Deficiency of the SMOC2 matricellular protein impairs bone healing and produces age-dependent bone loss

Secreted extracellular matrix components which regulate craniofacial development could be reactivated and play roles in adult wound healing. We report a patient with a loss-of-function of the secreted matricellular protein SMOC2 (SPARC related modular calcium binding 2) presenting severe oligodontia, microdontia, tooth root deficiencies, alveolar bone hypoplasia, and a range of skeletal malformations. Turning to a mouse model, Smoc2-GFP reporter expression indicates SMOC2 dynamically marks a range of dental and bone progenitors. While germline Smoc2 homozygous mutants are viable, tooth number anomalies, reduced tooth size, altered enamel prism patterning, and spontaneous age-induced periodontal bone and root loss are observed in this mouse model. Whole-genome RNA-sequencing analysis of embryonic day (E) 14.5 cap stage molars revealed reductions in early expressed enamel matrix components (Odontogenic ameloblast-associated protein) and dentin dysplasia targets (Dentin matrix acidic phosphoprotein 1). We tested if like other matricellular proteins SMOC2 was required for regenerative repair. We found that the Smoc2-GFP reporter was reactivated in adjacent periodontal tissues 4 days after tooth avulsion injury. Following maxillary tooth injury, Smoc2−/− mutants had increased osteoclast activity and bone resorption surrounding the extracted molar. Interestingly, a 10-day treatment with the cyclooxygenase 2 (COX2) inhibitor ibuprofen (30 mg/kg body weight) blocked tooth injury-induced bone loss in Smoc2−/− mutants, reducing matrix metalloprotease (Mmp)9. Collectively, our results indicate that endogenous SMOC2 blocks injury-induced jaw bone osteonecrosis and offsets age-induced periodontal decay.


Results
SMOC2 mutation in a patient produces specific dental abnormalities and mild skeletal dysplasia. We initially reported that SMOC2 deficiency severely disrupts tooth formation, as patients with a loss-of-function of this secreted matricellular protein display striking phenotypic defects in primary and permanent dentitions including microdontia, oligodontia, dysplastic root formation and alveolar bone hypoplasia 12 .
Continual clinical monitoring of a 9 year-old affected girl showed similarly severe defects of permanent incisors, and a potentially progressing skeletal dysplasia 12 . While frontal and lateral radiographic views (Fig. 1A,B) of the skull showed no severe skull malformations, lumbar vertebrae were markedly flattened (displaying platyspondyly; Supplementary Fig. S1A-C,E) and the presence of a hyperlordotic curved spinal column suggested skeletal defects might worsen with age, just as wider iliac wings suggested a small degree of skeletal dysplasia ( Supplementary Fig. S1D) not observed in other regions ( Supplementary Fig. S1F). Tooth alveolar bone hypoplasia appeared generalized ( Fig. 1A; Bloch-Zupan et al. 12 ). For most teeth, cone-beam computed tomography (CBCT) imaging confirmed that upper (maxillary) teeth and corresponding alveolar bone were relatively less affected (Fig. 1C), compared with the mandibular dentition which presented severe oligodontia, microdontia and dysplastic roots (Fig. 1D). The right first permanent mandibular molar displayed however macrodontia (Fig. 1C,D, red arrowhead). Extracting CBCT images showed the relations of alveolar bone, tooth, and adjacent root structure. Full-frontal views showed severe dental anomalies in lower jaw with alveolar bone hypoplasia (Fig. 1E). Image extraction of isolated teeth showed that a micro-root structure consistently accompanies patient microdontia (Fig. 1E,F).

Smoc2-GFP reporter analysis suggests fetal origins of SMOC2-deficiency defects.
