LIS1-Related Isolated Lissencephaly: Spectrum of Mutations and Relationships With Malformation Severity

Objective  With the largest data set of patients with LIS1-related lissencephaly, the major cause of posteriorly predominant lissencephaly related to either LIS1 mutation or intragenic deletion, described so far, we aimed to refine the spectrum of neurological and radiological features and to assess relationships with the genotype.

Design  Retrospective study.

Subjects  A total of 63 patients with posteriorly predominant lissencephaly.

Interventions  Of the 63 patients, 40 were found to carry either LIS1 point mutations (77.5%) or small genomic deletions (20%), and 1 carried a somatic nonsense mutation. On the basis of the severity of neuromotor impairment, epilepsy, and radiological findings, correlations with the location and type of mutation were examined.

Results  Most patients with LIS1 mutations demonstrated posterior agyria (grade 3a, 55.3%) with thin corpus callosum (50%) and prominent perivascular spaces (67.4%). By contrast, patients without LIS1 mutations tended to have less severe lissencephaly (grade 4a, 41.6%) and no additional brain abnormalities. The degree of neuromotor impairment was in accordance with the severity of lissencephaly, with a high incidence of tetraplegia (61.1%). Conversely, the severity of epilepsy was not determined with the same reliability because 82.9% had early onset of seizures and 48.7% had seizures more often than daily. In addition, neither the mutation type nor the location of the mutation were found to predict the severity of LIS1-related lissencephaly.

Conclusion  Our results confirm the homogeneity profile of patients with LIS1-related lissencephaly who demonstrate in a large proportion Dobyns lissencephaly grade 3a, and the absence of correlation with LIS1 mutations.

Genetically inherited disorders of neuronal migration in humans represent important causes of epilepsy and mental retardation. Of these, classical lissencephaly is a severe brain malformation caused by an arrest of neuronal migration from 9 to 13 weeks of gestation and characterized by absent or reduced gyration and an abnormally thick, poorly organized cortex with 4 primitive layers.¹ It encompasses a continuous spectrum of malformations from complete agyria to variable degrees of pachygyria to subcortical band heterotopia.² Lissencephaly may occur either as a component of the contiguous gene deletion disorder known as Miller-Dieker syndrome or as an isolated form known as isolated lissencephaly sequence.^3-5 Clinical manifestations range from profound mental retardation, intractable epilepsy, and spasticity with reduced life span⁴ to milder forms with infrequent seizures and intellectual disability.⁵ The clinical severity correlates generally with the degree of agyria and cortical thickness.¹

Approximately 80% of classical lissencephaly (isolated lissencephaly sequence) cases show abnormalities in either the LIS1 or the DCX genes.⁶ Most cases are due to variations (deletions or mutations) in the LIS1 gene on 17p13.3,^6-8 whereas most subcortical band heterotopias are due to mutations in the DCX gene on Xq22.3.^9,10 Both proteins are known to be associated with and to be required for correct neuronal migration.^11-13 The radiological pattern of LIS1-related lissencephaly differs from doublecortin-related lissencephaly. One of the main distinctive criteria between the 2 conditions is the gradient of lissencephaly, which is known to be more severe in posterior regions (giving a posterior to anterior gradient) in LIS1-related lissencephaly, whereas it is more severe in the anterior brain regions in lissencephaly cases.^2,6

The LIS1 gene comprises 11 exons, 10 of which encode the LIS1 protein, named PAFAH1B1, which contains a LisH homology domain, a coiled coil domain, and 7 WD40 repeats.¹⁴ Large deletions of the LIS1 gene account for most (40%) isolated lissencephaly sequence cases.⁶ To date, 61 intragenic LIS1 mutations have been described,^6,7,15-21 with most (84%) being truncating mutations scattered throughout the entire gene.¹⁸ Missense mutations are less frequent (16%),^18,21 whereas recent data suggest a high prevalence of small genomic deletions/duplications suspected to account for a large number of 49% of all LIS1 alterations.²²

Data on genotype phenotype correlation in LIS1-related lissencephaly are conflicting. Initially, a putative correlation was suggested, with the less severe lissencephaly associated with missense mutations and late truncating mutations localized at the 3′ end of the LIS1 gene.¹⁵ A recent study did not confirm this relationship.^15,21 Moreover, recent data on patients with small deletions in LIS1 suggest that they tend to have the same severity of malformation as those with a frameshift mutation, suggesting that the functional consequences of haploinsufficiency due to either truncating or frameshift mutations or exon(s), deletion(s), or duplication(s) could be comparable. However, owing to the small size of the population (21 in Uyanik et al²¹ and 15 in the Cardoso et al¹⁵) and the lack of major differences between the patients, the statistical analysis might not have been powerful enough to give significant results.

