This review addresses the fundamental compartmentation of the cerebellar cortex into transverse zones and parasagittal stripes. On the other hand, the review does not address secondary aspects of patterning such as the lobulation of the cortex (e.g., Legué et al., 2015) and the regional restriction of the afferent terminal fields and various interneurons (e.g., Consalez and Hawkes, 2013). The key perspective is the light cast on patterning by mouse mutant phenotypes.

Briefly, the cerebellar cortex is highly patterned into hundreds, if not thousands, of genetically predetermined quanta (reviewed in Apps and Hawkes, 2009; Armstrong and Hawkes, 2014; Leto et al., 2016; Apps et al., 2018). The earliest pattern to form is an anteroposterior array of transverse zones defined by boundaries between Purkinje cell (PC) domains (Ozol et al., 1999). Five zones are reproducibly identified to date in all birds and mammals: the anterior zone (AZ), central zone (CZ—further subdivided into anterior and posterior domains: Marzban et al., 2008), posterior zone (PZ) and nodular zone (NZ). Similar transverse boundaries are also found in the granular layer.

Each transverse zone is subdivided from medial to lateral into parasagittal stripes. This patterning is most clearly seen in the adult in the PC expression patterns of numerous molecules. For example, the AZ and PZ show alternating striped arrays of PC expression for zebrin II/aldolase C (= zebrin II+, Brochu et al., 1990; Ahn et al., 1994) or the zebrin II- subset (e.g., phospholipase (PL)Cβ4 (Sarna et al., 2006) or the transcription factor Early B-cell Factor 2 (Ebf2, Croci et al., 2006). In contrast, the CZ and NZ are uniformly zebrin II + but stripes are seen in other expression patterns [e.g., heat shock protein (HSP)25±, Armstrong et al., 2001 or HNK-1, Marzban et al., 2004]. In all zones, the zone-and-stripe patterns are highly reproducible between individuals and conserved across species (e.g., Sillitoe et al., 2005), but interspecific size differences are typical, reflecting a species’ particular lifestyle. For example, mosaic evolution results in the CZ being especially large in bats (Kim et al., 2009), where it subserves echolocation and vision, whereas it is especially small in blind species such as the naked mole rat (Marzban et al., 2011) or star-nosed mole (Marzban et al., 2015; see also Legué et al., 2015). In parallel to the patterning of the PCs, the granular layer is similarly patterned into transverse domains that align with the PC transverse zones (e.g., Sillitoe et al., 2003).

In brief, the zone-and-stripe pattern forms during development via two separate but interacting pathways (reviewed in Leto et al., 2016). In brief, the inhibitory neurons—in particular the PCs—are born in the ventricular zone of the fourth ventricle (VZ: E10-E13 in mice). Postmitotic PCs migrate first to the cerebellar plate which subsequently disperses to form a stereotyped array of clusters (E14-E18), which are the topographic organizing centers that pattern the afferent connections and the interneurons (reviewed in Consalez and Hawkes, 2013; Apps et al., 2018). In parallel to the PC patterning, the granule cells are generated from granule cell precursors (G) in the upper rhombic lip (URL), which divide extensively and spread to cover the cerebellar anlage as the external granular layer (EGL) by E18. Thus, by birth the fundamental zone-and-stripe architecture has been laid down. Subsequently the embryonic cerebellum transforms into the adult: between ∼E18-P20 the granule cells in the EGL and the PC clusters all disperse to form the elaborate adult zone-and-stripe array.

Dozens of mutations are known that impede normal cerebellar development and many show abnormal morphology. The underlying disruptions are of two kinds: “patterning” defects—mutations that disturb normal pattern formation; and “patterned” defects—those which are secondary consequences of the underlying abnormal patterns. A pertinent analogy might be the bud and the flower. Mutations acting on bud formation might result, for example, in missing petals; in contrast, in mutations acting on the transformation of the bud to the flower, all the normal components would be present but in some way distorted. Almost all cerebellar mutants are of the second kind—all normal elements are present but disorganized. Indeed, it is arguable that mutations of the first kind have never been unambiguously identified.

