SCFA producing bacteria shape the subtype of ADHD in children

There is little data on population-based identification of the gut microbiota with ADHD subtypes in children, yet whether the degree ADHD is characterized by short-chain fatty acids (SCFAs) remains unclear. We enrolled 59 ADHD children including 21 inattentive subtypes (ADHD-I), 20 combined subtypes (ADHD-C), 18 hyperactive-Impulsive subtypes (ADHD-H) and 23 healthy controls. The microbiota was characterized by 16S rRNA gene sequencing, and SCFA concentrations were determined by gas chromatographic analysis. Compared to the controls, we observed a decrease of 14 genera belonging to Ruminococcaceae, Lachnospiraceae, Verrucomicrobiaceae and Rikenellaceae family in ADHD-I, while Megamonas, Coprococcus_2 and Paraprevotella were significantly increased in ADHD-C. In addition, a lower abundance of Faecalibacterium, and a higher proportion of Marvinbryantia, Intestinimonas, Prevotella_9 and Eggerthella were detected in the ADHD-H. Analysis of fecal SCFAs showed that elevated levels of acetate and propionate were in ADHD subtypes. Furthermore, most of the bacterium associated with SCFAs overlapped with the differential bacterium in ADHD subtypes. Conclusion: Our data support the clinical distinction among different ADHD subtypes in children may also be reflected in alterations of specific gut microbiota, most of which are SCFA producing bacteria.

The gut microbiota plays a key role in brain development and function and in the etiology of neurodevelopmental disorders via the gut-brain axis, a bidirectional communication between the gut and the brain relying on immune, neuroendocrine and neural pathways [5][6]. The microbiota is increasingly recognized for its ability to promote enteric and circulating serotonin production, control the differentiation and function of immune cells, and regulate expression of the 5-hydroxytryptamine receptor (5-HT1A), brain-derived neurotropic factor (BDNF), and NMDA receptor subunit 2 (NR2A), in turn, impact the brain [7][8][9]. Dysbiosis (alterations in microbiota composition and function) of the human microbiome has been reported in subjects diagnosed with several neurodevelopmental diseases, such as schizophrenia, autism and ADHD [10]. Furthermore, microbiota shift in early adolescence with neurodevelopmental disorders has been proved to ameliorate some, but not all, of the behavioral dysfunctions [11].
In addition to neural, endocrine, and immune pathways, gut microbial metabolites may be an additional mechanism in signaling with the potential to affect host physiology [12]. A growing body of evidence shows that microbial metabolites have a major influence on neurodevelopmental disorders.
The best known bacterial fermentation products are short-chain fatty acids (SCFAs) such as butyrate, acetate and propionate, which exert several effects, including maintenance of gut barrier function, providing a source of energy for colonocytes and bacterial communities, and possessing neuroactive properties [13]. Previous studies have shown that administration of a high dose of propionate in rats induced behavioral alterations (increased levels of locomotor activity) related with neurodevelopmental disorders [14]. Acetate and propionate, considered as the common preservatives in food products, has been demonstrated to aggravate hyperactivity in ADHD children, but when artificial food colourings and benzoate preservatives diminished, these symptoms may be ameliorated [15]. Moreover, butyrate, well known for its capacity to inhibit histone deacetylase, exerts beneficial behavioral effects via an epigenetic mechanism [16].
Although some published studies have initially observed gut microbiome profile changes in ADHD children, it is worth further investigation whether there are differences in the composition of gut microbiome between different clinical manifestations of ADHD children. Moreover, no populationbased identification of children with ADHD has been performed to assess the association between microbiota and a spectrum of SCFAs [17][18]. By subtyping our subjects based on their specific type of ADHD, the purpose of this study was therefore to detect the gut microbiota profiling and the levels of microbial metabolite SCFAs, and gave analysis on the relation between these two indices.
Materials And Methods Study Participants ADHD children

