The latest technological advancements in the genetic field has allowed researchers the opportunity to detect specific gene loci involved in neuropsychiatric and developmental disorders. In the 1980's, researchers began to consider the possibility of a strong genetic component underpinning ASD. Since then, many genes were found to be associated to autistic traits. Genome-wide association studies (GWAS) highlighted the association between autism and genes, from rare ones, such as PARK2 (Yin et al., 2016), to those which are more common, like SHANK3 (Nemirovsky et al., 2015; Sanders et al., 2015; Guo et al., 2017), FOXP1, FHIT, (The Autism Spectrum Disorders Working Group of The Psychiatric Genomics Consortium, 2017) and independent polymorphisms, like rs57709857-A (The Autism Spectrum Disorders Working Group of The Psychiatric Genomics Consortium, 2017). Although this technique efficiently reveals genes involved in autism within a shorter period of time, findings should be interpreted with caution given the great overlap present between genes involved in ASD and other medical/neuropsychiatric conditions. The current literature presents evidence of shared genetic, biological, and physiological etiology between mental disorders, such as ASD and ADHD (Craig et al., 2015; Panagiotidi et al., 2017), schizophrenia (Ishizuka et al., 2017; Prata et al., 2017), major depressive disorder (Glessner et al., 2017), and medical conditions like epilepsy (for a review see Strasser et al., 2017) and disorders of the immune system (Cieślinska et al., 2017; Li et al., 2017). Aside from these considerations, other limiting aspects of GWAS studies include the need for greater sample sizes and difficulties in replicating studies involving a population with ASD. These complexities may be due to the heterogeneous nature of the disorder, differences among ethnicities, and other environmental factors which render genome-wide studies useful but not resolutive.

Candidate gene studies is an alternative approach that requires large sample sizes to explore haplotypes, single genes, or even a few polymorphisms at a time. Some optimistic developments in research of OXTr and AVPr receptor classes have been required through the use of candidate gene studies. For instance, this method has elucidated specific associations of OXTr and AVPr polymorphisms to specific structures of the brain (e.g., amygdala) (Schneider-Hassloff et al., 2016), as well as to downstream physiological pathways such as the hypothalamic-pituitary axis (Auer et al., 2015). Furthermore, this methodology is more applicable to the new criteria of ASD diagnosis found in the latest edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), which regards disorders to exist on a continuous spectrum, ranging from the general population to the severely impaired clinical group.

Autism has been shown to be a heterogeneous and complex condition with multiple etiologies, resulting in a diverse array of clinical presentations across individuals. With revisions in the clustering of symptoms and criteria for diagnosis, DSM-5 has prescribed to the notion of “spectrum,” giving importance not only to singular features but also to the extent to which these symptoms impair daily functioning. Ever since this latest DSM edition has been applied to clinical practices, a decrease in number of ASD diagnoses has been observed (Kulage et al., 2014; Sturmey and Dalfern, 2014; Smith et al., 2015). In distinguishing the degree of severity of ASD symptoms, this paradigm shift has established a continuum of autistic traits that ranges from the mild to the severely impaired. DSM-5 more accurately reflects how social behaviors widely differ from person to person, making it possible for autistic traits to be found in the general population too, especially in the case where individuals share common environmental factors (Skylark and Baron-Cohen, 2017; Suzuki et al., 2017). Recently, it has been observed that ASD and autistic traits also share similar genetic and biological elements (Bralten et al., 2017), but it is necessary to take a look at these information emerging from literature holistically, before speculating the mechanisms underlying the development of ASD. The pursuit of research that investigates common autistic characteristics across both clinical and non-clinical populations might be helpful in developing more streamlined diagnostic practices.

