Mucosal Invariant T cells are Diminished in Very Early-Onset Inflammatory Bowel Disease

Abstract

Objectives

Very early-onset inflammatory bowel disease arises in children less than six years old, a critical time for immunologic development and maturation of the intestinal microbiome. Non-conventional lymphocytes, defined here as mucosal-associated invariant T cells and innate lymphocytes, require microbial products for either development or expansion, aspects that could be altered in very early-onset inflammatory bowel disease. Our objective was to define conventional leukocyte and non-conventional lymphocyte populations in controls and patients using multiparameter flow cytometry to test the hypothesis that their frequencies would be altered in a chronic inflammatory state associated with significant dysbiosis.

Methods

Multiparameter flow cytometry was used in a control cohort of 105 subjects to define age-effects, not previously comprehensively examined for these cell types in humans. Differences were defined between 263 unique age-matched patients with very early-onset inflammatory bowel disease and 105 controls using Student’s t test. Subjects were divided into two age groups at the time of sampling to control for age-related changes in immune composition.

Results

Intermediate monocytes were consistently decreased in patients with VEO-IBD compared to controls. Mucosal-associated invariant T cells were significantly lower in patients with long-standing disease. Levels were less than half of those seen in the age-matched control cohort. The innate lymphoid cells type 2 population was expanded in the youngest patients.

Conclusions

Mucosal-associated invariant T cells are diminished years after presentation with inflammatory bowel disease. This durable effect of early life intestinal inflammation may have long term consequences. Diminished mucosal-associated invariant T cells could impact host defense of intestinal infections.

Introduction

Inflammatory bowel disease (IBD) is a chronic, relapsing inflammatory condition of the gastrointestinal tract with complex etiologies (1). Children with very early-onset inflammatory bowel disease (VEO-IBD), defined as onset less than 6 years of age, have a distinct natural history, with severe disease, higher rates of medical therapy failure, and with a higher genetic burden (2–5). Monogenic conditions occur at a rate of 10–20% (6–8). The increasing incidence of IBD in this early-onset population is the highest in Westernized countries and this rapid increase cannot be accounted for by the monogenic conditions alone (9). The cause of the increasing incidence of VEO-IBD in Westernized countries is not fully understood, however, changes in the microbiome is likely contributory.

We designed this study to examine conventional leukocyte populations as well as a set of innate cells where microbial metabolites and/or microbial stimulation is known to impact growth, development, and expansion. Mucosal-associated invariant T (MAIT) cells express a semi-invariant T cell receptor (an invariant Vα7.2-Jα33 TCRα) and recognize a major histocompatibility complex class Ib molecule called MR1 (10). MAIT cell development in mice is highly dependent on critical time windows where early life exposure to riboflavin-synthesizing bacteria is essential (11, 12). MAIT cells accumulate in the mucosa of patients with IBD where they secrete proinflammatory cytokines (13). Thus, there is a sense that these cells may be central to the inflammatory process, however, MAITs also appear to have a powerful role in healing of the skin and could therefore represent a response to inflammation rather than a cause (12).

The development of VEO-IBD arises as a time when the microbial communities are being established in young children (14). These patients represent a unique opportunity to study the development and evolution of non-conventional lymphocytes in states of inflammation. We found that circulating MAIT cells are markedly decreased in older patients who presented with VEO-IBD.

Methods

Patients and controls

This was a single center cross sectional study that included children with VEO-IBD, IBD that presented <6 years of age, and age-matched healthy controls. Informed consent was obtained for all samples according to an IRB-approved protocol from 2015–2020. The diagnosis of VEO-IBD was established based on a combination of clinical, biochemical, and histologic criteria with both acute and chronic changes observed on pathologic analysis and absence of infectious etiology. The controls were either healthy children having blood work obtained for routine monitoring or children undergoing endoscopy with normal endoscopic and histologic results. For participants with VEO-IBD, clinical data were extracted from the medical record into an established database.

Clinical Data

The following variables were obtained from the electronic medical record: date of birth, sex, date of IBD diagnosis, age at time of diagnosis, IBD diagnosis (Crohn disease (CD), ulcerative colitis (UC), IBD-Unclassified (IBD-U)), disease phenotype by Paris classification (15), macroscopic location of disease, diagnostic gastrointestinal pathology results, and growth parameters measured by weight- and height-for-age Z scores according to the WHO (for children age 0–24 months) and CDC growth curves (for children >/=2 years old). Disease severity was calculated based on the Pediatric Crohn Disease Activity Index (PCDAI) for patients diagnosed with Crohn disease and the Pediatric Ulcerative Colitis Activity Index (PUCAI) for patients with indeterminate colitis or colonic Crohn disease. Medications and erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) were extracted from the electronic medical record according to the date on which the sample was obtained.

