Colonoids From Patients With Pediatric Inflammatory Bowel Disease Exhibit Decreased Growth Associated With Inflammation Severity and Durable Upregulation of Antigen Presentation Genes

Abstract

Background

Defining epithelial cell contributions to inflammatory bowel disease (IBD) is essential for the development of much needed therapies for barrier repair. Children with very early onset (VEO)-IBD have more extensive, severe, and refractory disease than older children and adults with IBD and, in some cases, have defective barrier function. We therefore evaluated functional and transcriptomic differences between pediatric IBD (VEO and older onset) and non-IBD epithelium using 3-dimensional, biopsy-derived organoids.

Methods

We measured growth efficiency relative to histopathological and clinical parameters in patient enteroid (ileum) and colonoid (colon) lines. We performed RNA-sequencing on patient colonoids and subsequent flow cytometry after multiple passages to evaluate changes that persisted in culture.

Results

Enteroids and colonoids from pediatric patients with IBD exhibited decreased growth associated with histological inflammation compared with non-IBD controls. We observed increased LYZ expression in colonoids from pediatric IBD patients, which has been reported previously in adult patients with IBD. We also observed upregulation of antigen presentation genes HLA-DRB1 and HLA-DRA, which persisted after prolonged passaging in patients with pediatric IBD.

Conclusions

We present the first functional evaluation of enteroids and colonoids from patients with VEO-IBD and older onset pediatric IBD, a subset of which exhibits poor growth. Enhanced, persistent epithelial antigen presentation gene expression in patient colonoids supports the notion that epithelial cell-intrinsic differences may contribute to IBD pathogenesis.

Graphical Abstract

epithelial cells, inflammatory bowel disease, VEO-IBD, enteroids, colonoids

INTRODUCTION

Patient-derived enteroids (from small intestine) and colonoids (from colon) are generated directly from adult intestinal and colonic epithelial stem cells and are also referred to as “intestinal adult stem cells” or ASCs.^{1, 2} Distinct from pluripotent stem cells, enteroids and colonoids represent an important tool to understand phenotypic and functional aberrations in patients with inflammatory bowel disease (IBD) because they retain many of the transcriptional and epigenetic signatures of the tissue from which they were derived.^3–6 Functional analyses of enteroids and colonoids directly from affected patient populations, in conjunction with gene expression and genomic analyses, will permit a greater understanding of etiopathogenesis^{7, 8} and provide a foundation for development of new therapies.

Studies evaluating ASC-derived enteroids and colonoids from adult patients with IBD reveal changes in epithelial gene expression that could underlie poor healing of the epithelial barrier.^3–6 Indeed, persistent gene expression changes have been described in patients in remission,⁹ and confirmation of this finding in patient-derived enteroids reveals pathways and mechanisms, such as stem cell changes, that may not have otherwise been explored.³ Though studies in adult patients provide new and important information that could inform treatment decisions or development of new therapies for adults with IBD, few studies to date focus upon evaluating enteroids and colonoids in pediatric patients with IBD.

The current study addressed 2 unresolved questions using patient-derived enteroids/colonoids to understand IBD pathogenesis. First, we addressed the feasibility of establishing enteroid and colonoid lines from pediatric patients with older onset IBD (age 6–18 years) and very early onset IBD (VEO-IBD), diagnosed younger than 6 years of age. Based upon published studies in adult patients with ulcerative colitis, we hypothesized that pediatric patients with IBD (both VEO and older onset) with active disease would establish at a similar rate to patients with inactive disease and controls.⁷ Second, we evaluated 2 transcriptional aspects of pediatric patient-derived colonoids: (1) the durability in colonoid culture of transcriptional differences between patients with IBD and controls, as was published previously in adults with IBD,³ and (2) differences in stem cell and other gene signatures observed previously by others in epithelium of adults with IBD.^{3, 4, 7} Results herein suggest that alterations in the epithelial compartment in children with IBD can be maintained ex vivo and implicate genes and pathways that may contribute to disease pathogenesis.

MATERIALS AND METHODS

Subject Enrollment and Demographics

The institutional review board at the Children’s Hospital of Philadelphia (CHOP) approved the protocol (2014–010826), and all parents of patients provided written informed consent. Patients were recruited from the Center for Pediatric IBD at CHOP between September 2015 and October 2018. Study groups included (1) healthy control patients less than 18 years of age who were evaluated for gastrointestinal complaints but found to have normal endoscopic and histologic findings, (2) patients with VEO-IBD, including IBD diagnosed at 6 years of age or younger, and (3) older onset pediatric IBD, defined as children diagnosed with IBD at older than 6 years of age (but younger than 18). Inclusion criteria for IBD groups included patients with confirmed diagnosis of IBD by standard methods of endoscopic, radiologic, laboratory and clinical evaluation. Exclusion criteria included patients with a prior diagnosis of other intestinal or systemic inflammatory disease (including chronic allergic or inflammatory diseases). Indications for colonoscopy in VEO-IBD and pediatric IBD included diagnosis and disease surveillance. All patients with VEO-IBD underwent immunologic and genetic evaluation. Genetic studies were performed via whole exome sequencing and included trio analyses. Control patients were age-matched to disease patients and underwent colonoscopy for the following reasons: abdominal pain, poor growth, rectal bleeding, or diarrhea.

