Comprehensive genotype-phenotype analysis in POLR3-related disorders

Abstract

RNA polymerase III (RNA Pol III)-related disorders (POLR3-RDs) are a group of clinical entities characterized by causal variants in genes encoding RNA Pol III subunits, including POLR3A, POLR3B, POLR1C, POLR1D, POLR3D, POLR3E, POLR3F, POLR3GL, POLR3H, and POLR3K. These typically cause developmental phenotypes affecting the central nervous system; the eyes; connective tissues including bones, teeth, and endocrine axes; and the reproductive system. Similar phenotypes can be caused by variants in separate subunit genes (multigenic). In contrast, variants in the same gene can cause different phenotypes (pleiotropy), making genotype-phenotype correlation challenging. POLR3-RDs, though individually rare, have never been analyzed collectively. To bridge this gap, we developed an extensive database encompassing all published and unpublished cases of POLR3-RDs and conducted the first comprehensive genotype-phenotype correlation study across their entire spectrum. This work contributed new cases, representing 13% of all documented cases in the literature, along with 31 novel variants, accounting for 8% of all identified variants. This database was constructed by systematically reviewing the literature and integrating data from patients under the care of our international network of collaborators. The dataset includes genotype curation, bioinformatics, prior publications, and individual patient outcome information. By leveraging these comprehensive data, we were able to establish clear genotype-phenotype correlations for some pathogenic variants, which will help provide optimal clinical care and genetic counseling (including insights into disease phenotypes and progression) and offer valuable guidance for future clinical trial design and patient stratification.

Keywords: 4H leukodystrophy; POLR3-HLD; POLR3-RD; POLR3-related disorders; POLR3-related leukodystrophy; RNA Pol III; RNA polymerase III; genotype-phenotype correlations; hypodontia; hypogonadotropic hypogonadism; hypomyelination.

Figure 1

Assembly of patient data by collation of affected individuals from the literature with internal database entries The literature was searched using Medline, Embase, and Web of Science for literature pertaining to POLR3∗ genes and disease until October 31, 2024. The search details can be reviewed in Figure S1. Source data from the primary search yielded 1,422 records, from which 195 relevant records were filtered. Internally, we started with 333 patient files. These were collated together and reviewed for duplicates through a process of matching patients and previous publication records. The process was conservative, meaning that all patient entries are unique and mapped correctly to prior publications, but several patients that were potential duplicates have been excluded as a result of having the same variants and associated clinical data (e.g., birth year, sex, and/or case description).

Figure 2

POLR3-RD database breakdown and summary of affected individuals Patients with POLR3-RD in the analysis cohort are broken down by their causal genes and their status in the literature or as assessed by our team (for previously unpublished patients). Numeric values represent frequencies (counts) and percentages of the total. (A) Total patients with POLR3-RD, including published and previously unpublished patients broken down by their causal genes. (B) Previouly published affected individuals broken down by their causal genes. (C) Total variants in the literature regardless of publication status, broken down by their causal genes. (D) Novel variants that have not yet been published, broken down by genes affected. (E) POLR3-RD affected individuals grouped by diagnosis category.

