RNA sequencing shows transcriptomic changes in rectosigmoid mucosa in patients with irritable bowel syndrome-diarrhea: a pilot case-control study

Abstract

Our aim was to conduct a pilot case-control study of RNA expression profile using RNA sequencing of rectosigmoid mucosa of nine females with -diarrhea-predominant irritable bowel syndrome (IBS-D) with accelerated colonic transit and nine female healthy controls. Mucosal total RNA was isolated and purified, and next-generation pair-end sequencing was performed using Illumina TruSeq. Analysis was carried out using a targeted approach toward 12 genes previously associated with IBS and a hypothesis-generating approach. Of the 12 targeted genes tested, patients with IBS-D had decreased mRNA expression of TNFSF15 (fold change controls to IBS-D: 1.53, P = 0.01). Overall, up- and downregulated mRNA expressions of 21 genes (P = 10⁻⁵ to 10⁻⁸; P values with false detection rates are shown) were potentially relevant to IBS-D including the following: neurotransmitters [P2RY4 (P = 0.001), vasoactive intestinal peptide (VIP, P = 0.02)]; cytokines [CCL20 (P = 0.019)]; immune function [C4BPA complement cascade (P = 0.0187)]; interferon-related [IFIT3 (P = 0.016)]; mucosal repair and cell adhesion [trefoil protein (TFF1, P = 0.012)], retinol binding protein [RBP2 (P = 0.017)]; fibronectin (FN1, P = 0.009); and ion channel functions [guanylate cyclase (GUCA2B, P = 0.017), PDZ domain-containing protein 3 (PDZD3, P = 0.029)]. Ten genes associated with functions related to pathobiology of IBS-D were validated by RT-PCR. There was significant correlation in fold changes of the selected genes (Rs = 0.73, P = 0.013). Up- or downregulation of P2RY4, GUC2AB, RBP2, FNI, and C4BPA genes were confirmed on RT-PCR, which also revealed upregulation of farnesoid X receptor (FXR) and apical sodium-coupled bile acid transporter (IBAT/ASBT). RNA-Seq and RT-PCR analysis of rectosigmoid mucosa in IBS-D show transcriptome changes that provide the rationale for validation studies to explore the role of mucosal factors in the pathobiology of IBS-D.

the cause of loose consistency of bowel movements in patients with irritable bowel syndrome with diarrhea (IBS-D) is attributed in part to acceleration of colonic transit, which is documented in ∼45% of patients with IBS-D (11), providing less time for reabsorption of water in the colon (16). Circulating postprandial levels of serotonin are increased in patients with IBS-D (17; 21). Increased duodenal and rectosigmoid expression of secretory transmitters and fecal excretion of granins are also documented (18; 19; 29) although it is unclear whether these result in intestinal fluid and electrolyte secretion or accelerated transit with reduced colonic fluid absorption. There is one study that suggested that the ileal mucosa of patients with IBS, compared with that of healthy controls, responds to infusion of bile acids with greater fluid and electrolyte secretion (18). Mutations in the gene GUCY2C that controls the secretory guanylate cyclase C receptor have been reported in an extensive family who presented with chronic symptoms suggestive of IBS-D (20).

The mucosal functions of patients with IBS-D and controls appear to differ (7). IBS has been associated with changes in the rectosigmoid mucosal expression of immune and nonimmune protective factors, such as barrier function and mucus secretion (2). The literature to date (2; 4; 5; 8; 12; 22; 36; 39–41; 44; 45) is summarized in Table 1. However, there are only a few examples, typically derived from assessments of single genes in relatively small numbers of patients and controls, where the inherited genotype has been associated with altered mucosal gene expression. For example, differential expression of the TNFSF15 gene, which is associated with IBS, has been linked with functional alterations of mucosal immune and protective functions (36; 45). Although genotype can clearly influence the expression of related proteins in the colonic mucosa, function is more closely reflected in the expression of mRNA in the tissue. There is also evidence of alterations in jejunal mucosal abnormalities in the expression and distribution of apical junction complex proteins, specifically, increased protein expression of claudin-2, reduced occludin phosphorylation, enhanced redistribution from the membrane to the cytoplasm (24), increased myosin kinase expression, reduced myosin phosphatase, and, consequently, enhanced phosphorylation of myosin (24). In addition, the Barcelona group also reported that expression of zonula occludens 1 (ZO-1) was reduced in IBS-D at both gene and protein level, with protein redistribution from the tight junction to the cytoplasm (25). In a prior study, we have demonstrated a borderline significant difference in the ZO-1 intensity score in the small bowel mucosa (P = 0.06) of patients with IBS-D, compared with healthy controls with lower intensity in HLA-DQ2/8-positive relative to HLA-DQ2/8-negative patients with IBS-D (39).

Table 1.

