Cross-sectional Neuromuscular Phenotyping Study of Patients With Arhinia With SMCHD1 Variants

Abstract

Background and Objectives

Facioscapulohumeral muscular dystrophy type 2 (FSHD2) and arhinia are 2 distinct disorders caused by pathogenic variants in the same gene: SMCHD1. The mechanism underlying this phenotypic divergence remains unclear. In this study, we characterize the neuromuscular phenotype of individuals with arhinia caused by SMCHD1 variants and analyze their complex genetic and epigenetic criteria to assess their risk for FSHD2.

Methods

Eleven individuals with congenital nasal anomalies, including arhinia, nasal hypoplasia, or anosmia, underwent a neuromuscular examination, genetic testing, muscle ultrasound, and muscle MRI. Risk for FSHD2 was determined by combined genetic and epigenetic analysis of 4q35 haplotype, D4Z4 repeat length, and methylation profile. We also compared expression levels of pathogenic DUX4 mRNA in primary myoblasts or dermal fibroblasts (upon myogenic differentiation or epigenetic transdifferentiation, respectively) in these individuals vs those with confirmed FSHD2.

Results

Among the 11 individuals with rare, pathogenic, heterozygous missense variants in exons 3–11 of SMCHD1, only a subset (n = 3/11; 1 male, 2 female; age 25–51 years) met the strict genetic and epigenetic criteria for FSHD2 (D4Z4 repeat unit length <21 in cis with a 4qA haplotype and D4Z4 methylation <30%). None of the 3 individuals had typical clinical manifestations or muscle imaging findings consistent with FSHD2. However, the patients with arhinia meeting the permissive genetic and epigenetic criteria for FSHD2 displayed some DUX4 expression in dermal fibroblasts under the epigenetic de-repression by drug treatment and in the primary myoblasts undergoing myogenic differentiation.

Discussion

In this cross-sectional study, we identified patients with arhinia who meet the full genetic and epigenetic criteria for FSHD2 and display the molecular hallmark of FSHD—DUX4 de-repression and expression in vitro—but who do not manifest with the typical clinicopathologic phenotype of FSHD2. The distinct dichotomy between FSHD2 and arhinia phenotypes despite an otherwise poised DUX4 locus implies the presence of novel disease-modifying factors that seem to operate as a switch, resulting in one phenotype and not the other. Identification and further understanding of these disease-modifying factors will provide valuable insight with therapeutic implications for both diseases.

Facioscapulohumeral muscular dystrophy (FSHD) is one of the most common muscular dystrophies.¹ Burst-like expression of the typically silent DUX4 retrogene in a small fraction of skeletal myocytes perturbs a diverse set of molecular pathways and triggers myotoxicity,^2-6 leading to the prototypical clinical phenotype of FSHD.

Two broad genetic mechanisms lead to de-repression of DUX4 and subsequent muscle toxicity in FSHD. The most common (∼95%) genetic cause of FSHD, referred to as FSHD type 1 (FSHD1; OMIM 158900), is the contraction of the 4q35 macrosatellite D4Z4 repeat array from the normal 11–100 repeat units (RUs) to between 1 and 10 RUs.⁷ These contractions of the D4Z4 array lead to chromatin relaxation, hypomethylation, and inappropriate transcription of DUX4 from the most telomeric D4Z4 repeat unit.⁸ Only when the contracted D4Z4 repeat array is in cis with a permissive (4qA) haplotype that contains the necessary DUX4 polyadenylation signal is a stable, full-length DUX4 mRNA transcript (DUX4-fl) produced. Once translated, the DUX4 protein product accumulates and is myotoxic. The second de-repression mechanism is digenic and accounts for the remaining (∼5%) FSHD cases: FSHD type 2 (FSHD2; OMIM 158901). Missense and loss-of-function variants in genes encoding epigenetic repressors and cofactors acting at the 4q35 D4Z4 array, including SMCHD1,⁹ DNMT3B,¹⁰ and LRIF1,¹¹ result in chromatin relaxation and hypomethylation, which in turn increase the probability of DUX4 transcription. Thus, they can act as disease modifiers in FSHD1¹² or can cause FSHD2. Most patients with FSHD2 have at least one permissive 4qA allele with RU length between 11 and 20 (∼70%); in addition, the great majority of the remainder (>95%) have RU lengths between 21 and 30.¹³ Thus, variants in epigenetic repressors and cofactors such as SMCHD1 are not sufficient to cause FSHD2 on their own; they require the presence of at least one permissive 4qA haplotype on a moderately shortened D4Z4 repeat array to cause the disease.¹⁴

