A20, which is encoded by the tumor necrosis factor alpha-induced protein 3 (TNFAIP3) gene, is a negative regulator of the TNF-nuclear factor-κB (NF-κB) signaling pathway. Haploinsufficiency of A20 (HA20) causes autoinflammatory and autoimmune disorders (Kadowaki et al., 2018). Dysbiosis of the intestinal microbiome has been observed in some inborn errors of immunities (IEIs)/primary immunodeficiency disease, such as chronic granulomatous disease (CGD) (Sokol et al., 2016), X-linked inhibitor of apoptosis (XIAP) deficiency (Sokol et al., 2016; Ono et al., 2021), tetratricopeptide repeat domain 7A (TTC7A) deficiency (Sokol et al., 2016), common variable immunodeficiency (CVID) (Jørgensen et al., 2016; Varricchi et al., 2021), and Wiskott–Aldrich syndrome (Zhang et al., 2020). Despite IEIs having a monogenetic nature, they vary in severity relating to the microbiome (Liu et al., 2021). Little is known about the mechanism of dysbiosis in patients with IEI, but the association of gut microbiota and regulatory T cells (Tregs) was reported (Liu et al., 2021). Tregs regulate A20 (Kadowaki et al., 2020); as a result, the gut microbiota of patients with HA20 plays a vital role in association with Tregs. Alternatively, the mouse lacking A20 expression in dendritic cells showed intestinal dysbiosis (Talpin et al., 2019). This is the first report of the microbiome in patients with HA20.

Thirty-one fecal samples from 16 patients and 15 healthy family members were sequenced. However, samples from three patients who received antibiotic treatment within 4 weeks of stool collection were excluded from statistical analysis. Also, two samples from children under the age of 3 years were excluded from statistical analysis because their gut microbiome is naturally different from the older’s microbiome (Yatsunenko et al., 2012). Nobody but an excluded younger brother (P22-sib) has food allergies. The median (min–max) age of 13 patients with HA20 and their healthy family members of 13 individuals were 19 (3–43) and 36 (4–49) years (p = 0.09, Wilcoxon rank-sum test), respectively. Patients with HA20 consist of four mild, four moderate, and five severe patients. Eleven of 13 patients and eight of 13 presented with stomatitis and abdominal symptoms, respectively. One patient was treated with probiotics and 5 with immunosuppressants ( Supplementary Table 1 ).

Ten thousand reads were used from an average (and SD) of 53,935 (± 9,383) raw reads. An average of 9,848 reads passed the filter, and every 3,000 Passed Filter reads were used for OTU analysis. The number of OTUs in patients’ samples was considerably less than their healthy members (median 123 vs. 150, p = 0.049, Wilcoxon rank-sum test) ( Figure 1A ). Likewise, the Shannon index in patients’ samples was significantly less than their healthy members (median 4.68 vs. 5.26, p = 0.009, Wilcoxon rank-sum test) ( Figure 1B ), whereas the principal component analysis showed a significant difference in the composition of the intestinal microbiome in unweighted UniFrac analysis (p = 0.008, PERMANOVA) ( Figure 2A ), although not significant in weighted UniFrac analysis (p = 0.059) ( Figure 2B ). Thus, the stratified analysis suggested that any sex, stomatitis, abdominal symptoms, and use of probiotics do not affect the microbiome diversity ( Tables 1 and 2 ). Lower age was significantly associated with a lower Shannon index although insignificantly with OTUs. Both disease severity and immunosuppressant use were associated with a lower number of OTUs with significance, and they showed lower Shannon index although not significantly.

At the phylum level, there was no significant difference between patients with HA20 and their family members ( Figure 3 and Supplementary Table 2 ). Conversely, at the genus level, five genera (Streptococcus, Lactobacillus, Butyricicoccus, Haemophilus, and Enterococcus) increased, and another four (Ruminococcus, Clostridium, Parabacteroides, and Alistipes) decreased in patients with HA20 compared with their family members (LDA score greater than two at p < 0.05 with LEfSe analysis) ( Figure 4A ). At the OTU level, seven OTUs significantly increased in patients with HA20 compared with their family members (p < 0.001: closest species/strain name Haemophilus parainfluenzae, Streptococcus mutans, and Veillonella sp. oral taxon 158, p < 0.05: Bifidobacterium dentium, Lactobacillus salivarius, Enterococcus avium, and Eubacterium eligens, Wilcoxon rank-sum test). Most of these OTUs are indigenous bacteria and are undetected in the intestinal microbiota. As a representative, only one of two healthy child siblings under 10 years and none of 11 healthy adult members had H. parainfluenzae, when all six child patients with HA20 and two of seven adult patients with HA20 had it ( Figure 4B ). None of these seven OTUs showed a significant correlation with either stomatitis or abdominal symptoms. However, two of seven OTUs showed a significant difference between patients with and without autoantibodies (four and five patients, respectively; four patients were not analyzed) ( Figure 4C ).

This study demonstrated that patients with HA20 show intestinal dysbiosis. Despite HA20 being a genetically occurring disease, previous studies report bacterial composition changes in other IEIs. XIAP deficiency is a rare IEI, and it is characterized by recurrent hemophagocytic lymphohistiocytosis and refractory inflammatory bowel disease (IBD) similar to Crohn’s disease. Furthermore, patients with XIAP deficiency showed intestinal dysbiosis and IBD, which are rescued by allogeneic hematopoietic cell transplantation (Ono et al., 2021). These data further establish the hypothesis that monogenic disease is the cause of gut dysbiosis. From previous studies, there are 118 of 354 IEIs that exhibit gastrointestinal symptoms (Hartono et al., 2019), and the interaction between bacteria and the immune system might vary with the disease. For instance, immunoglobulin A secreted at the mucosal surface of the gastrointestinal tract was associated with selective IgA deficiency (Berbers et al., 2019), and the absence of various subsets of T cells in intestinal lamina propria was associated with X-linked severe combined immunodeficiency (Clarke et al., 2018).

