3.3.1. Spinocerebellar Ataxia Type 7

Spinocerebellar ataxia type 7 (SCA7) is an autosomal-dominant neurodegenerative syndrome that outcomes from polyglutamine expansion of ataxin 7 (ATXN7). Remarkably, although ATXN7 has a widespread expression in SCA7 patients, the pathology primarily hits the cerebellum and the retina [124]. A recently published paper suggested that RELN might be a formerly unidentified factor accountable for the tissue specificity of SCA7 [125].

#### 3.3.2. Autism and the Heterozygous Reeler Mouse

The disorders of the autistic spectrum (ASD), which are characterized by social, behavioral, and language insufficiencies, comprise Asperger syndrome, autism, and pervasive developmental disorder-not otherwise specified (PDD-NOS). Less than 20% of these disorders, acknowledged as "syndromic autism", derives from monogenetic diseases, most commonly fragile X syndrome and tuberous sclerosis. The remaining 80% of ASD cases are classified as "non-syndromic autism" and are widely investigated to find candidate genes that may contribute to pathology [126].

#### Genetics

At present autism cannot be considered, strictly speaking, a genetic disease, as one or more causative gene(s) has (have) not been found yet. The first gene association study implicating *RELN* in autism dates to 2001 [127]. However, subsequent gene population surveys yielded contrasting results [128–131]. Nonetheless, a more recent meta-analysis showed that at least one single nucleotide polymorphism (SNP) in *RELN* could be significantly associated with the risk of autism [132]. Therefore, results of SNP analysis appear to be compatible with the idea that heterozygous mutations in *RELN* may contribute to the onset of the disorder. Genetic studies on autism led to two main outcomes: 1. the more predominant existence of rare or de novo inherited mutations of a number of genes in autistic patients; 2. the discovery of certain common gene variants that contribute to the risk of autism but are also present, albeit at lower frequency, in the normal population [133]. When more than two de novo mutations occur in a gene, the latter becomes a very likely causative candidate of a disorder. There are four unique documented de novo mutations of *RELN* associated with autism [134–136], thus implicating *RELN* as a possible cause of ASD. However, although nonsense mutations are more frequent in autistic patients than in controls after whole-exome sequencing, there is not a striking gross increase of de novo mutations in the former [135]. To study autism heritability, one can also employ a different approach that distinguishes total narrow-sense heritability from that due to common gene variants. By this method, it was concluded that narrow-sense heritability of autism is ∼52.4%, and that the main contribution heritability was due to common gene variants, whereas rare de novo mutations contribute only for about 2.6% of cases, but substantially influence individual liability to the condition [137]. Thus, *RELN* may primarily have a role in the individual *predisposition* to manifest autism rather than being one of the contributory causes of the disorder.

Further support for a RELN involvement in autism derived from the detection of reduced expression of the *RELN* transcript and protein in autistic individuals. Decreased RELN levels were apparent in the superior frontal cortex [10] and cerebellum of autistics as compared to controls [10,11,138]. In these areas, the levels of *RELN* mRNA were lower, as was the *DAB1* transcript, whereas *VLDLR* mRNA levels augmented.

#### Imaging

Imaging findings in autism have been recently reviewed [139]. Numerous observations join to prove that there is an atypical development of the brain in autistic children. Early cross-sectional studies demonstrated that the brain of these children had a higher volume than that of regularly developing subjects. However, growth curves in the two groups eventually met at later childhood. More specifically, in the 6–35-year interval, there was an initial period of brain overgrowth, and then growth slowed down or even stopped during early and late infancy to which a phase of fast reduction of the brain volume eventually followed [140]. Neuroimaging data also indicated that differences in the brain of autistic people started to be detectable within the first two years after birth, *before* clinical symptoms became obvious. There are conflicting views about the probability that an accelerated growth rate of the brain in this postnatal window goes together with the occurrence of early neurodevelopmental perturbations [139]. In relation to this, it is relevant that we still do not know when the initial neuropathological signs of autism occur, also from the paucity of studies on autistic children during the first year of life.

The mechanisms at the basis of the abnormal growth of the autistic brain are also unclear. Although most imaging studies have focused onto the gray matter of the cerebral cortex, there are data indicating that an increased amount of cerebrospinal fluid in the subarachnoid space [141] and/or a greater volume of the white matter [142] occurred in parallel to the enlargement of the autistic brain. As regarding the cerebral cortex, its surface, but not thickness, increased in the autistic brain [143].

