NRF2, KEAP1 and GSK-3 levels in autism spectrum disorder: a case control study

Abstract

Recent studies show that oxidative stress has an important role in the etiology of autism. In our study, Nrf2, which is the main regulator of cellular antioxidant response, and Keap1 and Gsk-3β, which are the main proteins that regulate this pathway, were compared between children with autism and healthy controls. To the best of our knowledge, our study is the first in which Nrf2, Keap1 and Gsk-3β levels were evaluated together in children with ASD. In our study, Nrf2 level was found to be lower and Keap1 level higher in children with autism. Although GSK-3β is increased in many psychiatric and neurodegenerative diseases, it was found to be low in the autistic group in accordance with the literature. In conclusion, considering the versatile modulation of the Nrf2 pathway, this article does not provide any mechanistic insight into the pathway, but it suggests that Nrf2, Keap1 and Gsk-3β, which have central roles in oxidative stress, may play a role in the pathophysiology of autism.

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental disorder that first appears in early childhood, restricts daily functioning, and is characterized by an ongoing deterioration in social interaction, limited and stereotyped interests and behaviors (Association, 2013). Although until the 1980s autism was defined as a rare disorder of 4 out of 10,000, many epidemiological studies now showed that its prevalence was increasing. The United States Center for Disease Control (CDC) reported the prevalence of autism as 1/54 in its latest report published in 2020 (Maenner et al., 2020).

Despite its increasing prevalence, limited understanding of the pathophysiology of ASD prevents the creation of effective treatments for core symptoms (Smile and Anagnostou, 2015). In studies to determine the etiology of ASD, it is stated that there is a multifactorial etiopathogenesis scheme in which genetic, environmental and immunological factors take place together. Oxidative stress has an important role in etiology as a mechanism linking genetic, environmental and immunological factors (Chauhan and Chauhan, 2006). Increasing evidence in the literature suggests that oxidative stress plays a role in the development and clinical manifestation of autism (Mcginnis, 2004, Chauhan et al., 2004).

Nrf2 is a transcription factor that regulates the coordinated activation of a number of antioxidant genes against oxidative stress, and the Keap1 (Kelch-like erythroid cell-derived protein with CNC homology [ECH]-associated protein 1) – Nrf2 (The nuclear factor erythroid 2-related factor 2 – ARE (Antioxidant Response Element) axis serves as a “master regulator” in response to oxidative/electrophilic stresses and chemical attack through the induction of a large cytoprotective gene pool (Guo et al., 2015). Nrf2 activity is tightly regulated by two main inhibitors, Keap1 and Gsk-3β (Glycogen synthase kinase-3β) (Cuadrado, 2015).

Under homeostatic conditions, Nrf2 is retained in the cytosol, inhibited from translocation to the nucleus, and degraded by ubiquitination. Of the two mechanisms leading to this process, the first is mediated by Keap1 and the second is mediated by GSK3β. The former is involved in major oxidant/electrophile aggressions, while the latter is involved in minor fluctuations of Nrf2 levels as a function of punctual demands (Cuadrado, 2015).

Under basal conditions, the Keap1 homodimer interacts directly with Cullin3 (Cul3) and forms the Keap1-Cul3-RBX1 (Ring box protein-1) E3 ligase complex; this complex induces polyubiquitination of Nrf2 and protein degradation by the proteasome (Kobayashi et al., 2004). Upon exposure of cells to oxidative stress, reactive oxygen species react with cysteine residues in Keap1, causing its structure to change and inactivation. Likewise, cellular factors such as p62, p21, DJ-1, PALB2, and BRCA1 may disrupt the binding between Nrf2 and Keap1 via direct protein-protein interactions. Among them, oxidation of the cysteine residues of Keap1 and the binding of p62 to Keap1 have received much attention. p62 competes with Nrf2 for binding to Keap1, and binding of p62 to Keap1 leads to the degradation of Keap1 and the consequent Nrf2 stabilization (Li et al., 2019) Nrf2 is released from the Keap1-Cul3-RBX1 complex and translocates into the nucleus (Jeong et al., 2006, Pandey et al., 2017). Nuclear Nrf2 then binds to ARE and induce the downstream of the regulation of gene expressions, such as HO-1, glutathione peroxidase and NAD(P)H, quinone oxidoreductase 1(NQOl) and other antioxidant and detoxification enzymes were activated, thus playing a role in anti-oxidative stress, anti-inflammation, anti-apoptosis, and additional cellular protection (Li et al., 2020). After induction of Nrf2-dependent genes, their transport back to the cytosol and termination of Nrf2 signaling are also mediated by both Keap1 and Gsk-3β. Keap1 migrates to the nucleus to dissociate Nrf2 from the ARE, forming the Nrf2-Keap1 complex in the nucleus and then transporting Nrf2 from the nucleus to the cytosol, while GSK3β terminates Nrf2 signaling via a molecular cascade involving phosphorylation of Nrf2 (Shaw and Chattopadhyay, 2020).

