Supplemental File 2 Biological interpretation of the candidate gene list Here we provide with a detailed discussion of the functional implications of the 49 genes that are annotated as related to SZ in Table 1 (either as candidate genomic loci or differentially expressed in SZ brain). These genes can be categorized according to the biological processes and molecular pathways they act within, in order to focus on the complete molecular scenario and the scientific rationale adopted in this paper.
Some of these genes are not included nor interact with the core set of genes proposed for the “domestication syndrome” by Wilkins et al (2014). Nonetheless, they play a role in NC development and in language function, and appear to be related to SZ molecular aetiopathogenesis, as either candidate genes, or as part of the gene expression profiles displayed by SZ brains. Accordingly, they strike us as promising candidates for the abnormal presentation of the “domestication syndrome” in this condition, plausibly accounting for some aspects of the SZ phenotype, particularly, in the domain of language.
1. FOX genes and their interactors Two genes encoding forkhead box proteins are included in our list, namely FOXD3 and FOXP2. These encodes member of a large family of transcription factors that finely regulate the spatial and temporal gene expression during development and beyond.
- FOXD3 is required during NC development and regulates dorsal mesoderm development in the zebrafish (Chang and Kessler, 2010). The gene locus maps within one of the AMH-specific differentially-methylated genomic regions (Gokhman et al., 2014), suggesting that changes in the gene functional status may have occurred during recent human evolution. It interacts with NODAL, a candidate for language-readiness encoding a key regulator of the establishment of bilateral symmetry in the embryo (Zhou et al., 1993; Lowe et al., 1996; Krebs et al., 2003). It is also needed for the regulatory role of FoxD3 on NC development (Chang and Kessler, 2010). Finally, FOXD3 is downregulated by DISC1, a well-known candidate for SZ, to control the timing and pattern of NC migration and differentiation of cranial NC cells (Drerup et al., 2009).
- FOXP2 is mostly involved in neurogenesis, neuron differentiation and migration, and cell morphology changes, in the developing telencephalon (Tsui et al., 2013; Chiu et al., 2014; Garcia-Calero et al., 2015). Mutations in this gene are known to cause speech and language impairment (Vargha-Khadem et al., 1995; Shriberg et al., 2006; Zhao et al., 2010), while some gene polymorphisms are associated to SZ (Tolosa et al., 2010). FOXP2 might contribute as well to the transition from declarative to procedural memory (Schreiweis et al., 2014), and the modulation of functional brain asymmetries for speech perception (Ocklenburg et al., 2013). Interestingly, several core candidates for domestication are FOXP2 targets, including EDN3, FGF8, MAGOH, and RET (Spiteri et al., 2007; Vernes et al., 2011).
- POU3F2 encodes a transcription factor that activates FOXP2 expression and regulates dopamine and serotonin synthesis, neuronal migration and identity in the neocortex (McEvilly et al., 2002; Dominguez et al., 2013; Nasu et al., 2014). Sequence and copy number variations of POU3F2 are found in SZ patients (Huang et al., 2005; Potkin et al., 2009). Interestingly, the human FOXP2 bears a derived allele of a binding site for POU3F2 which is less efficient than the Neanderthal version in activating transcription of the gene (Maricic et al., 2013). Hence, a change in the regulatory pattern of FOXP2 expression plausibly occurred during recent human evolution.
2. SOX genes This is a large family of highly conserved transcription factors, involved in various aspects of eukaryotic development. Three members of this family, which are core candidates for the “domestication syndrome” according to Wilkins et al. (2014), are included in our list: SOX2, SOX9, and SOX10.