To summarize, that an initial brain overgrowth may be a reliable biomarker for autism remains highly questionable. Thus, it seems more profitable to focus onto regional brain structural differences in a more effective search for new neuroanatomical findings of clinical relevance [139].

Before entering the description of regional MRI investigations in autism, it is important to stress that, at present, there are no specific and/or causative objective findings for the condition, but, instead, the very same regions altered in autism may be interested in other psychiatric conditions.

The individual constituents of the neural circuitries causal of ASD are well defined. They include regions of the fronto-temporal, fronto-parietal, and dorsolateral prefrontal cortex (PFC); parts of the limbic system; the fronto-striatal circuitry, and the cerebellum. Neuroimaging studies on these regions have employed different approaches such as the definition of a region-of-interest (ROI), voxel- or vertex-vise methods. Traditional ROI studies have reported atypical findings in brain areas that participate to social cognition such as the medial PFC, the anterior cingulate cortex (ACC), the inferior frontal cortex, the superior temporal sulcus, the amygdala, and the anterior insula.

The cerebellum was larger than in controls in several MRI studies on autistic patients older than three years [144], but not in younger children [143]. Differently from the cerebellum, the size of the vermis was smaller [145–147] or larger [146] or did not display any relevant difference [147], and such discrepancies possibly depend from the different clinical presentations of the condition [147,148]. It is also unclear whether there are differences in size of individual vermal lobules, as claimed by some authors [145], but not others [147]. Similarly, there were no differences between cerebellar hemispheres in one study [147], whereas another group has found the hemispheric size as the only significant structural dissimilarity between verbal and nonverbal subjects [149].

The still fuzzy picture emerging from the imaging studies onto the autistic brain makes it very difficult to compare the human and mouse data in the search for common biomarkers. To our knowledge, there are only two MRI studies on the brain of the heterozygous *Reeler* mouse. In the first, Badea and co-worker [29] reported that the total volume of the brain, the ventricular volume and the hippocampal volume correspondingly raised of about 6%, 82%, and 7% compared to normal control mice. However, after statistical analysis, they showed that these volumes were like those of *reln*+/+ normal mice. They also measured the areas of different parts of the brain in comparison with wild-type mice and found no differences in hippocampus and cerebellum, but an enlargement of the lateral ventricles. A more recent paper confirmed the ventricular enlargement, but found a reduction of the cerebellar volume, whereas the volume of the motor cortex as well as its thickness was unchanged [150]. Therefore, given the paucity of data in mouse and the still unclear MRI pattern in human autistic subjects, one can only conclude that, at present, the cerebellum could be a part of the brain deserving further imaging investigations for translational purposes.

## Histopathology

A series of histological alterations of the whole brain occur in the autistic brain. In the first histological surveys, the only cortical area showing qualitative structural abnormalities was the ACC that, in autistic patients, lacked architectural refinement and had only a coarse lamination [151]. However, in the following decades substantial amounts of data have been collected and the list of cerebral structures displaying histopathological changes in ASD has grown substantially to include a series of cortical regions, the amygdala, the cerebellum and the brainstem, see e.g., [152,153]. Below we will briefly summarize the most significant histopathological findings in human patients and compare them to those in the heterozygous *Reeler* mouse. However, the interpretation of both human and mouse finding needs often much caution, because not all studies used sound quantitative approaches and/or proper stereological procedures.

#### a) Changes Affecting the Whole Brain

The diffuse alterations observed in the brains of autistic subjects at postmortem included cortical dysplasia and neuronal heterotopia, with the formation of aggregates of neuronal cell bodies in anomalous positions [154]. Other alterations, i.e., differences in size of the neuronal nucleus and perikaryon, occurred at the cytological level. These differences started being evident in young children and became more apparent in adults, but then tended to re-equilibrate with time [155]. Remarkably, there might be some compensations between different areas, as in some parts of the brain neurons were bigger, but smaller in others.

In the autistic brain there was also an increase of the neuropil extension in certain but not all cortical areas that have been investigated so far [156]. It is unclear which neuropil component(s) is (are) responsible of these volumetric variations. Fewer dendrites were, in fact, immunostained for microtubule-associated protein 2 (MAP2) in the PFC [157] and a reduction of dendritic spines was reported in hippocampus [158], but other studies reached completely opposite conclusions after examination of the pyramidal neurons from layers 2 and 5 in the frontal, temporal, and parietal cortex [159]. The issue of dendrite and dendritic spines density is quite important in the general framework of this discussion, because these parameters have been widely investigated, primarily aiming to validate the heterozygous *Reeler* mouse as a translational model.