Figure 3

Structural relationships of missense variants associated with specific diagnostic categories at the coding sequence and protein levels (A–C) Lollipop charts of the most frequently mutated genes, POLR3A, POLR3B, and POLR1C, demonstrating that missense variants typically occur throughout these genes and do not show spatial bias at the level of messenger RNA. (D–J) Two-dimensional dendrogram representation of the apo protein structure (PDB: 7A6H) reveals that residue positions affected by missense variants causing non-HLD outcomes are sometimes associated with spatial bias/clusters (e.g., ASDN) but often occur in clusters with promiscuous variants. (A) POLR3A missense variants occur throughout the coding sequence and are associated with multiple diagnoses. POLR3A missense variants causing specific disease outcomes have low spatial bias. The most frequently observed POLR3A missense variants are c.2554A>G (p.M852V), associated with multiple outcomes, and c.2015G>A (p.G672E), associated uniquely with HLD outcomes. (B) POLR3B missense variants occur throughout the coding sequence and are associated with multiple diagnoses. POLR3B missense variants causing specific disease outcomes have low spatial bias. The most frequently observed POLR3B missense variants are c.1568T>A and c.1244T>C (p.M415T), both associated with multiple outcomes. (C) POLR1C missense variants occur throughout the coding sequence and are associated primarily with HLD outcomes. The most frequently observed POLR1C missense variants are c.836G>A (p.R279Q) and c.193A>G (p.M65V). The former is associated with a position affected in multiple disease outcomes. (D) A two-dimensional circular dendrogram representation of the apo RNA Pol III structure developed by Girbig et al. (PDB: 7A6H) was assembled using alpha carbon positions of individual residues, and clustering was done using Ward’s minimum variance method. As pictured, variants closer to each other along the perimeter of the dendrogram are more closely associated in the three-dimensional RNA Pol III structure. Cluster height is represented as (r = log2(cluster height)), and increasingly peripheral nodes merge clusters with lower spatial variance (i.e., residues that are closer together merge at more peripheral nodes). Residues of specific RNA Pol III subunits are colored according to the legend pictured on the right of the diagram. (E) RNA Pol III residues affected by missense variants implicated in HLD occur broadly throughout the polymerase and display low spatial bias. (F) RNA Pol III residues affected by missense variants implicated in WRS are rare. Many occur in clusters populated by residues that are also implicated in association with multiple outcomes. (G) RNA Pol III residues affected by missense variants implicated in HSP/SA often occur in clusters populated by residues that are also implicated in association with multiple outcomes. (H) Missense variants causing ASDN occur in POLR3B. These variants typically fall into one of three unique spatial clusters of POLR3B residues that are thought to be involved in nucleic acid interactions. (I) Missense variants associated with SVF outcomes are typically found in clusters populated by residues that are also implicated in association with multiple outcomes. (J) Residue positions affected by missense variants causing multiple outcomes (i.e., promiscuous variants) are pictured. Underlying data for D-J can be found in Table S6.

Figure 4

Hypomyelinating leukodystrophy variant frequency by gene (n) and type (proportion) Frequency (no. of unique variants) and relative proportion (percentage) of types of variants among patients with POLR3-related hypomyelinating leukodystrophy. Variants can have multiple type classifications; the relative proportion represents the frequency of variants of a particular gene and type divided by the total number of instances of classification for that gene. (A) HLD is associated with 121 unique POLR3A variants in the database, of which the majority are missense variants (n = 88). The next most frequent variant types are splice (n = 16) and stop-gain (n = 12) variants. (B) POLR3B is associated with 97 unique variants. The most frequent are missense (n = 57) followed by splice (n = 19) and stop-gain (n = 12) variants. (C) POLR1C variants (n = 43) are mainly missense (n = 35). The next most frequent are splicing (n = 4) and frameshift (n = 4) variants. (D) POLR3D variants (n = 2) include an intronic variant in a polypyrimidine tract and another that is a missense variant. (E) POLR3K is known for two missense variants and a large deletion involving the entirety of exon 3 (the final exon), predicted to cause frameshift/stop gain. The underlying basis of the low diversity among POLR3D and POLR3K variants is the low frequency of total observations in the population and not a relationship between these subunits and a particular type of variant.

Figure 5

Variant frequency by gene (n) and type (proportion) among other presentations of POLR3-related disorders Frequency (no. of unique variants) and relative proportion (percentage) of types of variants among patients with POLR3-related disorders. Variants can have multiple type classifications; the relative proportion represents the frequency of variants of a particular gene and type divided by the total number of instances of classification for that gene. (A) HSP/SA is associated with 57 unique POLR3A variants. Observations of POLR3A missense variants (n = 22) are proportionally lower among patients with HSP/SA than among patients with HLD, while stop-gain variants (n = 15) are more frequent overall. The next most frequent variants among POLR3A are frameshift (n = 11) or splicing (n = 11) variants. (B) SVF is associated exclusively with POLR3A variants (n = 40), and all cases have at least one allele that is either POLR3A c.1771−6C>G or POLR3A c.1771−7C>G. Examination of the total variant population reveals that most non-canonical variants are missense (n = 19), splicing (n = 7), stop gain (n = 6), or frameshift (n = 5). These exemplify typical variants found in trans in patients with SVF. (C) Unique POLR3A variants (n = 32) associated with WRS are typically missense (n = 10), splice (n = 10), or stop gains (n = 6). (D) There are 3 unique POLR3B missense variants associated with WRS. (E) There are 22 unique POLR3B missense variants associated with cases of ASDN. These are all inherited de novo, making them unique as compared to all other variants causing POLR3-RD.