GSK3β, which is involved in various cellular processes such as neurogenesis, synaptic plasticity, and apoptosis, is kept in an inactive state under normal conditions due to its inhibition by AKT-mediated phosphorylation. However, Gsk-3β phosphorylates Nrf2 when AKT is inactive. This phosphorylation activates β-TrCP (β-transducing repeat containing protein) and initiates β-TrCP-mediated proteasomal degradation of Nrf2. Generally, Keap1-mediated proteasomal degradation of Nrf2 occurs in the cytosol, while Gsk-3β-mediated proteasomal degradation of Nrf2 occurs in the nucleus (Zang et al., 2020).

Nrf2 levels have been shown to vary significantly in various diseases, including cancer, neurodegenerative disease, cardiovascular disease, diabetes, autoimmune disease and psychiatric diseases (Copple, 2012), but its relationship with ASD has not been adequately studied. An important part of our knowledge on the relationship between Nrf2 and OSB comes from studies using Nrf2 activator sulforan (SFN) and other Nrf2 activators. Recent studies have shown that SFN found in vegetables such as broccoli and cauliflower activates Nrf2 and has a significant positive effect on individuals with ASD (Jeong et al., 2006). It has been shown that SFN plays a role in the development of behavior and social response in children with ASD, and there is a significant positive correlation between SFN use and ASD behavior and cognitive function (Singh et al., 2014, Lynch et al., 2017, Mcguinness and Kim, 2020).

The connections of Keap1 and Gsk-3β, which are important components of the Nrf2 signaling pathway, with OSB are not clear enough. In this study, it was aimed to evaluate the role of Nrf2 and its modulators, which play a central role in oxidative stress, in autism.

Methods

The research sample was carried out in Gaziantep University Şahinbey Research and Application Hospital Child and Adolescent Psychiatry Outpatient Clinic. All procedures in this study were performed in accordance with the ethical standards of the institutional research committee and the principles of the Declaration of Helsinki, and the study was approved by the Gaziantep University Ethics Committee. (Date/print: 2020/73). Patients aged 3–12 years who were diagnosed with ASD according to DSM-5 criteria were included in the study as the case group, and children aged 3–12 years without any psychiatric or medical disease were included in the study as the healthy control group. In both the case and control groups, individuals with a history of alcohol or cigarette use, individuals with anti-inflammatory and antioxidant use in the last 2 weeks, any known acute or chronic metabolic disease, genetic disease, neurological disease, autoimmune disease, allergic disease or other individuals with a medical illness were not included.

Conducting the research and method

Both patient and control group children were evaluated by a child psychiatrist, and the sociodemographic data form was filled by the clinician. The diagnosis of ASD was made through a clinical interview based on the DSM-5. The severity of ASD symptoms was evaluated according to Childhood Autism Rating Scale (CARS) scores. CARS is a valid and reliable 15-item scale widely used in the diagnosis of autism. Its validity and reliability in our country was determined by İncekaş Gassaloğlu and colleagues (İncekaş Gassaloğlu et al., 2016).

Collection of blood samples and laboratory study

Blood collection was done from the volunteers in Gaziantep University Medical Faculty Hospital blood collection unit. The blood samples were taken into a yellow-capped vacuumed plastic gel tube and centrifuged at 4000 rpm for 10 min, then the serum was separated and stored at −80°Celsius until the analysis period. After the samples were completed, Nrf2, Keap 1, GSK3ß were studied by the ELISA method with ready-to-use kits. The ELISA reagents were allowed to come to room temperature before the samples were run.

Nrf2 detection

40 µL of sample, 10 µL of anti-NRF2 antibody and 50 µL of streptavidin-HRP were added to the sample wells. 50 µL of standard (40, 20, 10, 5, 2.5 µL) and 50 µL of streptomycin were added to the standards. Secondary antibody with biotin and enzyme reagent were added to the sample and standard wells. It was then incubated at 37 °C for 60 min. ELISA plates were washed five times and chromogen A and chromogen B solutions were added, respectively, and incubated at 37 °C for 10 min in the dark. The stopping solution was then added. Only chromogen A, chromogen B and stop solution were added to the blank wells. Reading was done at 450 nm with an ELISA reader (Biotek Instruments, USA). Results were expressed as pg/mL.

Determination of Keap-1

After all reagents (Shangai Coon Koon Biotech Co.,Ltd) , standards and samples were prepared, 50 µL of standards were added to the standard well of the Micro ELIZA strip plate. Next, 10 µL of sample followed by 40 µL of diluent was added to a sample well. 100 µL of HRP-conjugate reagent was added to each well, then sealed with the sealing plate membrane and after gentle shaking incubated for 60 min at 37 °C. The plate was automatically washed five times by injecting 350 µL of washing solution into each well of the plate after the sealed plate membrane removed. For color formation, 50 µL of chromogen solution A was first added to each well, then 50 µL of chromogen solution B was added to each well. It was then incubated for 15 min at 37 °C away from light. After incubation, 50 µL of stop solution was added to each well to stop the reaction. The absorbance (OD) of each well within 15 min was measured with the ELISA Biotek ELx800, USA, instrument at a wavelength of 450 nm. Serum Keap-1 results were expressed as pg/mL.

