Sex Differences in Genomics in Lupus: Girls with Systemic Lupus Have High Interferon Gene Expression While Boys Have High Levels of Tumor Necrosis Factor-Related Gene Expression

Abstract

Objectives

Systemic lupus erythematosus (SLE) is a chronic disease occurring up to 15 times more frequently in females than males. This bias extends to possible differences in disease flares and response to therapy. This study was initiated to investigate the differences between girls and boys with childhood-onset SLE (cSLE) at the molecular level.

Methods

We analyzed the Gene Expression Omnibus National Center for Biotechnology Information microarray data available for 88 girls and 16 boys with treatment-naïve cSLE and compared the results to healthy controls. Transcriptional profiles were generated using the platforms of Affymetrix U133A and U133B gene chips and Bioconductor/R programming packages were used to process and compare the data.

Results

Girls with cSLE overexpress an interferon-alpha signature that appears absent in boys. Boys with cSLE were observed to overexpress tumor necrosis factor-related genes that were absent in girls. Boys and girls were observed to overexpress several genes related to granulopoeisis.

Conclusions

Our results suggest a potential application of genomics to differentially predict response to therapy between females and males with SLE.

Introduction

Systemic lupus erythematosus (SLE) is a multi-system autoimmune disease that primarily affects women of child-bearing age, with the highest incidence and prevalence in young post-pubertal women. The worldwide incidence in children less than 19 years has been estimated at 0.36–2.5/100,000/year, and the prevalence at 1.89–25.7/100,000 (1) across all ethnic groups. It has long been suggested that sex hormones play a role in the pathogenesis of SLE. There is a distinct increase in the number of females affected during puberty, with a ratio of 4:1 affected females to affected males pre-pubertal that increases to between 7:1 to 15:1 post-pubertal (2). Moreover, it is estimated that 15–20% of patients with SLE are diagnosed prior to their 19^th birthday (childhood-onset SLE), with a more severe disease course and severe organ involvement (1).

Sex hormones seem to be an important factor in differentiating between SLE in males and females. SLE may flare during pregnancy or monthly menstrual cycles (3). There may also be differences in clinical symptoms and response to standard-of-care treatment. Arthritis appears to be less common in men than women as a presenting symptom, but serositis, discoid lesions, and renal involvement (in particular, diffuse proliferative nephritis) are more common in men (4–6). Men with lupus in the Dallas-Fort Worth cohort had greater disease severity than women, with more cardiovascular and renal comorbidities (7), though this was equivocal in a cohort of lupus patients from the United Kingdom (8). Men with Klinefelter’s syndrome may have a higher incidence of lupus than those who do not (9). On a molecular level, it was recently shown that in a mouse model of lupus, female mice expressed higher levels of pro-inflammatory molecules than males, and higher numbers of immune cells such as T cells were seen in the spleen and gut of female mice compared to males (10).

Pro-inflammatory cytokines such as interferon (IFN) and tumor necrosis factor-alpha (TNF) are believed to play a role in the pathogenesis of SLE. With the advent of microarray, the landmark study by Bennett et al (11) demonstrated overexpression of IFN-related genes in 29 of 30 childhood-onset SLE (cSLE) patients, two of which strongly correlated with disease activity. Moreover, transcription profiling with microarray before and after treatment with intravenous corticosteroid in 3 patients demonstrated ablation of this IFN signature.

The role of TNF in SLE is not as well-defined as it has been shown to have protective as well as deleterious effects. Serum levels of TNF have been shown to correlate with disease activity in patients with lupus nephritis (12). However, a well-known adverse effect of patients on anti-TNF inhibitors is the potential to develop SLE. Palucka et al (13) demonstrated that 5 patients with juvenile arthritis on treatment with an anti-TNF inhibitor overexpressed IFN-regulated genes.

A major challenge in improving the care of SLE patients is the current lack of specific biomarkers for care. A decade of research in cancer genomics has convincingly demonstrated the utility of correlating gene expression signatures with clinical outcomes (reviewed in 14). There are currently no data on genomic differences in expression between males and females with SLE that may explain differences in disease onset, severity, and/or organ involvement.

Methods

We analyzed the existing Gene Expression Omnibus National Center for Biotechnology Information (GEO NCBI) data (15) of untreated cSLE patients to evaluate gene expression profiles for boys versus girls with cSLE. There were 16 untreated boys with cSLE and 6 healthy male controls that were compared against 88 untreated girls with cSLE and 16 healthy female controls. Using GEO2R, an interactive web tool available on the NCBI website, gene expression data from these patients were analyzed. Transcriptional profiles were generated using the platforms of Affymetrix U133A and U133B gene chips, totaling over 44,000 probe sets. Bioconductor/R programming packages were used to parse the data, and the linear models for microarray analysis (limma) R package was used to process and compare these data. Consistent with accepted parameters for biomarker discovery, a threshold of 1.5 log fold-change and an adjusted p-value using the Benjamini-Hochberg method of <0.05 was applied (16).