Flow cytometry

Ficoll-Paque Plus (GE Healthcare) was used to isolate peripheral blood mononuclear cells from whole blood. Zombie Aqua was used to distinguish live cells. T cells were identified from physical gate followed by CD3 staining. Subsets were determined using CD4, CD8, and CD45RA. CD3% was defined as the percent of the physical lymphocyte gate. CD4%, and CD8% are presented as percentages of CD3+ cells. CD4/CD45RA is presented as the percent of CD4+ cells. Innate lymphoid cells (ILCs) were defined first using physical parameters on DAPI negative cells, followed by CD45 and lineage negative (CD19, CD3, CD1a, CD11c, CD14, CD34, CD123, BDCA2, and FceRI) gating. CD127 positive cells were used as the anchor for all ILCs. CD127+/CRTH2+ was used to define ILC2 and the percentage refers to the fraction of CD127+ cells that are CRTH2+. CD127+/CRTH2- was used to define ILC1 (c-kit- and CD56-) and ILC3 (c-kit+ and either NKp44+/−). Percentages refer to the CRTH2- population. NKT cells were defined as CD3+/CD16dim. The percentage refers to percent of lymphocytes. The iNKT cells were defined as CD3+/TCRVα18 and the percentage is based on the CD3 population. MAIT cells were defined as CD3+/CD8+/CD161+/TCRVα7.2+. The percentage refers to cells within CD3+/CD8+ set. Cells were analyzed on an LSR Fortessa (BD Biosciences) in the CHOP Flow Cytometry Core Facility. The sources of the antibodies are listed below.

Statistical analyses

Student t-tests were used to compare VEO-IBD cohorts vs. controls in each age bracket using Prism software. There is no correction for multiple comparisons. Most populations had a normal distribution. The p values are shown in each graph when significant.

Results

Recruitment

Consecutive subjects with VEO-IBD were recruited from a quaternary clinic dedicated to VEO-IBD. Subjects had pathologic confirmation of IBD based on standard histopathologic criteria with re-evaluation by a single dedicated pathologist when biopsies were obtained prior to being seen. There were 105 control subjects between the ages of 1–20 years. We began to analyze non-conventional lymphocytes after beginning the study and 34 control subjects had ILC, MAIT, and iNKT analyses. In the overall VEO-IBD cohort, 263 subjects had conventional leukocyte studies by flow cytometry and 65 subjects had non-conventional lymphocytes studied.

Age matching

We ensured appropriate age matching of our control cohort after dividing the group according to age at sampling ≥10 years and those <10 years of age. The age ranges were determined after analysis of control data and identification of a clear distinction between those two age groups (see below). Among VEO-IBD cases studied at <10 years of age, Cohort A, the mean age was 4.6 years, while the matched control set had a mean age of 4.9 years (p value NS). Among the VEO-IBD cases studied at ≥10 years of age, Cohort B, the mean age was 15 years, compared to 17 years in the matched control cohort (p value NS). In Cohort A, 206 had conventional T cell analyses and 55 had non-conventional lymphocyte analyses performed. In Cohort B, 57 had conventional T cell analyses and 10 had flow cytometry for non-conventional lymphocyte subsets.

Cohort characteristics

There were 263 unique subjects with VEO-IBD (age on onset: 1 month to 5.9 years of age) included in this study. Disease duration at the time of the sampling included time of diagnosis to 16 years. 52% of patients were diagnosed with CD, 5% with UC, and 43% with IBD-U. Disease location was predominately colonic in the VEO-IBD cohort, (65%). 13% had ileocolonic disease, 22% had pan-enteric disease and small numbers of patients had macroscopic disease limited to the stomach, perianal region, or other discrete locations. The mean PCDAI score was 11 with a median of 7.5 (range 0–52.5) and the mean PUCAI score was 16.2 with a median of 10 (range 0–85). The median weight Z score at the time of the analyses was −0.3 [IQR: −1.3, 0.4], median height for age Z score was −0.6 [IQR: −0.15, −0.1], weight/length ages 0 to 24 months: 0.6 [IQR: −0.6, 0.9] and BMI Z score for ages 2 years-20 years: 0.2 [IQR: −0.5, 0.8]. All those cases included for this study had exome sequencing performed. All those included for study has exome sequencing performed. Clinical characteristics of Cohorts A and B are given in Table 1.