Clinical Data

Clinical information was obtained from the electronic medical records of the study population. Patients were categorized into 3 groups based on disease status and age at diagnosis, as described previously. The following variables were obtained: date of birth, sex, date of IBD diagnosis, age at time of diagnosis, IBD diagnosis (Crohn’s disease [CD], ulcerative colitis [UC], IBD-unclassified [IBD-U]), disease phenotype by Paris classification,¹⁰ macroscopic location of disease, diagnostic gastrointestinal pathology results, IBD-related medication history, and IBD-related surgical history as outlined in Table 1. Crohn’s disease was diagnosed when any of the following features were present in the diagnostic evaluation: endoscopic skip lesions or ileal inflammation in the presence of a normal cecum, histologic evidence of epithelioid granulomas or chronic ileitis, radiologic evidence of thickened small bowel loops, or evidence of intestinal fistulizing disease or perianal disease. Ulcerative colitis was diagnosed in cases of diffuse continuous mucosal ulceration of varying severity extending through the colon proximally from the rectum. Inflammatory bowel disease–unclassified was diagnosed in cases consistent with ulcerative colitis but with at least one of the following additional features: rectal sparing, macroscopic duodenal or esophageal ulcers without comorbidities, or numerous gastric aphthous lesions without additional etiology. Clinical disease severity was assigned by physician global assessment at the time of each colonoscopy. Colonoscopy evaluation was scored by simplified endoscopic score for Crohn’s disease and Mayo Clinic Endoscopic Subscore for ulcerative colitis for patients with IBD-U.^{10, 11} Therapies at the time of colonoscopy were categorized as none for newly diagnosed patients, immunosuppressive naïve, which included oral/rectal 5-ASA, enteral nutrition therapy, and antibiotics; and immunosuppressive therapies included biologics, budesonide, and immunomodulators.

TABLE 1.

Patient Demographics

	VEO	Older Onset	Control
Patients, n	22	16	15
Male, n (%)	17 (77)	8 (50)	11 (73)
Age at dx, mean (±SD), years	2.8(±1.8)	10.6 (±3.3)	NA
Age at sample collection, mean(±SD), years	7.1(±4.4)	13.2 (±4.3)	5.0 (±5.3)
Diagnosis, n (%)
CD	11 (50)	9 (56)	NA
UC	1 (5)	4 (25)	NA
IBDU	10 (45)	3 (19)	NA
Disease Location, n (%)
Colonic only	16 (73)	9 (56)	NA
Ileocolonic	6 (27)	5 (31)	NA
Ileal Only	0 (0)	2 (13)	NA
Perianal	5 (23)	3 (19)	NA
Prior Surgery, n (%)
Colectomy	3 (14)	0 (0)	NA
Ostomy	8 (36)	0 (0)	NA
Current Ostomy	7 (88)	0 (0)	NA
Prior Ostomy	1 (12)	0 (0)	NA
Total No. Colonoscopies/Biopsies	26	16	19
Therapy at Time of Biopsy, n (%)
None	7 (27)	6 (38)	NA
Immunosuppressive Naïve	6 (23)	3 (19)	NA
Biologic Only	4 (15)	4 (25)	NA
Immunomodulator Only	1 (4)	0 (0)	NA
Biologic + Immunomodulator	4 (15)	1 (6)	NA
Budesonide Only	0 (0)	1 (6)	NA
Biologic + Budesonide	1 (4)	1 (6)	NA
s/p Hematopoietic Stem Cell Transplant	3 (12)	0 (0)	NA
Physician Global Assessment at Colonoscopy, n (%)
Quiescent	11 (42)	6 (38)	NA
Mild	5 (19)	8 (50)	NA
Moderate	10 (38)	2 (12)	NA
Severe	0 (0)	0 (0)	NA
SES-CD Score, n (%)
Quiescent	2 (17)	5 (56)	NA
Mild	2 (17)	2 (22)	NA
Moderate	2 (17)	2 (22)	NA
Severe	6 (50)	0 (0)	NA
Mayo Endoscopic Subscore for UC/IBD-U, n (%)
Quiescent	2 (14)	1 (14)	NA
Mild	2 (14)	2 (29)	NA
Moderate	4 (29)	2 (29)	NA
Severe	6 (43)	2 (29)	NA

