Abstract

   Women with endometriosis (EMS) appear to be at a higher risk of
   developing other autoimmune diseases predominantly multiple sclerosis
   (MS). Though EMS and MS are evidently diverse in their phenotype, they
   are linked by a common autoimmune condition or immunodeficiency which
   could play a role in the expansion of endometriosis and possibly
   increase the risk of developing MS in women with EMS. However, the
   common molecular links connecting EMS with MS are still unclear. We
   conducted a meta-analysis of microarray experiments focused on EMS and
   MS with their respective controls. The GEO2R web application discovered
   a total of 711 and 1516 genes that are differentially expressed across
   the experimental conditions in EMS and MS, respectively with 129 shared
   DEGs between them. The functional enrichment analysis of DEGs predicts
   the shared gene expression signatures as well as the overlapping
   biological processes likely to infer the co-occurrence of EMS with MS.
   Network based meta-analysis unveiled six interaction
   networks/crosstalks through overlapping edges between commonly
   dysregulated pathways of EMS and MS. The PTPN1, ERBB3, and CDH1 were
   observed to be the highly ranked hub genes connected with
   disease-related genes of both EMS and MS. Androgen receptor (AR) and
   nuclear factor-kB p65 (RelA) were observed to be the most enriched
   transcription factor in the upstream of shared down-regulated and
   up-regulated genes, respectively. The two disease sample sets compared
   through crosstalk interactions between shared pathways revealed
   commonly up- and down-regulated expressions of 10 immunomodulatory
   proteins as probable linkers between EMS and MS. This study pinpoints
   the number of shared genes, pathways, protein kinases, and upstream
   regulators that may help in the development of biomarkers for diagnosis
   of MS and endometriosis at the same time through improved understanding
   of shared molecular signatures and crosstalk.

   Keywords: endometriosis, multiple sclerosis, pathway analysis,
   enrichment analyses, autoimmune disease, immunodeficiency,
   meta-analysis

Introduction

   Endometriosis (EMS) is an estrogen-dependent inflammatory disorder
   which affects approximately 5–10% of women in the reproductive age
   worldwide (Bulun, [29]2009). The endometrial tissue which normally is
   present inside the uterus is displaced outside in patients suffering
   from EMS resulting in pelvic pain and infertility (Capobianco and
   Rovere-Querini, [30]2013). Immunological factors are known to
   contribute significantly to the pathogenesis and pathophysiology of
   endometriosis (Berkkanoglu and Arici, [31]2003; Podgaec et al.,
   [32]2007, [33]2010; Fairbanks et al., [34]2009; Nielsen et al.,
   [35]2011; Capobianco and Rovere-Querini, [36]2013). The prime
   regulators of the innate immune response are macrophages which come
   into play in case of injury, damage and infection. Macrophages possess
   functionally diverse contrasting roles, as on one hand, they play a
   protective role through differentiation and regeneration of cells while
   on the other hand they stimulate the immune response leading to
   destruction of infected cells (Vogel et al., [37]2013).
   Pro-inflammatory cytokine (interferon-γ) activated macrophages are
   known to have an essential role in the onset and progression of
   endometriosis. The macrophages misinterpret the displaced ectopic
   endometrial tissue as an injury and hence, instead of removing the
   endometrial cells, they activate pathways that repair and enhance their
   survival leading to sustained endometrial tissue (Podgaec et al.,
   [38]2010; Capobianco and Rovere-Querini, [39]2013). It has been
   reported that the women suffering from EMS are more prone to acquire
   other inflammatory autoimmune disorders especially multiple sclerosis
   (MS) (Nielsen et al., [40]2011; Mormile and Vittori, [41]2014).
   Multiple sclerosis (MS) is a chronic neuroinflammatory autoimmune
   disease of the central nervous system associated with neurodegeneration
   (Hickey, [42]1999; Compston and Coles, [43]2002; Szczucinski and Losy,
   [44]2007). Like endometriosis, macrophages have been observed to be
   directly associated in the pathogenesis of MS (Oreja-Guevara et al.,
   [45]2012; Vogel et al., [46]2013). Additionally, T-helper 1
   (Th1)/T-helper 2 (Th2) imbalance has been associated with both EMS and
   MS wherein the pro-inflammatory Th1 profile dominates over the Th2
   anti-inflammatory response. This is similar to other autoimmune
   diseases where the immune system launches an attack on its own cells
   and tissues (Trapp et al., [47]1998; Peterson et al., [48]2001; Diestel
   et al., [49]2003; Aktas et al., [50]2005). Among the increased Th1
   cytokines, it has been reported that Interferon-γ (IFN-γ) is strongly
   associated with the pathomechanisms of MS (Oreja-Guevara et al.,
   [51]2012; Vogel et al., [52]2013). Though the association between these
   two heterogeneous diseases EMS and MS, is not clearly recognized, it
   may be attributable to differential gene expression and sharing of
   common dysregulated pathogenic pathways involved in the development of
   both diseases. A 'crosstalk' event between two pathways, thus,
   elucidates how one or more components of one pathway affect another
   through interactions with shared components, protein-protein
   interactions and transcriptional regulations (Lu et al., [53]2007; Guo
   and Wang, [54]2009; Housden and Perrimon, [55]2014). Therefore, an
   examination of possible crosstalks and shared components among common
   dysregulated pathways together with associated genes in both
   endometriosis and MS may be able to assist in the understanding of the
   disease mechanism.