Table 1.

Clinical characteristics: Patients with VEO-IBD divided according to age at the time of sampling

	Cohort A	Cohort B
	<10y of age at sampling	≥10y of age at sampling
N	206	57
Crohn Disease	43%	60%
L1 TI/cecal	1%	0%
L2 Colonic	43%	44%
L3 Ileocolonic	14%	23%
L4 Upper	8%	2%
Panenteric	31%	31%
Perianal	1%	0%
No macroscopic disease	3%	0%
Ulcerative colitis	4%	9%
Pancolitis	100%	100%
IBD-Unclassified	46%	22%
Colonic only	81%	69%
Ileocolonic	9%	12%
Panenteric	10%	16%
Unknown	7%	9%
Disease Location
Colonic	60%	49%
Ileocolonic	11%	18%
Panenteric	18%	24%
Other	11%	9%
Disease Severity as Measured by PCDAI and PUCAI
Inactive	37%	35%
Mild	26%	29%
Moderate	4%	4%
Severe	2%	3%
Not available	33%	26%
Not assessed due to ostomy	4%	7%
Monogenic causality of VEO-IBD	3%	15%

Eighteen patients (7%) had a monogenic cause identified. There were 106 newly diagnosed patients who were not on immunosuppressive therapy and 157 patients who were on immunosuppressive therapy of the time of the analyses. These therapies included anti-TNF, infliximab and adalimumab (n=86), immunomodulators, methotrexate and 6-MP (n=71), ustekinuab (n=8) and corticosteroids (n=89).

Development of ILC

Very little is known about the development and appearance of ILC and MAITs in blood during childhood. Therefore, we first set out to establish normal ranges in young children for these cell types. As seen in Supplemental Figure 1, throughout childhood, the conventional CD4 T cells and naïve CD4/CD45RA cells declined. We similarly examined NKT cells, ILC1, ILC2, ILC3 (NKp44+/−), and MAIT frequencies across the same age range. We found that ILC1 were the most frequent of the CD127+ ILC types in blood, comprising 50–80% of the ILC population overall in blood. Their frequency was stable as was that of ILC2 cells, the least frequent in the peripheral blood. MAITs were present in very low numbers in the first six years of life and were still rising in early adulthood. Monocytes represent another innate cell type highly impacted by microbial exposures (16). In our control cohort, there was no change in monocyte subset distribution according to age (data not shown).

Based on these normative data and a natural infection point for MAITs at 10y of age, we divided our study cohort into those < 10 years of age at sampling, Cohort A, and those ≥10 years of age at sampling, Cohort B. We also elected to present primarily percentages of cell types to minimize age-effects.

Conventional lymphocyte frequencies in VEO-IBD

To gain perspective on the overall composition of the lymphocyte populations in VEO-IBD, we analyzed conventional T cells initially. The patient and control cohorts were divided into two age groups that reflect their age at the time the sample was obtained. Among conventional T cells, only CD4/CD45RA naïve T cells were different in VEO-IBD Cohort A compared to controls (Figure 1). The mean frequency of CD4/CD45RA among the CD4 population was 77% in the VEO-IBD group, compared to 80% in the control cohort (p=0.007). Lower CD4/CD45RA T cells are often seen with chronic antigen stimulation. In VEO-IBD Cohort B, only CD4 T cell percentages were different between those with VEO-IBD and controls, with higher CD4 T cell percentages in patients (Figure 2). The mean frequency of CD4 T cells in the lymphocyte gate was 63% in the VEO-IBD cohort and 57% in the control cohort (p=0.008). Therefore, the disease does not lead to comprehensive changes overall in the T cell population frequencies.

Figure 1. Comparison of VEO-IBD Cohort A (sampled at less than 10 years of age) and age-matched controls.

The lymphocyte populations are indicated above each graph. The median is indicated with the horizontal line. N=206 for conventional lymphocytes and N=55 for nonconventional lymphocyte subsets in Cohort A. N=64 for conventional lymphocytes and N=20 for non-conventional lymphocytes in age-matched controls. Where there are significant differences, the p value based on a t test is indicated. CD4 CD45RA naïve T cells and intermediate monocytes are both significantly lower in the VEO-IBD Cohort A population compared to controls.

Figure 2. Comparison of VEO-IBD Cohort B (sampled at ≥10 years of age) and age-matched controls.