- SOX2 encodes one component of the SHH-GLI signalling pathway, important for NC cells fate (Oosterveen et al., 2012; Oosterveen et al., 2013; Peterson et al., 2012), but also for the changes that leaded to the globularization of the human skull/brain (Boeckx and Benítez-Burraco 2015). Accordingly, SOX2 interacts with many genes important for language-readiness, including GLI3, PQBP1¸ DLX5, FOXP2, and RUNX2 and the BMP signalling. Interestingly, SOX2 is involved in the enhancer effect of human endogenous retroviruses (HERVs) on brain genes related to SZ, specifically PRODH (Suntsova et al., 2013). Interestingly too PQBP1 encodes a protein involved in neurite outgrowth associated with intellectual disability (Wang et al., 2013), developmental delay and microcephaly (Li et al., 2013).
- SOX9 is considered a master regulator of craniofacial development and it has been related to congenital skeletal malformations (Mansour et al., 2002; Gordon et al., 2009; Lee and Saint-Jeannet, 2011). The activation of Sox9, through phosphorilation, is required for the very early stages of NC formation, and occur as a downstream effect of Wnt and BMP pathway activation (Liu et al., 2013b). SOX9 brain upregulation observed in the datasets considered in this paper is also confirmed by the data obtained in an independent study (Shao and Vawter, 2008).
- SOX10 is involved in the maintenance of precursor cell pools and regulation of cellular migration and differentiation in the NC; furthermore it is implicated in DISC1-dependant oligodendrocyte differentiation (Hattori et al., 2014). SOX10 is also found hypermethylated in the brain of SZ patients (Iwamoto et al., 2005), plus several polymorphisms of SOX10 have been related to the age of onset of the disease (Yuan et al., 2013). Indeed, also SOX10 expression is regulated by the SZ-susceptibility gene DISC1 in early NC cell migration stages (Drerup et al., 2009). Interestingly, SOX10 interacts with PAX3, a gene involved in the “domestication syndrome” (Wilkins et al., 2014), and a candidate for language-readiness (Boeckx and Benítez-Burraco, 2014b). Finally, SOX10 is also capable of protein-protein interaction with POU3F2 (Smit et al., 2000).
3. DLX and MSX homeotic genes DLX and MSX genes are two families encoding homeobox-containing transcription factors that actively lead vertebrate development, being required for normal craniofacial, limb and brain morphogenesis. Five members of these two families are included in our list: DLX5, DLX6, DLX1, DLX2, and MSX1.
- DLX5 and DLX6 genes function in early NC development, and late specification of NC-derived structures (McLarren et al., 2003; Ruest et al., 2003), and play key roles in skull and brain development (Jones and Rubinstein, 2004; Kraus and Lufkin, 2006). Dlx5/6(+/-) mice show reduced cognitive flexibility that seemingly results from an abnormal pattern of γ rhythms, caused by abnormalities in GABAergic interneurons; this phenotype recapitulates some clinical findings of SZ patients (Cho et al., 2015). Both genes also interact with BMP signalling in driving the development of the mandible, as a result of NC cell migration to the pharyngeal arches (Vincentz et al., 2016). DLX5 also interacts with FOXP2 and RUNX2, key candidates for language evolution (see Boeckx and Benítez-Burraco, 2014a for details).
- DLX1 also contributes to the patterning and morphogenetic processes in the NC-derived craniofacial mesenchyme (Mallo, 2001; Ishii et al., 2012). Because of this role in cranial morphogenesis, also DLX1 is a core candidate for globularization and the emergence of language-readiness in our species (Boeckx and Benítez-Burraco, 2014a). Decreased expression of DLX1 has been observed in the thalamus of SZ patients, who show a reduced volume that correlates with language dysfunction (Kromkamp et al., 2003; Li et al., 2015). Indeed, thalamic function should reasonably contribute to the evolution of human cognition (see Boeckx and Benítez-Burraco, 2014a).