In the numerous attempts to define the etiology of autism, much progress has been made possible due to research on animal models, which has exponentially increased since the use of the transgenic mouse. To date, this technique has allowed researchers to manipulate genetic information (Halladay et al., 2009), allowing control over genetic factors that contribute to the development of neurodevelopmental disorders, potentially giving rise to new medical treatments (Avraham et al., 2017). Rodent models have been adopted in ASD studies to investigate mechanisms for prevention (Vuillermot et al., 2017), and potential therapeutic treatments (Chadman, 2017; Guoynes et al., 2017), as well as to better understand the involvement of the gut-brain axis (for a review see Nithianantharajah et al., 2017) and the immune system (Schwartzer et al., 2017) in ASD etiology. To illustrate this, we draw specific attention to the pivotal role that animal models have played in elucidating how maternal immune activation, a well-ascertained environmental risk factor, contributes to ASD. Using immune activating agents, such as polyinosinic-polycytidylic acid (PolyIC), these studies induce an activation of the maternal immune system in pregnant rodents (Medzhitov, 2001). Following that, parameters of ASD-like behaviors in the offspring, such as pup ultrasonic vocalizations (Malkova et al., 2012; Hsiao et al., 2013; Schwartzer et al., 2013; Choi et al., 2016) and brain abnormalities, are compared against parallel ASD characteristics in humans (for reviews, see Meyer et al., 2009; Boksa, 2010). Cross-species evaluation has generated interest in examining the evolutionarily conserved effects of maternal immune activation on ASD-like behaviors (Belzung and Lemoine, 2011; Stewart and Kalueff, 2015) in rat and non-human primate models as well. While mice models have been the long-standing preferred species in biomedical research due to the extensive range of possible genetic manipulations, rat models feature a more complex array of brain activities and enriched social behaviors (Yee et al., 2012; Vanderschuren and Trezza, 2013; Ku et al., 2016), that serve as an advantage for studies relating to autism, since this disorder primarily features a deficiency in social cognition (Couture et al., 2010). Despite the numerous advantages that the rodent species has to offer, the social signals (i.e., vocalizations, facial expressions, body gestures) and underlying patterns of brain activity in rodents are limited (Chang et al., 2013). Thus, non-human primate models, such as the rhesus monkey, bridges the link from rodent models to humans (Watson and Platt, 2012; Platt et al., 2016), although such studies may be rife with ethical concerns.

With respect to OXT and AVP systems, evidence in literature highlights the contribution of these two peptides in social processes (Modi and Young, 2012; Peñagarikano et al., 2015; Zimmermann et al., 2016). Engineering of animal models specific to genes implicated in these systems is crucial in aiding the investigation of ASD etiology from this research front (Hammock and Young, 2006; Murakami et al., 2011; Peñagarikano, 2017). Indeed, OXT and OXTr knockout (KO) mice have been found to exhibit abnormal social behaviors akin to ASD-like traits (Ferguson et al., 2000; Winslow and Insel, 2002) and futher studies have begun to uncover the specific biological pathways involved. For example, Fujiwara et al. (2016) have recently investigated the role of HPC-1/syntax-in1A (STX1A), a N-ethylmaleimide-sensitive fusion attachment protein receptor complex which regulates the synaptic transmission of OXT (Südhof and Rothman, 2009). Employing a social novelty preference test, they found that STX1A KO mice showed unusual social behavior that is thought to be a component of a stereotypic ASD-like behavioral profile (Burket et al., 2011), also displayed by OXTR KO mice (Fujiwara et al., 2010). However, intracerebroventricular administration of oxytocin (OXT) partially rescued this phenotype (Fujiwara et al., 2016). This is a prime depiction of how animal model studies illuminate biological pathways related to the ASD phenotype as well as gives insight into preclinical research options that potentially advances the treatment of this disorder.