As part of our ongoing lissencephaly research, we have recruited 63 unrelated patients with posteriorly predominant lissencephaly. Of these, 40 patients were found to carry either mutations or small genomic deletions, and 1 patient had a nonsense somatic mutation in the LIS1 gene. The goal of this study was to evaluate in detail the spectrum of neurological and radiological features of LIS1-related lissencephaly. In combination with molecular genetic findings, we have provided additional data concerning the relationship between the severity of the phenotype and the nature of the mutation in the LIS1 gene and compare this with the phenotypes of patients without mutations in LIS1 gene. Our data reinforce the concept that neither the mutation type nor location can predict the severity of the clinical and radiological phenotype in the LIS1-related lissencephaly gene.

Of the 63 patients with classical posteriorly predominant lissencephaly referred to our laboratory for molecular screening, 40 were identified with LIS1-related lissencephaly. Clinical data and blood samples were obtained with informed consent from parents and/or patients.

The DNA was extracted using standard protocols. Mutation analysis was performed for LIS1 for all patients with classical lissencephaly with no deletion detected by fluorescence in situ hybridization (for 17p13.3). Mutation detection was performed by direct sequencing of genomic DNA, as described previously.²³ In all patients, the mutation was confirmed to be de novo by direct sequencing of both parents' DNA.

We subsequently performed Quantitative Multiplex PCR (polymerase chain reaction) of Short Fluorescent fragments (QMPSF) in patients who had negative results of mutation screening to detect genomic deletions or duplications in the LIS1 gene, based on the simultaneous amplification of short genomic fragments using dye-labeled primers under quantitative conditions.²⁴ The PCR products were visualized and quantified as peak areas using an automated DNA sequencer with the gene-scan mode in which both peaks' heights and areas are proportional to the quantity of template present for each target sequence. We designed 2 distinct QMPSF assays that contain the following targeted exons of the LIS1 gene: assay 1 for exons 3, 4, 5, 6, 8, 9, and 11, and assay 2 for exons 2, 7, and 10. Primer pairs were designed for each of these exons to generate, in both assays, PCR fragments ranging from 75 to 360 base pairs and chosen in such a way that they do not encompass known polymorphisms. The PCRs were run on 100 ng of genomic DNA in 25 μL of a dilution with 0.2 mmol/L each of deoxynucleotide, 2 mmol/L magnesium chloride, 2.5 U of AmpliTaq Gold DNA polymerase (Applied Biosystems-Roche, Foster City, California), and 0.5 to 2 mmol/L of each primer, 1 primer of each pair carrying a 6-FAM label.

We excluded the case of somatic mosaicism to facilitate the genotype-phenotype analysis and to avoid the phenotype variation caused by such mosaicism.²¹ Patients without mutations and deletion in the LIS1 gene were also screened for TUBA1A mutations, as previously described.²⁵ The investigators were unaware of the mutation type at the time of the initial neuroimaging review.

All patients were followed up regularly in various departments of pediatric neurology and were known personally to at least 1 of the authors. Detailed information regarding family history, prenatal and perinatal events, age of onset of first seizure, motor development, cognitive function, and neurological examination were collected.

Brain magnetic resonance imaging (MRI) and computed tomographic scans were available for all patients and reviewed independently by 2 authors (N.B. and N.B.-B.). They were graded according to the following lissencephaly patterning scale: grades 1 through 6 denote the overall severity of lissencephaly and refer to the maximum number of abnormalities in the posterior region (ie, posterior to anterior gradient of lissencephaly), as seen on neuroimaging.²⁶ Cerebellar, white matter, and corpus callosum abnormalities were also assessed.