Summary of changes in mucosal expression in IBS in the published literature

Function or Phenotype	Mucosal Expression in Rectosigmoid Biopsies in IBS	Associated Genotype	Reference
Cellular defense functions, nonimmune-mediated	Upregulation of mucin gene (MUC20) and dual oxidase 2 (DUOX2); reduced caspase-1 (CASP1) and lysozyme (LYZ)	Not tested	Aerssens et al., 2008
Mucosal cell immune activation: proinflammatory IL-8, IL-1 β	Upregulation of TLR2 and TLR4 mRNA in epithelial cells;	Not tested	Belmonte et al., 2012
	Upregulation of CCL-11, CCL-13	No association with genotype	Swan et al., 2013
Serotonergic functions	Reduced SERT-P mRNA expression (normal in some studies)	5-HTTLPR	Coates et al., 2004, Camilleri et al., 2007, Wang et al., 2012, Kerckhoffs et al., 2012
Immune activation	Reduced TNFSF15 mRNA in PI-IBS	TNFSF15 (C allele rs6478108)	Swan et al., 2013,
	Increased TNFSF15 mRNA in healthy	TNFSF15(G allele rs4263839)	Zucchelli et al., 2011
	Increased IL13 mRNA	IL13	Wouters et al., 2013
Immune recognition and TJ protein	Not tested	CDH1, TLR9, IL6	Villani et al., 2010
TJ protein mRNA expression	Reduced mRNA ZO-1 in IBS-D; downregulation of ZO-1 and occludin expression with no changes in mRNA expression	HLA DQ2/8	Vazquez et al., 2012
			Bertiaux-Vandaele et al., 2011

IBS, irritable bowel syndrome; TLR, Toll-like receptor; PI-IBS, postinfectious IBS; TJ, tight junction; IBS-D, diarrhea-predominant IBS; ZO, zonnula occludens.

The development of next-generation RNA sequencing provides a novel opportunity to examine the entire transcriptome. Analysis of differential expression in well-matched patients and controls has the potential to assess differences in the transcriptome that may modify the function of the tissue (e.g., mucosa) in the disease of interest.

Our hypothesis was that RNA expression of mucosal biopsies from patients with IBS-D (with previously demonstrated accelerated colonic transit) differs from that of healthy controls. The aim of the study was to conduct RNA sequencing in stored mucosal samples obtained from the rectosigmoid region of nine female patients with IBS-D (with previously demonstrated accelerated colonic transit) and nine female healthy controls. We approached this aim using two strategies: first, we examined differential expression in the following 12 “targeted” genes associated with IBS in the literature: 5-HTTLPR, TNFSF15, FAAH, CNR1, KLB, FGFR4, GPBAR1, TLR9, C11orf30, ORMDL3, NPSR1, and HLA2/8; second, we sought associations with differential expression of any other gene mRNA, independent of any a priori hypothesis, that is, an “unbiased analysis” intended for hypothesis generation. The results obtained for 10 of the genes of interest were validated by a quantitative RT-PCR. An ancillary study used pathway analyses to explore how any identified alterations in mRNA expression might mediate the pathobiology of IBS.

MATERIALS AND METHODS

Design.

We conducted a case-control study of a convenient number of participants in this pilot study. We used stored samples from patients who had consented to the use of biospecimens (for future research) in prior studies (2; 38) conducted at Mayo Clinic in Rochester, MN. These samples were obtained in nine female patients with IBS-D and nine female healthy controls. RNA sequencing was conducted by the Mayo Medical Genomics Facility and bioinformatics by the Mayo Bioinformatics Core Facility to interpret differential expression in patients and controls. The study was approved by Mayo Clinic Institutional Review Board (IRB no. 13-001595, date of approval 3/13/2013).

Selection of participants.

We examined, using RNA sequencing, the RNA expression of rectosigmoid mucosal biopsies from nine female patients with IBS-D (age range, 25–63 yr; BMI range, 20–40.5 kg/m²) with rapid colonic transit (geometric center, GC24: 4.19–4.92) and from nine female healthy controls who had no gastrointestinal symptoms on screening with a gastrointestinal symptom questionnaire (age range, 19–54 yr; BMI range, 19.4–40.2 kg/m²) and with normal colonic transit (GC24: 1.15–2.63). All participants lived in the same geographic region within 120 miles of Rochester, MN. The samples of equal numbers of cases and controls were obtained from people of same sex and similar BMI range.

Sample storage.

Participants' rectosigmoid biopsies were preserved in a solution of RNALater, stored at −80°C. Total RNA was isolated using the RNeasy Plus Mini (Qiagen, Valencia, CA). Quality of the extracted RNA was assessed by the Mayo Gene Expression Core using Agilent Bioanalyzers. All samples had RNA Integrity Numbers (RIN) greater than 7.0.

In-depth RNA sequencing of colonic mucosa.

RNA sequencing was performed as paired-end 51 base reads on an Illumina HiSeq 2000 with three samples multiplexed per lane, using TruSeq SBS Sequencing Kit Version 3. Base calling was performed using Illumina's RTA version 1.12.4.2.

Bioinformatics analysis workflow.

Analysis of each sample (alignment statistics, in-depth quality control metrics, gene and exon expression levels, fusion transcripts, and single nucleotide variants) was done using Mayo Clinic's MAPRSeq v1.0, a computational workflow for analysis of RNA-Seq. Differential expression between samples from patients with IBS-D and healthy controls was computed using edgeR algorithm (32). EdgeR [or empirical analysis of digital gene expression in R (an open source programming environment)] is a bioconductor software package for examining differential expression of replicated count data; it involves determining whether counts for a transcript or exon are significantly different across experimental conditions. An important summary statistic is the number of reads in a class; for RNA-Seq, this read count has been found to be linearly related to the abundance of the target transcript (27). Further details of the analysis, including bioinformatics, are included in Supplemental Materials.