PC pattern formation occurs in four distinct phases during which complexity steadily increases: PC birth in the ventricular zone of the VZ; migration to the cerebellar plate; dispersal of the cerebellar plate into a stereotypes array of PC clusters; and dispersal of the cluster array to reveal the mature zone-and-stripe architecture. In the interests of space, the patterning events that precede the formation of the cerebellar anlage (∼pre-E10) are not considered here (for an overview of PC patterning with numerous useful illustrations, see Leto et al., 2016).

Precisely when the zone-and-stripe pattern is first specified in cerebellar development is uncertain. There are several candidate stages at which aspects of the adult architecture might be specified: in the VZ; in the cerebellar plate; in the transformation of the cerebellar plate into the embryonic cluster array; and during the metamorphosis of the cluster array to the adult stripes. In all likelihood, stages in pattern formation involves several.

The patterning of the granule cell pathway happens in parallel to the PC VZ→cerebellar plate→cluster array→stripe formation pathway (a thorough review of granular layer patterning, together with useful cartoons, can be found in Consalez et al., 2021). All granule cells of the cerebellar cortex derive from Atonal Transcription Factor 1 (Atoh1: a.k.a Math1) lineage in the URL, and deletion of Atoh1 completely ablates the granular layer (Akazawa et al., 1995; Ben-Arie et al., 1997; reviewed in Consalez et al., 2021). GCPs migrate from the URL, starting at E10-E12, and proliferate and spread to cover the entire surface of the cerebellar anlage as the external granular layer (EGL) by E18. Starting around E18 postmitotic granule cells descend ventrally from the EGL, guided by Bergmann glial fibers through the dispersing PC clusters and the developing molecular layer, and settle in the nascent granular layer. Populating the granular layer proceeds until P20-P30, by which time the GCPs are spent, and the EGL has disappeared.

Within the URL, before the EGL is formed, there is already elaborate patterning as revealed by the expression of multiple transcription factors (Yeung et al., 2014), and by E18, the patterning of the EGL is established. Four EGL transverse domains that derive from distinct GCP populations are revealed by differences in gene expression, birthdates, lineage restriction and the consequences of mutations (reviewed in Consalez et al., 2021). Each transverse domain is aligned with its counterpart in the PC layer—the EGL-AZ (eAZ), eCZ, ePZ, and eNZ. This alignment presumably arises because the PC cluster architecture restricts the dispersal of different GCP lineages or the underlying PC clusters induce gene expression in the overlying EGL (e.g., Smeyne et al., 1995). In particular, multiple mutations result in defects with phenotypic abnormalities that are restricted at EGL transverse domain boundaries. These phenotypes are of two kinds: intrinsic defects that directly affect the GCPs [e.g., meander tail (mea/mea), Hamre and Goldowitz, 1997] and extrinsic defects due to abnormalities in the local environment (e.g., defective GCP production in sonic hedgehog (Shh) null mice, Lewis et al., 2004; PC defects in the Retinoid receptor-related Orphan Receptor (ROR)α mutant staggerer (Rora^sg), reviewed in Gold et al., 2003; Vitalis and Mariani, 2018). These are often difficult to distinguish in practice, and both intrinsic and extrinsic effects are often present concomitantly (e.g., weaver, Smeyne and Goldowitz, 1989).

The eAZ/eCZ boundary restricts the expression of several genes. For example, expression the homeodomain transcription factor Otx1 (Frantz et al., 1994), fibroblast growth factor alpha, Zic1 (Aruga et al., 1998), and protein tyrosine phosphatase rho (McAndrew et al., 1998), are all restricted predominantly to the EGL anterior to the eAZ/eCZ boundary. In contrast, expression of Otx2 (Frantz et al., 1994) and the homeodomain transcription factor Lmx1a is restricted to the EGL posterior to the eAZ/eCZ boundary (and see below for the eCZ). The eAZ/eCZ is also a lineage-restriction boundary in embryonic stem cell chimeras, and hence in this case at least the EGL boundary is not secondary to induction by the underlying PC clusters (Hawkes et al., 2008).

