemia. One such gene to demonstrate such interaction is VGLL4. VGLL4 binds to TEAD4 and forms a critical transcription factor complex that regulates vascular endothelial growth factor A, a well-known antigenic cytokine that gets activated in response to ischemia to restore blood flow (35). rs11707097-AA genotype is associated with decreased VGLL4 expression in post-ischemia in females only, suggesting that the genotype determines its sex-specific down-regulation due to ischemia. While the eQTLs identified here may indicate that women with certain genotypes are at an increased risk of ischemic injury, some genes had few homozygous genotypes. Nevertheless, SNP-by-sex interaction response QTLs clearly warrant further investigation.

Limitations and Conclusion

This investigation is unique in several ways. Sample collection was performed at the same physical location on the LV apex by the same surgeon at the same time in the operation, thereby reducing sample variability. The time from sample collection to RNA preservative was about 1 min, resulting in extremely high-quality RNA, and whereas similar studies often obtain tissue post-mortem or from cardiac transplantation, tissue was collected from live, non-failing hearts. While the sampled hearts had concentric hypertrophy from long-standing aortic stenosis, all samples were collected during non-urgent, elective procedures, and risk factors were optimally medically managed beforehand.

Despite the strengths of this investigation, there remain limitations. Ischemia induced by cold cardioplegia is not equivalent to in vivo ischemia, akin to coronary artery ligation in rodents. The results of this experiment may therefore underestimate the extent of ischemic injury. Additionally, down-regulation of gene expression may be due to cold temperature. However, the release of cardiac-specific biomarkers to levels that would qualify as a myocardial infarction in the non-surgical population suggests ischemic changes in response to aortic cross-clamping and CPB. Furthermore, while we were not able to distinguish individual cell types using single-cell sequencing, our cell-type enrichment analysis demonstrates that intra-individual samples exhibit similar cell-type composition profiles that are unaffected by ischemia. Lastly, the small amount of tissue collected prohibited protein-level validation, and different isoforms were not compared due to limitations in current differential isoform analysis software.

Moving forward, we envision to use this cohort to explore the genetics of gene expression by fine-mapping the eQTL loci, colocalizing the eQTL signal with cardiovascular traits to identify cases where the trait is compatible with a particular eQTL signal and doing mediation analyses (e.g. by Mendelian randomization) to identify if the eQTL signal ‘mediates’ the cardiovascular trait signal.

In conclusion, we have demonstrated that the human LV demonstrates significant sex-specific changes in gene expression as a result of cold cardioplegia-induced ischemia during CPB. The results of this investigation provide a valuable foundation in examining the sexual dimorphism of surgical outcomes in CPB and may, by extension, help elucidate the mechanisms underlying sex differences in IHD. The genes, pathways and variants revealed by this experiment should serve as targets for further exploration and, more generally, support the necessity of sex-specific experimentation in the advancement of human health and precision medicine for all.

Materials and Methods

Patients and tissue samples

One hundred fourteen patients undergoing elective aortic valve replacement surgery were prospectively enrolled after obtaining written informed consent approved by the Partners Healthcare Institutional Review Board (Boston, MA). Of the 114 patients in this study, 68 (60%) were male and 46 (40%) were female (Table 3). The median age for men was 71 years and for women was 76 years, with all women being post-menopausal.

Table 3.

Patient demographics and clinical characteristics

Demographic Total cohort Females Males P-value
N = 114 N = 46 N = 68
Age (years) 73 (65–80) 76 (68–84) 71 (63–78) 0.03
BMI (kg/m2) 29.8 (26.4–34.6) 32.8 (28.6–37.7) 28.5 (26.0–32.4) 2.0 × 10−3
Diabetes N (%) 49 (43.0) 19 (41.3) 30 (44.1) 0.85
CAD N (%) 56 (49.1) 20 (43.5) 36 (52.9) 0.350
Post-operative day 1 CKMB (μg/L) 25.7 (19.5–33.3) 25.6 (19.5–33.1) 26.5 (19.2–33.3) 0.95
Ischemic time (minutes) 74 (61–97) 67 (56–83) 82 (66–103) 2.1 × 10−3
LV ejection fraction 60 (55–65) 60 (55–65) 60 (55–65) 0.36
Open in a new tab

Median values with inter-quartile range. Diabetes: diagnosed by a physician; BMI: body mass index; CAD: coronary artery disease, diagnosed by angiography and requiring coronary revascularization; CKMB: creatine kinase MB fraction; LV: left ventricle; Aortic cross-clamp: ischemic time of the left ventricle. P-values were determined by Wilcoxon signed-rank test for age, BMI, CKMB, ischemic time and LV ejection fraction and by Chi-squared test for CAD and diabetes. A P < 0.05 denotes a significant difference in the mean value between males and females. N = number of affected patients, % = percent of affected patients.