Control
Participants in the control group were matched on age and gender with ADHD participants (n = 23).
Controls were also screened using DSM-IV to exclude oppositional defiant disorder, conduct disorder, Tourette disorder or any other Axis I psychiatric comorbid disorders and use of any psychotropic medications. Controls were required to have no family history of psychiatric illness in first-degree relatives. Other exclusion criteria were the same as those for the ADHD group.
In addition, all participants of this study were under a normal diet, and no antibiotics, probiotics, or prebiotics have been taken in the 3 months prior to the sample collection. None of the subjects were on anti-inflammatory or antioxidant drugs. The study protocol was designed in accordance with the guidelines outlined in the Declaration of Helsinki and was approved by ethics committee of Beijing Children's Hospital. Written informed consent was obtained from the parents or guardians of all subjects, prior to the study. The characteristics of the participants are presented in Table 1.
Sample preparation, sequencing, and data processing Stool samples used were obtained in a sterile container, bought to the laboratory and immediately PCR amplification was performed in a 20 μL reaction system using TransGen AP221-02: TransStart FastPfu DNA Polymerase (TransGen Biotech, China) with an ABI GeneAmp® 9700 sequence detection system (ABI, Foster City, CA, USA). The PCR products were purified using an AxyPrep DNA Gel Extraction Kit (AXYGEN, USA) and then mixed equally before pyrosequencing. The PCR products of the V3-V4 region of the 16S rRNA gene were sequenced by the Next Generation Sequencing Core using an Illumina MiSeq PE300 as previously described [19].
To obtain high-quality sequences, the raw sequences were filtered and trimmed as previously reported [20][21]. After the sequences were optimized, operational taxonomic units (OTUs) assignment was performed for all sequences with a distance limit of 0.03 (equivalent to 97% similarity) using the U search software (version 7.1). According to the results of the taxonomy analysis, Rarefaction and Shannon-Wiener curves, the Chao and Ace estimator for community richness, the Shannon and Simpson index for community diversity, and the Good's coverage for sequencing depth were assessed for each sample using Mothur software (version v1.30.1).
Taxonomical assignments of OTUs were performed using Mothur software in accordance with the SILVA database at an 80% confidence level. Finally, the sequences were phylogenetically assigned to taxonomic classifications using an RDP Classifier (version 2.2) at a 70% confidence level. After phylogenetic allocation of the sequences down to the phylum, class, order, family and genus levels, the relative abundance of a given phylogenetic group was defined as the number of sequences affiliated with that group divided by the total number of sequences per sample.
To clarify the similarities of fecal microbiota between the experimental groups, Venn diagrams and species rank abundance distribution curves (Whittaker plots) were generated using R-project for statistical computing. In addition, non-metric multidimensional scaling (NMDS) and hierarchical cluster analysis were performed to determine whether the OTUs identified using the KW filter discriminate between different groups by examining relationships between ecological communities. Bacterial taxonomic analyses and comparison including bacterial phylum and genus were conducted between any different ADHD subtype and control groups using Wilcoxon rank sum test.

Data analysis
One-way analysis of variance (ANOVA) was performed to compare means in different groups with normally distributed data using SPSS version 19.0 for Windows; for data with a non-normal distribution, the differences between groups were assessed using the Mann-Whitney U-test and the Wilcoxonsigned-rank test. Significant differences between ADHD subtypes and the controls were assessed using Dunnett's test. Spearman's correlation between fecal SCFAs and associated genera was calculated and scaled by coefficients of each respective linear model. All values are expressed as the mean ± SD, and P < 0.05 was considered to be statistically significant.

Demographic and clinical comparisons
Demographics and clinical description of the participants with ADHD-I, ADHD-C, ADHD-H and controls are presented in Table 1. No significant differences were present between any two subtypes or relative to controls in terms of age, gender distribution, or total IQ. However, the higher score of inattention index and lower scores of hyperactivity/impulsivity symptoms in ADHD-I group than that of The alpha and beta diversity values of the bacterial communities between any ADHD subtypes and control groups were assessed using various indices based on the OTU level. To determine alpha diversity, the bacterial phylotype richness, reflected by the ACE and Chao indexes was calculated.
Meanwhile, the bacterial diversity expressed as Shannon and Simpson indexes. Detailed data on the estimators in each group are presented in Table 2. Compared to healthy controls, richness indices (ACE and Chao) and diversity index (Shannon) were all significantly lower only in ADHD-I, not in ADHD-C and ADHD-H subjects. However, another diversity index (Simpson) was significantly higher in ADHD-I and ADHD-C groups than that in the control. Rarefaction and Shannon-Wiener curves for each group indicated that the total bacterial diversity was well represented. More than 99% of coverage in all of the samples indicated that the sequencing depth was sufficient to reflect the whole bacterial diversity and the real composition of gut microbiota.
NMDS was performed to identify for dissimilarities in the microbial composition between three subtypes of ADHD patients and the controls (Fig 1F).. Analysis using the KW test showed that the microbial clusters from any ADHD subtypes were all located a shift to the left, which indicated compositional differences from healthy children. However, no significant differences in microbial composition were found among three subtypes of ADHD patients.
The gut microbiota composition in three subtypes of ADHD patients and healthy controls from the phylum to the genus is illustrated in Fig 1A-