The lymphocyte populations are indicated above each graph. The median is indicated with the horizontal line. N=57 for conventional lymphocytes and N=10 for nonconventional lymphocyte subsets in Cohort B. N=41 for conventional lymphocytes and N=15 for non-conventional lymphocytes in age-matched controls. Where there are significant differences, the p value based on a t test is indicated. CD4 T cells are higher in the VEO-IBD cohort compared to controls and both intermediate monocytes and MAIT cells are decreased in the VEO-IBD Cohort B compared to controls.

We analyzed frequencies of conventional lymphocytes in association with ESR, CRP, PUCAI, PCDAI, medications, location of disease, height and weight (as Z-scores). There was no association of any individual cell population with these clinical parameters (data not shown).

ILC and MAIT cells in VEO-IBD

We therefore turned our attention to the non-conventional lymphocytes. We wished to test the hypothesis that chronic inflammation altered the composition of these cells in peripheral blood. All patients were diagnosed with IBD in early childhood, prior to 6 years of age. Therefore, Cohort B had disease of longer duration. In Cohort A, the ILC2 population was significantly higher in frequency in the patients with VEO-IBD [14.7% (SEM 1.5%)] compared to controls [7.0% (SEM 1.4%)] with a p=0.005 (Figure 1). There was a trend towards lower MAIT cell percentages in Cohort A but it did not reach statistical significance (p=0.06). Similarly for MAIT counts, the difference was non-significant with VEO-IBD sampled at <10y having a mean of 19 cells/mm³ and controls having a mean MAIT count of 32 cells/mm³, p=0.07. In Cohort B, where there are expected to be higher frequencies of MAITs, we found a marked decrease in MAITs compared to controls. The percentage of MAITs was 4.4% (SEM 2%) in VEO-IBD Cohort B while it was 11.7% (SEM 2%) in the control cohort (p=0.03) (Figure 2). The corresponding MAIT counts were 34 cells/mm³ for the VEO-IBD patients and 55 cells/mm³ for the controls, p=0.2. Thus, early life gastrointestinal inflammation was associated with markedly diminished MAIT cells in the peripheral blood, years after diagnosis.

Analyses of non-conventional lymphocytes with clinical parameters

To understand whether there are variables that impact the non-conventional lymphocyte frequencies in peripheral blood, we performed additional analyses with various clinical parameters. Due to the small N, we did not age stratify, but separately ensured age-matching. First, gowth effects were assessed by using the height and weight Z-scores and analyzing using linear regression. There were no significant assocaitions with growth parameters. We also did not see any relationship between PUCAI and PCDAI scores on the day of the visit and ILC or MAIT percentages. Similarly, ILC and MAIT frequencies were not associated with CRP or ESR (data not shown). These data suggest that non-conventional lymphocyte frequencies are not acutely influenced by disease activity. When patients were divided according to disease location (colonic, small bowel, panenteric), there were no differences in ILC or MAIT percentages, although in some cases the N was very small (Figure 3). Patients with an ostomy had significantly higher MAITs (p=0.0007). Finally, we analyzed potential medication effects and when stratified by medication, mean ILC and MAIT percentages were roughly equivalent across all treatments. These data analyses in some cases had very small sample sizes.

Figure 3. Clinical covariate analyses.

The non-conventional lymphocyte populations are indiated above each graph. The error bars represent standard deviation. The top analyzes disease location. Due to the sample size the subgroups are not stratified by age but the mean ages are comparable across groups. N=3 for ostomy, N=31 for colonic involvement, N=11 for panenteric and N=2 fpr small bowel disease exclusively. The second row analyzes medication effects. The subgroups are not age stratified but the mean ages are comparable across groups. N=20 for untreated, N=15 for steroids, N=11 for 6MP/methotrexate/azathioprine, N=2 for sirolimus, N=2 for vedolizumab, N=12 for TNF inhibitors, and N=8 for antibiotics. Where there are significant differences, the P value appears at the top of the graph. Patients who had undergone an ostomy had higher MAIT frequencies.

Additional cell types in VEO-IBD

As comparators, we examined other innate cells such as NKT cells, iNKT cells, and monocytes. NKT and iNKT cells were no different in patients and controls (data not shown). In both Cohort A and B, the intermediate monocytes were lower in patients with VEO-IBD compared to controls (Figure 1 and Figure 2). Intermediate monocytes are envisioned as a transitional cell type that bridges classical monocyte differentiation into non-classical monocytes. Increases in intermediate monocytes are seen with endotoxin exposure (17). We therefore analyzed monocyte subsets according to ESR and CRP and found no association of any subset with these inflammatory markers (data not shown).