Two punch biopsies from the anterolateral apical LV wall, one pre- (baseline) and one post-ischemia, were obtained. Baseline samples were collected immediately after commencement of CPB, which included intermittent cold blood cardioplegia for myocardial protection. Post-ischemic samples were obtained immediately before removal of the aortic cross-clamp at the end of the de-airing procedure. As the complexity of CABG surgery varies depending on patient characteristics, this is reflected in the range of ischemic times (range: 41–195 min). Tissue samples were immediately placed in RNAlater® (Ambion, ThermoFisher Scientific, Waltham, MA) and after 48 h at +4°C were stored at −80°C until RNA extraction. After being preserved in RNAlater® immediately after collection, tissue samples were RNA sequenced in batches at a later time-point on the Illumina HiSeq 2000 (Illumina, San Diego, CA). RNA extraction and sequencing are discussed more thoroughly in the Supplementary File, but, briefly, total RNA was isolated with Trizol and RNA quality was assessed using the Agilent Bioanalyzer 2100 (Agilent, Santa Clara, CA). Median RNA integrity number score was 6.8 (interquartile range, 6.3–7.5) across all samples. RNAs possessing poly-A tails were selected by annealing them to poly-T oligos bound to beads (Invitrogen, Life Technologies, Grand Island, NY) so as to remove ribosomal RNA. Messenger RNAs (mRNAs) were reverse transcribed using random hexamers (Invitrogen). The resulting cDNA was made into a double strand amplified by polymerase chain reaction. Sequencing adapters were attached to the double-stranded DNA, and strands of appropriate length for next-generation sequencing were selected by electrophoresis. mRNA samples from RNA-seq were prepared using the TruSeq RNA Sample Preparation Kit (Illumina). Template molecules were used for cluster generation and sequencing on the Illumina HiSeq 2000 (Illumina).

Paired-end reads 50, 90 or 100 base pairs long were generated, producing ~66 million reads per sample. Adapter sequences were removed using Skewer (36), and reads were trimmed using Sickle (37) at a quality threshold of 5. Reads were then aligned to the UCSC hg19 reference human genome using STAR (v2.5.2b) (38). Aligned reads were subsequently counted using featureCounts (v1.5.1) (39). Since differential gene expression was measured, not absolute expression, adjusting for gene length was not necessary.

Biological sex was verified by measuring the pre-ischemic expression levels of XIST, an X-linked gene expressed significantly higher in females than in males at baseline (6). XIST expression pre-ischemia was plotted for each sample. Four male samples were found to express XIST at levels both non-zero and greater than the lowest expressing female and were therefore removed from further analysis.

Differential expression and functional analysis

We consider two forms of differential expression: 1) difference in expression level in response to ischemia in each sex (e.g. in the post-ischemia sample, gene A is expressed higher in females than in males) and 2) differential response to ischemia by sex (e.g. gene A increases its expression from baseline to post-ischemia in females but not in males).

All statistical analyses were performed in R (v3.3.1). Gene counts were pre-processed by applying a variance-stabilizing transformation and count normalization using voom (40). The voom-transformed gene counts were analyzed for differential expression with limma (v3.30.11) (41). For a greater discussion of the use of limma, please refer to the Supplementary File.

Genes were filtered conservatively for inclusion in accordance with practices established by the authors of edgeR—a gene was required to have a count of at least six in at least two patients to be included in downstream analysis (42). Corrected for the smallest library size (1 961 608 reads), genes expressed at a count-per-million (cpm = raw count × 1 000 000/library size) less than 3.05 were removed from downstream analysis. Supplementary Material, Figure S1, shows the effect of filtering on stabilizing mean–variance trend. A linear model was constructed to calculate the mean expression of a gene, denoted y, in each sex at baseline and at post-ischemia while adjusting for age, diabetes and ischemic time. Adjusting for these cofactors effectively eliminates these factors’ contributions to changes in gene expression and allows one to compare changes in gene expression due to ischemia by sex:

graphic file with name M1.gif

where β denotes regression coefficients and the factor ‘sex at condition’ has four levels: female expression at baseline, female expression at post-ischemia, male expression at baseline and male expression at post-ischemia.

