Bioenergetic variation is related to autism symptomatology

76 individuals with ASD who met inclusion and exclusion criteria had ETC complex function measured with 61 undergoing cognitive and behavioral evaluations. Clinical characteristics of the participants are provided in Table 1.

Comorbid conditions were both derived from a parent reported medical questionnaire and from review of conditions diagnosed in the medical records. Regression (defined as loss of already obtained skills) was defined in detail in our questionnaire. Questions regarding regression included the timing, specific skills lost, duration of the regression, trigger and whether there were multiple regressions. Since this paper is not specifically on regression, we have not summarized these details in the table. This method for assessing medical comorbidies has been used in several of our previous studies (Frye et al. 2016b, d; Delhey et al. 2017; Frye et al. 2017b). Since there are numerous medical conditions, we summarized them into categories. Gastrointestinal disorders included constipation, diarrhea, stomach/abdominal pain, gastroesophageal reflux disease, vomiting, and feeding problems. Allergic disorders included allergies, asthma, breathing problems, other lung problems, and allergic skin conditions. Psychiatric disorders included depression, bipolar, anxiety disorder, obsessive compulsive disorder, attention deficit with or without hyperactivity. Immune disorder included ear infections, sinusitis, throat infections and other immune problems. Neurological disorders included hearing problems, headaches, vision problems, microcephaly, macrocephaly, seizures, opthalmoplegia, strabismus, cerebral palsy, tics, and muscle weakness. General health disorders included dental problems, fatigue, lack of sweating, exercise intolerance, heat intolerance, bone and joint problems, and blood problems and anemia. Cardiovascular disorders included heart conditions, syncope, congenital heart disease, and heart failure. Growth disorders included growth hormone deficiency, failure to thrive, short stature, tall stature, overweight and obesity. Endocrine disorders includes thyroid or other endocrine problems. Genitourinary disorders includes kidney, bladder, and genital problems.

Inclusion criteria were (i) age 3 to 14 years of age and (ii) ASD diagnosis. Exclusion criteria were (i) chronic treatment with medications that would detrimentally affect mitochondrial function such as antipsychotic medications; (ii) vitamin or mineral supplementation exceeding the recommended daily allowance, and (iii) prematurity.

The ASD diagnosis was defined by one of the following: (i) a gold-standard diagnostic instrument such as the Autism Diagnostic Observation Schedule and/or Autism Diagnostic Interview-Revised; (ii) the state of Arkansas diagnostic standard, defined as agreement of a physician, psychologist and speech therapist; and/or (iii) Diagnostic Statistical Manual (DSM) diagnosis by a physician along with standardized validated questionnaires and diagnosis confirmation by the Principal Investigator. We have validated that this criteria captures an accurate diagnosis of ASD in our previous studies by re-evaluating a portion of the participants with the Autism Diagnostic Interview-Revised and determining that they all were well within the diagnostic criteria for ASD (Frye et al. 2016b, d; Delhey et al. 2017; Frye et al. 2017b).

Historical controls of similar age and gender included 68 healthy individuals without neurological disease as described in previous studies (Goldenthal et al. 2015). Controls ranged in age from 3 to 21 years of age [mean (SD) 10.1 years (4.6 years)] with 33 (49%) being female. These controls were evaluated with the same methodology used in this study so that the data from the controls were truly comparable to the data collected in the current study. In a previous report, it was found that there was no correlation between enzyme activities and age and no difference in protein activities across ethnicity or race in both controls and mitochondrial disease patients (Goldenthal et al. 2012).

Contemporaneous controls of similar and age and gender included 14 generally healthy children without significant chronic health conditions. These controls ranged in age from 1 to 15 years of age [Mean (SD) 9.1 years (4.8 years)] with 6 (42%) being female. These individuals underwent the Vineland Adaptive Behavior Scale 2nd Edition (VABS) evaluation as well as an evaluation for mitochondrial function so their data could be included in the analysis examining the relationship between the VABS and variation in mitochondrial function in order to investigate whether the patterns found for children with ASD were in line with those found in typically developing individuals.

As mentioned in our previous study (Frye et al. 2016d), our research staff was trained by a multispecialty team consisting of two licensed psychologists and a speech therapist prior to performing assessments. During the study a research psychologist supervised research staff and provided feedback and retraining if necessary.

Observer rated measures included language and the Ohio Autism Clinical Impression Scale (OACIS). Parents completed the Aberrant Behavior Checklist (ABC), Social Responsiveness Scale (SRS) the VABS Survey Interview Form and Childhood Behavioral Checklist (CBCL).