Determination of Gsk-3β

50 µL of standard solutions (16, 8, 4, 2, and 1 ng/mL) and 50 µL of streptomycin-HRP were added to the wells. 40 µL of sample and 10 µL of anti-GSK3ß antibody and 50 µL of streptavidin-HRP were added to the sample wells. It was incubated for 60 min at 37 °C after gentle shaking. It was washed 5 times with washing solution. First, 50 µL of chromogen A solution and then chromogen B solution were added sequentially. After shaking it carefully, it was incubated for another 15 min at 37 °C in the dark. Finally, 50 µL of stop solution was added to each well. Reading was done at 450 nm with an ELISA reader (Biotek Instruments, USA). Results were expressed as ng/mL.

Statistical analysis

Descriptive statistics of the data obtained from the study are given with mean, standard deviation for numerical variables, and frequency and percentage analysis for categorical variables. The normal distribution test of Nrf2, Keap1 and Gsk-3β variables was analyzed using the Shapiro Wilk test. It was determined that the variables other than the Keap1 variable did not conform to the normal distribution (P < 0.05). In the comparison of these variables according to categorical variables, independent samples t-test/Mann–Whitney U test was used for categorical variables containing two groups, and Analysis of Variance/Kruskal Wallis test was used for categorical variables containing three or more groups. Analysis results are given as mean ± standard deviation for the variable conforming to the normal distribution, and as for the variables not conforming to the normal distribution are given as median (Q1–Q3). In addition, the differences between categorical variables were tested with χ² analysis. Also the relationships between numerical variables were examined by Spearman correlation analysis. Analyzes were carried out with the help of SPSS 22.0 program. P < 0.05 significance level was chosen.

Results

The sample group consisted of 40 children with ASD and 40 healthy controls. The mean age of the participants was 7.55 ± 2.48 (min: 3, max: 12). There was no statistically significant difference between the case and control groups in terms of age, gender, time of delivery, type of delivery and postnatal incubator care (P > 0.05). There was a statistically significant difference between the two groups in terms of the presence of psychiatric disease and autism in the first and second degree relatives of the participants, but there was no statistically significant difference when compared with the presence of Attention Deficit Hyperactivity Disorder (ADHD) (P > 0.05). When the parents in the case and control groups were evaluated in terms of chronic medical disease and psychiatric disease, no statistically significant difference was found (P > 0.05). The results are detailed in Table 1.

Table 1.

Socio-demographic information of the participants.

Variables	Control (n = 40 %)	Case (n = 40, %)	P
Age	7.73 ± 2.48	7.38 ± 2.5	0.531
Female	10 (25)	6 (15)	0.264
Male	30 (75)	34 (85)
Maternal Age	34.53 ± 6.45	33.15 ± 6.24	0.336
Paternal Age	38.75 ± 6.03	37.35 ± 5.83	0.295
Psychiatric disease in the family	25 (%62.50)	15 (%37.50)
None	15 (%37.50)	25 (%62.50)
Present			0.025
ASD in the family
None	39 (%97.50)	32 (%80.00)
Present	1 (%2.50)	8 (%20.00)	0.013
ADHD in the family
None	31 (%77.50)	26 (%65.50)
Present	9 (%22.50)	14 (%35.00)	0.217
Medical illness in mother
None	32 (%80.00)	32 (%80.00)
Present	8 (% 20.00)	8 (% 20.00)	1.000
Psychiatric illness in mother
None	32 (%80.00)	31 (%77.50)
Present	8 (% 20.00)	9 (%22.50)	0.785
Medical illness in father
None	32 (%80.00)	32 (%80.00)
Present	8 (% 20.00)	8 (% 20.00)	1.000
Psychiatric illness in father
None	36 (%90.00)	34 (%85.00)
Present	4 (%10.00)	6 (%15.00)	0.499

Abbreviations: ASD: autism spectrum disorder, ADHD: attention-deficit hyperactivity disorder.

According to the CARS scores filled by the clinician, it was seen that 18 (45%) patients met mild–moderate autism scores and 22 (55%) patients met severe autism scores. While 65% of the cases were accompanied by mental retardation, 35% had normal mental capacity. When it was evaluated in terms of accompanying ADHD, 82.50% of the cases had ADHD comorbidity, while only 17.50% did not have ADHD. While 31 (77.50%) of the patients received special education, 9 (22.50%) of them did not receive special education. Considering the duration of special education, it was seen that the case group received special education for an average of 4.19 ± 2.61 years. There was no significant difference in the comparison between gender and the degree of autism in the case group.

Comparison of oxidative stress parameters is reported in Table 2. A statistically significant difference was found between the Nrf2, Keap1 and Gsk-3β values between the case and control groups (P < 0.05). Accordingly, while Keap1 values were higher in the case group, Nrf2 and Gsk-3β values were found to be lower. A statistically significant positive correlation was found between Nrf2 and Gsk-3β in the case group (P < 0.05).

Table 2.

Comparison of serum Nrf2, Keap1, Gsk-3β levels between cases and controls.

	Control	Case	P
Nrf2	493.06 (463.68-535.86)	464.74 (423.25–496.60)	0.009*
Keap1	842.43 ± 171.65	1013.65 ± 179.38	0.001*
Gsk-3β	7.87 (7.42–8.21)	6.83 (6.15–7.53)	0.001*

Serum Nrf2 and Keap-1 concentrations are given in pg/mL, and Gsk-3β concentration in ng/mL. Abbreviations: Nrf2: Nuclear Factor E2-Related Factor 2, Keap1: Kelch-likeerythroidcell-derived protein with CNC homology [ECH]-associated protein 1, Gsk-3β: Glycogen synthase kinase-3ß.

