**Multivariate Analysis of Fecal Metabolites from Children with Autism Spectrum Disorder and Gastrointestinal Symptoms before and after Microbiota Transfer Therapy**

#### **Fatir Qureshi 1,2, James Adams <sup>3</sup> , Kathryn Hanagan <sup>4</sup> , Dae-Wook Kang 5,**† **, Rosa Krajmalnik-Brown 5,6,7 and Juergen Hahn 1,2,8,\***


Received: 18 August 2020; Accepted: 25 September 2020; Published: 2 October 2020

**Abstract:** Fecal microbiota transplant (FMT) holds significant promise for patients with Autism Spectrum Disorder (ASD) and gastrointestinal (GI) symptoms. Prior work has demonstrated that plasma metabolite profiles of children with ASD become more similar to those of their typically developing (TD) peers following this treatment. This work measures the concentration of 669 biochemical compounds in feces of a cohort of 18 ASD and 20 TD children using ultrahigh performance liquid chromatography-tandem mass spectroscopy. Subsequent measurements were taken from the ASD cohort over the course of 10-week Microbiota Transfer Therapy (MTT) and 8 weeks after completion of this treatment. Univariate and multivariate statistical analysis techniques were used to characterize differences in metabolites before, during, and after treatment. Using Fisher Discriminant Analysis (FDA), it was possible to attain multivariate metabolite models capable of achieving a sensitivity of 94% and a specificity of 95% after cross-validation. Observations made following MTT indicate that the fecal metabolite profiles become more like those of the TD cohort. There was an 82–88% decrease in the median difference of the ASD and TD group for the panel metabolites, and among the top fifty most discriminating individual metabolites, 96% report more comparable values following treatment. Thus, these findings are similar, although less pronounced, as those determined using plasma metabolites.

**Keywords:** fecal metabolites; ASD; microbiome; gastrointestinal symptoms; Fisher Discriminant Analysis

### **1. Introduction**

Autism spectrum disorder (ASD) encompasses a large group of early onset neurological conditions that result in impairments in social behavior and communication, which are estimated to affect 1 in 54 children under the age of eight in the United States [1]. Despite this high rate of occurrence, the understanding of the pathophysiology of ASD is still poor, and it is believed that at least in some cases ASD begins prenatally as a result of complex interactions between environmental and genetic factors [2,3]. Although diagnosis of this disorder is only made through behavioral evaluations, many systems of the body are strongly affected by this condition. A diverse range of physiological mechanisms have been observed to be perturbed in ASD including the immune, endocrine, and gastrointestinal (GI) systems [4,5]. Notably, the prevalence of GI symptoms co-occurring with ASD (~46%) lends significant credence to investigating the relationship of ASD to the GI system [6].

In recent years, there have been growing efforts to study the effect of the microbiome on the Gut-Brain Axis in the context of ASD etiology. Some studies have shown that the gut microbiome of individuals with GI issues varies significantly from those without such complications [7–9]. However, the microbiota of individuals with ASD without the presence of GI issues have also consistently been found to be distinct from their typically developing (TD) peers [10,11]. Certain genera such as Prevotella and Coprococcus have been shown to be significantly less prevalent in the gut of children with ASD [12,13]. Furthermore, it has been proposed that the microbiota differences in children with ASD give rise to metabolomic differences that can be quantitatively evaluated to distinguish them from their TD peers [14,15].

Some previous work involving fecal metabolites identified isopropanol, p-cresol, acetyl-carnitine, free carnitine and neurotransmitters-gamma-Aminobutyrate (GABA) as metabolites that have significantly different concentrations between the ASD and TD cohorts [14,16,17]. There have also been mixed results regarding the fecal concentrations of short chain fatty acids. While some studies show that the fecal concentration of acetic, propionic and butyric acids were higher in children with ASD [18–20], other investigations found that the concentration of these short chain fatty acids were lower or comparable to their TD peers [21–23].

As the role of the microbiome in ASD is being in more detail, the question is raised as to whether using fecal microbiota transplant (FMT) can mitigate the severity of GI and other symptoms of ASD. In one notable study, offspring of germ-free mice subject to microbiome transfer from individuals with ASD exhibited more ASD-like behaviors and produced different metabolome profiles when compared to offspring of germ-free mice subject to microbiome transfer from TD controls [24]. The use of FMT has shown considerable potential in its capability to alleviate not only symptoms associated with GI complications, but also in some cases to reduce the severity of certain behavioral symptoms in children with ASD. For example, Kang et al. demonstrated in an open-label study that through a modified FMT (called, Microbiota Transfer Therapy (MTT)), there was an 80% reduction in GI symptoms and a 24% initial reduction in core ASD symptoms, with greater improvement in ASD symptoms at a two-year follow-up [25,26]. Probiotic intervention has also shown potential to have a positive influence on ameliorating both behavioral and GI symptoms in individuals with ASD [27,28].

Past work in analyzing metabolites prior and subsequent to MTT therapy have also yielded promising results. Children with ASD who underwent MTT presented changes in their plasma metabolite profiles to resemble more closely those of their typically developing peers [29,30]. The work presented in this paper builds on the analysis of this same study [25], but focuses on fecal metabolites instead of plasma metabolites. Univariate assessment of the fecal metabolites examined in this study have previously shown limited capability for differentiating between ASD and TD cohorts when corrected for multiple hypotheses [30]. Thus, here we explored the use of multivariate techniques to detect underlying relationships that may have been otherwise missed.

#### **2. Materials and Methods**

#### *2.1. Study Design*

The purpose of this study was to examine the differences in gut metabolites between children with ASD and GI problems vs. typically developing children without GI problems, and determine the effects of gut microbiota transfer therapy on the fecal metabolites of the ASD group. The study involved 38 children, aged 7–16 years, 18 of these professionally diagnosed with ASD by a healthcare provider (verified with the Autism Diagnostic Interview-Revised) and 20 determined to be typically developing. The participants with ASD were required to have moderate to severe GI problems, and the range of GI issues included constipation, diarrhea, and alternating diarrhea/constipation. GI symptoms were assessed biweekly with the Gastrointestinal Severity Rating Scale (GSRS) and daily with a Daily Stool Record using the Bristol Stool Form scale [25]. The study consisted of 2 weeks of antibiotic therapy, 1 day of bowel cleans, and a high major initial dose and 7–8 weeks of lower maintenance doses of FMT treatment followed by evaluation at 8 weeks post treatment. The TD group did not undergo MTT, but instead was used as a comparison group whose measurements were taken at the same time as the ASD group before treatment. The MTT experimental protocol and details of the study population are outlined in Kang et al. [25].

The pre-treatment protocol consisted of two weeks of oral vancomycin, which is a broad spectrum non-absorbable antibiotic. This treatment was intended to reduce pathogenic bacteria and prime the GI system for MTT. The dose of vancomycin administered was individualized to the weight of each participant at 40 mg/kg, with a maximum dose of 2 g [25]. Participants were then subjected to one day of fasting and a bowel cleanser (MoviPrep) in order to remove the vancomycin and further reduce levels of intestinal bacteria. Standardized Human Gut Microbiota (SHGM) consisted of a full spectrum of highly purified microbiota from healthy, carefully screened donors. The ASD cohort was split into two groups, each one following a different initial high dose (2.5 × 10<sup>12</sup> cells/day) SHGM treatment. One MTT treatment consisted of a single dose administered rectally (*n* = 6) while the other involved doses administered orally on two days (*n* = 12). Both techniques were followed by a lower concentration SHGM maintenance dosage (around 2.5 × 10<sup>9</sup> cells) given orally, with treatment ending 8 weeks after the initial high dose [25]. However, the protocol differed slightly for both groups of ASD children as those that received SHGM rectally waited for one week prior to beginning low dose SHGM.

#### *2.2. Metabolite Measurements*

Once the study had concluded, aliquots of the fecal samples were shipped overnight on dry ice to Metabolon (Durham, NC, USA). Both the control and autism samples were blinded and randomized prior to being shipped. Metabolon utilized ultrahigh performance liquid chromatography-tandem mass spectroscopy (UHPLC-MS/MS) instruments for obtaining metabolomic information on 669 metabolites. A detailed overview regarding this protocol can be found in Long et al. [31]. By using this technique, it is possible to determine a signal intensity corresponding to a metabolite's presence in a sample. Subsequently, the signal intensity is used to derive the relative abundance of each metabolite. For this objective, peak area integration using the area under the curve was utilized. In the case of missing values, imputation was performed by taking the lowest value of each compound measurement divided by the square root of 2.

Fecal samples were taken at four time points from the participants with ASD (Figure 1). Parents were instructed to freeze these sample immediately after collection for up to 3 days, and the samples were then transported to Arizona State University on dry ice where they were stored in a −80 ◦C freezer. Initial fecal samples were collected from all participants at Week 0. Samples were also taken from ASD participants at the Week 3 mark from the beginning of the treatment (after about five days of microbiota transplant) and at the end of MTT treatment (Week 10). The ASD group was sampled again 8 weeks after administration of SHGM ceased (Week 18). In total, 18 ASD participants collected samples at all time points aside from Week 3, where only 17 samples were collected. The TD group received no treatment and 20 were sampled at the beginning (Week 0).

#### *2.3. Statistical Analysis*

The data collected for each of the metabolites underwent various forms of statistical analysis to assess differences between the ASD + GI and TD cohort. By comparing the differences observed for metabolites before and after MTT, it might be possible to gain some understanding of the role that this therapy could play in altering metabolic processes. Both univariate and multivariate techniques were used in this regard, and the implementation of the analysis routines was done in MATLAB. *J. Pers. Med.* **2020**, *10*, x FOR PEER REVIEW 4 of 28

**Figure 1.** Timeline for the experimental protocol, which can be divided into three main phases. From Week 0 to Week 3, the Autism Spectrum Disorder (ASD) cohort is primed for Microbiota Transfer Therapy (MTT). From Week 3 to Week 10 the ASD cohort receives low dose fecal microbiota (FM) or is prepared for low dose FM, and finally from Week 10 to Week 18 no treatment is given to the individuals. A closeup is provided of Week 2–3 as this is when Standardized Human Gut Microbiota (SHGM) is initialized. **Figure 1.** Timeline for the experimental protocol, which can be divided into three main phases. From Week 0 to Week 3, the Autism Spectrum Disorder (ASD) cohort is primed for Microbiota Transfer Therapy (MTT). From Week 3 to Week 10 the ASD cohort receives low dose fecal microbiota (FM) or is prepared for low dose FM, and finally from Week 10 to Week 18 no treatment is given to the individuals. A closeup is provided of Week 2–3 as this is when Standardized Human Gut Microbiota (SHGM) is initialized.

#### *2.3. Statistical Analysis*  2.3.1. Preprocessing

The data collected for each of the metabolites underwent various forms of statistical analysis to assess differences between the ASD + GI and TD cohort. By comparing the differences observed for metabolites before and after MTT, it might be possible to gain some understanding of the role that this therapy could play in altering metabolic processes. Both univariate and multivariate techniques were used in this regard, and the implementation of the analysis routines was done in MATLAB. 2.3.1. Preprocessing In order to ensure continuous distribution of values across all participants, metabolites with too In order to ensure continuous distribution of values across all participants, metabolites with too many values below the detection limit at their initial stool sample (Week 0) were removed. The detection limit for a metabolite was determined to be the minimum value recorded for that metabolite. If less than 40% of all measurements were above the detection limit, the metabolite was removed from subsequent analysis. This step accounted for the possibility that a measurement could be almost entirely below the detection limit in one cohort, while simultaneously being above the limit in the other cohort. The remaining metabolites were then normalized such that for each metabolite the median value was 1.0 in the Week 0 TD cohort.

#### many values below the detection limit at their initial stool sample (Week 0) were removed. The detection limit for a metabolite was determined to be the minimum value recorded for that 2.3.2. Univariate Analysis

metabolite. If less than 40% of all measurements were above the detection limit, the metabolite was removed from subsequent analysis. This step accounted for the possibility that a measurement could be almost entirely below the detection limit in one cohort, while simultaneously being above the limit in the other cohort. The remaining metabolites were then normalized such that for each metabolite the median value was 1.0 in the Week 0 TD cohort. 2.3.2. Univariate Analysis Univariate analysis identifies metabolites that are differentially expressed between the ASD and Univariate analysis identifies metabolites that are differentially expressed between the ASD and TD cohorts. Using this information, it is possible to examine common correlations and relationships across different measurement quantities. In turn, this has the potential to identify underlying mechanisms of ASD etiology as well as provide guidance for the development of a multivariate model that can more accurately distinguish between both groups. As there are 669 metabolites under investigation, there is significant concern related to overfitting of statistical models if many or all these measurements are used to develop a multivariate model. By reducing the number of measurements to a smaller subset, it is possible to alleviate some of the concerns related to overfitting.

TD cohorts. Using this information, it is possible to examine common correlations and relationships across different measurement quantities. In turn, this has the potential to identify underlying mechanisms of ASD etiology as well as provide guidance for the development of a multivariate model that can more accurately distinguish between both groups. As there are 669 metabolites under investigation, there is significant concern related to overfitting of statistical models if many or all Metabolites were individually analyzed for their ability to classify between the ASD and TD cohorts at their Week 0 measurements. The area under the receiver operator curve (AUROC) served as an assessment of the potential of a metabolite to distinguish between ASD and TD groups. This metric is defined as the false positive rate against the false negative rate at different ASD/TD classification thresholds. An AUROC of 1.0 indicates the capacity for perfect separation, while an AUROC of 0.5

these measurements are used to develop a multivariate model. By reducing the number of measurements to a smaller subset, it is possible to alleviate some of the concerns related to overfitting. indicates that there is no ability to distinguish between the groups. Metabolites with an AUROC value above 0.6 were selected as candidates for use in multivariate analysis in this work.

Univariate analysis techniques evaluated whether significant changes had occurred among the metabolites between the beginning and end of the study for the ASD cohort. For this purpose, the metabolite measurements at Week 0 and Week 18 were compared using a parametric or non-parametric test, depending upon their distribution. An Anderson-Darling test for normality was used at both time points to determine the distribution of each set of measurements. Subsequently, either a Wilcoxon signed-rank test or a paired t-test was performed on the ASD group, comparing measurements from Week 0 to Week 18. A relatively normal distribution employs the parametric paired t-test; otherwise, the non-parametric Wilcoxon signed-rank test is used. The resulting *p*-value indicates how significantly the concentration of the metabolite changed for the cohort over the course of the study.

As there were a considerable number of quantities measured per study participant, it was imperative that multiple hypothesis correction tests were utilized. Subsequently, a false discovery rate (FDR) for each individual metabolite was computed using leave-n-out (*n* = 1, 2, 3) cross validation (see Table A1). Leave-n-out is an iterative process and involves removing n individual data points from the total dataset and rerunning the univariate analysis on this subset. This procedure is repeated so that all possible combinations with n removed individuals are assessed. The FDR is calculated as the proportion of univariate results that were not deemed significant.

#### 2.3.3. Multivariate Analysis

Fisher discriminant analysis (FDA), metabolites that had been identified as having an AUROC value above 0.6 were used to develop a multivariate model for distinguishing between the ASD and TD cohorts. FDA is a dimensionality reduction technique that seeks to separate classes of data by determining a projection where such separation is maximized [32]. This is achieved by maximizing the ratio between the between-class variability *S<sup>B</sup>* and the within-class variability *S<sup>W</sup>* for a weight vector *W*.

$$J(W) = \frac{W^T S\_B W}{W^T S\_W W}$$

In the case of K classes with n number of samples and m measurements, *S<sup>B</sup>* is defined as follows, where *x* denotes the global mean, *x<sup>k</sup>* denotes the local class mean, and *n<sup>k</sup>* is the number of samples within class *k*:

$$S\_B = \sum\_{k=1}^{K} n\_k \left(\overline{\boldsymbol{x}}\_k - \overline{\boldsymbol{x}}\right) \left(\overline{\boldsymbol{x}}\_k - \overline{\boldsymbol{x}}\right)^T$$

In contrast, the within-class covariance matrix *S<sup>W</sup>* is defined as the following, where *x<sup>i</sup>* corresponds to an individual sample:

$$S\_W = \sum\_{k=1}^K n\_k \sum\_{i \in k} (\overline{\boldsymbol{\pi}}\_i - \overline{\boldsymbol{\pi}}\_k) \left(\overline{\boldsymbol{\pi}}\_i - \overline{\boldsymbol{\pi}}\_k\right)^T$$

Thus, FDA simultaneously maximizes the scatter between classes and minimizes the scatter within each class to find *k*-1 vectors that maximize the objective function. Subsequently, the eigenvector corresponding to the *k*-1 largest eigenvalue of *SBS<sup>W</sup>* corresponds to the optimum weight vector.

For this dataset, the objective of FDA is to separate the ASD and TD cohorts with a combination of metabolites. The initial stool samples (Week 0) were used to develop these models, so that the model classifies individuals before any treatment.

The previously mentioned preprocessing and univariate analysis steps were performed to reduce the set of metabolites considered for FDA. An FDA model could potentially be created with all 669 metabolites, but this model would likely overfit the data. To account for this, only metabolites that passed the preprocessing step and achieved a univariate AUROC of over 0.6 remained in consideration. This resulted in 165 metabolites under further investigation.

An exhaustive search was performed through all possible combinations of 2, 3, or 4 metabolites of the reduced set of 165 metabolites to determine the models which best separate the ASD and TD groups at Week 0. The AUROC was used here as well, measuring how well the multivariate models classify the two groups. Using kernel density estimation, the probability density function of each model was computed. Iterating through all combinations, the models were assessed, and the combination of metabolites was determined for each number of variables. For each number of metabolites, the 1000 models that had achieved the highest possible AUROC were recorded. To derive the five-metabolite models, all 1000 four metabolite models that had achieved the highest AUROC were augmented with each of the remaining 161 metabolites that had an AUROC greater than 0.6. The top 1000 five metabolite models that had the highest AUROC were then subjected to leave-one-out cross validation.

