Coupling deep phenotypic quantification with next-generation phenotyping for 192 individuals with germline histonopathies

Abstract

Mendelian histonopathies are rare neurodevelopmental disorders (NDDs) caused by germline variants in histone-encoding genes. Here, we perform a more expansive pan-histonopathy interrogation than previously possible. We analyze data from 192 individuals affected by histonopathies. This analysis includes representation of the 185 published individuals with HIST1H1E syndrome, Bryant-Li-Bhoj syndrome, and Tessadori-Bicknell-van Haaften NDD; as well as from seven unpublished individuals, five of whom harbor variants in genes not previously associated with disease (HIST1H2AL/H2AC16, H2AFZ/H2AZ1, HIST1H3D/H3C4, and HIST3H3/H3-4). By intersecting clinician-reported phenotypic data with next-generation phenotyping of published 2D facial photographs (n = 98), we sought to address the lack of established craniofacial gestalts or characteristic phenotypic patterns for this community. While these analyses may suggest a histone core versus linker protein basis of delineation, they more strikingly highlight data gaps that confound the identification of phenotypic patterns at this time. Based on this, we developed an updated standardized clinical survey, which allowed us to identify the second known individual with a germline histonopathy and a cancer diagnosis. Notably, the community-wide cancer incidence is currently 1%, which falls below the recommended 5% cut off for routine surveillance. Ultimately, this work highlights the ways in which histonopathy-associated phenotypes change throughout the lifespan, necessitating longitudinal re-evaluation; that every identified individual shapes our understanding of these syndromes in a way that improves care for this community; and the value of ongoing translational work to address the outstanding question of cancer predisposition for individuals living with germline histonopathies.

Keywords: Mendelian genetics; comorbidity; histones; neurodevelopmental disorders; next-generation phenotyping; translational genetics.

Figure 1

Germline histonopathy-implicated genes (A) Schematic overview of the different classes of histone proteins. (B) Genes encoding the different classes of histone proteins (color-coded based on A), subdivided based on whether a gene encodes an RC histone (top) or RI histone (bottom). Purple highlight indicates genes reported in this paper as linked to germline histonopathies. Bolded are previously unreported. (C) Integration of gnomAD constraint metrics and Monarch Initiative data on model systems with phenotypes when the human gene orthologue is perturbed. Unfilled fields indicate that data are not available. Gray = HIST1H1E syndrome-associated gene; green = BLBS-associated gene; blue = TBvH NDD-associated gene. Deep orange indicates constraint metric values flagged in gnomAD to be statistically significant. Pale orange indicates constraint metric values with a Z score ≥2 away from the mean. Abbreviations: RC = replication-coupled; RI = replication-independent; Sc = Saccharomyces cerevisiae; Ce = Caenorhabditis elegans; Dr = Danio rerio; Dm = Drosophila melanogaster; Mm = Mus musculus.

Figure 2

Quantification of clinician-reported craniofacial data Color-coding for (A)–(C): darkest color indicates a clinician-reported response of yes, intermediate color indicates a clinician-reported response of no, and palest color indicates that a feature was either not reported or not assessed. (A) Phenotypic analysis of all 191 affected individuals, subdivided by neurologic (N) features, non-neurologic systemic (S) features, growth (G) features and craniofacial (CF) features. Purple color indicates analysis unbiased by OMIM classification (pan-histonopathy). (B) Analysis of published phenotypes for affected individuals diagnosed with HIST1H1E syndrome (H1, gray), BLBS (H3.3, green), or TBvH NDD (H4, blue) (color-coded based on Figure 1A). Interrogation subdivided by neurologic (N) features, non-neurologic systemic (S) features, growth (G) features, and craniofacial (CF) features. (C) Reporting of phenotypes for previously unreported affected individuals. Top row includes the identifier for the affected individuals as well as their affected gene. First column lists prevalence of a feature across individuals previously reported.

Figure 3

GMDB inter-cohort analysis Inter-cohort analysis using mean pairwise distance distribution comparing images from individuals with the same syndrome (blue), different syndrome (red), and respective histonopathies (orange). Each row shows a pair of comparison, such as HIT1H1E/H3.3, HIST1H1E/H4, and H3.3/H4. Each column shows the results at different age groups such as age younger than 5 and 5 years old (A), age between 6 and 12 years old (B), and age older than 12 years old (C). The x axis is the distance. Therefore, small distance indicates high similarity. The black vertical line is the threshold that classifies whether it is the same disorder or different disorders.