Considering the relationship between age and Nrf2, Keap1, Gsk-3β levels, a direct proportional, low and statistically significant relationship was found between age and Gsk-3β levels (P < 0.05). Considering the relationship between gender and Nrf2, Keap1, Gsk-3β levels, Keap1 values show a statistically significant difference according to gender (P < 0.05). Accordingly, men have higher Keap1 values. Nrf2 and Gsk-3β values do not differ statistically significantly according to gender (P > 0.05) (see Table 3).

Table 3.

Relationship between age and gender and Nrf2, Keap1, Gsk-3β levels.

Variables		Nrf2	Keap1	Gsk-3β
Age	P	0.176	0.088	0.023*
	Female	478.55 (454.11–534.79)	834.07 ± 184.32	7.10 (6.15–8.09)
Gender	Male	484.92 (432.84–518.89)	951.53 ± 191.24	7.48 (6.69–7.96)
	P	0.674	0.030*	0.661

Serum Nrf2 and Keap-1 concentrations are given in pg/mL, and Gsk-3β concentration in ng/mL.

When patients with mild–moderate autism and severe autism were compared in terms of Nrf2, Keap1, Gsk-3β levels, no statistically significant difference was found (P > 0.05). When the Nrf2, Keap1 and Gsk-3β levels of the cases were compared with whether they received special education or not, a statistically significant difference was found between special education and Nrf2 (P < 0.05). Nrf2 levels were found to be higher in the group receiving special education. When the relationship between drug use and Nrf2, Keap1, Gsk-3β levels in children with autism was examined, it was found that there was a statistically significant difference between Keap1 and drug use, and Keap1 levels were higher in the drug use group. No statistically significant difference was found between family history of psychiatric disease or ADHD and Nrf2, Keap1, Gsk-3β levels (P > 0.05). When family history of ASD and Nrf2, Keap1, Gsk-3β levels were compared, a statistically significant difference was found between family history of ASD and Keap1 (P > 0.05), and Keap1 level was found to be higher in those with a family history of ASD (see Table 4).

Table 4.

The relationship between autism severity, special education, drug use, family history of ASD and Nrf2, Keap1, Gsk-3β levels.

Variables		Nrf2	Keap1	Gsk-3β
Autism Severity	Mild/ moderately	457.30 (427.51–493.42)	1008.61 ± 207.84	6.65 (3.99–7.12)
	Extremely autistic	475.36 (418.99–497.66)	1017.77 ± 157.30	7.06 (6.42–7.75)
	P	0.946	0.875	0.062
Special Education	No	418.99 (387.01–431.77)	972.12 ± 206.88	6.61 (3.99–6.87)
	Yes	476.42 (444.54–497.66)	1025.70 ± 172.48	6.93 (6.20–7.63)
	P	0.028*	0.437	0.157
Use of Psychotropic Agents	No	442.41 (399.81–506.15)	925.02 ± 164.11	6.62 (3.99–7.37)
	Yes	470.05 (429.64–491.29)	1061.37 ± 171.53	6.90 (6.42–7.63)
	P	0.900	0.020*	0.180
ASD in the family	No	484.92 (438.16–521.01)	908.90 ± 191.72	7.49 (6.63–7.99)
	Yes	487.05 (444.54–516.77)	1079.03 ± 152.55	6.87 (6.09–7.81)
	P	0.867	0.013*	0.337

Serum Nrf2 and Keap-1 concentrations are given in pg/mL, and Gsk-3β concentration in ng/mL.

Table 5 provides information on the regression analysis of oxidative stress parameters. In linear regression analysis, we found that age, gender, family history of psychiatric disease, maternal age and paternal age did not have a statistically significant effect on Nrf2 values. In the linear regression analysis performed between Keap1 and age, gender, family psychiatric disease, maternal age and paternal age, it was determined that age, gender and psychiatric disease in the family had a significant effect on the change in Keap1 values (B: 17,783 P: 0.048 for age; B: 131,537, P: 0.019 for gender; B: 84,862, P: 0.043 for family history of psychiatric illness). Other variables did not have a statistically significant effect on Keap1 values. In addition, 16.4% of the change in Keap1 value is explained by the variables of age, gender and psychiatric disease in the family. In linear regression analysis, age, gender, family history of psychiatric disease, maternal age and paternal age did not have a statistically significant effect on Gsk-3β values.

Table 5.

Linear regression analysis between Nrf2 and age, gender, family psychiatric disease, maternal age, and father’s age.