Generation of Enteroids and Colonoids from Patient Biopsies

Two mucosal biopsies per site from endoscopically uninvolved areas of terminal ileum and/or colon were obtained from patients during colonoscopy performed for disease surveillance or for diagnosis and transported in cold Dulbecco’s Modified Eagle’s Medium (DMEM) (Corning, Manassas, VA, USA) immediately to the laboratory for crypt isolation. Depending on the extent of evaluation performed at the time of endoscopy, the biopsy set consisted of specimens that were collected from the terminal ileum, right colon, and left colon. Supplemental Table 2 lists all analyses performed on specific patient samples. All crypt isolations and cultures were completed within 2 hours of biopsy according to published protocols.^12–15 Briefly, biopsies were washed in cold Phosphate-Buffered Saline (PBS) (Invitrogen, Carlsbad, CA) and placed in cold ethylenediaminetetraacetic acid chelation buffer for 30 minutes followed by mechanical dissociation (via scraping) of crypts. Filtered crypts were resuspended in 30 to 50 μL Matrigel (Corning) based upon pellet size and crypts visualized after plating to ensure optimal density. We determined that at least 50 crypt units per biopsy region were considered a successful isolation, and Matrigel plating volumes were adjusted to ensure that crypts were plated at similar densities to mitigate density growth bias. Colonoids were grown in human IntestiCult Organoid Growth Media (Stemcell Technologies, Vancouver, BC, Canada). Imaging of live enteroids was performed on days 1 to 7 to monitor plating efficiency and growth. Enteroids and colonoids were passaged at days 5 to 8 depending on culture growth and density. Mechanical passaging was performed as described previously,¹⁶ with the following modification: cold Matrigel suspension was pipetted up and down 5 times with a 5-milliliter pipette fitted with a p200 pipette tip. This was done in a 15-milliliter conical tube in a total of 5 milliliters of cold DMEM/F12 (Invitrogen). We evaluated cultures 24 hours after plating to determine the percentage of isolated crypts that grew into enteroids or colonoids (crypt to spheroid conversion). At least 50% of isolated crypts from control patients converted to spheroids within the first 24 hours, which would continue to grow robustly on subsequent days. We therefore generated a scoring system whereby a culture received a 0 if less than 50% of crypts converted to spheroids at 24 hours postplating, designated as “poor” growth, and a 1 if greater than or equal to 50% of crypts converted to spheroids, designated as “normal” growth.

Histology Scoring

Hematoxylin and eosin (H&E)-stained slides and accompanying pathology reports were evaluated by author PAR, with scoring modified from previously described protocols.¹⁷ Briefly, scoring is described as follows: inflammation, 0; absent neutrophils, 1; mild (neutrophils limited to the lamina propria or isolated cryptitis), 2; moderate (presence of microabscesses), 3; severe (diffuse colitis with ulceration or granulation tissue), 3.

For apoptosis, scoring is described as follows: 0, absent; 1, present, not increased (<1 apoptotic figure per 10 crypts); 2, present, increased (>1 apoptotic figure per 10 crypts); For cryptitis, scoring is described as follows: 0, absent; 1, present.

RNA Sequencing

Colonoids were expanded and passaged once before washing with ice-cold DMEM/F12 to remove Matrigel and snap-frozen for subsequent RNA isolation. RNA isolation, quality control, and sequencing were performed by GeneWiz (South Plainfield, NJ, USA) using NEBNext Ultra RNA library prep with rRNA depletion, RNA ScreenTape (Agilent, Santa Clara, CA), and Illumina HiSeq (New England Biolabs, Ipswitch, MA) platforms. RNA-sequencing (RNA-seq) data are available in the Gene Expression Omnibus (GEO) (Illumina, San Diego, CA) under the accession number GSE135170.

RNA-Sequencing Preprocessing and Analysis

Data quality was assessed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Paired-end reads were aligned to the human reference genome (GRCh37) using the STAR (v.2.4.1c) 2-pass method.¹⁸ Read counts were then obtained from the aligned data using RNA-Seq by Expectation Maximization¹⁹ and input into R (v3.3.3) for further analysis.²⁰ Data were normalized, and differentially expressed genes were identified using the edgeR package²¹ (Supplemental Table 3). Genes with a total count across all samples that was ≥50 were considered to be expressed and retained for downstream analysis. Differentially expressed genes were analyzed using Ingenuity Pathway Analysis (IPA)²² to identify enriched pathways (Supplemental Table 4). P values were adjusted for multiple testing, and a pathway was regarded to be significantly enriched using a Benjamini-Hochberg P value cutoff of 0.05. Gene ontology enrichment analysis was applied using DAVID, and an FDR cutoff of 0.05 was used for significance (Supplemental Table 5).^{23, 24}

Quantitative Reverse Transcription Polymerase Chain Reaction

RNA was isolated from colonoids using the RNeasy Plus Micro Kit (Qiagen 74034, Hilden, Germany). The cDNA was synthesized from 500 ng of RNA using the high capacity cDNA Reverse Transcription Kit (Applied Biosystems 4368813, Foster City, CA). The cDNA was amplified using the Step One Plus Real Time PCR System (Applied Biosystems), TaqMan primers (HLA-DRA Hs00219575_m1 and GAPDH Hs02758991_g1), and TaqMan Fast Gene Expression Universal PCR Master mix (Applied Biosystems 4352042). Relative gene expression was calculated as R = 2^(-ddCt).