#### *2.4. Cross-Validation*

Leave-one-out cross-validation was performed on the optimal FDA models to evaluate robustness and statistical independence. Cross-validation ensures that, rather than merely fitting a model to presented data, the model obtained is also capable of classifying new data. Although cross-validation generally has a lower accuracy than what is computed just by fitting a model to data, the cross-validation accuracy will better reflect generalizability to new data sets, i.e., data not used for developing a classifier. Leave-one-out cross-validation proceeds iteratively, as a single individual's data is removed from the total dataset, then an FDA model is computed with measurements from the remaining individuals [33]. The measurements from the removed individual are now used as a test case to determine if the model prediction regarding classification is correct. This process is repeated for measurements from each of the individuals in the dataset: their data are removed, a model is developed with the remaining data, then they are classified with this model, until the data for each individual has been removed once. A confusion matrix is computed which includes the true positive rate (TPR), or sensitivity, and the true negative rate (TNR), or specificity. Additionally, for each model, the Type II (false negative) error β was modulated between 0.01, 0.05, 0.1, and 0.2 during cross-validation. The Type II error determined the threshold value for separating the two groups. By alternating the values of β, it was possible to evaluate the cross-validated performances along different positions of the ASD distribution. Lowering β meant raising the Type I error while lowering the Type II error and the converse also holds true. Thus, each of the four models (2, 3, 4, or 5 metabolites) had cross-validation performed four times, with corresponding computation of TPRs and TNRs.

#### *2.5. Model Evaluation*

The models obtained after cross validating at different thresholds for data collected at Week 0 were used to make predictions about the ASD group at the other MTT time points. Specifically, measurements at Week 3, Week 10 and Week 18 were used to monitor the change in classification performance over the course of the MTT protocol. Data from these time points were rescaled with respect to the TD Week 0 median and standard deviation. The probability density functions were compared between the time points, and the discriminate scores for each model as well as of their constituent metabolites were determined. Changes resulting from MTT were quantified using the Type II error, with respect to the threshold associated with the probability density function (PDF) of the ASD + GI cohort's discriminant scores at each time point. Thus, both univariate assessments were performed as well as the total assessment of the multivariate models' discriminant score. Additionally, correlation analysis between significant metabolite pairs was performed to determine possible underlying relationships.

#### **3. Results**

#### *3.1. Univariate Analysis*

In total, there were 669 fecal metabolites that were measured in the study. Through the preprocessing step, 86 metabolites were determined not to have the prerequisite number of observations above the detection limit for further analysis. In order to classify ASD and TD cohorts at their Week 0 measurements, the area under the AUROC was used as an assessment of the potential of a metabolite to distinguish between ASD and TD groups. The remaining 583 metabolites were ranked according to their univariate AUROC, and 165 metabolites with an AUROC of at least 0.6 were identified. No single metabolite perfectly separated the cohorts (which would correspond to an AUROC = 1.0), as the metabolite with the highest AUROC, carnitine, achieved a value of 0.77 (Table 1).

Using the 165 metabolites with an AUROC greater than 0.6, additional univariate testing was performed to assess the degree to which measurements shifted following MTT. ASD metabolite samples measured at Week 0 were compared to their values following MTT at Week 18 using either a paired t-test or a Wilcoxon signed-rank test depending upon the distribution determined for the data via the Anderson-Darling normality test. It was found that 10.9% of the metabolites significantly changed (*p* < 0.05) following the MTT therapy when comparing the ASD group before and after treatment (see Table A1). The metabolites that had a threshold AUROC value of 0.6 were subsequently used for model discovery for the 2-, 3-, 4- and 5- metabolite models.

#### *3.2. FDA Models*

The FDA models with the greatest AUROC values for each number of constituent metabolites are listed in Table 2. The probability density function (PDF) of discriminant scores for the 2-, 3-, 4-, and 5-metabolite models that achieved the highest accuracy following cross validation are shown in Figure 2. There were two distinct models that were identified, using five separate metabolites, as having achieved the same accuracy after cross-validation. With the exception of one metabolite which differed between them (Adenosine and Indole), the constituents of these panels are identical. These two metabolite models are both shown in Table 2 and will be referred to as OFM-A and OFM-I, optimized fecal model-adenosine and optimized fecal model-indole, respectively (OFM-I/A). For all optimized metabolite panels, the TPR and TNR values for each are presented when the β value was modulated. The 5-metabolite models had higher AUROCs than the 2-, 3-, and 4-metabolite models, so they are the focus of the following analysis, due to their higher accuracy (0.95 specificity and 0.94 sensitivity). Modulating β revealed that the optimal cut-offs between the ASD and TD distributions for the OFM-I and OFM-A models was β = 0.05 for both the OFM-A and OFM-I.

For the 1000 best models with five metabolites, the AUROC ranged from 0.97 to 1.00 which are high values. The reason for using the 1000 best models is that there are not only one or two best models as judged by AUROC alone. Each of these models was subjected to cross validation, with OFM-I/A being derived from those that achieved the highest accuracy. The metabolites ultimately utilized for the development of a five-component model were all found to be in the top quartile of prevalence in the 1000 top models (Figure 3). Notably, among the top fecal metabolite models, adenosine and hydroxyproline appeared in 36.3% and 62.4% of models, respectively. Only three metabolites were present in more than 25% of the top 1000 models that were not among those included in the OFM-I/A panels. These metabolites were Adenine, 2-aminobutyrate and 1,7-dimethylurate (corresponding to the 5th, 57th, and 86th highest AUROC rank, respectively).





**Table 2.** Fitting and cross-validation results for the best combinations of two, three, four, and five metabolites used as part of Fisher Discriminant Analysis (FDA). The cross-validated true positive rate (TPR) and true negative rate (TNR) are shown for classification thresholds associated with different values of β calculated from the fitted probability density functions (PDFs). The results for two distinct 5-metabolite models are presented as they were able to achieve the same accuracy following cross-validation. The notable TPRs and TNRs are highlighted for the 5-metabolite models.


#### *3.3. Correlation Analysis*

Correlation analysis was performed on the OFM metabolites as these were the ones that had been identified as being able to distinguish between the ASD and TD cohorts with the highest accuracy after cross-validation. It can be observed that many of the top 50 metabolites (AUROC ≥ 0.66) were significantly correlated with the OFM metabolites (Table 3). In contrast, the individual OFM metabolites for both models had little to no correlation with each other, apart from hydroxyproline with adenosine and 2-hydroxy-3-methylvalerate; these findings were expected since, if individual OFM metabolites were highly correlated with each other, then they would not be useful in the model due to their correlation.

#### *3.4. Assessing E*ff*ects of MTT*

Univariate assessment of the top 50 metabolites as ranked by AUROC demonstrated that 14% of these 50 metabolites showed significant differences in their Week 0 and Week 18 ASD measurements and that 47 of these 50 metabolites achieved a lower AUROC eight weeks following treatment (Table 1). In addition to classification at baseline, the multivariate models developed can be used to observe changes in fecal metabolome composition over the course of the study. Most metabolites in the OFM-I and OFM-A models changed significantly after MTT and have values closer to the TD group after MTT (see Table 4). The average difference between the median of the five metabolites for TD group at Week 0 and the ASD measurements at Week 18 compared to measurements at Week 0 diminished by 88% and 82% for the OFM-I and OFM-A models (see Table 4), so the ASD group became much more similar to the TD group.

*J. Pers. Med.* **2020**, *10*, x FOR PEER REVIEW 11 of 28

Indole

Hydroxyproline 0.05 0.94 0.95 Theobromine 0.10 0.89 0.95 2-hydroxy-3-methylvalerate 0.20 0.78 0.95

**Figure 2.** PDFs of ASD and TD discriminant scores at Week 0. The probability density function of the FDA score provides a visualization of a model's ability to distinguish between the ASD and TD cohorts. The (**a**) two-metabolite model has most of its FDA scores highly concentrated near the region where thresholds would be applied. The (**b**) three-metabolite model is not as highly concentrated, but there is a significant amount of overlap between the scores of the ASD and TD participants, which is visible in both plots. The four (**c**) and five (**d**,**e**) metabolite models better separated the cohorts, with little overlap in the discriminant scores of the ASD and TD groups. For the 1000 best models with five metabolites, the AUROC ranged from 0.97 to 1.00 which are high values. The reason for using the 1000 best models is that there are not only one or two best **Figure 2.** PDFs of ASD and TD discriminant scores at Week 0. The probability density function of the FDA score provides a visualization of a model's ability to distinguish between the ASD and TD cohorts. The (**a**) two-metabolite model has most of its FDA scores highly concentrated near the region where thresholds would be applied. The (**b**) three-metabolite model is not as highly concentrated, but there is a significant amount of overlap between the scores of the ASD and TD participants, which is visible in both plots. The four (**c**) and five (**d**,**e**) metabolite models better separated the cohorts, with little overlap in the discriminant scores of the ASD and TD groups.

models as judged by AUROC alone. Each of these models was subjected to cross validation, with

*J. Pers. Med.* **2020**, *10*, x FOR PEER REVIEW 12 of 28

**Figure 3.** Frequency of appearance of each of the metabolites with AUROC > 0.6 in the top 1000 fivemetabolite Fisher discriminant analysis (FDA) models. The metabolites are ranked from highest to lowest area under the receiver operating characteristic curve (AUROC) as shown in Table 1. The metabolites included in the FDA models which achieved maximal separation following crossvalidation are shown in red: (A) indole (B) imidazole Propionate (C) adenosine (D) theobromine (E) hydroxyproline, (F) 2-hydroxy-3-methylvalerate. *3.3. Correlation Analysis*  **Figure 3.** Frequency of appearance of each of the metabolites with AUROC > 0.6 in the top 1000 five-metabolite Fisher discriminant analysis (FDA) models. The metabolites are ranked from highest to lowest area under the receiver operating characteristic curve (AUROC) as shown in Table 1. The metabolites included in the FDA models which achieved maximal separation following cross-validation are shown in red: (A) indole (B) imidazole Propionate (C) adenosine (D) theobromine (E) hydroxyproline, (F) 2-hydroxy-3-methylvalerate.

in the model due to their correlation.

Correlation analysis was performed on the OFM metabolites as these were the ones that had been identified as being able to distinguish between the ASD and TD cohorts with the highest accuracy after cross-validation. It can be observed that many of the top 50 metabolites (AUROC ≥


**Table 3.** The correlation coefficients between the optimized fecal model-indole/adenosine (OFM-I/A) metabolites and top metabolites 50 metabolites are examined and presented in order of their AUROC. Only those correlations that are significant (*p*-value < 0.05) are presented.

(25th/75th percentile)

**4. Discussion** 

(0.47, 2.38)


**Table 3.** *Cont.*

\*\* Theobromine was not found to be significantly correlated with any of the top 50 metabolites. \*\*\* 2-hydroxy-3-methylvalerate is not among the top 50 metabolites as ranked by AUROC but present in both OFM-I/A panels. Hydroxyproline (25th/75th percentile) 0.96 (0.72, 4.83) 1.06 (0.42, 3.67) 1.27 (0.24, 3.34) 0.80 (0.54, 3.60) 0.60 (0.29, 1.20) Theobromine 0.89 0.47 0.47 0.47 0.46

(0.16, 0.47)

(0.43, 0.47)

(0.34, 0.64)

(0.47, 1.68)

The OFM-I/A models were applied to the ASD samples at all distinct time points to assess their accuracy for classifying a sample as belonging to the ASD or TD cohort. The effectiveness of OFM-I/A for classification changed significantly before and after MTT. The type II error rate was initially observed to be 5% for both models, indicating that the ASD and TD distributions are quite distinct, but was observed to rise to 56% eight weeks after MTT was completed (Table 4), thereby indicating that distinguishing between the ASD and TD cohort is not reliably possible after MTT. The PDF curves are shown in Figure 4 to demonstrate the changes in the ASD cohort over time with respect to the values of the FDA score. The distributions indicate that the ASD cohort became more metabolically similar to the TD cohort after treatment, since the curves are shifted towards the TD curve. Notably, the distribution of scores for the ASD cohort become somewhat bimodal at the later time points for both models. The discriminant score for both models decreased substantially as time progressed, indicating that the metabolites of the ASD group were becoming more similar to that of the TD group. 2-hydroxy-3 methylvalerate (25th/75th percentile) 0.53 (0.43, 0.75) 0.43 (0.18, 0.56) 0.34 (0.06, 0.50) 0.52 (0.21, 0.63) 0.44 (0.21, 0.61) Indole (25th/75th percentile) 1.13 (0.25, 1.83) 0.66 (0.32, 1.75) 0.85 (0.18, 1.86) 0.86 (0.26, 1.52) 0.39 (0.15, 0.59) Adenosine (25th/75th percentile) 0.67 (0.36, 0.88) 0.77 (0.50, 1.01) 0.73 (0.47, 0.90) 0.57 (0.26, 0.86) 0.40 (0.18, 0.86) OFM-I Median discriminant score (25th/75th percentile) 3.90 (2.33, 5.72) 1.90 (0.72, 9.52) 1.84 (0.90, 3.71) 1.73 (0.71, 2.62) 0.46 (0.21, 1.35) Type II error 5% 53% 50% 56% - OFM-A Median discriminant score (25th/75th percentile) 3.51 (2.28, 5.73) 2.87 (1.13, 9.43) 2.18 (1.07, 4.18) 1.36 (0.54, 2.44) 0.62 (0.35, 1.05) Type II error 5% 53% 39% 56% -

**Figure 4.** PDF curves for the (**a**) OFM-I and (**b**) OFM-A model when assessing the ASD cohort over the course of the study. The overlap between the TD cohort and the ASD cohorts increases at Week 10 and Week 18, indicating that the metabolite profile of the ASD group after MTT treatment has become more similar to the ones of the TD group. **Figure 4.** PDF curves for the (**a**) OFM-I and (**b**) OFM-A model when assessing the ASD cohort over the course of the study. The overlap between the TD cohort and the ASD cohorts increases at Week 10 and Week 18, indicating that the metabolite profile of the ASD group after MTT treatment has become more similar to the ones of the TD group.

classification [35]. However, this value was not chosen here as the AUROC is employed here as a prescreening tool for reducing the number of metabolites for classification, and not for determining a metabolite that by itself can distinguish between the two groups. The highest AUROC value for an individual metabolite was 0.77, corresponding to carnitine, indicating that the ASD group is

Preliminary analysis using univariate methods revealed that none of the individual fecal

**Table 4.** Change in the difference between OFM-I/A metabolites measured in the TD and ASD cohort over the course of the study. The discriminant score was calculated by first taking the absolute value of the difference between measurements at each time point and the median of the TD group, then normalizing the difference by the standard deviation of the TD Week 0 measurements, and then adding the normalized absolute difference for each of the five metabolites. The background color distinguished the individual metabolites from the multivariate models.


#### **4. Discussion**

Preliminary analysis using univariate methods revealed that none of the individual fecal metabolites achieved a high AUROC value by itself. Interpretations regarding the threshold value needed for an AUROC to be deemed an effective classifier vary by discipline. AUROC values between 0.9–1.0 are desirable for diagnostic tests and are seen to be reflective of excellent classification [35]. However, this value was not chosen here as the AUROC is employed here as a pre-screening tool for reducing the number of metabolites for classification, and not for determining a metabolite that by itself can distinguish between the two groups. The highest AUROC value for an individual metabolite was 0.77, corresponding to carnitine, indicating that the ASD group is somewhat heterogeneous. In contrast, all optimized multivariate models using three or more elements were able to achieve an AUROC greater than 0.9, highlighting that a multivariate analysis can provide better classification than that which can be determined using univariate analysis alone. Nonetheless, 94% of the top 50 univariate metabolites report lower AUROC (Table 1) eight weeks following MTT, which indicates greater similarity between the ASD and TD measurements after treatment.

Analysis of all possible significant metabolites at Week 0 resulted in the OFM-I and OMF-A models, consisting of five metabolites. Four of these five metabolites were identical between the two models and both achieved AUROC values greater than 0.99. Interestingly, the two metabolites that differed between them, Adenosine and Indole, are associated with different metabolic processes and have no significant correlation. Furthermore, cross-validation revealed that using the OFM-I/A models at the Week 0 timepoint resulted in a 0.95 TPR and 0.94 TNR. Subsequently, there was an overall 94.7% accuracy for correctly classifying an individual into the ASD/TD groups after leave-one-out cross-validation.

Many of the metabolites identified as being differentially expressed between the ASD and TD cohorts have also been previously examined for their relationship to ASD. Specifically, among the top five metabolites ranked by their AUROC value, carnitine, indole and sphingosine have all been found to be differentially expressed in some capacity among individuals with ASD [14,17,18,36]. In a meta-analysis, 10–20% of individuals with ASD were found to have disorders with synthesizing carnitine, which was the metabolite that had achieved the largest AUROC value [36]. Plasma carnitine concentration has also previously been shown to be lower among cohorts with ASD [36]. It should be noted that following MTT the AUROC values of carnitine reduced to 0.68, which indicates that the ASD and TD carnitine distributions were less different after MTT (see Table A1). Sphingolipids such as sphingosine have been found to play an active role in the crosstalk between microbiota and intestinal cells [37]. The significant change in concentration for metabolites such as N-palmitoyl-sphingosine (d18:1/16:0) may have been associated with the changed microbiome composition resulting from MTT (Table 1). Approximately 68% of the variance observed in the fecal metabolome can be explained by the gut microbiome [38], which underscores the potential impact FMT can have on reshaping metabolite concentrations.

Among the metabolites which form part of the OFM-I/A models, theobromine exhibited a significant change between its measurements at Week 0 compared to the Week 18 value for the ASD cohort when a sign ranked test was applied at both timepoints (Table 1). Theobromine is not a microbial metabolite, and its source in fecal samples likely stems from dietary intake and from human metabolism of caffeine [38]. Consequently, this may account for the reason why it was not observed to be correlated with any other metabolite and why the median discriminate score often took the value of the detection limit. However, the metabolization of theobromine is primarily via hepatic demethylation and oxidation, which are processes that have at least been hypothesized to be perturbed in ASD [39,40]. The median concentration for this metabolite was also not found to change following the bowel cleanse measurement (see Table A1). We conducted a secondary analysis of a five-metabolite model without theobromine and found that it results in significantly lower sensitivity and specificity, so including theobromine seems to be important for developing a classification model.

Nonetheless, in the case of all metabolites present in the OFM-I/A models, the average difference between the Week 0 TD measurements and ASD group decreased greatly (82–88%) by the end of the study (Table 4). Hydroxyproline, which is another of the OFM metabolites, has been previously determined to be expressed in significantly higher concentration in the plasma of children with ASD, consistent with the higher levels in feces [41] and in the present study. Indole, which was also one of the OFM metabolites, has been found in higher concentration in fecal samples in children with ASD and other neurodevelopmental conditions [17], consistent with the results of this study, and is an important metabolite for tryptophan metabolism [42]. Thus, the shift to a lower discriminant score following the completion of the treatment is consistent with measurements of ASD fecal metabolites becoming more like those of their TD counterparts following MTT.

The OFM-I/A models in their totality demonstrated similar behavior when contrasting measurements taken at Week 0 and Week 18 of the ASD cohort. This study found that some metabolic changes had begun by Week 3 (after vancomycin, bowel cleanse, and approximately five days of FMT). It is also notable that the distributions of FDA scores within both the five-metabolite models at later timepoints (Week 10 and Week 18) are bimodal. This suggests that some individuals may respond differently to MTT than others. This finding was similar to the analysis performed on plasma metabolites where a steep decline in median discriminant score was also observed at Week 3 and Week 10 [29].