Alterations in neuronal differentiation and migration may also occur in the autistic nervous system and, thus, the consequences of a dysregulation of these processes may be at the basis of whole brain changes in autism [154]. In spite of this, there are only a few investigations on the expression of RELN in the brain of autistic patients and, after quantitative analysis, there was no alteration in the density of layer 1 RELN+ neurons in the superior temporal lobe of the autistic brain, although these neurons represent about 70% of the total layer 1 population [160].

#### b) Brain Regional Changes

#### Forebrain

Most of the histopathological observations on the brain of autistic patients focused on the cerebral cortex and hippocampus and, broadly speaking, alterations were almost exclusively restricted to neurons. The parameters considered have been size, number, and density of the different neuronal populations, often in relation to the cortical layers or the hippocampal subfields. A point of attention in considering these studies is that, in several cases, comparative brain volume evaluations between autistic patients and controls are missing, while they are, instead, essential to settle whether modifications in cell density reflect true differences in total cell counts.

#### Cerebral Cortex

The autistic pathology affects several regions of the cerebral cortex.

The PFC, which plays a major role in cognitive control, displayed a general overgrowth with an increase in the number of neurons, whereas glial cells were apparently unaffected. Among the GABAergic interneurons, there was a numerical increase of the PV-immunoreactive chandelier neurons, whereas the calbindin- and calretinin-expressing neurons were unaltered [153,161–163]. However, after qRT-PCR, the levels of the RELN and GAD67 mRNAs diminished in the post-mortem PFC from autistic patients in comparison to healthy controls [164].

In the inferior frontal cortex, changes affected the small-sized pyramidal neurons that did not display numerical alterations but were of smaller size [163].

The fusiform gyrus, which intervenes in facial recognition and social interactions, had a reduced neuronal density in layer 3, whereas neurons in layers 2, 5, and 6 were less numerous, being also smaller in layers 5 and 6 [165], but these alterations were not confirmed in [166].

In the frontoinsular cortex and ACC, both intervening in emotional regulation and self and others awareness, lamination was rudimental. In the former, von Economo neurons (VENs) of layer 5 increased in number [167–169], whereas in ACC neuronal density augmented in layers 1–2 of area 24a of the left hemisphere but diminished in layers 5–6 of area 24c; size, instead, diminished in all layers of area 24b [151,170].

The anterior midcingulate cortex also displayed a numerical increase of VENs, as well as of the pyramidal neurons of layer 5 that, however, were of smaller size [171].

Lastly, the entorhinal cortex, which has a role in memory, navigation and perception of time, displayed characteristic terminal swellings, referred to as spheroids [172] that were also observed in all hippocampal subfields (see below).

Another issue of interest is the possibility that there are alterations in the minicolumnar organization of the cerebral cortex in early age onset autism [156], as it might be the case in *Reeler* mice. Specifically, it appeared that minicolumns were smaller, more numerous and with lower neuronal density in several cortical areas of autistic (and Asperger's syndrome) patients, although these observations still need to be confirmed in full [156]. Under this perspective, it may be useful to here recall some of the results on the localization of the neural cell adhesion molecule 2 (NCAM2) in the *Reeler* mouse because the molecule has also been proposed as a predisposition gene for the development of autism [173]. In mutants, NCAM2 immunopositive and negative patches formed a mosaic filled with dendritic aggregates originating from two different populations of neurons in a fashion suggestive of a minicolumnar organization [174,175]. However, one must consider these findings with much caution, as that minicolumns are indeed the fundamental modular units of neocortical organization is currently still a matter of debate, see e.g., [176] for review.

Whereas the human cerebral cortex has been widely investigated in autism, investigations on the cerebral cortex of the heterozygous *Reeler* mouse have been relatively few but reported a reduction in the levels of GAD67 [53,177] like that observed in humans. Since the studies on hippocampus led to highly comparable results in the two species, it would be of importance to undertake rigorous investigations on the number and density of the cortical neurons in *reln*+/<sup>−</sup> mice, with attention to the different neurochemical populations of inhibitory interneurons. The results of these studies will be relevant, from one side, to validate the mouse model and, from the other, to confirm some numerical observations in humans that, as mentioned at beginning of this section, need validation using approaches more reliable than those often employed at histopathology.

