RNA-seq implicates deregulation of the immune system in the pathogenesis of diverticulitis

Abstract

Individuals with diverticula or outpouchings of the colonic mucosa and submucosa through the colonic wall have diverticulosis, which is usually asymptomatic. In 10-25% of individuals, the diverticula become inflamed, resulting in diverticulitis. Very little is known about the pathophysiology or gene regulatory pathways involved in the development of diverticulitis. To identify these pathways, we deep sequenced RNAs isolated from full-thickness sections of sigmoid colon from diverticulitis patients and control individuals. Specifically for diverticulitis cases, we analyzed tissue adjacent to areas affected by chronic disease. Since the tissue was collected during elective sigmoid resection, the disease was in a quiescent state. A comparison of differentially expressed genes found that gene ontology (GO) pathways associated with the immune response were upregulated in diverticulitis patients compared with nondiverticulosis controls. Next, weighted gene coexpression network analysis was performed to identify the interaction among coexpressed genes. This analysis revealed RASAL3, SASH3, PTPRC, and INPP5D as hub genes within the brown module eigengene, which highly correlated (r = 0.67, P = 0.0004) with diverticulitis. Additionally, we identified elevated expression of downstream interacting genes. In summary, transcripts associated with the immune response were upregulated in adjacent tissue from the sigmoid colons of chronic, recurrent diverticulitis patients. Further elucidating the genetic or epigenetic mechanisms associated with these alterations can help identify those at risk for chronic disease and may assist in clinical decision management.

NEW & NOTEWORTHY By using an unbiased approach to analyze transcripts expressed in unaffected colonic tissues adjacent to those affected by chronic diverticulitis, our study implicates that a defect in the immune response may be involved in the development of the disease. This finding expands on the current data that suggest the pathophysiology of diverticulitis is mediated by dietary, age, and obesity-related factors. Further characterizing the immunologic differences in diverticulitis may better inform clinical decision-making.

diverticular disease, encompassing both diverticulosis and diverticulitis, describes a spectrum of changes occurring most frequently within the sigmoid colon of adults in Western countries (10). This condition initiates as diverticula or outpouchings of the colonic mucosa and submucosa through the colonic wall where blood vessels penetrate the muscle layer (10, 30). This asymptomatic condition is known as diverticulosis, and its incidence increases with age. In approximately 10–25% of individuals with diverticulosis, inflammation of the diverticula leads to the development of diverticulitis (10). Complications associated with diverticular disease, including surgery, account for a significant healthcare burden in the United States and, overall, contributed to more healthcare visits than inflammatory bowel disease (IBD) in 2009 (22). Yet the pathobiology of this disease is very poorly understood. Which patients will follow a benign course vs. those who will develop diverticulitis is poorly predicted by clinical criteria. A better understanding of the pathogenesis of diverticulitis is required to uncover more effective disease management strategies.

Historically, a low-fiber diet was considered a causal factor in the development of diverticulosis (20), but this notion has since been refuted (23). Obesity is a well-established risk factor for diverticulitis (25, 29), and smokers have an increased risk of diverticular perforation/abscess relative to nonsmokers (9). Although both environmental factors and genetic susceptibility have been implicated in the risk for diverticulitis, little is known about the biological mechanisms associated with development of diverticulosis and, subsequently, diverticulitis. Genetics may play a major role in the disease phenotype with the estimated heritability of diverticular disease found to be 40-53% by two twin studies (5, 28). Although genome-wide association studies (GWAS) have the advantage of identifying at-risk alleles for complex diseases, the ability of GWAS to study the mechanism of disease pathogenesis is limited since many of the identified single nucleotide polymorphisms map to noncoding regions of the genome (35). More recently, examining the overlap of GWAS with expression quantitative trait loci (eQTL) has helped understand how single nucleotide polymorphisms located within a regulatory element affect gene expression (19). Unfortunately, studies examining the genetic contribution to diverticular disease are sparse and GWAS has yet to be reported.

We conducted RNA-seq analysis on transcripts isolated from full-thickness adjacent sigmoid colon tissues from individuals with chronic, recurrent diverticulitis during disease quiescence (n = 20) and nondiverticulosis controls (n = 5) to identify pathways associated with disease. We found 1,381 differentially expressed genes, of which 314 had a |fold change| >1.60. Gene ontology (GO) pathway analysis identified that a majority of these genes were associated with upregulation of the immune response in diverticulitis patients relative to controls. Network analysis found that highly connected hub genes included the immune regulators: RAS protein activator-like 3 (RASAL3), SAM and SH3 domain-containing 3 (SASH3), protein tyrosine phosphatase, receptor type C (PTPRC), and inositol polyphosphate-5-phosphatase D (INPP5D). Overall, these data suggest that immunoregulatory defects may be involved in the pathogenesis of diverticulitis.