The OFM-I/A metabolites demonstrated limited correlation among themselves. This was to be expected as FDA seeks to maximize the amount of discriminating information with a minimal number of utilized metabolites. For this reason, within this subset of fecal metabolites, those with few correlations tended to appear more frequently in the top 1000 models. Specifically, there was a high proportion of top 1000 models featuring theobromine (28.5%), which was ranked as the fourth most common metabolite present in the models. Notably, this metabolite was not significantly correlated with any of the other top 50 metabolites as discussed above. In total, 44 of the 50 metabolites with the highest AUROC were correlated with the OFM metabolite panel, suggesting that there are at least six common types of metabolic abnormalities associated with ASD. Although adenosine and indole were not found to be correlated, they were included in the OFM-A and OFI-I models, respectively, and both

metabolites are related to distinct biological pathways, with adenosine being associated with purine metabolism, while indole is associated with tryptophan metabolism. It was observed in one study that metabolites associated with these two pathways were the most different in urine of ASD and TD children [43].

While there are several metabolic pathways that are related to the top 50 metabolites identified, about 45% of the top metabolites were connected to phenylalanine and tyrosine metabolism, fatty acid metabolism or sphingolipid metabolism. Differences in tyrosine metabolites such as decreased concentration of phenylalanine and increased concentration of p-cresol have been previously observed in studies examining the gut metabolite composition in TD and ASD children [14,44]. The role of the microbiota in this pathway is also very significant. Tyrosine metabolism pathway downregulation was observed in an ASD cohort to be associated with an increased prevalence of *Bacteroides vulgatus* while upregulation was associated with *Eggerthella lenta* [44]. Similarly, the relationship between sphingolipid metabolism and microbiome crosstalk has been suggested, and differences in short chain fatty acids of children with ASD and their TD peers have been noted [24,36].

Several metabolites related to mitochondrial metabolism and regulation such as carnitine, betaine, and adenine were also determined to have particularly high AUROC values [35,45–47]. Carnitine serves as the cofactor that transports long-chain fatty acids to the mitochondria matrix, and betaine plays a role in increasing mitochondrial membrane potential [45,47]. There has been considerable investigation into the relationship between ASD and mitochondrial dysfunction. It is estimated that around 4–7% of children with ASD are affected by mitochondrial disease, but it is speculated that up to 80% may have abnormalities in mitochondrial function [48,49].

Prior work has shown similar relationships between ASD fecal metabolite profiles as were observed in this study. GABA, an important neurotransmitter, was one of the metabolites identified as having a lower concentration in the ASD group prior to MTT, which is consistent with prior work [16,17]. Similarly, the fecal concentrations of free carnitine have been previously observed to be higher in children with ASD, which was also observed in this study [14]. The fecal metabolite measurements are consistent with prior work as the average indole measurements for the ASD cohort were more than twice the value of their TD counterparts at Week 0 (see Table A1) [17]. It has also been observed that fecal metabolites associated with glutamate metabolism such as 2-Keto-glutaramic acid and l-Aspartic acid were downregulated in children with ASD [44]. These metabolites were not measured in this study. Nonetheless, one of the metabolites associated with glutamate metabolism, carboxyethyl-GABA, was identified in significantly lower concentration in the ASD + GI cohort at baseline.

While there are indeed some similarities between the analysis of fecal metabolites and prior assessment of plasma samples taken from these participants, there are key distinctions. None of the metabolites identified as being utilized in the optimum multivariate models were previously identified as being significant for classification in the multivariate plasma models. The general performance of the fecal metabolites when subjected to univariate analysis had generally lower AUROC values than plasma metabolites [29]. However, despite having lower AUROC scores, multivariate analysis achieved high accuracy in distinguishing ASD and TD children. That being said, we were able to achieve greater separation using three plasma metabolites than with five fecal metabolites. This may be due to the greater homogeneity in plasma samples vs. stool samples. There have also been far more studies conducted examining plasma metabolite concentrations in individuals with ASD than studies focused on fecal samples [50,51].

Although the models were able to classify between the ASD + GI and TD cohorts with high accuracy, this study also has several limitations. The study focused exclusively on children with ASD with initially moderate to severe GI problems, which were compared to TD children with no GI issues. Therefore, assessments regarding ASD were confounded with GI problems in this analysis. ASD subgroups differentiated by variations in GI abnormalities were ignored in the analysis as subgroups were too small for a robust statistical assessment [52,53]. Furthermore, the study-cohort was not large and the ASD cohort was further split up into two different initial treatments. Future

studies with a larger sample size examining cohorts with and without GI symptoms would allow for an assessment on the effectiveness of MTT in ameliorating behavioral symptoms in addition to GI related pathology.

#### **5. Conclusions**

This study investigated differences in fecal metabolites between a group of children diagnosed with ASD and GI symptoms and their typically developing peers with no history of GI symptoms. The univariate analysis demonstrated that individual fecal metabolites had limited potential to distinguish between ASD+GI and TD cohorts, unlike the previous study of plasma; this may be due to greater heterogeneity in stool compared to plasma. However, multivariate statistical analysis resulted in five-metabolite models that had high accuracy even after cross-validation. Both the OFM metabolite panels were shown to be capable of achieving 95% specificity and 94% sensitivity.

Following MTT, 14% of the top 50 metabolites that were found to have the greatest difference in concentration between the TD and ASD group shifted such that their distributions were significantly different eight weeks after the treatment ended. Furthermore, 94% of these metabolites reported lower AUROC following treatment, indicating diminished capacity to distinguish between the ASD and TD group. When considering a normalized average of the metabolites in the OFM models, the difference between the ASD and TD groups decreased by 82–88% at 18 weeks. These findings are similar, although less pronounced, as those determined using plasma metabolites, and both suggest that MTT resulted in shifting the metabolic profile of the ASD group towards becoming more similar to the TD group. Future work should be performed to validate the effect of MTT on fecal metabolites using a larger study cohort and a placebo arm.

**Author Contributions:** J.A., D.-W.K., R.K.-B. and J.H. conceived of the current study and contributed to its design and coordination. R.K.-B. received grant for FMT trial funding, helped design FMT trial, directed microbiome analysis J.A., D.-W.K. and R.K.-B. collected and interpreted the clinical data. F.Q. and K.H. analyzed the data and generated the tables and figures. Statistical results were interpreted by J.A., F.Q. and J.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors gratefully acknowledge partial financial support from the National Institutes of Health (grant 1R01AI110642).

**Acknowledgments:** We would like to acknowledge the contributions of Troy Vargason, Uwe Kruger and Kathryn Hollowood Jones for the initial help with the analysis. We would also like to thank Sharon McDonough-Means, Thomas Borody, Alex Khoruts, Michael Sadowsky, Alessio Fasano, Devon Coleman, and Elena Pollard for their help with the treatment portion of the Microbiota Transfer Therapy study.

**Conflicts of Interest:** The authors have filed an invention disclosure on the results presented here.





**Table A1.***Cont.*




**Table A1.***Cont.*


**Table A1.** *Cont.*


**Table A1.** *Cont.*

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Mitochondria May Mediate Prenatal Environmental Influences in Autism Spectrum Disorder**

**Richard E. Frye 1,\*, Janet Cakir <sup>2</sup> , Shannon Rose <sup>3</sup> , Raymond F. Palmer <sup>4</sup> , Christine Austin <sup>5</sup> , Paul Curtin <sup>5</sup> and Manish Arora <sup>5</sup>**


**Abstract:** We propose that the mitochondrion, an essential cellular organelle, mediates the longterm prenatal environmental effects of disease in autism spectrum disorder (ASD). Many prenatal environmental factors which increase the risk of developing ASD influence mitochondria physiology, including toxicant exposures, immune activation, and nutritional factors. Unique types of mitochondrial dysfunction have been associated with ASD and recent studies have linked prenatal environmental exposures to long-term changes in mitochondrial physiology in children with ASD. A better understanding of the role of the mitochondria in the etiology of ASD can lead to targeted therapeutics and strategies to potentially prevent the development of ASD.

**Keywords:** autism spectrum disorder; mitochondria; oxidative stress; prenatal environment; immune dysfunction

### **1. Introduction**

Autism spectrum disorder (ASD) is a behaviorally defined disorder [1], with the most recent Center for Disease Control and Prevention estimates suggesting that it affects 1 in 54 children in the United States [2]. Recent studies suggest that inherited single-gene and chromosomal defects account for a minority of ASD cases [3], and that ASD most likely arises from a complicated interaction between genetic predisposition and environmental exposures [4,5]. Given the high recurrent risk in siblings, the prenatal maternal environment has undergone careful study with many prenatal risk factors identified [6,7]. Despite the epidemiological connection between many prenatal risk factors and the development of ASD, the biological mechanisms which link prenatal environmental influences and the increased risk of developing ASD are just beginning to be uncovered.

Three physiological abnormalities which have been increasingly recognized to be associated with ASD are immune system dysfunction, mitochondrial dysfunction, and oxidative stress and redox regulation [1]. Previous reviews examining prenatal physiological abnormalities related to ASD have concentrated on prenatal immune stressors as key and consider mitochondrial dysfunction and oxidative stress to have secondary roles of this "Bad Trio" [8]. In contrast, the current review concentrates on the mitochondria as the central player. Of course, the particular component of the "Bad Trio" that is the initiating culprit may be different for different patients and it is possible that multiple stressors on the various portions of the "Bad Trio" simultaneously may also initiate the pathway to disease.

**Citation:** Frye, R.E.; Cakir, J.; Rose, S.; Palmer, R.F.; Austin, C.; Curtin, P.; Arora, M. Mitochondria May Mediate Prenatal Environmental Influences in Autism Spectrum Disorder. *J. Pers. Med.* **2021**, *11*, 218. https://doi.org/ 10.3390/jpm11030218

Academic Editor: Guido Krenning

Received: 5 February 2021 Accepted: 17 March 2021 Published: 18 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **2. The Mitochondria: Dysfunction Can Be Self-Perpetuating**

Mitochondria are essential for a wide range of functions in almost every cell in our body (Figure 1). Best known for their role in the production of adenosine triphosphate (ATP) by oxidative phosphorylation, mitochondria are intimately involved in other cellular functions such as redox metabolism, calcium buffering, lipid homeostasis, and steroid synthesis [9–13]. Mitochondria also have a role in important non-energy-producing metabolic pathways, such as the urea cycle, amino acid and porphyrin production, and as a pathway for the activation of apoptosis. Mitochondria also have important roles in cell signaling, most notably being an essential part of the inflammasome, a complex that initiates immune activation, by releasing damage-associated molecular pattern (DAMP) molecules such as cardiolipin, n-formyl peptides, reactive oxygen species (ROS), and mitochondrial DNA (mtDNA) [14]. Lastly, normal mitochondrial function results in the production of ROS, which can cause cellular injury if not controlled.

**Figure 1.** The mitochondria can be negatively affected by many environmental and biological factors associated with autism spectrum disorder (left orange panel) and has many critical roles in cellular physiology (right blue panel). ETC: electron transport chain; DAMP: damage-associated molecular pattern

> Since ATP produced by mitochondria is essential for many cellular systems, abnormal mitochondrial function can disproportionally adversely affect cellular physiology. However, there are several pathways in which abnormal mitochondrial function can result in a self-perpetuating destructive cycle causing sustained pathophysiology. Most notably, interactions between mitochondria, redox metabolism, and the immune system can be mutually detrimental; such detrimental interactions have been documented in ASD (Figure 2) [15].

> Mitochondria are both a major producer and target of ROS. Dysfunctional mitochondria produce high amounts of ROS which can result in dysfunction of the electron transport chain (ETC) enzymes, particularly complex I and III, as well as aconitase, the first enzyme in the citric acid cycle (CAC). To compound this problem, reduced glutathione (GSH), the main intracellular and mitochondrial antioxidant, requires ATP for its *de novo* production. As such, a decrease in ATP production resulting from reduced mitochondrial function will result in lower GSH production, resulting in poorer control of ROS. In fact, a lower GSH redox ratio has been correlated with lower aconitase activity in post-mortem brain from individuals with ASD [16]. Oxidative damage to cellular lipids, proteins, and nucleic acids [17] has been associated with ASD; this is especially important since mtDNA is vulner-

able to oxidative damage, and studies have shown that children with ASD demonstrated mtDNA damage in a pattern consistent with oxidative damage [18].

**Figure 2.** Self-perpetuating destructive cycles which can result in mitochondrial dysfunction.

There are several pathways in which dysfunctional mitochondria can cause a wide variety of abnormalities in immune system function. First, cellular damage due to oxidative stress can activate inflammatory pathways [15,16]. Second, as an essential part of the inflammasome, mitochondria release DAMP molecules such as cardiolipin, n-formyl peptides, ROS, and mtDNA [14]. Third, regulatory immune cells are highly dependent on ATP derived from mitochondrial oxidative phosphorylation, while inflammatory cells are highly glycolytic [19]. Thus, once an immune response has been started, the inflammatory response may be difficult to regulate if mitochondrial dysfunction exists. Fourth, inflammation and immune activation upregulate metabolism and recruit physiological processes, but without mitochondrial support, such resources will not be available. Lastly, the immune system produces ROS as a defense mechanism against potential invaders. Such increases in ROS can result in a further detrimental effect on already dysfunctional mitochondria. In fact, mitochondria appear to have a particularly important role in innate immunity [20], which is an area of immune dysfunction that is implicated in ASD. In addition, an increase in proinflammatory cytokine production has been associated with mitochondrial dysfunction in a subset of children with ASD [21].

#### **3. Prenatal Risk Factors for ASD Modulate Mitochondrial Function**

Many prenatal factors associated with an increased risk of ASD are associated with mitochondrial dysfunction; these include nutritional agents, both intrinsic and extrinsic stressors, common medications given during pregnancy, modulators of mitochondrial function, and genetic conditions which might affect the fetus (Figure 1).

Folate is an essential vitamin which is well-known to be important during pregnancy to prevent neural tube defects. Folate is also an essential co-factor for adequate mitochondrial function [22]. Several studies have demonstrated that folate supplementation during pregnancy reduces the risk of ASD. Abnormalities in the folate pathway are associated with ASD, including maternal polymorphisms in the reduced folate carrier [23]. Further, mothers of children with ASD have been shown to have the folate receptor alpha autoantibody [24],

an abnormality that prevents folate transport across the placenta, resulting in an ASD phenotype in an animal model [25].

Abnormalities in several nutrient metals have been linked to an increased risk of ASD. Studies have associated an increased risk of ASD with atypical pre- and postnatal Zn and Cu metabolism [26] and found that atypical levels of essential (Zn, Mn) and nonessential (Pb) metals during prenatal development and early life in individuals with ASD are associated with long-term physiological and developmental alternations [27]. Studies have associated low maternal iron (Fe) with increased ASD risk [28]. Prenatal Zn and Cu are essential for the function of the cytoplasmic superoxide dismutase (SOD) which is essential for controlling cellular oxidative stress, while Cu is essential for ETC complex IV function. Mn is essential for mitochondrial SOD function, and Fe is an essential component of cytochromes, which are critical components of the ETC.

Carnitine is an essential nutrient for mitochondrial function and fatty acid metabolism [11,29,30]. Abnormalities in carnitine metabolism have been linked to gestational diabetes [31], a risk factor for ASD [32]. A genetic defect in carnitine synthesis is a risk factor for ASD [33], carnitine metabolism is known to be disrupted in ASD [29,30] and a mouse model of ASD is associated with alternations in carnitine biosynthesis [34]. Due to its importance in energy metabolism in neural stem cells of the developing mammalian brain, carnitine deficiency has been proposed to be a prenatal risk factor for ASD [35].

As noted above, increased oxidative stress as well as inflammation is associated with mitochondrial dysfunction. Abnormalities in maternal trans-sulfuration metabolism and chronic oxidative stress are found in mothers of children with ASD during [36] and following [37] pregnancy and, in general, infection during pregnancy is a risk factor for ASD [38]. In fact, the maternal immune activation (MIA) mouse model of ASD demonstrates longterm mitochondrial dysfunction in brain [39] and leukocytes [40] after birth. Pregnancies resulting in a child with ASD have demonstrated increases in proinflammatory cytokines IL-1α [41] and IL-6 [41] in blood, while maternal elevation in IL-17a has been strongly implicated in the MIA model [42]. In laboratory studies, IL-17 [43,44] induces mitochondrial dysfunction through activation of the mitochondria-induced apoptosis pathway and IL-1 suppresses mitochondrial function [45], while IL-6 promotes mitochondrial biogenesis and fatty acid oxidation [46].

Prenatal exposure to many environmental toxicants is linked to an increased risk of developing ASD, including cigarette smoke, phthalates, air pollution, and pesticides such as organophosphate insecticides (e.g., chlorpyrifos) and organochlorine pesticides (e.g., dicofol and endosulfan) [7]. These toxicants have been associated with mitochondrial dysfunction. Organophosphate induces mitochondrial ultrastructure changes and inhibits ETC and CAC enzyme function, while organochlorine pesticides influence mitochondrial dysfunction indirectly by increasing ROS and reactive nitrogen species by altering antioxidant systems including SOD and GSH [47]. Phthalates, common plasticizers present in everyday products, have also been linked to increased oxygen consumption, mitochondrial mass, and fatty-acid metabolism in neonatal rat cardiomyocytes [48] and detrimental changes in mitochondrial membrane potential in human semen [49]. Mitochondrial-derived peptides in cord blood have been associated with prenatal exposure to non-freeway traffic-related air pollution [50]. Cigarette smoke causes mitochondrial dysfunction in the lung epithelium [51].

Several medications commonly used during pregnancy have been linked to an increased risk of developing ASD including acetaminophen [52] and selective serotonin reuptake inhibitors [53]. Acetaminophen has known toxicity by increasing reactive metabolites leading to suppression of mitochondrial function [54]. Fluoxetine, a commonly used selective serotonin reuptake inhibitor, has been shown to inhibit multiple mitochondrial enzymes in several laboratory studies [55] and may result in long-term changes in energy metabolism with neonatal exposure [56]. Other medications commonly used in pregnancy with less certain association with ASD are associated with mitochondrial dysfunction. For example, commonly used antibiotics, including quinolones, aminoglycosides, and

β-lactams, can cause mitochondrial dysfunction [57], and commonly used anesthesia may be particularly detrimental to the developing brain, partially through mitochondrial mechanisms [58].

