Introduction

According to the Diagnostic and Statistical Manual of Mental Disorders, DSM-V, autism spectrum disorders (ASD) are highly variable neurodevelopmental disorders characterized by persistent impairments in social interaction and communication, as well as restricted, repetitive behaviors1. Generally, these disorders appear before the age of 3 and affect boys 4 times more than girls2. Over the last years, the studies carried out have shown dramatic increases in the prevalence of ASD. For example, in the United States, the estimated ASD prevalence has risen to affect 1 in 54 children who are 8 years of age, according to a recently published report by the Centers for Disease Control and Prevention (CDC)2. However, the prevalence in Lebanon in 2016 was 1 in 66 children in Beirut and Mount regions3 but recently, this prevalence is estimated to affect between 49 and 513 children per 10,0004.

In parallel to the considerable clinical heterogeneity of these disorders, several studies have shown that ASD are multifactorial disorders due to the involvement of genetic and environmental factors5. Moreover, peripheral systems and organs are also affected and several comorbidities, such as gastrointestinal (GI) disorders, are considerably reported in ASD subjects6,7,8. To this end, the researchers were encouraged to explore the potential role of gut microbiota in ASD. Indeed, GI symptoms in children with ASD are 4.4 times more prevalent than in neurotypical children, with higher rates of constipation, diarrhea, and abdominal pain9. In fact, the gut microbiota has a substantial function in the metabolism, the immune system homeostasis and the central nervous system (CNS) activities through the microbiota-gut-brain (MGB) axis10,11.

Gut microbiota directly affects the production of several chemicals responsible for the brain functioning like GABA, dopamine and serotonin and it regulates the hypothalamic-pituitary-adrenal axis (HPA) by being the origin of microbial metabolites such as short-chain fatty acids that activates this axis12,13. Thus an altered microbial population leads to defective immune, endocrine, and nervous systems through a hyper-activation of cells of the immune system such as T-helper cells and an altered blood-brain-barrier integrity14. For instance, many studies have shown the association of ASD with altered levels of Lactobacillus, Prevotella, Bifidobacterium, Ruminococcus, Suterella and Alcaligenaceae15,16,17,18,19.

The studies conducted gave contradictory results. For example, a study to link ASD and abnormal gut functions conducted on 20 ASD patients and 20 neurotypical children, outlined less diverse gut microbiome and reduced abundance of Coprococcus, Prevotella and unclassified Veillonellaceae in ASD patients17. However, a study conducted using qPCR in the Simons Simplex Collection on 59 ASD subjects and 44 neurotypical siblings, showed no variation in the microbial population20. Additionally, most of the gut microbiota studies have been performed on West European and American populations, while so far, there are no studies on the Lebanese population. Therefore, there is an undeniable scientific interest in extending our knowledge to a Mediterranean population, especially that the ASD prevalence in Lebanon is increasing4.

Thus, the aim of our study is to identify the gut microbial community of ASD patients compared to their unaffected siblings and to controls among the Lebanese population in order to understand the impact of microbiota dysbiosis on ASD or its associated disorders Microbial compositions were estimated by the 16S rRNA sequencing (NGS) after a DNA extraction from samples collected from the patients, their siblings and control individuals from all the Lebanese regions.

Results

Socio-demographic data

Twenty-three subjects with ASD were included (two girls, twenty-one boys). They were aged 11 ± 5.9. Out of the 23 siblings, 13 (56.5%) were males and 10 (43.5%) were females. They were aged 11.3 ± 6. Out of 23 controls, 14 (60.8%) were males and 9 (39.2%) were females. They were aged 12.7 ± 6.1.

Comparison of gut microbial composition

The gut microbial composition of individuals with ASD, their siblings, and a control population was assessed at all levels of the taxonomic hierarchy in this study. At the Phyla level, our findings revealed that the relative abundance of Proteobacteria was significantly higher in the ASD group as compared to the control group (2.85% vs. 0.73%, p value of 0.0012) (Fig. 1). Conversely, we observed a relatively lower abundance of Phylum Bacteroidetes in individuals with ASD (15.5% vs. 23%, p value of 0.05) as compared to their siblings. Despite the fact that the control group had a higher proportion of Bacteroidetes (17%) compared to the ASD group, this variation was not statistically significant. Our findings showed that Tenericutes had a significantly higher relative abundance in individuals with ASD as compared to the control group (0.32% vs. 0.13%, p value of 0.04). Moreover, the ASD group had a relatively higher abundance of Tenericutes in comparison to their siblings (0.32% vs. 0.15%, with a p value of 0.08). We calculated the Firmicutes/Bacteroidetes (F/B) ratio for each group and compared it between the ASD and control groups, as well as between the ASD and sibling groups. However, no statistically significant differences were observed between these groups (p value between 0.3695 and 0.3684, respectively). Nonetheless, the Firmicutes/Proteobacteria (F/P) ratio was significantly higher in the control and sibling groups as compared to the ASD group (p value between 0.009 and 0.09, respectively). At the genus level, a single-factor statistical comparison revealed significant differences (p value < 0.05) in the abundances of Ruminococcus, Catenibacterium, Paraprevotella, Turicibacter, Gemmiger, and Anaerostipes across the three groups (Fig. 2).