Recently, missense variants in SMCHD1 were found to also cause a very rare yet distinct disorder: congenital arhinia (absence of an external nose).^15,16 Arhinia is often accompanied by ophthalmic defects and central hypogonadism, a triad called Bosma arhinia microphthalmia syndrome (BAMS).¹⁷ Remarkably, at least 3 SMCHD1 variants have been found to cause FSHD2 in some families and arhinia in others.^15,18 No individuals or families with both arhinia and clinical FSHD2 have been reported.¹⁵ In one previously reported family (family T),¹⁵ an SMCHD1 variant was passed from a father with myopathy to a son with arhinia; however, further testing in the father showed normal methylation in chromosome 4q35, excluding the possibility of FSHD2 and suggesting a different etiology for his myopathy. Deep phenotyping studies of patients with FSHD2 have not uncovered any dysmorphic facial features, olfactory defects, or reproductive disorders suggestive of BAMS phenotypic spectrum.¹⁸ Deep neuromuscular phenotyping studies in individuals with arhinia have not been conducted.

We performed detailed clinical evaluation and muscle imaging in 11 individuals with either congenital arhinia or related craniofacial phenotypes (e.g., congenital nasal hypoplasia or anosmia). We also performed standard genetic testing for FSHD2 in all affected individuals and investigated the connection between FSHD clinical stigmata, genotype, and full-length DUX4 mRNA expression.

Methods

Participants and Clinical Research Studies

In our cross-sectional study, 13 individuals (11 with pathogenic SMCHD1 variants and 2 without SMCHD1 variants) aged 14–52 years underwent a detailed neuromuscular examination, a blood draw, a skin punch biopsy, muscle ultrasound, and muscle MRI studies between May 2017 and December 2018 at the NIH clinical center (n = 12) or offsite (n = 1). Basic clinical characteristics and SMCHD1 variants were previously reported in 11 of the patients with arhinia^19-21 and alternate patient IDs from prior studies are included in the current study. Two individuals underwent a muscle biopsy. The muscle samples were processed per standard procedures for histologic analysis and the remainder of the muscle was used to establish primary myoblast cultures. All phenotyping was performed by investigators who were blinded to the results of the genetic testing at the time of evaluation. Fibroblast cultures from an additional 6 patients with arhinia and SMCHD1 variants were studied in collaboration with their local physicians; they completed a detailed medical questionnaire, provided a blood sample (for DNA extraction), and underwent a skin punch biopsy. Primary myoblasts from 2 individuals with FSHD2 (made available as a gift from the University of Rochester) were analyzed.

Standard Protocol Approvals, Registrations, and Patient Consents

Written informed consent and age-appropriate assent was obtained from all participants. Ethical approval was obtained from the institutional review boards at the National Institute of Neurologic Disorders and Stroke/NIH (Protocol 12-N-0095) and from the National Institute of Child Health and Development/NIH (Protocol 2012-CH-0050).

Muscle MRI and Ultrasound

Skeletal muscle MRI and ultrasound was performed in all 13 individuals. Muscle MRI was performed using conventional T1-weighted spin echo and short tau inversion recovery (STIR) of the lower extremities on a 3.0T Verio (Siemens Healthineers) or 1.5T Aera (Siemens Healthineers). Noncontrast images were obtained from pelvis, thighs, and lower legs in the axial plain. Slices were 5–10 mm thick. The gap between slices were 8–10 mm thick. Muscle ultrasound was performed using an upgraded Acuson S2000 (Siemens Heatlhineers) with an 18-MHz linear probe. Muscle ultrasound echogenicity was assessed on the right or left side by sampling rectus femoris, vastus lateralis, tibialis anterior, gastrocnemius (medial and lateral heads), hamstrings, deltoid, biceps, and triceps muscles. Transverse images were acquired at approximately the mid belly of the muscles and after ensuring perpendicular imaging of the underlying bone to decrease the potential for oblique angles. The images were qualitatively graded based on a modified Heckmatt scale²² (grade 0 = normal, grade 1 = increase in muscle echogenicity while bone echo still distinct, grade 2 = marked increase in muscle echogenicity with obscuration of internal structure and reduced bone echo, 3 = severely increased muscle echogenicity with loss of bone echo) by 3 independent neuromuscular specialists trained in muscle ultrasound. Any discrepancies in ratings were resolved by discussion among the raters and the consensus ratings were used.

SMCHD1 Sequencing

Rare sequence variants in SMCHD1 were identified by either whole-exome sequencing or by targeted (Sanger) sequencing of exons 3–13 in a CLIA-certified laboratory (University of Iowa). Pathogenic variants in DNMT3B or LRIF1 were not identified.