Flow Cytometry

Fresh growth media was added to each well 24 hours before analysis. An average of 2 wells were treated with 50 ng/mL of recombinant human interferon-γ (IFNγ) (PeproTech, Rocky Hill, NJ), with an equal number of wells remaining unstimulated. Twenty-four hours later, colonoids were extracted from the Matrigel and dissociated to single cells by incubation in TrypLE Express (Invitrogen) with ROCK inhibitor (Y-27632, Hello Bio, Princeton, NJ) and DNase DNAseI (Roche, Basel, Switzerland) with periodic vortexing. Cells were resuspended in PBS and filtered through a 40-μM filter. Cells collected from the single-cell isolation were stained with Zombie Aqua (Biolegend, San Diego, CA), CD80 (BD Biosciences Clone L307.4, San Jose, CA), HLA-DR (BD Biosciences), HLA-ABC (BD Biosciences), and CD86 (Biolegend). Antibodies were diluted in Roswell Park Memorial Institute (Invitrogen). After staining, cells were rinsed with PBS and analyzed with the LSR Fortessa at CHOP’s flow cytometry core. Results were analyzed using FlowJo Version 10 software. Gates were drawn to show all cells. Zombie Aqua positive cells were excluded as dead (Supplemental Fig. 2). The geometric mean, mean, median, and percentages of the cells were calculated for each sample.

Statistical Analyses

Frequency counts for patients exhibiting normal or poor crypt conversion between the 3 groups, frequency compared with endoscopic disease activity, and frequency compared with pathology scoring were compared using Fisher exact test. Real-time polymerase chain reaction and flow cytometry analyses were based on 2-sided, paired t tests or Mann-Whitney tests as indicated, and a P value of less than 0.05 was considered statistically significant.

RESULTS

Clinical Characteristics

Patient demographics and cohorts are listed in Table 1. Patients with VEO-IBD were diagnosed at younger than 6 years of age (range: 0.3–5.8 years), and age at time of collection ranged from 0.36 to 14.8 years. Three patients were identified to have monogenic defects. Patients with older onset pediatric IBD were diagnosed between age 6.4 years and 16.9 years, and age at time of collection ranged from 6.5 to 20.7 years. Patients with VEO-IBD and pediatric older onset were 63% and 53% male, respectively, and the majority of patients had isolated colonic disease (73% for VEO-IBD and 56% for pediatric older onset). The majority of the colonoscopies were performed as disease surveillance (19 of 26 in VEO-IBD and 10 of 16 in older onset IBD). Three patients underwent multiple colonoscopies during the study period for clinical indications, and biopsies were obtained for colonoid generation from endoscopically uninvolved areas. Of the colonoscopies from the VEO-IBD cohort, 13 of 26 were performed on treatment-naïve or nonimmunosuppressive therapy patients. Similarly, 9 of 16 pediatric IBD patients were treatment naïve or on nonimmunosuppressive therapy at time of colonoscopy. The majority of VEO-IBD and older onset IBD patients on immunosuppression were treated with biologic therapy, namely anti-TNFα medications.

Poor Enteroid and Colonoid Growth From a Subset of Pediatric Patients With IBD

We generated crypt enteroids (from terminal ileum mucosal biopsies) and colonoids (from right and/or left colon mucosal biopsies) from patients with older onset pediatric IBD, VEO-IBD, and age-matched controls. We evaluated enteroid and colonoid cultures 24 hours after plating to determine the percentage of isolated crypts (stem cell–containing units) that grew into enteroids or colonoids (crypt to spheroid conversion). We noted that enteroids and colonoids from control patients with at least 50% of isolated crypts converting to spheroids within 24 hours continued to grow robustly on subsequent days. Therefore, enteroids or colonoids were considered to have “normal” growth if ≥50% of crypts converted to enteroids or colonoids at 24 hours postplating. Although 100% of enteroids and colonoids from control patients exhibited normal crypt conversion, a lower percentage of patients with VEO-IBD (46% enteroids and 33% colonoids) and pediatric older onset IBD (79% enteroids and 73% colonoids) exhibited normal crypt conversion (Fig. 1A and 1B, Table 2).

TABLE 2.

Crypt Conversion Score by Groups

	Group
Growth Score, n(%)	Control	VEO	Older Onset	Total	All groups	VEO vs Older Onset
Enteroid
Poor	0 (0)	7 (53.9)	3 (21.4)	10 (25.0)	0.005	0.120
Normal	13 (100)	6 (46.1)	11 (78.6)	30 (75.0)
Colonoid
Poor	0 (0)	8 (66.7)	4 (26.7)	12 (31.6)	0.002	0.057
Normal	11 (100)	4 (33.3)	11 (73.3)	26 (68.4)

Data are presented as frequency counts (percentages). P1 values are reported from the Fisher exact test comparing 3 groups (control, VEO, and older onset). P2 values are reported from the Fisher exact test comparing 2 groups (VEO and older onset).