To understand the physiopathology of these dental and mineralized tissues anomalies, we turned to mouse models. While Smoc2 expression at late (fetal) stages of mouse development suggested its disruption could produce tooth growth deficiencies 12 , by performing lineage tracing using a heterozygous Smoc2 modified allele with an in-frame green fluorescent protein (GFP) reporter insertion 11 we confirmed the presence of this reporter in dental progenitor cells. In embryonic (E) 14.5 Smoc2-GFP mice, GFP immunostaining labeled first molar dental follicle mesenchymal populations ( Fig. 2A, red arrowheads). At E18.5 and post-natally, GFP was highly enriched in the mesenchyme surrounding the labial and lingual cervical loops of the lower incisor (Fig. 2B,C) and molar cervical loops (Fig. 2D), but absent from epithelial cervical loop stem cell zones allowing rodent incisors to continuously grow and produce enamel 19 . Other GFP-labeled sites at fetal stages were the telencephalon (Supplementary Fig. S2A), perioptic mesenchyme ( Supplementary Fig. S2B, red arrowhead), basal nasal epithelial cells ( Supplementary Fig. S2C), and hair follicle vibrissae ( Supplementary Fig. S2D). Altogether, GFP analysis showed dynamic Smoc2 expression often localizing to proliferative zones with a property of continuous selfrenewal, like the intestinal crypts 11 Fig. S3D) − Smoc1 being a homologous gene whose inactivation produces lethality with severe ocular and limbs malformations 21 . On a gross level, skeletal morphology appeared normal at fetal and perinatal stages ( Supplementary Fig. S4), but detailed examination revealed specific defects. Enamel (the hardest mineralized tissue) of mouse teeth normally has an orange/yellow color, whereas in Smoc2 −/− mutants it had a whiter appearance ( We performed X-ray micro-computed tomography (μCT) imaging and surface rendering to assess tooth defects. Both incisor and molar number was normal in most of the mutants analyzed, but discrete cusp morphology changes were observed, still allowing normal dental occlusion (Fig. 3E,F; Supplementary Fig. S5A,B,E,F). Two specific dental anomalies were observed, namely a reduction of the 1st-3rd molar field length (  year-old girl with a homozygous SMOC2 mutation (also see Ref. 12 ). (A,B) Respective frontal and lateral radiographic views of the skull. Cranial bones and sutures, sella turcica, and orbit structures are normal except for the maxilla and mandible. (C) A full-frontal view of the dentition. The lower dentition displays more severe oligodontia and microdontia compared with the upper teeth. Affected teeth are typically 15-20% smaller than normal, except for the lower right first molar (red arrowhead), which is bigger. (D) Extracted CBCT images of full-lateral views. Defects include macrodontia (red arrowhead) and microdontia (yellow arrowhead). (E,F) Extracted frontal and lateral views with the lower dentition displaying more severe oligodontia and microdontia compared with the upper teeth. Image extraction of isolated teeth (using the Analyze 11.0 software) shows that a micro-root structure consistently accompanies microdontia. A false blue background was added using the Analyze 11.   Fig. 3). mRNA-seq analysis was also performed on E18.5 microdissected mandibular bone. This analysis (Supplementary Table S6) revealed a down-regulation of dentin sialophosphoprotein (Dspp), a major dentin extracellular matrix protein instructing dentin, root, and periodontal growth 27   www.nature.com/scientificreports/ cytokine signaling and bone healing 28 , targets potentially involved in both mouse and patient tooth and skeletal alterations and deficits in healing response. We are currently examining mechanisms to understand these targets.  www.nature.com/scientificreports/ Consortium for a similar C56BL/6 Smoc2 loss-of-function 29 . Upon tail-clipping for genotyping, we observed more truncated tails (compared to WT), suggesting injury-induced growth deficits (data not shown). Tail growth requires a combination of cartilage, bone, vascular, and skin repair-not amenable to investigate which target tissue might produce defects. To specifically test the capacity of dental alveolar bone to heal, we used an experimental maxillary molar avulsion injury model. Two-month-old males were employed to avoid potential effect of estrogen variations on bone healing found in females. Smoc2 −/− mutants and age-matched controls were deeply anesthetized before the left 1st maxillary molar crown was removed (leaving the underlying molar roots intact) creating an experimental alveolar bone lesion. Six-week post-injury, μCT scanning images revealed that Smoc2 −/− mutants exhibited extended bone loss around the injured first molar compared to controls (Fig. 4A,D WT; 4B,E Smoc2 −/− , yellow arrowhead). Bone loss in mutants included extensive 1st, and 2nd molar roots resorption. This appeared as a sort of accelerated "osteonecrosis" phenotype propagated as 2 nd molar root resorptions in 5 out of 7 mice (Fig. 4E, red arrowhead; for other representative images, see Supplementary