Al‐Gazali Skeletal Dysplasia Constitutes the Lethal End of ADAMTSL2‐Related Disorders

Lethal short‐limb skeletal dysplasia Al‐Gazali type (OMIM %601356), also called dysplastic cortical hyperostosis, Al‐Gazali type, is an ultra‐rare disorder previously reported in only three unrelated individuals. The genetic etiology for Al‐Gazali skeletal dysplasia has up until now been unknown. Through international collaborative efforts involving seven clinical centers worldwide, a cohort of nine patients with clinical and radiographic features consistent with short‐limb skeletal dysplasia Al‐Gazali type was collected. The affected individuals presented with moderate intrauterine growth restriction, relative macrocephaly, hypertrichosis, large anterior fontanelle, short neck, short and stiff limbs with small hands and feet, severe brachydactyly, and generalized bone sclerosis with mild platyspondyly. Biallelic disease‐causing variants in ADAMTSL2 were detected using massively parallel sequencing (MPS) and Sanger sequencing techniques. Six individuals were compound heterozygous and one individual was homozygous for pathogenic variants in ADAMTSL2. In one of the families, pathogenic variants were detected in parental samples only. Overall, this study sheds light on the genetic cause of Al‐Gazali skeletal dysplasia and identifies it as a semi‐lethal part of the spectrum of ADAMTSL2‐related disorders. Furthermore, we highlight the importance of meticulous analysis of the pseudogene region of ADAMTSL2 where disease‐causing variants might be located. © 2023 The Authors. Journal of Bone and Mineral Research published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research (ASBMR).


Introduction
S hort-limb skeletal dysplasia Al-Gazali is a rare neonatal lethal skeletal dysplasia, which has previously only been described in three unrelated individuals. (1,2) The main skeletal features are severe growth retardation, short limbs, generally sclerotic bones, broad and underdeveloped ribs and clavicles, short long bones, mild platyspondyly, and severe brachydactyly. Craniofacial morphology includes relatively large head with a wide anterior fontanelle, flat facial profile, low-set ears, short nose, and thin lips. (1,2) Other characteristics are hypertrichosis, mildly hypoplastic thorax, and joint stiffness. Radiological characteristics include increased bone density, short tubular bones, wide diaphyses, and round metaphyses. In the previously reported individuals, no genetic etiology was identified. Phenotypically similar but milder skeletal dysplasia caused by biallelic variants in ADAMTSL2 is geleophysic dysplasia type 1 (GPHYSD1, OMIM #231050). Twenty-nine pathogenic variants in the ADAMTSL2 gene have been reported in 43 individuals with GPHYSD1. (3,4) GPHYSD1 is a rare autosomal recessive disorder generally characterized by severe short stature, small hands and feet, restricted joint movement, joint contractures, distinctive facial features, tip-toe walking, and thickened skin. Skeletal features include broad proximal phalanges, cone-shaped epiphyses, delayed bone age, osteopenia, short tubular bones, and ovoid vertebral bodies. (5,6) In addition, recurrent respiratory infections, tracheal stenosis, and progressive cardiac disease are common and often lead to an early death. (5,7) In the current nosology and classification of genetic skeletal disorders, GPHYSD1 belongs to the group of acromelic dysplasias. (8) Acromelic dysplasias collectively form a group of rare connective tissue disorders that share distinct musculoskeletal features, such as short stature, brachydactyly, joint stiffness, pseudomuscular build, and thick, tight skin. (3) Most of the genes related to acromelic dysplasias encode secreted extracellular matrix (ECM) proteins that are important for microfibrillar network assembly and regulation of TGFβ and BMP signaling pathways. (3) Human ADAMTSL2 (NM_014694.4, ENST00000393060.1, hg19) is composed of 19 exons, of which the first is non-coding, and exons 10-19 share sequence homology with a pseudogene region. Low expression of ADAMTSL2 in commonly clinically available tissues, such as dermal fibroblasts and blood, has been reported in adults (GTEx, https://gtexportal.org/). Expression of ADAMTSL2 was also shown to be undetectable in cultured human fetal cardiac fibroblasts. (9) Through international collaboration, a cohort of nine patients with a clinical diagnosis of lethal skeletal dysplasia Al-Gazali type and disease-causing variants in the ADAMTSL2 gene was collected. We present a summary of clinical, radiological, and genetic data and show that in the less severe cases, survival into early childhood is possible. However, the condition is associated with high perinatal mortality most often caused by respiratory and cardiac failure.