   Over the last two decades, a meta-analysis approach has been well
   exploited to uncover the shared molecular signatures between related
   diseases by integrating the publicly accessible microarray datasets
   (Silva et al., [56]2007; Higgs et al., [57]2012; Tuller et al.,
   [58]2013; Jha et al., [59]2016). Recent studies have focused on
   identifying crosstalk among dysregulated pathways using expression
   profiles of genes from control vs. disease samples (Zhang et al.,
   [60]2013; Niu et al., [61]2014, [62]2015; Chen et al., [63]2015). In
   the present study, we aimed to identify commonly dysregulated genes and
   pathways which probably co-occur in both EMS and MS to elucidate the
   relationship between the two diseases. We performed here a
   meta-analysis using gene expression data from microarray experiments of
   EMS and MS with their respective controls to predict the Differentially
   Expressed Genes (DEGs) involved in the respective diseases (Arasappan
   et al., [64]2011; Taminau et al., [65]2014). Widely used enrichment
   analysis methods such as Kyoto Encyclopedia of Genes and Genomes
   (KEGG), Gene Ontology (GO), and Protein-Protein Interactions (PPI) were
   adopted for the prediction of dysregulated pathways and subsequent
   possible crosstalk between EMS and MS. The findings from this study
   increase our understanding of the molecular mechanisms affecting both
   EMS and MS. Moreover, it brings forth the commonly shared genes,
   molecules and pathways co-existing in both EMS and MS which may be
   further explored as newer therapeutic targets.

Materials and methods

Data acquisition

   Widely accessible gene expression datasets related to endometriosis
   (EMS) and MS were obtained from the Gene Expression Omnibus (GEO)
   database of NCBI ([66]http://www.ncbi.nlm.nih.gov/geo/; Barrett and
   Edgar, [67]2006; Barrett et al., [68]2013). The keywords
   “endometriosis” and “multiple sclerosis” with “homo sapiens” or “human”
   were employed to mine the dataset. Studies evaluated on Affymetrix
   human gene expression dataset (irrespective of platform) containing
   samples from both normal and diseased tissue (more or less equally
   distributed) of women were taken. It was also ensured that these
   studies included only tissue samples that were not cultured in vitro.
   Similarly, tissue samples treated with any drugs before extraction were
   also excluded. The expression profiles of both primary and secondary
   cell cultures were also not considered for this analysis. Overall 14
   datasets (7 each for EMS and MS) which met these criteria were selected
   from published studies and downloaded from GEO database. The expression
   datasets of EMS combined tissue samples from 7 GEO profiles, i.e.,
   [69]GSE11691, [70]GSE25628, [71]GSE51981, [72]GSE6364, [73]GSE7305,
   [74]GSE7307, and [75]GSE7846 were selected. Likewise, expression
   datasets of MS involved tissue samples from 7 GEO profiles, i.e.,
   [76]GSE16461, [77]GSE21942, [78]GSE26484, [79]GSE38010, [80]GSE41848,
   [81]GSE41849, and [82]GSE41890. These samples were further separated
   into groups according to tissue source. Datasets were not subjected to
   any additional normalization, as all the data obtained had already been
   processed/normalized and were cross-comparable. The related information
   regarding the dataset pertaining to the microarray platform used,
   sample type and sample size are listed in Table [83]1.

Table 1.

   Published datasets related to endometriosis (A) and multiple sclerosis
   (B) used in this study.
   GEO accession Platform No. of probes No. of samples (control/disease)
   Tissue stage References