The core symptoms of autism revolve around deficits in normal social communication (Snow et al., 2009; Sala et al., 2011). Two neuropeptides of interest, oxytocin (OXT) and arginine-vasopressin (AVP), have been well-documented to mediate social and cognitive processes in clinical and non-clinical settings (Neumann, 2008; Ebstein et al., 2009, 2012), as well as across species (Veenema and Neumann, 2008; Ebstein et al., 2009; Zhang et al., 2017). They have been established to facilitate biologically adaptive social behaviors, such as trust (Kosfeld et al., 2005) positive communication (Ditzen et al., 2009), maternal care (Bielsky et al., 2005; Bosch and Neumann, 2008), and affiliation (Lim et al., 2005; Veenema et al., 2010). Because of their integral function in shaping social conduct, researchers began to postulate the involvement of these neuropeptides in the etiology of ASD (Insel et al., 1999; Hammock and Young, 2006; Watanabe et al., 2017). Indeed, animal model studies have shown that administration of OXT and AVP were able to rescue autistic traits (Sala et al., 2011) and increase social recognition memory (Bielsky and Young, 2004; Winslow and Insel, 2004), while in humans, intranasal OXT has been reported to heighten the understanding of others' mental states (Domes et al., 2007) - a deficit found amongst individuals with ASD (Caronna et al., 2008). As these neuropeptides exert their behavioral effects upon binding to three specific G-protein-coupled (Gimpl and Fahrenholz, 2001; Thibonnier et al., 2002) receptors; OXTr, AVPR1a, and AVPr1b (Manning et al., 2008), the next advance in this field was made when scientists looked toward OXTr (Wu et al., 2005; Jacob et al., 2007; Lerer et al., 2008) and AVPr (Kim et al., 2002; Wassink et al., 2004; Yirmiya et al., 2006) as potential candidate genes (for a review, see Zhang et al., 2017).

Known for its role in facilitating social behavior, OXTr has been implicated in neuropsychiatric disorders, ranging from social anxiety disorders and schizophrenia (Wu et al., 2005; Jacob et al., 2007; Lerer et al., 2008; Yrigollen et al., 2008; Ebstein et al., 2009; Gregory et al., 2009; Liu et al., 2010; Guastella and MacLeod, 2012; Meyer-Lindenberg and Tost, 2012) to autism (Wu et al., 2005; Ylisaukko-oja et al., 2006; Jacob et al., 2007; Lerer et al., 2008; Yrigollen et al., 2008; Gregory et al., 2009; Liu et al., 2010; Campbell et al., 2011). Based on numerous genome-wide association studies, OXTr has been proposed to be a potential risk gene in ASD. Supporting this proposition is a 0.7 Mb deletion in 3p25.3, which has been found to occur with autism (see Figure 1). Since OXTr was encompassed in the deleted region, such mutations implicating OXTr were suggested to be involved in ASD etiology (Wu et al., 2005; Jacob et al., 2007; Lerer et al., 2008). However, this phenomenon appears to be rare as other studies have not reported OXTr deletions typifying autism (Sebat et al., 2007; Marshall et al., 2008; Glessner et al., 2009; Sanders et al., 2015). As compared to deletions, the association of OXTr SNPs with ASD has been more commonly observed. Several SNPs of OXTr have been linked to ASD, including rs237887, rs2268491 and rs225429, rs7632287 (Ylisaukko-oja et al., 2006; Jacob et al., 2007; Lerer et al., 2008; Yrigollen et al., 2008; Ebstein et al., 2009; Liu et al., 2010; Campbell et al., 2011; Walum et al., 2012; LoParo and Waldman, 2015). However, some studies did not relate any association (Tansey et al., 2010; Wermter et al., 2010; Verhallen et al., 2017), while others reported ethnic differences in the association of OXTr SNPs with autism (Wu et al., 2005; Jacob et al., 2007; Lerer et al., 2008; Liu et al., 2010).

