**3.** *RELN***-Related Human Neurological Conditions and Their Mouse Counterparts**

Several human neurological conditions have a direct or indirect link with *RELN* and its encoded protein, as well as with the components of the RELN signaling pathway (Figure 1 and Table 1). We will briefly describe these conditions below, aiming to put in the better perspective those features that may be useful for well understanding the translational relevance of the *Reeler* mouse.

**Figure 1.** Summary of the most relevant human pathologies modeled in the *Reeler* mouse. The monogenic conditions provoked by the *RELN* gene, i.e., ADLTE and LIS2, are in red, those related to genes encoding for the proteins of the Reln intracellular cascade or only tentatively linked to RELN are indicated in blue. Autism and schizophrenia, which have a complex multifactorial etiology, are in black with an interrogative mark to underline the still tentative association of the two disorders with *RELN*. Abbreviations: ADLTE autosomal-dominant lateral temporal epilepsy, LIS2 lissencephaly 2, PAFAH1B1 platelet-activating factor acetyl hydrolase IB subunit α, *RELN* reelin gene (human), *reln* reelin gene (mouse), SCA37 spinocerebellar ataxia type 37, SCA7 spinocerebellar ataxia type 7, VLDLR very low-density lipoprotein receptor.



Note that only LIS2 and autosomal-dominant lateral temporal epilepsy (ADLTE) have a demonstrated link with *RELN*. *RELN* may be relevant for *LIS1*.

#### *3.1. Neurological Conditions Caused by RELN Mutations*

Several diseases are based on mutations of *RELN* or of genes encoding for proteins associated with the RELN signaling pathways. Among these, lissencephaly 2 (LIS2) and autosomal-dominant lateral temporal epilepsy (ADLTE) are of relevance to the present discussion as they have a clear genetic link with *RELN*.

#### 3.1.1. Human Lissencephalies and the Homozygous *Reeler* Mouse

Human lissencephalies are a group of cortical malformations that are consequent to neuronal migration disorders. Broadly speaking, the structural phenotype in lissencephalies ranges from a thickened cortex and complete absence of sulci (agyria) to a thickened cortex with a few, shallow sulci (pachygyria) [23]. The main feature of classic lissencephaly, formerly referred to as type I lissencephaly but today named lissencephaly 1 (LIS1), is a marked thickening of the cerebral cortex with a posterior to anterior grade of severity. An anomalous neuronal migration in the interval between the ninth to the thirteenth week of pregnancy causes LIS1, resulting in an assortment of agyria, mixed agyria/pachygyria, and pachygyria. An abnormally thick and ill ordered cortex with four highly disorganized layers, diffuse neuronal heterotopia, enlarged cerebral ventricles of anomalous shape, and, often, hypoplasia of the corpus callosum are typical of LIS1 [24]. The basal ganglia are normal, except that the anterior limb of the internal capsule is usually not noticeable, and, most often, the cerebellum is normal as well.

Lissencephalies are now classified based on brain imaging results and molecular investigation [25], as they have been associated with mutations in several genes such as *LIS1* (*PAFAH1B1*; MIM#601545), *DCX* (Doublecortin; MIM#300121), *ARX* (Aristaless-related homeobox gene; MIM#300382), *RELN* (Reelin; MIM#600514), *VLDLR* (MIM#224050) and *TUBA1A* (αtubulin 1a) [26]. Some rare forms of lissencephaly (LCH) are associated with a disproportionately small cerebellum.