Fig. 1
figure 1

Taxonomic composition of community at Phylum level using Stacked Bar plot.

Full size image
Fig. 2
figure 2

Taxonomic composition of community at Genus level using Stacked Bar plot.

Full size image

Species diversity and richness

The microbial communities within the three study groups (autistic, sibling, and control) were analyzed for their alpha diversity (Figs. 3, 4). Based on the visual examination of the bar plot, differences in diversity were observed between the ASD and sibling groups compared to the control group. However, when subjected to statistical analysis using one-way ANOVA, these differences were found to be statistically not significant at both the phylum and genus levels.

Specifically, in the observed index, the phylum diversity appeared to be lower in the ASD and sibling groups relative to the control group. Conversely, the ACE index, which considers unobserved or rare species, showed the lowest diversity in the ASD group, while the siblings group exhibited relatively higher diversity. Shannon and Simpson’s indices showed very little difference between the groups at the phylum and genus levels.

Although the visual inspection of the bar plot suggests potential diversity disparities, the lack of statistical significance from the one-way ANOVA indicates that these observed differences may be attributed to random variation or sampling effects.

The beta diversity index was assessed using the Bray-Curtis method, employing Principal Coordinate Analysis (PCoA) based on the relative abundance of Operational Taxonomic Units (OTUs) at the genus level (Fig. 5). PERMANOVA, a non-parametric approach, was used to evaluate the null hypothesis of no multivariate differences by employing permutation tests to compute p values21,22. The analysis revealed a variance with a p value of 0.053.

Fig. 3
figure 3

Gut microbial diversity in the studied groups. Alpha diversity was assessed at the Phylum level using four frequently employed methods: Observed, ACE, Shannon, and Simpson. The boxplots depict the median and interquartile (IQR) range, with whiskers reaching to the outermost data points within 1.5 times the IQR.

Full size image
Fig. 4
figure 4

Gut microbial diversity in the studied groups. Alpha diversity was assessed at the Genus level using four frequently employed methods: Observed, ACE, Shannon, and Simpson. The boxplots depict the median and interquartile (IQR) range, with whiskers reaching to the outermost data points within 1.5 times the IQR.

Full size image
Fig. 5
figure 5

Beta diversity index was assessed using the Bray-Curtis method, employing principal coordinate analysis to analyze the relative abundance of OTUs. Axis.1 and Axis.2 explained 23.3% and 20.7% of the variances, respectively. Statistical significance (p < 0.05) was assessed using permutation tests, integral to the PERMANOVA framework.

Full size image

Gut microbial differential abundance

To identify unique microbial profiles distinguishing each group based on differential abundance at various taxonomic levels, we employed LEfSe analysis using the Linear Discriminant Analysis (LDA) score (Fig. 6). The analysis revealed significant findings, indicating that the genus Catenibacterium showed enrichment in the ASD group, while the genus Turicibacter exhibited depletion in both the ASD and sibling groups compared to the control group.

Fig. 6
figure 6

Gut microbial markers were measured by LEfSe analytical tool with an LDA cut-off value > 2.0 in autistic, control, and sibling groups.

Full size image

Random forest analysis

Random forest analysis was conducted to identify predictive features that could potentially serve as biomarkers for a specific group (Fig. 7). Through the analysis, several genera exhibiting distinct abundance in specific groups were identified and ranked based on Mean Decrease Accuracy scores. The genus Catenibacterium ranked second, showing abundant enrichment exclusively in the ASD group, while the genus Turicibacter ranked seventh, displaying depletion in both the ASD and sibling groups.

Fig. 7
figure 7

Significant features identified by Random Forest. The features are ranked by the mean decrease in classification accuracy when permuted.

