c) Cerebellum

In the cerebellum of schizophrenic patients, there was a loss of distal and terminal dendritic branches and a decrease in density of the dendritic spines of the Purkinje neurons [236]. Again, these histological alterations are not specific for the condition, being evident also in autism. In addition to such changes, in the schizophrenic cerebellum there were altered levels of expression of the general presynaptic marker SYP1, of complexin II, a marker of the excitatory synapses, but not of complexin I that, instead, labels the inhibitory synapses [237]. Thus, some of the structural changes described in the heterozygous *Reeler* mouse in relation to autism also in cerebellum recollected the human schizophrenic phenotype.

## **4. Does the Behavior of Heterozygous** *Reeler* **Mice Recall the Human Conditions Related to RELN?**

The recapitulation of the behavioral modifications typical of human autism, schizophrenia, or epilepsy in heterozygous *Reeler* mice still is a subject of debate. The dissimilar outcome of behavioral experiments performed in different laboratories is not surprising, because neuropsychiatric behaviors in humans primarily regard social interaction, communication, and restricted interest, and these behaviors are, obviously, very difficult to measure objectively in mice [238].

It is perhaps worth mentioning here that most of our knowledge on the effects of Reln in the cognitive or behavioral field derives from work on mouse hippocampus. This is not surprising as this part of the brain, as discussed previously, has been the primary focus of numerous investigations also in human patients affected by neuropsychiatric disturbs. Several behaviors comparable to those observed in these human disturbs also occur in *reln*+/<sup>−</sup> mice [239–241], as well as the deficits in reversal learning after visual discrimination tasks that were hypothesized to follow a diminished visual attention [240]. In addition, testing *reln*+/<sup>−</sup> mice for anxiety-related behavior, motor impulsivity and morphine-induced analgesia yielded a different behavioral profile from that of wild-type littermates in that they displayed, starting form adolescence, a decreased inhibition and emotionality. To these modifications, a small increase of impulsive behavior and different pain thresholds also occurred in adult mice [242]. Heterozygous mice were also tested in a complex series of PPI protocols (unimodal and cross-modal) to conclude that they exhibited a multifaceted configuration of changes in startle reactivity and sensorimotor gating, with both resemblances to and dissimilarities from schizophrenia [243]. At least partly in line with these latter observations, other studies failed, incompletely or in full, to validate the behavioral analogies between neuropsychiatric patients and *reln*+/<sup>−</sup> mice [181,244–248]. For example, Salinger and co-workers were unsuccessful to find differences between *reln*+/<sup>−</sup> and *reln*+/+ mice after testing gait, emotionality, social aggression, spatial working memory, novel-object detection, fear conditioning, and sensorimotor reflex modulation [244]. In another survey [246], heterozygous *Reeler* mice were evaluated for cognitive plasticity in an instrumental reversal learning task, impulsivity in an inhibitory control task, attentional function in a three-choice serial reaction time task, and working memory in a delayed matching-to-position task to conclude that there were no differences in comparison to *reln*+/+ littermates in prefrontal-related cognitive trials. However, *reln*+/<sup>−</sup> mice were deficient in two operant tasks. From these observations, the authors concluded that heterozygous *Reeler* mice were *not* a good model for the essential prefrontal-dependent cognitive shortfalls detected in schizophrenia, although they could be useful to model learning deficits in a more general sense.

In another paper it was reported that heterozygous and wild-type mice displayed comparable levels of general activity, coordination, thermal nociception, startle responses, anxiety-like behavior, shock threshold; identical cued freezing behavior, and comparable spatial learning in Morris water maze tasks, albeit a significant decrease in contextual fear conditioned learning was observed in *reln*+/<sup>−</sup> mice only [181]. These authors have then hypothesized that the pharmacological administration of Reln in heterozygous mice could restore the response to PPI. They were unable to find differences in the acoustic startle reflex among treated and untreated animals, but Reln-treated *reln*+/<sup>−</sup> mice showed a substantial increase in the percent inhibition to 78-, 86- and 90-dB pre-pulse [180].

One study has specifically focused onto the *reln*+/<sup>−</sup> mouse behavioral phenotype in young (P50–70) and fully adult (older than P75) animals to conclude that they were not useful to model schizophrenia [245]. An ample series of behavioral test was used (Irwin test; rotarod; spontaneous locomotor activity; social behavior; light-dark transition; startle response and pre-pulse inhibition; hot plate). Heterozygous mice were like their wild-type littermates at either age, though completely adult male *reln*+/<sup>−</sup> mice were involved in social exploration for a longer time. In addition, performance on the rotarod deteriorated with age.