DNA Methylation Analysis

The DNA methylation status of the D4Z4 region was assessed using digestion with the restriction enzymes EcoRI and BglII, followed by a second digestion with the methylation-sensitive enzyme FseI. Following linear gel electrophoresis, the digestion product was hybridized with the probe p13E-ll. The relative amounts of methylated and hypomethylated DNA were quantitated using Image Lab software (University of Iowa). In a subset of samples (n = 6), bisulfite sequencing (BSS) assay specific for the FSHD2-affected D4Z4 arrays on 4q35 and 10q26 was also performed as previously described.²³ Briefly, genomic DNA was purified from fibroblasts or myoblasts using the Wizard genomic DNA purification kit (Promega) and bisulfite converted using EpiTect Bisulfite kit (Qiagen) per the manufacturer's instructions. Bisulfite converted genomic DNA (100 ng) was amplified with primers BSS167F and BSS1036R, followed by the nested primers BSS475F and BSS1036R. The product was gel purified and cloned into the pGEMT-easy vector (Promega) for Sanger sequencing. Sequence data were analyzed using Web-based analysis software BISMA with the default parameters.²³

4q35 Haplotype Analysis and Array Size Determination

Standard genomic PCR was performed on nonconverted DNA to identify the 4qA, 4qA-L, and 4qB chromosome. Specific 4q and 10q haplotypes were identified and assigned by simple sequence length polymorphism analysis as previously described.^24,25

For 4q35 and 10q26 D4Z4 array size determination, peripheral blood leukocytes or cultured lymphoblast DNA were embedded in agarose plugs and digested with 3 different restriction enzymes (EcoRI, EcoRI/BlnI, and XapI). Restriction fragments were separated by pulsed-field gel electrophoresis and visualized by Southern blot with a p13E-11 probe. In some cases, the Southern blots were rehybridized with a D4Z4 probe.²⁶ Optical genome mapping was performed in one sample, as previously described.¹³ The following genetic and epigenetic criteria for FSHD2 were used: rare SMCHD1 variant, D4Z4 repeat unit length <21 repeat units in cis with a 4qA haplotype, and D4Z4 methylation <30%.

Studies in Patient Fibroblasts and Myoblasts

Dermal fibroblasts from 11 individuals with arhinia or related phenotypes (partially overlapping with the 13 individuals in the neuromuscular phenotyping studies) and 3 individuals with FSHD2 (1,090-1, 19FB042, 19FB047) were derived from skin punch biopsy samples as previously described.²⁷ Control fibroblasts (n = 3, GM07525, GM23962, and GM23967) were purchased from Coriell Institute for Medical Research. Briefly, primary fibroblasts were plated at a cell density of 1 × 105 cells per well in 6-well plates in Dulbecco’s Modified Eagle Medium (DMEM) medium containing 15% fetal bovine serum (FBS). The following day, 5-Aza-2′-deoxycytidine (ADC; Sigma-Aldrich A3656) was added into the media with a final concentration of 5 μM for 48 hours, followed by a combination treatment of 5 μM ADC plus 200 nM Trichostatin A (TSA; Sigma-Aldrich T1952) for another 24 hours.²⁸ Dimethyl sulfoxide was used as a vehicle for the untreated controls.

Primary myoblasts were established from muscle biopsy explants from 2 individuals with arhinia as previously described.²⁹ In brief, a small piece of muscle (1–3 mm) was placed in a small amount of DMEM supplemented with penicillin/streptomycin/amphotericin B (Thermo Fisher; 15290026) and cut into small fragments, with each piece transferred to a flask and allowed to attach. The muscle cells were allowed to grow over 1–2 weeks in proliferating medium (DMEM, 20% FBS, insulin 10 µg/mL, l-glutamine, human basic fibroblast growth factor 25 ng/mL, and epidermal growth factor 10 ng/mL).

Primary myoblasts were plated at a density of 3 × 105 per well in 6-well plates and grown in the 0.02% collagen-coated plates with myocyte growth media (Ham F-10 medium supplemented with 20% FBS, 0.5% chick embryo extract, 1% antibiotics and antimycotics, and 1.2 mM CaCl₂). The next day, the media were replaced with differentiation medium (DMEM medium containing 2% horse serum) and cultured for 4 days. The media was then replaced back to myocyte growth media. The cells were harvested at day 3 and day 6 postdifferentiation.