Several mutant EGL phenotypes are restricted at the eAZ/eCZ boundary. The most prominent phenotype is the presence of an agranular AZ. This is seen in the meander tail (mea/mea) mutant (Ross et al., 1990), in rostral cerebellar malformation (Unc5c^rcm, Ackerman et al., 1997; Eisenman and Brothers, 1998), in the failure of the zebrin II- PC subpopulation in the cadherin receptor mutant cerebellar deficient folia (Ctnna2^cdf, Beierbach et al., 2001) and in the heterozygous weaver mutant (wv/+ , Eisenman et al., 1998; Armstrong and Hawkes, 2001). In contrast, a null mutation of the transcriptional activator gene Neurogenic Differentiation 1 (Neurod1) results in a selective loss of GCPs posterior to the eAZ/eCZ boundary (Miyata et al., 1999; Cho and Tsai, 2006). As noted above, in the posterior EGL the GCPs are Lmx1a+, and consequently Lmx1a mutations (for instance, the deletion in the autosomal recessive mutant dreher (Lmx1a ^drJ)) manifest EGL defects focused on the posterior vermis (Chizhikov et al., 2010: interestingly, in null mutations of Lmx1a, GCPs over-migrate into the anterior vermis, leading to posterior vermis hypoplasia). In addition, in the dreher mouse both roof plate and rhombic lip defects induce extreme deformities in the overall morphology including dramatic vermis distortion (Millonig et al., 2000; Chizhikov et al., 2010), presumably secondary to the switching of the Lmx1a + GCPs to an anterior (Lmx1a-) fate. However, despite the gross cortical deformities the underlying zone-and-stripe architecture is preserved (e.g., the HSP25 ± and zebrin II ± stripe arrays are intact, Sillitoe et al., 2014). Similarly, Ebf2 is expressed transiently in the URL and the Atoh1 + migratory stream between E12.5-E13.5 (Croci et al., 2006, 2011; Chung et al., 2008). Genetic fate mapping indicates that granule cells derived from Ebf2 + precursors are restricted largely to the eAZ [Badaloni et al., 2019: the AZ is populated predominantly by early born GCs (Machold and Fishell, 2005), so this may account for their anterior location]. There is no obvious phenotype in the Ebf2−/− EGL. Finally, in Dab1 null mutants a clear separation of anterior and posterior EGLs is found at the eAZ/eCZ boundary presumably secondary to the failure of PC cluster dispersal (Gallagher et al., 1998).

The evidence for a distinct eCZ/ePZ boundary and a specific eCZ domain is less clear cut. First, in chimeras of wildtype and the mutation small eye (Pax6^Sey/Sey) a lineage boundary is seen in lobules VII where the Pax6 mutation preferentially affects the EGL anterior to the eCZ/ePZ boundary (Swanson and Goldowitz, 2011). Secondly, Otx1/Otx2 expression reveals two boundaries in the EGL—the eAZ expresses only Otx1 and the ePZ expresses Otx2. However, between the two the eCZ is revealed as a transverse domain of Co-expression (Frantz et al., 1994).

No positive markers of the third EGL transverse domain, the ePZ, are known. However, the ePZ can be defined as located between the posterior border of the eCZ and the anterior border of the eNZ. This is also a lineage boundary in chimeras at the same approximate location (Hawkes et al., 2008) and an expression boundary for zebrin II in the heterozygous mutant Lurcher (Tano et al., 1992), which has constitutive activation of the delta 2 ionotropic glutamate receptor (Grid2^Lc, Wollmuth et al., 2000). The posterior boundary of the ePZ is revealed by several positive markers of the eNZ. The clearest example is the restricted expression of the homeodomain transcription factor gene Tlx3, which clearly delineates the fourth EGL sub-domain aligned with the PC NZ (Logan et al., 2002). Expression of a neurotrophin-3-lacZ transgene is similarly restricted to the adult granular layer of the NZ, both during development and in the adult (Tojo et al., 1995), and the same boundary is seen when an Atonal-cre was used to eliminate Neurod1 (Pan et al., 2009). Finally, a transverse discontinuity in the EGL is seen at the same approximate location in the heterozygous weaver mutant (wv/+, Eisenman et al., 1998; Armstrong and Hawkes, 2001).