In the domain of social behavior, AVPR1a (Lim et al., 2005; Veenema et al., 2010) and AVPR1b are two integral players from the AVP system. Well-known to be located on chromosome 12q14-15, the gene for AVPR1a possesses three microsatellites in the 5u flanking region and a fourth one in an intronic region (Thibonnier et al., 2000; Thibonnier, 2004) (see Figure 2). Expression patterns of AVPR1a in the brain significantly predict differences in social behavior of voles within and across species (Hammock and Young, 2002, 2006). Of clinical relevance, AVPR1a polymorphism has been shown to be involved in social recognition (Bielsky et al., 2005), amygdala activation (Meyer-Lindenberg et al., 2009) and implicated in ASD (Kim et al., 2002; Wassink et al., 2004; Yirmiya et al., 2006). In 2002, Kim et al. (2002) was the first group to document the significant transmission disequilibrium of a microsatellite on AVPR1a and autism, before Wassink et al. (2004) and Yirmiya et al. (2006) independently reproduced this finding. (Israel et al., 2008) found that deficits in socialization skills mediated this link. For the past decade, the central wellspring of research in AVPr lies in the promoter region of two microsatellites, RS1 and RS3, where length of microsatellites has been found to modulate social behavior in human and lower mammals (Israel et al., 2008; Knafo et al., 2008). This observation was supported by corresponding mRNA analyses, in which longer AVPR1a RS3 repeats were found to be associated with higher levels of AVPR1a mRNA expression (Knafo et al., 2008). This asserts the functional role of different AVPR1a microsatellites in determining social behavior. Of less prominence in the study of autism is the receptor AVPR1b, which has been previously linked to aggression, anxiety, pro-sociality and autistic traits (Dempster, 2007; Chakrabarti et al., 2009; Dempster et al., 2009; Zai et al., 2012; Wu et al., 2015). Only recently, Francis et al. (2016) reported a significant association between AVPR1b and ASD.

Having chronicled the consensus and contention surrounding OXTr and AVPr in ASD, we have established the foundation for this review. Thereafter, we will endeavor to highlight specific polymorphisms of OXTr and AVPr that have been either recently or repeatedly associated with autism. Exploring the breadth of the current literature, we will provide suggestions on how inconsistencies can be approached and reconciled, as well as recommend possible future directions that might advance this field of research further.

There is robust evidence to posit that oxytocin receptor gene plays a determining role in social behaviors, ranging from individual differences in non-clinical populations to ASD. Many studies have been conducted in the past two decades to better understand which polymorphisms have a bigger impact in the development of autism. A pioneer study by Horvath and colleagues found a significant correlation between A carriers in OXTr SNPs rs53576 and rs2254298 and autism (Horvath et al., 2001). Subsequently, other authors expanded our understanding of this relationship (LoParo and Waldman, 2015), finding a correlation between oxytocin receptor gene variants (rs53576, rs2254298 mainly) and autistic traits (Wu et al., 2005; Jacob et al., 2007; Lerer et al., 2008).

For OXTr rs53576, it has been demonstrated that A carriers, who are considered more vulnerable compared to G carriers, display reduced hypothalamic volume, that is shown to be related to susceptibility in ASD development (Liu et al., 2010; Tost et al., 2010; Jack et al., 2012; Wang et al., 2013). Many researchers have contributed studies which revealed the influence of this polymorphism on different social behavior. Specifically, there are evidences of correlations between rs53576 and prosocial disposition (Tost et al., 2010; Kogan et al., 2011; Yamasue, 2013), emotional withdrawal in psychotic disorders (Haram et al., 2015), negative affect (Lucht et al., 2009; Saphire-Bernstein et al., 2011), and reaction to stress and empathy (Rodrigues et al., 2009; Chen et al., 2011; Weisman et al., 2015; McDonald et al., 2016). Indeed, rs53576 has received much attention by researchers in the field of autism, with results suggesting that its involvement is not limited to ASD alone, but influences a wide spectrum of phenotypes related to general sociability and affiliative behaviors (LoParo and Waldman, 2015; Zhang et al., 2017). Generally, these results suggest that rs53576 is more related to sensitivity to social environment than to ASD itself (Tost et al., 2010; Jack et al., 2012; Wang et al., 2013).