Gene Expression Analysis by RT-qPCR

Full-length DUX4 mRNA was measured by nested reverse transcription quantitative real-time PCR (RT-qPCR) as previously described.³⁰ Briefly, total RNA was extracted from the culture cells homogenized by TRIzol (Thermo Fisher) and purified using RNeasy Mini kit (Qiagen) per manufacturer instructions. First strand cDNA synthesis was performed with RNA (2 μg), oligo dT primer, and SuperScript III Reverse Transcriptase (Invitrogen), then 200 ng cDNA was used for quantitative PCR (qPCR) analysis of DUX4-fl, and the relative expression was normalized to levels of 18S rRNA. The data represent the mean ± SEM of 3 individual experiments using different plating of cell lines, each assayed by 3 qPCR replicates per experiment.

Statistics

The RT-qPCR relative expression data were analyzed by 1-way analysis of variance in conjunction with Dunnett multiple comparisons test using GraphPad Prism 7.0. Statistical comparisons in the human fibroblasts experiment were made between relative expression of the genes in drug-treated vs vehicle-treated cells. Statistical comparisons in the primary myoblast cultures were made across the 3 time points in each cell line. Results with p values < 0.05 were considered statistically significant.

Data Availability

De-identified data not published within this article will be made available upon reasonable request from any qualified investigator.

Results

Eleven individuals (7 female, 4 male; age 18–52 years) with congenital arhinia (n = 9) or nasal hypoplasia (n = 2) underwent a detailed neuromuscular evaluation. Ocular abnormalities included microphthalmia/anophthalmia (n = 5) and achromatopsia with abnormal optic discs (n = 1). All individuals had complete anosmia or impaired olfaction and the majority (n = 10) had hypogonadotropic hypogonadism (testosterone <50 ng/dL, estradiol <10 pg/mL) and had been off hormone replacement for 1 month before clinical evaluation. Additional clinical details of systemic and endocrine findings in the affected individuals in this cohort were reported previously.²¹

Genetic Testing and FSHD2 Risk Assessment

All individuals underwent genetic testing to identify SMCHD1 variants and assess the methylation of the D4Z4 locus as well as 4q35 haplotyping studies to determine their genetic risk for developing FSHD (Table 1). Eleven individuals had rare (minor allele frequency <0.1%) heterozygous missense variants in SMCHD1, all classified as pathogenic and associated with the arhinia phenotype in the ClinVar database. Three individuals met the formal genetic criteria for FSHD2, defined as a pathogenic variant in SMCHD1, D4Z4 hypomethylation (<30%), a low normal 4q35 D4Z4 repeat length (11–20 RUs), and a permissive 4q35 haplotype (4A159, 4A161, 4A161-L, or 4A168).^9,31-33 All 3 had at least 1 permissive 4qA haplotype with a D4Z4 length of <21 RU (Table 1).

Table 1.

Genetic and Epigenetic Studies of Patients With Midline Craniofacial Abnormalities

Neuromuscular Signs and Symptoms

Only one individual, P1/A1, who met full FSHD2 genetic criteria, reported neuromuscular symptoms, prominently manifesting in her 40s as chronic fatigue, muscle pain, and exertional muscle weakness. The other 2 individuals who met formal FSHD2 criteria (P5/AM2 and P7/I1) remain clinically asymptomatic. No individual had a history of gross or fine motor delays, but some reported difficulty in keeping up with peers in physical activities, including a history of slow running speed since childhood. These mild limitations were nonprogressive and remained subclinical.

Detailed neuromuscular evaluation identified normal strength in 5/11 while the remaining 6 individuals had mild, proximal-predominant weakness, primarily affecting the neck flexors, arm abductors, hip flexors, and knee flexors (Medical Research Grade 4/5 or higher), in a pattern distinct from FSHD (Table 2). Assessment of facial muscles in this cohort was limited due to craniofacial abnormalities and multiple reconstructive surgeries in nearly all affected individuals. None of the individuals had typical findings of FSHD³⁴ such as scapular winging, biceps weakness, lower abdominal weakness (Beevor sign), or ankle dorsiflexor weakness. The 3 individuals who met FSHD2 genetic and epigenetic criteria did not differ from the others in the severity or distribution of their weakness; all demonstrated mild, symmetric weakness involving proximal and truncal muscles. Serum creatine kinase levels were within the normal range when tested (Table 2). Nerve conduction studies and EMG were not performed. For comparison, 2 individuals (P12 and P13) without arhinia or SMCHD1 variants but with related craniofacial phenotypes (Binder type nasomaxillary dysplasia [n = 1] and bifid nose [n = 1]) were similarly evaluated and both had normal strength (Tables 1 and 2).

Table 2.