- DLX2 is also involved in craniofacial development (Jeong et al., 2008; Gordon et al., 2010), but also controls aspects of brain development (Jones and Rubenstein, 2004; Johnson et al., 2009; McKinsey et al., 2013). Both Dlx1 and Dlx2 control steps of neuronal proliferation within the cortex, via Zfhx1b, which is a candidate for neurocristophaties, like Mowat-Wilson syndrome and Hirschsprung disease (Van de Putte et al., 2003 and 2007; McKinsey et al., 2013). Mutations in DLX2 give rise to craniofacial, limb, and bone anomalies (Kraus and Lufkin, 2006), but also to autism and psychosis (Liu et al., 2009). Noticeably, dlx2a knockdown in the pharyngeal arches of zebrafish downregulates sox9a (Sperber et al., 2008).
- While MSX genes control the spatial organization of the NC-derived craniofacial skeleton (Attanasio et al., 2013; Khadka et al., 2006; Han et al., 2007; Gitton et al., 2011), MSX1 encodes a transcriptional repressor specifically involved in odontogenesis (Alappat et al., 2003; Cohen, 2000), whose variants are associated with orofacial clefting and tooth agenesis (Liang et al., 2016). Noticeably, MSX1 is a direct downstream target of DLX5 during early inner ear development (Sajan et al., 2011). Methylation changes in MSX1 are found in the hippocampus of SZ patients, as a part of the circuit-specific DNA methylation changes affecting the GAD1 regulatory network, which may explain GABAergic dysfunction in the disorder (Ruzicka et al., 2015).
4. FGF/FGFR signalling Fibroblast growth factors (FGF) genes encode multifunctional proteins binding any four tyrosine kinase receptors (FGFR). Most members of this gene family, known to interact with several other key developmental pathways, such as HH, BMP/TGFβ, and Wnt, are comprised in our genelist, specifically, FGF8, FGF7, FGFR1, and FGFR2.
- FGF8 is a core domestication candidate in mammals (Wilkins et al., 2014) and it is involved in the regionalization of brain tissue (Fukuchi-Shimogori and Grove, 2001). Mutations in FGF8 give rise to holoprosencephaly, a condition characterized by forebrain and midline facial anomalies, plus severe neurocognitive impairment (Sarnat and Flores-Sarnat, 2001; Solomon et al., 2012). FGF8 also interacts with many genes important for the globularization of the human skull/brain and the evolution of language-readiness, including FOXP2 (Spiteri et al., 2007), and RUNX2 (Trokovic et al., 2003; Komori, 2011; Stuhlmiller and García-Castro, 2012), and it is regulated by the SHH-GLI signalling pathway (Kobayashi et al., 2010; Rash and Grove, 2011).
- FGF7 also cooperates within the FGF and the BMP signalling during cranial NC development (Endo et al., 2012). FGF7 locus is listed among the top five percent regions showing signals of positive selection in AMHs compared with ancient hominins (Green et al., 2010).
- FGFR1 and FGFR2 are found mutated in both syndromic and (rarely) nonsyndromic craniosynostosis, featuring peculiar craniofacial features that suggest NC impairment (Lattanzi et al., 2012; Lattanzi, 2016). FGFR2 has been found associated to SZ risk susceptibility (O’Donovan et al., 2009). Consistent with what we have found in the tested datasets (Figure 2), elevated mRNA levels of FGFR1 are observed in the prefrontal cortex of young schizophrenics (Volk et al., 2016). Mutant mice with impaired Fgfr1 exhibit SZ-like behaviours that seemingly result from a general loss of neurons and postnatal glial dysfunction. The Integrative Nuclear FGFR1 Signalling (INFS) has been hypothesised to be one of the neurodevelopmental pathways on which the products of multiple SZ candidates converge, ultimately affecting the development of multiple neural circuits and neurotransmitter systems (Stachowiak et al., 2013).
5. EDN/EDNR signalling Endothelin genes emerged at the advent of vertebrates and encode proteins involved in the patterning of oropharyngeal and other NC-derived craniofacial structures (Kuraku et al., 2010). Four member of this signalling pathway are found in our list: EDN3, EDNRB, EDN1, and EDNRA.