#### Hippocampus and Amygdala

In humans, beside the widespread occurrence of spheroids, other general changes in hippocampus [151,158,172,178,179] consisted in a reduction of neuronal size and dendritic arbors, and these smaller neurons appeared to be more densely packed. There were also a series of specific modifications affecting the excitatory pyramidal neurons that were more numerous in CA1, but less abundant in all other adjacent hippocampal regions. The GABAergic inhibitory interneurons, instead, displayed a higher density, specifically the calbindin-immunoreactive neurons in the dentate gyrus, the parvalbumin-immunoreactive neurons in CA1 and CA3, and the calretinin-immunoreactive neurons in CA1 [151,158,172,178,179].

In the heterozygous *Reeler* mouse the hippocampus displayed reduced levels of GAD67, [53,177] that could be somewhat restored after stereotaxic injections of Reln [67,180]. These experiments indicated the existence of a causal link between the decrease in GAD67 expression and Reln haplodeficiency. In keeping with such a possibility, Reln supplementation could, at least partly, reverse such a decrease. Other experiments, in line with this interpretation, have confirmed that the decrease in the levels of GAD67 in heterozygous mice can be overturned, e.g., after administration of nicotine, which reduces the GAD67 promoter methylation and increases its transcription [177].

In heterozygous mice, pyramidal neurons displayed a reduction in the average length and width of their apical and basal dendritic spines [52], consistent with the decrease in the spread of the dendritic arbor of the same population of neurons in humans. Additionally, Reln supplementation was effective in promoting a full (apical) or partial (basal shaft) spine recovery [180]. These morphological observations are in line with a previous report showing that, in the forebrain, spines were hypertrophic in mice conditionally overexpressing Reln [66]. At electrophysiological recordings CA1 pyramidal neurons displayed reduced spontaneous inhibitory postsynaptic currents [181], an observation that was fully coherent with the reduction of the inhibitory input from the GABAergic interneurons observed histologically.

Synaptic plasticity is fundamental for hippocampal function. In CA1 of heterozygous mice, LTP was impaired [182] as well as long-term depression (LTD) [181], which returned to normal levels after the administration of Reln [180]. Additionally, in *reln*+/<sup>−</sup> and *reln*−/<sup>−</sup> mice, post tetanic potentiation (PTP), a form of short-term plasticity that depends on neurotransmitter release, was reduced in CA1 [180,183] and could be reversed by Reln [180].

Collectively these data indicate that the experimental administration of the glycoprotein was able to reverse the morphological, neurochemical, and physiological hippocampal deficits consequent to a reduction of brain Reln. Translationally, they are very important because they offer some cues for further investigations onto the autistic brain. It would be of interest to map the post-mortem distribution of hippocampal RELN in patients compared to healthy controls, to ascertain whether the pattern of immunoreactivity will be consistent between humans and mice.

In the autistic patients, several studies reported that the amygdala, which is involved in emotional learning, increased in size and displayed an augmented density of neurons within the medial, central, and cortical nuclei [151,184–187]. Neurons were, however, less numerous, although numerical variations could be age dependent. To our knowledge, Boyle et al have investigated in detail the amygdala of the homozygous *Reeler* mouse with a marker-based phenotypic approach [188], but there are no data on heterozygous mice.

#### Cerebellum

Analysis of cerebellar alterations in autism has attracted many efforts of the basic researchers and clinicians. The most consistent anatomic findings in autistic patients were a reduction in size of certain lobules of the cerebellar vermis (but see 3.3.2b Imaging) and a decrease in the number [186,189–192] and size [193,194] of the Purkinje neurons. The inhibitory GABAergic basket and stellate interneurons that connect with these neurons did not show quantitative differences compared to normal cerebella [195]. This observation is indicative of a late developmental death of the Purkinje neurons, as they differentiate well before the interneurons. In addition to structural observations, Western blots demonstrated a reduction of about 40% in the level of expression of RELN in autistic patients related to age and sex corresponding controls [11]. Quantitative RT-PCR also showed a drop of RELN and GAD67 mRNAs in the post-mortem autistic brain [164].