Protocols were approved by the appropriate institutional review board human committee.

Thirty-one heterozygous LIS1 mutations and 8 small genomic deletions in the LIS1 gene were identified. Of the 31 LIS1 mutations, 12 were nonsense, 8 frameshift, 6 missense, and 5 splicing defect mutations confirmed by reverse transcriptase–PCR. The mutations described here were found scattered throughout the gene (except in exons 3 and 9). All of these mutations were confirmed to be de novo by parental DNA screening. One nonsense somatic mutation (c.531G>A; p.G177X) present in 30% of the blood was also found but, unfortunately, other tissues were not available for testing.

Eight small intragenic deletions ranging from 1 exon to almost the entire gene (exons 2-11) were found by QMPSF in samples for which DNA sequencing had failed to detect any gene alterations (Figure 1).

The age range at the time of the study was 1 to 39 years (median, 6 years). The sex ratio was not evenly distributed (male: female, 18: 22). Clinical features recorded in 38 patients consisted of hypotonia and/or developmental delay in 17 patients (44.7%) and seizures in 21 patients (55.3%).

All patients except for 4 exhibited severe developmental delay (90%) with severe motor impairment including axial hypotonia and spastic quadriparesis in 24 patients (60%), virtually no language development in all cases, and moderate to severe behavioral disturbances including autistic features and sleep disorders in 12 cases (30%). Head circumference was normal in almost half of the patients, and 16 (40%) showed postnatal microcephaly.

All patients experienced seizures within the 3 first years of life, with onset before 7 months in most (30 of 36 patients; 83.3%) and infantile spasms in 28 (77.7%), either isolated or combined with various seizure types including generalized tonic seizures.

At the time of evaluation, all patients had ongoing seizures and refractory epilepsy, with up to 10 seizures daily in 19 patients (47.5%). For 21 other patients, the seizure frequency ranged from 1 per week (50%) to 1 per month (50%), despite polytherapy. Individual data are detailed in Table 1

The severity of lissencephaly, graded according to Dobyns and Truwit,²⁶ ranged from 1 to 5 (Table 1). Posterior agyria and anterior pachygyria (grade 3a) was the prominent radiological phenotype, affecting 21 patients (55.3%). Grade 2a (diffuse agyria with shallow sulci in anterior regions) was observed in 9 patients (23.7%) and grade 4a (posteriorly predominant pachygyria) in 6 (15.8%). Only 1 patient demonstrated grade 1a lissencephaly (generalized agyria). The 2 remaining patients had grade 5a (pachygyria posteriorly with subcortical band heterotopia); of these, 1 had the nonsense somatic mutation (Figure 2).

Additionally, 22 of 34 patients (64.7%) had corpus callosum abnormalities that consisted of mild hypogenesis of the rostrum and splenium, with flattening of the body in 17 and thick and dysmorphic corpus callosum in 5 others. Two patients had cerebellar hypoplasia, with brain stem hypoplasia in 1 case (Figure 3, A and C). The white matter was roughly normal, although delayed myelination (3 cases) and prominent perivascular (Virchow-Robin) spaces were observed in 23 cases. Twenty-eight patients (73.7%) had enlarged ventricles, mostly prominent posterior horns of the lateral ventricles. For the 2 remaining patients, both addressed for posteriorly predominant agyria, MRI scans were not available for our reevaluation of Dobyns grade.

First, we compared patients exhibiting missense mutations (n = 6) with those showing nonsense/frameshift mutations (n = 18; MRI data of 20 patients) and with small deletions (n = 8). It is of interest that in the nonsense/frameshift and small deletions groups, most patients had Dobyns grade 3, 11 (61.1%), and 6 (75%). When compared with the 6 patients with missense mutations, it appears that half of the patients in this group tend to have more severe lissencephaly (50%; grade 2a). However, owing to the small number of patients in this group, we cannot consider that the repartition of the lissencephaly severity score is significantly different.