- EDN3 is a core candidate for domestication according to Wilkins et al. (2014), and encodes an endothelium-derived vasoactive peptide which binds the product of EDNRB (other of Wilkins et al.’s candidates), playing a leading role in the development of NC-derived cell lineages, such as melanocytes and enteric neurons (Kurihara et al., 1999). Both EDN3 and EDNRB are candidates for two paradigms of neurocristopathies, namely Waardenburg syndrome and Hirschsprung disease (Sánchez-Mejías et al., 2010).
- EDN1 is involved in the dorsal-ventral patterning and growth of the craniofacial skeleton, through constant interaction with WNT and BMP pathways (Alexander et al., 2014). Homozygous mutations of EDN1 cause auriculocondylar syndrome, a rare craniofacial disorder involving first and second pharyngeal arch derivatives (Gordon et al., 2013). The EDN1 receptor, encoded by EDNRA, is a key determinant in the patterning of cephalic and cardiac NC derivatives, as demonstrated in knockout animal models (Clouthier et al., 1998; Yanagisawa et al., 1998). In particular, the development of branchial arch-derived structures is impaired in EdnrA-mutant mice, through the downregulation of Dlx6 (Charite et al., 2001).
6. PAX genes The paired box family comprises largely conserved homeodomain transcription factors, playing key roles in both early developmental stages and tissue specification. Three PAX genes are found among our candidates.
- PAX3 is among the earliest genes activated in NC progenitors and contributes to craniofacial patterning (Maczkowiak et al., 2010; Bae et al., 2014; Plouhinec et al., 2014). It interacts with other core candidates for the domestication syndrome, like SOX10, mentioned above (Lang and Epstein, 2003), and TCOF1 during the development of the enteric nervous system (derived from vagal NCC; Barlow et al., 2013). PAX3 is a candidate for Waardenburg syndrome (Tassabehji et al., 1992; Chen et al., 2010). PAX3 is mentioned by Wilkins et al (2014) among the candidates for the “domestication syndrome” in mammals.
- PAX6 is specifically involved in brain development: it regulates the migration of NCCs from the anterior midbrain (Matsuo et al., 1993), neural cell precursor adhesion (Tyas et al., 2003), and diencephalic patterning by modulating HH signalling (Caballero et al., 2014). This gene is functionally related to both FOXP2 and RUNX2 and is an effector of POU3F2, discussed above (see Benítez-Burraco and Boeckx 2015 for details).
- PAX7 is required for NC formation (Basch et al., 2006), and is actively involved in the development of craniofacial NC derivatives in vertebrates, along with multiple non NC-derived lineages (Murdoch et al., 2012; Monsoro-Burq, 2015). Along with its homologue PAX3, it specifically drives craniofacial skeleton and muscle development, its locus being associated with orofacial cleft in distinct populations (Böhmer et al., 2013; Butali et al., 2014; Leslie et al., 2015 and 2016; Gowans et al., 2016). Also, PAX7 is involved in neural tube patterning and polarization, particularly in the visuomotor system (Thompson et al., 2007). Interestingly, PAX3 and-7 expression in the developing neural tube is controlled by RAB23 (Li et al., 2007), a partner of HH signalling, involved in craniofacial morphogenesis and mutated in Carpenter syndrome, a syndromic craniosynostosis (Jenkins et al., 2011). Indeed itacts in concert with SHH as a dorsoventral patterning gene, necessary for spinal cord and hindbrain morphogenesis (Luo et al., 2006).