FIGURE 1.

Decreased crypt conversion in patients with VEO and older pediatric onset IBD. We evaluated intestinal and colonic crypts from endoscopically uninvolved biopsies from pediatric patients with VEO and older onset IBD and compared them with control patients. Enteroids or colonoids were considered to have “normal” growth if ≥50% of crypts converted to enteroids or colonoids at 24 hours postplating. (A) A significantly lower percentage of VEO-IBD (n = 13) or older onset (n = 14) enteroids grew normally compared with controls (n = 13). P = 0.005 by Fisher exact test. The percentage of patients with VEO and older onset IBD (n = 14) with normal enteroid growth was not significantly different. Data are presented as frequency counts (percentages of patients). (B) A significantly lower percentage of patients with VEO-IBD (n = 12) or older onset (n = 15) yielded normal growth colonoids compared with controls (n = 11). P = 0.001 by Fisher exact test. (C) Patients with IBD with normal crypt conversion (enteroid and colonoid combined) exhibited either no or mild inflammation, whereas patients with poor crypt conversion exhibited a range of inflammation scores, from none to severe, based upon histological inflammation scoring. P = 0.038 comparing all groups by Fisher exact test. (D) There were no significant differences in apoptosis scores (absent, no apoptosis noted; present but not increased, <1 apoptotic figure per 10 crypts; present and increased, >1 apoptotic figure per 10 crypts) between patients with normal vs poor crypt conversion. (E) We observed a significant increase in cryptitis in patients with IBD with poor crypt conversion compared with normal crypt conversion. Data represented as percentage presence of each histopathological feature; P value is reported from Fisher exact test. (n = 5 normal and n = 8 poor crypt conversion for enteroids; n = 8 normal and n = 8 poor crypt conversion for colonoids)

Patients With Moderate to Severe Histopathology Exhibit Poor Crypt Conversion

We evaluated histopathology from the included patients with VEO and older onset IBD and compared them with crypt conversion scores. Biopsies for histological analyses were obtained during the same procedure and similar locations as those used for the generation of enteroids/colonoids. Histologic grading for inflammation and apoptosis, in addition to presence of lymphoid hyperplasia, lymphoid aggregate, cryptitis, crypt distortion, intraepithelial neutrophils, and crypt abscess, were determined according to published protocols.¹⁷ Enteroids and colonoids from patients with IBD (both VEO-IBD and older onset) with normal crypt conversion had no (50%) or mild (50%) inflammation (black bars, Fig. 1C). Conversely, enteroids and colonoids with poor crypt conversion were from patients with no (23.1%), mild (38.5%), moderate (15.4%), or severe (23.1%) inflammation on corresponding biopsies (white bars, Fig. 1C). Though we did not observe a difference in histologic apoptosis between normal and poor crypt conversion groups by these analyses (Fig. 1D), we did detect a significant increase in the presence of cryptitis in patients with poor compared with normal crypt conversion (Fig. 1E). A complete list of histopathological scoring is available in Table 3. Taken together, these data suggest that a subset of patients with pediatric IBD exhibit poor crypt conversion associated in part with moderate to severe histological inflammation and presence of cryptitis.

TABLE 3.

Histopathological Parameters by Crypt Conversion (Enteroid and Colonoid)

	Growth Score
	Poor (n = 13)	Normal (n = 16)	Total (n = 29)	P
Inflammation	n (%)	n (%)	n (%)
None	3 (23.1)	8 (50.0)	11 (37.9)	0.04
Mild	5 (38.5)	8 (50.0)	13 (44.8)
Moderate	2 (15.4)	0 (0.0)	2 (6.9)
Severe	3 (23.1)	0 (0.0)	3 (10.3)
Lymphoid hyperplasia
Absent	8 (61.5)	10 (62.5)	18 (62.1)	0.99
Present	5 (38.5)	6 (37.5)	11 (37.9)
Lymphoid aggregate
Absent	7 (53.8)	4 (25.0)	11 (37.9)	0.14
Present	6 (46.2)	12 (75.0)	18 (62.1)
Cryptitis
Absent	5 (38.5)	13 (81.3)	18 (62.1)	0.03
Present	8 (61.5)	3 (18.8)	11 (37.9)
Crypt abscessa
Absent	4 (50.0)	6 (75.0)	10 (62.5)	0.61
Present	4 (50.0)	2 (25.0)	6 (37.5)
Granulomas
Absent	12 (92.3)	14 (87.5)	26 (89.7)	0.99
Present	1 (7.7)	2 (12.5)	3 (10.3)
Chronic architectural changes
None	7 (53.8)	12 (75.0)	19 (65.5)	0.27
Mild	3 (23.1)	4 (25.0)	7 (24.1)
Moderate	2 (15.4)	0 (0.0)	2 (6.9)
Severe	1 (7.7)	0 (0.0)	1 (3.4)
Crypt distortion
Absent	8 (61.5)	12 (75.0)	20 (69.0)	0.69
Present	5 (38.5)	4 (25.0)	9 (31.0)
Plasma cells
Absent	1 (7.7)	1 (6.3)	2 (6.9)	0.99
Present	12 (92.3)	15 (93.8)	27 (93.1)
Apoptosis
Absent	12 (92.3)	15 (93.8)	27 (93.1)	0.70
Present	0 (0.0)	1 (6.3)	1 (3.4)
Relatively Increased	1 (7.7)	0 (0.0)	1 (3.4)
IEN
Absent	10 (76.9)	12 (75.0)	22 (75.9)	0.99
Present	3 (23.1)	4 (25.0)	7 (24.1)
EOS
Absent	0 (0.0)	1 (6.3)	1 (3.4)	0.88
WNL LP	11 (84.6)	13 (81.3)	24 (82.8)
Inc LP	1 (7.7)	2 (12.5)	3 (10.3)
Epithelium	1 (7.7)	0 (0.0)	1 (3.4)
Mucus depletiona
Absent	5 (71.4)	8 (100.0)	13 (86.7)	0.20
Present	2 (28.6)	0 (0.0)	2 (13.3)
Villous surface transformationa
Absent	6 (100.0)	8 (100.0)	14 (100.0)	–
Small intestine villous bluntingb
Absent	3 (60.0)	7 (87.5)	10 (76.9)	0.51
Present	2 (40.0)	1 (12.5)	3 (23.1)

