Exploring the Association of Autism Spectrum Disorders and Constipation through Analysis of the Gut Microbiome

Over the past two decades, research into the role of the gut microbiome in regulating the central nervous system has rapidly increased. Several neurodevelopmental diseases have been linked to the unbalance of gut microbiota, including autism. Children on the autism spectrum often suffer from gastrointestinal symptoms, including constipation, which is four times more prevalent than it is in children without autism spectrum disorders (ASD). Although studies in animals have shown the crucial role of the microbiota in key aspects of neurodevelopment, there is currently no consensus on how the alteration of microbial composition affects the pathogenesis of ASD, let alone how it exerts an impact on the following comorbidities. In our study, we were able to control the effects of constipation on gut dysbiosis and distinguish neuropathological-related and gastrointestinal-related bacteria in ASD patients separately. By analyzing published data, eight additional bacteria significantly altered in autistic individuals were identified in our study. All of them had a decreased relative abundance in ASD patients, except Lactobacillaceae and Peptostreptococcaceae. Eighteen and eleven bacteria were significantly correlated with ASD symptoms and constipation, respectively. Among those, six bacteria were overlapped between the groups. We have found another six bacteria highly associated with constipation status in ASD patients only. By conducting Welch’s t-test, we were able to demonstrate the critical roles of microbes in ASD core and gastrointestinal symptoms and raised the hypotheses of their confounding and mediating effects on the relationship between the two symptoms.


Introduction
Autism spectrum disorders (ASD) are a group of complex disorders characterized by reduced social skills, restricted interests, and repetitive behaviors [1]. The prevalence of ASD is 1 in 36 children in the United States, among which boys are four times more likely to be diagnosed with ASD compared to girls [2,3]. Several comorbidities, including gastrointestinal (GI) symptoms, are frequently reported in autistic children [4,5]. Among those GI symptoms, constipation was observed to be the most common problem in ASD patients, with a prevalence of up to 50% or more [5][6][7][8]. The prevalence estimates of constipation in the general population were 24% and 39.6% only for chronic constipation and sub-chronic constipation, respectively [9].
Although the rate of ASD is rising, the causes are not well understood. Despite the fact that rare genetic variants have been found to contribute to the disease, they could only account for one third of all autistic cases [10]. Increasing evidence has shown that environmental factors play an important role in the etiology of ASD [11]. Recent studies have revealed differences between the gut microbiomes (the genomic profiles of gut microbiota) of ASD patients and healthy individuals, suggesting that ASD is associated with an unbalanced gut microbiota [12]. However, it is still not clear whether this unbalance, termed dysbiosis, is a factor causing ASD, or if the disease is causing the microbial alterations, let alone the exact microbial composition that is associated with the pathology of ASD and the following GI problems. Furthermore, although constipation is more prevalent in autistic children compared with normal individuals, the cause-effect relationship between ASD, constipation, and gut microbiota is not well established.
The aims of this study were to characterize microbes significantly correlating with ASD core symptoms and constipation separately, and to identify unique bacteria associated with constipation status in ASD patients. To achieve these purposes, we performed analyses on published data collected from autistic and non-autistic individuals, in which numbers of individuals suffering from constipation were clearly stated. We have found groups of bacteria that could reasonably explain the high prevalence of constipation among autistic individuals. Bacteria highly associated with constipation status in autistic, but not in neurotypical (NT), individuals were also identified. We further proposed hypotheses of confounding and mediating effects of the microbes in the interplay of ASD and constipation. One's dietary habits and intestinal metabolic environmental changes could play a significant role in modulating the relation between ASD, gut microbiome, and constipation. Our results not only provide insights on the cause-effect relationship between ASD, constipation, and microbes, but also shed light on future microbiome-based therapies for alleviating or even preventing the symptom of constipation in autistic children.

Methods
We adapted our data from the study of Strati et al. [13], in which 40 autistic children (average age: 11.1 ± 6.8; gender: 31 males and 9 females) and 40 age and gender-matched neurotypical healthy subjects (average age: 9.2 ± 7.9; gender: 28 males and 12 females) were recruited. Details of individual characteristics, sample collection process, diagnostic criteria for ASD and constipation, and the measurement of microbiomes can be found in Appendix A and the previous reference [13]. Sample accessions IDs, metadata, and sequencing read information are available in Appendix B.