MATERIALS AND METHODS

Study design and specimen collection.

This was a retrospective cohort study performed at the Pennsylvania State University College of Medicine with Institutional Review Board approval. Chronic, recurrent diverticulitis patients gave consent between April 2010 and August 2014. At the time of elective sigmoid resection, colonic tissue was collected into the Penn State Hershey Colon and Rectal Diseases Biobank. Diverticulitis was confirmed by preoperative computed tomography scans and histopathologic evaluation of surgical specimens. Immediately after resection, surgical tissues were transported from the operating room to the surgical pathology laboratory where several full-thickness segments of tissue were obtained and stored in RNAlater (Invitrogen, Carlsbad, CA). Adjacent tissue was obtained 5-10 cm away from areas of chronically diseased tissue. Diverticulitis patients with a concurrent diagnosis of IBD or other inflammatory conditions, recent use of immunosuppressives, cancer, dysplasia, or dysmotility disorders were excluded. Controls were identified from patients undergoing sigmoid resection without evidence of diverticulosis for the following conditions: endometriosis, rectal prolapse, slow transit, and sigmoid volvulus.

RNA isolation.

Approximately 250 mg of tissue stored in RNAlater were pulverized in a MultiSample BioPulverizer (Biospecs Products, Bartlesville, OK) atop a bath of liquid nitrogen. RNA was first isolated using TRIzol (Ambion, Waltham, MA) and chloroform. Following centrifugation, the aqueous phase was subsequently purified using an RNeasy Mini Kit (Qiagen, Valencia, CA) following the manufacturer’s instructions. Total RNA was analyzed using the Agilent Bioanalyzer 2100 and samples had an RNA integrity number ≥7.

RNA-seq.

The cDNA libraries were prepared using the SureSelect Strand Specific RNA Library Preparation Kit (Agilent Technologies, Santa Clara, CA) as per the manufacturer’s instructions. RNA-sequencing was performed as previously described (24). The Illumina CASAVA pipeline v1.8 was used to extract the demultiplexed sequencing reads. FastQC (v0.11.2) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) was used to validate the quality of the raw sequence data. Additional quality filtering used FASTX-Toolkit (http://hannonlab.cshl.edu/fastx_toolkit) set at a quality score cutoff of 20. Next, the filtered reads were aligned to the human reference genome (build hg38) using Tophat (v2.0.9) (31) allowing two mismatches. Read counts were calculated using HTSeq (1) as provided with the Ensembl gene annotation package (release 78). RPKM (Reads per Kilobase per Million mapped reads) values were calculated after applying GC-content and quantile normalizations using an R package “cqn” (The R Project for Statistical Computing, Vienna, Austria). Volcano plots were constructed using R. Gene Ontology (GO) pathway analysis was performed using the Generally Applicable Gene Set Enrichment (GAGE) pathway analysis package for R (15).

Reverse transcription and quantitative PCR.

cDNA synthesis was performed from 100 ng of RNA using the SuperScript III First-Strand Synthesis kit (ThermoFisher Scientific, Leesport, PA) according to the manufacturer’s instructions. RASAL3, PTPRC, INPP5D, and SASH3 gene expression levels were assessed by reverse transcription and quantitative PCR (RT-qPCR). The following primers were used to detect RASAL3 (forward: TGGCTATCTCTCTGCTCCCAGACC; reverse: CCTCAAAGCACCCTGCCTCAAT), PTPRC (forward: GCCAGCACCTACCCTGCTCAGAAT; reverse: GGGGCACCAAGTGGATTAACACAA), INPP5D (forward: CTCCTCCTGCCCAGCTTCCTATG; reverse: TTTCTTCCAGCCTCAGCACTTGGT), and SASH3 (forward: TACTCAATGGCAAGGTGGGCTCTT; reverse: AGCTCATGCAGGGTCTTAGGCTTG). cDNA was diluted to 10 ng/µl for qPCR in reactions containing SensiFAST SYBR & Fluorescein master mix (Bioline, Taunton, MA) as previously described (24). Thermocycling conditions were 94°C for 3 min, 45 cycles of 94°C for 10 s, and 68°C for 40 s. TUBB3 (β-tubulin 3) served as an internal reference gene, and its expression was measured using the primers (forward: ACAACGAGGCGCTCTACGACATCT; reverse: AAGGAGGTGGTGACTCCGCTCA) under the same thermocycling conditions. Values were normalized to TUBB3 and fold change was calculated using the formula 2^−ΔCt. The data were subsequently log transformed, and a two-tailed t-test was performed.