#### Lissencephaly 2

#### a) Humans

Lissencephaly 2 (LIS2) also referred to lissencephaly syndrome, Norman–Roberts type or Norman–Roberts syndrome (OMIM #257320) is associated with *LIS1* but displays several specific clinical features. In 2000, Hong and colleagues were the first to describe an autosomal recessive form of lissencephaly that, at MRI, also exhibited severe alterations of the cerebellum, hippocampus, and brainstem. More specifically, these alterations consisted of a thickening of the cerebral cortex with a simplified convolutional pattern that was particularly evident in the frontal and temporal lobes, whereas the parietal and occipital lobes were almost normal. The hippocampus was unfolded and flattened, lacking definable upper and lower blades. The corpus callosum was thin and the lateral ventricles enlarged. The cerebellum was clearly smaller than in the normal brain, hypoplastic, and devoid of folia. Authors also showed that the responsible gene mapped to chromosome 7q22 and that the condition was associated with two independent mutations in *RELN*, resulting in low or undetectable amounts of RELN after Western blots analysis of the patients' serum [9]. Two other unrelated groups of patients, later, presented the same type of LIS2 [27]. They were children that, at MRI, displayed a 5–10 mm thick cerebral cortex, a malformed hippocampus and a very hypoplastic cerebellum, almost completely devoid of folia. As LIS2 is a rare disease, there are very limited histopathological data on the condition. To our knowledge, the only post-mortem description of a male fetus with Norman–Roberts syndrome reported the occurrence of a four-layered cerebral cortex (Figure 2A,B), a well-developed cerebellum with organized folia, and heterotopia of the dentate nucleus [28].

**Figure 2.** Structural alterations in human, LIS2, and homozygous *Reeler* mouse (**A**–**D**)**;** modifications of the neocortex architecture in human LIS2 (**B**); and *Reeler* mutation (**D**); compared to healthy controls (**A**,**C**). After MRI imaging, the human LIS2 cortex is thicker than normal, whereas there are apparently no thickness changes in mouse. Note that in both species the pathological neocortex only consists of four layers, with an upside-down layer disposition mainly affecting the pyramidal neurons that are also irregularly oriented compared to their usual positioning in normal individuals/mice. Pyramidal neurons are in different color and sizes according to their position in cortical layers. Stellate spiny cells of layer 4 are orange. Inhibitory interneurons are black with a red nucleus. Cajal-Retzius cells of layer 1 are red.

(**E**–**H**): Structural alterations in the *Reeler* mouse cerebellum; (**E**) sagittal sections of the P15 cerebellum in a normal *reln*+/+ mouse; and (**F**) a *Reeler reln*−/<sup>−</sup> mouse: the *Reeler* cerebellum is much smaller and devoid of folia, with a smooth surface. (**G**) Misalignment of the Purkinje neurons in the P60 cerebellum of the *Reeler* mouse. After calbindin 28 kDa immunostaining, the Purkinje neurons are well aligned in a monolayer below the molecular layer in *reln*+/+ mice. They, instead, form a large internal cellular mass within the white matter in *reln*−/<sup>−</sup> mutants (**H**). Abbreviations: DAPI = 4- ,6-Diamidine-2- -phenylindole; GL = granular layer of the cerebellar cortex; ICM = internal cellular mass; ML = molecular layer of the cerebellar cortex; P = postnatal day.

#### b) *Reeler* Homozygous Mice

Alterations in *Reeler* homozygous recessive mice fully recapitulate those in human LIS2 (Figure 1). Due to obvious technical and practical reasons, the amount of MRI data in mouse is by far less abundant than in patients, whereas mice have provided extensive histopathological information. The first MRI description of the neuroanatomical phenotypes in homozygous (and heterozygous) mice using morphometry and texture analysis, led to conclude that the structural features of the *Reeler* brain most closely copied the MRI phenotype of LIS2 patients [29]. Indeed, the *reln*−/<sup>−</sup> mice had a smaller brain, but larger lateral ventricles compared to wild-type littermates. Sharp differences existed in the olfactory bulbs, dorsomedial frontal and parietal cortex, certain districts of the temporal and occipital lobes, and the ventral hippocampus where gadolinium-based active staining demonstrated a general disorganization with differences in the thickness of individual hippocampal layers. The cerebellum also resulted profoundly affected by the mutation and appeared strongly hypoplastic. A subsequent study, based on the use of manganese-enhanced MRI (MEMRI) to better detect the cortical laminar architecture, compared the MEMRI signal intensity in the cerebral cortex of normal and mutant mice. The authors of this survey observed that signal was low in cortical layer 1, increased in layer 2, decreased in layer 3 until mid-layer 4, and increased again, peaking in layer 5, before decreasing through layer 6. In *Reeler* there were, instead, no appreciable changes in signal intensity, an observation consistent with the absence of cortical lamination after histological examination [30]. A more recent and very elegant study has employed diffusion tractography imaging (DTI) to perform an in vivo origin-to-ending reconstruction of the mouse somatosensory thalamo-cortical connections and demonstrated an extensive remodeling in *Reeler* mutants because of the highly disorganized cortical lamination [31].