Language was assessed by an ability-appropriate instrument, the CELF-preschool-2, the CELF-4 or the Preschool Language Scale-5 (PLS-5). The CELF is a standardized, well-validated instrument that has been used in studies on ASD (Verly et al. 2014; Edgar et al. 2015). The PLS-5 is a standardized, well-validated instrument that measures subtle changes in verbal communication, particularly in pre-verbal children (Volden et al. 2011). The standardized summary score of each instrument (mean 100, standard deviation 15) was used as the outcome measure.

The OACIS is an observer-rated scale sensitive to clinically meaningful changes in ASD symptoms (Butter and Mulick 2006) which has been shown to have good inter-rater and cross-cultural reliability (Choque Olsson and Bolte 2014) and has been used in several ASD clinical trials (Frye et al. 2015). The VABS is a reliable and valid measure of the ability to perform age-appropriate everyday skills, including communication, daily living, social and motor skills, through a 20–30 min structured interview with a caretaker (Frye et al. 2013a). The ABC is a 58-item questionnaire (Frye et al. 2013a) that measures disruptive behaviors and has convergent and divergent validity (Kaat et al. 2014). The SRS is a 65-item questionnaire that measures the severity of social skill deficits across five domains (Constantino 2002) which has been shown to have good correspondence to the gold-standard instrument (Murray et al. 2011). The CBCL measures general behavioral problems in children and has been described as robust for use in measuring behavior in children with ASD (Hanratty et al. 2015).

The buccal cells were collected using Catch-All Buccal Collection Swabs (Epicentre Biotechnologies, Madison, WI). Four swabs were collected by firmly pressing a swab against the inner cheek while twirling for 30s. Swabs were clipped and placed in 1.5 ml microcentrifuge tubes that were labeled and placed on dry ice for overnight transportation to the Goldenthal laboratory.

Buccal extracts were prepared using an ice-cold buffered solution (Buffer A, ABCAM) containing protease inhibitor cocktail and membrane solubilizing non-ionic detergent and cleared of insoluble cellular material by high speed centrifugation at 4 °C. Duplicate aliquots of the protein extract were analyzed for protein concentration using the bicinchoninic acid method (Pierce Biotechnology, Rockford, IL). Samples were typically stored at −80 °C for up to 1 week prior to enzymatic analysis.

Dipstick immunocapture assays measured ETC Complex I activity using 50 μg of extracted protein (Ezugha et al. 2010; Goldenthal et al. 2012; Yorns et al. 2012). Signals were quantified using a Hamamatsu immunochromato reader (MS 1000 Dipstick reader). Raw milliAbsorbance results were corrected for protein concentration and data were expressed as percentages of the values obtained with control extracts run on the same assay. ETC Complex IV and CS activity were assessed using standard spectrophotometric procedures in 0.5 ml reaction volume. Specific activities of respiratory complexes were initially expressed as nanomoles/min/mg protein. For comparison to controls, Complex I and IV activities were normalized to CS activity.

Analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC). Graphs were produced using Excel version 14.0 (Microsoft Corp, Redmond, WA). This analysis is modeled on our recently analysis of bioenergetic variation in a genetic syndrome (Frye et al. 2016a).

Normal control values were based upon the established historical controls from the Goldenthal laboratory published previously. The two-tailed 95% normative range was considered within 1.96 +/− standard deviation (SD) of the mean, meaning that abnormal values were above 97.5% or below 2.5% of the normal distribution as defined by the historical control mean and standard deviation for each ETC complex activity measured. For comparison to normal historical controls, ETC complex activity was normalized to CS activity.

Abnormalities in ETC complex activity were analyzed in several ways. First, the number of values outside of the normal range was calculated for each ETC complex measured. The binomial probability was calculated to determine if the number of values outside the normal range were significantly different than expected by chance. Second, the ASD group mean complex activity values were compared to the entire control population using t-tests. Third, a difference in the group variance was compared using F-tests.

To better understand the relationship between cognitive and behavioral indexes and mitochondrial enzyme activity we used a general linear model with the cognitive and behavioral indexes as the dependent measure and mitochondrial enzyme activity as predictor variables. ETC complex activity was not normalized in this analysis so the effect of CS could be differentiated from the effect of individual complex activity. Since there is data suggesting that mitochondrial enzyme activity both abnormal and below normal is related to ASD, a second order polynomial is used to model the predictor variables in order to account for a potential curvilinear relationship. The generalized linear model module ‘genmod’ in SAS was used with an alpha of 0.05 with a backward stepwise elimination of non-significant variables. Since adaptive behavior is known to affect social (i.e., SRS (Hus et al. 2013)) and behavior (e.g. ABC (Rattaz et al. 2015)) symptoms in ASD, the composite score of the VABS was used as a covariate for the SRS, ABC, CBCL and OASIS outcome measures.