Full size image

Multi-factor analysis (MaAsLin2)

We used MaAsLin2, a microbial association analysis tool, to examine the connections between microbial features and experimental metadata (Fig. 8). Notably, Catenibacterium, Turicibacter, Faecalibacterium, and Phascolarctobacterium were identified as significantly associated (p value of 0.0003, 0.002, 0.001, and 0.003, respectively) with an FDR value less than 0.05. In addition to earlier findings, the application of metagenome sequencing allowed for the identification of 13 genera with a statistically significant p-value in terms of differential abundance of OTUs. Notably, the genera Catenibacterium, Turicibacter, and Faecalibacterium were among the included genera in this list.

Fig. 8
figure 8

Microbial association analysis to examine the connections between microbial features and experimental metadata for Autism Vs Control groups.

Full size image

Discussion

Our study aims to investigate the role of the microbiome in the manifestation of ASD by examining three distinct groups: individuals with ASD, their unaffected siblings, and a control group15,17,18,19,23,24,25.

Our comprehensive analysis, incorporating various statistical methods such as LEfSe, Random Forest, Differential Abundance Analysis, and MaAsLin, consistently highlighted the pivotal roles of Catenibacterium and Turicibacter in ASD. Notably, the LEfSe analysis revealed a significant enrichment of Catenibacterium in individuals with ASD, while Turicibacter exhibited a depletion in the ASD group compared to the control group. The results from the Random Forest analysis, Differential Abundance Analysis, and MaAsLin further supported these findings, with Catenibacterium exhibiting differential abundance in the ASD group compared to both the control and sibling groups (p value of 5.35E-15), and Turicibacter showing significant differential abundance in the control group compared to ASD and sibling groups (p value of 0.036). The convergence of these findings from multiple analytical approaches underscores the robustness and significance of the association between Catenibacterium and Turicibacter abundance with ASD. Additionally, we hypothesize the involvement of genetic factors and shared environment as potential contributors to these differences especially for Turicibacter which showed a depletion in the ASD and sibling groups compared to controls. Wu et al.26 noted a decrease in the abundance of Turicibacter closely related to ASD. In fact, Turicibacter are a gram-positive anaerobic bacteria associated with butyric acid which stimulates insulin secretion in the pancreas. It has been shown that Turicibacter play important roles such as providing anti-obesity effects, reducing metabolic stress, and inhibiting inflammatory reactions27. Turicibacter has been associated with the production of short-chain fatty acids (SCFAs), which are essential for gut-brain communication and have been linked to a number of neurological and psychiatric disorders28. SCFAs can influence the central nervous system by modulating the synthesis of neurotransmitters, changing the blood-brain barrier integrity, and impacting the immune system function13.

At the genus level, our study revealed significant differences (p value < 0.05) in the abundances of Ruminococcus, Catenibacterium, Paraprevotella, Turicibacter, Gemmiger, and Anaerostipes across the three groups. These findings are consistent with previous research that has implicated these genera in various aspects of ASD. For instance, Zhang et al. (2018) reported decreased levels of Faecalibacterium and Roseburia, both members of the Ruminococcus genus, in the ASD group compared to the control group29. Chen et al.30 observed an elevated abundance of Ruminococcus torques in stool samples from children with ASD. In addition, Soltysova et al.31 reported decreased levels of Ruminococcus, Turicibacter, and Anaerostipes in children with ASD. Song et al.32 found higher levels of Clostridium and Ruminococcus species in ASD subjects compared to healthy controls. Moreover, a study by Liu et al.33 associated a variation in the glycosyltransferase B3GALNT2 gene, associated with congenital muscular dystrophy with brain malformations, with the increased abundance of Catenibacterium mitsuokai in a subset of children with ASD34. Furthermore, Chen35 indicated that Catenibacterium was consistently discerned as a potential biomarker by at least two analysis methods. Catenibacterium is Gram-positive and an obligatory anaerobe, known for its role in carbohydrate fermentation, it ferments glucose for the production of acetic, lactic, butyric and iso-butyric acids. The presence of Catenibacterium has been positively associated with obesity-related insulin resistance. Furthermore, it has been shown that consuming a Western diet increases the abundance of Catenibacterium in the gut microbiota36. Alterations in the gut microbiota composition, including changes in Catenibacterium abundance, could potentially lead to dysregulation of the gut-immune-brain axis, which has been proposed as a contributing factor in the pathogenesis of ASD37.