Neuromuscular Phenotype of Individuals With Midline Craniofacial Abnormalities

Muscle Ultrasound and MRI

Muscle ultrasound of affected muscles generally showed a pattern of mildly to moderately increased echogenicity with a granular appearance (Figure 1, A and B). Most affected muscles showed very mild increases in echo intensity (grade 1 or 2 on the modified Heckmatt scale).²² However, deltoid, hamstrings, and medial gastrocnemius muscles were on average more severely affected (higher echogenicity) than the other muscles. The increased echogenicity was diffuse and seldom severe enough to completely obscure the bone echo. There was no discernible difference in pattern and severity of muscle echogenicity in patients with arhinia based on genetic risk for FSHD2. Fasciculations or a streak-like pattern of echogenicity, typically seen in neurogenic disease, were not observed.

Figure 1. Muscle Ultrasound in Individuals With Arhinia or Related Phenotypes.

(A) Muscle ultrasound echogenicity was assessed and graded based on the modified Heckmatt scale by 3 independent clinicians. The consensus ratings are depicted in a heatmap format. Individuals who are at risk for facioscapulohumeral muscular dystrophy type 2 (FSHD2) based on genetic and epigenetic criteria are highlighted in bold. (B) Ultrasound images of rectus femoris, hamstring, and medial gastrocnemius muscles of 4 patients are shown as representative examples. The increased echogenicity, if present, was mild and rarely severe enough to obscure the bone echo. We failed to detect a marked difference in pattern and distribution of increased echogenicity between individuals at risk for FSHD2 and those without genetic and epigenetic risk for FSHD2.

On lower extremity muscle MRI, T1 axial images were either normal or demonstrated only minimal signs of muscle fatty replacement (Figure 2A). The only notable exception was selective and severe involvement of the medial gastrocnemius muscle in P5/AM2, the oldest individual in our cohort who met the FSHD2 genetic and epigenetic criteria. STIR signal hyperintensity is thought to highlight areas of muscle disease activity prior to tissue remodeling and appearance of T1 signal hyperintensity in FSHD.³⁵ However, STIR imaging of the muscles in our cohort failed to identify areas of increased signal regardless of genetic risk for FSHD2 (Figure 2B).

Figure 2. Muscle MRI in Individuals With Arhinia or Related Phenotypes.

(A) T1-weighted muscle MRI of 4 patients in the cohort do not show a consistent pattern of muscle involvement. The only abnormality is limited to T1 hyperintensity in the medial gastrocnemius muscle of one individual (P5/AM2), who is genetically at risk for facioscapulohumeral muscular dystrophy type 2 (FSHD2). (B) Short tau inversion recovery (STIR) imaging of the muscle did not identify areas of increased signal regardless of genetic risk for FSHD2. T1-weighted images from the same cut are provided for comparison.

Muscle Pathology

Two individuals with arhinia underwent muscle biopsies: P1/A1, who is at risk for FSHD2 based on genetic testing, and P4/AM1, who is not at risk for FSHD2. Histologic analysis identified mild, nonspecific myopathic changes in both muscle biopsies. AM1's muscle biopsy also showed slightly increased prevalence of internal nuclei (Figure 3A). Longitudinal sections confirmed the presence of central nuclei in chains (Figure 3B), a finding commonly seen in regenerating muscle fibers. P1/A1's muscle biopsy showed variability in myofiber size (Figure 3C) with presence of both hypertrophic and atrophic myofibers without internal nuclei. There were no lobulated fibers, dystrophic changes, or inflammatory infiltrates, which have previously been reported in some FSHD muscle biopsies. Ultrastructural studies with electron microscopy were unremarkable and showed normal sarcomeric structure and mitochondria (Figure 3D).

Figure 3. Muscle Histology in Individuals With Congenital Arhinia and SMCHD1 Pathogenic Variants.

(A) Hematoxylin & eosin stain of P4/AM1 muscle biopsy of vastus lateralis shows normal myofiber size. Internal nuclei are only slightly increased in frequency (∼4%) (arrows). (B) Formalin-fixed, paraffin-embedded longitudinal sections reveal centrally placed nuclei in chains (arrow) (P4/AM1). (C) Muscle biopsy in P1/A1 shows variability in myofiber size with presence of both hypertrophic (asterisk) and atrophic (arrow) myofibers. (D) Electron micrographs demonstrate normal sarcomeric ultrastructure. The few mitochondria visible in between myofibrils appear morphologically normal (P1/A1). Scale bars: (A–C) 100 µm; (D) 3 µm.