Data are presented as frequency counts (percentages). _aData are only available for colonoid. _bData are only available for enteroid. P values are reported from the Fisher exact test.

Transcriptome Analysis of Pediatric IBD Patient-derived Colonoids Revealed Common and Distinct Gene Expression Patterns Compared With Studies Published in Adults

We performed RNA sequencing to evaluate transcriptome-wide changes between passage 1 colonoids from patients with older pediatric onset IBD (n = 3) and VEO-IBD (n = 4) compared with controls (n = 7). We identified 1941 genes that were differentially expressed (P < 0.05) between the 3 groups (Supplemental Table 1). Principal component analysis of the 1941 genes that were differentially expressed (P < 0.05) demonstrated differences between control and older onset IBD along principal component 1 and control from VEO-IBD along principal component 2 (Fig. 2A), and the top 100 genes contributing to each principal component have been marked in Supplemental Table 1 with an “x.” The top 100 differentially expressed genes are represented as a heat map in Supplemental Figure 1. We next compared differential gene expression between colonoids from controls vs all pediatric IBD patients (older onset and VEO-IBD) and control vs VEO-IBD only with cutoff of P < 0.05 and log fold change >2.0 (Fig. 2B). Top upregulated genes in IBD patients vs control include LYZ, which has been identified in prior published studies in colonoids from adult patients with ulcerative colitis and Crohn’s disease,³ and HLA-DRB1 and HLA-DRA, which have been identified in intestinal epithelial cells in patients with IBD.^{25, 26} In our pediatric patients with IBD, we also found genes previously not identified in adult patients, including Kallikrein-related peptidase (KLK) genes. Additional genes that were upregulated in VEO-IBD patients included NMUR2, KCNJ3, FAM71A, KIAA1462, CSTA, and NLRP6. Interestingly, the most downregulated gene in all IBD colonoids compared with control colonoids was CD177, a glycosylphosphatidylinositol-anchored glycoprotein expressed in neutrophils and colon crypt cells that is associated with decreased disease in IBD^{27, 28} (Fig. 2C). Very early onset IBD colonoids additionally exhibited downregulated NFAT5. Loss of function mutations in NFAT5 have been associated with evidence of immunodeficiency and autoimmune panenteric disease in a patient with pediatric IBD and in a mouse model.²⁹

FIGURE 2.

Colonoids from patients with pediatric IBD exhibited transcriptomic differences compared with control patients. (A) Principle component analysis using differentially expressed genes with P < 0.05 of VEO-IBD (+, n = 4), older onset IBD (●, n = 3) and control (■, n = 7) colonoids at passage 1. (B) Venn diagram of differential gene expression between colonoids from control patients vs pediatric IBD (older onset and VEO-IBD combined) and control vs VEO-IBD only with cutoff P < 0.05 and log fold change ≥2.0 increase in gene expression. Top 10 upregulated genes in both VEO and older onset are listed, in addition to top upregulated in VEO-IBD. (C) Venn diagram of differential gene expression between colonoids from control patients vs pediatric IBD (older onset and VEO-IBD combined) and control vs VEO-IBD only with cutoff P < 0.05 and log fold change ≥2.0 decrease in gene expression. Top 10 downregulated genes in older onset and VEO-IBD combined are listed, in addition to top downregulated in VEO-IBD. (E) Quantitative reverse transcription polymerase chain reaction confirmation of one of the top upregulated genes, HLA-DRA, in control (n = 5) vs IBD (n = 8). Data presented as relative gene expression. P = 0.0485 by Mann-Whitney test; data expressed as mean ± SEM.