In alignment with prior investigations38,39, our study revealed a statistical significance in the abundance of Proteobacteria between the ASD and control groups. However, there remains a lack of established explanations for the observed significant difference in Proteobacteria abundance between individuals with ASD and their siblings. Proposed hypotheses suggest genetic factors, shared environmental exposures, and interactions between genetics and the environment as possible causes of this difference20,40,41. The gram-negative Proteobacteria are one of the most abundant phyla in the human gut microbiota42. By absorbing oxygen and reducing the redox potential in the gut environment, they are thought to play a key role in preparing the gut for colonization by the anaerobes necessary for gut health43.

Bacteroides species have beneficial effects including supplying energy to the host by facilitating fermentation, promoting T cell-dependent immune responses, providing protection from pathogens and supplying nutrients to other microbial residents of the gut44. Our study revealed a significant finding regarding the relative abundance of Bacteroidetes in individuals with ASD compared to their siblings, demonstrating a relatively lower abundance of Phylum Bacteroidetes in individuals with ASD (15.5% vs. 23%, p value of 0.05). This observation aligns with previous research highlighting alterations in the gut microbiota composition in individuals with ASD23. Furthermore, although the control group exhibited a higher proportion of Bacteroidetes (17%) compared to the ASD group, this difference was not statistically significant. These results support the notion that dysbiosis of the gut microbiota, including alterations in Bacteroidetes abundance, may be associated with ASD pathology23,45.

Our study examined the Firmicutes/Bacteroidetes (F/B) ratio in individuals with ASD compared to the control and sibling groups. Surprisingly, we found no statistically significant differences in the F/B ratio between the ASD and control groups (p value of 0.3695) or between the ASD and sibling groups (p value of 0.3684). These findings contradict previous studies by Zhang et al.29 and Zou et al.46 that reported an elevated Bacteroidetes/Firmicutes ratio in children with ASD. However, consistent with our results, De Angelis et al. observed no significant differences in the Bacteroidetes/Firmicutes ratio between children with ASD and controls40. It is important to highlight that the majority of Bacteroidetes species produce propionic acid and other short-chain fatty acids as metabolic byproducts. Therefore, a study demonstrated that injecting propionic acid into the cerebral ventricles of rats led to the emergence of ASD-related biological and pathological traits23. The relationship between the Bacteroidetes/Firmicutes ratio and ASD is complex, with conflicting findings across studies. While changes in gut microbiota composition have been reported in individuals with ASD, it remains unclear whether the Bacteroidetes/Firmicutes ratio serves as a consistent marker in this population.

In our study, we observed that Tenericutes exhibited relatively low numbers in the gut microbiota of individuals with ASD. A systematic review and meta-analysis of eighteen studies assessing a total of 493 ASD children and 404 controls showed low numbers of Tenericutes in individuals with ASD and the controls47. However, our findings demonstrated a significantly higher relative abundance of Tenericutes in the ASD group compared to the control group (0.32% vs. 0.13%, p value of 0.04). Furthermore, the ASD group displayed a relatively higher abundance of Tenericutes in comparison to their siblings (0.32% vs. 0.15%, p value of 0.08). It is worth noting that Tenericutes are a diverse group of bacteria with varying abundances across different environments, and their specific role in ASD remains unclear. While Tenericutes have been identified as a dominant phylum in certain animal gut microbiota48, their association with the etiology of ASD has not been established in the available literature49. Further research is needed to elucidate the potential significance and implications of Tenericutes in the context of ASD.

In our study, the impact of sex differences on microbiota configurations cannot be overlooked, particularly in the context of autism spectrum disorder (ASD). Research has shown that biological differences related to sex can significantly influence the structure and function of the microbiome, affecting brain development and behavior relevant to ASD. Notably, variations in gut permeability and microbiome-related gene expressions between sexes may influence neurological outcomes. Furthermore, differences in the microbiota-gut-brain axis could explain the observed sex bias in ASD prevalence and manifestations, suggesting a differential impact of gut dysbiosis across genders50.

Conclusion

The current study expands the understanding of the role of the microbiome in ASD by uncovering specific bacterial taxa associated with ASD, including a lower abundance of Turicibacter and a significant enrichment on Proteobacteria in the ASD and siblings’ groups compared to the controls along with a lower abundance of Bacteroidetes in the ASD group compared to controls and siblings and an increased abundance of Catenibacterium and Tenericutes in the ASD group. These observations may highlight the importance of the interplay between environmental and host genetic factors in shaping the gut microbiome. This study also emphasizes the importance of the identification of specific microorganisms’ changes that can be targeted as potential biomarkers such as Turicibacter and Catenibacterium, although further validation with larger cohorts is needed and further investigations are needed to unravel their functional implication in the context of ASD.