We downloaded results of mean relative abundance (%) ± standard deviation (SD) of bacterial taxa at genus levels in autistic and neurotypical subjects, both constipated and non-constipated, from Strati et al. [13]. We then conducted Welch's t-test to investigate the difference of the relative abundance between groups. A p-value less than 0.05 is considered statistically significant. The Benjamini-Hochberg procedure was conducted to control the false discovery rate.

Results
We first compared the difference of relative abundance of bacteria between ASD patients and healthy individuals. Out of 152 bacteria taxa detected in all subjects [13], 13 were found to be significantly different between the groups (Figure 1). Out of these 13 bacteria taxa, five were mentioned in the original study generated via LEfSe analysis [14]. The other eight bacteria taxa (Eggerthella, Bacteroides, Lactobacillaceae; Unknown, Akkermansia, Peptostreptococcaceae; Unknown, Sporobacter, Flavonifractor, Barnesiella), although not found in the original paper, have all been identified as autism-related bacteria in previous studies [12,[15][16][17][18][19]. Notably, there was no Lactobacillaceae; Unknown detected in NT individuals, whereas the presence of this bacteria was found in ASD patients.

Figure 1.
The overall difference of gut microbiota between autism spectrum disorders (ASD) and neurotypical (NT) individuals was assessed. A negligible number (0.000001) was added to the mean abundance of every bacterium to avoid the value of magnitude of fold change being infinitely large or small. This number was calculated based on the one hundredth of the minimal relative abundance in either group. Each point represents a bacterium with its magnitude fold change in relative abundance (log2 of ASD/NT) on the x-axis and the value of statistical significance (−log10 of p value) on the y-axis. The dashed red line shows where p = 0.05, with points above the line having p < 0.05 and points below the line having p > 0.05. Points with magnitude of fold change less than 1 are shown in grey. Points outside of the solid lines are bacteria with mean abundance of 0 in either ASD (left) or NT (right) group. Bacteria significantly altered in abundance in addition to the findings in Strati et al. were colored or circled in pink.
In order to understand which bacteria are highly associated with ASD core symptoms, we next compared the relative abundance of bacteria between the ASD and neurotypical individuals that were both free from constipation (non-constipated ASD individuals (AD-NC) vs. non-constipated neurotypical individuals (NT-NC)). Out of 152 bacteria genera detected in these two groups, 18 (hereinafter termed Ga) were significantly different in relative abundance between the two groups. By excluding the ones with unknown genera, 15 were left, in which 12 (80%) appeared to be lower in the ASD group ( Figure  2A). The presence of Sporobacter in NT-NC was not seen in AD-NC. Interestingly, the degree of magnitude in decreasing of bacteria was substantially higher than increasing (Figure 2B). Phylum-level analysis showed a distinctive alteration of microbial composition in AD-NC characterized by a lower ratio of Bacteroidetes/Firmicutes (NT-NC = 4.46, AD-NC = 0.60). The proportion in relative abundances (%) of Firmicutes in these two phyla in AD-NC is 3.42 fold more than NT-NC ( Figure 2C).
We then investigated the bacteria associated with constipation status in NT individuals only. When comparing the group of non-constipated and constipated NT subjects, the relative abundance of 11 bacteria taxa (hereinafter Gc) were found to be significantly different, suggesting that these imbalanced bacteria play a role in causing constipation ( Figure 2D). All of the 11 bacterial taxa had a significant decrease in the NT constipated individuals ( Figure 2E). In order to understand which bacteria are highly associated with ASD core symptoms, we next compared the relative abundance of bacteria between the ASD and neurotypical individuals that were both free from constipation (non-constipated ASD individuals (AD-NC) vs. non-constipated neurotypical individuals (NT-NC)). Out of 152 bacteria genera detected in these two groups, 18 (hereinafter termed Ga) were significantly different in relative abundance between the two groups. By excluding the ones with unknown genera, 15 were left, in which 12 (80%) appeared to be lower in the ASD group ( Figure 2A). The presence of Sporobacter in NT-NC was not seen in AD-NC. Interestingly, the degree of magnitude in decreasing of bacteria was substantially higher than increasing ( Figure 2B). Phylum-level analysis showed a distinctive alteration of microbial composition in AD-NC characterized by a lower ratio of Bacteroidetes/Firmicutes (NT-NC = 4.46, AD-NC = 0.60). The proportion in relative abundances (%) of Firmicutes in these two phyla in AD-NC is 3.42 fold more than NT-NC ( Figure 2C). Int. J. Environ. Res. Public Health 2021, 18, x FOR PEER REVIEW 4 of 13 We then investigated the bacteria associated with constipation status in NT individuals only. When comparing the group of non-constipated and constipated NT subjects, the relative abundance of 11 bacteria taxa (hereinafter Gc) were found to be significantly different, suggesting that these imbalanced bacteria play a role in causing constipation ( Figure 2D). All of the 11 bacterial taxa had a significant decrease in the NT constipated individuals ( Figure 2E).