Patients and clinical records
Clinicians with patients with clinical diagnosis of Al-Gazali skeletal dysplasia were invited to recruit the families to this study. All referring physicians are members of the International Skeletal Dysplasia Network that facilitated this collaboration. Nine affected individuals from seven non-consanguineous and one consanguineous family were included in this study. Clinical findings of patient 1 from a Swedish family and patient 2 from a Japanese family were described previously. (2) In family 3, two fetuses were clinically diagnosed with Al-Gazali skeletal dysplasia, but DNA sample was available only from one fetus. Clinical data were collected from patient records by their referring physicians. Genomic DNA was extracted from blood, frozen fetal tissues, or primary cells as summarized in Supplemental Table S1. DNA from the first patient identified with Al-Gazali skeletal dysplasia was not available (personal communication with Prof L Al-Gazali).

Genetic analysis
Different sequencing approaches were performed for each patient in this study as shown in Supplemental Table S1. Initially, testing by exome sequencing was completed on DNA samples from family trios 1, 2, and 3, which led to identification of only two heterozygous a disintegrin and metalloproteinase with thrombospondin motifs-like 2 (ADAMTSL2) variants in two of the families. DNA samples from patient 1 and parental samples from family 2 were reanalyzed by whole genome sequencing on Illumina (San Diego, CA, USA) HiSeq X and NovaSeq 600 with median coverage of 30Â at Science for Life Laboratory (SciLifeLab, Stockholm, Sweden). The amount of DNA from patient 2 was not sufficient for further genome sequencing analysis. DNA samples from family 4 were analyzed by whole  Table 1 are annotated in black and white. (A) Overall appearance: short limbs, short neck, narrow thorax, protuberant abdomen, small hands and feet, and talipes equinovarus (mild in patient 2). (B) Typical facial features in patients 1, 5, and 6: round facies, hypertrichosis, periorbital puffiness, and anteverted nares. (C) Typical head morphology in patients 1, 2, and 5: brachycephaly, short neck, flat face, micrognathia, and small low-set ears with thick helix. (D) Typical hand morphology in patients 1, 5, and 6: trident appearance of the hand and severe brachydactyly. (E) Close view of lower extremities in patients 1, 5, and 6: equinovarus deformity and short toes. Clinical photos of patients 1 and 2 in A and C are reproduced with permission from AJMG part A.    genome sequencing on Illumina HiSeq X (SciLifeLab). Bioinformatic analysis and variant calling has been performed as described previously. (10) Variant annotation involves Combined Annotation Dependent Depletion (CADD) score, (11) VEP annotations, (12) and population allele frequencies. (13) Sanger sequencing of the entire ADAMTSL2 gene was performed on DNA samples from patient 3 and parental samples from family 7 (patient sample was not available for analysis). Illumina DNA PCR-Free Library Prep was used for resequencing the maternal sample from family 7 (Illumina). DNA sample from patient 5 was analyzed by targeted sequencing analysis using TruSight One sequencing panel, including 4813 genes associated with a known clinical phenotype (Illumina). DNA samples from families 6 and 8 were analyzed by whole exome sequencing (WES).

Splice variant analysis
Blood from the father of patient 8 was used to analyze the splice variant. RNA was extracted using PAXgene Blood RNA kit (PreAnalytiX, Hombrechtikon, Switzerland

Western blot
Western blot analysis was performed according to standard procedures. Briefly, protein from HDFn of patient 1 and controls was extracted using RIPA lysis buffer (Thermo Fisher Scientific) supplemented with Halt protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific). Protein concentration was measured using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). An amount of 20 μg protein was used for Western blot. Protein lysates were run on 4% to 15% gradient Mini-PROTEAN TGX precast gels (Bio-Rad) and transferred onto PVDF membrane. Membrane was blocked in 5% BSA for 1 hour at room temperature (RT) after primary antibody incubation at 4 C overnight. Proteins were visualized using Clarity Western ECL substrate (Bio-Rad), and membrane was developed using LI-COR (Lincoln, NE, USA) C Di-Git blot scanner. Quantification was done using ImageJ.

Immunohistochemistry
Cells were cultured for 14 days in 24-well black plates with flat coverslip bottom (ibidi). Fixation was done for 10 minutes at 37 C using Image-iT Fixative Solution (Thermo Fisher Scientific).

Statistical analysis
Statistical analysis was performed using GraphPad Prism 9 software (San Diego, CA, USA). Statistical significance was determined using Student's t test (p < 0.05). Data were presented as boxplots showing median, interquartile range, and minimum and maximum values.