To identify genes with significant difference in expression level in response to ischemia in each sex, the significance of the following model-fitted contrasts was assessed:

‘female ischemia response’ = ‘female expression at post-ischemia’ – `female expression at pre-ischemia’.

‘male ischemia response’ = ‘male expression at post-ischemia’ – ‘male expression at pre-ischemia’.

To determine if the change in the expression of a gene in response to ischemia differed between the sexes, the model was fit to the following contrast, which produced an estimated LFC that measured the difference between the change in gene expression within females and that within males in response to ischemia:

Differential response to ischemia by sex = (female expression at post_ischemia – female expression at pre_ischemia) – (male expression at post_ischemia – male expression at pre_ischemia)

The significance (Padj < 0.05) of each contrast was determined by empirical Bayes moderated t-statistics, and adjustment for multiple testing was performed using Benjamini–Hochberg false discovery rate (FDR) (43). A positive/negative LFC is interpreted as an increase/decrease in gene expression due to ischemia in females relative to that in males. A significant ‘Differential response to ischemia by sex’ contrast is compatible with multiple scenarios (see Fig. 4 for illustration):

1) Gene expression significantly increases/decreases, in a direction opposite of the other sex. 2) Gene expression significantly increases/decreases in both sexes but more significantly in one sex. 3) Gene expression significantly increases/decreases only in one sex.

Functional analysis of the genes with sex-specific differential expression was performed using g:Profiler (44), which searches functional terms from GO, KEGG, REACTOME, CORUM, miRBase and Human Phenotype Ontology databases. Significant functional terms were identified at Benjamini–Hochberg FDR = <10% (43). Genes were queried in descending order of magnitude of fold change between males and females due to ischemia. g:Profiler analyzes ordered queries by sampling increasingly larger numbers of genes from the top of the list, which allows one to identify functional terms associated with larger changes in gene expression and the list of genes as a whole (44). GO and KEGG associations elucidated genes involved in biological pathways. REACTOME enrichments identified genes involved specific biological processes and reactions, and CORUM associations revealed genes implicated in protein complexes. miRBase terms denoted miRNAs thought to target associated genes, and TRANSFAC terms indicated transcription factor binding sites possessed by listed genes. HP enrichments revealed genes associated with a specific human phenotype.

eQTL and cell-type deconvolution

Using DNA isolated from whole blood, SNP genotyping was performed using the Illumina Omni2.5 with exome content genotyping array (Illumina). SNPs were excluded if the minor allele frequency was less than 5%, the genotype missing rate was greater than 5% or were not in Hardy–Weinberg equilibrium (P < 10−6). These filters were applied to the non-imputed genotypes of males and females separately, with only the SNPs surviving in both groups retained for testing. The associations between SNP genotypes and gene expression were tested using Matrix eQTL, separately for baseline and post-ischemia profiles. We inferred hidden factors affecting gene expression for both sexes combined and both sexes independently using Probabilistic Estimation of Expression Residuals (PEER) (45). A cis-eQTL analysis (within 100 000 base pairs from the transcription start site) was performed to identify and compare cis-eQTLs present at baseline and post-ischemia. To identify eQTL with different effects sizes in males and females at post-ischemia, an interaction cis-eQTL analysis was performed using matrix eQTL (46). A linear model was constructed to test the interaction of sex and SNP genotype on PEER-corrected residuals while adjusting for age, ischemic time, diabetes, sequence center, number of reads and genetic principal components (PC1, PC2 and PC3) (46). A negative interaction term indicates a larger eQTL effect size in females, whereas a positive value indicates a larger eQTL effect size in males. We performed this analysis separately for samples collected at baseline and at post-ischemia. We retained those genes with a significant SNP-by-sex interaction eQTL (Benjamini–Hochberg FDR = 5%) only in the post-ischemia analysis for downstream analysis. Multiple-testing adjustment was done for all cis-eQTL analyses: EigenMT (47) to correct for the number of SNPs within genes and Benjamini–Hochberg to correct for the number of genes.

To test the shared effects between differentially expressed genes due to ischemia and the differential response to ischemia by sex, we quantified the proportion of true positives estimated from the enrichment of significant P-values (pi1) (48).