In keeping with the results of imaging studies, at gross anatomical examination the *reln*−/<sup>−</sup> mouse brain was atrophic, as total volume in mutants decreased of about 19% when compared to normal mice [29]. Such a reduction was particularly evident in the cerebellum (Figure 2E,F) that also displayed a very limited degree of foliation. Therefore, also the gross anatomy of the *Reeler* brain closely resembled that of the LIS2 human brain.

In general, it seems that the histological anomalies in mutants depended on an abnormal migration of neurons, rather than an alteration in cell fate determination or axonal guidance. Among these anomalies, the most distinguishing ones are that the cerebral and cerebellar cortices lose their layered structure, in accordance with the MEMRI observations [30]; numerous neuronal nuclei disappeared or, at least, became hardly recognizable in several brain regions; and neurons often displayed an ectopic position. Table 2 summarizes the most important structural anomalies of the *reln*−/<sup>−</sup> CNS without taking into consideration the histological alterations in the cerebral cortex, hippocampus and cerebellum, as we will analytically discuss the phenotype of these brain areas in the following sections. Detailed descriptions of the morphological phenotype of the *Reeler* mouse CNS can be found e.g., in [29,32].


**Table 2.** Main histopathological changes in the homozygous *Reeler* mouse.

The Table does not list the histopathological observations on cerebral cortex, hippocampus, and cerebellum.

Very early observations demonstrated the occurrence of dendritic anomalies in cortical and hippocampal neurons of *Reeler* mice [48,49]. The discovery of *Reln* confirmed the dendritic pathology, as not only Reln but also the molecules of its signaling pathway resulted to be necessary for the correct maturation and differentiation of the dendritic branches and spines in hippocampal and neocortical pyramidal neurons [50,51].

Due to the complexity of the phenomena involved in dendritic maturation, one can argue that dendritic anomalies represented a consequence of the deep cytoarchitectonic derangement occurring in *Reeler* mice rather than a primary effect of the lack of Reln, but observations on heterozygous mice were not supportive of this interpretation [52,53]. Interestingly, the block of the Reln signaling by means of specific antibodies resulted in an increased complexity of branching in the apical dendrites of layer 2/3 cortical pyramidal neurons, whereas their basal arborizations remained unaffected [54].

There are many important issues related to the structure and role of the dendritic tree of neocortical and hippocampal pyramidal neurons that make the *Reeler* mouse an important tool for the study of (forebrain) neurodevelopment. Inputs to layer 5 neurons are processed by separate compartments, with the basal dendrites receiving bottom-up information and the apical dendrite being the recipient of a feedback input from higher cortical areas, see e.g., [55]. This framework is, however, even more complex because apical dendrites span most cortical layers before reaching layer 1, where the apical tuft is located [56]. Today we know well that the type and distribution of ion channels at the neurolemma ultimately determine the electrophysiological properties of a neuron.

Essential to the function of the long apical dendrite of the pyramidal neurons is the progressively increasing density of hyperpolarization-activated cyclic nucleotide–gated (HCN) channels, proceeding from proximal to distal segments [57]. Such a gradient critically contributes to the functional distinction between dendritic compartments. Although Reln signaling specifies this gradient [58], 17β-estradiol, which stimulates Reln expression, promoted the enrichment of HCN1 in the distal dendritic compartment of CA1 neurons without the intervention of Reln [59]. The evidence that Reln was involved in the trafficking and targeting of ion channels in cortical and hippocampal neurons suggested that their intrinsic electrophysiological properties could indeed be different in the *Reeler* mouse. However, an early study by Bliss and Chung [60] demonstrated that, despite the layering derangement, the basic synaptic organization of the hippocampus was largely unchanged in mutants.

More recently, Silva et al. [61] carried out an accurate survey dealing with the intrinsic electrophysiological properties of cortical neurons in *Reeler* mice. These authors showed that the firing pattern and synaptic responses of the pyramidal neurons were normal, but with an inverted radial distribution. Notably, these authors concluded that, although mispositioned, neurons maintained the membrane properties appropriate to their function.