Network analysis.

Network analysis was performed from log2 RPKM values using the weighted gene coexpression network analysis (WGCNA) package in R (13). Student’s t-test was performed on log2 RPKM values to identify a set of genes with P < 0.05 to be subjected to WGCNA (1,381 genes). The dataset was evaluated for outliers, and one diverticulitis patient was subsequently removed from the analysis. Following the standard WGCNA protocol, with the use of a soft thresholding power of 14, clustering was first performed using topological overlap matrix (TOM)-based dissimilarity followed by a clustering of consensus module eigengenes (MEs). This analysis resulted in multiple MEs comprised of genes that are highly coexpressed. The grey ME consisted of genes that were unassigned within an ME and were not further evaluated. Pearson correlation was performed to identify the relationship between MEs and clinical traits. The final data were exported into Cytoscape v3.4.0 (27) to visualize the network between the top weighted gene interactions for each module. GO pathway analysis was performed for each module using the “GAGE” package for R (15).

Statistical analysis.

Statistical analyses were performed using R software v3.3.3. Two-tailed statistical tests were determined to be significant at P ≤ 0.05. For demographics and clinical indexes, two-tailed χ²-tests were used to compare categorical variables while two-tailed t-tests were used to compare continuous variables. For the RNA-seq analysis, q values were obtained by the false discovery rate method (2) and were considered significant if q ≤ 0.05.

RESULTS

Immune-associated differentially expressed genes distinguish diverticulitis patients and controls.

Transcriptomic profiling is an unbiased approach that has been used effectively to identify deregulated signaling pathways in intestinal diseases, such as colorectal cancer (17). We performed RNA-seq analysis on full-thickness sigmoid colon tissue obtained 5–10 cm away from areas of diseased tissue to identify pathways that could be involved in the development of disease while avoiding the inflammation associated with diverticulitis. We analyzed the transcriptome of diverticulitis patients (n = 20) and nondiverticulosis controls (n = 5) (Table 1). In total, 1,381 genes were differentially expressed (P < 0.05), of which 314 genes displayed a |fold change| >1.60 (Fig. 1A). Several genes displaying the greatest difference in expression were involved in immune system processes, including T-box 21 (TBX21), B lymphoid tyrosine kinase (BLK), C-C motif chemokine receptor 7 (CCR7), and T-cell leukemia/lymphoma 1A (TCL1A).

Table 1.

Demographics and clinical information

	Diverticulitis, n (%)	Controls, n (%)	P Value
Total n	20	5
Sex
Male	10 (50.0%)	2 (40.0%)	0.6889
Female	10 (50.0%)	3 (60.0%)
Race
Caucasian	20 (100%)	7 (100%)	1.0000
Age at surgery, yr, means ± SD (range)	57.3 ± 14.2 (36.5–78.0)	57.1 ± 10.6 (43.9–71.0)	0.9769
Body mass index, kg/m2, means ± SD (range)	26.8 ± 4.8 (20.0–42.0)	28.2 ± 7.6 (19.0–40.0)	0.6085
Smoking status
Positive history	13 (65.0%)	1 (20.0%)	0.1904
Former smoker*	9 (40.0%)	1 (20.0%)
Current smoker	5 (25.0%)	0 (0%)
Negative history	7 (35.0%)	4 (80.0%)

SD, standard deviation. Continuous variables analyzed by two-tailed t-test and categorical variables analyzed by two-tailed χ²-test.

^*Smoking cessation of ≥1 year before surgery, mean time between smoking cessation and surgery 18.1 ± 14.6 yr (range 1–43 yr).

Fig. 1.

Immune-associated genes are upregulated in diverticulitis patients. A: volcano plot of differentially expressed genes in diverticulitis (n = 20) and control (n = 5) sigmoid colon tissues with each gene represented by a point. Red represents genes with P < 0.05. Orange represents genes with |fold change| >1.60. Green represents genes with P < 0.05 and |fold change| >1.60. B: gene ontology categories of upregulated genes. C: gene ontology categories of downregulated genes.

To uncover pathways represented amongst the 314 differentially expressed genes, we subjected the list to GO pathway analysis. Of the significantly upregulated pathways (q ≤ 0.05), we highlighted three categories of pathways that were most prominently featured in our analysis, including those involved in the immune response, apoptosis and cell death, and cellular processes (Fig. 1B). Over 50% of the identified pathways were classified under the immune response with both the innate and adaptive immune systems represented. In addition, we identified significantly downregulated pathways in diverticulitis patients compared with controls (Fig. 1C). These data were also categorized into three groups: nervous system, cellular processes, and muscle and tissue system, with similar representation in each category.