The tissue biopsy and resulting transcriptome profile are not restricted to a single cell type. We attempted single-cell RNA-seq using two different techniques (Fluidigm C1 and 10x Genomics Chromium) but were not successful given the size of human myocytes. Therefore, we performed a cell-type enrichment analysis to obtain an estimate of the identity and relative proportions of different cell types present in each sample and to compare them to other tissues including heart LV characterized by the GTEx Project. The data used for the analyses described in this manuscript were obtained from the GTEx Portal on 12/1/2017 (49).

We performed cell-type enrichment analysis using xCell (50) on the transcriptomes of all 11 688 GTEx v7 samples and the pre- and post- ischemia LV biopsies. Sample enrichment scores were obtained for 64 immune and non-immune cell types. For all cell types, we tested for sex differences in a) cell abundances in pre-ischemic samples and b) the ratio of cell abundances in post- versus pre-ischemia. We applied principal component analysis to the matrix of cell types and sample enrichment scores to ensure that the observed differences in expression between females and males are not a result of variable expression across heterogeneous cell types. Furthermore, we explored the contribution of each cell type to differential gene expression between sexes in response to ischemia.

Supplementary Material

Supplementary_File_ddz014
Click here for additional data file. (22.9KB, docx)
Supplementary_Figure_and_Table_Legend_ddz014.docx
Click here for additional data file. (14KB, docx)
S1_Fig_ddz014
Click here for additional data file. (173.8KB, png)
S1_File_ddz014
Click here for additional data file. (14.1KB, docx)
S1_Table_Revised_ddz014
Click here for additional data file. (2.2MB, xlsx)
S2_Fig_ddz014
Click here for additional data file. (878.3KB, png)
S2_Table_Revised_ddz014
Click here for additional data file. (86KB, xlsx)
S3_Fig_ddz014
Click here for additional data file. (140.6KB, png)
S3_Table_Revised_ddz014
Click here for additional data file. (13.6KB, xlsx)
S4_Fig_ddz014
Click here for additional data file. (631.4KB, png)
S4_Table_new_ddz014
Click here for additional data file. (355.7KB, xlsx)

Acknowledgements

We acknowledge the contribution to sample collection by the Perioperative Genomics Center and its staff: James Gosnell, RN; Kujtim Bodinaku, MD; Svetlana Gorbatov, MPH.

Conflict of Interest statement. None declared.

Funding

National Institutes of Health (R01HL118266 to J.D.M.).