Model		Unstandardized Coefficients		Standardized Coefficients	T	Sig.
		B	Std. Error	Beta
Nrf2	(Constant)	512.039	61.800		8.285	0.000
	Age	4.992	2.928	0.204	1.705	0.092
	Sex	−20.734	18.106	−0.138	−1.145	0.256
	Psychiatric disease in the family	−0.707	13.644	−0.006	−0.052	0.959
	Maternal age	1.718	2.211	0.180	0.777	0.440
	Paternal age	−2.404	2.429	−0.236	−0.990	0.325
Keap1	(Constant)	528.452	186.676		2.831	0.006
	Age	17.783	8.844	0.227	2.011	0.048*
	Sex	131.537	54.691	0.272	2.405	0.019*
	Psychiatric disease in the family	84.862	41.214	0.219	2.059	0.043*
	Maternal age	−6.275	6.679	−0.205	−0.940	0.350
	Paternal age	5.216	7.336	0.159	0.711	0.479
Gsk-3β	(Constant)	5.991	1.412		4.243	0.000
	Age	0.109	0.067	0.196	1.632	0.107
	Sex	0.141	0.414	0.041	0.340	0.735
	Psychiatric disease in the family	−0.233	0.312	−0.085	−0.746	0.458
	Maternal age	0.024	0.051	0.110	0.477	0.635
	Paternal age	−0.017	0.055	−0.072	−0.302	0.764

Age, sex, family psychiatric disease, maternal age, paternal age, time of birth, and Nrf2, Keap1 and Gsk3ß were included in the logistic regression analysis for predicting autism and being a healthy control. Among the evaluated parameters, a significant correlation was found between Keap 1 and Gsk-3β and autism (OR[95% CI]: 1.01 [1.01–1.02] p: 0.001 for Keap 1, OR[95% CI]: 0.05 [0.01–0.49] p: 0.010 for Gsk-3β) (see Table 6).

Table 6.

Logistic regression analysis of age, gender, family psychiatric disease, maternal age, father’s age, time of birth, and Nrf2, Keap1, Gsk-3β levels.

Variables	OR[95% CI]	P
Age	1 [0.62–1.61 ]	0.994
Sex (Reference = female)	1.56 [0.08–32.34 ]	0.772
Family psychiatric disease (Reference = present)	0.18 [0.02–1.46 ]	0.109
Maternal age	1.11 [0.83–1.48 ]	0.478
Paternal age	0.86 [0.63–1.17 ]	0.332
Time of birth (Reference = Postterm)	2.68 [0.11–67.49 ]	0.549
Nrf2	0.98 [0.96–1.01 ]	0.179
Keap1	1.01 [1.01–1.02 ]	0.001*
Gsk-3β	0.05 [0.01–0.49 ]	0.010*

Discussion

The most important findings of our study are the statistically significant differences between the case and control groups in terms of Nrf2, Keap1 and Gsk-3β values. Keap1 values were higher and Nrf2 and Gsk-3β values were lower in the case group.

Looking at the studies investigating the role of Nrf2, in the etiology of autism, one of the most important mechanisms regulating oxidative stress, Furnari et al. stated that they observed that Nrf2 knockout mice had less social play contact with other mice (Furnari et al., 2015). In a study in dermal fibroblasts of 12 children with ASD and healthy controls, impaired antioxidant response mediated by a defective Nrf2 pathway in ASD fibroblasts was demonstrated. In response to oxidative damage, ASB fibroblasts were found not to show a Nrf2 nuclear translocation similar to that observed in control cells. This study, which is one of the rare studies investigating Nrf in ASD, showed a decrease in Nrf2 levels and a lower oxidative phosphorylation capacity in autism, consistent with our findings (Stamova et al., 2013). A recent study revealed that monocytes of ASD subjects had decreased Nrf2 expression/activity with increased inflammation, similar to our findings (Nadeem et al., 2020). In contrast, another study showed a slight (<2 fold) but significant increase in Nrf2 mRNA levels in peripheral blood mononuclear cells isolated from ASD subjects. In this study, although oxidative stress was confirmed by increased Nrf2 mRNA expression in blood samples from children with ASD, no major difference in erythrocyte plasma membrane oxidative modifications could be detected. The results of the study were associated with the idea that despite the increase in mRNA level, there may be a disruption in translocation of Nrf2 to the nucleus (Bolotta et al., 2018). Although there are few studies to the contrary, the low level of Nrf2 in autism, which has a central role in protecting the organism from oxidative stress, is an expected situation as supported by our research.

Keap1, which has an important place in the regulation of Nrf2 activity, has recently been investigated in many neurological and psychiatric diseases. Kerr et al provided the first in vivo evidence that Keap1 inhibition can exert neuroprotective effects by ameliorating deficiencies in Nrf2 activity (Kerr et al., 2017).