7. SLIT/ROBO signalling SLIT factors are component of the Slit/Robo signalling, which contributes to define spatio-temporal trajectories in NCCs migration (Jia et al., 2005; Giovannone et al., 2012). Members of the SLIT like SLIT1 and SLIT2 play an important role in distinct developmental step of the forebrain, plus they are mutated in clinical conditions involving cognitive and language deficits (discussed in detail by Boeckx and Benítez-Burraco, 2014b). Using a terminology which is common among linguists, we believe that these genes pertain to a specific facet of our language-ready brain, namely, the externalization component of language. This allows us to communicate thoughts to our conspecifics, whereas the gene network clustered around RUNX2 pertains to the core component of our language-readiness, to what linguists call the syntax-semantic interface. Several genes belonging to this group are found in our list of candidates: SLIT1, SLIT2 (mentioned above), ROBO1, ROBO2, and HES1 - ROBO1 and ROBO2 encode highly conserved transmembrane receptors that cooperate with SLIT factors. Mutations in ROBO genes have been linked to multiple human neurodevelopmental disorders. Noticeably, genome-wide analyses have identified both ROBO1 and ROBO2 as candidate loci for SZ risk (Potkin et al., 2009, 2010).
- HES1 encodes a transcription suppressor that interacts with the SLIT/ROBO signalling pathway during neurogenesis (Borrell et al., 2012). It is involved in the development of both GABAergic and dopaminergic neurons (Kameda et al., 2011; Long et al., 2013). HES1 is an important gene for the evolution of language, because of its specific interactions with ROBO1 and RUNX2 (Boeckx and Benítez-Burraco, 2014b). The expression of HES1 is reduced in patients carrying deletions encompassing the locus of EXOC6B, involved in retinoic acid (RA) metabolism, and presenting with intellectual disability, language delay, facial asymmetry, and vertebral and/or craniofacial abnormalities (Wen et al., 2013). Indeed, HES1 is regulated by RA, which acts as an inductor of neural differentiation, through SOX9 (Müller et al., 2009). Also other genes important for our language-readiness are related to RA signalling, including FOXP2 (Benítez-Burraco and Boeckx, 2015).
8. BMP/TGFβ signalling BMP superfamily signalling is implicated in almost all aspects of bone, cartilage and joint biology and is found altered in many human skeletal disorders (Salazar et al. 2016). Among our candidates we have found several genes belonging to this pathway: BMP2, BMP7 and NOG. Both BMP2 and BMP7 are core candidates for the globularizarion of the human skull/brain and the emergence of language-readiness (see Boeckx and Benítez-Burraco, 2014a for details).
- BMP2 is a well-known osteogenic member of the BMP/TGFβ pathway. Even though mutations in this gene have not been found to date in craniofacial phenotypes, significant and reproducible association of the BMP2 locus with sagittal craniosynostosis has been demonstrated in a genome-wide association study (Justice et al., 2012). In addition, the BMP2-SMAD osteogenic cascade is overactive in cells and tissues isolated from craniosynostosis patients (Lattanzi et al., 2013). The upregulation of BMP signalling in the NC causes the abnormal ossification of cranial bones (Komatsu et al., 2013). Additionally, BMP2 plays relevant roles in GABAergic and dopaminergic neurogenesis (Shakèd et al., 2008), and in neural migration and patterning (Waite and Eng, 2003), supressing differentiation in the NC (Kleber et al., 2005).
- BMP7, expressed in the NC, also regulates osteogenesis (Cheng et al., 2003), and mediates skull and brain development (Yuge et al., 2011; Segklia et al., 2012). Mutations in this gene have been reported to cause developmental delay and learning disabilities (Wyatt et al., 2010).
- NOG encodes a signalling protein that inhibit BMP signalling in the extracellular compartment. It is released by the notochord, regulates neural tube and somite patterning (Marcelino et al., 2001), and is involved in dopaminergic neuron production (Chiba et al., 2008).
9. HH signalling HH signalling is required for cranial NC morphogenesis and chondrogenesis (Wada et al., 2005), and the correct interplay among different germ layer-derived structures involved in the formation of the different tissue of the head (Xavier et al., 2016). Interestingly, HH members and interactors are involved in the molecular pathogenesis of facial clefts and aglossia (Yuan et al., in press). Three members of this signalling pathway are found among our candidates: SHH, GLI3 and PTCH1.
