This study involved two men from two unrelated Chinese families, with duplication of exons 56–61 in the DMD gene, which was confirmed by multiplex ligation-dependent probe amplification (MLPA). Informed consent was obtained from each participant to perform the investigation and genetic studies.

Genomic DNA (gDNA) was extracted from blood samples using standard procedures. MLPA was performed to detect the copy number of SMN1 (OMIM #600354) using the SALSA P060 SMA Kit and DMD using the SALSA P034/P035 DMD Kit (MRC Holland, Amsterdam, the Netherlands).

Intragenic mutation screening in SMN1 (exon 1–8) was performed by long-range PCR (LR-PCR) and nested PCR as described previously in our laboratory (Kubo et al., 2015; Sun et al., 2020). LR-PCR and nested PCR were amplified using KOD FX Neo Polymerase (TOYOBO, Osaka, Japan) and 2× Taq PCR Mastermix (TIANGEN, Beijing, China), respectively. The PCR products were confirmed by 1.5% agarose gel electrophoresis and sequenced on the ABI 3130 Genetic Analyzer using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Waltham, USA).

Genomic DNA was extracted and libraries were prepared using Illumina library construction and capture kits (Illumina, San Diego, USA) according to standard instructions (Li et al., 2021). 100× coverage of pair-end reads were obtained on an Illumina Novaseq 6,000 Sequencer (Illumina, San Diego, USA) and mapped to the hg19/GRCh37 human reference genome. Exonic and splice site variants with a minor allele frequency of less than 0.01 in public databases (dbSNP database, gnomAD, Exome Aggregation Consortium, and 1,000 Genomes) were selected. Copy number variants analysis was conducted by comparing the coverage of depth between the target sample and other male control samples in the same pipeline (Zhai et al., 2021).

DNA was extracted from the fresh peripheral blood, electrophoresed to confirm its integrity, and measured using Nanodrop 2000 and Qubit 4.0 to confirm its quantity (Thermo Fisher, Massachusetts, USA). Nanopore LRS was performed using the SQK-LSK109 Kit with R9.4 flow cells on GridION X5 (ONT, Oxford, UK) according to the manufacturer’s instructions (Miao et al., 2018).In brief, 5 μg DNA was sheared to ∼ 5–25 kb fragments using Megaruptor (B06010002, Diagenode, Liège, Belgium) and size-selected to 10–30 kb with a BluePippin (Sage Science, Beverly, USA). After end repair and dA-tailing of DNA fragments, a SMRTbell™ library was purified and sequenced on R9.4 flowcells using GridION X5. The sequencing generated 2,418,823 base-called reads containing 43,641,144, and 812 bases with an average read length of 18,042 bp. The long reads were aligned to the GRCh37/hg19 reference genome using NGM-LR v0.1.4 (https://github.com/philres/nextgenmap-lr) with default parameters. Structural variations (SVs) were called by Sniffles (version 1.0.11) (Sedlazeck et al., 2018). Ribbon and IGV were used to visualize the alignment results.

Genomic DNA was sequenced using PacBio SMRT target sequencing of the whole DMD gene (Grandomics, Beijing, China) according to the standard manufacturer’s conditions. To cover the DMD gene, DNA probes of 120 bases were designed and synthesized by Boke Biotechnologies (Boke, Beijing, China). Probes corresponding to repetitive sequences were removed using the Repeat Masker dataset during the design of the probes. The PacBio SMRT sequencing library was constructed using a Template Prep Kit (PacBio, Menlo Park, USA) according to the instructions. 3 μg genomic DNA was sheared to 1∼6 kb fragments by a g-Tube (#520079; Covaris, Bankstown, Australia) centrifugation. A SMRTbell™ library was prepared and sequenced on R9.4 flowcells using GridION X5. The SMRTlink 8.0 (PacBio) was used to remove low-quality reads and adapters resulting from raw sequencing data. Circular Consensus Sequence (CCS) reads were generated using the PacBio SMRT analysis software, and barcode splitting was performed by Lima. We used PBMarkDUP (PacBio) to remove potential copies in CCS reads and PBMM2 (https://github.com/PacificBiosciences/pbmm2) for comparing CCS reads to the reference genome hg19. SVs were detected using by PBSV (V9.0, https://www.pacb.com/support/software-downloads/). SVs were annotated by Annovar (http://nar.oxfordjournals.org/content/38/16/e164).