Second, we divided the nonsense/frameshift mutations according to their location within the LIS1 protein, considering the boundary at the level of the end of the coiled-coil domain (exon 5), as described previously.^15,27 Mutations up to the 3′ region of the end of the coiled-coil domain are termed early nonsense/frameshift (n = 5; MRI data of 7 patients), while the remaining located in the WD40 repeats are termed late nonsense/frameshift mutations (n = 13). We found no difference between the 2 groups, most of which also had a grade of 3 (3 [50%] and 8 [61.5%]).

Finally, we compared all patients with either mutations or deletions in the LIS1 gene (n = 37) with patients without abnormalities in the LIS1 or TUBA1A genes (n = 24). Patients without mutations tended to have less severe Dobyns grades, with most patients without mutations having a grade of 4 (of 10; 41.6%), compared with 6 (16.2%) of those with mutations (Table 2).

Here, we describe the largest series of 40 patients with LIS1-related lissencephaly with detailed clinical and imaging analysis, combined with molecular studies of the LIS1 gene. Consistent with previous studies, the present findings confirm that the phenotypic variability among patients with the LIS1 mutation may not be correlated with the LIS1 mutant genotype.²¹ In addition, we observed that most patients with posteriorly predominant lissencephaly not mutated for LIS1 tend to exhibit a less severe phenotype.

LIS1-related lissencephaly is responsible for severe neuromotor impairment with moderate microcephaly and a high frequency of early seizures, mainly consisting of infantile spasms, evolving into drug resistance in half, with low interindividual variability. Most patients with LIS1 mutations have Dobyns grade 3a, consistent with previous data,^2,6,15,21 often combined with prominent perivascular (Virchow-Robin) spaces that we presume to be secondary to the migrational disorder and corpus callosum hypogenesis. It is of interest that only 1 patient had cerebellar hypoplasia reminiscent of the phenotype lissencephaly with cerebellar hypoplasia type a, suggesting that this feature is rarely associated with LIS1-related lissencephaly.²⁸

The current mutation analysis of the LIS1 gene revealed 31 intragenic mutations distributed over the entire gene, with about a third at the 5′ end and the remaining mutations located in the WD40 repeat domains encoded by exons 5 through 11. This repartition differs from the series by Uyanik et al²¹ that reported a high prevalence of mutation in exons 10 and 11, but is consistent with previous studies.^6,15 Most of the mutations (23 of 31) reported here are newly described, with few recurrent mutations, suggesting that no mutational hot spot characterizes mutations in the LIS1 gene. Six are missense, and 5 of these are newly described, with a total of 15 missense mutations identified in the LIS1 gene.^2,6,18,21 Most of them are clustered in WD2 and 3 are encoded by exons 6 and 7. The other newly described mutations (nonsense/frameshift and splicing defect mutations) follow the repartition of mutation types usually reported for the LIS1 gene. Moreover, we confirm the high prevalence of small deletions because 8 of 40 patients with typical posteriorly predominant isolated lissencephaly sequence with QMPSF gave results comparable with recent data from Mei et al,²⁴ who used a multiplex ligation-dependant probe amplification assay. In our series, of the 2 patients that showed a milder phenotype with grade 5 lissencephaly, 1 patient was found to have a somatic nonsense mutation. Somatic mutations in LIS1 are uncommon (2.5% of patients in our series and 14% in Uyanik et al²¹) and are supposed to lead to milder phenotype.¹⁹ However, our data suggest that such milder form can also be observed with germline mutation.