Discussion

This study examined non-conventional lymphocyte subsets and monocytes in patients with VEO-IBD and controls in an effort to assess the impact of early life intestinal inflammation on these populations. Key ILC types where we expected to see an impact from early life intestinal disease were the ILC3 and MAIT populations. ILC3s have a critical role in host defense through secretion of IL-22, which stimulates antimicrobial peptide production, improves gut barrier function, and regulates tissue remodeling (18). Various activation pathways have been described for ILC3 including bacterial metabolites such as short-chain fatty acids produced from metabolism of fiber acting on the receptor Ffar2 (19–22). ILC3 are the dominant population of ILCs throughout the gastrointestinal tract and the tissue density of these cells declines in parallel with the degree of inflammation. We surprisingly saw almost no effect of VEO-IBD on the peripheral blood frequency of ILC3. Two previous studies of ILC3 in IBD focused on the tissue space in adult patients and found a loss of ILC3 with active disease (23, 24). In children with VEO-IBD we saw no effect on peripheral blood frequencies of ILC3.

We noted expanded ILC2 in Cohort A with VEO-IBD compared to controls. ILC2s are localized in lymphoid clusters in the intestinal mesentery of the small bowel (25). Tuft cell production of IL-25 may be under the influence of the microbial metabolite succinate (26). ILC2s can also interface with sensory nerve fibers and can mitigate intestinal inflammation (27)(28). The expanded ILC2 we found in Cohort A, the youngest children, may represent an effort to restrain inflammation early in the disease process.

Our most robust finding was decreased MAIT cells in the patients with VEO-IBD in the older children, Cohort B. MAIT cells are unlike ILC in that they develop in the thymus and are reliant on riboflavin (11, 12). After leaving the thymus, they undergo expansion and can reach 10% of the total T cell population in human blood samples around age 25 years, before declining again in the elderly (29–31). MAIT cells have been previously identified as diminished in IBD (13, 32, 33). Our study uniquely investigated the VEO-IBD population where we found a trend towards lower MAIT frequency in the youngest children where the MAIT population had not yet expanded (Cohort A). The MAIT frequency in Cohort B had markedly fewer MAIT cells than in the control cohort. There was not a clear difference in the MAIT frequencies between active and inactive disease states at the time of the blood draw, nor disease location, however, our study was not sufficiently powered to be able to discern small difference in this rare cell population. We did note that MAIT frequencies were preserved in children with an ostomy and whether this represents re-distribution of MAIT cells or some other mechanism will require further study.

Our study is innovative in its approach to the analysis of non-conventional lymphocytes, a set of cell types with pivotal roles in barrier function, healing, and host defense. We also chose to highlight the VEO-IBD cohort, where therapeutics have been disappointing and our understanding of the unique pathogenesis is lacking. Our study suggests significant alterations on the composition of non-conventional lymphocytes with diminished MAIT frequencies, a population dependent on microbial cues for expansion. Although our study offers unique analyses of an understudied cohort, there are limitations to our approach. We utilized a multi-parameter flow cytometry approach of peripheral blood predicated on the systemic nature of IBD. Other tissue compartments may have distinct findings. Although our sample size was large by the standards of VEO-IBD, a larger sample size and longitudinal approach would allow improved tracking of findings with disease activity. Finally, our population is enriched with monogenic causes of IBD, a variable we did not specifically dissect in our analyses due to small numbers. The monogenic cases were not specifically redacted but in no case did they have absent MAIT cells.

In conclusion, our study found markedly decreased MAIT cells in the VEO-IBD cohort assessed at later ages. Surprisingly, global effects on all lymphocytes were not seen. We were also surprised that there were limited effects on peripheral blood ILCs, given that their role in intestinal inflammation has been highlighted in murine models. The effect of MAIT depletion over the lifespan has not been evaluated and may be of importance in disease perpetuation.

What is known:

• Innate lymphoid cells are impacted by microbial stimuli and serve critical functions at barriers

• Adult populations with Crohn disease have altered ILC3 cell distribution

• In mice, exposure to a diverse microbial community is required early in life for mucosal-associated invariant T cell expansion and seeding of tissues

What is new:

• Patients with very early-onset inflammatory bowel disease had markedly lower mucosal-associated invariant T cells compared to controls even years after their inflammatory process developed

• ILC2 cells were expanded in the youngest patients, a cell type normally suppressed by dietary short chain fatty acids