The first clinical study investigating Nrf2 and Keap1 levels in ADHD was conducted in Turkey by Ayaydın et al. In this study, Keap1 levels were found to be significantly higher in 41 children with ADHD and in 34 control group children, but no significant difference was observed between the groups in terms of Nrf2 in the study (Ayaydin et al., 2020b). The results of a study conducted by Lukic et al. with 30 patients with major depressive disorder and 35 healthy controls revealed that both Nrf2 and Keap1 levels were elevated in the cytoplasm of peripheral blood mononuclear cells of patients with major depressive disorder. It has been noted that although both Nrf2 and Keap1 are upregulated in the PBMC cytoplasm, the negative role of Keap1 may be impaired under prooxidative conditions. It has been shown that oxidized Keap1 is unable to bind Nrf2. In the study, high Nrf2was associated with the inability of Keap1 to bind Nrf2 and decreased degradation of Nrf2 (Lukic et al., 2014). A study using postmortem brain samples showed decreased expressions of Keap1 and Nrf2 in the parietal cortex of patients with major depressive disorder and bipolar disorder compared to the control group (Zhang et al., 2018). In a study investigating peripheral Keap1 and Nrf2 levels in ASD, Keap1 and Nrf2 levels were found to be significantly higher. According to total CARS scores, no correlation was found between ASD severity and Keap1 and Nrf2 (Ayaydin et al., 2020a). This study by Ayaydın et al. (2020a, b) is the first clinical study investigating peripheral Keap1 and Nrf2 levels. In the study in which serum heme oxygenase-1 (HO-1), Keap1 and Nrf2 levels were measured in 43 patients with autism and 41 healthy controls, HO-1 levels were found to be significantly lower in patients compared to controls, while Keap1 and Nrf2 levels were found to be significantly higher. The fact that Keap1 levels were higher in the autistic group in our study supports the results of the study of Ayaydın et al., the only study in which peripheral Keap1 levels were measured in children with autism. Our study suggests that the high Keap1 levels we obtained in the autistic group may be responsible for the decreased Nrf2 levels and impaired oxidative stress. However, unlike the aforementioned study, Nrf 2 level was found to be low in our study. It is thought that our research makes an additional contribution to the literature in terms of supporting other studies in the literature that reveal the impaired activity of Nrf2 in autism, which cannot support an appropriate response to oxidative stress. In addition, considering that Nrf2 activity is regulated by two main inhibitors, Keap1 and Gsk-3β, our study is thought to be important in terms of autism oxidative stress relationship, as it is the first study in which Nrf2, Keap1 and Gsk-3β were evaluated together, unlike the study by Ayaydın et al.

Another important modulator involved in the regulation of Nrf2 activity is Gsk-3β, and recent studies suggest that changes in Gsk-3β activity may be an important aspect of the pathophysiology of psychiatric disorders, including depression, ASD, schizophrenia, mental retardation, and bipolar disorders (Kim and Snider, 2011, Pandey et al., 2010, Karege et al., 2007). The phospho-Gsk-3β/Gsk-3β ratio in platelets was found to be decreased in patients with depressive symptoms, and this was thought to indicate higher activity of Gsk-3β (Pláteník et al., 2014). In a postmortem study with suicidal and non-suicidal subjects, Gsk-3β activity was found to be significantly increased in depressed suicidal and non-suicidal subjects, but not in non-depressed suicidal individuals (Karege et al., 2007). In another study, it is stated that overexpression of Gsk-3β in the brain causes hyperactivity and mania (Prickaerts et al., 2006). In a study including 23 healthy control subjects, 9 bipolar subjects treated with lithium, and 13 bipolar subjects not receiving lithium treatment, total Gsk-3β level was found to be higher in bipolar manic patients than in healthy controls (Li et al., 2007). Compared to healthy controls, it has been reported that total Gsk-3β is decreased in platelets and increased in peripheral blood mononuclear cells at various stages of the disease in patients with bipolar disorder (De Sousa et al., 2015).

The fact that Gsk-3β, which is a critical regulator in broad aspects of nerve development, is associated with many ASD-related genes, such as FMR1, PTEN, APC, DISC1, SHANK3, MeCP2, and Reelin. Abrahams and Geschwind (2008) and Kim and Snider (2011) suggests that it is also important in the pathophysiology of ASD, and the relationship between ASD and Gsk-3β has recently begun to be investigated.

Studies have shown that Gsk-3β knockout mice exhibit autistic repetitive behaviors and social and learning difficulties (Feng et al., 2021). The first study to describe the links between Gsk-3β and social behaviors was an animal study by Mines et al (2010) with Fragile X Mental Retardation 1 gene knockout mice. This study shows that impaired inhibitory regulation of Gsk-3β may contribute to some social disorders and may be partially remedied by lithium administration (Mines et al., 2010).

Autopsies of patients with ASD showed that Gsk-3β expression was significantly reduced in the cerebral cortex, hippocampus, and striatum (Feng et al., 2021). In a postmortem study in ASD, no significant difference was observed between Gsk-3β levels in the cerebellum or superior frontal cortex of autistic individuals compared to matched controls (Fatemi et al., 2005). Results in a mouse autism model study showed that Gsk-3β expression was severely reduced in the prefrontal cortex, hippocampus, and cerebellum. It has also been shown that Gsk-3β immunoreactive Purkinje cells in the cerebellum are significantly reduced (Chen et al., 2009).