Validation of HLA Gene Expression Demonstrates Durable Upregulation in Colonoids

We analyzed colonoids further for surface expression of MHC class II. Surface expression is governed by a complex pathway of protein-protein interactions and intracellular compartment transfers. Increased transcript levels could occur without surface expression; however, surface expression is critical for functional antigen presentation. Prior publications demonstrated class II antigen presentation gene expression in intestinal epithelial cells in adult^{11, 12, 16} and pediatric patients with IBD⁴. In addition, recent studies identified functional MHCII machinery specifically in intestinal stem cells in mice, which in turn governs stem cell proliferation vs differentiation in response to infection¹⁷. The present analysis was also critical to rule out immune cell contamination of colonoid cultures as a confounding variable for gene expression analyses by evaluating later passages. We confirmed upregulation of the HLA pathway gene HLA-DRA via qPCR in early passage colonoids (Fig. 2D). Next, we performed flow cytometry on colonoids with or without IFNγ stimulation. Colonoid lines were passaged up to 10 times before analysis (Fig. 3A, Supplemental Table 1). We measured expression for HLA-DR and HLA-ABC, which correspond to class II and class I antigen presentation, respectively. We also evaluated costimulatory molecules CD80 and CD86, which are required for stimulation of T cells. Expression of HLA-DR and HLA-ABC was comparable in untreated colonoids in both groups. Interferon-γ was utilized to test inducibility of gene expression in these cells that normally do not express MHC class II gene. There was a trend toward an increase in the percentage of HLA-DR-expressing cells with IFNγ treatment in IBD colonoids accompanied by a significant increase in HLA-DR expression levels (P = 0.008) (Fig. 3B and C, Fig. 3B; P = 0.058), whereas IFNγ treatment increased HLA-ABC expression in both groups, with P = 0.023 for control and P = 0.065 for IBD (Fig. 3D). There were no significant differences between groups for costimulatory molecules CD80 and CD86 (Fig. 3E and F). Taken together, these data demonstrate that colonoids from IBD patients passaged multiple times retain an IFNγ-inducible increase in HLA-DR expression.

FIGURE 3.

Upregulation of antigen presentation genes in pediatric IBD colonoids. We performed flow cytometry on dissociated cells from control (n = 4) and pediatric IBD (n = 6) colonoids (at passages ranging from 2–10 (see Supplemental Table 1). (A) Schematic of IFNγ stimulation and flow cytometry analysis. (B, C) Percentage of cells expressing HLA-DR and mean HLA-DR expression. Mann-Whitney tests demonstrated a statistically significant increase in IBD colonoids treated with IFNγ (P = 0.008 for C). (D) Analysis of mean HLA-ABC expression demonstrated a statistically significant increase in control colonoids treated with IFNγ (P = 0.023). (F, G) Analysis of costimulatory molecules demonstrated no difference in expression levels of CD86 or CD80 in control or IBD colonoids. Data are expressed as mean ± SEM. MFI, mean fluorescent intensity.

DISCUSSION

There is a significant need to understand the contribution of intestinal epithelial cells to the etiology of IBD to identify novel approaches that promote healing of the epithelial barrier. Patient-derived enteroids and colonoids are critical tools for defining epithelial cell-intrinsic mechanisms that may contribute to disease pathogenesis. The current study provides the first evaluation of enteroids/colonoids from patients with VEO-IBD in comparison with older onset pediatric IBD and control patients. Our studies illustrate phenotypic distinctions between pediatric IBD and published analyses of adult patients: namely that both VEO-IBD and older onset IBD patient-derived enteroids and colonoids exhibit decreased growth efficiency in culture compared with control patients, whereas adult IBD patient colonoids grow similarly to adult controls.⁷ Our analysis of histopathology from patients with poor growth enteroids and colonoids suggests that inflammation is associated with inability for epithelial growth in pediatric IBD, although there was a subset of patients with poor growth that did not exhibit inflammation via histological scoring. Whether this is an intestinal stem cell specific phenomenon is not known. Of note, studies in enteroids from adult patients with active Crohn’s disease exhibited upregulation of stem cell marker gene OLFM4 compared with patients in remission. We found a relative decrease in OLFM4 in colonoids from patients with VEO-IBD vs control colonoids (Supplemental Table 6), suggesting that differences in stem cell dynamics in VEO vs adult IBD should be explored further.

Differences in gene expression, DNA methylation, permeability, and tight junction proteins between IBD and non-IBD patients have been reported using purified epithelial cells and/or patient-derived enteroids or colonoids.^{3, 4, 6, 7, 30} Consistent with prior studies, we observed gene expression and protein changes that were maintained ex vivo in patient colonoids. We found that a subset of differential gene expression identified in studies that analyzed colonoids from adults with ulcerative colitis was recapitulated in colonoids from pediatric patients with IBD, including increased expression of LYZ and HYAL1 and decreased expression of PLA2G2A, AQP8, and ZG16 (Supplemental Table 6). Additional genes, including KLK genes and HLA genes were upregulated in pediatric patients with IBD (both older onset and VEO).