Clinical findings
The main clinical findings in Al-Gazali skeletal dysplasia patients included moderate intrauterine growth restriction (IUGR) detected in the second trimester, narrow thorax, short limbs and club feet, and dysmorphic features including flat facial profile, hypertelorism, depressed nasal bridge, protruding eyes, hypertrichosis, relative macrocephaly, low-set ears with abnormally thick helix, brachycephaly, stiff joints, and thick and tight skin (Fig. 1). Five of nine patients presented with internal abnormalities, including stenosis of the aorta and pulmonary artery, lung hypoplasia, right atrial, and bilateral ventricular dilatation ( Table 1).
The main skeletal findings in our patients were generalized bone sclerosis, large anterior fontanelle, hypoplastic thorax with broad ribs, mild platyspondyly, absent or deficient ossification of pubic bones, short tubular bones, and severe brachydactyly (Fig. 2, Table 2).
Only two of the patients (5 and 8) have survived for more than 1 year. At the age of 12 months, patient 8 had significant hyperopia, possibly mild hearing loss, sleep apnea, narrow airway, pulmonary stenosis, and generalized stiffness. Radiological findings of patient 8 included severe brachydactyly, mild undermodeling of long bones of the arm (not shown), hypoplasia of ischial and pubic bones, and mild broadening of the ribs, but these features were milder compared with patient 5 (Fig. 2B, C). Patient 5 exhibited all radiological hallmarks of Al-Gazali skeletal dysplasia as well as unique fibular dysplasia and was not available for further follow-up of the clinical symptoms after the age of 14 months. Clinical and radiographic summaries for all patients are provided in Tables 1 and 2.

Genetic and molecular findings
The location and type of variants discovered in our patient cohort are shown in the schematic ADAMTSL2 protein structure together with the previously reported variants associated with GPHYSD1 (Fig. 3A). All patients tested, except the fetus from family 3, were compound heterozygous for variants in ADAMTSL2 (NM_014694.4, hg19) ( Table 3). The patient sample in family 7 was not available for genetic investigation; however, analysis of parental samples identified a missense variant, c.1998C>G, p.(Cys666Trp), in the paternal sample and c.939+5G>A variant in the maternal sample. The maternal variant was predicted to affect splicing according to Alamut Visual Plus (v1.2, SOPHiA Genetics, St. Sulpice, Switzerland). Maternal RNA was not, however, available for further analysis. Two missense variants detected in our patient cohort, p. (Arg113His) in patient 3 and p.(Arg645His) in patient 8, were previously reported in patients with GPHYSD1. (5,16,17) One of the variants identified in patient 8 (c.-151+2T>C) was paternally inherited and predicted to affect splicing (Alamut Visual Plus, v1.2, SOPHiA Genetics). Paternal blood sample from patient 8 was available for RNA analysis. It was hypothesized that the nucleotide change in the canonical splice site leads to no expression of the affected allele. Evaluation of the paternal WES data identified four heterozygous single nucleotide variants (SNVs) in ADAMTSL2, which were targeted on cDNA level to confirm their hemizygous state. Figure 3B shows Sanger sequencing validation of one of the heterozygous variants (chr9:136421044 C > T, c.1641C>T, p. (His547=)), which appears as hemizygous on cDNA level and supports the pathogenicity of the c.-151+2T>C alteration.

ADAMTSL2 expression analysis by ddPCR
ADAMTSL2 has a very low expression in cultured dermal fibroblasts (median TPM = 0.0512, Fig. 4A) according to public databases (GTex, https://gtexportal.org/) and is poorly detectable using conventional qPCR methods. Thus, ddPCR was used to estimate the expression levels in primary dermal fibroblasts from patient 1, who was compound heterozygous for p. (Trp53Arg) and p.(Cys713Arg), and control samples. Gene expression was measured at two different time points to evaluate whether ADAMTSL2 expression was dependent on different ECM conditions. RNA was collected at 95% confluency (3-4 days, immature ECM) and after 7 days in culture (mature ECM), as shown in Fig. 4B. Gene expression of ADAMTSL2 at immature ECM conditions showed a minor increase in the patient 1 sample compared with controls but showed no statistical significance (Fig. 4C). At the mature ECM conditions, the patient 1 sample had comparable ADAMTSL2 expression to the control samples (Fig. 4D). Overall, similar ADAMTSL2 expression levels were observed in patient and control samples regardless of the ECM conditions.