Presently, studies on rs53576 have begun to investigate the role of epigenetic modification. For instance, Rijlaarsdam and colleagues have has recently discovered a significant interaction between OXTr methylation and child autistic traits, as well as social communication problems (Rijlaarsdam et al., 2017). Specifically, greater OXTr methylation was found to be correlated with social communication problems in rs53576 G-allele homozygotes, rather than A-allele carriers. Such findings encourage us to question the link that has been previously established between ASD susceptibility and the A-allele. This underscores the pertinence of accounting for epigenetic data in understanding the role of OXTr polymorphisms in ASD.

Another well-investigated polymorphism of OXTr is rs2254298, which is located in the third intron, with a guanine to adenine variation that is supposedly the “risk-inducing” form. As with rs53576, several studies attempted to understand the impact of rs2254298 on the development of ASD. Neuroimaging studies revealed a correlation between this variation and increased amygdala volume in general (Furman et al., 2011; Brüne, 2012) and when combined with environmental factors of early life stress (Marusak et al., 2015). Starting from the first genetic study on the etiology of autism in a Chinese population (Wu et al., 2005), others followed suit to find a significant correlation between rs2254298 and ASD diagnosis in Caucasian and Japanese samples (Jacob et al., 2007; Liu et al., 2010; LoParo and Waldman, 2015). Other findings centered on OXTr rs2254298 show that autistic traits are related to this polymorphism, but only when it is comprised in a haplotype (Lerer et al., 2008; Mosconi et al., 2009; Nyffeler et al., 2014; Francis et al., 2016). Besides autism, rs2254298 was also linked to social impairment in general (Parker et al., 2014), empathy (Wu et al., 2012) anxiety and depression (Mosconi et al., 2009; Thompson et al., 2011; Bittencourt Jacondino et al., 2014).

On the contrary, some authors found no significant correlation between OXTr rs2254298 and etiology of autism (Tansey et al., 2010; Campbell et al., 2011; Montag et al., 2017). An enlightening meta-analysis by Bakermans-Kranenburg and van Ijzendoorn emphasized the need to account for epigenetic mechanisms when examining SNPs like rs2254298 due to the lack of available research in the literature relating epigenetics not only to ASD development but to general social behavior (Bakermans-Kranenburg and van Ijzendoorn, 2014). Indeed, present data on genetic polymorphisms, up to the year 2014, fail to explain a considerable part of social behavior in humans. Another recent large study found no association between common variants of OXTr and AVPr, and social integration (Chang et al., 2014), although the authors suggested that the results could have been small. Additionally, the study was run on healthy non-clinical populations, not taking ASD diagnosis into account.

The results of the abovementioned studies pushed researchers to uncover the relationships between other OXTr variants in the promoter region and ASD in greater depth. Among these, rs2268493 has been found to correlate with ASD diagnosis (Yrigollen et al., 2008; Di Napoli et al., 2014), specifically with social withdrawal and repetitive behavior, both as an individual SNP and as part of a haplotype (Ebstein et al., 2009; Francis et al., 2016). There are also evidences of its relation with social cognitive performances in patients with schizophrenia (Davis et al., 2014) and in children diagnosed with ADHD (Ayaz et al., 2015); both disorders share features and symptoms with ASD. Campbell and colleagues found a statistically significant correlation of rs2268493 with autism but the significance was not retained after multiple comparison analysis (Campbell et al., 2011).

In the 3′-untranslated region of the OXTr gene, there is another SNP investigated in association with ASD. rs1042778 was found to be statistically correlated to ASD diagnosis in Caucasian and Jewish populations (Jacob et al., 2007; Lerer et al., 2008; Campbell et al., 2011), but the same result was not confirmed in the Japanese population (Liu et al., 2010). Furthermore, there are suggestions of its implication in aggression (Israel et al., 2009; Johansson et al., 2012; Malik et al., 2012) and prosocial behavior (Israel et al., 2009). With respect to rs13316193, there have been a few studies that found an association of this SNP, contained in a haplotype, with ASD (Ebstein et al., 2009; Liu et al., 2010). As an individual SNP, it has been associated with poor social abilities (Ebstein et al., 2012). Other SNPs have been tested in the last decade, but results are still few and contradictory (Apicella et al., 2010; Tansey et al., 2010), suggesting that these data need to be interpreted with caution and that more research is necessary. Other variants of OXTr that have been found to be related to autism and autistic traits are rs237897 (Lerer et al., 2008; Yrigollen et al., 2008; Wu et al., 2012; Kranz et al., 2016), rs237889 and rs2268494 (Lerer et al., 2008), rs237887 and rs237885 (Liu et al., 2010). Campbell and colleagues found a correlation between rs7632287 and multiple domains of ASD (Campbell et al., 2011) and childhood social problems such as pair-bonding between mother and child (Walum et al., 2012).

