Osteoporosis and skeletal dysplasia caused by pathogenic variants in SGMS2

Mechanisms leading to osteoporosis are incompletely understood. Genetic disorders with skeletal fragility provide insight into metabolic pathways contributing to bone strength. We evaluated 6 families with rare skeletal phenotypes and osteoporosis by next-generation sequencing. In all the families, we identified a heterozygous variant in SGMS2, a gene prominently expressed in cortical bone and encoding the plasma membrane–resident sphingomyelin synthase SMS2. Four unrelated families shared the same nonsense variant, c.148C>T (p.Arg50*), whereas the other families had a missense variant, c.185T>G (p.Ile62Ser) or c.191T>G (p.Met64Arg). Subjects with p.Arg50* presented with childhood-onset osteoporosis with or without cranial sclerosis. Patients with p.Ile62Ser or p.Met64Arg had a more severe presentation, with neonatal fractures, severe short stature, and spondylometaphyseal dysplasia. Several subjects had experienced peripheral facial nerve palsy or other neurological manifestations. Bone biopsies showed markedly altered bone material characteristics, including defective bone mineralization. Osteoclast formation and function in vitro was normal. While the p.Arg50* mutation yielded a catalytically inactive enzyme, p.Ile62Ser and p.Met64Arg each enhanced the rate of de novo sphingomyelin production by blocking export of a functional enzyme from the endoplasmic reticulum. SGMS2 pathogenic variants underlie a spectrum of skeletal conditions, ranging from isolated osteoporosis to complex skeletal dysplasia, suggesting a critical role for plasma membrane–bound sphingomyelin metabolism in skeletal homeostasis.

2 Ettlingen, Germany). Once acquired, the Raman spectra were baseline corrected (rubber band, 5 iterations) to account for fluorescence, and the following previously described (1). Raman parameters were calculated: i. The mineral/matrix ratio was expressed as the ratio of the integrated areas of the v2PO4 (410-460 cm −1) to the amide III (1215-1300 cm−1) bands. This parameter reports the amount of mineral per amount of organic matrix per volume analyzed and has been shown to correlate with ash weight measurements (2) and to be directly proportional to bending stiffness and failure moment (3).
ii. Nanoporosity (approximated by the ratio of the integrated areas of the spectral slice 494-509 cm-1 of polymethylmethacrylate (PMMA) and Amide III band, a recently introduced parameter (4) that was not considered in our previous work (5) served as a proxy for tissue water content. In bone tissue, water generally exists within pores and bound to the matrix (6). Within the voxel analyzed (in our case 1x1x1.5 µm 3 ), PMMA will occupy space that is void of either mineral or organic matrix, thus in the present case the pores containing PMMA are in the sub-micron scale (hence the term "nanoporosity"). Matrix-bound water is not a monolayer but rather a series of coordinated water molecules, the outermost of which may be removed by the tissue processing (dehydration through a series of alcohols, acetone, and subsequent PMMA embedding). Thus, the water content inferred in the present study may be loosely coordinated water as well as water that was present in canaliculi. Bone tissue water content has been shown to influence the mechanical properties of bone (6) and was recently found altered in a mouse model for osteogenesis imperfecta (7).
iii. The relative glycosaminoglycan (GAG) content was defined as the GAG / total organic matrix ratio, calculated from the ratio of the integrated areas of the CH3 band ~ 1365-1390 cm−1 (characteristic of GAGs) (8) to the amide III bands (1215-1300 cm−1), respectively. Previously, this band was identified as proteoglycans (5) which are protein backbones decorated with GAGs. Since we actually use the CH3 band from the GAG chain, we use here the more precise denomination, GAG. iv. Finally, the relative pyridinoline content (a major trivalent collagen cross-link) was calculated as the absorbance height at 1660 cm-1/area of the amide I (1620 -1700 cm-1) (2,9). Enzymatic collagen crosslinks have been shown to affect both stiffness and toughness (10).