Table 2 outlines the results of the statistical analyses. Overall 62% of ASD individuals showed mitochondrial enzyme activity outside the control range (Fig. 1). Complex I activity was outside of the control range for 39% of ASD participants, which is significantly more than expected by chance alone. Variation in Complex I activity was significantly greater in ASD participants as compared to the control group and mean Complex I activity for ASD participants was significantly lower than the control group. Complex IV activity was outside the normal range for 11% of the participants, which is significantly more than expected by chance alone. Mean Complex IV activity was significantly lower in ASD participants as compared to the control group, and variation was significantly greater in the ASD group as compared to the control group. The ratio of Complex I to Complex IV was abnormal in 24% of the ASD participants. The average ratio was not significantly different than controls but the variation was greater in the ASD group as compared to the control group.

Table 3 outlines the overlap of complex function abnormalities within the 47 ASD participants with atypical mitochondrial function. Overall, most, 57% (27/47), of the ASD participants with mitochondrial enzyme abnormalities had only one enzyme affected, while 32% (15/47) had two enzymes affects and 7% (3/47) had all three enzymes affected. Complex I and Citrate Synthase (CS) were isolated a majority of the time while Complex IV abnormalities were commonly associated with either a Complex I or CS abnormality. Table 4 provides the details of the overlap of the underactivity and overactivity of specific mitochondrial enzymes. Although some interesting patterns emerge, the number of samples limit conclusions that can be made from these patterns.

Mitochondrial enzymes demonstrated moderate correlations with one another (Fig. 2). CS activity was positively correlated with activity of Complex I (r = 0.47, p = 0.0001) and Complex IV (r = 0.65, p < 0.0001) and Complex I activity was positively correlated with activity of Complex IV (r = 0.50, p < 0.0001).

The VABS Communication, Daily Living, Social and Motor subscales and the overall Adaptive Behavior Composite were all related to Complex IV activity in a non-linear manner (Fig. 3). Communication [Linear χ² (1) = 11.58, p < 0.001; Quadratic χ² (1) = 10.58, p = 0.001; Fig. 3a], Daily Living [Linear χ² (1) = 11.30, p < 0.001; Quadratic χ² (1) = 10.35, p = 0.001; Fig. 3b], Social skills [Linear χ² (1) = 12.22, p < 0.001; Quadratic χ² (1) = 10.60, p = 0.001; Fig. 3c] and Motor skills [Linear χ² (1) = 9.11, p < 0.01; Quadratic χ² (1) = 6.53,p = 0/01; Fig. 3d] were related to Complex IV activity such either relatively higher or lower Complex IV activity was related to poorer skills. Complex IV activity also demonstrated a similar non-linear relationship with the Adaptive Behavior Composite [Linear χ² (1) = 12.73, p < 0.0005; Quadratic χ² (1) = 11.35, p < 0.001].

Complex I was also concurrently related to VABS in a linear manner (Fig. 4) for the Adaptive Behavior Composite [Linear χ² (1) = 5.00, p < 0.05] and VABS subscales except motor skills. As see in Fig. 4, better Communication [Linear χ² (1) = 5.72, p < 0.05], Daily Living [Linear χ² (1) = 3.71, p = 0.05], and Social skills [Linear χ² (1) = 10.60, p = 0.001] were linearly related to higher Complex I activity.

SRS total and subscales were related to CS activity in a non-linear manner such that relatively worse social function was associated with higher CS activity (Fig. 5) when level of adaptive function was taken into account [Awareness Linear χ²(1) = 10.60, p = 0.001, Quadratic χ²(1) = 6.97, p < 0.01, Adaptive Behavior χ²(1) = 18.84, p < 0.0001; Cognition Linear χ²(1) = 4.15, p < 0.05, Quadratic χ²(1) = 6.02, p = 0.01, Adaptive Behavior χ²(1) = 32.28, p < 0.0001; Communication Linear χ²(1) = 6.21, p = 0.01, Quadratic χ²(1) = 6.10, p = 0.01, Adaptive Behavior χ²(1) = 30.75, p < 0.0001; Motivation Linear χ²(1) = 4.75, p < 0.05, Quadratic χ²(1) = 3.59, p = 0.05, Adaptive Behavior χ²(1) = 20.14, p < 0.0001; Mannerisms Linear χ²(1) = 7.41, p < 0.01, Quadratic χ²(1) = 12.18, p = 0.0005, Adaptive Behavior χ²(1) = 25.91, p < 0.0001; Total Linear χ²(1) = 8.74, p < 0.005, Quadratic χ²(1) = 10.27, p = 0.001, Adaptive Behavior χ²(1) = 34.93, p < 0.0001].

Mitochondrial enzyme activity was not related to the other behavioral or language assessments.