Arginine-vasopressin is involved in regulatory mechanisms of social behavior like attachment, aggressive behavior and anxiety due to the activity of AVP receptors in the hypothalamic-pituitary-adrenal axis (Tansey et al., 2011). In a recent study, Shou and colleagues found that children with ASD, as compared to typically developing ones, showed a positive correlation between AVP levels in the plasma, increased volume in the left amygdala and left hippocampus, and decreased volume in the bilateral hypothalamus, suggesting a contribution of these structural and functional changes to the etiology of ASD (Shou et al., 2017). Studies about social behavior have shown the involvement of AVPR1a and AVPR1b gene polymorphisms (Avinun et al., 2011; Zai et al., 2012). Unlike oxytocin receptor gene, arginine-vasopressin receptor is composed by different encoder genes, such as AVPR1a, AVPR1b and AVPR2. In particular, two nucleotides located in the 5′ promoter region of AVPr; RS1 and RS3, have been the focus of research in this field, with evidence showing that these microsatellites modulate a wide range of social behaviors in both ASD and neurotypical populations (Kim et al., 2002; Wassink et al., 2004; Francis et al., 2016).

Arginine vasopressin receptor 1a (AVPR1a) is largely expressed throughout the brain (Kantojärvi et al., 2015), but also in the liver, kidneys, and vasculature. It is considered to be a crucial receptor for regulation of a set of social behaviors in humans (Ebstein et al., 2009, 2012; Francis et al., 2016), such as altruism and social integration (Israel et al., 2008; Knafo et al., 2008), prepulse inhibition (Levin et al., 2009), emotional and cognitive empathy (Israel et al., 2008; Knafo et al., 2008; Moons et al., 2014; Uzefovsky et al., 2015; Wang et al., 2016), processing and recognition of facial expressions (Israel et al., 2008; Knafo et al., 2008; Moons et al., 2014; Uzefovsky et al., 2015; Wang et al., 2016) and non-human mammals (Murakami et al., 2011; Freeman et al., 2014; Paré et al., 2016; Lesse et al., 2017). Even though vasopressin distribution patterns are quite similar across species, AVPR1a partitions differ both between and within species. In humans, vasopressin production occurs in the hypothalamus, bed nucleus of the stria terminalis, and the medial amygdala, while binding sites are located in the lateral septum, thalamus, basal amygdaloid nucleus, and brainstem (Meyer-Lindenberg et al., 2009; Zhang et al., 2017). Starting from the beginning of the century, many authors highlighted an association between variation at AVPR1a and neuropsychiatric phenotypes, with a wide focus on autism (Kim et al., 2002; Yirmiya et al., 2006; Tansey et al., 2011; Francis et al., 2016), parental behavior (Avinun et al., 2012) and social behavior in general (Yirmiya et al., 2006; Meyer-Lindenberg et al., 2009; Ebstein et al., 2012). Human AVPR1a promoter region contains two microsatellite repeats, which are highly polymorphic in the general population, and have been studied in association with individual variation in social behavior (Ebstein et al., 2012; Procyshyn et al., 2016): RS1, a (GATA)14 repeat sequence, and RS3, a complex (CT)4TT(CT)8(GT)24 repeat sequence (Thibonnier et al., 2000). Studies report an association between repeat number of these microsatellites and expression of AVPR1a in the hippocampus and the brain in general (Knafo et al., 2008). As a consequence, this can affect the modulation of social behavior on non-clinical autistic phenotypes (Carver and Harmon-Jones, 2009; Moons et al., 2014) and autistic traits (Israel et al., 2008; Procyshyn et al., 2016).