Results

The mean age of the 88 treatment-naïve girls with cSLE was 13.3 years (range 8–17 years). Sixty percent were Hispanic, 27% black, 10% white, and 3% other. The mean age of the 16 treatment-naïve boys with cSLE was 14.1 years (range 9–17 years). Fifty-six percent were Hispanic, 25% black, and 19% other. Table 1 shows a sample of the differences in gene expression profiles between treatment-naïve girls and boys with cSLE. It is evident that there is minimal overlap between these profiles, and that the IFN-alpha signature identified by Bennett et al (11) is overwhelmingly present in treatment-naïve girls, but absent in treatment-naïve boys with cSLE. Interestingly, the boys with cSLE demonstrated differential expression of genes related to TNF more so than IFN production. Both sexes demonstrated some differential expression of granulopoeisis-related genes, but the signature was stronger in the girls. The differential expression of granulopoeisis-related genes has been reported in the past (11) in SLE patients with active and inactive disease.

Table 1.

Differential gene profiling in treatment-naïve girls as compared to boys with lupus in top 100 genes from both sexes. These lists were generated from lists of the top 100 genes with the greatest fold-change and most significant p-value from healthy controls.

Females	Gene ID	Gene Name
Interferon signature:	SIGLEC1	Sialoadhesin
(range fold change: 1.752–5.661,	IFI44L	interferon-induced protein 44-like
range log10 p-values: 7.06–13.9)	IFI27	interferon, alpha-inducible protein 27
	IFIT1	interferon-induced protein with teratricopeptide repeats 1
	IFIT3	interferon-induced protein with tetratricopeptide repeats 3
	IFI44	interferon-induced protein 44-like ligand
	IFI6	interferon, alpha-inducible protein 6
Granulopoesis-related:	ELANE	neutrophil expressed elastase
(range fold change: 1.16–2.33,	S100	S100 calcium binding protein
range log10 p-values: 6.02–6.5)	MMP8	neutrophil collagenase
	DEFA1	defensin A1
	DEFA4	defensin A4
	DEF3	defensin 3
	GRN	granulin
Males
Tumor necrosis factor-related:	GRK1	G protein coupled receptor kinase 1
(range fold change: 1.15–2.12,	TP63	tumor protein p63
range log10 p-values: 4.08–4.82)	PTPRD	protein tyrosine phosphatase receptor type D
	SERPINA7	serine peptidase inhibitor A7
	TMEM59	transmembrane protein 59
	TRIM2	tripartite motif containing 2
Granulopoesis-related:	ELANE	neutrophil expressed elastase
(range fold change: 0.7–1.45,	DEF3	defensin 3
range log10 p-values: 3.8–4.3)	MPO	myeloperoxidase

Discussion

Girls with SLE overexpress an IFN-alpha signature that appears absent in boys while boys with SLE were observed to overexpress TNF-related genes that were absent in girls, though at a lower level than the IFN-alpha overexpression in girls. Although these data remain to be validated, they suggest a novel theory in the pathogenesis of SLE, namely that SLE may have a different driving cytokine pathway in females than males.

This is an intriguing concept, in particular its relation to development of treat-to-target and individualization of therapy for SLE patients in the future. It has been demonstrated by various groups that differential gene expression in peripheral blood mononuclear cells exists between active and inactive disease states (17,18) using the “nearest shrunken centroids” method for data analysis. Rus et al demonstrated that several genes highly correlated with disease activity scores (SLEDAI) with correlation coefficients ranging from 0.4 to 0.7 (17). Kawasaki et al used quantitative real-time polymerase chain reactions to validate the correlation of 27 genes identified by DNA microarray with disease activity (18). Both groups identified genes that were up- and down-regulated differently in active disease as compared to inactive disease. Most commonly, genes related to TNF receptors and converting enzyme were identified in both studies as being highly correlated with SLEDAI. However, the data remain scarce on studies in gene expression profiling between males and females with active and inactive SLE.

Bennett et al (11) were the first to note differential gene expression post-treatment with high-dose corticosteroids. In fact, the authors described how high-dose corticosteroids ablate the IFN signature in SLE patients. However, the data remain scarce on response to treatment, clinical outcome, and their correlation with disease activity/remission. Although a weakness of our analysis was that we have not yet validated our data, we suspect a difference in gene expression also exists between males and females with SLE and their response to therapy, given their divergent gene expression profiles at the time of diagnosis when the patients are still treatment-naïve. The parameters employed in this study follow guidelines for setting thresholds in biomarker discovery (16) as it has been shown that using more stringent thresholds for fold change and/or log10 transformed p-values can result in false discovery rates and the discarding of potentially clinically significant differentially expressed genes. Moreover, the small number of healthy controls in this study is a limitation. Further studies in transcriptional profiling and subsequent epigenetic studies in males and females pre- and post-immunosuppressive therapy would be of further interest and aid in delineation of response to treatment profiles.