To provide further insight into putative genotype phenotype correlations, we examined the relationship between the mutation type and location, but no significant correlation could be observed. Several explanations can be proposed. First, in spite of detailed neurological and radiological analyses, there is little variability among patients with the LIS1 mutation who exhibit, in most cases, Dobyns grade 3 to 4, tetraplegia and refractory epilepsy. This observation, coupled with the limited size of each group studied, leads to a lack of power in the statistical tests. Second, the limits set between the early and late nonsense/frameshift mutations proposed by Cardoso et al¹⁸ and used here might not reflect the physiological functions of the LIS1 protein (PAFAH1B), which is known to be involved in multiple protein to protein interactions, playing different roles in the complex process of cortical development through participation in different protein complexes. Biochemical data suggest that all domains of the LIS1 protein are involved as the noncatalytic subunit of a heterodimeric complex that regulates brain levels of the platelet activating factor (PAFAH)²⁹ and as a microtubule-associated protein involved in cell proliferation and neuronal migration.^30-33 Moreover, data on the biochemical outcomes of mutations showed they variably affect either the stability of the LIS1 protein, its folding, and/or its interactions with its partners. The full length of the LIS1 protein is required for a correct interaction with most of LIS1 partners, in particular nuclear distribution factor E homolog 1 (NDE1)^14,34,35 and NDE-like 1 (NDEL1).³⁶ The dimerization properties of LIS1 are mediated via WD3, WD6, and the N-terminal region.³⁷ For the interaction of LIS1 with the PAFAH catalytic subunits α1 and α2, the WD domains WD2, WD3, and WD7 appear to be most important. Finally, WD2, WD5, WD6, WD7 and the N-terminal region are involved in microtubule binding.³⁷ Thus, many regions of the LIS1 protein seem critical for its multiple functions. Finally, the absence of a relationship between phenotype and genotype could be related to the nonsense-mediated mRNA decay that is suspected to play a major role in the processing of the LIS1 transcript, as suggested by Western blot.¹⁵ For this reason, early and late nonsense/frameshift mutations described here would be subjected to nonsense-mediated mRNA decay, leading to haplo insufficiency in each case.^38,39

Finally, patients without these mutations tend to have less severe lissencephaly, with most in our series having Dobyns grade 4. Recent data suggest that other genetic causes of lissencephaly could account for posteriorly predominant lissencephaly, including TUBA1A mutations. Although accurate analysis of the phenotype of the patients without the mutations allowed us to confirm that they did not demonstrate the distinctive features of TUBA1A-related lissencephaly consisting of dysmorphic basal ganglia and cerebellar hypoplasia;^25,40TUBA1A mutations were excluded in all patients without LIS1 mutations. We hypothesize that either partner of the functional cascade of LIS1 or TUBA1A genes could explain these posteriorly predominant lissencephalies.

In conclusion, with the largest series of LIS1-related lissencephaly, combined with patients found negative for LIS1 mutation and deletion, our results confirm the homogeneity of patients with LIS1-related lissencephaly, a large proportion of whom have Dobyns lissencephaly grade 3a, with no correlation with LIS1 mutations. On the other hand, patients with posteriorly predominant lissencephaly without the LIS1 mutation tend to have a less severe phenotype, suggesting either additional molecular basis or unexplored alteration in the LIS1 gene.

Correspondence: Nadia Bahi-Buisson, MD, PhD, Pediatric Neurology, Hopital Necker Enfants Malades, 149 rue de Sevres, 75015 Paris, France (nadia.bahi-buisson@nck.aphp.fr).

Accepted for Publication: November 12, 2008.

Author Contributions:Study concept and design: Saillour, Chelly, and Bahi-Buisson. Acquisition of data: Saillour, Carion, Quelin, Leger, Boddaert, Toutain, Mercier, Barthez, Milh, Joriot, des Portes, Philip, Broglin, Roubertie, Pitelet, Moutard, Pinard, Cances, Kaminska, Chelly, and Bahi-Buisson. Analysis and interpretation of data: Saillour, Quelin, Elie, Pinard, Kaminska, Beldjord, and Bahi-Buisson. Drafting of the manuscript: Saillour, Leger, and Bahi-Buisson. Critical revision of the manuscript for important intellectual content: Carion, Quelin, Boddaert, Elie, Toutain, Mercier, Barthez, Milh, Joriot, des Portes, Philip, Broglin, Roubertie, Pitelet, Moutard, Pinard, Cances, Kaminska, Beldjord, and Bahi-Buisson. Administrative, technical, and material support: Carion. Study supervision: Milh, Chelly, Beldjord, and Bahi-Buisson.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the French National Programme Hospitalier de Recherche Clinique (2003-32) and the Société d'Etudes et de Soins pour les Enfants Paralysés et Polymalformés.

Additional Contributions: The authors wish to thank all the families and clinicians whose cooperation made this study possible. The authors are grateful to Fiona Francis, PhD, for her contribution. We thank Monika Eisermann, MD, Christine Soufflet, MD, and Perrine Plouin, MD, for their helpful discussions and their assistance with the electroencephalographic analysis.