The apparent discrepancy between the data demonstrating the role of Reln in the modulation of ion channels and the relative lack of anomalies in the intrinsic properties of cortical neurons in mutant mice might have several explanations. Other factors, such as neuronal activity [62] could be more effective than Reln for the modulation of membrane channel targeting. Furthermore, the complex machinery of the long apical dendrite is required when layer 5 neurons settle appropriately but might be useless for the same neurons displaced to more superficial cortical layers. Finally, future investigations based on refined electrophysiological techniques, such as direct dendritic recordings, will help to establish if indeed the cortical neurons in *Reeler* mice display subtler changes of their firing/intrinsic properties that those so far ascertained.

Reln signaling is also able to modulate key molecules of the cascade leading to synaptic plasticity, such as the NMDA receptors [63,64]. Therefore, several studies concentrated on the changes of synaptic plasticity in *Reeler* mutants. Ishida et al. [65] reported that the induction of long-term potentiation (LTP) was impaired in the CA1 region of the hippocampus, claiming that the malpositioning of some neuronal populations could account for such an alteration. On the other hand, both the overexpression of Reln in transgenic mice [66] and Reln supplementation strongly increased LTP [67]. Later, a defective LTP was observed in the hippocampus of *vldr*-deficient mice, but slice perfusion with Reln was able to enhance LTP in CA1 [68].

Most cortical neurons are spiny, glutamatergic pyramidal cells, whose migratory path during prenatal development follows an inside-out radial pattern from the ventricular zone to the final position [69]. Reln signaling is essential for the localization of pyramidal neurons to appropriate cortical layers, as reviewed in [70]. Consequently, the lack of Reln caused a disruption of the layered cortical organization, including abnormal positioning [71,72], as well as an increased percentage of inverted pyramidal cells [73,74] (Figure 2C,D).

Inhibitory GABAergic interneurons represent a minority population within the neocortex. Yet, their morphological, neurochemical and functional diversity likely plays a key role for the cortical function, see e.g., [75]. Unlike pyramidal neurons, interneurons originate in the ganglionic eminence of the ventral telencephalon and follow a tangential migratory path to the cortex [69]. While the malpositioning of the pyramidal neurons in *Reeler* mice is evident, it is not clear if the Reln signaling cascade also affects the migration of the interneurons. An answer to this latter issue came from observations on *Reeler* mutants crossed with mice expressing the green fluorescent protein (GFP) in inhibitory neurons. Thus, the results of these observations confirmed that also the cortical interneurons displayed an abnormal laminar position and morphology [76]. However, we still do not know whether interneurons' ectopy directly depends from Reln signaling or is rather the consequence of the malpositioning of principal pyramidal projection neurons. The debate on this issue is still open, as contradictory views exist in the literature. Namely, while some observations [77,78] argue against a direct role of Reln,

Hammond et al. [79] showed that only early-generated cortical interneurons were misplaced as a consequence of the ectopy of the pyramidal neurons, whereas the correct layering of late-generated interneurons seemed to be directly modulated by Reln signaling.

Other basic neurodevelopmental features, such as cortical [80] and cerebellar neurogenesis, seem to be as well regulated by the glycoprotein. Consequently, the minicolumnar organization of the cerebral neocortex appeared to be deeply affected by Reln deficiency [81] and some physiological counterparts of cortical connectivity, such as trans-synaptic signal propagation, were also impaired [82]. However, the outcome of Reln deficiency on the microcircuitry sustaining the cortical machinery is controversial and, surprisingly, the deep architectonic disorganization that follows the lack of the protein occurs in the absence of dramatic functional anomalies. Both early and more recent studies point out that the absence of Reln did not prevent the development of functionally appropriate cortical connections and maps [31,83–86]. In addition, when studied at the fine-scale electron microscopic level, the basic synaptic organization of misplaced cortical neurons was unchanged [87]. Therefore, although the laminar organization is thought to be critical for cortical computation [88,89], evidences obtained in *Reeler* mice led Guy and Staiger [90] to challenge the importance of cortical lamination, affirming that "future studies directed toward understanding cortical functions should rather focus on circuits specified by functional cell type composition than mere laminar location".