Specifically, results revealed an association between “short-alleles” of variant RS1 and a decrease in transcription of AVPR1a gene in the amygdala, leading to poorer prosocial behavior (Poulin et al., 2012; Kantojärvi et al., 2015; Yang et al., 2017) and higher amygdala activation during a facial recognition task (Meyer-Lindenberg et al., 2009). Kim et al. (2002) have been the first to conduct molecular genetic studies on AVPR1a and ASD, finding that one of the alleles of RS1 showed increased transmission in a family based study of 115 trios (mother, father, and child). More studies on the same microsatellite highlighted similar results (Wassink et al., 2004; Yirmiya et al., 2006; Yang et al., 2010a; Tansey et al., 2011; Kantojärvi et al., 2015). On the contrary, it is the “long-alleles” variant in RS3 that is associated with an increased in activity of the amygdala, which has been linked to social withdrawal. Studies show associations between this microsatellite and cognitive empathy and social skills in humans (Bosch and Neumann, 2008; Knafo et al., 2008; Harony and Wagner, 2010; Avinun et al., 2011; Uzefovsky et al., 2015), especially in the context of mother-infant play (Avinun et al., 2012). The “long-alleles” variant in RS3 has been linked to a risk of developing ASD too (Ebstein et al., 2009; Kantojärvi et al., 2015; Procyshyn et al., 2016). These results confirmed previous findings on animal models (Insel et al., 1994).

The studies reviewed show that AVPR1a modulation over brain functioning and ASD traits, with specific involvement in the domains of social communication, emotion and cognition, is often considered a consequence of the haplotype RS1-RS3 (Meyer-Lindenberg et al., 2009). Some authors have started investigating other SNPs belonging to the promoter region of AVPR1a gene to investigate if the etiology of autism might be modulated by a joint mechanism. An Irish study conducted in 2011 tagged four SNPs of the AVPR1a gene (rs3803107, rs1042615, rs3741865, rs11174815) but results showed that only rs11174815 was weakly correlated to autism (Tansey et al., 2011). Another study found a significant correlation of rs7294536, rs3759292, and rs10877969 on autistic phenotypes in trios in a Korean population (Yang et al., 2010b). Findings from these two studies suggest that the underlying actions of these receptors remain elusive, and that results may be influenced by specific cultural environments.

While the relationship between AVPR1a and ASD is relatively consistent, there is still need for further studies on the impact of AVPR1b on the outcome of autism. At the present, our knowledge is limited to the involvement of this receptor in the modulation of neuroendocrine response to stress (Muráni et al., 2010; Roper et al., 2011) and developmental disorders (Dempster et al., 2009; van West et al., 2009). Evidence about the role of AVPR1b in social and affiliative behaviors come from mice models (Wersinger et al., 2004; Caldwell et al., 2008, 2010; Pagani et al., 2015). There are few results that shed light on the involvement of AVPR1b in autistic traits in humans (Zai et al., 2012; Luppino et al., 2014; Wu et al., 2015), although it is possible to find evidence in literature that associates AVPR1b rs28632197 with comorbid conditions such as panic disorder (Keck et al., 2008). Exception being a recent report by Francis and colleagues on a correlation between AVPR1b rs35369693 and autism (Francis et al., 2016).

As already mentioned, ASD is well known to be sexually biased, occurring 4 times more in boys than girls. Studies on animal models, specifically in mammals, showed that the influence of OXT and AVP on brain activity and behavior is sexually-dimorphic, with stronger effects of oxytocin on females and of arginine-vasopressin in males. Hence, oxytocinergic and vasopressinergic systems may be related to the sex-biased occurrence of ASD.