To determine whether differentially expressed genes could segregate diverticulitis patients from controls, we subjected the immune response pathway, consisting of 55 genes, to hierarchical clustering analysis (Fig. 2). Overall, the diverticulitis patients segregated from the controls, with the exception of one patient. Thus differences in the immune-associated transcriptome delineate diverticulitis and from controls.

Fig. 2.

Hierarchical clustering analysis of genes comprising the immune system pathway segregates diverticulitis patients and controls. Heat map depicting 55 genes clustered by diverticulitis patients (yellow bar) and controls (black bar). The color key indicates log2 RPKM values.

Immune-associated module eigengenes correlate with the presence of diverticulitis.

To define the interactive network of coexpressed genes, WGCNA was performed on the 1,381 genes that were differentially expressed between diverticulitis and controls (P < 0.05). These genes were designated into five MEs based on coexpression (Fig. 3A). The grey ME includes genes that were not assigned to another ME and thus were not analyzed further. GO pathway analysis was performed to identify the pathway contribution for each ME. Similar to our previous GO analysis, immune-associated genes comprised the brown, yellow, and turquoise MEs. The blue ME included genes involved in the transcription process. Pearson correlation (r) was performed for each ME and clinical traits of interest (Fig. 3A). The brown ME was positively correlated with diverticulitis (r = 0.67, P = 0.0004) and smoking status (P = 0.01). Additionally, the turquoise ME was trending towards significance with age at surgery (P = 0.08).

Fig. 3.

Immune-associated module eigengenes (MEs) correlate with the presence of diverticulitis. A: Pearson correlation was performed on MEs to evaluate the association between expression and clinical traits: diagnosis of diverticulitis, sex, smoking history (positive or negative history), body mass index (BMI), and age at surgery (SxAge). The heat map is color-coded by correlation coefficient identified by the color key with the P value listed in parenthesis. B: Weighted gene coexpression network analysis identified gene coexpression networks for each ME, with the top interactions (weight ≥0.20) illustrated from the brown ME. Fold change gene expression of diverticulitis patients (n = 20) relative to nondiverticulosis controls (n = 5) was overlaid on the pathway as designated by the color key. C: quantitative RT-PCR from an independent cohort of diverticulitis patients (n = 9) and controls (n = 9) demonstrating log2 relative expression of RASAL3, PTPRC, INPP5D, and SASH3. Boxplots display median and interquartile range. *P < 0.05, **P < 0.01.

Network interaction analysis identifies upregulation of hub genes and downstream genes in patients with diverticulitis.

An undirected network was constructed from the top weighted interactions for each module to identify hub genes and potential downstream effects. The brown ME is presented as it was highly correlated with diverticulitis (Fig. 3B). We identified four hub genes, or genes with high connectivity, in the brown ME: RASAL3, SASH3, PTPRC, and INPP5D (more commonly known as SHIP1). For all four hub genes, gene expression in the diverticulitis cohort was higher relative to nondiverticulosis controls. Additionally, higher expression of genes downstream of the hub genes was also seen in the diverticulitis patients compared with controls. Biological replication of hub genes was performed in an independent cohort of diverticulitis patients (n = 9) and controls (n = 9), which confirmed upregulated expression of these targets (Fig. 3C). Overall, these data suggest that these hub genes may be involved in regulating the immune response and that this signaling network is deregulated in diverticulitis patients.

DISCUSSION

Previously, our group uncovered a genetic association between tumor necrosis factor superfamily member 15 (TNFSF15) and surgical diverticulitis (4), suggesting that intrinsic defects in the host immune response may be involved in diverticulitis pathogenesis. To evaluate in a nonbiased fashion the potential immunologic involvement and the global transcriptome contribution to diverticulitis, we performed RNA-seq on colonic tissues obtained from diverticulitis and control individuals. We identified an enrichment of pathways involved in the immune response, including both the innate and adaptive immune systems in patients with diverticulitis. These data were unexpected since the tissue used for this study was collected from patients during disease quiescence, away from an area of disease involvement. Identification of deregulated highly interactive hub genes associated with immunoregulation provides evidence to suggest that defects in the host immune response may be involved in the pathogenesis of diverticulitis.