Indeed, age appeared to be a further issue of complexity. In fact, adult *reln*+/<sup>−</sup> mice did not display discernible changes in activity, motor coordination, anxiety, or environmental perception compared to wild-type littermate controls. However, juvenile animals displayed not as much of anxiety- and risk assessment-related behaviors in the elevated plus-maze [182,241]. In addition, in one of these two studies it was demonstrated that young *reln*+/<sup>−</sup> mice had a hippocampal-dependent shortfall in associative learning and impulsivity–anxiety-related behavior [182]. Additionally, one study, starting from the clinical observations that reported the occurrence of vocal and motor anomalies in autistic patients, has described that *reln*+/<sup>−</sup> mice had a general delay in the development of their repertoire of neonatal vocal and motor behaviors [249].

Finally, one must consider that gender apparently influenced some behaviors, although very few studies have focused on this issue. Among these studies, young heterozygous female mice were described to be more active in the light/dark transition test than the heterozygous males that were, instead, more aggressive than females during social interaction [241].

#### **5. Usefulness of the** *Reeler* **Mouse in Translational Studies: Concluding Remarks**

The analysis of the literature discussed above requires one trying to draw some conclusions about the true usefulness of the *Reeler* mouse in translational studies.

At first, it may perhaps be useful to remember that, as discussed, *RELN* is causative of LIS2 and a small percentage of ADLTE, whereas only tentative associations up to now hold for the other conditions here considered (see Figure 1 and Table 1).

Remarkably, both LIS2 and ADLTE are rare diseases. Very few cases of LIS2 (around ten) so far come about in the literature (see OMIM #257320). Similarly, patients with lateral temporal epilepsy (LTE) are only about 10% of all temporal epilepsies, and the real prevalence of ADLTE, which has been up to now reported in Europe, USA, and Japan, is unknown, but it may account for about 19% of familial idiopathic focal epilepsies [250,251]. When one considers the human conditions related to the Reln signaling pathway, one still encounters a group of rare diseases. The actual frequency of *VLDLR*-associated cerebellar hypoplasia is unknown, but initial reports regarded not more than twenty-five affected individuals in Canada and USA [113,252], although the condition occurs worldwide, *PAFAH1B1*-associated lissencephaly is very rare as the prevalence of classic lissencephaly ranges from 11.7 to 40 per million births [120]. To date, sixty-six affected individuals and seven asymptomatic individuals with the ATTTC repeat insertion within *DAB1* have been reported in ten relatives from the south of the Iberian Peninsula, and no individuals with SCA37 from other geographic areas have been described [116]. SCA7 has a prevalence of less than 1:100,000 and accounts for about 2% of all SCAs [253].

Therefore, one must deal with the paradox of the relatively little interest for translational studies on the very conditions for which the *Reeler* mouse and/or mice with mutations of the genes of the Reln pathway fully meet the criteria for construct and face validity. Thus, the homozygous Reeler mice appear to be more interesting to the neurobiologist than to the clinician and their study will surely be still rewarding in terms of our comprehension of neurodevelopment, as the model already helped to establish that many functional and circuit features of cortical neurons are relatively independent from positional cues and cortical lamination, e.g., [179].

Very differently, the prevalence of autism in the worldwide population is around 1% [254] and that of schizophrenia is just below [221]. As the two conditions are very diffuse in the human population, there is an obvious translational interest for the heterozygous *Reeler* mouse as a model for the two disorders. However, such an interest is again paradoxical, as the validity of these mice to extract information about the human pathologies remains dubious. The first explanation for this uncertainty lies, beyond any doubt, in the substantial lack of construct validity, which is the direct consequence of the complex genetic background of autism and schizophrenia. As regarding face validity, the present survey of the literature clearly points out that there are several similarities but also dissimilarities between the human and the mouse phenotypes. Among dissimilarities, one must consider the heterogeneity of results of the behavioral experiments in mouse. Further complexity derives by the vast array of clinical symptoms in humans. Structurally, most of the imaging and post-mortem findings in humans are not specific for each of the two conditions. However, one must consider that both the human and mouse phenotype converges to indicate the cerebral cortex, hippocampus, and cerebellum as the primary foci of the pathologies and the inhibitory interneurons as major players in the context of the circuitry involved.

A serious drawback to a full validation of the heterozygous *Reeler* mouse as a model of autism and/or schizophrenia lies in the observation that the alterations so far described in mouse are very subtle in both structural, functional and neurochemical terms. The relatively low resolution of current neuroimaging procedures, and the difficulty to obtain post-mortem samples amenable for neurochemical, electrophysiological, and fine (ultra) structural analyses make it very difficult to establish whether the alterations in heterozygous *Reeler* mice have a true biological significance that goes beyond mere statistics [255,256]. In the affirmative, one could take advantage of these alterations to discover novel biomarkers that will be helpful for an earlier and more precise diagnosis in the human practice.

**Author Contributions:** All authors contributed to write the manuscript.

**Funding:** The original research work carried out at the Department of Veterinary Sciences described in this paper was funded by Local Grants of the University of Turin to LL and AM and FFABR 2017 from the Italian MIUR to CC. AG was in receipt of a Local Grant of the Catholic University of the Sacred Heart, Milano.

**Conflicts of Interest:** The authors declare no conflict of interest.