The presence of transverse domains in the EGL is clear but evidence for parasagittal stripes is much less so. One candidate is in the young rabbit, where a monoclonal human granulocyte antibody B4,3 (= anti-CD15) reveals a striped expression pattern in the EGL (Marani and Tetteroo, 1983). Presumably this is a secondary induction via the underlying PC clusters or stripes. Another novel mediolateral pattern is seen in the evidence that one subgroup of GCPs in humans—concentrated in the hemispheres—is especially sensitive to Shh signaling and, perhaps as a result, is also more likely to give rise to meduloblastomas (Tan et al., 2018): how, if at all, this GCP mediolateral segregation relates to the anteroposterior EGL compartmentation, is unknown.

Mutant mice with cerebellar phenotypes have proved powerful tools to explore the mechanisms of pattern formation (Sidman et al., 1965). Abnormal patterns include missing cells, either never born (e.g., the Zfp423Δ9-20 allele, Casoni et al., 2017; Lmx1a−/−, Chizhikov et al., 2010) or died (e.g., Ebf2−/−, Croci et al., 2006; Chung et al., 2008; selective granule cell death in Neurod1−/−: Miyata et al., 1999; Cho and Tsai, 2006), ectopias due to abnormal cell dispersal (e.g., for PCs, mutations in the Reln→Dab1 pathway or in wv/wv—Armstrong and Hawkes, 2001; for granule cells, examples include abnormal EGL migration in Lmx1a−/−, Chizhikov et al., 2010), and probably expression defects (“pattern defects” vs. “expression defects”; e.g., expression patterns induced by the local environment). These include 5′-nucleotidase expression by Bergmann glial cells induced by the local zebrin II + PCs (Scott, 1963; Eisenman and Hawkes, 1989) or abnormal tyrosine hydroxylase expression in rolling mouse Nagoya (rol/rol) a mutation in the α1 subunit of the calcium channel gene Cacna1a (Sawada et al., 1999), and the spontaneous dilute-lethal (Myo5a^d–l) mutant (Sawada et al., 1999).

It is remarkable how much light has been shed on cerebellar patterning and the mechanisms of pattern formation through the analysis of mutant mouse strains. The mutant phenotypes reveal stages of patterning during development that become progressively complex. During PC generation and the formation of the cerebellar plate (E10–E13) there is at least some PC subtype specification, but whether the cerebellar plate contains a protomap is not known. From E14-E18, the cerebellar plate disperses into the embryonic PC cluster array. In parallel, during granule cell development, mutant phenotypes reveal reproducible boundaries within the EGL that arise often from distinct lineages in the URL (e.g., rcm, cdf, mea etc.). The transverse domains in the EGL and the zones in the PC layer become aligned during embryogenesis, but the mechanism that guides this is not well understood. As a result, by E18 the fundamental zone-and-stripe pattern is established. The embryonic bud is transformed into the adult flower during the first 3–4 postnatal weeks due to PC dispersal triggered by Reln→Dab1 signaling and granule cell migration from the EGL to the maturing granular layer. Finally, granular layer maturation reveals a further stage in cerebellar patterning—the subdivision of stripes into thousands of patches (e.g., Hawkes and Turner, 1994; Hawkes, 1997; Ozol and Hawkes, 1997; Sillitoe et al., 2003)—probably associated with inductive influences during the maturation of the mossy fiber projection maps (e.g., Shambes et al., 1978; Schilling et al., 1994).

The author confirms being the sole contributor of this work and has approved it for publication.

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

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.