EVALUATION OF SGMS2 GENE EXPRESSION
The mRNA expression of Sgms2 in different tissues was evaluated in 22-weeks old female C57BL/6N mice (Charles River). Mice where anesthesized with Ketador/Dexdomitor (Richter Pharma/Orion Pharma), bled, and euthanized by cervical dislocation. Soft tissues were dissected, snap-frozen in liquid nitrogen, and stored at -80C until RNA preparation. Flushed mid-diaphysel tibial bone (cortical bone) and vertebral body L6 were stored in RNAlater at -80C until RNA preparation.
Liver, kidney, spleen, lung, muscle, heart, aorta and thymus were homogenized in RLT buffer with 1% 2mercaptoethanol using a TissueLyser, followed by RNA preparation using the RNeasy mini kit (Qiagen).
Bone tissues (cortical bone and vertebra), fat (retroperitoneal, gonadal and brow fat), brain cortex and hypothalamus were homogenized in TRIzol reagent (Life Technologies, Carlsbad, CA, USA) using a Tissue Lyser. After centrifugation to remove cell debris, the TRIzol homogenate was mixed with an equal volume of chloroform, centrifuged and the aqueous phase recovered. The aqueous phase was mixed with an equal volume of 70% ethanol and added to an RNeasy spin column. Thereafter the isolation followed the protocol from the RNeasy mini Kit (Qiagen).
Murine bone marrow macrophages and osteoclasts were cultured from bone marrow from male C57BL/6 mice as previously described (11,12). Bone marrow macrophages were cultured in 30 ng/ml M-CSF or in 30 ng/ml M-CSF and 4 ng/ml RANKL for 72 h to induce osteoclast differentiation. Cells were lysed in RLT buffer with 1% 2-mercaptoethanol, RNA was isolated using the RNeasy micro kit (Qiagen).
Primary osteoblasts were isolated by sequential digestion of calvarial bone dissected from 3-5 days old C57Bl/6N pups as previously described (13,14). Primary osteoblasts were cultured in the absence or presence of 100 ng/ml BMP-2 for 7 days prior to lysis in RLT buffer with 1% 2-mercaptoethanol, and RNA isolation using the RNeasy micro kit (Qiagen).
RNA was converted to cDNA using High-Capacity cDNA Reverse Transcription kit (Applied Biosystems).
Quantitative real-time PCR analysis for Sgms2 was performed in the StepOnePlus Real-Time PCR system using pre-designed TaqMan Assay (Assay no Mm00512327_m1) and 18S as internal control (cat no 4310893E).

FACS analysis of peripheral blood monocytes
Peripheral blood monocytes can be divided into subtypes based on their surface expression of CD14 and

In vitro osteoclast differentiation on plastic and bone
Two separate in vitro osteoclast differentiation experiments were performed. In the first experiment, monocytes from F1-1 and two age and gender matched controls and F2-1 and one age and gender matched control were differentiated into osteoclasts on plastic and bone. In the second experiment, more age and gender matched healthy controls were included and the resorptive capacity analyzed. For all in vitro osteoclast differentiation experiments, CD14 + monocytes were purified from PBMCs using CD14 MicroBeads and MACS columns according to the manufacturer's instructions (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). Cells were then seeded in 96-well plates (3x10 5 cells/cm 2 ) with or without

Analysis of in vitro osteoclast bone resorption
CD14 + monocytes were cultured on bone discs as described above for 14 days. Bone resorption pits on bone discs were visualized by reflective light microscopy. The amount of collagen type I fragments (CTX-I) released was measured in the cell culture media at the time of media renewal (every three to four days), using a commercial ELISA (IDS Immunodiagnostics Systems cat no AC-07F1). TRAP5b was analyzed in culture media using commercial ELISA (IDS Immunodiagnostics Systems cat no SB-TR201A).

Staining for the actin ring
Actin organization was studied at day 8 on bone. Cells were fixed in 4% paraformaldehyde, washed with PBS and then permeabilized using 0.1% Triton X-100 in PBS for 10 min at 4°C. After washing again with PBS, the cells were stained with rhodamine conjugated phalloidin (Life Technologies cat no R415), 5 U/ml in 2 % BSA/PBS, for 20 min at 4°C. Following washing in PBS the cells were mounted in Prolong Gold Mountant (Life Technologies cat no P-36931).