Interestingly, mitochondrial abnormalities have been documented in genetic syndromes associated with ASD such as PTEN mutations [59] and tuberous sclerosis [60,61], Fragile X [62,63], Rett [64–66], Phelan–McDermid [67], 15q11-13 duplication [68,69], Angelman [70] and Down [71,72] syndromes as well as septo-optic dysplasia [73]. Thus, a fetus with these genetic changes may already have vulnerable mitochondria that might be sensitive to environmental stressors.

Common nutritional deficiencies are prenatal ASD risk factors which have been suggested to modulate mitochondrial function. Decreased vitamin D in the first [74] or second [75] trimester as well as lifetime [76] is associated with more severe ASD [74,75] or increased ASD risk [76]. Vitamin D deficiency is associated with oxidative stress and reduced mitochondrial respiration that is mediated through the vitamin D receptor [77,78].

Several lines of evidence suggest that ASD is associated with disruption of the microbiome in individuals with ASD as well as their mother during pregnancy [79]. Although a recent landmark study provided preliminary evidence that transplanting the microbiome in individuals with ASD can improve gastrointestinal and ASD symptoms [63], studies in pregnant women are more difficult to conduct, leading to animal studies addressing this possibility. Environmentally induced rodent models of ASD, including the MIA [80] and the valproic acid exposure models [81–83], have an altered microbiome. Manipulations which can affect the microbiome have been shown to mitigate the effects of these maternal exposures. Treating pups born from maternal valproic acid exposure demonstrate reduced ASD-like behaviors [84,85] as well as normalization of mitochondrial abnormalities [84], while treatment of pups born from MIA with *Bacteroides fragilis* normalizes gut permeability and microbial composition and reduces ASD-like behaviors [86]. Interestingly, several prenatal environmental exposures linked to ASD, including air pollution [87], glyphosate [88], prenatal antibiotic use [89], maternal stress [90] and organophosphate herbicides [91], have evidence for disrupting the microbiome, suggesting a potential biological pathway for their effects. The microbiome can influence mitochondrial function through several mechanisms, although the most compelling is through the production of short chain fatty acids [92].

#### **4. Unique Abnormalities in Mitochondrial Function Are Prevalent in ASD**

The possibility that environmentally induced mitochondrial dysfunction could have a role in ASD is particularly compelling because abnormal mitochondrial function is one of the most prevalent metabolic disorders found in individuals with ASD, with prevalence ranging from 5% for classically defined mitochondrial disease to 8–47% for biomarkers of mitochondrial dysfunction [11,16], to 62–65% for abnormal ETC/CAC enzymology [93,94], and to 80% for abnormal ETC activity in lymphocytes and granulocytes [95,96]. Particularly compelling is that the great majority of the time, genetic defects cannot explain the mitochondrial abnormalities, suggesting that the abnormality could be acquired as a result of an environmental exposure. There is also a tremendous practical appeal in this hypothesis, as environmental determinants may be particularly amenable to modification.

In classic mitochondrial disease, ETC activity is, by definition, depressed. However, what is unique about abnormalities in mitochondrial function in individuals with ASD is that ETC activity is significantly increased in many cases. The first case reported was a boy with ASD who demonstrated a significant increase in ETC complex I activity while his sister, who was diagnosed with a classic mitochondrial disease known as Leigh syndrome, showed depressed ETC activity [97]. Interestingly, both siblings manifested the same mtDNA mutation, but the sister had a greater genetic mutational load (i.e., higher heteroplasmy). Subsequently, a case-series of five patents with ASD with muscle ETC complex IV activity about 200% of normal was reported [98]. The association of elevated ETC complex IV activity with ASD has subsequently been confirmed in fresh frozen superior

temporal gyrus [99], buccal swabs enzymology [93] and lymphoblastoid cell lines (LCLs) using high-resolution respirometry [100].

#### **5. The Significance of Mitochondrial Dysfunction in ASD: Sensitivity to Physiological Stress**

Since increased ROS is a key mechanism by which environmental stressors, such as toxicants [101–108] and inflammation [1,11], can disrupt mitochondrial function, an assay has been developed which systematically increases ROS *in vitro* [109]. This assay, called the Mitochondrial Oxidative Stress Test (MOST), systematically increases ROS *in vitro* using 2,3–dimethoxy–1,4-napthoquinone (DMNQ), an agent that generates intracellular superoxide and hydrogen peroxide but does not directly deplete thiols [110]. The model was initially developed using LCLs from boys with ASD and age-matched healthy controls (CNT) where DMNQ was shown to increase ROS in CNT and ASD LCLs [110].

In this model, mitochondrial respiratory rates were significantly higher in ASD LCLs as compared to CNT LCLs. Most compelling, indices of mitochondrial health, reserve capacity and maximal respiratory rate, decreased to a greater extent in ASD LCLs when challenged with DMNQ in the MOST assay. Increases in ROS, using the MOST assay, resulted in a depletion of these respiratory parameters at lower DMNQ concentrations in the ASD LCLs, despite these parameters being higher at baseline. This suggests that the ASD LCLs demonstrated a greater vulnerability to ROS.

To determine if these bioenergetics changes were specific to a subset of ASD LCLs, a cluster analysis was used to separated ASD LCLs into those with normal bioenergetics (AD-N) and those with atypical bioenergetics (AD-A) [109]. When the AD-A LCLs were compared to the CNT LCLs, the bioenergetic differences were found to be large. Baseline respiratory rates were ~200% higher in AD-A as compared to the CNT LCLs. Most notably, maximal respiratory capacity and reserve capacity markedly decreased as DMNQ increased, such that reserve capacity was rapidly depleted as DMNQ increased despite being much higher at baseline. This pattern of abnormal respiration in this subset of LCLs has been confirmed over eight studies [61,109,111–116].

Despite these experiments, the question remained whether this pattern of mitochondrial dysfunction was specific to ASD or simply a consequence of higher levels of chronic ongoing intrinsic oxidative stress. Since neurotypical siblings (SIBs) of children with ASD manifest similar chronic ongoing elevation in oxidative stress in their LCLs, mitochondrial function was compared between 10 LCLs from boys with ASD and their 10 male SIBs and 10 age-matched CNT males [117]. Mitochondrial function was similar between SIBs and CNTs, but both were different from ASD. In the ASD LCLs, mitochondrial respiration was elevated at baseline and reserve capacity declined more precipitously with increasing DMNQ, as compared to SIB and CNT LCLs. Most notably, the severity of mitochondrial abnormalities in ASD LCLs was related to the severity of stereotyped behaviors and restricted interests as measured on the gold-standard Autism Diagnostic Observation Schedule evaluation years earlier when the blood samples were original collected. Thus, atypical mitochondrial activity in the ASD LCLs is not simply a product of abnormal redox metabolism but rather associated with atypical mitochondrial function specifically. Furthermore, this atypical mitochondrial function is related to more severe core ASD behaviors, suggesting an association with molecular mechanisms of ASD.

Individuals with ASD demonstrate three developmental trajectories: in the early onset subtype, symptoms are obvious from early in infancy, perhaps at birth; in the plateau subtype, infants develop normally throughout the first year of life but then plateau in the rate of gaining skills, followed by the development of ASD symptoms; lastly is a subset that demonstrates neurodevelopmental regression (NDR). The NDR category is intriguing: children attain all their normal developmental milestones but then lose previously attained skills followed by the development of ASD-like behaviors. Often, NDR is associated with a trigger such as seizure [118] and/or fever [119]. NDR is not uncommon in individuals with mitochondrial disease when an illness occurs [120]. Thus, it is not surprising that a

meta-analysis showed that NDR was more common in children with ASD that were also diagnosed with mitochondrial disease [11].

Given that the subset of ASD LCLs with elevated respiratory rates have increased vulnerability to physiological stress, a recent study hypothesized that children with ASD and NDR would demonstrate increase mitochondrial respiratory rates. Mitochondrial function was measured in cryopreserved peripheral blood mononuclear cells (PBMCs) from children with ASD, with and without NDR, as well as CNT using the Seahorse XF96 respirometer [121]. Viability was measured and 600 k viable PBMCs were plated per well. As hypothesized, mitochondrial respiration was elevated in children with ASD and NDR. Specifically, the maximal oxygen consumption rate, maximal respiratory capacity, and reserve capacity were higher in the individuals with ASD and NDR as compared to the other groups. Additionally, comparing ASD twins discordant on NDR demonstrated that the twin with NDR showed a significantly elevated maximal oxygen consumption rate. Thus, the NDR ASD phenotype may be a hallmark of abnormal mitochondrial physiology.

#### **6. Unique ASD Mitochondrial Abnormalities May Be Linked to Both Environmental and Genetic Factors**

Interestingly, the unique elevations in mitochondrial respiration reported in human tissue have been associated with both genetic and environmentally induced animal models of ASD. Elevated mitochondrial respiration has been documented in genetic syndromes associated with ASD, including Phelan–McDermid [122], Fragile X [62,63], 22q13 dup [67], and Rett [66] syndromes, the PTEN haploinsufficient mouse model of ASD [59] and the Drosophila model of the ASD associated CYFIP1 mutation [123]. Similar changes have been associated with prenatal environmental exposures, specifically prenatal exposure to inflammation as in the MIA mouse, a model of ASD induced by prenatal immune environmental stress [39], and prenatal exposure to toxins as in the maternal valproic acid exposure mouse model of ASD [84].

Consistent with evidence from these prenatal environmental animal models of ASD, several studies have examined the relationship between alternations in long-term mitochondrial function and prenatal environmental stressors in ASD.

Prenatal exposure to air pollution, as measured by average and maximum PM2.5, has been found to be related to mitochondrial respiration in childhood, as measured in PBMCs. The relationship was significantly different for those children with and without a history of NDR. For those with a history of NDR, higher prenatal PM2.5 exposure was associated with higher mitochondrial respiration rates, while for those without a history of NDR, higher prenatal PM2.5 exposure was related to lower mitochondrial respiration [27]. Additional research has linked prenatal air pollution exposure to mitochondrial-derived peptides in cord blood that are associated with long-term changes in mitochondrial physiology [50].

Both prenatal exposures to nutritional and toxic metal were measured in deciduous teeth using laser ablation inductively coupled plasma mass-spectrometry. Prenatal exposure to Mn and Zn was associated with mitochondrial respiration, but only in children with ASD and NDR [29]. The prenatal Cu to Zn ratio was associated with two independent measures of language development in all children with ASD (both those with and without NDR) [29]. This latter study extends previous findings linking prenatal nutrient metal (Zn, Mn, Cu) exposure and ASD [26,27,124–127].

To determine whether the long-term changes associated with ASD could be induced by exposure to environmental toxicants, one study exposed LCLs to low levels of ROS for a prolonged time (96 h) to simulate chronic ROS exposure which might occur with prolonged exposure to environmental toxicants or other physiological stressors [61]. Prolonged exposure to low levels of ROS induced bioenergetic changes in respiratory parameters involved in ATP production. After prolonged ROS exposure, the LCLs demonstrated increased respiratory rates at baseline, similar to the AD–A LCLs and the ETC activity seen in children with ASD in multiple tissues. These data suggest that a prolonged exposure to a pro-oxidant microenvironment can have chronic effects on the mitochondria.

#### **7. Long-Term Induced Changes in Mitochondrial Function: Adaptive or Maladaptive**

Mitochondria undergo long-term adaptive changes in physiology as a result of environmental stressors through a process known as mitoplasticity, which is one compelling mechanism for long-term changes in mitochondrial function associated with prenatal exposures [9]. Several studies may provide insight into the molecular mechanisms associated with long-term changes in mitochondrial function.

Matched typically developing siblings have similar abnormalities in oxidative stress compared to their ASD siblings, but the mitochondria of the ASD siblings appear to have difficulty regulating the mitochondrial ROS (mtROS). mtROS is regulated at the inner mitochondrial membrane through several mechanisms. Uncoupling proteins regulate mtROS by leaking protons across the inner mitochondrial membrane, a process known as proton leak. Studies have demonstrated increased proton leak respiration along with high oxidative stress in both cytoplasm and mitochondria in ASD LCLs [16,128]. Studies have associated ASD with an increase in mechanisms of proton leak. Three previous studies have found an increase in uncoupling protein (UCP2) gene expression [61,116] and protein concentration [109] in LCLs derived from children with ASD, particularly in the LCLs with high respiratory rates (i.e., AD–A LCLs). The adenine nucleotide translocator (ANT) has a significant role in the regulation of inner mitochondrial membrane proton leak. Microdeletion including the SLC25A5 (ANT2) has been associated with non-syndromic intellectual disability with ASD [129]. Heteroplasmic levels of the mtDNA 3243A > G mutation associated with ASD are also associated with significant changes in ANT gene expression [130]. Consistent with this evidence, increased protein leak is also a feature of mitochondrial abnormalities in the Fragile X syndrome mouse, where it is found to directly affect synaptic growth [131]. Thus, dysregulation in inner membrane proton leak seems to be associated with an ASD phenotype.

One of the key processes which maintains optimal mitochondrial function in the face of physiological stress is mitophagy. Several studies have linked ASD to a failure to induce this important process. Mutations in WDFY3 which result in ASD with intellectual disability have been linked to bioenergetic abnormalities in the brain through decreased mitophagy [132], and PARK2, a gene known to be involved in mitophagy, has been identified as a candidate gene for ASD [133]. Examination of the post-mortem temporal cortex has demonstrated ETC activity abnormalities along with alternation in levels of fission (Fis1, Drp1) and fusion (Mfn1, Mfn2 and Opa1) protein essential for regulating mitophagy [134]. Neurons deficient in TSC1/TSC2, a model of tuberous sclerosis, demonstrate impaired mitophagy through a mTORC1 mechanism [60,135]. This is consistent with studies which have demonstrated that ASD LCLs with elevated respiratory rates (i.e., AD–A LCLs) fail to upregulate genes associated with mitophagy through a mTORC1 pathway related mechanism [61]. Recently, ASD-derived fibroblasts were found to have elevated respiratory rates, atypical mitochondrial morphology, and alteration in the mitophagy pathway [136]. Interestingly, a loss of mitophagy can lead to a build-up of cytosolic ROS and mtDNA damage and an increase in proinflammatory cytokines, including IL–1α, IL–1β, IL–18, IFNα, MIF, IL–23, and IL–17 [137].

Alternative to regulatory changes within the body is the possibility that long-term alternation in the microbiome environment (technically outside of the body) could account for long-term changes in mitochondrial function. The connection between the microbiome and mitochondrial dysfunction in ASD is supported by the compelling propionic acid (PPA) rodent model of ASD. Adult rats intraventricularly injected with PPA [138], juvenile rates intraperitoneally injected with PPA [139], and prenatally PPA exposed rodents [140,141] demonstrated ASD-like behaviors. The importance of this model is that fact that enteric microbiome producers of PPA, particularly *Clostridia* sp, are relatively overrepresented in the ASD microbiome [142]. PPA exposure results in several physiological abnormalities, including mitochondrial dysfunction, particularly disruption of the fatty acid metabolism, as manifested by unique elevations in acyl-carnitines, as well as redox abnormalities as manifested by GSH alternations [142]. Parallel to the unique acyl-carnitine elevations in the

rodent model, the same pattern was reported in a case series of ASD patients [143] with the parallel mitochondrial and GSH abnormalities later reported in a more detailed study [144]. Further *in vitro* studies with ASD LCLs demonstrated the potential detrimental effects of PPA [114] and the protective effect of another prominent enteric microbiome derived short chain fatty acid butyrate [115]. Thus, long-term changes in the microbiome could also account for long-term changes in mitochondrial function in individuals with ASD.

#### **8. Conclusions**

One of the critical knowledge gaps in environmental medicine is how an environmental exposure can result in disease when the symptoms arise years after the exposure. Mitochondrial metabolism is a promising mechanism through which the environment can cause long-term effects. For ASD, we have discussed the many prenatal environmental exposures which are linked to ASD and can cause detrimental changes in mitochondrial function. This is compelling because mitochondrial dysfunction is prevalent in ASD. ASD is also associated with a unique type of mitochondrial dysfunction which may be related to environmental exposures. Studies support the notion that specific prenatal exposures linked to ASD, most notably air pollution and prenatal nutritional metal exposure, is associated with long-term alterations in mitochondrial physiology. A better understanding of these biological processes may lead to prevention and treatment strategies.

Prenatal mitochondrial function is important for brain development, as recent animal studies have shown that mitochondrial dysfunction during gestation alters white matter brain connectivity [145] and non-radial interneuron migration [146]. Mitochondrial function is essential for optimal neuronal function and is essential for transport of essential nutrients into the brain such as folate [30]. Mitochondria are also essential for important core metabolic pathways as well as for optimal function of the immune system, which is now known to have an important role in brain development [30]. Thus, mitochondrial dysfunction during the prenatal period can have a profound effect on long-term development.

In this manuscript, we discuss studies with findings that link prenatal environmental exposures to long-term mitochondrial function and ASD. These studies remain intriguing yet preliminary. These findings need to be confirmed and further studies need to be conducted to expand on these findings. One important caveat is a consideration of the measurement of mitochondrial function, as mitochondrial function can differ from tissue to tissue and various assays reflect different aspects of mitochondrial function. For example, the use of PBMCs to measure mitochondrial function is a relatively newly developed biomarker which will require further validation in the future [147].

The applications of these findings may be far reaching, as interventions to correct metabolic and mitochondrial abnormalities are under development, potentially providing targeted treatments for preventing ASD and other diseases from developing if implemented prenatally.