Figure 4

GMDB craniofacial analysis of previously published images GestaltMatcher analysis of published photos of individuals with germline histonopathies stratified by age. Pairwise rank dendrograms for 5 years and younger (A); 6–12 years old (B); 13 years and older (C). Pairwise rank dendrograms are annotated with cluster 1 (black/gray) and cluster 2 (purple), cluster 3 (bright purple). The pairwise rank matrix represents the similarity rankings of images, where each column corresponds to a specific test image and each row shows the rank of similarity for the remaining images. Labels indicate the image ID from GMDB and the group classification (e.g., HIST1H1E, H3.3, or H4). The hierarchical clustering highlights phenotypic patterns and relationships among the patient groups. Bar charts report distribution of OMIM-characterized syndromes in each age group (total possible), in each cluster, and in the unclustered images.

Figure 5

Quantification of non-craniofacial clinician-reported phenotypes Color-coding for (A)–(C): darkest color indicates a clinician-reported response of yes, intermediate color indicates a clinician-reported response of no, and palest color indicates that a feature was either not reported or not assessed. (A) Phenotypic analysis of all 191 affected individuals, subdivided by neurologic (N) features, non-neurologic systemic (S) features, growth (G) features, and craniofacial (CF) features. Purple color indicates analysis unbiased by OMIM classification (pan-histonopathy). (B) Analysis of published phenotypes for affected individuals diagnosed with HIST1H1E syndrome (H1, gray), BLBS (H3.3, green), or TBvH NDD (H4, blue) (color-coded based on Figure 1A). Interrogation subdivided by neurologic (N) features, non-neurologic systemic (S) features, growth (G) features, and craniofacial (CF) features. (C) Reporting of phenotypes for previously unreported affected individuals. Top row includes the identifier for the affected individuals as well as their affected gene. First column lists prevalence of a feature across individuals previously reported. (D) Mixed qualitative/quantitative longitudinal assessment of head size, limited to affected individuals for whom >1 measurement was available. 1 = clinician-reported macrocephaly (HP:0000256), 0 = clinician-reported head size within expected range for age and sex, −1 = clinician-reported microcephaly (HP:0000252).

Figure 6

H3-3B in cancer (A) A map of all the known H3-3B germline variants (top) and H3-3B somatic mutations aggregated in cBioPortal or PeCan (bottom). Yellow germline variant highlights the individual with both a BLBS diagnosis and a history of an ACTH-producing carcinoid tumor (HP:0030445) of the lung. Yellow somatic mutations highlight residues affected in lung tumors (HP:0100526). The red somatic mutation indicates the putative driver mutation in an atypical lung tumor (HP:0100526) reported in cBioPortal. (B) Tumor types ranked by the percentage of cases with somatically mutated H3-3B (for alteration frequency ≥1) based on data aggregated in cBioPortal. Yellow = lung cancers. (C) Co-occurrence or mutual exclusivity log2 odds ratio of somatic H3-3B mutations with somatic mutations in genes frequently reported to harbor driver mutations in solid tumors (sourced from PeCan), including genes with co-occurring mutations in high-grade gliomas (HP:0009733); in chromatin-remodeling genes found to be recurrently mutated in pulmonary carcinoid tumors (HP:0030445) etiologically similar to the one resected from individual BLBS-97; or genes implicated in the clonal expansion of solid tumors. Circles = co-occurring; boxes = mutually exclusive.

Figure 7

Existing data support the interrogation of cancer risk in individuals with germline histonopathies (A) Some of the observed phenotypes in individuals with all three OMIM-identified germline histonopathies, including overgrowth and premature aging, overlap with phenotypes that are cardinal features of cancer-predisposition syndromes. (B) Review of the existing functional evidence performed in dermal fibroblasts from affected individuals and zebrafish models of germline histonopathies motivate exploration of cancer as a comorbidity (left) and are potentially consistent with the two cancer diagnoses reported in the 191 person pan-histonopathy cohort (right).