Association of MEs with clinical traits identified various potential associations with smoking status and patient age at surgery. We found that the brown ME also correlated with smoking status. This result is not surprising since smoking is an environmental risk factor for diverticulitis (32), with one predicted mechanism being deregulation of the immune system (21, 33). There are currently no data indicating how smoking cessation influences the colonic immune system. Therefore, we included both former and current smokers in our analysis, which might be contributing to the significance of this ME. However, we also found a potential association with age at surgery and the turquoise ME. This ME includes immune-associated genes and those involved in the protein kinase cascade. Although diverticulitis is commonly associated with the older population, a small percentage (<5%) develop this disease at a younger age (<40 yr old) (34). More robust studies are needed that compare diverticulitis patients diagnosed before 40 yr old to age-matched controls. This analysis would be imperative to confirm this association and to identify the transcriptomic contribution to age-related diagnosis of diverticulitis.

Network interaction of the brown ME identified four immune-associated hub genes: RASAL3, SASH3, PTPRC, and INPP5D. Neither RASAL3 nor SASH3 have been studied in inflammatory diseases, although RASAL3 regulates natural killer T-cell expansion and function in mice (26). INPP5D, also known as SHIP1, is a leukocyte-specific molecule that negatively regulates phosphatidylinositol 3-kinase (PI3K) activation. SHIP1 dephosphorylates phosphatidylinositol-3,4,5-triphosphate [PtdIns(3,4,5)P₃] to phosphatidylinositol-3,4-biphosphate [PtdIns(3,4)P₂] (11, 14). SHIP-deficient mice are prone to a Crohn’s disease (CD)-like mucosal inflammation (12). Similarly, reduced levels of INPP5D/SHIP1 were found from intestinal biopsies from CD patients (18). This is contrary to the results seen in our study, which demonstrate increased INPP5D/SHIP1 gene expression in diverticulitis patients (fold change: 1.50), and thus warrants further investigation to understand the potential differences in disease processes. Increased expression of INPP5D/SHIP1 may result in enhanced dephosphorylation of PtdIns(3,4,5)P₃, disrupting various signaling pathways, including the Akt pathway, which is involved in apoptosis (6). In fact, overexpression of INPP5D/SHIP1 induced caspase-3 and -9 activity in vitro (3). Although we did not detect significant upregulation of CASP3 (fold change: 1.11) or CASP9 (fold change: 1.18), we identified upregulation of apoptosis pathways in diverticulitis patients relative to controls (Fig. 1B). Overall, we describe an upregulation of INPP5D/SHIP and the precise role for this gene in diverticulitis requires further investigation.

PTPRC, also known as CD45, is an abundant cell surface glycoprotein involved in protein tyrosine phosphorylation (8). The primary substrates for PTPRC are the Src-family tyrosine kinases (SFKs), which are involved in initiation of leukocyte-specific immune responses (8). A previous report suggested that upregulation of PTPRC results in high levels of inflammatory cytokines (7). A similar mechanism was described in our study. Diverticulitis patients demonstrated higher levels of PTPRC transcripts relative to controls (fold change: 1.52). Upregulation of the PTPRC pathway may contribute to the chronic nature of disease in these individuals, increasing the risk for recurrent diverticulitis.

In summary, we used RNA-seq analysis on full thickness sigmoid colon tissues to identify molecular pathway that differed between diverticulitis patients and controls. Previously, the pathophysiology of diverticulitis was thought to involve diet, age, and obesity (16). The findings from the current study provide additional, nonbiased evidence that implicates a deregulation of the immune system in the pathogenesis of diverticulitis. Further characterizing these specific immunologic differences may allow the development of objective biologic criteria that could eventually help in clinical decision-making.

GRANTS

This publication was supported by the Carlino Fund for IBD Research and in part by the National Center for Advancing Translational Sciences Grants UL1 TR000127 and TL1 TR000125 (to K. M. Schieffer).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

K.M.S., G.S.Y., and W.A.K. conceived and designed research; K.M.S., S.E., D.M.K., A.S., and Y.I.K. performed experiments; K.M.S., D.M.K., A.S., Y.I.K., and G.S.Y. analyzed data; K.M.S., D.M.K., A.S., Y.I.K., G.S.Y., and W.A.K. interpreted results of experiments; K.M.S. prepared figures; K.M.S. and G.S.Y. drafted manuscript; K.M.S., C.S.C., L.H., S.D., D.M.K., A.S., Y.I.K., G.S.Y., and W.A.K. edited and revised manuscript; K.M.S., C.S.C., L.H., S.D., D.M.K., A.S., Y.I.K., G.S.Y., and W.A.K. approved final version of manuscript.