**Author Contributions:** All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This research was supported by the Arkansas Biosciences Institute (Little Rock, AR, USA), The Jonty Foundation (St Paul, MN, USA), The Autism Research Institute (San Diego, CA, USA), the Gupta Family Foundation (Atherton, CA, USA), The Jane Bostford Johnson Foundation (New York, NY, USA) and the Jager Family Foundation (Chicago, IL, USA).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


### *Article* **Mitochondrial Fatty Acid** β**-Oxidation and Resveratrol Effect in Fibroblasts from Patients with Autism Spectrum Disorder**

**Rita Barone 1,2,\* ,†, Jean Bastin 3,†, Fatima Djouadi <sup>3</sup> , Indrapal Singh 4,5, Mohammad Azharul Karim 4,5 , Amrit Ammanamanchi 4,5, Patrick John McCarty 4,5, Leanna Delhey <sup>6</sup> , Rose Shannon <sup>6</sup> , Antonino Casabona <sup>7</sup> , Renata Rizzo <sup>1</sup> and Richard Eugene Frye 4,5,\***


**Abstract:** Patients with autism spectrum disorder (ASD) may have an increase in blood acyl-carnitine (AC) concentrations indicating a mitochondrial fatty acid β-oxidation (mtFAO) impairment. However, there are no data on systematic mtFAO analyses in ASD. We analyzed tritiated palmitate oxidation rates in fibroblasts from patients with ASD before and after resveratrol (RSV) treatment, according to methods used for the diagnosis of congenital defects in mtFAO. ASD participants (*N* = 10, 60%; male; mean age (SD) 7.4 (3.2) years) were divided in two age-equivalent groups based on the presence (*N* = 5) or absence (*N* = 5) of elevated blood AC levels. In addition, electron transport chain (ETC) activity in fibroblasts and muscle biopsies and clinical characteristics were compared between the ASD groups. Baseline fibroblast mtFAO was not significantly different in patients in comparison with control values. However, ASD patients with elevated AC exhibited significantly decreased mtFAO rates, muscle ETC complex II activity, and fibroblast ETC Complex II/III activity (*p* < 0.05), compared with patients without an AC signature. RSV significantly increased the mtFAO activity in all study groups (*p* = 0.001). The highest mtFAO changes in response to RSV were observed in fibroblasts from patients with more severe symptoms on the Social Responsiveness Scale total (*p* = 0.001) and Awareness, Cognition, Communication and Motivation subscales (all *p* < 0.01). These findings suggested recognition of an ASD patient subset characterized by an impaired mtFAO flux associated with abnormal blood AC. The study elucidated that RSV significantly increased fibroblast mtFAO irrespective of plasma AC status, and the highest changes to RSV effects on mtFAO were observed in the more severely affected patients.

**Keywords:** autism spectrum disorder; energy metabolism; fatty acid oxidation; acyl-carnitines; resveratrol

#### **1. Introduction**

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by early social communication deficits and repetitive motor behavior, sensory abnormalities, and restricted interests, with a global prevalence of about 1% [1] and, according to the most

**Citation:** Barone, R.; Bastin, J.; Djouadi, F.; Singh, I.; Karim, M.A.; Ammanamanchi, A.; McCarty, P.J.; Delhey, L.; Shannon, R.; Casabona, A.; et al. Mitochondrial Fatty Acid β-Oxidation and Resveratrol Effect in Fibroblasts from Patients with Autism Spectrum Disorder. *J. Pers. Med.* **2021**, *11*, 510. https://doi.org/ 10.3390/jpm11060510

Academic Editor: Farah R. Zahir

Received: 3 May 2021 Accepted: 31 May 2021 Published: 4 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

recent estimates, of about 2% in the United States (US) [2]. ASD biology is particularly complex, including individual genetic contributions interacting with multiple environmental factors. Genetic risk points to a complex inheritance, with additive contributions from common variants or through rare variants with larger effect sizes [3]. The vast majority of ASD is idiopathic, with a specific cause identified in only 4–20% of patients, including a genetic etiology. Association of ASD behavioral phenotypes to specific genetic subtypes is envisaged; however, patients with molecularly defined ASD are not easily clinically identified because clinical and neurobehavioral correlates of a given genetic contribution vary widely [3,4].

Immune dysregulation/inflammation, oxidative stress, and mitochondrial dysfunction are all key pathologic underpinning of ASD [5,6]. Complex biological changes play a role in ASD clinical heterogeneity, hindering the discovery of universal biomarkers for diagnosis and treatments. Currently, ASD diagnosis is based on measurements of behavioral symptoms according to the Diagnostic Statistical Manual of Mental Disorders Version 5 (DSM-5) diagnostic criteria [7]. Recent data point to identifying patient subsets to better define the contribution of certain biological changes in individual patients. In this context, considerable evidence highlights energy metabolism abnormalities in ASD, pointing to acquired mitochondrial dysfunction in a proportion of patients [8,9]. Impaired mitochondrial metabolism may influence neuronal development and synaptic plasticity, which play a major role in neurodevelopment and contribute to ASD. Low free carnitine [10] and abnormal levels of blood acyl-carnitines (AC) were repeatedly found in ASD clinical studies [11,12]. We recently tested the metabolic profile in dried blood spots to support early recognition of young children at risk for ASD diagnosis. We found a significant increase in blood short-chain and long-chain AC and, to a lesser extent, medium-chain AC. Using machine learning analyses, we found a high classification performance of this AC signature to support diagnosis at younger ages (<5 years) [13]. Interestingly, a similar pattern of increased AC had been detected in patients with ASD from the US [12] and in rodent models, where ASD-like behavior was induced by propionic acid [14]. Recently, a global metabolome analysis of plasma and feces found differential levels of AC among the most discriminant metabolites in ASD compared with typically developing (TD) populations. Furthermore, blood C2–C14 AC levels were positively correlated with a more severe impairment of social behavior, supporting a key role for mitochondrial dysfunction in ASD pathophysiology [15].

The mitochondrial fatty acid β-oxidation (mtFAO) is a major energy-producing metabolic pathway, which uses fatty acids to produce adenosine triphosphate (ATP). First, the import of long-chain fatty-acids (LCFA) into mitochondria requires the activity of a multi-enzymatic carnitine-dependent shuttle, with formation of AC intermediates. Within mitochondria, AC are converted back to acyl-CoA, and the various β-oxidation enzymes isoforms progressively shorten the acyl-CoA to produce acetyl-CoA, NADH, and FADH<sup>2</sup> via the Lynen helix [16]. Re-oxidation of produced NADH and FADH<sup>2</sup> by the mitochondrial respiratory chain ultimately results in the production of large amounts of ATP. In addition, very recent data unveiled the involvement of mtFAO in instructing non-energy-related functions, such as chromatin modification or neural stem cell activity [17,18].

Thus, AC enables the transport of fatty acids across mitochondrial membranes, and these conjugated fatty acids typically accumulate when β-oxidation is disturbed. This led to widespread clinical applications in the screening for inborn mtFAO disorders in newborns, implemented in many countries, which is based on the detection of specific accumulating AC in the blood of newborns [19]. In line with this, it can be hypothesized that the increase in plasma AC levels in patients with ASD points to an impairment of mtFAO. To date, the implication of mtFAO in the pathophysiology of ASD is largely unknown, but there are consistent findings documenting AC accumulation, and thereby possible mtFAO impairments, as biomarkers and therapeutic targets in a subset of patients with ASD. In the present study, we therefore analyzed the mtFAO flux and respiratory

chain activities on primary cultured fibroblasts from ASD patients with elevated plasma AC (w-AC) or without elevated plasma AC (w/o-AC).

Resveratrol (RSV) (3,5,4-trihydroxy-trans-stilbene) is a natural polyphenol produced in plants and enriched in grapes and red fruits. Pre-clinical studies showed that RSV ameliorated social behavior and sensory alterations in the rat model of ASD induced by valproic acid [20,21]. At the molecular level, RSV decreased neuroimmune dysregulation through the inhibition of neuronal toll-like receptors and COX-2 signaling [22] and by downregulation of the chemokine receptor in the BTBR T<sup>+</sup> Itpr3tf/J mice model [23]. Interestingly, RSV was also shown to stimulate mtFAO in control human fibroblasts and could restore normal mtFAO rates in fibroblasts from patients with mild forms of inborn mtFAO deficiencies [24]. This further supports a putative therapeutic effect of RSV in patients with ASD [25] and led us to test whether RSV might induce up-regulation of mtFAO in fibroblasts from ASD patients.

Altogether, the present study had several objectives. At first, we aimed to determine if fibroblasts from ASD patients exhibited mitochondrial mtFAO and respiratory chain deficiencies, compared with fibroblasts from control individuals. In parallel, we sought to determine whether cellular mtFAO rates correlated with the AC status of the ASD patients. Then, we tested the effects of RSV on fibroblasts mtFAO rates in patients with ASD and we evaluated whether the mtFAO response upon RSV treatment was depended on clinical characteristics.

#### **2. Materials and Methods**

#### *2.1. Participants*

Protocols used in this study were registered in clinicaltrials.gov as NCT02000284 and NCT02003170 and approved by the Institutional Review Board at the University of Arkansas for Medical Sciences (Little Rock, AR, USA). Parents of participants provided written informed consent. All participants were recruited from the Arkansas Children's Hospital Autism Multispecialty clinic directed by Dr Richard E. Frye (senior author). The ASD diagnosis was documented by at least one of the following: (i) a gold-standard diagnostic instrument, such as the Autism Diagnostic Observation Schedule and/or Autism Diagnostic Interview-Revised (ADI-R); (ii) the state of Arkansas diagnostic standard, defined as agreement of a physician, psychologist, and speech therapist who specializes in ASD; and/or (iii) Diagnostic Statistical Manual of Mental Disorders diagnosis by a physician along with standardized validated questionnaires including the Social Responsiveness Scale (SRS), the Social Communication Questionnaire and the Autism Symptoms Questionnaire, all of which have excellent correspondence to the gold-standard instruments, along with diagnosis confirmation by the referral investigator (senior author). In our recent clinical trial [26], we found that methods (ii) and (iii) were consistent with the ADI-R diagnostic criteria for ASD.

In general, fibroblast samples were obtained for clinical use and then transferred to the research laboratory. For individuals that underwent sedated procedures, most commonly muscle biopsy, the samples were obtained under sedation by the surgeon. For individuals that did not undergo other procedures, Dr Richard E. Frye (senior author) personally obtained the sample by a punch biopsy with local anesthesia.

Overall, fibroblast samples were available from 10 children diagnosed with ASD for this study, 6 males and 4 females aged 7.4 ± 3.2 years (mean ± SD) (range: 3–13). Seven patients (70%) were diagnosed with regressive ASD. Five control fibroblasts from children of similar age who did not manifest any known medical disease or genetic abnormalities were obtained from Coriell Institute for Medical Research (Camden, NJ, USA).

#### *2.2. Neurodevelopmental and Behavioral Measurements*

Neurodevelopment assessment was accomplished by the Vineland Adaptive Behavior Scale (VABS) 2nd edition. The VABS is a valid tool based on structured interview with a caretaker allowing to measure age-appropriate abilities in everyday skills including social

and motor skills, communication, and daily living. The VABS provides standard scores (m = 100, SD = 15) and higher scores indicate better functioning. The Aberrant Behavior Checklist (ABC), a 58-item parent-reported questionnaire, was used to measure behavioral symptoms across 5 subscales (social withdrawal, hyperactivity, stereotypy, inappropriate speech, and irritability) (0–30 raw scores, higher is worse). Multiple ASD clinical trials have used it and it has both convergent and divergent validity [26]. ASD symptoms severity was assessed by the SRS, a 65-item questionnaire completed by a parent or close family member that measures the severity of social skill deficits across five domains (awareness, cognition, communication, motivation and restricted interests, and repetitive behavior) (clinical cut-off ≥ 60).

#### *2.3. Cell Culture and Metabolic Evaluation*

Fibroblasts were derived from skin biopsies obtained with written informed consent from the parents. Control and patient fibroblasts were cultured at 37 ◦C, 5% CO<sup>2</sup> in RPMI with Glutamax™ (Carlsbad, CA, USA) supplemented with 10% (*v*/*v*) fetal bovine serum and 0.2% (*v*/*v*) primocin (InvivoGen, San Diego, CA, USA). For treatment, the media were removed and vehicle (0.04% DMSO) or RSV (75 µM *trans*-RSV, Cayman Chemical, Bertin technologies, Montigny-le-Bretonneux, France), were added to a fresh medium for the last 48 h of culture before mtFAO measurement, as previously described [27]. The mtFAO flux was measured in cultured fibroblasts from w-AC (*N* = 5) and w/o-AC (*N* = 5) patients, and in control fibroblasts (*N* = 5) obtained from healthy subjects with equivalent age and gender distribution. Metabolic evaluation was blinded to patient group, specifically fibroblast assignments were retained confidential and were unveiled at the end of the mtFAO assay for statistical analysis. The FAO flux was determined by quantifying the production of <sup>3</sup>H2O from (9,10-3H) palmitate, as described previously [27]. The FAO assay was run in triplicate and was repeated twice for each cell line. The FAO assay on resveratroltreated fibroblasts was performed once. Results (mean ± SD) were expressed in nmol of tritiated-palmitate oxidized per hour per milligram of protein. The electron transport chain (ETC) activity was tested in frozen muscle biopsies and/or cultured fibroblast cultures as previously reported (Baylor Medical Genetics Laboratory, Houston TX, USA) [28]. Values corrected and uncorrected for citrate synthase (CS) activity were considered.

#### *2.4. Statistical Analyses*

Data were presented as means and standard deviations (SDs) for continuous variables. Statistical analyses on FAO rates were conducted in the three groups of fibroblasts (w-AC, w/o-AC, and controls). The data were preliminarily subjected to the Shapiro–Wilk test to verify the presence of a normal distribution of the sample, and to the Levene's test to verify the homogeneity of the variances between the groups. Two-way analysis of variance (ANOVA) with repeated measures was used to verify whether FAO was influenced by the group itself (inter-group effect) and presence/absence of RSV (intra-group effect). We used the ANOVA to compare the three groups and all combinations of paired groups. Possible differences between paired groups in the ANOVA analyses were then evaluated by *t*-test with the Bonferroni correction. Student's *t*-test was used to compare the mean values of ETC activity and the clinical scores between the ASD study groups. Correlations among study variables were analyzed by the Pearson correlation analysis. Differences with *p* < 0.05 were considered significant. We assumed that the data were normally distributed based on previous studies which have examined these measures in larger sample sizes as well as by examination of our current dataset. Thus, parameter statistical analyses especially the technique used here within which are robust to small differences in parameter distribution—were considered appropriate for analysis.

Additionally, to interpret differences between groups, especially in the context of small sample sizes where biological variability can prevent differences from being statistically significant, we compared differences between groups with the minimally clinical important difference (MCID), a value which indicates whether the difference could be considered

clinically significant. These values for the VABS, SRS, and ABC are provided in previous clinical studies [26]. Data were analyzed using the SPSS Statistics software, version 23 (SPSS, Inc., Chicago, IL, USA, IBM, Somers, NY, USA). considered clinically significant. These values for the VABS, SRS, and ABC are provided in previous clinical studies [26]. Data were analyzed using the SPSS Statistics software, version 23 (SPSS, Inc., Chicago, IL, USA, IBM, Somers, NY, USA).

parameter statistical analyses—especially the technique used here within which are robust to small differences in parameter distribution—were considered appropriate for

Additionally, to interpret differences between groups, especially in the context of small sample sizes where biological variability can prevent differences from being statistically significant, we compared differences between groups with the minimally clinical important difference (MCID), a value which indicates whether the difference could be

#### **3. Results 3. Results**

analysis.

#### *3.1. Blood AC Levels in Patients with ASD 3.1. Blood AC Levels in Patients with ASD*

*J. Pers. Med.* **2021**, *11*, x FOR PEER REVIEW 5 of 13

Participants were divided in two groups: w-AC (n:5; age 7.2 ± 3.7) or w/o-AC (n:5; age 7.6 ± 3.1). Mean age was not significantly different in the two study groups (*p* = 0.429). Patients with high plasma AC levels had consistent AC elevation defined as at least three AC significantly elevated (*p* < 0.05) in repeated analyses [12] (Figure 1). Participants were divided in two groups: w-AC (n:5; age 7.2 ± 3.7) or w/o-AC (n:5; age 7.6 ± 3.1). Mean age was not significantly different in the two study groups (*p* = 0.429). Patients with high plasma AC levels had consistent AC elevation defined as at least three AC significantly elevated (*p* < 0.05) in repeated analyses [12] (Figure 1).

**Figure 1.** Average blood acyl-carnitine (AC) levels in studied patients with autism spectrum disorder (ASD) with (AC+, orange bars) and without (AC−, blue bars) elevations in acyl-carnitines. Short-chain (C5, C5DC), medium-chain (C8:1), and long-chain (C14:2, C16:1) acyl-carnitines (AC) were significantly elevated in a subset of studied patients with ASD. Patients with high plasma AC levels had consistent AC elevation defined as at least three AC clinically significantly elevated (outside the normal reference range) in repeated analyses. **Figure 1.** Average blood acyl-carnitine (AC) levels in studied patients with autism spectrum disorder (ASD) with (AC+, orange bars) and without (AC−, blue bars) elevations in acyl-carnitines. Shortchain (C5, C5DC), medium-chain (C8:1), and long-chain (C14:2, C16:1) acyl-carnitines (AC) were significantly elevated in a subset of studied patients with ASD. Patients with high plasma AC levels had consistent AC elevation defined as at least three AC clinically significantly elevated (outside the normal reference range) in repeated analyses.

#### *3.2. mtFAO Activity in Fibroblast Cultures from Patients with ASD*

There was no overall significant difference in FAO rates between controls and the ASD fibroblasts (Figure 2; F1,8 = 1.107; *p* = 0.368) indicating that under basal conditions, FAO was not significantly impaired in ASD children's fibroblasts, regardless of the AC status. However, significant differences were found between the two ASD patient groups (F1,8 = 5.374; *p* = 0.049). Pairwise *t*-test revealed that the effect found between the ASD

groups in the ANOVA was driven by differences between the two groups of children with ASD, before RSV supplementation. In fact, under basal conditions, the w-AC patients' fibroblasts exhibited significantly decreased FAO values (5.20 ± 0.42) compared with those measured in the w/o-AC group (5.9 ± 0.53) (*p* = 0.044). Basal FAO rates (5.85 ± 1.12) appeared higher in the control compared with the w-AC patients, but no significant differences were found in paired comparisons, likely because of higher variability of FAO rates in controls. Overall, although not significantly impaired compared with controls, baseline FAO values were significantly lower in patients w-AC elevations compared with patients with normal AC blood levels (Figure 2). groups in the ANOVA was driven by differences between the two groups of children with ASD, before RSV supplementation. In fact, under basal conditions, the w-AC patients' fibroblasts exhibited significantly decreased FAO values (5.20 ± 0.42) compared with those measured in the w/o-AC group (5.9 ± 0.53) (*p* = 0.044). Basal FAO rates (5.85 ± 1.12) appeared higher in the control compared with the w-AC patients, but no significant differences were found in paired comparisons, likely because of higher variability of FAO rates in controls. Overall, although not significantly impaired compared with controls, baseline FAO values were significantly lower in patients w-AC elevations compared with patients with normal AC blood levels (Figure 2).