Fecal SCFAs in different ADHD subtypes
As shown in Fig 3, compared to healthy controls, total SCFA propionate concentrations in feces were increased in ADHD subtypes, especially the concentration of acetate and propionate, were increased in ADHD-H subtypes. No significant differences were seen in butyrate, isovalerate and valerate concentrations between any subtype of ADHD and control groups (Fig 3)..  (Fig 4). More than half of the bacterium associated with SCFAs overlapped with the differential bacterium among ADHD and control groups, and this indicated that SCFAs might play the vital role in the occurrence of ADHD.

Discussion
To date, there are limited studies to identify that the fecal microbiota composition exist differences among different ADHD subtypes. In this study, we sequenced the total bacteria DNA of stool samples from 59 ADHD children including 21 ADHD-I, 20 ADHD-C, 18 ADHD-H and 23 control individuals, and examined the gut microbial composition with different ADHD subtypes. Compared to healthy controls, ADHD-I subtypes showed more reduced gut microbial diversity and more differences in microbial composition than the ADHD-C and ADHD-H subtypes.
Further analysis revealed that a significantly lower Verrucomicrobia phylum were shown only in ADHD-I children. Moreover, the majority of decreased Verrucomicrobia ascribed to the reduced Akkermansia genus. Some research has shown that Akkermansia has a beneficial role on the intestinal mucosal layer and enhances the barrier function of the gut epithelium, and may mediate obesity, diabetes, and inflammation [22]. Thus, in conditions of low Akkermansia abundance, the maintenance of a healthy gut barrier may be possible destroyed and pathogenic factors of other bacteria, like Lipopolysaccharide (LPS), could consequently harm the host [23].
At the family, Ruminococcaceae significantly decreased in ADHD-I and ADHD-H subtypes.
Ruminococcaceae constitute the major taxonomic group of the human gut microbiota and include many putative anti-inflammatory and thus potentially protective genera. The key members of Ruminococcaceae showed a confirmed decrease and were negatively correlated to disease duration in some neurodegenerative diseases, such as Parkinson's disease (PD), multiple system atrophy (MSA), multiple sclerosis (MS), and amyotrophic lateral sclerosis (ALS) [24]. The animal model for autism spectrum disorders also demonstrated that dysregulation of Ruminococcaceae are responsible for the anxiety-like phenotype [25]. The modulating social behavior of Ruminococcaceae may have been related to degrade complex polysaccharides to short-chain fatty acids, including acetate, butyrate, and propionate, that could induce alterations in T-cell differentiation and microglial function, and affect expression of neuroinflammation and neuroplasticity related genes in the prefrontal cortex or hippocampus [26][27].
Although the family Ruminococcaceae was found no difference between ADHD-H subtypes and controls, the genus Faecalibacterium which belongs to Ruminococcaceae was decreased in ADHD-H subtypes. The decreased Faecalibacterium in ADHD-H subtypes were in agreement with other previous ADHD studies [28][29]. Low Faecalibacterium levels are associated with the elevated markers of inflammation, which impact on the development of atopic diseases including asthma, eczema, and allergic [30][31]. Inflammatory cytokines released during conditions of low Faecalibacterium levels in the gut have the ability to cross the blood-brain barrier, which makes them likely candidates to play a role in the development of ADHD [32]. At present, Faecalibacterium is suggested as a potential future probiotic due to some strain-dependent anti-inflammatory features, like butyrate production [33].
However, the LEfSe approach in the present study also showed that an increased abundance of the family Prevotellaceae wasexhibited in ADHD-C and ADHD-H subtypes. Prevotellaceae is a well-known and often discussed genus in the phylum of Bacteroidetes. As a 'fermenting' group of bacteria, Prevotellaceae produce butyrate, which has been identified as a specific inducer of Treg cell differentiation [34]. The relative abundance of Prevotellaceae was also significantly higher in autoimmune disease, such as Graves' disease (GD), irritable bowel syndrome (IBS) and Crohn's disease [35][36]. Apparently, the high abundance of Prevotellaceae contributed to the inflammatory processes, with leading to a breakdown of the mucosal barrier [37].