Comparison of expressions in rectosigmoid mucosa from IBS-D and healthy participants.

Gene expression variation was examined in the following genes, which have been associated with IBS in the literature based on genotype association studies (9; 39): 5-HTTLPR, TNFSF15, FAAH, CNR1, KLB, FGFR4, TGR5 (GPBAR1), PRDM1 (BLIMP1), TLR9, C11orf30, ORMDL3, NPSR1, and HLA2/8. The association of TNFSF15 with IBS symptom phenotype was significant with false detection rate (FDR) correction, and the strongest association was with constipation-predominant IBS (IBS-C) (45). Of these 13 genes, the association with colonic transit has been reported in 10, and the association of KLB was significant with FDR correction (43). In the published literature, no significant association with colonic transit was found with 5-HTTLPR, TNFSF15, or FGFR4 as individual genetic risk factors. However, FGFR4 significantly influenced the effect of KLB on colonic transit (43), and HLADQ2/8 significantly altered the effect of gluten on gut permeability (38) and was associated with accelerated small bowel transit in patients with IBS-D (37), in addition to the association with colonic transit (39). The second experimental approach assessed differential expression of any other gene mRNA, independent of any a priori hypothesis, that is an “unbiased” or hypothesis-generating analysis. An ancillary investigation used pathway analysis to assess the genes identified as differentially expressed by edgeR (adjusted P < 0.05).

mRNA expression using quantitative RT-PCR.

To confirm the results of the RNA-Seq study, we submitted the mRNA from these biopsies to quantitative RT-PCR using the Qiagen RT² assays, which use a Sybr Green detection method. Thus we evaluated 20 genes of interest, including 10 (identified in Fig. 2) of those with significant P values on RNA-Seq that are associated with functions related to pathobiology of IBS-D, as well as other genes related to tight junction proteins (CLDN1, OCLN, ZO-1), bile acid homeostasis, receptor or absorption [FGFR4, GPBAR1, FXR, IBAT/ASBT (SLC10A2)], serotonin transporter protein (SLC6A4), and immune modulation (TNFSF15). The latter genes were selected based on recent data acquired in other studies of IBS-D in our laboratory or in the literature (1; 3; 37; 38; 45).

Fig. 2.

Correlation (Spearman) analysis comparing the mRNA fold change (IBS-D over healthy controls) for 10 genes selected from the RNA-Seq analysis.

Briefly, total RNA was purified from human rectosigmoid tissue using the Qiagen RNAeasy Plus kit (Qiagen), including on-column DNAse treatment to remove genomic DNA. RNA quality was assessed by Agilent Bioanalyzer, and 0.5 μg RNA (RIN>7) was reverse transcribed using the RT² First Strand Kit (Qiagen). Resulting cDNA was amplified and gene expression quantified by Custom RT² PCR Array (Qiagen) using a 7300 Real-time PCR Machine (Applied Biosystems, Foster City, CA). In this custom profile, we included 20 candidate genes, two housekeeping genes for normalization, and three control genes to assure reaction quality (see Supplemental Materials, gene expression method by RT-PCR). Each assay utilizes a proprietary and experimentally verified algorithm for designing gene-specific qPCR primers with uniform PCR efficiency and amplification conditions. The assay was conducted as recommended by the supplier (Qiagen). For data analysis, the RT² Profiler PCR Array Data Analysis v3.5 software package was used. This package uses ^ΔΔCT-based fold change calculations and Student's t-test (two-tail, equal variance: http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php?target=upload) to assess the differences in mRNA fold changes between IBS-D and healthy control groups. In addition, we determined the Spearman correlation between by RNA-Seq and RT-PCR.

RESULTS

Sequence statistics based on sample analysis demonstrating quality control of the sequencing method.

The sequence statistics showed an average of 118,760,446 (min/max 99,967,896/150,273,460) total reads per sample was obtained, which is above the minimum of 50 million reads necessary for RNA sequencing analysis: 84.5% mapped to the genome and 12.9% to the splice-site junction reference sequence set; 2.6% did not map for various technical reasons. Thus extensive expression data were available from all cases and controls.

mRNA expression of genes previously associated with IBS-D: hypothesis-based analysis.

Gene expression variation was examined in the following genes, which have been associated with IBS in the literature: 5-HTTLPR, TNFSF15, FAAH, CNR1, KLB, FGFR4, TGR5 (GPBAR1), PRDM1(BLIMP1), TLR9, C11orf30, ORMDL3, NPSR1, and HLA2/8 (Table 2 shows gene expression variations in genes analyzed by DE Seq and edgeR).

Table 2.