There was no overall significant difference in FAO rates between controls and the ASD fibroblasts (Figure 2; F1,8 = 1.107; *p* = 0.368) indicating that under basal conditions, FAO was not significantly impaired in ASD children's fibroblasts, regardless of the AC status. However, significant differences were found between the two ASD patient groups (F1,8 = 5.374; *p* = 0.049). Pairwise *t*-test revealed that the effect found between the ASD

*J. Pers. Med.* **2021**, *11*, x FOR PEER REVIEW 6 of 13

*3.2. mtFAO Activity in Fibroblast Cultures from Patients with ASD* 

**Figure 2.** Mitochondrial fatty acid oxidation (FAO) rates (blue bars) and effects of resveratrol (RSV) (red bars) in fibroblasts of ASD patients and healthy control individuals. Figure shows participants with autism spectrum disorder (ASD) that also have acyl-carnitine (AC) elevations (ASD w-AC) and participants with ASD without AC elevations (ASD w/o-AC) as well as control participants. Under basal conditions, the β-oxidation flux (nmol 3HFA/h/mg protein) measured in fibroblasts of patients with ASD w-AC were significantly decreased compared with the ASD w/o-AC. RSV significantly increased mtFAO values in all the study groups. The results are means (±SD) of three different experiments. \* *p* < 0.05. **Figure 2.** Mitochondrial fatty acid oxidation (FAO) rates (blue bars) and effects of resveratrol (RSV) (red bars) in fibroblasts of ASD patients and healthy control individuals. Figure shows participants with autism spectrum disorder (ASD) that also have acyl-carnitine (AC) elevations (ASD w-AC) and participants with ASD without AC elevations (ASD w/o-AC) as well as control participants. Under basal conditions, the β-oxidation flux (nmol <sup>3</sup>HFA/h/mg protein) measured in fibroblasts of patients with ASD w-AC were significantly decreased compared with the ASD w/o-AC. RSV significantly increased mtFAO values in all the study groups. The results are means (±SD) of three different experiments. \* *p* < 0.05.

#### *3.3. Electron Transport Chain Complex Activity 3.3. Electron Transport Chain Complex Activity*

Measurements of activities of the ETC complexes were performed on frozen muscle biopsies (*N* = 8) and cultured fibroblasts (*N* = 10). Percentages of normal ETC function, uncorrected or corrected for citrate synthase, were compared between the w-AC and w/o-AC groups (Figure 3). Measurements of activities of the ETC complexes were performed on frozen muscle biopsies (*N* = 8) and cultured fibroblasts (*N* = 10). Percentages of normal ETC function, uncorrected or corrected for citrate synthase, were compared between the w-AC and w/o-AC groups (Figure 3).

The graph values represent percentages of normal ETC function, uncorrected (Figure 3A,C) or corrected for citrate synthase (Figure 3B,D). Activities of the complex II in the muscle (M; Figure 3B) and the complex II/III in fibroblasts (FB; Figure 3D) were significantly reduced in the patients with ASD with plasma AC elevations as compared with patients with ASD without plasma AC elevations (*p* < 0.05). The graph values represent percentages of normal ETC function, uncorrected (Figure 3A,C) or corrected for citrate synthase (Figure 3B,D). Activities of the complex II in the muscle (M; Figure 3B) and the complex II/III in fibroblasts (FB; Figure 3D) were significantly reduced in the patients with ASD with plasma AC elevations as compared with patients with ASD without plasma AC elevations (*p* < 0.05).

**Figure 3.** Electron transport chain (ETC) activity in muscle (M; Panels (**A**,**B**)) and fibroblast (FB) culture (Panels (**C**,**D**)) without (Panels (**A**,**C**)) and with (Panels (**B**,**D**)) correction for citrate synthase (CS) activity, in patients with autism spectrum disorder (ASD) that also have acyl-carnitine (AC) elevations (ASD w-AC, blue bars) and participants with ASD without AC elevations (ASD w/o-AC, red bars). Mitochondrial DNA copy number (mtDNA) is also provided in the graphs for muscle samples. **Figure 3.** Electron transport chain (ETC) activity in muscle (M; Panels (**A**,**B**)) and fibroblast (FB) culture (Panels (**C**,**D**)) without (Panels (**A**,**C**)) and with (Panels (**B**,**D**)) correction for citrate synthase (CS) activity, in patients with autism spectrum disorder (ASD) that also have acyl-carnitine (AC) elevations (ASD w-AC, blue bars) and participants with ASD without AC elevations (ASD w/o-AC, red bars). Mitochondrial DNA copy number (mtDNA) is also provided in the graphs for muscle samples.

#### *3.4. Clinical Characteristics 3.4. Clinical Characteristics*

Standardized clinical assessment in participants with ASD is reported in Table S1. On average, VABS indicated mildly impaired functioning in all three domains [daily living SS (70.1 ± 16.3), communication SS (73.1 ±16.3); and socialization SS (70.5 ± 16)] in all participants with ASD. The levels of functional disability for ASD symptoms, communication, and socialization measured by the VABS subscales were lower in the participants without AC elevations as compared with those with AC elevations, and these differences exceeded the MCID, suggesting that they were clinically observable. However, these differences were not statistically significant, probably because of the small sample size. On average, patients had disruptive behavioral symptoms, such as irritability and hyperactivity, at least of moderate severity (defined as scores greater than or equal to 13 on the ABC). Patients in the w-AC group had higher scores on the ABC social withdrawal, inappropriate speech, and stereotypy subscales compared with the w/o-AC group, and these differences exceeded the MCID, suggesting that they were clinically observable. Patients in the w/o-AC group had higher scores on the ABC hyperactivity and irritability, and they exceeded the MCID. However, all these differences were not Standardized clinical assessment in participants with ASD is reported in Table S1. On average, VABS indicated mildly impaired functioning in all three domains [daily living SS (70.1 ± 16.3), communication SS (73.1 ±16.3); and socialization SS (70.5 ± 16)] in all participants with ASD. The levels of functional disability for ASD symptoms, communication, and socialization measured by the VABS subscales were lower in the participants without AC elevations as compared with those with AC elevations, and these differences exceeded the MCID, suggesting that they were clinically observable. However, these differences were not statistically significant, probably because of the small sample size. On average, patients had disruptive behavioral symptoms, such as irritability and hyperactivity, at least of moderate severity (defined as scores greater than or equal to 13 on the ABC). Patients in the w-AC group had higher scores on the ABC social withdrawal, inappropriate speech, and stereotypy subscales compared with the w/o-AC group, and these differences exceeded the MCID, suggesting that they were clinically observable. Patients in the w/o-AC group had higher scores on the ABC hyperactivity and irritability, and they exceeded the MCID. However, all these differences were not statistically significant, probably because of the small sample size.

statistically significant, probably because of the small sample size. All participants with ASD had social impairment severity in the clinical range (T-scores > 60 on SRS total), with no significant differences between the study groups. All participants with ASD had social impairment severity in the clinical range (Tscores > 60 on SRS total), with no significant differences between the study groups. How-

ever, those with ASD with AC elevations demonstrated less impaired social awareness and cognition but greater repetitive behaviors (mannerism) on the SRS as compared with those with ASD but without AC elevation, and these differences exceeded the MCID. awareness and cognition but greater repetitive behaviors (mannerism) on the SRS as compared with those with ASD but without AC elevation, and these differences exceeded the MCID.

However, those with ASD with AC elevations demonstrated less impaired social

*J. Pers. Med.* **2021**, *11*, x FOR PEER REVIEW 8 of 13

Then we evaluated whether fibroblast mtFAO at baseline and upon RSV treatment was depended on clinical characteristics (Table S2). We found that patients who had the highest mtFAO activity in response to RSV were the most impaired on the SRS–total (*r* = 0.65; *p* = 0.044), SRS–awareness (*r* = 0.72; *p* = 0.019), and SRS–cognition (*r* = 0.76; *p* = 0.011). Moreover, the highest changes to RSV with respect to baseline were observed in patients most impaired on the SRS–total (*r* = 0.87; *p* = 0.001; Figure 4A) as well as on SRS subscales including awareness (*r* = 0.89; *p* < 0.001; Figure 4B), cognition (*r* = 0.88; *p* < 0.001; Figure 4C), communication (*r* = 0.78; *p* = 0.008; Figure 4D), and motivation (*r* = 0.77; *p* = 0.009; Figure 4E). Then we evaluated whether fibroblast mtFAO at baseline and upon RSV treatment was depended on clinical characteristics (Table S2). We found that patients who had the highest mtFAO activity in response to RSV were the most impaired on the SRS–total (*r* = 0.65; *p* = 0.044), SRS–awareness (*r* = 0.72; *p* = 0.019), and SRS–cognition (*r* = 0.76; *p* = 0.011). Moreover, the highest changes to RSV with respect to baseline were observed in patients most impaired on the SRS–total (*r* = 0.87; *p* = 0.001; Figure 4A) as well as on SRS subscales including awareness (*r* = 0.89; *p* < 0.001; Figure 4B), cognition (*r* = 0.88; *p* < 0.001; Figure 4C), communication (*r* = 0.78; *p* = 0.008; Figure 4D), and motivation (*r* = 0.77; *p* = 0.009; Figure 4E).

**Figure 4.** Correlation analyses among clinical characteristics of patients with ASD as measured by the Social Responsiveness Scale (SRS) and resveratrol (RSV) effect on fatty acid oxidation (FAO) in fibroblasts. Highest changes of mtFAO in response to RSV were significantly associated with clinical impairment (higher scores) on the SRS–total (**A**) as well as on SRS awareness (**B**), cognition (**C**), communication (**D**), and motivation (**E**) subscales. **Figure 4.** Correlation analyses among clinical characteristics of patients with ASD as measured by the Social Responsiveness Scale (SRS) and resveratrol (RSV) effect on fatty acid oxidation (FAO) in fibroblasts. Highest changes of mtFAO in response to RSV were significantly associated with clinical impairment (higher scores) on the SRS–total (**A**) as well as on SRS awareness (**B**), cognition (**C**), communication (**D**), and motivation (**E**) subscales.

#### **4. Discussion 4. Discussion**

In the present study, we systematically applied palmitate oxidation rate measurements in fibroblasts to investigate ex vivo mtFAO levels in a series of patients with ASD. We found that under basal conditions, mtFAO in fibroblasts was not significantly different in children with ASD compared with healthy controls. However, the mtFAO flux was significantly decreased in patients with elevated blood AC compared with patients with normal blood AC. Significant differences in basal mtFAO rate between ASD groups may be consistent with recurrent findings of elevated blood AC in subgroups of children In the present study, we systematically applied palmitate oxidation rate measurements in fibroblasts to investigate ex vivo mtFAO levels in a series of patients with ASD. We found that under basal conditions, mtFAO in fibroblasts was not significantly different in children with ASD compared with healthy controls. However, the mtFAO flux was significantly decreased in patients with elevated blood AC compared with patients with normal blood AC. Significant differences in basal mtFAO rate between ASD groups may be consistent with recurrent findings of elevated blood AC in subgroups of children with ASD [12,13].

with ASD [12,13]. As mentioned above, screening of inborn mtFAO deficiencies in newborns is based on acyl-carnitines analysis in dried blood spots, which needs to be followed by confirmatory testing to define the ultimate diagnosis [19]. Importantly, patients diagnosed in this way at birth are mostly asymptomatic, and the AC pattern is not predictive of sympto-As mentioned above, screening of inborn mtFAO deficiencies in newborns is based on acyl-carnitines analysis in dried blood spots, which needs to be followed by confirmatory testing to define the ultimate diagnosis [19]. Importantly, patients diagnosed in this way at birth are mostly asymptomatic, and the AC pattern is not predictive of symptomatology [29]. In many cases, the symptom-free period will last for years, and the disease manifestations will only develop in the adolescence or adulthood, generally triggered by exercise, cold, fever, or other metabolic stress situations. This is true in particular for the

VLCAD deficiency, one of the most common inborn mtFAO disorders. In the mild—most frequent—form of this disorder, the typical C (14:1) AC accumulation detected at birth can be associated with near-normal FAO flux values in the patients' fibroblasts and with the absence of clinical manifestations [30–32]. In adult patients with the muscular form of carnitine-palmitoyl transferase II (CPT-II) deficiency, high plasma levels of long-chain AC can be measured in the absence of disease symptoms, and mtFAO values in patient fibroblasts may also appear marginally modified [33]. In sum, evidence from genetic mt-FAO defects suggests that an inefficient mtFAO flux with accumulation of AC biomarkers may arise without obvious symptoms affecting mtFAO-dependent organs and even in the presence of a relatively high residual mtFAO activity. The occurrence of developmental delay, autistic-like behavior, or ASD in genetic defects of mtFAO, particularly VLCAD [34] and LCHAD [35], suggests that impaired mtFAO may contribute to dysfunctional energy metabolism in subsets of patients with ASD. Furthermore, interestingly, post-hoc analysis of newborn screening data in a large (>9000 individuals) cohort showed that high levels of ACs could be associated with an increased risk of ASD [36]. Specific abnormalities in AC also characterize an animal model of ASD in which propionic acid, a microbiomeproduced short chain fatty acid, results in ASD-like behavior, mitochondrial dysfunction, and neuroinflammation [37].

In addition to reduced mtFAO oxidation rate, patients with ASD and plasma AC elevations had significantly lower ETC complex II and complex II/III activities in muscle and fibroblasts, respectively, when compared with patients with normal AC blood levels. This is consistent with a previous study that demonstrated a partial deficit in complexes I/III and I/III RS functions in muscle of patients with ASD and elevated blood AC [12]. As a whole, the current findings support recognition of an ASD patient subset characterized by impaired mtFAO efficiency associated with abnormal blood AC underlying decreased energy production in different cell types.

It is noteworthy that our finding of a >200% increase in the ETC complex I/III activity in patients with ASD is consistent with recent findings illustrating that an elevation of the respiratory chain activity may be abnormal and associated with ASD. An increase in the baseline respiratory chain activity in lymphoblastoid cell lines from subsets of patients with ASD is associated with increased vulnerability to environmental toxicants and to physiological stress [38,39]. It was shown that patients with ASD and mitochondrial dysfunction have higher rates of neurodevelopmental regression with loss of acquired abilities following fever or infections [40] as not uncommonly seen in patients with ETC mitochondrial disorders [41]. Such abnormalities are consistent with acquired mitochondrial dysfunction in ASD patients with neurodevelopmental regression [42] and with studies which demonstrated that mitochondrial dysfunction in ASD can be associated with environmental factors, such as fever [40], immune activation and oxidative stress [43], microbiome metabolites [37], air pollution [44], and prenatal nutritional deficiencies [45]. Some other symptoms of mitochondrial disease, including hypotonia and motor delay, seizures, and gastrointestinal disturbances are frequently encountered in ASD as well [5].

Previous research established that ex vivo measurements of palmitate oxidation rate provide a sensitive approach to test the efficiency of compounds potentially capable to stimulate mtFAO, such as fibrates and RSV [24,27,46]. Pre-clinical studies using different ASD models showed the effects of RSV on several pathways involved in ASD, such as decreasing microglia-induced neuroinflammation and oxidative stress, by reducing oxygen and nitrogen reactive species and neurotransmitter imbalance [22,47]. More recently, RSV was found capable of ameliorating social behavioral deficits in the oxytocin receptor gene knockout by up-regulation of the silent information regulator 1 (Sirt1) gene and early growth response factor 3 (Egr3) gene expressions in the amygdala of Oxtr-KO mice [48]. In 2020, a double-blind, placebo-controlled clinical trial investigated the effects of RSV as an adjunctive treatment in decreasing disruptive behavioral symptoms measured by SRS in patients with ASD. RSV add-on therapy to Risperidone did not yield any significant improvement on the irritability subscale compared with placebo but led to significant

improvement on the hyperactivity/non-compliance subscale after a 10-week period [49]. No severe adverse effects were observed and no significant difference in the frequency of adverse effects was documented between RSV-treated versus placebo group, thus supporting further clinical studies using RSV monotherapy in patients with ASD.

For the first time, the current study explored the effects of RSV on mtFAO in fibroblasts from patients with ASD. It showed that RSV significantly increased mtFAO activity in fibroblasts from patients with ASD as well as in controls. RSV effects on FAO are mediated by peroxisome proliferator activated receptor gamma co-activator-1-alpha (PGC-1α). PGC-1α serves as a co-activator of various transcription factors [PPARs, NRF 1 and 2 (nuclear respiratory factors), hepatic nuclear factor 4], thus regulating the expression of several enzymes involved in the mtFAO pathway, mitochondrial biogenesis, oxidative phosphorylation, and energy production [24]. Importantly, we showed an association between RSV effects on mtFAO activity in fibroblasts from children with ASD and pertinent clinical characteristics. We found that the highest changes to RSV with respect to baseline were observed in patients with more severe ASD symptoms on the SRS scale, such as impaired awareness, cognition, social communication, and motivation.

The main limitations of the present study include the small size of the samples used for ex vivo analyses of mtFAO. This suggests that the study is underpowered, so nonsignificant differences cannot be claimed to signify that no difference exists. Clearly, further studies with larger sample sizes will be needed to follow up this work. Secondly, it should be noted that results from pre-clinical studies such as the present one may partially apply to the in vivo condition and will ultimately require to be tested in clinical trials. As ASD is heterogeneous in nature, there is a need to characterize patients also considering what possible differences in metabolic profiling might lead to different clinical responses to the same therapies. Taken into account that ASD is unequivocally associated with mtFAO impairment in patient subsets, the effects of RSV in ameliorating mtFAO in patient cells may be relevant for understanding the therapeutic response.

#### **5. Conclusions**

In conclusion, we found a significant difference in fibroblasts basal mtFAO rates between ASD groups that differ by the presence or absence of elevated blood acyl-carnitines. The study showed for the first time that the mtFAO activity in fibroblasts of ASD children increased significantly after RSV and that the highest changes to RSV effects on mtFAO occurred in the most severely affected patients. In the light of the present findings, future clinical trials might well consider whether possible RSV effects in ameliorating behavioral symptoms are associated with baseline symptoms severity in patients with ASD. Moreover, further studies should take into account how and to what extent different metabolic profiles may influence response to therapies in patients with ASD.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/jpm11060510/s1, Table S1: Standardized clinical assessment scores in participants with ASD. Table S2: Correlation analysis (*r*) among clinical severity and mitochondrial fatty acid oxidation (mtFAO) activity in fibroblasts from patients with ASD under basal conditions and after resveratrol (RSV+) treatment.