Among the works discussed in the present review, there are few evidences showing sex-specific effects of OXTr and AVPr (and their variations), that enhanced the hypothesis of a sexually dimorphic modulation of autistic spectrum traits, and social behaviors in general, by these two neurotransmitters. Regarding OXTr, a study by Chen and colleagues found a significant sex-dependent effect in the interaction between rs2254298 and social functioning. Female A-carriers tended to show greater anxiety while men reported more autism-associated traits (Chen et al., 2011).

Furthermore, research by Chang and colleagues on social connectedness highlighted an opposite association between men and women with respect to rs4686302, and a direct correlation between rs53576 and women only, suggesting that sex can determine differential effects from the same variation (Chang et al., 2014). Other studies revealed sex-based differences in the amygdala, showing greater gray matter volume in male A-carriers for rs53576 (Tost et al., 2010) and a negative correlation with scores in reward tasks (Ebstein et al., 2012). The same amygdalial structural observation was also detected in homozygous G-carrier women for rs2254298 (Furman et al., 2011). Sex has also been shown to predict emotional empathy, with women scoring higher than men (Wu et al., 2012; Uzefovsky et al., 2015), although some studies did not find any differences (Rodrigues et al., 2009).

With regards to diagnosis of ASD, significant correlations were found between the Japanese male population and rs2254298 (Liu et al., 2010; Watanabe et al., 2017) and rs53576 (Watanabe et al., 2017). In the case of DNA methylation, as compared to healthy individuals, Gregory and colleagues found a significant sex-specific decrease in the expression of OXTr in males with autism, correlated to an increase in methylation at a specific site. Heterozygous deletion of the same OXTr site in mothers included in the sample did not result in diagnosis of ASD while deletion in males did (Gregory et al., 2009).

Switching to the vasopressin system, a few studies explored sex differences of AVP, AVPr and its variation on autism and social behaviors. Animal models showed that AVP in the central nervous system (CNS) activates brain regions (i.e., amygdala, hyppocampus) that coordinate specific-social behaviors, such as intermale aggression (Veenema et al., 2010); with regards to the receptors, it is possible to find in literature evidence that AVPR1a is necessary and sufficient for social recognition (Bielsky et al., 2005), and that deficits are worse in the male sample (Winslow and Insel, 2004). Furthermore, AVP in the CNS coordinates a range of different social behaviors. Rodents, specifically male samples of AVPR1B knockout mice, have also been used to elucidate mechanisms related to social dominance (Caldwell et al., 2010).

With regards to humans, prepulse inhibition (PPI) has been studied in both sexes, due to its involvement in social skills deficits. In 2009, two studies explicated the correlation between PPI and AVPR1a RS3, finding higher responses in long-carriers (Levin et al., 2009), which were even more evident in men (Ebstein et al., 2009). As for AVPR1B, rs28373064 was found to be more linked to prosociality in men compared to women (Wu et al., 2015) and this mechanism might be modulated by emotional empathy. A study conducted on children with ASD highlighted a strong association between AVP blood plasma levels and the connection between the left amygdala and left supermarginal gyrus, but only in boys (Shou et al., 2017). The amygdala is well-known to be a sexually-dimorphic major nucleus, receiving neurons from the hypothalamus that contains AVP, so this might shed light on mechanisms that contribute to the etiology of ASD.

With regards to separate analysis by sex, some incongruence is still present in literature (Knafo et al., 2008), underlining the need for lower variability and bigger sample sizes (Procyshyn et al., 2016). The included studies represent only a small portion of the evidences that underlie the sexually-dymorphic nature of OXT and AVP in the etiology of autism. Looking at these neurotransmitters together, a study by Miller and colleagues in an ASD population found higher levels of OXT in blood plasma in girls, and of AVP in boys; this difference was also reflected in a diverse association with autistic symptoms, especially with anxiety levels and repetitive behaviors (Miller M. et al., 2013). OXTr and AVPr genes have been found to interact with blood plasma levels of oxytocin and vasopressin in generating responses to emotional stress and these effects are displayed differently in men and women (Moons et al., 2014).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.