- SHH is a well-defined member of the HH signalling pathway, leading the intense trafficking inside the primary cilium compartment. It promotes the survival of NC cells (Delloye-Bourgeois et al., 2014) and the development of multipotent NC progenitors (Calloni et al., 2007). Shh promotes the expression of the SZ-candidate disc1 in the zebrafish brain (Boyd et al., 2015) and defines the midbrain floor plate of the embryo, from which midbrain dopaminergic neurons, known to be functionally aberrant in SZ, originate (Joksimovic et al., 2009). SHH-directed neural patterning depends on SOX2 input (Oosterveen et al., 2012; Oosterveen et al., 2013; Peterson et al., 2012), whereas SHH modulates the FGF8/WNT signalling source in the forebrain (Kobayashi et al. 2010; Rash and Grove, 2011).
- GLI3 encodes a key mediator of HH signalling in vertebrates, acting as a repressor in dorsal brain regions (Haddad-Tóvolli et al., 2012). It controls cortical size by regulating the primary cilium-dependent neuronal migration (Wilson et al., 2012). GLI3 is involved in the genetic aetiology of congenital anomalies, including syndromic craniosynostosis with cognitive impairment (McDonald-McGinn et al., 2010; Lattanzi, 2016). Mice carrying a null Gli3 allele feature an abnormal skull morphology due to the abnormal development of the NC, also involving FGF signalling (Veistinen et al., 2012; Tabler et al., 2016). Gli3 regulates calvarial suture development by controlling Bmp signalling, which integrates a Dlx5/Runx2 cascade (Tanimoto et al., 2012). Accordingly, differential Gli-binding affinity together with Sox2 input have been proven to provide context and positional information in Shh-directed neural patterning (Oosterveen et al., 2012 and 2013; Peterson et al., 2012). As discussed by Boeckx and Benítez-Burraco (2015), GLI3 is expected to have played a significant role in the anatomical and physiological events leading to globularization. Nearly 98% of Altaic Neanderthals and Denisovans gained a non-synonymous change in GLI3 that is described as mildly disruptive (Castellano et al., 2014).
- PTCH1 encodes the transmembrane receptor for HH in the primary cilium and is needed for NC-dependent orofacial development. Mutations in the gene have been associated to orofacial clefting (Metzis et al., 2013), but also to holoprosencephaly (Ming et al., 2002), a classic example of perturbed midline embryo development in which brain and craniofacial defects coexist. Mutations in the gene also cause Gorlin-Goltz syndrome, a condition involving macrocephaly and corpus callosum agenesis (Ponti et al., 2014). Together with GLI factors, PTCH1 regulates also SLIT genes expression (Liu et al., 2014).
10. RUNX2 The osteogenic master gene RUNX2 is a core candidate gene for the globularization of the human skull/brain (Boeckx and Benítez-Burraco, 2014b; Benítez-Burraco and Boeckx, 2015). This gene encodes a transcription factor playing a pivotal role in skull morphogenesis, as it represents the downstream effector of most signalling pathways (BMP/TGFβ, FGF, HH, eph-ephrin and WNT, among others), orchestrating craniofacial development (Lattanzi, 2016). It signals the bony specification of NC cells leading to the formation of the frontal and craniofacial primordium (Yoshida et al., 2008). In addition, RUNX2 is also involved in the development of brain structure, including thalamus and hippocampal GABAergic neurons (Pleasure et al., 2000; Benes et al., 2007; Reale et al., 2013). Molecular genetics aspects of this gene accounts for a dose-related effect it plays in the pathogenesis of craniofacial disorders: RUNX2 deficiency leads to reduced skull ossification found in cleidocranial dysplasia, while increased RUNX2 copy number, due to chromosome 6p21.1 triplication or quadruplication, is associated to syndromic craniosynostosis (Lattanzi, 2016). Importantly, a selective sweep in RUNX2 occurred after our split from Neanderthals (Green et al., 2010). RUNX2 interacts with SOX2 (reviewed above), and it is also predicted to bind TCOF1 (a core candidate for the “domestication syndrome”) based on ChIP analyses (Young et al., 2007). Several other core candidates for domestication (MTIF, GDNF, and RET) show expression changes in cell lines after RUNX2 overexpression (Kuhlwilm et al., 2013). RUNX2 is involved in protein-protein interactions with members of the FGF pathways, including FGF8 (via FGFR1), which is responsible for NC induction and the patterning of the pharyngeal region, and create a permissive environment for NCC migration (Trokovic et al., 2003; Komori, 2011; Stuhlmiller and García-Castro, 2012). In SZ patients, this gene was found differentially expressed in diverse brain areas: it is significantly downregulated in the hippocampus (Benes et al., 2007), whereas we found it upregulated in the temporal cortex (GSE53987 dataset; Figure 2; see manuscript text for details).
