Untying the Gordian knot – the return of Reelin

RELN. Amongst the various genes implicated in neurodevelopmental disorders, Reelin (RELN) has always been one of the more controversial genes. While bi-allelic variants have been implicated in lissencephaly with cerebellar hypoplasia, the role of autosomal dominant variants has been controversial and is currently considered disputed. Reelin is a relatively large gene – accordingly, missense variants are frequent. However, a recent study suggests that the picture might be more complicated and that both monoallelic and bi-allelic variant in Reelin may contribute to neurodevelopmental disorders. Here are my thoughts.

Figure 1. Ectopic expression of Reelin causes neuronal aggregation (Figure 2 in Hirota and Nakajima under a Creative Commons Attribution License (CC BY). In the developing neocortex (LEFT), Reelin (red) is mainly secreted from the Cajal-Retzius (CR) cells in the marginal zone (MZ), into which the migrating neurons extend their apical leading processes. The primitive cortical zone (PCZ) is densely occupied by neuronal somata (gray). In neuronal aggregation induced by ectopically expressed Reelin (RIGHT), the leading processes of the migrating neurons assemble in the Reelin-rich central region, and the cell bodies are aligned in the PCZ-like peripheral region (gray). In both cases, the migrating neurons exhibit an “inside-out” cell arrangement in which the late-born cells (green) pass by the early-born neurons (magenta).

Reelin. In 1951, Falconer described an unusual mouse phenotype. This particular spontaneous mouse mutant showed a characteristic swaying and staggering gait and was therefore called the reeler mouse. Subsequent histological studies showed that this mouse model had an unusual disorganization of cortical patterning and that the size of the cerebellum is significantly decreased. Fast forward several decades to the discovery of Reelin. The Reelin protein is coded by RELN, the causative gene behind the mouse reeler mutant. Reelin is a large secreted protein that is produced by a specific cell type, the Cajal-Retzius cells in the developing brain. Reelin is critical in establishing the pattern of the human cortical layers. In contrast to the cortex in reptiles and other animals, mammals have and inside-out pattern in the cortex. When neuronal progenitor cells travel into the cortex, the young cells pass through a layer of older cells, which is mediated by Reelin. If Reelin is absent, this complex patterning no longer takes place.

RELN genetics. The human genetics regarding RELN has been complicated with an initial impactful discovery followed by almost two decades of disappointment. In 2000, Hong and collaborators described bi-allelic disease-causing variants in RELN in lissencephaly, a disorder of brain development where the gyri and sulci in the cortex are absent. The particular hallmark of RELN-related lissencephaly was the combination of lissencephaly with cerebellar hypoplasia, a phenotype that paralleled the findings in the reeler mouse. This presentation has been very characteristic, but also extremely rare. Up until the recent study by Di Donato and collaborators, only 11 individuals had ever been reported. However, RELN had also been implicated in other disease mechanisms. For example, heterozygous RELN variants had been implicated in temporal lobe epilepsy. However, in contrast to the lissencephaly phenotype, this finding had been controversial.

Controversy. In order to understand the controversy regarding autosomal dominant RELN variants in temporal lobe epilepsy, it is important to understand how we typically assess whether variants in a given gene are disease-associated. In the ClinGen framework, specific variant classes (de novo, inherited) are given specific points. Given that inherited variants could theoretically always represent benign population variants, particularly the points for inherited missense variants is low. This is particularly relevant as Reelin is a very big protein. With 65 exons, it has almost three times as many exons as the SCN1A gene. However, a look at gnomAD also shows something else: the RELN gene is constrained in the general population with regards to loss-of-function variation, a typical feature of candidate genes in neurodevelopmental disorders with haploinsufficiency as the major disease mechanism. However, without any de novo variants in RELN described, invoking such a model is difficult.

New RELN publication. The recent publication by Di Donato and collaborators unties the Gordian knot, demonstrating how various disease mechanisms regarding RELN can co-exist. The authors demonstrate there are basically several disease mechanisms in RELN-related disorders, summarize the existing literature and provide information about previously unreported individuals. In brief, in addition to the bi-allelic RELN­-related disorders resulting in lissencephaly, the authors report a milder lissencephaly phenotype that can be due to in-frame RELN variants that likely are causative through a dominant-negative disease mechanism. In addition, they convincingly report protein-truncation variants that usually cause neurodevelopmental disorders such as autism. In summary, the publication by Di Donato revives the discussion regarding heterozygous RELN variants as a cause for neurodevelopmental disorders.

Is Reelin re-established? Is the publication by Di Donato and collaborators sufficient to change the current assessment of heterozygous RELN variants as a disputed disease mechanism in ClinGen. Not quite, unfortunately. The authors have developed their own terminology with regards to the type of variants they classify as pathogenic, which somewhat blurs the distinction to the ACMG-AMP classification. As the publication by Di Donato and collaborators focuses on the similarity of phenotypes in the various subsets of RELN-related disorders, this nomenclature make sense – however, for gene curation, all different disease mechanisms are typically regarded separately and require their own validity assessment. I am sure that heterozygous RELN variants can at least be removed from “disputed” category and re-classified as a potential candidate. Nevertheless, RELN remains a very large gene with a large number of expected missense variants. Implicating missense variants in disease that are not de novo will therefore remain challenging.

Ingo Helbig is a child neurologist and epilepsy genetics researcher working at the Children’s Hospital of Philadelphia (CHOP), USA.