We next searched for bacteria that were highly associated with both ASD core symptoms and constipation. Six taxa (hereinafter Ga_c) were overlapping between Ga and Gc (Bacteroidales; Unknown, Barnesiella, Parabacteroides, Odoribacter, Bilophila, Butyricimonas). All of these bacteria belong to the phylum Bacteroidetes except Bilophila. All six bacterial taxa significantly decreased in relative abundance in both Ga and Gc (Figure 3). We next searched for bacteria that were highly associated with both ASD core symptoms and constipation. Six taxa (hereinafter Ga_c) were overlapping between Ga and Gc (Bacteroidales; Unknown, Barnesiella, Parabacteroides, Odoribacter, Bilophila, Butyricimonas). All of these bacteria belong to the phylum Bacteroidetes except Bilophila. All six bacterial taxa significantly decreased in relative abundance in both Ga and Gc (Figure 3). By comparing gut microbiota compositions in non-constipated and constipated ASD children (AD-NC vs. AD-C), we found nine bacteria taxa (Gc_a) showing significant difference in relative abundance ( Figure 4A). The abundance was substantially lower in all bacteria except Enterobacteriaceae; Unknown in AD-C ( Figure 4B). Seven out of the nine bacteria taxa had genus information and all belong to the Firmicutes phylum. Even after p-value correction using the Benjamini-Hochberg method, one of the bacteria (Faecalibacterium) was still significant. To further understand the bacteria correlating with the constipation status in ASD only, we removed Gc from Gc_a. Six imbalanced bacteria taxa were uniquely associated with constipation status in ASD patients (Gc_a_uniq), among which five had genus information (Turicibacter, Roseburia, Dialister, Staphylococcus, Butyricicoccus). These five bacteria genera belong to the Firmicutes phylum and all anticorrelate with constipation status ( Figure 5). On the other hand, out of the eight imbalanced bacteria taxa that were uniquely associated with constipated NT individuals, five (62.5%) belong to the Bacteroidetes phylum, and only two (25%) were in the Firmicutes phylum. All of these bacteria anticorrelate with constipation status. The decreasing of Faecalibacterium, Gemmiger, and an unknown bacterium were associated with constipation status in both ASD and NT individuals ( Figure 5).  By comparing gut microbiota compositions in non-constipated and constipated ASD children (AD-NC vs. AD-C), we found nine bacteria taxa (Gc_a) showing significant difference in relative abundance ( Figure 4A). The abundance was substantially lower in all bacteria except Enterobacteriaceae; Unknown in AD-C ( Figure 4B). Seven out of the nine bacteria taxa had genus information and all belong to the Firmicutes phylum. Even after p-value correction using the Benjamini-Hochberg method, one of the bacteria (Faecalibacterium) was still significant. To further understand the bacteria correlating with the constipation status in ASD only, we removed Gc from Gc_a. Six imbalanced bacteria taxa were uniquely associated with constipation status in ASD patients (Gc_a_uniq), among which five had genus information (Turicibacter, Roseburia, Dialister, Staphylococcus, Butyricicoccus). These five bacteria genera belong to the Firmicutes phylum and all anticorrelate with constipation status ( Figure 5). On the other hand, out of the eight imbalanced bacteria taxa that were uniquely associated with constipated NT individuals, five (62.5%) belong to the Bacteroidetes phylum, and only two (25%) were in the Firmicutes phylum. All of these bacteria anticorrelate with constipation status. The decreasing of Faecalibacterium, Gemmiger, and an unknown bacterium were associated with constipation status in both ASD and NT individuals ( Figure 5). icicoccus). These five bacteria genera belong to the Firmicutes phylum and all anticorrelate with constipation status ( Figure 5). On the other hand, out of the eight imbalanced bacteria taxa that were uniquely associated with constipated NT individuals, five (62.5%) belong to the Bacteroidetes phylum, and only two (25%) were in the Firmicutes phylum. All of these bacteria anticorrelate with constipation status. The decreasing of Faecalibacterium, Gemmiger, and an unknown bacterium were associated with constipation status in both ASD and NT individuals ( Figure 5).