There are numerous studies showing that the WNT/β-catenin pathway, which has been shown to cause inhibition of Gsk-3β, is hyperactive in ASD (Kwan et al., 2016, Mulligan and Cheyette, 2016, Caracci et al., 2016). A review study investigating the ERK signaling pathway known to inhibit Gsk-3β in autism and bipolar disorder revealed that the ERK cascade was inappropriately activated in a subset of autistic patients, whereas the activity of the ERK cascade was hypoactive in bipolar patients, at least in one subset of patients. Considering the alleged hyperactivity of the ERK pathway in autism and its inhibitory effect on Gsk-3β, it has been stated that Gsk-3β activation can be targeted (Kalkman, 2012). The relationship between PTEN and ASD, which causes decreased phosphorylation of Gsk-3β and therefore increased Gsk-3β activity, was also investigated by McBride et al. and PTEN mutations were detected in 5–17% of children with autism (Mcbride et al., 2010). PTEN mutation can cause an increase in head circumference, which is a feature of autism, and in addition, people with PTEN mutation show spontaneous social and cognitive impairments. This relationship between PTEN mutation and autism has been supported in many recent studies (Buxbaum et al., 2007, Zhou and Parada, 2012, Garcia-Junco-Clemente and Golshani, 2014). In a study that included 41 children with ASD and 31 typically developing children, children with autism were shown to have lower Gsk-3β activity in isolated T lymphocytes. Current evidence suggests that low Gsk-3β activity negatively affects long-term depression (LTD), one of the pathways for synaptic plasticity in the brain (Onore et al., 2017). Considered a neuroprotective mechanism, LTD (Chrobak and Soltys, 2017) is involved in synapse elimination and synaptic pruning, and the molecular pathways involved in synaptic pruning are largely the same as those required for induction of LTD (Piochon et al., 2016). LTD deregulation is a synaptic abnormality frequently found in mouse models of autism (Hansel, 2019). To address the normal function of Gsk-3β in the central nervous system, a study recording from hippocampal neurons discovered that various Gsk-3β inhibitors completely inhibited the induction of LTD (Peineau et al., 2007). In vitro studies showing that overexpression of Gsk-3β induces neuronal death while its inhibition protects cortical neurons from apoptosis supports the relationship between Gsk-3β and LTD (Hetman et al., 2000).

Studies support the effect of Gsk-3β suppression in dendritic pruning defects and synaptic disorders commonly seen in ASD patients (Tang et al., 2014).

Administration of lithium, a Gsk-3β inhibitor, to newborn rats displaying ASD-like behaviors suppressed the patients’ symptoms and ameliorated the defects in neurogenesis (Wu et al., 2014).

In our study, we found lower Gsk-3β levels in children with ASD compared to healthy controls. As far as we know, there is no study in the literature that measures the level of Gsk-3β in the serum of patients with autism. Recently, numerous animal studies, postmortem studies, pharmacological studies, and studies using peripheral blood cells investigating the relationship between autism and Gsk-3β have demonstrated that Gsk-3β plays an important role in the pathogenesis of ASD and that Gsk-3β is a promising therapeutic target for the treatment of ASD. In a significant part of the studies conducted in this area, it is seen that Gsk-3β activity is low in autism, pathways such as PI3K/Akt, WNT and ERK, which cause Gsk-3β inhibition, are hyperactive, and proteins such as PTEN, which cause Gsk-3β activation, are decreased. However, the relationship between Gsk-3β and ASD is complex, and studies with different results suggest different clinical assumptions. While VPA, which is known to inhibit Gsk-3β, causes autism-like symptoms in animal studies, lithium, one of the best known inhibitors of Gsk-3β, can improve in animal models with ASD. Gsk-3β is involved in a wide variety of signaling pathways and developmental processes, and therefore clinical translation of Gsk-3β therapy is thought to be very difficult. Although cumulative data show that Gsk-3β plays a central role in ASD, the relationship between ASD and Gsk-3β is complex and is associated with multiple pathways in which Gsk-3β plays a role. There is a need for studies that comprehensively investigate the signaling pathways associated with Gsk-3β.

In our study, a statistically significant positive correlation was found between Nrf2 and Gsk-3β in the case group. Considering the relationship between Nrf2 and Gsk-3β, it has been revealed in many studies that Gsk-3β is one of the most important regulatory mechanisms of Nrf2 and that Gsk-3β reduces Nrf2 levels (Cuadrado, 2015, Shaw and Chattopadhyay, 2020, Zang et al., 2020). While it is expected that low Nrf2 may be caused by high Gsk-3β, the low Gsk-3β in the autistic group suggests that a simple explanation will not be sufficient, and the relationship between autism and this signaling pathway should be investigated. Due to the versatile modulation of Nrf and Gsk-3β, studies are needed to measure proteins such as βTrcp, Cul1, Cul3, P62, P21, which play a role in Nrf2 modulation, and parameters such as PI3K/Akt, WNT, ERK and PTEN, which play a role in Gsk-3 β modulation. In addition, the fact that phosphorylated Nrf2 levels were not measured is a limitation of our study, and it is thought that monitoring the phosphorylated Nrf2 level in future studies may be helpful. We think that the fact that markers such as ADH3 and ALDH, which have a role in the development of autism and have a complex relationship with the Nrf2 pathway (Wei et al., 2022, Zuo et al., 2013) were not examined, is an important deficiency of our article.

In our study, a low and statistically significant correlation was found between age and Gsk-3β. When the literature is examined, it is seen that infants are more vulnerable to oxidative stress than adults, and that oxidative stress is higher in the neonatal, infancy period and in older ages (Ono et al., 2001, Liu and Choi, 2000, Liu, 2002).

An important finding of our study that can be associated with the above findings is the positive correlation between increased Gsk-3β serum levels and age. The fact that the early age group with a high risk of oxidative stress was not included in our study is a limitation of our study. Literature data indicating that the risk of oxidative stress is high in the neonatal and infant period also shows the importance of early treatment approach in patients with autism. In addition, studies indicating an age-related decrease in Nrf2 and a decrease in its transcriptional activity (Shih and Yen, 2007, Li et al., 2006) also support that oxidative stress increases with age. However, in our study, no relationship was found between Nrf2 and age.