Polyadenylated, Full-Length DUX4 mRNA Expression in Patient-Derived Cells

Aberrant expression of the DUX4 retrogene in mature skeletal muscle is the molecular hallmark of FSHD,³⁶ but DUX4 transcript and protein levels in muscle tissue or patient-derived cells are very low and technically challenging to detect.^2,37,38 We recently developed a cell-based assay that uses epigenetic drugs ADC and TSA to assess the inducibility of DUX4 expression in dermal fibroblasts in its native epigenetic context and showed that in contrast to unaffected control cells, FSHD-derived cells express high levels of DUX4 transcript after epigenetic drug induction.²⁸ Thus, this epigenetic DUX4 activation assay can be used as a research tool to assess whether genetic and epigenetic conditions required to express DUX4-fl mRNA exist in a specific somatically derived patient cell line, regardless of the underlying genomic mechanisms of DUX4 de-repression. We used this assay and compared DUX4 induction in fibroblasts from 11 individuals with arhinia with that of 3 patients with FSHD2 (1,090-1, 19FB042, 19FB047) and 3 healthy controls (GM07525, GM23962, and GM23967) (Figure 4A and eTable 1, links.lww.com/WNL/B800).

Figure 4. DUX4 Expression in Drug-Treated Primary Human Dermal Fibroblasts (HDF).

(A–C) HDF were treated with 5-aza-2ʹdeoxycytidine (ADC) for 72 hours and co-treated with Trichostatin A for the past 24 hours. The expression levels of (A) full-length DUX4 (DUX4-fl) mRNA, (B) human ankyrin repeat domain 1 (hANKRD1), and (C) human myogenic factor 5 (hMYF5) were assayed in drug-treated cells (+) and vehicle-treated (dimethyl sulfoxide) cells (−) by reverse transcription quantitative real-time PCR. The data represent the mean with SEM of 3 individual cell culture experiments each assayed by 3 quantitative PCR replicates per experiment. Statistical comparisons were between drug and vehicle treatment condition for each cell line using 1-way analysis of variance (*p < 0.05). FSHD2 = facioscapulohumeral muscular dystrophy type 2.

ADC and TSA treatment induced robust DUX4-fl mRNA expression in fibroblasts derived from the patients with FSHD2. DUX4-fl mRNA induction was also observed in fibroblasts treated with epigenetic drugs in the 3 patients with arhinia with the FSHD2 genetic and epigenetic profiles (patients P1/A1, P7/I1, and P5/AM2) (Figure 4A). Although the DUX4 expression level for each of the patients with arhinia was substantially lower than in 2 of the symptomatic patients with FSHD2 (1,090-1 and 19FB042), it trended higher when compared to healthy controls and patients with arhinia with a nonpermissive 4qA haplotype. Successful epigenetic gene activation by drug treatment in all samples was confirmed by upregulation of human ankyrin repeat domain 1 (hANKRD1)³⁹ (Figure 4B) and the epigenetically regulated early myogenesis marker human myogenic factor 5 (hMYF5); these genes were not upregulated in vehicle-treated controls (Figure 4C).

We also examined DUX4-fl expression in differentiated myotubes from 2 patients with arhinia who had undergone a muscle biopsy (Figure 5). DUX4 expression was induced in the myotubes differentiated for 3 days or 6 days in the patient with arhinia with an FSHD-permissive haplotype (patient P1/A1) but not in the patient with arhinia without genetic risk for FSHD2 (patient P4/AM1). The overall DUX4 transcript levels from the patient with arhinia were lower when compared to the levels in differentiated myoblasts from patients with FSHD2 and the results recapitulated the data from the drug treatment in fibroblasts. Overall, these data suggest that patients with arhinia meeting the genetic and epigenetic criteria for FSHD2 can in fact express DUX4 in skeletal muscle cells.

Figure 5. DUX4 Expression in Differentiated Myotubes.

(A, B) Primary myoblasts derived from facioscapulohumeral muscular dystrophy type 2 (FSHD2) (17MB020, 18MB051) or arhinia (A1, AM1) were differentiated into myotubes and harvested on day 3 or day 6 postdifferentiation. Reverse transcription quantitative real-time PCR analysis of the expression of DUX4-fl (A) and hMYH1 mRNA (B) in undifferentiated myoblasts (UD) and differentiated myocytes on day 3 (D3) and day 6 (D6) postdifferentiation. The data represent the mean with SEM of 3 individual cell culture experiments each assayed by 3 quantitative PCR replicates per experiment. Statistical comparisons were across the time points for each cell line using 1-way analysis of variance in conjunction with Dunnett multiple comparisons. ∗p < 0.05 compared to each undifferentiated control.