Overall, our study contributes to the growing body of evidence linking the microbiome to ASD. Understanding the specific bacterial taxa associated with ASD opens avenues for targeted interventions and personalized treatment strategies. Further research exploring the mechanisms of these bacteria, along with the integration of additional metadata, will enhance our understanding of the microbiome-ASD relationship and its clinical implications.

Materials and methods

Ethics statement and study approval

This study was complied with the ethical standards and guidelines stated by the Declaration of Helsinki in 1964 and its later amendments. The Holy Spirit University Ethical Committee has reviewed and approved the study protocol. Also, prior to the study, the parents of all participants signed an informed consent.

Subjects

Twenty-three subjects diagnosed with ASD according to the Diagnostic and Statistical Manual of Mental Disorders in its 5th edition1 were recruited from specialized institutions and non-governmental organizations (NGOs) distributed all over Lebanon. Two control groups, with twenty-three subjects in each, were included in this study: a non-autistic sibling group and an unrelated healthy control group.

In addition, the parents filled a questionnaire covering the socio-demographic factors such as the parental age and their economic status, the pre- and peri-natal factors like the maternal diet during pregnancy and the delivery method, and the post-natal factors such as early childhood complications.

Fecal sample collection and preservation

Stool samples were collected in a 120 ml medical disposable plastic specimen container for stool samples from twenty-three cases, twenty-three siblings and twenty-three controls. After defecation at home, the sample were kept at − 20 °C. They were the transported in a camping cooler filled with ice to the Holy Spirit University of Kaslik labs, where they were stored at − 80 °C for further processing.

DNA extraction and 16 S-rRNA sequencing

DNA was extracted from 200 mg fecal samples using the ZymoBiomics DNA Mini Prep Kit (ZymoResearch, Irvine, CA, USA) according to the manufacturer’s instructions. Then, the concentration and purity of the extracted DNA were measured using the Multiskan Sky Microplate Spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). DNA extracts were stored at -20 °C until further testing. Then, the DNA extracts were shipped to Macrogen Inc. (Seoul, Korea) for 16 S rRNA paired-end sequencing using Illumina MiSeq platform with Reagent Kit v3 (Illumina Inc., San Diego, CA, USA), targeting the V3–V4 region.

Bioinformatics analysis

After polymerase chain reaction (PCR) amplification and sequencing on Illumina platform, raw fastq files were processed using the Quantitative Insights into Microbial Ecology 2 (QIIME 2)51. The paired-end reads removed primer and adapter sequences using the QIIME 2 pipeline. Then, using the DADA2 program (https://qiime2.org/)52 with forward and reverse truncation set, the denoising pipeline was performed for the amplicon sequence variants construction including quality filtering, dereplication, dataset-specific error model learning, denoising, paired-end reads joining and chimeras removing. For each representative sequence, the feature-classifier53 and algorithm in QIIME251 were employed to annotate taxonomy classification based on the information retrieved from the GreenGenes (https://greengenes.secondgenome.com/) and Silva (https://www.arb-silva.de/) databases. Then, we performed multiple sequence alignment using the QIIME2 alignment MAFFT54 against these databases55,56 in order to analyze the sequence similarities among different amplicon sequence variants.

Statistical analysis

Relative frequencies at each phylogenetic level were calculated by ANCOM, from the DADA2 feature table for assigned sequence variant.

Alpha and beta diversity analyses were performed using the phyloseq package2. The results were plotted across samples and reviewed as box plots for each group or experimental factor. Further, the statistical significance of grouping based on experimental factors was also estimated using an either parametric or nonparametric test. Pairwise distance and pairwise differences in alpha diversity values (observed operational taxonomic units) were calculated within QIIME2, using the q2-longitudinal plugin. Computing species abundance using the q2-ANCOM plugin. Beta diversity differences were assessed using permutational multivariate analysis of variance (PERMANOVA), based on Bray-Curtis dissimilarity. The analysis was performed with 999 permutations to determine statistical significance. Species abundance was assessed, consecutively at each taxonomic level to detect significant changes in abundance.

In addition, principal coordinate analysis (PCoA) was used to investigate the similarities between bacterial communities. LEfSe (Linear discriminant analysis Effect Size) was used to identify the differentially abundant taxa. The Kruskal–Wallis rank sum test was done to detect features with significant differential abundance, followed by Linear Discriminant Analysis to evaluate the relevance or effect size of differential abundant features.