In our study, no significant difference was found between Nrf2 and Gsk-3β values and gender, but a significant difference was found between Keap1 values and gender. Accordingly, Keap1 values were found to be higher in males. Although it is important to understand the oxidative effect of gender in ASD, limited research has been conducted on potential gender differences in immunological profiles due to the higher diagnosis rate in men. In a study that comprehensively evaluated the level of oxidative stress in children with ASD, no significant gender differences were found in plasma levels of reactive oxygen metabolites and in the ratio of biological antioxidant potential/reactive oxygen metabolites (Morimoto et al., 2020). A comprehensive meta-analysis study stated that gender has a moderate effect on oxidative stress in ASD, and this result is plausible given that the male/female ratio for ASD is 4:1 (Chen et al., 2021).

The fact that Keap1 was found to be high in men in our study supports the literature knowledge that oxidative stress is higher in men. These data, together with extensive research in this area, may also help us to understand gender-related biological differences in the etiology of autism.

In our study, no significant difference was found between autism severity and Nrf2, Keap1 and Gsk-3β levels. In a study investigating markers of urinary oxidative stress in children with autism, a significant correlation was found between the severity of autism and urinary lipid peroxidation products. While oxidative stress markers were positively correlated with autism severity, antioxidants were negatively correlated (Damodaran and Arumugam, 2011). Bjørklund et al. in their review evaluating the relationship between ASD and oxidative stress, they stated that oxidative stress is an integral part of the pathophysiology of ASD and that oxidative stress is associated with ASD symptom severity (Bjørklund et al., 2020). Altun et al. (2018) found a positive correlation between CARS scores and another oxidative stress parameter, malondialdehyde, in their study in which 52 ASD and 48 healthy control children were included (Altun et al., 2018). In another study, no correlation was found between CARS scores and HO-1, Keap1 and Nrf2 parameters (Ayaydin et al., 2020a). In our study, no correlation was found between Nrf2, Keap1 and Gsk-3β and autism severity, consistent with this study. These results suggest that these markers play a role in the etiology of autism independent of autism severity.

When the Nrf2, Keap1 and Gsk-3β levels of the cases were compared with whether the cases received special education, a statistically significant difference was found between special education and Nrf2. Nrf2 levels were found to be higher in the group receiving special education. Special education has been shown to be effective in both improving social functioning and reducing clinically significant maladaptive behaviors. Special education provides striking gains in terms of IQ, nonverbal IQ, expressive and receptive language, and adaptive behavior, and the effects of these gains on neural activation have been demonstrated using neuroimaging methods (Lei and Ventola, 2017, Smith, 2013, Peters-Scheffer et al., 2011). The fact that Nrf2 was found to be higher in special education areas in our study suggests that special education may indicate positive effects on the biology of individuals. Future studies should try to characterize how special education interacts with an individual’s biology, which is crucial in autism.

When family history of ASD and Nrf2, Keap1 and Gsk-3β levels were compared, a statistically significant difference was found between family history of ASD and Keap1, and Keap1 level was found to be higher in those with a family history of ASD. The fact that the prevalence of autism is very high relative to population prevalence in families where a sibling of a child diagnosed with autism is affected, siblings and parents of an affected child are more likely than control groups to display subtle cognitive or behavioral traits that are qualitatively similar to those observed in probands, siblings of a child with autism at higher risk of sibling relapse (Ozonoff et al., 2011, Waye and Cheng, 2018, Abrahams and Geschwind, 2008, Lichtenstein et al., 2010) confirms that there is a strong hereditary component of autism. Considering these results showing a strong genetic etiology in autism, the fact that Keap1 level was higher in those with a family history of ASD in our study may indicate that oxidative stress is higher in the group with a family history. It is thought that studies on this subject will be valuable in terms of showing the biological differences in individuals with and without a history of ASD.

The strengths of our study are that the levels in our study were largely consistent with the literature data in the autistic group, the case and control groups included in the study were equalized in terms of age and gender, neurological, genetic or chronic medical diseases that could affect the oxidative stress levels were excluded in the participants, and CARS scale was performed by the same clinician. The limitations of our study are that our study was cross-sectional, the sample size was relatively small, the case group used psychotropic drugs, the nutritional characteristics of children with autism were not examined, and the Nrf2, Keap1 and Gsk-3β levels were measured from the serum.

Conclusions

In conclusion, since Nrf2 is the main regulator of the cellular defense mechanism against oxidative stress, a correct understanding of this pathway and identification of key proteins that regulate this pathway may allow us to understand the role of oxidative stress in autism. Considering the versatile modulation of the Nrf2 pathway, our article does not provide any mechanistic insight into the Nrf2 pathway, but suggests that Nrf2, Keap1 and Gsk-3β, which have central roles in oxidative stress, may play a role in the pathophysiology of autism. It would provide a better understanding of the pathophysiology of ASD and guide potential treatments, if future studies in ASD patients were replicated in a larger, well-matched population and designed by considering combinations of different biomarkers in the same pathway.

Funding Statement

We would like to thank Gaziantep University Scientific Research Center for financing this project.

Disclosure statement

No potential conflict of interest was reported by the authors.