Late-passage colonoids stimulated with IFNγ were characterized by increased HLA-DR expression in both control and IBD colonoids, suggesting long-term, durable changes in antigen presentation cell gene expression ex vivo. HLA class II is a requirement for T-cell activation. It is, therefore, possible that activation of this pathway in epithelial cells could contribute to a T cell–mediated pathologic process in IBD. Of note, IFNγ was chosen in our system as one of the most relevant pro-inflammatory cytokines in IBD upstream of Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling, suggesting that JAK-STAT activation may directly induce HLA-DR expression in the epithelium of patients with IBD. As such, these data may provide additional support for emerging data, suggesting the utility of JAK inhibitors in at least some patients with IBD.^{31, 32}

We found expression of costimulatory molecules CD80 and CD86 in a small subset of cells (<1%), suggesting that although intestinal epithelial cells express components of the antigen presentation machinery, there may be a smaller subset of cells that can functionally present an antigen. This is consistent with recent reports that intestinal stem cells marked by Lgr5, which constitute a similarly small subset of cells in the intestinal epithelium, express MHCII genes and can present antigen in cocultures with CD4+ T helper cells.³³ HLA class II genes are upregulated in epithelial cells in patients with IBD and celiac disease.^{27, 34, 35} Interestingly, low levels of HLA class II can be detected in intestinal biopsies from controls where they localize to multivesicular bodies within cells. However, in patients with ulcerative colitis or Crohn’s disease, they localize to the basolateral membrane, providing access to exogenous antigens for presentation.³⁶ The enhanced expression of HLA class II upregulation in patient-derived colonoids has not been described previously, and our data support that this phenotype can persist ex vivo. Whether or not epithelial cell–intrinsic upregulation of antigen presentation genes contributes to disease pathogenesis in pediatric or adult IBD remains to be determined.

CONCLUSION

The present study suggests that epithelial cells exhibit durable changes when cultured over several passages, underscoring the importance of this model system to contribute novel insight into epithelial contributions to IBD pathogenesis. This includes defining in future studies whether upregulation of HLA class II genes in epithelial cells contributes mechanistically to IBD, potentially via T-cell activation. Understanding functional differences in the epithelium from patients with IBD will provide new avenues for therapy specifically focused on epithelial healing, which remains a critical challenge in treating patients with IBD.

Author Contribution: KH, JK, ND, MC, EB, KS, and MDevoto contributed to the study concept and design. JK, KH, TK, LB, MDent, RB, KM, RS, MC, KW, RM, PC, and LS contributed to the acquisition of data. JK, MC, LM, KH, ND, JW, SU, KS, MDevoto, and AM contributed to the analysis and interpretation of data. JK and KH contributed to the drafting of the manuscript. JK, KH, KW, ND, MC, TK, SA, CG, KS, EB, and MDevoto, and AM contributed to the analysis of the manuscript for important intellectual content. JK, LM, KH, and ND contributed to the statistical analysis. JK, KH supervised the study.

Supported by: University of Pennsylvania Institute for Translational Medicine and Therapeutics (ITMAT) and University of Pennsylvania National Institutes of Health (NIH)/National Institutes of Diabetes and Digestive and Kidney Diseases (NIDDK) Center for Molecular Studies in Digestive and Liver Diseases (NIH P30DK050306) (pilot funds to JRK, KEH); National Institutes of Health (NIH) K23DK100461 (JRK); Crohns and Colitis Foundation Broad (JRK); Crohns and Colitis Foundation (CCF) Career Development Award (KEH); National Institutes of Health (NIH) K01DK100485 (KEH); National Institutes of Health (NIH) R03DK114463 (KEH); National Institutes of Health (NIH) R01DK124369 (KEH); Crohns and Colitis Foundation (CCF) Career Development Award (MAC); National Institutes of Health (NIH) K08DK106444-01 (ABM); National Institutes of Health (NIH) R03DK118310 (ABM); Asociacion Española Contra el Cancer (AECC) PAO16174030MORE (LM); National Institutes of Health (NIH) K01DK103953 (KAW); National Institutes of Health (NIH) R01DK121159 (KAW); Institutional funds from Children’s Hospital of Philadelphia Research Institute (JRK, KEH, KES); National Institutes of Health (NIH) R01DK111843 (M. Devoto); National Institutes of Health (NIH) 1R01AI123538 (CG); Children’s Hospital of Philadelphia Research Institute Gastrointestinal Epithelial Biology Center (KEH, ABM).

ACKNOWLEDGMENTS

Authors acknowledge the Children’s Hospital of Philadelphia’s VEO-IBD team: Janine McDermott, RN, BSN, Jamie Stevenson, RN, BSN, Nora Casper, RN, BSN, CPN, and Rachael Wohl, MSW, LSW. Authors thank Drs. Natalie Terry and Louis Ghanem for helpful comments. They are grateful for the support of the CHOP GI Epithelial Biology Center.