References

  • 1. Gulati M., Cooper-DeHoff R.M., McClure C., Johnson B.D., Shaw L.J., Handberg E.M., Zineh I., Kelsey S.F., Arnsdorf M.F., Black H.R. et al. (2009) Adverse cardiovascular outcomes in women with nonobstructive coronary artery disease: a report from the Women's Ischemia Syndrome Evaluation Study and the St James Women Take Heart Project. Arch. Intern. Med., 169, 843–850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Hellermann J.P., Jacobsen S.J., Gersh B.J., Rodeheffer R.J., Reeder G.S. and Roger V.L. (2002) Heart failure after myocardial infarction: a review. Am. J. Med., 113, 324–330. [DOI] [PubMed] [Google Scholar]
  • 3. Vaccarino V., Abramson J.L., Veledar E. and Weintraub W.S. (2002) Sex differences in hospital mortality after coronary artery bypass surgery: evidence for a higher mortality in younger women. Circulation, 105, 1176–1181. [DOI] [PubMed] [Google Scholar]
  • 4. Stramba-Badiale M. (2010) Women and research on cardiovascular diseases in Europe: a report from the European Heart Health Strategy (EuroHeart) project. Eur. Heart J., 31, 1677–1681. [Google Scholar]
  • 5. Muehlschlegel J.D., Christodoulou D.C., McKean D., Gorham J., Mazaika E., Heydarpour M., Lee G., DePalma S.R., Perry T.E., Fox A.A. et al. (2015) Using next-generation RNA sequencing to examine ischemic changes induced by cold blood cardioplegia on the human left ventricular myocardium transcriptome. Anesthesiology, 122, 537–550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Isensee J., Witt H., Pregla R., Hetzer R., Regitz-Zagrosek V. and Noppinger P.R. (2008) Sexually dimorphic gene expression in the heart of mice and men. J. Mol. Med. (Berl.), 86, 61–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Miller V.M., Kaplan J.R., Schork N.J., Ouyang P., Berga S.L., Wenger N.K., Shaw L.J., Webb R.C., Mallampalli M., Steiner M. et al. (2011) Strategies and methods to study sex differences in cardiovascular structure and function: a guide for basic scientists. Biol. Sex Differ., 2, 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Heidecker B., Lamirault G., Kasper E.K., Wittstein I.S., Champion H.C., Breton E., Russell S.D., Hall J., Kittleson M.M., Baughman K.L. et al. (2010) The gene expression profile of patients with new-onset heart failure reveals important gender-specific differences. Eur. Heart J., 31, 1188–1196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Chen Q., Williams R., Healy C.L., Wright C.D., Wu S.C. and O'Connell T.D. (2010) An association between gene expression and better survival in female mice following myocardial infarction. J. Mol. Cell. Cardiol., 49, 801–811. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Jankowski M., Bissonauth V., Gao L., Gangal M., Wang D., Danalache B., Wang Y., Stoyanova E., Cloutier G., Blaise G. et al. (2010) Anti-inflammatory effect of oxytocin in rat myocardial infarction. Basic Res. Cardiol., 105, 205–218. [DOI] [PubMed] [Google Scholar]
  • 11. Ondrejcakova M., Ravingerova T., Bakos J., Pancza D. and Jezova D. (2009) Oxytocin exerts protective effects on in vitro myocardial injury induced by ischemia and reperfusion. Can. J. Physiol. Pharmacol., 87, 137–142. [DOI] [PubMed] [Google Scholar]
  • 12. Guillou S., Tamareille S., Giraud S., Poitevin G., Prunier-Mirebeau D., Nguyen P., Prunier F. and Macchi L. (2016) Fondaparinux upregulates thrombomodulin and the endothelial protein C receptor during early-stage reperfusion in a rat model of myocardial infarction. Thromb. Res., 141, 98–103. [DOI] [PubMed] [Google Scholar]
  • 13. Herzog C., Lorenz A., Gillmann H.J., Chowdhury A., Larmann J., Harendza T., Echtermeyer F., Muller M., Schmitz M., Stypmann J. et al. (2014) Thrombomodulin's lectin-like domain reduces myocardial damage by interfering with HMGB1-mediated TLR2 signalling. Cardiovasc. Res., 101, 400–410. [DOI] [PubMed] [Google Scholar]
  • 14. Suzuki K., Nishioka J., Kusumoto H. and Hashimoto S. (1984) Mechanism of inhibition of activated protein C by protein C inhibitor. J. Biochem., 95, 187–195. [DOI] [PubMed] [Google Scholar]
  • 15. Sato J., Kinugasa M., Satomi-Kobayashi S., Hatakeyama K., Knox A.J., Asada Y., Wierman M.E., Hirata K. and Rikitake Y. (2014) Family with sequence similarity 5, member C (FAM5C) increases leukocyte adhesion molecules in vascular endothelial cells: implication in vascular inflammation. PLoS One, 9, e107236. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Burke J.E. and Dennis E.A. (2009) Phospholipase A2 biochemistry. Cardiovasc. Drugs Ther., 23, 49–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Capestrano M., Mariggio S., Perinetti G., Egorova A.V., Iacobacci S., Santoro M., Di Pentima A., Iurisci C., Egorov M.V., Di Tullio G. et al. (2014) Cytosolic phospholipase A(2)epsilon drives recycling through the clathrin-independent endocytic route. J. Cell Sci., 127, 977–993. [DOI] [PubMed] [Google Scholar]