Discussion
ASD is a complex neurological and developmental disorder often followed by various comorbidities, including GI symptoms. Studies have shown that autistic individuals harbor an altered bacterial gut microbiota [20][21][22][23]; however, there has been little consensus on specific bacterial species that were similarly altered across autistic individuals. Most studies were done without ruling out the effect existing comorbidities potentially had on the composition of the gut microbiome, which left the etiology of ASD elusive. To the best of our knowledge, we are the first to investigate the role gut microbiota play on ASD and its comorbidities independently. The data released by Strati et al. [13] enabled us to assess microbial populations associated with psychiatric impairment and GI symptoms separately.
Among the eight additional bacterial taxa that were found correlating with the 40 ASD patients tested, family Lactobacillaceae was present in the patients but not found in NT individuals. This family has been observed as one of the key families in differentiating ASD from healthy children [24]. By analyzing the microbes in ASD and NT subjects with or without constipation status, bacteria specifically contributing to ASD symptoms, constipation, and both were identified. Several studies have shown that gut dysbiosis could lead to ASD-like and GI symptoms through independent mechanisms. The early-life gut microbiome could influence later neurodevelopment to a great extent in animal models. These bacteria secrete neurotransmitters which interferes with hosts' central neural path-

Discussion
ASD is a complex neurological and developmental disorder often followed by various comorbidities, including GI symptoms. Studies have shown that autistic individuals harbor an altered bacterial gut microbiota [20][21][22][23]; however, there has been little consensus on specific bacterial species that were similarly altered across autistic individuals. Most studies were done without ruling out the effect existing comorbidities potentially had on the composition of the gut microbiome, which left the etiology of ASD elusive. To the best of our knowledge, we are the first to investigate the role gut microbiota play on ASD and its comorbidities independently. The data released by Strati et al. [13] enabled us to assess microbial populations associated with psychiatric impairment and GI symptoms separately.
Among the eight additional bacterial taxa that were found correlating with the 40 ASD patients tested, family Lactobacillaceae was present in the patients but not found in NT individuals. This family has been observed as one of the key families in differentiating ASD from healthy children [24]. By analyzing the microbes in ASD and NT subjects with or without constipation status, bacteria specifically contributing to ASD symptoms, constipation, and both were identified. Several studies have shown that gut dysbiosis could lead to ASD-like and GI symptoms through independent mechanisms. The earlylife gut microbiome could influence later neurodevelopment to a great extent in animal models. These bacteria secrete neurotransmitters which interferes with hosts' central neural pathways [25]. Similarly, evidence from a human study showed a strong linkage between infant gut microbes and childhood neurodevelopmental outcomes [26]. Other evidence suggests that alterations of bacteria may also lead to GI symptoms by influencing intestinal motility and secretory functions through changing the metabolic environment in the gut [27,28]. Taking the above evidence together, we could reasonably hypothesize that Ga_c are risk factors for both ASD and GI symptoms ( Figure 6). Due to the fact that ASD patients are at greater risk of constipation, the possibility that the presence of Ga_c positively confounds the association between ASD and constipation cannot be underestimated. The potential confounding effect of Ga_c may account for all or part of the association. Levels of the effect could be tested in animal models by adjusting their intestinal microbial composition. of the effect could be tested in animal models by adjusting their intestinal microbial composition. We found that Bacteroidales; Unknown ranked third (p = 0.0022) among Ga. The fold change of its relative abundance was 0.14 in AD-NC compared to NT-NC. Bacteroidales; Unknown was also the second most significantly reduced bacteria in Gc (p = 0.0016). The fold change of its relative abundance in NT-C versus NT-NC was 0.12. That the lack of Bacteroidales could possibly lead to ASD and constipation can be supported by a study done by Hsiao et al. [29]. They demonstrated that introducing human commensal Bacteroides fragilis to a mouse model of ASD could reverse many of the behavioral and gastrointestinal changes. Sporobacter was another bacterial taxon that was significantly altered in Ga. In fact, it was completely absent in AD-NC. Multiple studies have shown Sporobacter significantly decreased in ASD children [12,30], indicating that the decreasing of Sporobacter could be one of the biomarkers for ASD core symptoms. Together with the findings provided by Strati et al., these results lead us a step closer to understanding the crosstalk between the central nervous system and the gut microbiota.