The mechanism that causes dysbiosis in patients with HA20 is mediated by Tregs because it is a key factor of immunoregulatory disorders in patients with HA20 (Liu et al., 2021), and in the mouse model of Treg-associated monogenic autoimmune disorders, Aire knockout mouse showed significant dysbiosis (Dobeš et al., 2015). Mouse models, Aire-deficient mice (Gray et al., 2007), and Foxp3 knockout mice (Chinen et al., 2010) exhibit an autoimmune phenotype even in germ-free conditions, and it was concluded that these diseases are independent of commensal microbial regulation (Liu et al., 2021). Conversely, the Treg-depleted model showed more severe inflammation in the small intestine of specific pathogen-free mice than germ-free mice, which indicates that the disease severity is associated with the microbiome (Liu et al., 2021). These findings indicate the possible association between disease severity and gut dysbiosis. Additionally, data from this study showed a significant correlation between disease severity and dysbiosis. The use of immunosuppressants, an alternative indicator of disease severity, also correlated with dysbiosis.

Another possible mechanism of dysbiosis is metabolism. A mouse model lacking A20 in specific dendritic cells reported aberrant expression of antibacterial peptides and luminal dysbiosis (Talpin et al., 2019).

Ten of 18 patients with HA20 were treated with immunosuppressants (Kadowaki et al., 2020), and one refractory patient successfully received hematopoietic cell transplantation (Shiraki et al., 2021). However, less invasive therapy would be required. Successful fecal microbiota transplantation (FMT) is reported in patients with HIV (Serrano-Villar et al., 2021) and the mouse model of Behcet’s disease (Ye et al., 2018), while there is no evidence of FMT in patients with IEIs including HA20. Further study is required to clarify the mechanism of dysbiosis in patients with HA20 and IEIs, to determine the biomarker of the disease, and to discover if FMT works or not.

Several monogenic IEIs inform us how the disease-specific immune dysfunction causes disease-specific dysbiosis. Collectively, we can easily understand how the complex immune systems respond to various microorganisms. This study showed that S. mutans and L. salivarius significantly increased in both patients with HA20, those with autoantibodies compared with family members and patients without autoantibodies, respectively. Since the sample size was small for statistical analysis (13 vs. 13, and 4 vs. 5, respectively), statistical significance in two manners was rare enough to reject the null hypothesis. Also, although the close mechanism remains unclear, it was suggested that these two OTUs might be associated with autoimmune onset.

In the context of statistical analysis, we excluded patients younger than 3 years and/or patients with antibiotics use because they present a confounding bias for microbiota (Hopkins et al., 2002; Yatsunenko et al., 2012; Ramirez et al., 2020). However, our data suggested that lower age is the explanatory variable for lower Shannon index, not for OTU numbers. This potential confounding bias is statistically a problem but not clinically crucial. Considering clinical management, both lower age and antibiotics use affect disease severity as well as microbiome diversity. Indeed, the excluded patients based on antibiotics use [previously described as P4, P9, and P21 (Kadowaki et al., 2018)] were relatively younger (4, 1, and 3 years old, respectively) and more severe (P4 and P9 are moderate; P21 is severe) than the included patients. Early-onset patients could be regarded as more strongly influenced by genetics and so potentially more severe than late-onset patients.

This study has several limitations. Firstly, clinical heterogeneity exists among medications, such as probiotics and antibiotics. The potential confounding factor is underestimated because of the small sample size. Secondly, because of rare diseases, it is difficult to collect more fecal samples for statistical analysis of the potential confounding factor described. Multivariate analysis was unfeasible due to small samples. Thirdly, not all patients with HA20 received gastrointestinal endoscopy, and so alternative indicators such as the disease severity or abdominal symptoms do not necessarily reflect the exact intestinal condition.

We investigated the intestinal microbiome composition in patients with HA20 and identified that seven OTUs significantly increased in the patients. In particular, S. mutans and L. salivarius significantly increased in patients with HA20 and autoantibodies, and these OTUs were associated with autoimmune onset.

The datasets for this article are not publicly available as regards participant/patient anonymity. Requests to access the datasets should be forwarded to the corresponding author.

Written informed consent was obtained from patients or their parents. This study was conducted following the Declaration of Helsinki and approved by the ethics boards of Tokyo Medical and Dental University and Keio University.

ET, KT, and YK contributed to data management. KT, YK, WS, HM, WS, and MH conducted DNA sequencing and bioinformatics analysis. HO and HK managed sample recruitment. HO, TK, RN, KN, CK, TE, TS, and YI contributed to sample collection. ET, KT, and HK wrote the manuscript. TM and KH provided critical discussion. HK designed and managed the project. All authors read and approved the final manuscript.

This work was partly supported by MEXT/JSPS KAKENHI (Grant Number: JP17K10099) and Naoki Tsuchida Research Grant to HK.

KH is a scientific advisory board member of Vedanta Biosciences and 4BIO CAPITAL. YK is an employee of JSR Corporation.

The remaining author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.