Single-molecule optical mapping enables quantitative measurement of D4Z4 repeats in facioscapulohumeral muscular dystrophy (FSHD)

Abstract

Purpose: Facioscapulohumeral muscular dystrophy (FSHD) is a common adult muscular dystrophy. Over 95% of FSHD cases are associated with contraction of the D4Z4 tandem repeat (~3.3 kb per unit) at 4q35 with a specific genomic configuration (haplotype) called 4qA. Molecular diagnosis of FSHD typically requires pulsed-field gel electrophoresis with Southern blotting. We aim to develop novel genomic and computational methods for characterising D4Z4 repeat numbers in FSHD.

Methods: We leveraged a single-molecule optical mapping platform that maps locations of restriction enzyme sites on high molecular weight (>150 kb) DNA molecules. We developed bioinformatics methods to address several challenges, including the differentiation of 4qA with 4qB alleles, the differentiation of 4q35 and 10q26 segmental duplications, the quantification of repeat numbers with different enzymes that may or may not have recognition sites within D4Z4 repeats. We evaluated the method on 25 human subjects (13 patients, 3 individual control subjects, 9 control subjects from 3 families) labelled by the Nb.BssSI and/or Nt.BspQI enzymes.

Results: We demonstrated that the method gave a direct quantitative measurement of repeat numbers on D4Z4 repeats with 4qA allelic configuration and the levels of postzygotic mosaicism. Our method had high concordance with Southern blots from several cohorts on two platforms (Bionano Saphyr and Bionano Irys), but with improved quantification of repeat numbers.

Conclusion: While the study is limited by small sample size, our results demonstrated that single-molecule optical mapping is a viable approach for more refined analysis on genotype-phenotype relationships in FSHD, especially when postzygotic mosaicism is present.

Keywords: D4Z4; FSHD; facioscapulohumeral muscular dystrophy; macrosatellite; single-molecule optical mapping.

Figure 1

An overview of the genomic architecture of segmental duplications at the chromosome 4q35 region and the 10q26 region. In the GRCh38 reference genome, 4q35 incorrectly shows two D4Z4 arrays (8 units 4qB type and 1.5 unit 4qA type) with a 50 kb gap between AC225782.3 (4qB) and AC215524.3 (4qA) (panel A). 10q26 incorrectly shows two D4Z4 arrays (each with 7 D4Z4 repeat units) with a 50 kb gap between AL845259.22 (10qA) and AL7323751.8 (10qA) (panel B). Two new patch scaffold sequences were recently added in CRCh38 patch 7 with 4qA configuration (KQ983257.1) without gap and with 4qA-L configuration (KQ983258.1) without gap, and we additionally illustrated them (panel C and D). The segmental duplication (green boxes) in 4q35 has high sequence identity with the corresponding region in 10q26, KQ983257.1 and KQ983258.1, while the distal D4Z4 array separated by the gap in 10q26 is marked with an orange box. The incorrect assembly gap is not present in 4q35 (panel E) or 10q26 (panel F) in the GRCh37 genome assembly.

Figure 2

Illustration of the region with label dissimilarity between 10q26, 4q35, KQ983257.1 and KQ983258.1 (blue box) adjacent to the region of similarity (green box), based on in silico analysis on GRCh37, GRCh38, KQ983257.1 and KQ983258.1. Red vertical bars represent labels of enzyme recognition sites. By using fragments that spans the region of dissimilarity, we can confidently separate fragments originating from 10q26, 4q35 or those that are undetermined (uninformative). The panel A and B represent labels generated by the Nb.BssSI and Nt.BspQI enzymes, respectively. Although the reference genome GRCh38 contains two labels (repeat unit #2 and #5) within the D4Z4 repeat region for the Nt.BspQI enzyme (red box), we rarely observe them in real data, possibly due to the inclusion of a very rare allele in the GRCh38 or due to errors in genome assembly.

Figure 3

Molecular diagnosis of facioscapulohumeral muscular dystrophy by Southern blot on samples from cohort 1. The results for cohort 2 is available in online supplementary figure 1. E/H and p13E-11: double digested with EcoRI/HindIII and then labelled with probe p13E-11, and all the 4q and 10q segments are illustrated. E/B and p13E-11: double digested with EcoRI/BlnI and then labelled with probe p13E-11, and the 10q segments are digested so only 4q segments are illustrated. H and 4qA: digested with HindIII and then labelled with probe 4qA, and the 4qA alleles are illustrated. H and 4qB: digested with HindIII and then labelled with probe 4qB, and the 4qB alleles are illustrated. The asterisk ‘*’ denotes pathogenic allele with <10 repeat units and with a 4qA configuration. The plus sign ‘+’ denotes somatic mosaic allele.

Figure 4

Determination of somatic mosaicism by Nb.BssSI (panel A/B/C) and Nt.BspQI (panel D/E/F) enzymes on a patient with 2, 15 and 27 repeats (ID: P06). For both enzymes, our computational pipeline accurately identified the presence of somatic contraction, and determined that the contraction occurs on the parental allele carrying the 4qA configuration. For repeat quantification using the Nt.BspQI enzyme, mean±SD is annotated in the figure. Vertical bars represent labels of enzyme recognition sites.