Gene expression by edgeR analysis in healthy participants relative to IBS in the genes previously associated with IBS in the literature

Chr	Gene ID	Start	Stop	Coding Length	Fold Change	Log Fold Change	P Value
Neurotransmitter-related genes
chr1	FAAH	46859939	46879520	2095	0.874993	−0.1834	0.242634
chr6	CNR1	88849585	88875767	5955	0.837784	−0.23277	0.507953
chr17	SLC6A4	28523378	28562954	4533	1.016064	No data	No data
chr7	NPSR1	34697897	34917944	1824	0.338832	No data	No data
Genes related to immune function
chr3	TLR9	52255096	52260179	3870	0.924332	−0.10029	0.784644
chr9	TNFSF15	1.18E +08	117568408	6765	1.531795	0.691593	0.00985
Genes related to bile acid synthesis and receptor
chr4	KLB	39408473	39453153	6079	1.05282	0.074556	0.769264
chr5	FGFR4	1.77E +08	176525126	3143	0.964932	−0.08917	0.776776
chr2	GPBAR1	2.19E +08	219128582	2003	0.891095	−0.16474	0.61718
Genes related to membrane or barrier function
chr6	PRDM1	1.07E +08	106557814	5398	1.040026	0.09534	0.725122
chr11	C11orf30	76156069	76262589	5511	1.133884	0.208539	0.289325
chr17	ORMDL3	38077296	38083884	2139	0.977262	−0.03066	0.865844
Immunogenotype (histocompatibility locus)
chr6	HLA-DQA1	32605183	32611429	1542	0.910484	−0.13848	0.54487
chr6	HLA-DQA2	32709163	32714664	1152	1.176086	0.243684	0.59239
chr6	HLA-DQB1	32627241	32634466	1664	1.011884	0.086342	0.911969
chr6	HLA-DQB2	32723875	32731330	1122	1.219724	0.321712	0.296543

The significant difference in TNFSF15 is shown in boldface text, indicating higher fold change expression in health than in IBS-D.

There was decreased mRNA expression of TNFSF15 in rectosigmoid mucosa in nine patients with IBS-D compared with biopsies from nine healthy controls; the relative decrease in TNFSF15 expression was about 70% (Table 2). No other changes in expression of genes previously associated with IBS were found.

Differential expression of mRNA: unbiased hypothesis-generating analysis.

Differential expression analysis was run through edgeR algorithm, which identified 21 genes [13 upregulated, 8 downregulated in IBS-D relative to health (Fig. 1)] with an FDR-corrected P value of P < 0.05; the entire list of 21 genes with differentially expressed mRNA is shown in Table 3. The functions of these 21 genes are further described in Table 4. Many of the genes could be mapped to several biologically relevant pathways or mechanisms in IBS-D, as summarized here. Neurotransmitter functions were demonstrated when P2RY4 and VIP showed higher expression in IBS-D compared with healthy controls. Ion channel functions were upregulated in IBS-D relative to healthy controls; GUCA2B receptor associated with enterocyte chloride channel activation in response to uroguanylin and PDZD3 (also known as IKEPP, PDZK2, NHERF4); this PDZ protein associates with guanylate cyclase C and regulates cGMP production following receptor stimulation and mediates chloride secretion. Cytokines and complement were examined, and C4BPA complement cascade and CCL20 showed higher expression in IBS-D compared with healthy controls. Regarding immune function, TNFSF15 [P = 0.01 (Table 2)] and IFIT3 had lower expression in IBS-D relative to healthy controls. For mucosal repair and cell adhesion, TFF1 gastrointestinal trefoil protein and RBP2 retinol binding protein showed increased expression in IBS-D; conversely, FN1 fibronectin had lower expression in IBS-D relative to healthy controls.

Fig. 1.

Downregulated and upregulated mRNA expression in rectosigmoid mucosa in healthy controls compared with diarrhea-predominant irritable bowel syndrome (IBS-D). Gene expression fold change calculated by edgeR; negative numbers reflect lower mRNA expression of each gene of interest in patients with IBS-D compared with healthy controls.

Table 3.

Gene expression identified in hypothesis-free analysis by edgeR showing relative expression in IBS-D compared to health

Chr	GeneID	Coding Length	Fold Change	logFC	P Value	P (FDR)
Neurotransmitter-related genes
chrX	P2RY4	1639	5.80	2.53	7.03E-08	0.001094
chr6	VIP	1586	2.66	1.41	2.19E-05	0.02133
Ion-channel-related genes
chr1	GUCA2B	597	2.15	1.10	1.15E-05	0.017229
chr11	PDZD3	2363	2.19	1.13	3.47E-05	0.029047
Stress protein-related genes
chr6	HSPA1A	2429	0.54	−0.89	7.97E-07	0.006201
chr6	HSPA1B	2520	0.58	−0.80	1.69E-05	0.018718
Tissue repair, structural proteins, and barrier genes
chr21	TFF1	492	2.66	1.41	4.56E-06	0.011828
chr3	RBP2	688	3.17	1.66	1.03E-05	0.017229
chr15	WDR72	7309	0.27	−1.87	2.23E-06	0.008686
chr2	FN1	8975	0.54	−0.89	2.82E-06	0.008766
chr11	MMP1	2055	0.35	−1.50	0.000114	0.070962
chr1	ITLN1	1182	2.62	1.39	0.000183	0.098276
Glucose metabolism-related genes
chr2	IGFBP2	1429	2.40	1.26	1.46E-06	0.007553
chr9	ALDOB	2426	3.18	1.67	1.22E-05	0.017229
chr17	ALDOC	1651	2.09	1.06	1.41E-05	0.01827
Lipid transport-related genes
chr9	LCN15	762	3.35	1.74	2.61E-05	0.023844
chr9	ABCA1	10502	0.55	−0.87	7.62E-05	0.053883
Interferon-related genes
chr1	IFI44L	5874	0.23	−2.14	7.13E-06	0.015853
chr10	IFIT3	2688	0.31	−1.69	8.25E-06	0.016043
chr10	IFIT1	1860	0.36	−1.48	0.000173	0.098276
Chemokine-related genes
chr2	CCL20	842	2.70	1.43	1.70E-05	0.018718
chr1	C4BPA	2243	3.58	1.84	1.80E-05	0.018718
chr17	CCL3	796	0.31	−1.67	0.000196	0.098334
Immune function- and viral infection-related genes
chr21	MX1	3605	0.36	−1.48	4.67E-05	0.034633
chr21	MX2	2961	0.38	−1.38	0.000164	0.098144
chr12	OAS2	6169	0.41	−1.29	8.73E-05	0.059058
Apoptosis-related gene
chr17	BIRC5	2700	2.11	1.08	3.55E-05	0.029047
Energy metabolism-related gene
chr1	DUSP27	4034	0.19	−2.38	3.80E-05	0.029561