Analysis of pSMAD2 signaling and FBN1 microfibrillar network
Previous studies in primary cells of GPHYSD1 patients (5) and various mouse models suggest that the pathogenic mechanism of GPHYSD1 is an overactivation of TGFβ signaling pathway and/or affected microfibrillar network. In cultured dermal fibroblasts from patient 1 with Al-Gazali skeletal dysplasia, no differences in pSMAD2 expression were observed compared with controls (Fig. 5A). In addition, the microfibrillar network organization was explored by imaging FBN1 microfibrils ( Fig. 5B and Supplemental  Fig. S1). FBN1 had comparable signal intensity between patient 1 and controls. However, distribution pattern of FBN1 microfibrils in patient 1 showed more stretched and denser FBN1 microfibrils compared with controls. These differences were more evident in the lower-magnification images (Supplemental Fig. S1).

Discussion
In this study, we summarize clinical and genetic findings of nine individuals with Al-Gazali skeletal dysplasia, two of whom have previously been described by Grigelioniene and colleagues. (2) We report for the first time that biallelic variants in the ADAMTSL2 gene cause skeletal dysplasia Al-Gazali type.
ADAMTSL2 is a disease-causing gene in geleophysic dysplasia type I (GPHYSD1), the severe forms of which show early lethality and a similar clinical and radiographic presentation to Al-Gazali skeletal dysplasia. (5,16,18) In addition to ADAMTSL2, FBN1 and LTBP3 genes are also associated with GPHYSD2 and GPHYSD3, respectively. Disease-causing variants are spread across the ADAMTSL2 protein (Fig. 3A), whereas variants in FBN1 are localized in one specific TB5 domain. Five individuals with LTBP3associated GPHYSD and varying degrees of disease severity have also been described. (19) The evident genetic heterogeneity in geleophysic dysplasia and the variation of phenotype severity in affected individuals indicate the broad disease spectrum.
Herein, we suggest that Al-Gazali skeletal dysplasia is the most severe form of ADAMTSL2-related conditions and that there is a phenotypic continuum in geleophysic and Al-Gazali skeletal dysplasias. The distinction between two conditions can be made on the basis of radiological features. The hallmarks, such as generalized osteosclerosis, severe brachydactyly and a very short first metacarpal, undermodeling of the long bones, broad ribs, and mild misalignment of the elbow, are unique for Al-Gazali skeletal dysplasia, and not commonly found in GPHYSD1 (Supplemental Table 2). However, it is likely that the milder phenotype of Al-Gazali skeletal dysplasia, such as in patient 8 in this study, overlaps with the most severe features of geleophysic dysplasia. (18) The pseudogene region of the ADAMTSL2 gene is poorly covered in the major genomic databases using GRCh37/hg19 genome build (gnomAD v2.1.1), creating read alignment challenges when using exome or genome sequencing for clinical diagnostics. (20) In addition, ADAMTSL2 has been identified as one of the eight genes where discrepant variant calls are influenced by the choice of reference assembly (GRCh37 versus GRCh38). (21) ADAMTSL2 coverage is significantly improved in the GRCh38 genome build, as found in gnomAD v.3.1.2 (Fig. 6), indicating that the read alignment challenge is caused by selection of reference genome. GRCh37 is still widely used in whole genome or exome sequencing analysis in the clinical diagnostics setting. If diagnosis of Al-Gazali or GPHYSD1 is suspected based on radiographic findings but only single candidate variant is detected in ADAMTSL2, one should manually inspect the poorly mapped reads for potential disease-causing variants. Alternatively, more advanced sequencing methods, such as long-read sequencing, would enable more accurate variant calling. The identification of disease variants that are located in the pseudogene region of ADAMTSL2 highlights the importance of meticulous phenotyping and selection of the appropriate genetic analysis method, including the analysis of non-coding exon of the ADAMTSL2 gene.
The direct analysis of variant impact on ADAMTSL2 gene expression in patient samples is complicated because of low ADAMTSL2 mRNA levels in commonly clinically available tissues. In our study, we show that it is possible to detect ADAMTSL2 expression in primary human dermal fibroblasts using ultrasensitive detection method, ddPCR. Even though significant differences in ADAMTSL2 expression between patient and control samples were not observed, it is possible that the slight change of expression levels in different culturing conditions (immature and mature ECM) indicates the importance of ADAMTSL2 in ECM assembly and function. Our observation is limited by the very small sample size, and further studies are needed to confirm these findings in primary human cells.