**Author Contributions:** Conceptualization, R.B., J.B. and R.E.F.; Data curation, R.B., J.B., F.D., R.S. and R.E.F.; Formal analysis, J.B., F.D., R.S. and A.C.; Investigation, J.B., F.D., I.S., M.A.K., A.A., P.J.M. and L.D.; Methodology, R.B., J.B., A.C. and R.E.F.; Project administration, R.S. and R.E.F.; Writing original draft, R.B., J.B. and R.E.F.; Writing—review & editing, R.B., J.B., R.R. and R.E.F.; Funding Acquisition: R.R., R.E.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded, in part, by the N of 1 Foundation (Dallas, TX, USA), the Phoenix Children's Hospital Foundation (Phoenix, AZ, USA), the Arkansas Biosciences Institute (Little Rock, AR, USA), The Jonty Foundation (St Paul, MN, USA), the Gupta Family Foundation (Atherton, CA, USA), and the Jager Family Foundation (Chicago, IL, USA)to R.E.F. and by Piano di Ricerca di Ateneo 2020-22 linea 2 grant from the University of Catania (Catania, Italy) to R.B. and R.R.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki. Protocols used in this study were registered in clinicaltrials.gov as NCT02000284 and NCT02003170 and approved by the Institutional Review Board at the University of Arkansas for Medical Sciences (Little Rock, AR, USA).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** All data presented in the study are included in the article. Further inquiries can be directed to the corresponding authors.

**Acknowledgments:** The authors would like to thank John Slattery for his help with coordination participants, Sirish C. Bennuri for help coordinating samples, Stephen Kahler for his help with patient recruitment, and the families who have volunteered to be involved in the research despite their demanding lives.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Folate Receptor Alpha Autoantibodies in Autism Spectrum Disorders: Diagnosis, Treatment and Prevention**

**Natasha Bobrowski-Khoury <sup>1</sup> , Vincent T. Ramaekers <sup>2</sup> , Jeffrey M. Sequeira <sup>3</sup> and Edward V. Quadros 3,\***


**Abstract:** Folate deficiency and folate receptor autoimmune disorder are major contributors to infertility, pregnancy related complications and abnormal fetal development including structural and functional abnormalities of the brain. Food fortification and prenatal folic acid supplementation has reduced the incidence of neural tube defect (NTD) pregnancies but is unlikely to prevent pregnancyrelated complications in the presence of folate receptor autoantibodies (FRAb). In pregnancy, these autoantibodies can block folate transport to the fetus and in young children, folate transport to the brain. These antibodies are prevalent in neural tube defect pregnancies and in developmental disorders such as cerebral folate deficiency (CFD) syndrome and autism spectrum disorder (ASD). In the latter conditions, folinic acid treatment has shown clinical improvement in some of the core ASD deficits. Early testing for folate receptor autoantibodies and intervention is likely to result in a positive outcome. This review discusses the first identification of FRAb in women with a history of neural tube defect pregnancy and FRAb's association with sub-fertility and preterm birth. Autoantibodies against folate receptor alpha (FRα) are present in about 70% of the children with a diagnosis of ASD, and a significant number of these children respond to oral folinic acid with overall improvements in speech, language and social interaction. The diagnosis of folate receptor autoimmune disorder by measuring autoantibodies against FRα in the serum provides a marker with the potential for treatment and perhaps preventing the pathologic consequences of folate receptor autoimmune disorder.

**Keywords:** autism spectrum disorders; folate receptor alpha; folates; pregnancy; brain development; fetal development

#### **1. Background**

Folate, an umbrella term used for metabolically active forms of folic acid (B9), is an essential B-complex vitamin necessary for basic cellular metabolism including, but not limited to, essential cellular DNA synthesis, repair and methylation including regulation of synthesis and metabolism of monoamine neurotransmitters. As a nutrient found in green leafy vegetables, legumes and fruits, it is readily absorbed by the upper small intestine after breakdown from polyglutamates to monoglutamates. Folate in its active forms facilitates one-carbon transfer reactions and contributes to the synthesis of purines, pyrimidines and amino acids [1]. One of its most characterized roles is facilitating single carbon transfer to homocysteine to form methionine. This reaction is critical for maintaining intracellular Sadenosyl methionine, an essential compound for methylation reactions. Folate also has a codependent relationship with vitamin B<sup>12</sup> in that both vitamins must be present in adequate amounts for conversion to the physiologic forms that participate in metabolic reactions. If folate and B<sup>12</sup> are not adequate, cellular metabolism and replication is interrupted [2,3]. This is most critical during fetal and neonatal development because inadequate folate during this

**Citation:** Bobrowski-Khoury, N.; Ramaekers, V.T.; Sequeira, J.M.; Quadros, E.V. Folate Receptor Alpha Autoantibodies in Autism Spectrum Disorders: Diagnosis, Treatment and Prevention. *J. Pers. Med.* **2021**, *11*, 710. https://doi.org/10.3390/jpm11080710

Academic Editors: Richard E. Frye, Richard Boles, Shannon Rose and Daniel Rossignol

Received: 27 June 2021 Accepted: 21 July 2021 Published: 24 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

period can result in interruptions in brain development leading to structural abnormalities that produce functional deficits of the CFD syndrome. Low cerebro-spinal fluid (CSF) folate is a characteristic feature of CFD syndrome, as first described by Ramaekers and Blau [4]. On rare occasions, CFD can also result from mutations in the FRα gene [5–7], but the most common cause of low CSF folate in CFD is the presence of anti-folate receptor antibodies (FRAb) that can block folate transport across the choroid plexus [8,9]. A recent report has identified mutations in the *CIC* transcription factor gene in children diagnosed with CFD syndrome. Mutations in the *CIC* gene decrease the expression of FRα to reduce folate transport across the choroid plexus [10]. No abnormalities of the FRα gene are found in ASD, but a majority of these children are positive for FRAb and have low CSF folate [11,12]. This is *a priori* proof that FRα is the primary transporter of folate into the brain under physiologic folate status.

#### **2. Folate Requirements during Pregnancy**

Since the discovery of its role in megaloblastic anemia and spina bifida, folate supplementation during pregnancy and fortification of food products have become two of the most globally accepted methods of treating and preventing folate deficiency. The basic folate requirement increases 75 to 100% (approximately 300–400 µg per day) in pregnancy because folate has a critical role in the growth and development of the embryo/fetus, especially during early stages of development [13]. It is, therefore, common practice to recommend that women supplement their diet with folate before conception and throughout pregnancy. The prevention of folate deficiency during pregnancy is achieved by consumption of at least 0.4 mg/day of folic acid during the first trimester of pregnancy [14,15]. In light of the recently discovered FRAb that can block folate transport, women positive for these antibodies may need additional supplementation with folinic acid to provide adequate folate to the developing fetus [16,17].

#### **3. Folate and Fetal Brain Development**

The importance of folate during embryonic and fetal brain development has been demonstrated in genetic animal models and dietary manipulations of folate deficiency [18,19]. If either folate transport or folate concentration in circulation is adversely manipulated, embryonic and fetal development is significantly altered. Mouse knockout models of genes such as FOLR1 that encode for folate receptor alpha (FRα) produce lethality in litters along with orbito-facial abnormalities, congenital heart defects and/or neural tube defects [20]. In FOLR1 knockout mouse, these lethalities can be prevented with adequate folinic acid (N5-formyltetrahydrofolate, a reduced form of folate) supplementation. These dramatic results occur because folate transport is lacking in the KO mouse during the early stages of neurulation and in regions where abnormalities arise [21]. In rodent models, folate deficiency causes a decrease in progenitor cells and an increase in apoptosis, and this could lead to infertility or resorption of embryos or fetal malformations [22]. Behavioral deficits are seen in rat pups born to folate-deficient mothers [23] and on methyl donor deficient diet during pregnancy [24]. In a rat model of exposure to rat folate receptor antibodies during pregnancy, resorption of embryos and malformations of the cranio-facial region and the brain were reported [25]. When the antibodies were administered at lower doses, embryos were carried to term with normal appearing pups born. However, these pups showed severe behavioral deficits [23,26]. The behavioral phenotype can be rescued by treatment with folinic acid and dexamethasone prior to antibody exposure [27]. These studies provide strong evidence in support of the pathologic consequences of exposure to FRα antibodies and the protective role of folinic acid.

#### **4. Folate and Neonatal Brain Development**

After birth, it is crucial for the offspring to have an adequate amount of folate in their diet. Instead of rapid cell division as embryogenesis calls for, postnatal development requires folate for neural progenitor differentiation as well as proliferation [28]. It has yet

to be fully elucidated what the detailed mechanisms of folate action are, but the folate deficiency produced in animal models during early postnatal development illustrates the importance of folate in preventing developmental and cognitive deficits [23,27]. Researchers have also reported changes in neuronal excitability and maintenance that arise with a decrease in brain folate in a rat model [29]. Others have reported an increase in p53 and signs of homocysteine accumulation in the neurons and astrocytes [30]. There was a long-term effect on locomotor function and cognition in these animals. Therefore, folate is necessary for maintenance of neuronal function, as well. Based on this, further investigations into the mechanisms of folate metabolism in neurons and support cells of the brain are necessary. Thus far, folate has been linked to neuronal repair and differentiation after injury, myelin formation and maintenance and neuronal plasticity [30–32]. Figure 1 provides a summary of the effects of folate deficiency on fetal and post-natal brain development and the consequent sequelae that contribute to neurologic deficits. deficiency produced in animal models during early postnatal development illustrates the importance of folate in preventing developmental and cognitive deficits [23,27]. Researchers have also reported changes in neuronal excitability and maintenance that arise with a decrease in brain folate in a rat model [29]. Others have reported an increase in p53 and signs of homocysteine accumulation in the neurons and astrocytes [30]. There was a long-term effect on locomotor function and cognition in these animals. Therefore, folate is necessary for maintenance of neuronal function, as well. Based on this, further investigations into the mechanisms of folate metabolism in neurons and support cells of the brain are necessary. Thus far, folate has been linked to neuronal repair and differentiation after injury, myelin formation and maintenance and neuronal plasticity [30–32]. Figure 1 provides a summary of the effects of folate deficiency on fetal and postnatal brain development and the consequent sequelae that contribute to neurologic deficits.

After birth, it is crucial for the offspring to have an adequate amount of folate in their diet. Instead of rapid cell division as embryogenesis calls for, postnatal development requires folate for neural progenitor differentiation as well as proliferation [28]. It has yet to be fully elucidated what the detailed mechanisms of folate action are, but the folate

*J. Pers. Med.* **2021**, *11*, x FOR PEER REVIEW 3 of 16

**Figure 1.** Effects of folate deficiency on the fetus and on brain development. Multiple causes lead to systemic as well as fetal folate deficiency. Folate receptor autoantibodies can block folate transport to the fetus and to the fetal as well as neonatal brain. In addition to folate deficiency, immune-mediated inflammation can contribute to the pathology. This has multipronged effects on brain development and function. **Figure 1.** Effects of folate deficiency on the fetus and on brain development. Multiple causes lead to systemic as wellas fetal folate deficiency. Folate receptor autoantibodies can block folate transport to the fetus and to the fetal as well as neonatal brain. In addition to folate deficiency, immune-mediated inflammation can contribute to the pathology. This has multipronged effects on brain development and function.

#### **5. Folate Receptors: Expression and Function**

In humans, there are four genes that code for folate receptors (see Table 1). The most characterized of these receptors is folate receptor alpha (FRα). As extracellular receptors, FRα, FRβ, FRγ and FRδ function as transporters of folate across different target

tissues [33,34]. FRα can also act as a transcription factor [33]. Other transporters of folate include the reduced folate carrier (RFC), which requires high local concentrations (micromolar) of biologically active reduced forms of folates, and the proton-coupled folate transporter (PCFT), which can only transport folates and folic acid under acidic conditions and is the primary transporter involved in folate absorption in the gut [35].


**Table 1.** Summary of folate transporters.

#### **6. FR**α **Role in Maternofetal Transport of Folate**

The high demand for folate during pregnancy requires homeostatic mechanisms to ensure that sufficient folate is provided to the fetus throughout development. As the most characterized receptor in the folate transporter family of proteins, the accepted mechanism of FRα-mediated transport is translocation/endocytosis of the holo receptor subsequent to folate binding [35]. FRα is expressed on all epithelial cells including the choroid plexus. It is highly expressed in the reproductive tissues including the placenta and the fetus. To determine the mechanism of folate transport in the placenta during pregnancy, Yasuda et al. [41] manipulated osmolarity, concentrations of phosphatidylinositol-specific phospholipase C inhibition and concentrations of <sup>3</sup>H-folic acid *in vitro* culture of human placental brush border membrane vesicles and determined that FRα, RFC and PCFT could transport various forms of folate, but that approximately 60% of folate was binding to FRα. They also noted that the folate requirements of Wistar rats increased across gestation, and expression of the mRNA of the transporters increased as well.

#### **7. FR**α **Role in Folate Transport to the Brain**

FRα is accepted as the main transporter of folate into the brain. However, there have been limitations to studying how FRα transports folate across the blood–brain barrier. A potential mechanism of folate transport across the choroid plexus and into the brain has been described by Grapp et al. [42]. In their experiments using immortalized Z310n rat choroid plexus cells in culture and a mouse model, they determined that transport of folate required shuttling of folates via exosomes from the basolateral side of the choroid plexus to the brain parenchyma of the apical side. Alternative transporters such as RFC and PCFT may only play a role when there is a disruption of FRα expression and transport, and adequate folate concentration is made available locally at the receptor [43]. The shuttling across the epithelial lining of the choroid plexus is a mechanism presumed to be conserved in all tissues that express FRα [44].

#### **8. Folate Receptor Autoantibodies: Their Role in Disrupting Folate Transport**

In some conditions, there is disruption in folate utilization that is not related to a dietary deficiency but is most likely due to a disturbance in the folate's transport due to genetic or metabolic abnormalities. An emerging culprit of folate transport disruption is folate receptor autoimmune disorder, where autoantibodies against the FRα can interfere with folate transport to the fetus; it has been associated with subfertility, difficulty in conceiving, miscarriage and neural tube defects in the fetus [16,17,45,46].

In infants and young children, these antibodies can block folate transport to the brain. Approximately 70% of the children diagnosed with cerebral folate deficiency syndrome or autism spectrum disorder have low CSF folate and respond to folinic acid treatment [47,48]. The majority of the autoantibodies are of the IgG class and, therefore, can readily cross the placenta and affect the fetus. Two distinct types of antibodies have been identified. One binds to FRα at the active site where folate binds and, as a consequence, blocks folate binding (blocking Ab). Another type of antibody binds to an antigenic site not involved in folate binding (binding Ab) but can trigger an immune reaction and inflammation and render the receptor nonfunctional. In most cases, one or both types of antibodies are present [49,50]. Thus, functional blocking of folate transport and inflammation are an integral part of the pathology [44].

#### **9. Pathologic Consequences of Folate Receptor Antibodies**

The presence of folate receptor autoantibodies can disrupt the transport of folate, and the consequences of decreased folate uptake by cells can impact development of the fetus, especially the central nervous system. There is also a correlation of folate receptor antibodies with neural tube defect pregnancy [16]. In less severe cases, a subset of children born with exposure to maternal FRα autoantibodies *in utero* develop lowfunctioning autism with or without neurological deficits after birth. Recent studies show significant association of folate receptor autoantibodies with autism spectrum disorder in children [11,51,52].

#### **10. Diagnosis of Folate Receptor Autoimmune Disorder**

Early indications of cerebral folate deficiency that are potentially due to maternal folate deficiency or folate receptor autoantibodies can be deduced by measuring serum folate and homocysteine and folate receptor autoantibodies in the mother during pregnancy. Other than dietary folate deficiency, folate receptor autoantibodies in the pregnant mother can contribute to fetal folate deficiency. In the latter case, blocking of folate transport across the placenta and antibody-mediated inflammation could contribute to the pathology, as shown in the rat model of exposure to rat folate receptor antibodies during pregnancy [26,27]. In infants, the presence of folate receptor autoantibodies in the blood could provide a mechanism by which folate transport to the brain via the choroid plexus could be blocked, thus leading to cerebral folate deficiency [51,52]. Therefore, determining the presence of folate receptor autoantibodies in the blood of pregnant mothers and children becomes a necessary test to prove or rule out folate receptor autoimmune disorder.

Methodology for determination of the antibody titer in serum is well-established. Two distinct types of IgG and/or IgM antibodies have been described [50]. These antibodies can be blocking and/or binding antibodies. Both types of antibodies are capable of triggering an immune reaction due to antigen/antibody interaction, leading to local inflammation, and this could interfere with folate transport via the FR protein. Both types of assays can be performed in a laboratory setting as described below.

#### **11. Assay for Blocking Antibodies**

Blocking autoantibodies to FRα are determined using a functional binding radio assay. Patient's serum (200 µL) is acidified with 300 µL 0.1 M glycine/HCl pH 2.5/0.5% Triton X-100/10 mM EDTA and added to 12.5 mg charcoal pellets in a separate tube (250 µL of 5% charcoal/1% dextran in 0.1 M Na PO4 pH 7.4/0.5% Triton X-100/10 mM

EDTA, spun down and supernatant-aspirated) to remove any endogenous folate, and the pH of the supernatant fluid is neutralized with 40 µL of 1 M dibasic NaPO4 prior to using it in the assay. This assay is performed by adding purified apo human FRα protein (40 ng) to the processed serum and incubating overnight at 4 ◦C. The next day, <sup>3</sup>H-folic acid (700 pg) (Moravek) is added and incubated for 20 min at room temperature. Unbound <sup>3</sup>H folic acid is removed with dextran-coated charcoal (200 µL) and the <sup>3</sup>H folate bound to FRα determined by counting the sample in a liquid scintillation counter. The reduction in binding of <sup>3</sup>H-folic acid to the apo human FRα when compared to the negative control serum sample provides a measure of the blocking autoantibody present in the sample [50]. Blocking antibody can be IgG or IgM; the values are expressed as pico moles of <sup>3</sup>HPGA blocked per ml serum, and the titer can range from >0.2 to 0.5 (low titer), >0.5 to 1.0 (medium titer) or >1.0 (high titer).

#### **12. Assay for Binding Antibody**

Binding of the IgG autoantibody to folate receptor alpha (FRα) is determined by an ELISA-based method. FRα (1 µg in 100 µL) purified from human milk is added to each well of an ELISA plate to covalently bind the protein to maleic anhydride-coated wells (Thermo Fisher, Waltham, MA, USA). Following blocking of additional sites by treatment with normal goat serum (200 µL) overnight to prevent non-specific binding to the wells, serum samples (4 and 8 µL) (negative control, positive control and patient samples) are added to wells along with 100 µL fresh goat serum and incubated at 4 ◦C overnight to facilitate binding of autoantibodies to the FRα in the wells. Following washing of the wells to remove unbound proteins, the specific IgG autoantibody bound in each well is detected by incubating with a peroxidase-conjugated, anti-human IgG secondary antibody (1:6000 dilution) (Vector Labs) for 1 h at room temperature. After washing to remove the unbound secondary antibody, the bound peroxidase-conjugated secondary antibody is determined by incubation with ultra TMB (Thermo Fisher) for 10 min. The resultant blue colored reaction is converted to yellow with 100 µL of 1.0 M HCl, and then absorbance is read at 450 nm in an ELISA plate reader. In a second set of wells, known amounts of human IgG captured in protein A-coated plates are used to construct a standard curve [50]. Values are expressed as pico moles of IgG antibody per ml serum and can range from >0.1 to 0.5 (low titer); >0.5 to 2.0 (medium titer) and >2.0 (high titer).