Bold P values are significant with false detection rate (FDR).

Table 4.

Summary of functions of 21 genes with significant differential expression in IBS-D rectosigmoid mucosa

Gene Symbol	Aliases and Descriptions	Gene ID	Summary of Gene Function or Product
Neurotransmitters
P2RY4	purinergic receptor P2Y	5030	G protein-coupled receptor responsive to uridine nucleotides, partially responsive to ATP, and not responsive to ADP
VIP	vasoactive intestinal peptide	7432	Encodes G protein-coupled receptor, relaxes smooth muscle, stimulates cAMP to produce intestinal secretion
Ion Channels
GUCA2B	guanylate cyclase activator 2B (uroguanylin)	2981	Gene encodes uroguanylin expressed in the stomach and intestine, functions as endogenous ligand for guanylate cyclase-C receptor increasing cyclic GMP, chloride, and water secretion
PDZD3	PDZ domain-containing protein 3; alias IKEPP; PDZK2; NHERF4	79849	Encodes protein-protein interaction molecules of structural proteins and the tight junction protein, zonula occludens 1; functions in sodium-hydrogen exchange (NHE3) as well as guanylyl cyclase C (GCC, or GUCY2C)-induced chloride and water secretion (through cGMP signaling that leads to cystic fibrosis transmembrane conductance regulator phosphorylation)
Mucosal repair, apoptosis, and cell adhesion
FN1	fibronectin 1	2335	Encodes fibronectin, a glycoprotein at the cell surface and in extracellular matrix, is involved in cell adhesion and migration processes including wound healing and host defense
TFF1	trefoil factor 1	7031	Gene on chromosome 21 for stable secretory proteins expressed in GI mucosa and likely protect the mucosa from insults, stabilizes mucus layer, and affects healing of epithelium.
RBP2	retinol-binding protein 2	5948	RBP2 is present in the small intestinal epithelium and involved in the uptake and/or intracellular metabolism of vitamin A, necessary for growth, differentiation of epithelial tissues
BIRC5	baculoviral IAP repeat-containing 5	332	Inhibitor of apoptosis (IAP) gene family, expression is high during fetal development, low in adult tissues.
Glucose metabolism
IGFBP2	insulin-like growth factor-binding protein 2	3485	Metabolism of proteins; regulation of insulin-like growth factor (IGF) transport and uptake by insulin-like growth factor binding proteins (IGFBPs)
ALDOB	aldolase B and C, fructose-bisphosphatases	229	Catalyzes the conversion of hexose (fructose-1,6-bisphosphate) to 2 trioses (glyceraldehyde 3-phosphate and dihydroxyacetone phosphate). Vertebrates have 3 aldolase isozymes A, B, and C
			Defects in ALDOB cause hereditary fructose intolerance.
ALDOC		230	Gene encodes aldolase expressed specifically in the hippocampus and Purkinje cells of the brain
Stress protein-related genes
HSPA1A	heat shock 70-kDa protein 1A	3303	Intronless genes located in the major histocompatibility complex class III region; encodes a 70-kDa heat shock protein that stabilizes existing proteins against aggregation and mediates the folding of newly translated proteins in the cytosol and in organelles
HSPA1B	heat shock 70-kDa protein 1B	3304
Chemokine-related genes
CCL20	chemokine (C-C motif) ligand 20	6364	Encodes gene for protein that provides key homing signals for cells of the adaptive immune system, both proinflammatory and suppressive T cell subsets
C4BPA	complement component 4-binding protein α	722	Gene on human chromosome 1 encodes C4b-binding protein that controls activation of the complement cascade through the classical pathway
Immune function, defense against infections
IFI44L	interferon-induced protein 44-like	10964	A type I interferon-stimulated gene that encodes a protein that has modest activity against hepatitis C virus
IFIT3	interferon-induced protein with tetra-tricopeptide repeats 3	3437	Genes involved in expression of INF-induced genes, hallmark of host antiviral immunity
MX1	myxovirus resistance protein 1	4599	Encodes interferon-inducible Mx protein responsible for protection against influenza virus infection and a member of both the dynamin family and the family of large GTPases
Absorptive mechanisms
LCN15	lipocalin 15	389812	Gene encodes proteins with high affinity to small lipophilic molecules
Other functions
DUSP27	dual specificity phosphatase 27	92235	Encodes protein with role in MAPK signaling cascades
WDR72	WD repeat-containing protein 72	256764	Encodes a protein with seven WD40 domains. WD or β-transducin repeats typically fold together to form a circular solenoid protein domain called the WD40 domain. Mutations in amelogenesis imperfecta type IIA3

Other genes appeared to be less specific to the pathobiology of IBS-D, and these included: 1) cellular intermediary carbohydrate metabolism [such as aldolase B and C genes ALDOB, ALDOC, and insulin growth factor binding protein (IGFBP2)] showed increased expression in IBS-D. 2) Cellular energy metabolism gene DUSP27 showed reduced expression in IBS-D compared with controls. 3) Lipid transport LCN15 is overexpressed in IBS-D relative to controls. 4) Protection against viral infection and stress-related proteins MX1, HSPA1A, and HSPA1B showed reduced expression in IBS-D compared with controls.