Fig. S9). In 2 out of 7 mutants resorption of the 2nd molar crown was also observed. Alveolar bone/root mineral volume compared with total volume is tabulated in Supplementary Fig. S10, confirming significant reductions in jaw bone healing in Smoc2 −/− mutants. As SMOC2 is activated by NF-κB signaling 9 , mutants could display alterations in inflammatory response and bone homeostasis. We tested if this differential bone loss (temporally correlating with the inflammatory stage of bone healing) could be prevented by a short (10-day) systemic treatment with a broad acting NSAID anti-inflammatory COX2 inhibitor, ibuprofen. Ibuprofen was provided for 10 days following surgery at a recommended pain-relieving dosage for mice (30 mg/kg body weight) 30 to limit post-operative inflammatory response. Remarkably, augmented bone loss found 6 weeks after injury in Smoc2 −/− mutants disappeared after ibuprofen treatment. Hence, only 2 out of 7 mutants had a 2nd molar root resorption, with much less bone loss than untreated mutants (Fig. 4C,F, Supplementary Fig. S9). Histologically, at 6 weeks post-injury, groups of multinucleated osteoclast cells appeared at the resorbed roots in the remnants of mineralized tissue in Smoc2 −/− mutants (compare WT 4G vs. arrowheads in mutants in Fig. 4H,I: untreated and after ibuprofen treatment). Note that, in all WT animals analyzed (n = 7), ibuprofen treatment did not alter bone repair at 6 weeks as observed by μCT analysis (Supplementary Fig. S9).
Interestingly, both GFP immunofluorescence and RT-PCR confirmed increases in Smoc2-Gfp reporter expression at 4 days following tooth injury, suggesting that Smoc2 activation is part of a bone progenitor regenerative response ( Fig. 5A-C). We also found that Smoc2 −/− mutants display increased osteoclast activity, as monitored by TRAP (tartrate resistant acid phosphatase) staining 7 days after injury around the surgical extraction site (Fig. 5D,F WT; 5E,G Smoc2 −/− ). TRAP staining indicated that osteoclasts were also increased at the 2nd molar roots site where the resorption of distal root was observed in Smoc2 −/− mutants (Fig. 5G, red arrow, quantitated in Fig. 5H).

SMOC2 loss of function induces chronic periodontitis. Augmented inflammation or defects in
osteogenic or osteoclast-induced bone response could produce periodontal disease and alveolar bone loss-a predominant clinical consequence of chronic inflammation. To assess such possible defects, we performed μCT imaging on 1-year old mice, and found that ~ 40% of Smoc2 −/− mutants displayed spontaneous chronic periodontitis due to aging (n = 10, Supplementary Table S9). Smoc2 −/− mutants exhibited alterations characteristic of chronic inflammation marking periodontal disease. Thus, aging increased age-dependent bone and root resorption in the mutants (Fig. 6A,C Supplementary Fig. S13). We propose a coherent model explaining why Smoc2 deficiency causes increased bone and root loss following tooth avulsion injury, and how ibuprofen might restore these changes ( Fig. 7; see discussion).

Discussion
Mouse Smoc2 mutants reveal accumulative effects of matricellular proteins in growth maintenance. We and others 11 initially reported that the absence of SMOC2 in young mice did not dramatically alter skeletal patterning, nor disrupt organogenesis. One might interpret this as a limited (potentially redundant) role during normal development. Concurrent reports, though, indicated that SMOC2 promotes angiogenesis 14 , and increases TGF-β signaling regulating fibrosis 15,16 . Such effects might suggest SMOC2 roles in tissue maintenance, which could be offset during pathological fibrosis or cancer metastasis. Our data investigating aging and tooth www.nature.com/scientificreports/ www.nature.com/scientificreports/ injury response indicate a variety of stage-and cell type-specific SMOC2 functions. While SMOC2 overexpression in osteoblasts inhibits ossification 32 , we observe its deficiency produces age-dependent periodontal bone loss. Canine SMOC2 has been proposed as a key modulator of facial length, with reductions correlating with breed-specific brachycephaly 5 . In humans, SMOC2-mutated patients have maxilla and mandible defects that could be related to oligodontia and microdontia inducing alveolar bone reductions 6 . Mouse Smoc2 −/− mutants have a significant reduction in the size of the molar field, and also exhibit sporadic appearance of supernumerary molars, alterations in molar cusp patterning and smaller roots. We initially proposed facial changes could be secondary to reduced neural crest cell migration into the branchial arches 12 , as Smoc2 knockdown in zebrafish produces severe craniofacial hypoplasia 4 . Deletion of the homologous Smoc1 gene produces a severe overall growth deficiency, as well as limb syndactyly, and bone hypoplasia 21 . Other non-structural matricellular proteins, such as tenascins, are responsible when mutated for subtle tissue-specific phenotypes in neuroepithelial or osteogenic stem cell niches, and have been used to investigate pathological susceptibilities 33 .