Multiple insights into the role of ADAMTSL2 at organism and tissue level have been generated from in vivo studies. Adamtsl2 À/À mice die soon after birth because of severe lung dysfunction and bronchial epithelial dysplasia, accompanied by accumulation of fibrillin microfibrils in the bronchial epithelia. (22) The described phenotype in mice resembles that of GPHYSD and Al-Gazali skeletal dysplasia patients suffering from narrowing of the trachea and recurrent respiratory infections. Downregulation of Adamtsl2 expression in bronchi after E17.5 suggests that Adamtsl2 could have a crucial role during early development, thus possibly explaining early lethality in affected individuals. The skeletal phenotype of Adamtsl2 KO mice was analyzed in more detail in another study and showed significant shortening of forelimbs and hindlimbs, shorter long bones, slight deformation of vertebral bodies, and shortened calvaria. (23) However, the early perinatal lethality of Adamtsl2 KO mice did not permit examination of the postnatal skeletal phenotype in more detail. Nonetheless, on the cellular level, the formation of chondrocyte columns in the long bones was significantly affected in conditional Adamtsl2-Col2a1Cre mice, recapitulating the skeletal dysplasia phenotype in GPHYSD1. (23) A similar observation was made in the Prx1-Cre conditional Adamtsl2 knockout model, where Adamtsl2-Prx mice had shorter forelimb and distal bones and wider diaphyseal and metaphyseal regions. (24) Contrasting observations in regard to growth plate phenotype were noted in both studies. Whereas Adamtsl2-Prx mice had normal growth plate morphology, the Col2a-mediated chondrocyte-specific deletion led to growth plate abnormalities. (23,24) Moreover, Adamtsl2-Prx mice exhibited shorter and fibrotic Achilles tendons, with an excess of FBN1 microfibrils. The tendon phenotype of Adamtsl2 mice has further been studied using the tendonspecific Adamtsl2 deletion model (Adamtsl2-Scx). Adamtsl2-Scx mice also had shorter limbs but an overall milder bone phenotype compared with Adamtsl2-Prx mice. (22,24) The biological mechanism of how pathogenic variants in ADAMTSL2 lead to skeletal dysplasia are thought to be through abnormalities in TGFβ signaling and/or affected microfibrillar network composition and morphology. (3,5) It is not entirely clear if the overactivation of the TGFβ signaling pathway is a secondary consequence of affected ECM organization. We did not observe differences in TGFβ signaling as measured by pSMAD2 levels. This is consistent with findings in another type of acromelic dysplasia, Weill-Marchesani syndrome, where no changes in TGFβ signaling were noted in primary human fibroblasts. (4) Moreover, the overactivation of TGFβ signaling in bronchial tissue in mice was observed only during the embryonic period and was not rescued by TGFβ antibody treatment. (22) The molecular and in vivo data provide evidence that ADAMTSL2 binds to ECM components FBN1, FBN2, and LTBP1, and is important for proper ECM function. (5,22,24) Different distribution pattern of FBN1 microfibrils was found in cultured cells from patient 1 with Al-Gazali skeletal dysplasia. Previous studies in mice show evident abnormalities of various ECM components upon genetic manipulation of Adamtsl2 in connective tissues. (22)(23)(24) However, more extensive studies of ECM in human samples are needed to confirm the pathogenic effects of ADAMTSL2 variants on ECM assembly and function. In a review by Stanley and colleagues, a conceptual model was proposed to decipher the interaction between several ECM proteins within the acromelic dysplasia complex, where FBN1 microfibrils serve as a crucial ECM hub to bring ADAMTS(L) and LTBP proteins and control TGFβ and BMP signaling pathways. (3) Furthermore, Steinle and colleagues described five unrelated individuals with dermatosparaxic Ehlers-Danlos syndrome (dEDS) presenting with generalized joint hypermobility and fragility of the connective tissue caused by an autosomal dominant variant in ADAMTSL2 (c.1261G>A, p.(Gly421Ser)). (25) The contrasting clinical features of connective tissue laxity in dEDS versus tissue rigidity in GPHYSD1 and Al-Gazali skeletal dysplasias suggests the possible gain-of-function effect of the p.(Gly421Ser) variant and thus further expands the spectrum of ADAMTSL2-related disorders.
In conclusion, this study adds Al-Gazali skeletal dysplasia to the clinical and genetic spectrum of ADAMTSL2-related disorders and indicates the need for further molecular studies to understand the impact of different mutations on phenotypic variability and to explain the summative interaction of important molecular players within the ECM microenvironment.