Comparison of RNA-Seq fold expression with results on RT-PCR.

Figure 2 demonstrates a comparison of fold changes between patients with IBS and healthy subjects in mRNA expression levels, as measured by RNA-Seq and RT Q-PCR. This includes 10 selected genes identified by RNA-Seq that are, at least theoretically, associated with functions related to pathobiology of IBS-D. In general, there was a significant Spearman correlation between fold change expression by the two techniques [Rs = 0.73, P = 0.013 (Fig. 2)].

The RT-PCR results mirror those of RNA-Seq in terms of the up- or downregulation of genes in IBS-D relative to healthy controls. In addition (Table 5), the RT-PCR identified significant upregulation (P < 0.05) of P2RY4 (also significant on RNA-Seq) and of C4BPA, borderline significant upregulation of GUCA2B and RBP2, and significant downregulation of FN1. Thus results of 5 of the 10 genes selected from RNA-Seq analysis were confirmed on RT-PCR.

Table 5.

Summary of mRNA expression of genes of interest using RT-PCR

Chr	GeneID	Coding Length	Fold Change	95% CI	P Value
GENES IDENTIFIED IN RNA-SEQ AS SIGNIFICANT OR BORDERLINE
Neurotransmitter-related genes
chrX	P2RY4	1639	2.714	(0.94, 4.49)	0.024
chr6	VIP	1586	0.949	(0.00001, 1.94)	0.433
Ion-channel function-related genes
chr1	GUCA2B	597	2.046	(0.60, 3.50)	0.061
chr11	PDZD3	2363	1.492	(0.72, 2.27)	0.217
Tissue repair, structural proteins, and barrier genes
chr21	TFF1	492	1.616	(0.41, 2.82)	0.514
chr3	RBP2	688	1.699	(0.58, 2.82)	0.056
chr2	FN1	8975	0.417	(0.21, 0.62)	0.012
Interferon-related genes
chr10	IFIT3	2688	0.447	(0.11, 0.78)	0.209
Chemokine-related genes
chr2	CCL20	842	0.654	(0.00001, 1.45)	0.637
chr1	C4BPA	2243	2.398	(0.19, 4.61)	0.040
OTHER GENES OF INTEREST MEASURED BY RT-PCR
Tight junction protein-related genes
3	CLDN1	3481	0.5123	(0.24, 0.78)	0.10
5	OCLN	6438	1.0341	(0.48, 1.59)	0.76
15	ZO-1	7950	0.7658	(0.45, 1.09)	0.35
Bile acid homeostasis-, receptor-, and absorption-related genes
5	FGFR4	3122	0.8182	(0.61, 1.03)	0.17
2	GPBAR1	2003	0.8386	(0.56, 1.12)	0.34
12	FXR/NH1R4	2884	1.7404	(0.94, 2.54)	0.038
13	IBAT/ASBT	3777	1.7337	(0.28, 3.19)	0.09
Serotonin transporter gene
17	SLC6A4	6543	0.9596	(0.20, 1.72)	0.87
Immune function-related gene
9	TNFSF15	6687	0.4718	(0.16, 0.79)	0.057

P values in italics reflect borderline significance, and those in bold reflect significance.

Moreover (Table 5), the RT-PCR demonstrated borderline decreased expressions of the tight junction protein claudin 1 and an immune modulator (TNFSF15). Conversely, there was significantly increased expression of the nuclear receptor, FXR, and borderline increased expression of apical sodium-coupled bile acid transporter (IBAT/ASBT). Table 6 shows that the altered mRNA expression of these genes was not detected on RNA-Seq.

Table 6.

mRNA expression using RNA-Seq of additional genes of interest identified by the RT-PCR

Chr	GeneID	Coding Length	logFC	P Value	P (FDR)
Additional tight junction protein-related genes
Chr 3	CLDN1	3452	0.60	0.081	1
Chr 5	OCLN	6576	−0.029	0.827	1
Chr 15	ZO-1	7164	0.068	0.793	1
Bile acid homeostasis-, receptor-, and absorption-related genes
Chr 5	FGFR4	3143	−0.089	0.777	1
Chr 2	GPBAR1	2003	−0.165	0.617	1
Chr 12	NH1R4	2589	0.947	0.011	0.609
chr13	IBAT/ASBT	No data

SLC6A4 and TNFSF15 were already documented in Table 2.

DISCUSSION

This is the first report of an RNA-Seq study of the expression of genes in patients with IBS-D compared with controls. This analysis builds on a prior analysis that we conducted several years ago, in which we examined the rectosigmoid mucosal expression of RNA from 30 patients with IBS conducted using Affimetrix Gene chips. The former study showed functional alterations of several components of the host mucosal immune response to microbial pathogens. The current study shows highly relevant alterations of mRNA expression that reflect alterations in functions that predispose to the pathophysiology of IBS-D.