In addition to its role in the inflammation, we identified the majority of changed gut microbiota in ADHD children also participated in SCFA metabolism. Previous studies indicate that Ruminococcaceae and Lachnospiraceae are robust butyrate-producers, and expansion of gut-residing Lachnospiraceae may result in increased butyrate production [38]. Species such as Akkermansia have been identified as key propionate producing mucin degrading organisms [39]. SCFAs play a pivotal role in host gut, metabolic and immune function with regulation of various elements of gut-brain axis. Moreover, circulating SCFAs produced by gut microbiota can directly influence the integrity of the blood-brain barrier (BBB) by increasing production of the tight junction proteins claudin-5 and occluding which limits entry of undesirable metabolites into brain [40][41].
Recent clinical studies suggested that the changed concentrations of SCFAs, including acetate, propionate and butyrate, were likely to be an important risk factor in PD, autism spectrum disorders (ASD), epilepsy and some inheritable metabolic disorders [42][43]. Exogenous sodium propionate supplementation lead to rapid intracellular acidification, and subsequently caused transient and variable decreases in excitatory postsynaptic current [44]. Furthermore, intraventricular infusions of PPA produced reversible repetitive dystonic behaviours, hyperactivity, turning behaviour, retropulsion, caudate spiking, and the progressive development of limbic kindled seizures [45]. In addition, Butyrate altered the metabolic behavior of macrophages to increase oxidative phosphorylation and also promoted alternative macrophage activation. Oral antibiotics disrupt this process to promote sustained T cell-mediated dysfunction and increased susceptibility to infections, highlighting important implications of repeated broad-spectrum antibiotic use [46]. In this study, we detected the increased acetate and propionate concentrations in ADHD subtypes, however, there were no differences among ADHD subtypes.
These SCFAs have been shown to affect the host through multiple mechanisms including the regulation of histone acetylation and methylation, G-protein coupled receptors (GPCRs), facilitating the secretion of various hormones (e.g. GLP-1 and PYY) and neurochemicals (e.g. serotonin), and the induction of vagus nerve signaling [49]. Since HDACs regulate gene expression, inhibition of HDACs has a vast array of downstream consequences [47]. Another intriguing function of SCFAs was able to affect the activity of epigenetic enzymes or act as the substrates necessary for epigenetic modifications which lead to changes of DNA methylation status in inflammatory bowel disease (IBD), type 1 diabetes mellitus (T1D), and obesity [48][49]. Recently, aberrant DNA methylation and histone acetylation were indicated in ADHD patients [50]. It is needed to confirm whether associations between SCFAs and DNA methylation levels might exist.
In conclusion, the present study identified preliminary significant differences in microbial composition between different ADHD subtypes and controls, and these gut microbiota changes are associated with SCFA producing process, which indicated that SCFA alteration may involve in the pathological symptoms of ADHD. Future studies will be needed to validate the present findings and elucidate the possible link of SCFAs and ADHD in children.   The bacterial abundance of differential microbial genus in ADHD subtypes and controls (P< 0.05). ADHD = attention-deficit hyperactivity disorder, ADHD-I = predominantly inattentive ADHD, ADHD-C = combined ADHD, ADHD-H = predominantly hyperactive-impulsive ADHD.

Figure 3
The fecal concentrations of SCFAs in different ADHD subtypes and the controls. Significance was established at P<0.05.