The breakpoints of the DMD gene identified by LRS were confirmed by PCR and Sanger sequencing. Junction fragments were amplified from gDNA using 2× Taq PCR Mastermix (TIANGEN, Beijing, China). The primers used for sequencing were designed using Gene Tool and listed in Supplementary Table S4. The breakpoint sequences were confirmed by Sanger sequencing and aligned to the reference genome hg19 using the BLAT tool (UCSC).

We manually analyzed the flanking sequence of each breakpoint and identified microhomology. We selected 100bp reads upstream and downstream of each breakpoint to search for repetitive elements using the “Repeat Masker” program of the UCSC Genome Browser.

Family 1 was a non-consanguineous couple that visited our hospital because they gave birth to a female patient (IV:2) diagnosed with spinal muscular atrophy (SMA1; OMIM #253300) (Figure 1A). Molecular genetic testing of SMA in the IV:2 female and ECS in the couple (III:1 and the III:2) were performed simultaneously in order to reduce the risk of having an affected child associated with genetic disease. Compound heterozygous mutations, exons 7–8 deletion, and c.835G > C p(G279R) in the survival motor neuron (SMN1) gene were detected in the IV:2 female, which were inherited from her parents (Figure 1C, Supplementary Figure S1, S2). The quality control of the WESdata is summarized in Supplementary Table S1. Furthermore, no other pathogenic variant was found in the couple by single-nucleotide variants and indel analysis of whole-exome sequencing (WES). However, duplication of exons 56–61 in the DMD gene was found in the Ⅲ:2 male (Figure 1A) by WES-based CNV analysis (Supplementary Figure S3), which was confirmed by MLPA (Figure 1D). Since he was asymptomatic and there was no family history of DMD, other members of family 1 were tested (Figure 1A, Ⅱ:1, Ⅱ:3, and Ⅱ:9). MLPA analysis identified the same duplication in the mother (Ⅱ:1) and uncle (Ⅱ:9) of the male (Figure 1D). The man (Ⅲ:2) and his uncle (Ⅱ:9) were 33 and 44 years old, respectively; they had slightly elevated serum creatine kinase (CK) concentration (288 IU/L and 204 IU/L, reference range: 24–195 IU/L) and normal surface electromyography of the flexor and extensor muscles (EMG).

In family 2, the 23-year-old man (Ⅱ:6) was presented with a typical DMD (Figure 1B). He was unable to run or jump and became wheelchair dependent at the age of eleven. His CK concentration was 12,696 IU/L, which was 60 times higher than the upper normal limit. He had been dependent on 24 h ventilation for 7 years before the visit. The duplication of exons 56–61 in DMD was identified in the Ⅱ:6 male patient by MLPA analysis, which was inherited from his mother.

The results of Nanopore LRS and PacBio SMRT target sequencing in the male of family 1 revealed complex structural variations (SVs) on chromosome X, comprising a 251.4 kb inversion duplication (INVDUP, DMD gene); three duplications of 2.3, 13.5, and 66.6 kb (CFAP47 gene); and a 4.9 kb deletion. Sanger sequencing revealed that the 5′ and 3’ breakpoints of INVDUP were located at chrX:35945179 and chrX:35854874, respectively, which indicated that the duplication occurred outside the DMD gene (Table 1; Figure 2; Supplementary Figure S4–7; Supplementary Table S2-3). In the patient of family 2, PacBio SMRT target and Sanger sequencing confirmed that the duplication was a tandem repeat and that the junction of breakpoint was located at chrX:31567961 (Table 1; Figure 2; Supplementary Figure S4–7).

To determine the mechanism by which SVs were formed, the mutational signatures of the breakpoint junctions were investigated. The microhomology analysis performed between the paired flanking sequences identified 4 bp microhomology “TTGC” and 27 bp microhomology “CTG​CCT​CAG​CCT​CCC​GAG​TAG​CTG​GGA” were identified in the junctions of INVDUP and 2 bp microhomology “AT” in the junction of tandem repeat (Supplementary Figure S5). To obtain sufficient information, we used 100 bp flanking reads and performed the motif analysis. We found that these segments around the breakpoints are considered to be derived from repeating elements overlapped with short interspersed nuclear elements (SINEs, Alu family) and long interspersed nuclear elements (LINEs, L1 family) in the asymptomatic male (Ⅲ:2, family 1) and low-complexity repeats (AT rich) in the Ⅱ:6 male from family 2 (Table 1, Supplementary Figure S8).