Discussion

Despite its highly recognizable clinical phenotype, FSHD remains a genetically complex disorder. Recent advances in molecular genetics have disentangled some of the complex pathophysiologic mechanisms of FSHD and have put forth a unifying molecular mechanism for the disease: aberrant expression of the DUX4 retrogene in mature skeletal muscle.³⁶ The underlying cause of this aberrant expression can vary from D4Z4 repeat contraction in FSHD1 to secondary changes in the chromatin structure and DNA hypomethylation due to variants in epigenetic regulatory genes such as SMCHD1 in FSHD2. Regardless of the underlying mechanism, postnatal expression of DUX4 protein in skeletal muscle appears to be necessary and sufficient to cause the disease with its characteristic phenotypic features.³⁶

SMCHD1 variants can also cause arhinia, an allelic disorder with a seemingly nonoverlapping phenotype with FSHD2. This phenotypic pleiotropy could simply arise from the absence of concurrent genetic and epigenetic changes in chromosome 4q35 that are necessary for DUX4 transcription. However, when we clinically evaluated 11 individuals with SMCHD1-related arhinia or related phenotypes for FSHD genetic and epigenetic criteria, we were surprised to find 3 individuals (P1/A1, P5/AM2, and P7/I1) who by the strictest criteria were at genetic risk for FSHD2, yet did not manifest clinicopathologic features of FSHD2.

Considering the variability of molecular mechanisms of FSHD, the molecular diagnosis of FSHD remains an onerous task. Our epigenetic drug assay measuring the inducibility of DUX4 expression in fibroblasts, similar to our previous study in myoblasts,²⁸ provides a sensitive strategy to delineate DUX4 de-repression and expression in the specific epigenetic context of the patient. To further assess the risk for FSHD2 in patients with arhinia using this sensitive tool, we used this novel drug treatment assay. Illustrating the high sensitivity of this assay, fibroblasts from patients with arhinia with hypomethylated D4Z4 repeats and a permissive 4A161 haplotype (P1/A1, P5/AM2, and P7/I1) all expressed higher DUX4-fl upon induction compared to fibroblasts of patients with arhinia without a permissive haplotype and healthy controls (Figure 4 and eTable 1, links.lww.com/WNL/B800). DUX4-fl levels in drug-treated arhinia fibroblasts were on average lower than the levels in symptomatic FSHD2; however, it remains unclear whether DUX4 expression levels in this cell system can predict the onset or severity of FSHD.

Myoblasts from patient P1/A1, who met the FSHD2 genetic criteria, expressed DUX4 during myogenic differentiation, although the DUX4-fl levels were much lower than the levels in clinically manifesting FSHD2 (Figure 5). On the other hand, there was no DUX4 expression in the myocytes from patient with arhinia P4/AM1, who is not at genetic risk for FSHD2. Whereas expression of DUX4 in induced pluripotent stem cells derived from patients with arhinia has previously been reported,⁴⁰ inducibility of DUX4 in somatic cells derived from patients with arhinia has not. These findings suggest that in addition to methylation status, 4qA haplotypes, and repeat unit size, other genetic or epigenetic factors can modify DUX4 expression and influence FSHD phenotypic penetrance. Although our study does not identify these factors, our findings are compatible with the observations of incomplete penetrance in FSHD pedigrees, which includes asymptomatic carriers of pathogenic variants.⁴¹ In addition, the distinct dichotomy between FSHD2 and arhinia phenotypes further suggests that the genetic or epigenetic factors leading to arhinia may be protective against FSHD and vice versa and appear to operate as a switch, resulting in one phenotype and not the other. Transcription factors such as DUX4 commonly act as such molecular master switches and can direct different gene expression signatures based on their expression levels and the timing and duration of expression.⁴² Recent studies suggest that the dynamics of DUX4 expression may influence its cytotoxicity: constitutive expression is reported to be toxic while pulses of expression do not affect cell survival.⁴³ Thus we speculate that the factors that lead to the distinctly different phenotypes of FSHD and arhinia may play a role in determining the level, timing, and duration of DUX4 expression during development or determine its tissue specificity.

FSHD2 and arhinia do not seem to be governed by identical epigenetic changes at the 4qA locus. Hypomethylation of the 4qA locus in patients with arhinia in our cohort is comparable to SMCHD1-related FSHD2⁴⁴ (Table 1). In contrast, whereas distribution of RUs in clinically affected patients with FSHD2 is usually biased toward moderately shorter lengths (∼70% with 11–20 RUs, ∼30% with >20 RUs),¹³ 4qA allele RU lengths in our patients with arhinia with hypomethylation and at least one permissive 4qA allele (n = 7) are more evenly distributed (3/7; ∼42% with 11–20 RUs, 4/7; ∼58% with RU >20) (Table 1). Thus, unlike SMCHD1-related FSHD2, the SMCHD1-related arhinia phenotype is not enriched in individuals with low/normal RUs, suggesting differences in molecular mechanisms underlying the 2 diseases.