Macroscopically, the cerebellum of the *Reeler* mouse is smaller than that of age-matched littermates (Figure 2E,F); it is club-shaped with the main axis transverse to the mid plan of the body, and has an almost completely smooth surface, with just a few superficial grooves [91]. The architecture of the *Reeler* cerebellum is profoundly different from the normal pattern, firstly because of the impairment in the complicated series of migrations made by neurons to reach their destination in the mature organ. Trajectories of migrating neurons follow two opposite directions from the surface to the depth of the cerebellum and the other way around, depending from the species, the type(s) of neurons and the developmental stages (for details see e.g., [92]). Eventually, disturbances in the migration of the cerebellar neurons make that *Reeler* mice display a cerebellum that retains several features of immaturity.

The area of the cerebellar cortex in mutants was analyzed quantitatively during postnatal (P0–P25) development and resulted to be reduced compared to age-matched controls [93]. Reduction in the extension of the cortex was particularly evident in the molecular layer and the (internal) granular layer. Physiologically, as the cerebellum matured, the molecular layer became more and more populated by the parallel fibers, but, at P25, its increase in size was about one third in the mutants compared to *reln*+/+ mice [93]. Post-migratory granule cells, which are born in the temporary subpial external granular layer, progressively populate the (internal) granular layer during normal cerebellar development. This process was disturbed in *Reeler*, to the extent that, from P0 to P10, the granular layer of *reln*+/+ mice increased about five-folds in size, but only 2.6-fold in *reln*−/−, where it was significantly reduced in size to 0.62-fold that of normal mice after P10 [93]. In a different way from the cortex, the medullary body was larger in the mutants than in wild-type mice. Its progressively increasing area mainly reflected the ongoing myelination of the axons of the Purkinje neurons that abandon the cortex moving across the white matter to reach the cerebellar nuclei, as well as the expansion of the incoming afferent and departing efferent fiber systems. The mass of the medullary body augmented in relation to postnatal age irrespectively of the lack of Reln (*reln*−/<sup>−</sup> 2.59, *reln*+/+ 1.93-fold), but, at P25, *Reeler* mice had a larger medullary body than normal mice (1.88-fold) [93]. In brief, *Reeler* mice had a reduced cerebellar cortex but a bigger medullary body than their *reln*+/+ littermates. The cerebellar hypoplasia was thus a consequence of a reduction in cortical magnitude and cellularity and the latter, in turn, resulted to be associated to measurable differences in the degree of cell proliferation and apoptosis, as well as imbalances in the timing of postnatal cortical maturation [93]. The same study led to conclude that density of proliferating cells was the most significant predictive factor to determine the cortical cellularity in *Reeler* [93]. Therefore, beside the well-defined consequences onto neuronal migration, the lack of Reln also caused a calculable deficit in neuronal expansion. Ultrastructurally, the cerebellar

neurons underwent several different forms of programmed cell death during postnatal development and the deficit of Reln affected the kind and grade of neuronal death [94].

Perhaps the most striking histological feature in mutants is the lack of alignment of the Purkinje neurons to form a discrete intermediate layer in the cerebellar cortex (Figure 2G,H). Thus, in *Reeler*, only about 5% of the Purkinje neurons were in a normal position, 10% were still inside the cortex but in the granular layer, and the remaining 85% formed an internal cellular mass intermixed with the white matter [95–97]. Ultrastructurally, in *Reeler* there was a reduction in the density of the contacts between the Purkinje neurons and the parallel and climbing fibers, from P5 onward [98]. Functionally, both the normally placed Purkinje neurons and those ectopically dislocated in the granular layer displayed a 0–1 response to stimulation, indicating that, as in normal mice, they received a synaptic contact by a single climbing fiber. The Purkinje neurons in the internal cellular mass, instead, showed intensity-graded responses to electrical stimulation, as several climbing fibers provided them with a convergent input [95], likely as a failure of physiological pruning to occur [99]. Neurochemically, there were no obvious variations between normal mice and the mutants in the temporal expression of some widely diffused neuronal and glial markers (NeuN, vimentin, calbindin, GFAP, Smi32, GAD67) during postnatal development [93], but the Bergmann glia was misplaced in *Reeler* [100].

To conclude, the histological and electrophysiological observations in *Reeler* mutants suggest that similar structural and functional alterations may also occur in LIS2 patients, particularly in relation to the postnatal growth retardation, severe intellectual disability, and spasticity observed in affected subjects (see also https://www.orpha.net/).