Notably many of the histological alterations in the human autistic cerebellum are like those described in *reln*+/<sup>−</sup> mice. These animals displayed a progressive loss of Purkinje neurons already during the first weeks of life [35], and inferior numbers of these cells were observed in adult subjects as well [196]. Human MRI studies did not allow, at present, to ascertain whether the cerebellar vermis is hit by the pathology in its entirety or, rather, only at specific lobuli. Therefore, our group has, at first, focused its attention on five different lobules, which receive diverse types of afferent functional inputs, to analyze the number and topological organization of the Purkinje neurons in *reln*+/+ and *reln*+/<sup>−</sup> adult mice of both sexes [197]. We have thus shown that the Purkinje neurons: 1. Displayed a lower density in *reln*+/<sup>−</sup> males (14.37%) and *reln*+/<sup>−</sup> females (17.73%) compared to *reln*+/+ males; 2. Were larger in *reln*+/<sup>−</sup> males than in the other phenotypes under study, and smaller in females (regardless of the *reln* genetic background) than in *reln*+/+ males; 3. Were more messily arranged along the YZ axis of the vermis in *reln*+/<sup>−</sup> males than in *reln*+/+ males and, except in central lobule, *reln*+/<sup>−</sup> females.

Very recently, as many observations have associated a number of synapse-related genes in the genesis of autism and other neuropsychiatric conditions [198,199], we have examined the expression of synaptophysin 1 (SYP1) and contactin 6 (CNTN6) in the vermis of *reln*+/<sup>−</sup> and *reln*+/<sup>+</sup> adult mice of both genders [200].

SYP1 is a pre-synaptic marker and CNTN6 is a marker of the synapses made by the parallel fibers onto the Purkinje neurons' dendrites. Notably, there is evidence, although still to be validated in full, that SYP1 is involved in the structural alterations of the autistic synapses [189,201], and very recent observations have shown that copy number variations [202] or a truncating variant [203] of

*CNTN6* are found in autistic patients. In addition, *CNTN6* mutations may be a risk factor for several neurodevelopmental and neuropsychiatric disorders [150,199,204].

In line with these human studies, we have demonstrated that *reln*+/<sup>−</sup> mouse males displayed a statistically significant drop of 11.89% in SYP1 compared to sex-matched normal animals, whereas no modifications were detected comparing *reln*+/<sup>+</sup> and *reln*+/<sup>−</sup> females [200]. In *reln*+/<sup>−</sup> male and female mice, reductions in SYP1 levels were particularly evident in the molecular layer, whereas in heterozygous mice of both sexes a reduction in CNTN6 occurred in all the three cortical layers of the vermis. In addition, alterations in the levels of expression of SYP1 in the molecular layer of male *reln*+/<sup>−</sup> mice ensued across all lobules except lobule VII, but they were limited to lobule II for the granular layer and lobule VII for the Purkinje cell layer.

Thus, the widespread reduction of SYP1 and of CNTN6 in the molecular layer of *reln*+/<sup>−</sup> male mice well matched with the autistic phenotype in humans [150].

In the vermis (and the whole cerebellum), there is proof for a topographic segregation of the areas controlling motion versus those connected to cognitive and affective functions, and the diverse lobules are coupled with precise zones of the brain and spinal cord [205]. The CNS areas that handle sensorimotor inputs are directly or indirectly connected with the anterior lobe (lobules I–V of the vermis), lobule VIII, and, to a lesser grade, with lobule VI; on the contrary, cortical association areas that collect non-motor responses are linked to lobules VI and VII. Existing clinical data indicate that the vermis is the chief target of the limbic system, and physiological and behavioral observations implicate the vermis in the regulation of emotions [206]. Therefore, the neurochemical modifications of the cerebellar cortex in heterozygous mice are fully in line with the possibility that the social and communication aberrations typical of autism rest on anomalies of the limbic system and its connections [207,208].

At post-mortem, a numerical reduction of the Purkinje neurons in the posterior cerebellum was long ago described in autistic subjects [184,209], but it did not appear to disturb the vermis [189]. Hypoplasia in lobules VI and VII was initially detected in vivo using MRI [145], but subsequent observations proved the existence of two distinct autistic subtypes related to vermian hypoplasia or hyperplasia [146]. A systematic review and meta-analysis of the accounts of structural MRI has then established that the reduction in size of lobules VI–X (i.e., the lobules included in the posterior cerebellum) showed a remarkable heterogeneity that associated to differences in time of life and intelligence quotient (IQ) merely in lobules VI–VII [210]. Other observations showed that the posterior/inferior vermis, i.e., lobules VII, VIIIb (left), and IX, was more prone to pathological deviations [211], with a decrease of the gray matter after quantitative MRI [190,212,213]. Therefore, it appears that the cerebellar phenotype of the heterozygous *Reeler* mouse is fully compatible with that in humans and that a deeper structural and neurochemical characterization may be useful to direct the discovery of new biomarkers of translational interest.