Several literatures have shown a decrease in the Bacteroidetes/Firmicutes ratio in the fecal samples of autistic children [13,31,32]. In our study, we have found that the alteration was even greater between AD-NC and NT-NC, indicating that bacteria within the Firmicutes phylum contributed substantially to the core symptoms of ASD. Firmicutes is known We found that Bacteroidales; Unknown ranked third (p = 0.0022) among Ga. The fold change of its relative abundance was 0.14 in AD-NC compared to NT-NC. Bacteroidales; Unknown was also the second most significantly reduced bacteria in Gc (p = 0.0016). The fold change of its relative abundance in NT-C versus NT-NC was 0.12. That the lack of Bacteroidales could possibly lead to ASD and constipation can be supported by a study done by Hsiao et al. [29]. They demonstrated that introducing human commensal Bacteroides fragilis to a mouse model of ASD could reverse many of the behavioral and gastrointestinal changes. Sporobacter was another bacterial taxon that was significantly altered in Ga. In fact, it was completely absent in AD-NC. Multiple studies have shown Sporobacter significantly decreased in ASD children [12,30], indicating that the decreasing of Sporobacter could be one of the biomarkers for ASD core symptoms. Together with the findings provided by Strati et al., these results lead us a step closer to understanding the crosstalk between the central nervous system and the gut microbiota.
Several literatures have shown a decrease in the Bacteroidetes/Firmicutes ratio in the fecal samples of autistic children [13,31,32]. In our study, we have found that the alteration was even greater between AD-NC and NT-NC, indicating that bacteria within the Firmicutes phylum contributed substantially to the core symptoms of ASD. Firmicutes is known for its function of degrading polysaccharides and converting them into carbohydrates and other energy products [33]. Although carbohydrates serve as one of the main fuels for our brains, excess carbohydrates could cause numerous neurological conditions such as Alzheimer's disease, Parkinson's disease, depression, and others [34][35][36][37]. It is likely that an abnormal amount of Firmicutes could lead to a surplus amount of carbohydrates in the bloodstream, causing neurological and behavioral problems associated with ASD. This also explains the high rate of obesity among teens with autism [38] as a higher Firmicutes/Bacteroidetes ratio has been linked to excess body fat in previous studies [39][40][41].
The effects of microbial composition and ASD can be mutual. Not only can microbial dysbiosis contribute to ASD and constipation, but intestinal microbes could also be influenced through one's dietary habits and further cause constipation. In other words, the causal effects of ASD and constipation are mediated by the altered microbes ( Figure 6). Although the connection is poorly understood, diet does affect the composition of gut bacteria [42]. It is well known that most ASD patients are very sensitive to not just the flavor, but also the color, smell and texture of foods, resulting in a strong preference for a narrow selection of foods. Many prefer having carbohydrates and processed foods and dislike high-fiber food such as fruits and vegetables. This behavior may lead to lower beneficial Firmicutes and higher levels of mucosa-associated Proteobacteria ( Figure 5) [43]. Previous studies have shown that mice fed with a high-fat diet showed a marked decrease of Turicibacter in abundance. In our study, the decreasing of Turicibacter was the most significant (p = 0.0086) among all bacteria associated with constipation status in ASD. Dialister was another bacterium with a significant decrease in constipated ASD patients compared to non-constipated ones. This difference was not found between healthy constipated and non-constipated individuals. The dislike of high-fiber food in ASD may result in the reduction of Dialister. A human study has demonstrated that a daily dose of 60 g of whole-grain barley (WGB) would enrich the genera of Dialister [44]. Roseburia and Butyricoccus were another two beneficial bacteria which were found significantly reduced in constipated ASD patients in our study. This again might be due to selective eating habits. One study found that a high-animal-protein diet, which is high in fat, could result in reduced Roseburia in the tested subjects [45]. It is worth noting that the dietary component of pregnant women also highly correlates with the health status of the baby. It has been suggested that the lack of Bacteroidetes may contribute to malnourishment via a reduction in ability to ferment glycans and generate short-chain fatty acids (SCFAs) [46]. Maternal malnutrition jeopardizes the neurodevelopment of the fetus, which could cause prematurity or growth restriction of the baby [47]. It has been shown that preterm babies have a significantly greater risk of ASD [48].