  • 18. Ogura Y., Parsons W.H., Kamat S.S. and Cravatt B.F. (2016) A calcium-dependent acyltransferase that produces N-acyl phosphatidylethanolamines. Nat. Chem. Biol., 12, 669–671. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Du G., Huang P., Liang B.T. and Frohman M.A. (2004) Phospholipase D2 localizes to the plasma membrane and regulates angiotensin II receptor endocytosis. Mol. Biol. Cell, 15, 1024–1030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Han J., Park S.J., Thu V.T., Lee S.R., Long le T., Kim H.K., Kim N., Park S.W., Jeon E.S., Kim E.J. et al. (2013) Effects of the novel angiotensin II receptor type I antagonist, fimasartan on myocardial ischemia/reperfusion injury. Int. J. Cardiol., 168, 2851–2859. [DOI] [PubMed] [Google Scholar]
  • 21. Lee A.J., Cai M.X., Thomas P.E., Conney A.H. and Zhu B.T. (2003) Characterization of the oxidative metabolites of 17beta-estradiol and estrone formed by 15 selectively expressed human cytochrome p450 isoforms. Endocrinology, 144, 3382–3398. [DOI] [PubMed] [Google Scholar]
  • 22. Zacharia L.C., Dubey R.K., Mi Z. and Jackson E.K. (2003) Methylation of 2-hydroxyestradiol in isolated organs. Hypertension, 42, 82–87. [DOI] [PubMed] [Google Scholar]
  • 23. Yue T.L., Wang X., Louden C.S., Gupta S., Pillarisetti K., Gu J.L., Hart T.K., Lysko P.G. and Feuerstein G.Z. (1997) 2-Methoxyestradiol, an endogenous estrogen metabolite, induces apoptosis in endothelial cells and inhibits angiogenesis: possible role for stress-activated protein kinase signaling pathway and Fas expression. Mol. Pharmacol., 51, 951–962. [DOI] [PubMed] [Google Scholar]
  • 24. Wang Z., Liu D., Varin A., Nicolas V., Courilleau D., Mateo P., Caubere C., Rouet P., Gomez A.M., Vandecasteele G. et al. (2016) A cardiac mitochondrial cAMP signaling pathway regulates calcium accumulation, permeability transition and cell death. Cell Death Dis., 7, e2198. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Si J., Wang N., Wang H., Xie J., Yang J., Yi H., Shi Z., Ma J., Wang W., Yang L. et al. (2014) HIF-1alpha signaling activation by post-ischemia treatment with astragaloside IV attenuates myocardial ischemia-reperfusion injury. PLoS One, 9, e107832. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Novotny J.L., Simpson A.M., Tomicek N.J., Lancaster T.S. and Korzick D.H. (2009) Rapid estrogen receptor-alpha activation improves ischemic tolerance in aged female rats through a novel protein kinase C epsilon-dependent mechanism. Endocrinology, 150, 889–896. [DOI] [PubMed] [Google Scholar]
  • 27. Manson J.E., Hsia J., Johnson K.C., Rossouw J.E., Assaf A.R., Lasser N.L., Trevisan M., Black H.R., Heckbert S.R., Detrano R. et al. (2003) Estrogen plus progestin and the risk of coronary heart disease. N. Engl. J. Med., 349, 523–534. [DOI] [PubMed] [Google Scholar]
  • 28. Bianco A.C., Salvatore D., Gereben B., Berry M.J. and Larsen P.R. (2002) Biochemistry, cellular and molecular biology, and physiological roles of the iodothyronine selenodeiodinases. Endocr. Rev., 23, 38–89. [DOI] [PubMed] [Google Scholar]
  • 29. Olivares E.L., Marassi M.P., Fortunato R.S., Silva A.C., Costa-e-Sousa R.H., Araujo I.G., Mattos E.C., Masuda M.O., Mulcahey M.A., Huang S.A. et al. (2007) Thyroid function disturbance and type 3 iodothyronine deiodinase induction after myocardial infarction in rats a time course study. Endocrinology, 148, 4786–4792. [DOI] [PubMed] [Google Scholar]
  • 30. Novitzky D. and Cooper D.K. (2014) Thyroid hormone and the stunned myocardium. J. Endocrinol., 223, R1–R8. [DOI] [PubMed] [Google Scholar]
  • 31. Paolino B.S., Pomerantzeff P.M., Dallan L.A.O., Gaiotto F.A., Preite N.Z., Latronico A.C., Nicolau J.C., Bianco A.C. and Giraldez R. (2017) Myocardial inactivation of thyroid hormones in patients with aortic stenosis. Thyroid, 27, 738–745. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Aavik E., Lumivuori H., Leppanen O., Wirth T., Hakkinen S.K., Brasen J.H., Beschorner U., Zeller T., Braspenning M., Criekinge W. et al. (2015) Global DNA methylation analysis of human atherosclerotic plaques reveals extensive genomic hypomethylation and reactivation at imprinted locus 14q32 involving induction of a miRNA cluster. Eur. Heart J., 36, 993–1000. [DOI] [PubMed] [Google Scholar]