SMOC2 is required for bone repair and during periodontal aging. With aging, Smoc2 −/− mutant mice exhibit spontaneous alveolar bone deterioration and maxillary molar root resorption mimicking agedependent dental and periodontal changes such as periodontal disease encountered in elderly human patients. We observed that short-term bone injury produces an osteonecrotic-like response in Smoc2 −/− mutants. Matricellular proteins like CCN1 aid wound repair by directly interacting with integrin receptors promoting tissue attachment 34 . SMOC2 induces integrin-induced stress fiber attachment 9 and is required for a fibroblast to myofibroblast transition required for wound healing 15 . Injuries such as tooth extractions also induce an acute, highly regulated inflammatory response. Chronic inflammation produces increased bone loss in bisphosphonateinduced jaw osteonecrosis models 35 . Smoc2 deficiency attenuates the inflammatory response and increase of interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), and IL-1β levels caused by bleomycin injury-induced  www.nature.com/scientificreports/ dependent inflammatory cytokine and chemokine induction is followed by myofibroblast migration, collagen synthesis, and osteoclast-directed bone breakdown 16 . The early SMOC2-dependent pro-inflammatory response is necessary for post-extraction alveolar bone repair 37 . Both intramembranous and alveolar bone repair also require de novo coupling with angiogenesis 38 wherein SMOC2 functions in vascular patterning may be augmented during injury-induced revascularization 14 . SMOC2 is differentially increased in failing human heart, potentially marking injury-induced myocardial ischemia and/or fibrosis 39 . Smoc2 mutants could serve as a novel model to understand tissue repair, particularly in non-invasive therapy as anti-inflammatory (ibuprofen) treatment was shown to normalize defects in our experimental system.

SMOC2 may normally block osteoclast "over-activation" during bone repair.
Our results suggest that SMOC2 signaling is a key inflammatory regulator. SMOC2, like other SPARC matricellular proteins binds collagens which can affect matricellular ECM assembly and signaling. SPARC matricellular proteins have required functions during embryonic development. These roles persist in the adult, functioning as key clues for remodeling during inflammation and wound repair 40 . SMOC2 is rapidly reactivated during healing response, just as it is dysregulated during pathogenic fibrosis 15,16 . Evidence supports the premise that SMOC2 is a component of a pro-inflammatory secretome 10 . We show that aging Smoc2 −/− mutants exhibit periodontal bone loss-a sign of chronic inflammation. Injury triggers a highly abnormal cytokine/lymphokine response with Smoc2 −/− mutants displaying marked bone loss. A short term (10-day) systemic treatment with the NSAID ibuprofen restores bone regenerative response in mutants, this phenotypic rescue solidifying the connection with inflammatory signaling. Collectively, our results suggest a fundamental role of SMOC2 in tooth/periodontium injury response illustrated in Fig. 7. In the normal (WT) situation, the presence of SMOC2 prevents over-activation of osteoclasts, lowering Mmp9, and allowing reparative bone remodeling in a controlled effective manner. Smoc2 −/− mutants display increased expression of Mmp9, an abundant oral endopeptidase showing increased expression during periodontal decay 41,42 . We find SMOC2 deficiency promotes osteoclast activation, increasing Mmp9, TRAP activity, and COX2, and leading to dysregulation of inflammatory cells and cytokines. We hypothesize that the acute inflammation response following tissue injury persists in Smoc2 −/− mutants, thereby damaging healthy surrounding tissues and causing further bone and root destruction 43 . Ibuprofen can diminish osteoclast number 44,45 , prevent excessive inflammation 46   www.nature.com/scientificreports/ tooth injury, display normalized Mmp9 and COX2 expression, and "rescued" bone repair. Clinically, short-term ibuprofen treatment does not impair post-tooth extraction bone repair or is detrimental to dental implant osseointegration 48 . Recent studies even indicate NSAIDs can inhibit periodontal osteoclast activation following bacterial lipopolysaccharide-induced inflammation 49 . The effects of NSAIDs in bone repair need to be better defined 50 , in particular the potential positive effects of NSAID administration may be beneficial in chronic pathological situations such as long-term bacterial infection, chronic stress, and chronic force. SMOC2-regulated events required during alveolar bone repair and/or altered during periodontal bone decay could indeed be reactivated-and therapeutically controlled-in other progenitor cell populations, serving as potential clinical targets for further directed therapeutics in the aging population.