An alternative explanation for the lack of an FSHD2 phenotype in individuals with arhinia may be their relatively young age. Although the age at onset of symptoms in FSHD can vary widely, the majority of patients with FSHD develop symptoms in young adulthood.³⁴ Thus, it is possible that P7/I1, who is 23 years of age, may develop signs and symptoms of FSHD2 with age; however, P1/A1 and P5/AM2 are in their 40s and 50s, respectively, and did not manifest typical FSHD symptoms. Although our clinical evaluation of the facial muscles was limited due to multiple reconstructive surgeries and the underlying craniofacial abnormalities, we also failed to detect clinically pertinent skeletal muscle weakness outside the craniofacies in a pattern consistent with FSHD, which almost universally also involves other muscle groups: the scapular fixators, biceps, lower abdominal muscles, or anterior lower leg muscles. Instead of an FSHD-like distribution of muscle weakness, we found mild and subclinical changes in muscle strength and normal muscle imaging by MRI, including normal STIR sequences. Muscle imaging by ultrasound, which has a high sensitivity but low specificity for detecting mild changes in muscle based on increased echogenicity, identified a generalized but mild pattern of change in muscle echointensity (Figure 1) in most of the individuals with arhinia, irrespective of their genetic risk for developing FSHD2. These nonspecific findings may be secondary to concurrent endocrinopathies of this cohort (i.e., hypogonadism) and lower levels of activity, which can also secondarily result in nonspecific myopathic changes and mild proximal weakness, especially in those with severe visual impairment. Thus, the neuromuscular-related findings in individuals with arhinia likely do not represent a primary myopathic disorder; however, due to the small number of individuals in this study, which stems from the rarity of the condition, we cannot confidently exclude the presence of a mild, static, early-onset myopathy or the possibility that these individuals may develop clinical weakness with advanced age.

A limitation of our observational study is the small number of individuals available for neuromuscular examination and muscle imaging. This is primarily due to the extremely small number of individuals with arhinia (fewer than 100 cases reported in the literature in the past century). Cell studies were limited by availability of muscle biopsy and primary myoblast cultures in only a few patients in this cohort. Another limitation is lack of electrodiagnostic data (nerve conduction studies and EMG), which could have potentially improved the sensitivity of the neuromuscular examination and uncovered subclinical changes in muscle and nerve function. Despite these limitations, our study raises the intriguing hypothesis that the genetic or epigenetic factors leading to arhinia may be protective against FSHD and at the very least mitigate the severity of the disease or delay its onset.

We provide detailed neuromuscular phenotyping data and FSHD2 risk assessment in individuals with congenital nasal anomalies with pathogenic SMCHD1 variants. In this cohort, we identified 3 individuals at risk for developing FSHD2 based on the most stringent genetic and epigenetic criteria. These individuals had higher levels of DUX4-fl mRNA in patient-derived somatic cells compared to the healthy subjects but lack clinical signs and symptoms of FSHD2. The underlying molecular cause for this seeming protection is unclear. Given the complex effects of SMCHD1 protein on long range chromatin interactions and accessibility of chromatin-modifying enzymes to their target sequences,⁴⁵ additional studies that focus on the 3D chromatin structure in FSHD2 and arhinia may provide an opportunity to identify novel FSHD disease modifiers that influence DUX4 expression in skeletal muscle.

Glossary

ADC

5-Aza-2′-deoxycytidine

BAMS

Bosma arhinia microphthalmia syndrome

BSS

bisulfite sequencing

DMEM

Dulbecco’s Modified Eagle Medium

FBS

fetal bovine serum

FSHD

facioscapulohumeral muscular dystrophy

qPCR

quantitative PCR

RT-qPCR

reverse transcription quantitative real-time PCR

RU

repeat unit

STIR

short tau inversion recovery

TSA

Trichostatin A

Appendix. Authors

Study Funding

This work was supported in part by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences (1ZIAES103327-05; N.D.S.) and intramural funds of the National Institute of Neurologic Disorders and Stroke (1ZIANS003129; C.G.B.). N.D.S. is also supported as a Lasker Clinical Research Scholar (1SI2ES025429-01). P.L.J. is funded by NIH R01AR062587 and is supported by the Mick Hitchcock, PhD endowed chair for medical biochemistry at the University of Nevada, Reno. A.M.K. is funded by the German Research Foundation (DFG; SFB1315, FOR3004) and the Charité (3R). S.A.M. is funded by P50 NS053672 to the Iowa Wellstone Muscular Dystrophy Specialized Research Center.

Disclosure

The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.