11. Additional genes Other genes in the selected list are not directly related to any of the above mentioned pathways, although they are involved in neural development. A brief mention to the functional profile could hence enable clarifying their relation to SZ molecular pathogenesis, to domestication, and to language evolution.
- GDNF is a core candidate for the domesticated phenotype in mammals (Wilkins et al. 2014). It encodes a neurospecific gene involved in the differentiation of dopaminergic neurons and in synaptogenesis (Christophersen et al., 2007; Ledda et al., 2007). GDNF levels are lower in SZ patients compared with healthy controls (Tunca et al., 2015), hence this gene is considered a plausible functional candidate for SZ. It has been also proposed as a positional candidate, as (AGG) repeats in the 3' UTR of the gene have been negatively associated to SZ, suggesting a putative protective function (Michelato et al., 2004). Nonetheless, no significant association with GDNF polymorphisms has been found in an independent SZ cohort (Williams et al., 2007). Instead, polymorphisms in the GDNF receptor GFRA3 gene have been associated to the disease (Souza et al., 2010).
- RET is another of Wilkins et al’s candidates. Whole-genome copy-number-variation analyses have revealed that RET, which encodes a cadherin involved in NC development, is deleted in SZ patients (Glessner et al., 2010).
- CDC42 controls NC stem cell proliferation (Fuchs et al., 2009) and its inactivation gives rise to craniofacial defects (Liu et al., 2013a). Aberrant function of the CDC42 signalling pathway is thought to contribute to neuronal spine deficits in SZ. Accordingly the downregulation of CDC42 in the dorsolateral prefrontal cortex reduces dendritic spines of pyramidal cells (Datta et al., 2015). CDC42 polymorphisms, reducing the expression of the gene, have been found associated to increased SZ risk (Gilks et al., 2012). This gene is also dysregulated in the hippocampus of SZ patients (Franke et al., 2012). Interestingly, two components of the CDC42 signalling pathway are also altered in SZ: CDC42BPB (Narayan et al., 2008), which is also a target of FOXP2 (Spiteri et al., 2007), and CDC42EP4 (Datta et al., 2015), which is hypermethylated in AMHs compared to Denisovans (Gokhman et al., 2014). CDC42 shows functional connections with core candidates for globularization and the externalization of language, including FOXP2, RUNX2, SLIT2, and ROBO1, and with genes related to language disorders, like CMIP, and with other genes known to have changed after our split from Neanderthals and Denisovans, like ITGB4, ARHGAP32, and ANAPC10, as already discussed elsewhere (Boeckx and Benítez-Burraco, 2014b).
- CTNNB1 is a core component of Wnt/β-catenin signalling pathway that paces aspects of NC development, from NC induction, to lineage decisions, to differentiation (Hari et al., 2012). Mutations in this gene have been associated to SZ (Levchenko et al., 2015). Interestingly, CTNNB1 interacts with PCDH11X/Y, a gene pair also related to SZ, language evolution, and language acquisition (Speevak and Farrell, 2011; Crow, 2013).