Methods
Patients. The patients with a homozygous SMOC2 mutation were initially described 12  www.nature.com/scientificreports/ mapping relative to the mm10/NCBI37 mouse reference genome were performed using Tophat. Only when unique aligned sequence reads were obtained was gene expression quantification performed using HTSeq-0.6.1 (described at https ://www-huber .embl.de/users /ander s/HTSeq /doc/overv iew.html). For each transcript, reads per exon kilobase of model per million sequence reads (RPKM) were then converted to raw read counts, then added for each gene locus. Data normalization was performed as described 53 and resolved with the DESeq Bioconductor package. A Benjamini and Hochberg-based multiple testing model provided adjusted p values 54 . Transcripts alterations with a RPKM > 1, and an adjusted p < 0.05 were considered.

Molar avulsion injury.
On the day of wound creation (D0), seven 2-3 month-old male Smoc2 −/− mutants and age-matched control mice were deeply anesthetized with intraperitoneal administration of 9% ketamine HCl and 1% xylazine (10 μl/g body weight). A number 11 blade scalpel was used to loosen the maxillary left first molar. The tooth was then broken by removing the crown (leaving the roots intact) by rotating clinical forceps causing the top of the tooth to crack. After removal, the extracted crown was verified for integrity. All Smoc2 −/− mutants and age-matched controls exhibited no surgical complications following tooth extraction. At 6 weeks after extraction, seven mice per group were sacrificed and μCT performed. For NSAID treatment, seven Smoc2 −/− mutants and age-matched controls were given ibuprofen (Children's Motrin) at 30 mg/kg in the drinking water for 10 days 30 . Molecular responses in bone and root underlying either the 1st or 2nd molar were determined by RT-PCR 7 days after crown extraction (~ 30 mice per group).
Histological analysis and TRAP staining. Heads of 2-month-old Smoc2 mutants and age-matched controls were fixed in 4% PFA overnight, rinsed, and demineralized in 10% EDTA 55 . After several water washes, serial dehydration was performed in a series of graded ethanol solutions, followed by Histosol R clearing, and paraffin embedding at 60 °C. Staining of 8 µm sections with hematoxylin/eosin and Mallory's stain was performed according to standard procedures. To characterize osteoclast activity, a tartrate resistant acid phosphatase (TRAP) labeling kit (387A, Sigma-Aldrich) was used on deparaffinized sections assayed 7 days following molar crown extraction following manufacturer's instructions. In all experiments 6 adult mice for each group were used.
Ethics statement. All methods and experimental procedures were entirely carried out in accordance with all relevant French guidelines and/or regulations and thus were reviewed and approved by the IGBMC institutional safety committee. For procedures involving mice, all animals were maintained and manipulated according to protocols in agreement with the French Ministry of Agriculture guidelines for use of laboratory animals (IGBMC protocol 2012-097) and with NIH guidelines, described in the Guide for the Care and Use of Laboratory Animals. The Smoc2 −/− mouse line was created at the Mouse Clinical Institute (ICS) using Cre and Flp methods 51 , as described in Supplementary Fig. S1 and Supplementary Table S1.