METABOLOMICS
We performed untargeted metabolomic studies with Direct-infusion High-resolution mass spectrometry from dried whole blood spots of 12 patients, 8 healthy family members and in-house controls. Dried blood spot sample extraction (3 mm, ~3.1 μL whole blood) was performed by addition of 140 μL IS methanol, followed by a 20-minute ultra-sonication step. Samples were diluted with 60 μL 0.3% formic acid and filtered using a methanol preconditioned 96 well filter plate (Acro prep, 0.2 μm GHP, NTRL, 1 mL well; Pall Corporation, Ann Arbor, MI) and a vacuum manifold. The filtrate was collected in a 96 well plate (Advion, Ithaca, NY). A TriVersa NanoMate system (Advion, Ithaca, NY) controlled by Chipsoft software (version 8.3.3, Advion) was mounted onto the interface of a Q-Exactive high-resolution mass spectrometer (Thermo Scientific™, Bremen, Germany). The Q-Exactive high-resolution mass spectrometer was operated in positive and negative ion mode with automatic polarity switching. Scan range was 70 to 600 m/z, resolution was 140,000 at m/z = 200. Samples were analyzed in triplicate.
Data acquisition was performed using Xcalibur software (version 3.0, Thermo Scientific™, Bremen, Germany). Using MSConvert15 (ProteoWizard Software Foundation), raw data files containing scanning time, mass over charge (m/z) and peak intensity were converted to mzXML format in Profile mode. A peak calling pipeline was developed in R programming language. Mass peak identification and annotation was conducted by matching the m/z value of the mass peak with a range of two parts per million to metabolite masses present in the Human Metabolome Database, version 3.6 (15). Z-scores for each metabolite were calculated based on control samples measured in the same run and the metabolites were ranked based on Z-scores; a Z-score of 1 meaning the metabolite is elevated with one standard deviation compared to in-house controls (n=20-40). A metabolite level was regarded elevated when the Z score was >1 and decreased when the Z score was <-0.8 when compared to in-house controls. In addition, metabolites of interest were excluded if they were altered in unaffected controls (family members) provided by the same center as the patients as such alterations are likely due to artefacts (differences in type of Guthrie card or storage conditions). Multiple comparisons t-test with Bonferroni correction was used.

Figure S6. Findings in Raman spectroscopy analysis of transiliac bone biopsies. Transiliac bone biopsies
were obtained from patients F1-1 and F2-1. F1-1 had fluorescent labels unlike F2-1. Therefore, area selection for analysis was based on the double labels for F1-1 (precise tissue age) while for F2-1 it was based on Ca content in qBEI images (selected areas with low Ca content; ongoing bone formation).
Outcome values were compared with previously published regression models describing the variation of the spectroscopic parameters in healthy females as a function of chronological age as well as tissue age within the same donor (16). Three tissue ages were analyzed; in F1-1, TA1 = mid-distance between the second label and the mineralizing front, TA2 = mid-distance between the two labels, TA3 = 2 µm behind the first label. In F2-1, distances were matched from the mineralizing front in qBEI. The results indicate that: 1. Mineral/matrix: At TA1, there were no striking differences between the 2 cases and age-matched healthy. But both patients had lower values at TA2 and TA3, suggesting a deceleration of mineral accumulation kinetics.
2. Nanoporosity (a surrogate for tissue water content) did not show any consistent trend. At TA3, the higher values would be consistent with the lower mineral/matrix ratio as mineral mainly displaces water in the bone tissue.    Black squares, patients; black circles, controls. At sub-optimal concentration of RANKL (0.5 ng/ml), more osteoclasts were formed when the same number of monocytes were cultured from Patient F1-1 compared to two matched controls whereas monocytes from Patient F2-1 generated fewer osteoclasts on plastic compared to a matched control. At optimal concentration of RANKL (2 ng/ml) no obvious differences between osteoclasts from patients and healthy controls could be observed.   Incorporation of 14 C-choline into phosphocholine (PC) and sphingomyelin (SM) was analyzed by autoradiography (top). Lipids were stained with iodine vapor to verify that total lipid content between extracts was comparable (bottom). Note that part of the TLC analysis (bottom-row, left) is also shown in Figure 7C.        (18). Parameters and abbreviations are explained in Figure S5.