#### Lissencephaly 3

*TUBA1A* mutations [101,102] cause lissencephaly 3 (LIS3), another human condition that has a mouse counterpart. TUBA1A chiefly occurs in cortical, hippocampal, cerebellar and brainstem post-mitotic neurons, with expression falling soon after birth but persisting through adulthood [103]. The mouse phenotype consists, among others, in a failure of the cerebellar Purkinje neurons to migrate, so that, similarly to *Reeler*, they remain entrapped into the medullary body, where they form a series of streaks intermingled with the neurons of the cerebellar nuclei [104]. Several other mutations of *TUBA1A* exist in humans. They give rise to a predominant phenotype of LCH, which also shows irregularities of the corpus callosum and the basal ganglia/internal capsule [105].

## 3.1.2. Autosomal-Dominant Lateral Temporal Epilepsy and the Heterozygous *Reeler* Mouse

ADLTE, also referred to as autosomal dominant epilepsy with auditory features, partial epilepsy with auditory aura or partial epilepsy with auditory features, is a genetic epileptic syndrome, clinically showing typical focal seizures in response to specific sounds. ADLTE is genetically heterogeneous, and mutations in the leucine-rich, glioma inactivated 1 gene (*LGI1*) account for fewer than 50% of affected families. Very recent observations demonstrated that heterozygous *RELN* mutations give rise to a classic ADLTE syndrome, clinically identical to that associated with mutations of *LGI1*. Seven different heterozygous missense mutations in *RELN* were, in fact, described in some unrelated families of Italian ancestry with familial temporal lobe epilepsy-7 (ETL7–OMIM #616436) [106]. Incidence was 17.5% over the total number of families studied that were specifically suffering by lateral temporal lobe epilepsy [106]. By three-dimensional modeling, the same authors anticipated that the outcomes of these mutations would be protein structural defects and misfolding. Some of the affected individuals displayed a reduction up to 50% of their serum levels of the 310 kDa RELN isoform in comparison to healthy subjects and thus, very likely, the mutations also resulted in a loss of function. In a subsequent study on the same patients, 1.5 T MRI scans were not useful in detecting structural anomalies of the brain [107]. Similarly, in a very recent study on an 18-year old ADLTE patient, 3T MRI brain scans could not provide relevant information on indistinct grey-white matter connections, voxel-based morphometry, and cortical thickness [108]. However, analysis of functional connectivity with high-density electroencephalography (HdEEG) revealed greater local synchrony in

the left temporal (middle temporal gyrus), left frontal (supplementary motor area, superior frontal gyrus), and left parietal (gyrus angularis, gyrus supramarginalis) regions of the cerebral cortex and the cingulate cortex (middle cingulate gyrus) as compared to normal subjects [108].

As the discovery of RELN mutations in ADLTE is a quite recent finding, there are, at present, no observations on heterozygous *Reeler* mice focused to ascertain possible similarities with the human phenotype. Like ADLTE patients, *reln*+/<sup>−</sup> mice display a 50% reduction of Reln in their serum. Therefore, it would be interesting to investigate whether sound-triggered epileptic manifestations also occur in these animals. A very recent study has demonstrated that optogenetic stimulation of the parvalbumin (PV) immunoreactive GABAergic neurons of the mouse basal forebrain can modulate the cortical topography of auditory steady-state responses [109]. As the regional distribution of these neurons displayed relevant differences in *reln*+/<sup>−</sup> mice compared to wild-type animals [110,111], any phenotypic alteration may be of interest to shed additional light onto human ADLTE. Finally, a very latest report has provided proof of concept that HdEEG can be used to record electrical activity from the mouse brain in a model of juvenile myoclonic epilepsy [112]. Therefore, one can envisage applying such an approach to analyze the brain electrical pattern in *reln*+/<sup>−</sup> mice aiming to collect data for translational comparison with ADLTE.

#### *3.2. Human Conditions Caused by Mutations of Genes of the Reln Intracellular Pathway and Their Mouse Correlates*

In general, the brain phenotype of the human monogenetic conditions that are consequent to mutations of the genes coding for the proteins of the RELN intracellular signaling pathway is similar to that of the *reln*−/<sup>−</sup> mouse brain, except that, in certain cases, differently from *Reeler*, the human cerebellum is normal at MRI and gross anatomical observation (Figure 1). We will briefly describe these conditions below.