Observations from mucosal RNA-Seq analysis.

In the section of the current study that was guided by an a priori hypothesis, patients with IBS-D had reduced expression of TNFSF15, which is involved in innate immune responses (15; 33). Our prior study had shown an association of TNFSF15 genotype, particularly with IBS-C (45). Thus, genotype TNFSF15 (G allele) was associated with IBS [P = 2.2 × 10⁻⁵; odds ratio (OR) 1.37; 95% CI 1.19 to 1.58] and was predominantly due to a significant association with IBS-C (P = 8.7 × 10⁻⁷; OR 1.79; 95% CI 1.41 to 2.26). We did observe that there was a correlation of rs4263839 genotype (GG genotype) with gene expression of TNFSF15 (mRNA levels), both in peripheral blood leucocytes and in rectal mucosal biopsies from healthy individuals (45). IBS-C (and G allele carriers) would have greater tissue expression of TNFSF15, and we hypothesize that IBS-D would be associated with reduced tissue expression of TNFSF15, consistent with our demonstration of reduced expression of TNFSF15 mRNA in patients with IBS-D with rapid colonic transit; further studies need to be conducted in the future to validate current findings.

In the unbiased analysis section of our study, we made several novel observations that led to testable hypotheses that are relevant to the pathobiology of IBS-D and that require confirmation in the future in validation cohorts. It is important to note that the increased or decreased expressions in tissues are typically >1.5 (Fig. 1). Peart et al. (30) and Raouf et al. (31) declared genes to be differentially expressed if they showed a fold change of at least 1.5 and also satisfied P < 0.05 after adjustment for multiple testing, as was used in the present analysis.

First, there was increased expression of two neurotransmitter ligand or receptor genes, VIP and P2RY4, that are involved in secretomotor and peristaltic reflexes, intrinsic sensory afferents and mucosal blood flow. Second, there was increased expression of two genes associated with ion channels or ion secretion, GUCA2B and PDZD3, and both control the enterocyte secretion of chloride ions through the effects of guanylate cyclase C receptor in response to endogenous ligands such as intestinal uroguanylin, which is secreted with mucins from goblet cells in normal physiological processes (13; 23). These two mechanisms may lead to increased fluid secretion and conceivably contribute to looser stool consistency in patients with IBS-D. The ileal epithelium has been shown to be more sensitive to the secretory effects of bile acids in patients with IBS-D (28). The relevance of genetic control of guanylate cyclase C receptor on the impact of clinical symptoms was clearly demonstrated in a report of an entire family diagnosed as having chronic functional diarrhea; in that family, there was a dominantly inherited, fully penetrant disease in 32 members that resulted from a missense mutation in GUCY2C (20).

A third group of overexpressed molecules, TFF1 and RBP2, is related to tissue repair or barrier function; such expression would be important to restore normal epithelial structure, for example, in the one-third of patients with IBS-D who have evidence of malabsorption of endogenous bile acids (34; 42) that have detergent effects resulting in increased mucosal permeability. On the other hand, in the same group of barrier-related genes, WDR72 and fibronectin 1 (FN1) are underexpressed in IBS-D relative to controls and may be responsible for the increased mucosal permeability and, possibly, the associated visceral hypersensitivity of IBS (10).

Fourth, the overexpression of a lipid transport-related gene, lipocalin 15 (LCN15), may also be a compensatory change to the slight increase in fecal fat observed in patients with IBS-D compared with controls and patients with IBS and constipation (34; 42). A fifth observation is that the spectrum of genes associated with interferons, responses to viruses (including myxoviruses), responses to stress (such as heat shock proteins), and innate immune responses (e.g., TNFSF15) was underexpressed in patients with IBS-D compared with controls. The underexpression of antimicrobial mechanisms, such as caspase 1 and lysozyme in mucosal biopsies previously demonstrated by Aerssens et al. (2), is consistent with reduced resistance to microbes. These characteristics may conceivably predispose to postinfectious IBS (35) or the pauci-inflammatory variant of IBS, which is most commonly observed in IBS-D relative to other groups (14). The current study did not replicate the prior observations in the study by Aerssens et al. (2), in which five genes were expressed at higher level in mucosa from patients with IBS, with fold changes in expression <1.53 for all genes. Of these five genes with increased expression in IBS, two genes are of unclear function, one was apparently unrelated to gut function (creatine kinase, brain), two reflect mucin production (MUC20), and one gene (VSIG2) is involved in myeloid cell functions. Hence, the current data have failed to confirm upregulation of mucin gene expression, and, possibly, one gene may be associated with immune activation. It is worth noting that, although all patients in the two studies were recruited and studied in our laboratory, the report by Aerssens et al. (2) included a heterogeneous group of 36 patients (15 with IBS-C and 21 with IBS-D), whereas the current study included nine patients selected by the most accelerated colonic transit in an attempt to have as homogeneous a group as possible for the RNA-Seq analysis. Therefore, the participants in the two studies had very different phenotypes, and, because there are diverse peripheral mechanisms leading to IBS (7), we hypothesize that these are associated with differential expression of genes or posttranslational modification in the mucosa in the different subgroups of IBS. Further prospective studies of the subgroup of IBS related to epidemiologically proven relation to enteric infection are indicated to ascertain whether the observed reductions in expression occur more frequently in postinfectious cases of IBS and to appraise the time course of the disordered expression.