Among other criteria, specific diagnosis of folate receptor autoimmune disorder is confirmed using the above tests. After correcting for background, for the blocking antibody, values of 0.2 pmoles or greater are considered positive and for the binding antibody, 0.1 pmoles or greater are considered positive. Because folate receptor alpha is a peripheral membrane protein, the antibody titer measured in the serum should be considered as excess antibody appearing in the circulation after saturating the membrane-bound antigen. Fluctuations in antibody titer have been reported in the same individual over time and can range from low to medium titer or to undetectable levels. While the reason for these changes in antibody titer are not identified, it is likely that changes in FR antigen on cells, exposure to milk FR antigen in the gut and the specific B-cell population may be contributory factors.

#### **13. Treatment of FR**α **Autoimmune Disorder in ASD**

Among the developmental disorders, ASD is most prevalent and has continued to increase over the past decade. Based on available publications, the WHO reports the worldwide prevalence at 1 in 162 births [53]. In the USA and Canada, the prevalence is reported at 1 in 50, and this rate is predicted to increase over the next few years [54]. While the clinical phenotype of ASD may result from multiple genetic, epigenetic and environmental factors, nutrient deficiencies such as folate can play a significant role. Folate receptor antibodies and cerebral folate deficiency are prevalent in ASD. Treatment of FRα autoimmunity in ASD is based on our previous findings in infantile-onset CFD syndrome and low-functioning autism associated with neurological deficits [11]. In these children,

a repeat CSF analysis after three to six months of treatment with folinic acid showed normalization of 5-methyl-tetrahydrofolate levels [11,51]. Supplementation with high-dose dl-folinic acid (Leucovorin) (0.5–2 mg/kg body weight or 0.25–1.0 mg levofolinate) given in 1 or 2 divided daily doses increases 5-methyltetrahydrofolate concentration by more than 100-fold compared to physiological folate concentrations in plasma. Despite the autoantibody-induced blocking of the FRα-pathway to transport folate across the choroid plexus, the significant increase of 5-methyltetrahydrofolate and folinic acid in plasma will enable reduced folate carrier-1 (RFC-1), a high capacity/low affinity transmembrane folate transporter at the blood–brain barrier, to transport sufficient 5-methyltetrahydrofolate and folinic acid to the brain. In this context, it appears important to verify a normal vitamin D status because RFC-1 gene transcription depends on vitamin D availability within microvasculature cells at the blood–brain barrier [40].

Another important therapeutic intervention represents a diet strictly free of animalderived milk or milk products, which can be replaced by other vegetal milk products (for example soya-, almond- and rice-based and coconut milk). Although many previous studies on a casein/gluten-free diet have been conducted, there has been no final evidence yet to consider these dietary treatments as beneficial in the management of ASD. Many studies have been conducted for a maximum of only 3 months, although some studies on a small number of patients were conducted over 1 to 2 years and indicated that part of the core symptoms of autism had improved [55–58]. The conclusion was that a casein/gluten-free diet should be tried for at least 6 months to see a positive response in a subset of the ASD population. One suggested hypothesis was that opioid peptides derived from milk casein contributed to the pathogenesis of autism [59]. Because bovine milk contains a soluble form of the FRα protein with 91% homology with human FRα, we examined the binding properties of human FRα autoantibodies with different FRα antigens isolated from human placenta; human milk; and bovine, goat and camel milk. The highest cross-reactivity of the autoantibody was found for soluble FRα protein from bovine and camel milk (Figure 2). To determine if FRα in the milk consumed contributed to the autoimmune disorder, we studied the effects of a milk-free diet in children positive for the FRα antibody. Patients with infantile CFD syndrome associated with FRα antibodies were randomized to receive either a cow's milk-free diet or a normal, milk-containing diet. Among children on a normal diet, FRα antibodies increased from baseline toward higher titers during 6–12 months of evaluation. However, the children receiving a milk-free diet showed a significant drop in FRα antibody titers after 3–6 months that rose again after re-introduction of bovine milk. These studies confirmed down-regulation of the FRα antibodies following a strict animal milk-free diet [55]. In this group of patients with infantile CFD syndrome, a number also suffered from low-functioning autism with neurological deficits and showed a clinical response after a milk-free diet. These findings suggest that in predisposed individuals, the soluble FRα antigen derived from bovine and other animal-derived milk products acts as the antigen that triggers a gut immune response with the formation of specific B-cell clones that produce autoantibodies that enter the circulation, cross-reacts with the human FRα anchored to the choroid plexus and blocks folate transport from the circulation into the CSF [44,60]. Thyroid dysfunction is common in children with ASD. Even though FRα expression in the thyroid gland is decreased in older children and adults, it is highly expressed in the fetal and neonatal thyroid, and FRα antibodies can affect development of the thyroid gland [61]. A preferred strategy for individuals with autism spectrum disorder is to take a serum sample for determination of FRα autoantibodies after exposure to milk products for about 2–3 weeks. After this diagnostic blood test, autistic children can be placed on an animal milk-free diet. As soon as the FRα autoantibodies test positive, a milk-free dietary intervention can be continued along with high oral doses of folinic acid. Other treatment strategies to reduce FRα autoantibodies may be immunosuppression using steroids or intravenously administered immunoglobulins, but these therapeutic options should be reserved for emergency situations such as refractory epileptic seizures or dramatic movement disorders such as dystonia, choreoathetosis or ballism.

disorders such as dystonia, choreoathetosis or ballism.

FRα autoantibodies may be immunosuppression using steroids or intravenously administered immunoglobulins, but these therapeutic options should be reserved for emergency situations such as refractory epileptic seizures or dramatic movement

**Figure 2.** (**A**) Folate receptor concentrations in milk. (**B**) Immune cross-reactivity of blocking autoantibodies against various FR antigens. The blocking assay was performed by testing a known amount of blocking activity in serum samples from 8 different patients with molar equivalent amounts of FR antigens from milk. Blocking activity was determined as pico moles of 3HPGA blocked and was compared to blocking in human milk antigen. **Figure 2.** (**A**) Folate receptor concentrations in milk. (**B**) Immune cross-reactivity of blocking autoantibodies against various FR antigens. The blocking assay was performed by testing a known amount of blocking activity in serum samples from 8 different patients with molar equivalent amounts of FR antigens from milk. Blocking activity was determined as pico moles of <sup>3</sup>HPGA blocked and was compared to blocking in human milk antigen.

Treatment with high-dose folinic acid in a subgroup of ASD children positive for FRα autoimmunity, i.e., low-functioning autism with neurological deficits, showed clinical improvement of core autistic symptoms and normalization of previously lowered CSF 5 methyltetrahydrofolate [51]. A double-blind, placebo-controlled study conducted among children with ASD without additional neurologic deficits showed significant improvements in verbal scores in subjects positive for FRAb following treatment with folinic acid [48]. A recent, self-controlled clinical trial was conducted among children with low-functioning autism without additional neurological complications. In these, a high, 76% prevalence of FRα antibodies was found. These children also had multiple nutrient deficiencies attributed to selective eating habits and malnutrition. Combined correction of deficient nutrients and high dose folinic acid administration resulted in an overall significant recovery from severe autism to mild–moderate autism (Figure 3A). Comparison of the Childhood Autism Rating Scale (CARS) after 2 years of treatment (folinic acid supplementation and correction of abnormal nutrient values) with the CARS at baseline showed better outcomes for children having negative or low FRα antibody titers of the blocking type, up to 0.44 pmol FRα blocked/mL serum, versus the group whose FRα antibody titers were above 0.44. The baseline CARS score increased as a function of the age at which treatment was initiated. The outcome became poorer for the older subgroup of treated autistic children (Figure 3B). This outcome may be further compounded by the presence of maternal and paternal autoantibodies and embryonic exposure to these. Preliminary data suggested that in the event of maternal or parental FRα autoantibodies, the child´s outcome after treatment was also less favorable (Figure Treatment with high-dose folinic acid in a subgroup of ASD children positive for FRα autoimmunity, i.e., low-functioning autism with neurological deficits, showed clinical improvement of core autistic symptoms and normalization of previously lowered CSF 5-methyltetrahydrofolate [51]. A double-blind, placebo-controlled study conducted among children with ASD without additional neurologic deficits showed significant improvements in verbal scores in subjects positive for FRAb following treatment with folinic acid [48]. A recent, self-controlled clinical trial was conducted among children with low-functioning autism without additional neurological complications. In these, a high, 76% prevalence of FRα antibodies was found. These children also had multiple nutrient deficiencies attributed to selective eating habits and malnutrition. Combined correction of deficient nutrients and high dose folinic acid administration resulted in an overall significant recovery from severe autism to mild–moderate autism (Figure 3A). Comparison of the Childhood Autism Rating Scale (CARS) after 2 years of treatment (folinic acid supplementation and correction of abnormal nutrient values) with the CARS at baseline showed better outcomes for children having negative or low FRα antibody titers of the blocking type, up to 0.44 pmol FRα blocked/mL serum, versus the group whose FRα antibody titers were above 0.44. The baseline CARS score increased as a function of the age at which treatment was initiated. The outcome became poorer for the older subgroup of treated autistic children (Figure 3B). This outcome may be further compounded by the presence of maternal and paternal autoantibodies and embryonic exposure to these. Preliminary data suggested that in the event of maternal or parental FRα autoantibodies, the child´s outcome after treatment was also less favorable (Figure 3C).

3C).

**Figure 3.** (**A**). Compared to untreated autistic patients (*n* = 84) whose CARS remained unchanged, a group treated with folinic acid and with their nutritional deficits corrected showed a decrease in baseline CARS score from severe (*n* = 82; CARS mean ± SD: 41.34 ± 6.47) to moderate or mild autism (mean ± SD: 34.35 ± 6.25; paired *t*-test *p* < 0.0001). (**B**). As a function of age, the baseline CARS (blue bars) increased slowly with advancing age, while the CARS after a 2-year treatment period (orange bars) diminished significantly for all age subgroups. The increase of baseline CARS with advancing age will adversely influence the final outcome for older age groups, particularly above 6 years. (**C**). This graph represents the outcome of treatment as a function of the particular FRα antibody profile in the child (K), mother (M) and father (F). The presence of maternal FRα antibodies or presence of antibodies in both parents will negatively affect the treatment outcome (adapted from [51]).

Compared to infantile-onset CFD syndrome where FRα antibody testing remained negative in the parents, testing of the parents of children with autism revealed a prevalence of 34% in mothers and 29% in fathers versus 3% in healthy controls [51]. Another study also confirmed an equal prevalence of FRα autoimmunity in children with autism (76%) and even higher autoantibody prevalence in their unaffected siblings (75%), fathers (69%) and mothers (59%) [62]. The appearance of these antibodies may have a familial heritable origin, but the risk of developing ASD is likely influenced by other unknown factors because some siblings positive for these antibodies have been asymptomatic. Two of the suspected determinant factors for the development of autism are the appearance of antibodies at a critical stage of neurodevelopmental processes during the first 18 months of life and fetal exposure to maternal antibodies.

The outcome after folinic acid treatment of autism associated with FRα autoimmunity appears to be influenced by several factors such as the level of FRα antibody titer and age at which treatment was initiated as well as the FRα antibody profile amongst parents. In our studies on the treatment outcome after folinic acid therapy for two years we only included the group of children with infantile-onset autism in whom genetic abnormalities had been excluded because genetic defects might constitute a bias to statistical assessment regarding the influence of FRα autoimmunity (Figure 4).

**Figure 4.** The significance of parental antibody status on developmental outcomes in offspring (top panel) and postnatal antibody development in offspring (bottom panel).

1

#### **14. Treatment of FR Autoimmune Disorder in Pregnancy**

FRα autoimmunity has been associated with a high risk of neural tube defects (NTD) and other congenital malformations in offspring [16,17]. This has been confirmed by other independent studies [63,64]. Even though the incidence of NTD is high in the Irish population, evaluation of FRα autoantibodies in this population has failed to show a statistically significant correlation with NTD pregnancies [65]. However, the study showed a higher prevalence (35–40%) of FR antibodies in the mothers and both male and female controls. Because FRα antibodies were not assessed in the fathers of the children with NTD and only in unrelated males of the control group, the contribution of the possibility of both parents being positive for FRα antibodies to the NTD outcome cannot be ruled out. Parental studies in ASD have shown both maternal and paternal influences on the incidence and severity of ASD outcome [51]. The folic acid fortification of foods has reduced the prevalence of NTD by 30–50 percent. However, for women having a normal folate status but testing positive for FRα autoantibodies, we suspect that even the addition of a daily dose of 400–800 µg or 1000 µg for twin pregnancies may not prevent NTD or congenital malformations due to the persistence of embryonic and fetal folate deficiency in the presence of FRα antibodies.

One case report described a woman who, upon follow-up after three pregnancies, was found to have high titers of serum FRα autoantibodies of the blocking and binding types. She had previously had two miscarriages and a third pregnancy with monozygotic twins, during which she took 1 mg folic acid per day. However, one twin was reduced at 12 weeks because of an encephalocele, and the pregnancy was terminated because the other twin had hypoplastic left-heart syndrome and choroid plexus cysts. At this time, extensive genetic testing did not reveal any abnormalities. After finding FRα autoantibodies, treatment with a milk-free diet was able to reduce FRα antibody titers, but a fourth pregnancy by IVF also resulted in a miscarriage after 5 months. It was only after continuation of the milk-free diet combined with 4 mg folic acid, 2.5 mg leucovorin and 5 mg prednisone that FRα antibody titers fell to undetectable levels, and a fifth pregnancy was carried to term and resulted in the birth of a healthy baby boy [17]. FRα autoantibodies are significantly associated with subfertility and preterm birth [45,46], and preterm babies have a higher prevalence of ASD [66]. Therefore, testing for FRα antibodies in women of child-bearing age may help in preventing some of these disorders by early intervention.

#### **15. Prevention of ASD and Related Complications Due to FR Antibodies**

Prevention of ASD has not been reported yet. However, it seems extremely important to diagnose ASD at the earliest age possible in order to be able to perform further evaluation including testing for serum FRα autoantibodies. We believe that as soon as ASD is strongly suspected and FRα antibodies identified as early as possible prior to the age of three years, the outcome following prompt treatment with high-dose folinic acid in combination with an animal milk-free diet will have a favorable outcome if maternal FRα antibodies or the presence of antibodies in both parents were negative [51].

Currently, FRα antibody testing is only performed after ASD is suspected or has been diagnosed. However, this procedure postpones treatment and causes a significant delay, affecting prognosis unfavorably. Therefore, the early screening of autism at 18 or 24 months using the Modified Checklist for Autism in Toddlers (M-CHAT test) or other instruments can be used by health workers, although there is lack of optimal sensitivity and specificity for ASD when using these tests at an early age.

Another option would be to perform the screening test for FRα antibodies at an early age between 12 and 18 months, particularly among those children suspected to manifest one or more autistic signs or symptoms. The children testing positive for FRα antibodies could be placed on an animal milk-free diet and receive folinic acid supplementation with a mandatory follow-up of these children.

#### **16. Clinical Significance of the Findings**

Since the discovery of folic acid more than a century ago, the hematologic consequences of its deficiency and its role in DNA synthesis and treatment of megaloblastic anemia has been well defined. Emerging research aims to define its role in methylation reactions, epigenetic regulation of gene expression, reproductive function, pregnancy and fetal development. It is becoming evidently clear that folate plays a major role not only in fetal brain development but also in post-natal development and refinement of functional integration of the mature brain. Clinical improvement seen in response to folinic acid treatment in ASD, schizophrenia, depression and dementia attests to the role of folate in metabolic regulation of brain function, potentially by regulating the expression and processing of neurotransmitters. While conventional thinking would associate disruption of folate metabolism with nutritional folate deficiency and gene defects of folate pathways, the identification of folate receptor autoantibodies contributing to fetal and cerebral folate deficiency has thrown a monkey wrench into our current thinking regarding folate transport into the brain and its role in regulating brain function. Therefore, clinical recognition of cerebral folate deficiency is critical to our understanding of neuro–developmental as well as neuro–psychiatric disorders.

To prevent fetal folate deficiency, specific guidelines for treatment of future parents testing positive for FRα antibodies should consider the time frame and dose for folinic acid supplementation prior to conception and for mothers, the folinic acid dose to be used during pregnancy. It will be extremely important to assess optimal dosage in order to provide sufficient folate supplementation but avoid excess dosing, especially since data are lacking on the safety profile of high-dose folinic acid administered throughout pregnancy. Based on the safety profile of high-dose folinic acid used in the treatment of CFD and ASD, one may speculate that a daily dose of 5–10 mg may be in the safe range. The future availability of levofolinate could reduce this dose by half. For favorable outcomes in CFD and ASD, early testing for FRAb and treatment with folinic acid could potentially prevent the development of neurologic deficits.

#### **17. Concluding Remarks**

Decades of research into neural tube defect pregnancies have only managed to reduce their incidence through folic acid supplementation, but not prevent them altogether. ASD incidence, on the other hand, has continued to rise with no definitive contributing cause identified. Both public and private funding agencies have poured a major share of available funds toward identifying gene defects and genomic polymorphisms to no avail. An enormous sum of money has been expended in developing gene deletion mouse and rat models to identify the autism gene(s). It is now clear that ASD is not a congenital genetic disorder and does not follow Mendelian inheritance. Therefore, the answer to the pathogenesis of ASD must lie in epigenetic and environmental factors that broadly affect gene expression. Folate plays a pivotal role in DNA/RNA synthesis, methylation and epigenetic control of gene expression, and therefore, decreased folate availability during critical stages of development, albeit by the presence of FRAb-blocking folate transport as well as triggering inflammation, may play a significant role in the pathology of ASD.

**Author Contributions:** All authors contributed equally to the conceptualization, writing and finalization of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** Ongoing research in the Quadros Laboratory is supported by a grant from the Brain Foundation and by the Vembu Srinivasan family fund. N.B.-K. is supported by a graduate student fellowship from the NY City-funded Institute for Basic Science Research, Staten Island, NY, USA.

**Institutional Review Board Statement:** Not Applicable.

**Informed Consent Statement:** Informed consent was obtained from all subjects reported in this study.

**Data Availability Statement:** Not Applicable.

**Conflicts of Interest:** Two of the authors (J.M.S. and E.V.Q.) are inventors on a US patent for the detection of FRalpha autoantibodies issued to the Research Foundation of SUNY.

#### **References**