  • 33. Oyamada S., Bianchi C., Takai S., Chu L.M. and Sellke F.W. (2011) Chymase inhibition reduces infarction and matrix metalloproteinase-9 activation and attenuates inflammation and fibrosis after acute myocardial ischemia/reperfusion. J. Pharmacol. Exp. Ther., 339, 143–151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Kokkonen J.O., Lindstedt K.A. and Kovanen P.T. (2003) Role for chymase in heart failure: angiotensin II-dependent or -independent mechanisms? Circulation, 107, 2522–2524. [DOI] [PubMed] [Google Scholar]
  • 35. Teng A.C., Kuraitis D., Deeke S.A., Ahmadi A., Dugan S.G., Cheng B.L., Crowson M.G., Burgon P.G., Suuronen E.J., Chen H.H. et al. (2010) IRF2BP2 is a skeletal and cardiac muscle-enriched ischemia-inducible activator of VEGFA expression. FASEB J., 24, 4825–4834. [DOI] [PubMed] [Google Scholar]
  • 36. Jiang H., Lei R., Ding S.W. and Zhu S. (2014) Skewer: a fast and accurate adapter trimmer for next-generation sequencing paired-end reads. BMC Bioinformatics, 15, 182. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Joshi N.A. and Fass J.N. (2011), Sickle: A sliding-window, adaptive, quality-based trimming tool for FastQ files (Version 1.33) [Software]. Available athttps://github.com/najoshi/sickle.
  • 38. Dobin A., Davis C.A., Schlesinger F., Drenkow J., Zaleski C., Jha S., Batut P., Chaisson M. and Gingeras T.R. (2013) STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 29, 15–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Liao Y., Smyth G.K. and Shi W. (2014) featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics, 30, 923–930. [DOI] [PubMed] [Google Scholar]
  • 40. Liu R., Holik A.Z., Su S., Jansz N., Chen K., Leong H.S., Blewitt M.E., Asselin-Labat M.L., Smyth G.K. and Ritchie M.E. (2015) Why weight? Modelling sample and observational level variability improves power in RNA-seq analyses. Nucleic Acids Res., 43, e97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Ritchie M.E., Phipson B., Wu D., Hu Y., Law C.W., Shi W. and Smyth G.K. (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res., 43, e47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Robinson M.D., McCarthy D.J. and Smyth G.K. (2010) edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics, 26, 139–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Benjamini Y. and Hochberg Y. (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B Stat. Methodol., 57, 289–300. [Google Scholar]
  • 44. Reimand J., Arak T., Adler P., Kolberg L., Reisberg S., Peterson H. and Vilo J. (2016) g:Profiler-a web server for functional interpretation of gene lists (2016 update). Nucleic Acids Res., 44, W83–W89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Stegle O., Parts L., Piipari M., Winn J. and Durbin R. (2012) Using probabilistic estimation of expression residuals (PEER) to obtain increased power and interpretability of gene expression analyses. Nat. Protoc., 7, 500–507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Shabalin A.A. (2012) Matrix eQTL: ultra fast eQTL analysis via large matrix operations. Bioinformatics, 28, 1353–1358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Davis J.R., Fresard L., Knowles D.A., Pala M., Bustamante C.D., Battle A. and Montgomery S.B. (2016) An efficient multiple-testing adjustment for eQTL studies that accounts for linkage disequilibrium between variants. Am. J. Hum. Genet., 98, 216–224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Storey J.D. and Tibshirani R. (2003) Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. U. S. A., 100, 9440–9445. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49. The G.C. (2013) The genotype-tissue expression (GTEx) project. Nat. Genet., 45, 580–585. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Aran D., Hu Z. and Butte A.J. (2017) xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol., 18, 220. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary_File_ddz014
Click here for additional data file. (22.9KB, docx)
Supplementary_Figure_and_Table_Legend_ddz014.docx
Click here for additional data file. (14KB, docx)
S1_Fig_ddz014
Click here for additional data file. (173.8KB, png)
S1_File_ddz014
Click here for additional data file. (14.1KB, docx)
S1_Table_Revised_ddz014
Click here for additional data file. (2.2MB, xlsx)
S2_Fig_ddz014
Click here for additional data file. (878.3KB, png)
S2_Table_Revised_ddz014
Click here for additional data file. (86KB, xlsx)
S3_Fig_ddz014
Click here for additional data file. (140.6KB, png)
S3_Table_Revised_ddz014
Click here for additional data file. (13.6KB, xlsx)
S4_Fig_ddz014
Click here for additional data file. (631.4KB, png)
S4_Table_new_ddz014
Click here for additional data file. (355.7KB, xlsx)

Articles from Human Molecular Genetics are provided here courtesy of Oxford University Press

RESOURCES