The published data adapted in our study was collected from an Italian research group in which stool samples were produced by individuals on Mediterranean-based diets. Information regarding a more detailed diet preference in ASD and healthy subjects was not available in the original article. Whether the diet one consumes could alter the ASD-specific signature which ultimately leads to constipation has yet to be analyzed. It is worth noting that subjects with different ethnicities could yield different results due to their unique genetics and dietary habits. One limitation of using summary data lies within the missing resolution on the individual level, making it difficult to conduct pathway analysis. It would be interesting to include this type of analysis in our future studies, which could possibly reveal the underlying mechanisms of how microbial functionalities' change is associated with ASD and constipation. Another limitation resulting from using summary data is insufficient statistical power. Analysis of data on the individual level would be our next step to increase the power of our study. Lastly, the data analyzed in this study was generated from bacterial 16S rRNA, in which the resolution was at best at the genus level. Bacterial species under the same genus produce metabolites with various functions and might have different contributions to ASD and its comorbidities. In order to examine results at a higher resolution, analyzing metagenomic data would be a good approach.

Conclusions
To the best of our knowledge, we are the first to raise the hypotheses of confounding and mediating effects of gut microbiota to ASD and constipation. In summary, we were able to identify ASD-correlated bacteria which failed to be detected in the original study. Our results suggest multi-directional and complex interactions between ASD, microbiome, and constipation. Children with a low abundance of Bacteroidetes may have a higher chance of developing ASD and constipation simultaneously; however, further case-parent trio studies or comparisons between AD children and their NT siblings need to be done in order to confirm the causality. In the case of children who are already on the spectrum, low abundance of Firmicutes may be one of the major causes of constipation. In addition, the relative abundances of six unique diet-related bacteria were found to be significantly lower in constipated ASD subjects. Longitudinal studies are necessary to further confirm the relevance of these microbes in ASD and constipation through diet. Institutional Review Board Statement: Ethical review and approval were waived for this study, due to the secondary analysis of the existing and de-identified data.
Informed Consent Statement: Patient consent was waived due to the secondary analysis of the existing and de-identified data.

Data Availability Statement:
No new data were created or analyzed in this study. Data sharing is not applicable to this article. other 4 (10%) were classified as moderately severe based on CARS scores. There were no comorbidities other than constipation found in the autistic cohort with an exception of celiac disease in 2 patients (5%). Rome III criteria was used to define constipation. The criteria must be fulfilled for the past 3 months, with onset of symptoms at least 6 months before diagnosis.

Appendix A.3. Measurement of Microbiome
In order to understand the gut microbial composition in an individual, one's fecal samples were collected and sent for total DNA extraction. Bacterial 16S rRNA genes were subsequently amplified by PCR using a specific primer set, which contains barcodes (unique sequences) that could ligate to each individual's samples' genetic material before samples were mixed together. Once the final amplicon library is constructed, high throughput DNA sequencing (454 pyrosequencing on the GS FLX+ system using the XL+ chemistry) was carried out by following the manufacturer's instructions. The sequencing data obtained were demultiplexed (processing steps in which the barcode information shows which sequences came from which samples after being sequenced together) and submitted to the European Nucleotide Archive (ENA) with accession numbers PRJEB15418.
MICCA pipeline (v. 0.1) was used to preprocess the reads. The RDP classifier version 2.7 was used to assign reads to the corresponding taxonomy model. Template-guided multiple sequence alignment (MSA) was performed using PyNAST (v. 0.1) against the multiple alignment of the Greengenes database (release 13_05) filtered at 97% similarity.
For more details on total DNA extraction from stool samples, sequencing procedures on the bacterial 16S rRNA genes, and how the reads were processed were explained in the section of "Pyrosequencing and data analysis" under Methods of Strati et al., 2017.