- KIT, one of Wilkins et al’s candidates for domestication, encodes a tyrosine kinase receptor involved in the regulation of NC-derived processes of hematopoiesis, melanogenesis, and gametogenesis (Rothschild et al., 2003; Kasamatsu et al., 2008). In rats, mutations of Kit affect hippocampal synaptic potentiation and spatial learning and memory (Katafuchi et al., 2000).
- MITF (another core candidate for domestication) encodes a transcription factor important for the differentiation and NC-derived melanocyte development and is indeed mutated in auditory-pigmentary syndromes categorized as neurocrstopathies (e.g. Waardenburg syndrome) (Hershey et al., 2005). MITF is believed to be a target of RUNX2, being modulated upon RUNX2 overexpression in cells (Kulhwilm et al., 2013).
- CITED2 is involved in NC function, leading to the establishment of left-right axis, through interactions with BMP signalling and Nodal, reviewed above (Preis et al., 2006; Lopes Floro et al., 2011). It is also involved in craniofacial development (Bhattacherjee et al., 2009). From the evolutionary standpoint, 99% of AMHs bear an intergenic change near CITED2 compared to Altai Neanderthals and Denisovans, which is reported as highly disruptive (Prüfer et al., 2014). This gene is highlighted as important for the evolution of our language-readiness (Boeckx and Benítez-Burraco 2014a,b), based on its functional connections with FOXP2 and RUNX2 (Vernes et al., 2011; Nelson et al., 2013).
- VCAN encodes a protein that regulates the migration of NC cells (Dutt et al., 2006). The human gene shows a fixed aminoacidic change compared to the Neanderthal protein (Pääbo, 2014). It regulates neuronal attachment, neurite outgrowth, and synaptic function of hippocampal neurons, via EGFR-dependent and -independent signalling pathway(s) (Xiang et al., 2006). In turn, SNPs in EGFR have been found significantly associated with SZ (Benzel et al., 2007).
- NCAM1 encodes a protein involved in dendritic and axonal growth, synaptic plasticity, and cognition, and is a target of FOXP2 (Konopka et al. 2009) and RUNX2 (Kuhlwilm et al. 2013). The gene has been related to working memory performance in the healthy population (Bisaz et al. 2013), but also to SZ (Vawter et al. 2001; Atz et al. 2007). NCAM1 interacts with VCAM1, which shows a fixed change (D414G) in AMHs compared to Neanderthals/Denisovans (Pääbo, 2014, Table S1). Interestingly, VCAM1 is upregulated by CLOCK (Gao et al. 2014), a circadian clock gene also associated to SZ (Zhang et al. 2011; Jung et al. 2014).
- AXIN2 is expressed in the cranial NC and plays a key role in NC-derived frontal bone development and patterning (Yu et al., 2005; Li et al., 2015). It is a negative regulator of canonical Wnt pathway and contributes to the stability of CTNNB1 (Li et al., 2015). Interestingly, AXIN2 mutations causing non-syndromic oligodontia, resulting in speech alterations (Liu et al., 2015).
- POLR1A encodes a subunit of a RNA polymerase, which plays a role in the regulation of NC-derived skeletal precursor cells (Weaver et al., 2015). The gene is mutated in acrofacial dysostosis (Cincinnati type, OMIM#616462) involving microcephaly.
- SIX2 is involved in NC proliferation and migration during brain development, but also in the morphogenesis of the frontal craniofacial skeleton, interacting with BMP signalling in mice (Garcez et al., 2014). Mutations in this gene have been found in syndromic craniosynostosis (Hufnagel et al., 2016).
- ZIC1 encodes a homeodomain transcription factor that regulates cell cycle and cell migration acting on HH and on FGF signalling, contributing to NC development and, specifically, to craniofacial development (Milet et al., 2013; Plouhinec et al., 2014). It also participate in neurogenesis, left–right axis formation, myogenesis and skeletal patterning (Aruga et al., 2006; Merzdorf, 2007). Heterozygous ZIC1 mutations give rise to severe coronal synostosis with learning disabilities (Twigg et al., 2015).