Domain specific phenotypic expansion associated with variants in MACF1

Abstract

Purpose: While heterozygous de novo missense variants in the microtubule-binding GAR domain of Microtubule-actin cross-linking factor 1 (MACF1) cause Lissencephaly 9 with Complex Brainstem Malformations [MIM #618325], the phenotypic impact of variants outside this domain remains unclear.

Methods: Through collaborative efforts, we assembled a cohort of 10 affected individuals from 8 unrelated families with either biallelic or monoallelic non-GAR domain MACF1 variants who exhibit partially overlapping yet unique phenotypic traits. Combined with previously reported cases, we analyzed genotype and phenotype data from 29 individuals using Human Phenotype Ontology (HPO)-based unsupervised hierarchical clustering.

Results: Clustering revealed two distinct phenotypic signatures, suggesting domain-specific effects. Variants outside the GAR domain associate with broader neurodevelopmental phenotypes and variable craniofacial and skeletal expressivity. Additionally, enrichment analysis (p < 0.001) using OMIM HPO sets supported these findings. In contrast to the GAR domain's strong correlation with lissencephaly and brainstem malformations, biallelic non-GAR domain MACF1 variants were linked to diverse developmental anomalies.

Conclusion: These results expand the phenotypic spectrum of MACF1-related disorders and highlight the relevance of domain-specific variant effects. Comprehensive genetic and phenotypic assessments are essential for understanding the role of MACF1 in development, informing diagnosis, and guiding future research on cytoskeletal regulation in neurodevelopment.

Keywords: Actin-microtubule cross-linking; Domain-specific phenotypes; GAR domain; Lissencephaly; MACF1 variants; Neurodevelopmental disorder; Phenotypic variability.

Figure 1:

Pedigrees and their genotypes for families with MACF1 variants. Standard pedigree structures are utilized—filled circles and squares denote clinically affected individuals, and probands are indicated by black arrows. Nine families with 11 individuals affected with MACF1-related neurodevelopmental disorders that are identified through the BHCMG/BCM-GREGoR, the Baylor Genetics (BG) clinical diagnostic laboratory database, Undiagnosed Diseases Network (UDN), and GeneMatcher. All participants were evaluated by a clinical geneticist and/or neurologist. Detailed pedigrees and phenotypic data were collected from collaborating clinicians using a standardized template.

Figure 2:

Schematic diagram of MACF1 protein structure (A) Conceptual representation of MACF1 protein structure produced by translation of MANE transcript NM_001394062.1 with mapping location (circles) of the amino acid variants observed in this study. In total, we identified 23 different SNVs and indels that are distributed throughout the protein. A larger deletion (39.6kb) is shown in a dark blue line. Homozygous missense variants are represented in orange circles, while heterozygous missense variants are denoted in yellow. Greenish blue circles represent compound heterozygous variants, and the black circle is for heterozygous frameshift variant observed in this study. (B) Annotation grid of phenotypes highlighting domain specificity. Columns represent individual cases, mapped to the MACF1 protein based on the identified variant. Phenotype terms are listed in each row. Variants within the GAR domain are associated with Lissencephaly and cerebellar hypoplasia, whereas variants outside the GAR domain correspond to a broader neurodevelopmental phenotype, including craniofacial and skeletal abnormalities.

Figure 3:

Quantitative phenotypic analyses. (A) Phenotypic similarity heatmap for individuals with MACF1 variants. A heatmap was generated using proband phenotypic similarity scores and ordered based on Hierarchical Agglomerative Clustering of proband phenotypic similarity. Dendrograms showing clusters are present at the left and top sides of the heatmap. Highest phenotypic similarity score of 1 is shown in red that runs along as the diagonal. Two distinct clusters are noted suggesting clinical subgroups within MACF1-related disorder. (B) Significant HPO terms differentiating the 2 clusters is plotted. Applying a generalized linear model (GLM) using a quasipoisson distribution, unique HPO terms significantly associated with the two specific clusters were identified. Percentages of cases in each cluster for each of the significant HPO term is denoted in the columns.

Figure 4:

Comparative phenotypic analysis using OMIM gene term sets. Phenotypic similarity heatmap of individuals with MACF1 variants, analyzed alongside OMIM gene term sets (p < 0.001), reveals two distinct clustering patterns. Cluster 1 comprises cases with MACF1 variants outside the GAR domain, along with NCAPD3, which are linked to microcephaly and global developmental delay as shared phenotypes. Cluster 2 includes cases with MACF1 variants within the GAR domain, as well as OMIM genes associated with lissencephaly and cerebellar hypoplasia, consistent with microtubule-binding defects.

Figure 5:

Comparative phenotypic analysis using OMIM disease term sets. Phenotypic similarity heatmap of individuals with MACF1 variants, analyzed alongside OMIM disease term sets (p < 0.001), reveals four distinct clustering patterns. Cluster 1 includes cases with GAR domain variants that are closely aligned with Lissencephaly and Cortical Dysplasia, Complex, with Other Brain Malformations phenotypes. Cluster 2 includes de novo heterozygous cases with frameshift mutations or large deletions near the GAR domain, clustering with intellectual disability, developmental delay, and epilepsy phenotypes. Cluster 3 comprises cases previously described with congenital myasthenic syndrome, which align with myasthenic syndrome phenotypes. Cluster 4 consists of cases from our cohort without a defined OMIM disease phenotype, suggesting that these cases represent broad neurodevelopmental phenotypes.