Disorders in expression of factors protective against bacterial infection, such as cytokine and tissue repair overexpression in the current study, overexpression of mucin genes [MUC20 and V-set and immunoglobulin domain containing, 2 gene (VSIG2)] in the previous study by Aerssens et al. (2), as well as overexpression of Toll-like receptors demonstrated by Brint et al. (6), are all consistent with a role of microbiota and immune activation in the pathophysiology of IBS. Conversely, the underexpression of genes that protect the mucosa from luminal antigens or the inhibition of caspases that result in reduction in the apoptosis of immune cells may ultimately lead to the low-grade inflammation in IBS. Such reductions in gene expression identified in the Aerssens et al. study (2) were scavenger receptor cysteine-rich type 1 protein (M160), CD163 antigen-like 1, V-set and immunoglobulin domain-containing 4 (VSIG4), caspase 1, apoptosis-related cysteine peptidase (CASP1), IL-1β (convertase), neutrophil cytosolic factor 4 (NCF4), and lysozyme (LYZ). In the current study, downregulation of TNFSF15, IFIT3, MX1, HSPA1A, and HSPA1B is consistent with the same concept.

Observations from RT-PCR analysis of mRNA expression.

The RT-PCR analysis demonstrated numerical up- or downregulation of genes that were in the same direction as observed on the RNA-Seq analysis, and fold changes were confirmed by both methods for five genes (P2RY4, VIP, GUCA2B, C4BPA, and FN1); there was a significant correlation between the expressions by the two methods, suggesting that the RNA-Seq data are valid. In addition, the RT-PCR analysis identified significantly increased expression of FXR gene as well as interesting numerical fold changes (all P ≤ 0.10) in claudin 1, ASBT, and TNFSF15 genes. These findings are consistent with alterations in the mucosal expression of barrier, bile acid absorption, feedback regulation of synthesis, or immune activation. Importantly, ASBT expression is increased, and this is consistent with the literature that shows no evidence of mutation of the gene as a cause of bile acid malabsorption in patients with IBS (26). Conversely, the increased expression of the nuclear receptor, FXR, is consistent with the increased production of FGF-19 and feedback regulation of hepatic bile acid synthesis in patients with IBS-D. These interesting data provide hypotheses for replication in future studies. Figure 3 summarizes the overall findings in the current studies of mRNA expression. Pathway and ingenuity analyses provided confirmatory information on the relevance of these gastrointestinal mechanisms in IBS and identified common cellular and metabolic processes that may not be specifically related to the pathobiology of IBS.

Fig. 3.

Cartoon summarizing altered mRNA expression in cellular elements of the rectosigmoid mucosa in patients with IBS-D, based on the more stringent statistical analyses. TJ, tight junction; GC-C, guanylate cyclase; CFTR, cystic fibrosis transmembrane regulator; VIP, vasoactive intestinal peptide.

There are key limitations of this study. First, it is a pilot case-control study that requires replication. However, given the expense involved in such RNA-Seq studies, we perceive that this study provides the basis for planning future studies with regard to alterations in mucosal function, and it may allow greater focus on such mucosal mechanisms of secretion and barrier and immune functions in patients with IBS-D. Replication in a different cohort of patients with IBS-D and rapid colonic transit will inform researchers on the generalizability of the current findings. Regrettably, there are no other RNA-Seq studies in the current literature with which we can compare or contrast the results of the current study. However, the replication of several of the RNA-Seq fold changes by the RT-PCR data provides reassurance that the RNA-Seq data offer a rationale for future hypothesis-based studies that will be necessary to replicate these data. A second limitation is that we are unable to determine whether alterations of the transcriptome observed in the rectosigmoid mucosa of the patients with IBS-D are the cause or consequence of their condition. A third limitation is that the observations here pertain to a subgroup of patients with IBS-D and extremely fast colonic transit.

In conclusion, this pilot study demonstrates the potential utility of in-depth analysis, based on deep RNA-Seq of stored rectosigmoid mucosal biopsies, to develop greater understanding of the pathobiology of IBS-D and to help develop hypotheses for future studies. RNA-Seq and RT-PCR appear to be complementary in identifying changes in rectosigmoid mucosal mRNA expression of genes of interest. Rectosigmoid mucosa analysis by RNA-Seq and RT-PCR in IBS-D shows transcriptome changes affecting neurotransmitters, ion channels, cytokines, immune function, cell adhesion, or barrier function that provide the rationale for future validation studies to explore the role of mucosal factors in the pathobiology of IBS-D. Overall, these data continue to support the large body of evidence demonstrating the disturbances of peripheral functions in patients with IBS (7).

GRANTS

This study is supported by NIH RO1-DK92179.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: M.C. and G.F. conception and design of research; M.C., A.A.N., and E.W.K. analyzed data; M.C. and E.W.K. interpreted results of experiments; M.C. prepared figures; M.C. drafted manuscript; M.C., P.C., A.A., I.A.B., A.A.N., S.J.G., G.F., and E.W.K. edited and revised manuscript; M.C., P.C., A.A., and G.F. approved final version of manuscript; P.C., A.A., I.A.B., and S.J.G. performed experiments.