ATP6V1H Deficiency Impairs Bone Development through Activation of MMP9 and MMP13

ATP6V1H is a component of a large protein complex with vacuolar ATPase (V-ATPase) activity. We identified two generations of individuals in which short stature and osteoporosis co-segregated with a mutation in ATP6V1H. Since V-ATPases are highly conserved between human and zebrafish, we generated loss-of-function mutants in atp6v1h in zebrafish through CRISPR/Cas9-mediated gene knockout. Homozygous mutant atp6v1h zebrafish exhibited a severe reduction in the number of mature calcified bone cells and a dramatic increase in the expression of mmp9 and mmp13. Heterozygous adults showed curved vertebra that lack calcified centrum structure and reduced bone mass and density. Treatment of mutant embryos with small molecule inhibitors of MMP9 and MMP13 significantly restored bone mass in the atp6v1h mutants. These studies have uncovered a new, ATP6V1H-mediated pathway that regulates bone formation, and defines a new mechanism of disease that leads to bone loss. We propose that MMP9/MMP13 could be therapeutic targets for patients with this rare genetic disease.

Introduction ATP6V1H is a member of the family of vacuolar ATPases (V-ATPases), and a component of protein complexes responsible for acidification of intracellular compartments in all eukaryotic cells [1]. Specifically, V-ATPases regulate protein degradation, mediate pH homeostasis, and facilitate cellular function and development through acidification. In human, mutations in ATP6B1, which encodes the β-subunit of the apical proton pump mediating distal nephron acid secretion, cause distal renal tubular acidosis, a condition characterized by impaired renal acid secretion, resulting in metabolic acidosis [2]. In osteoclasts, V-ATPases acidify the space that is in contact with the bone surface; this process is essential for bone resorption. Consequently, components of V-ATPases are potential therapeutic targets for the treatment of osteolysis [3]. Moreover, Atp6i-deficient mice exhibit severe osteopetrosis due to loss of osteoclastmediated extracellular acidification; the osteopetrotic phenotype may be due to failure of osteoclast formation during development [4]. Using zebrafish model and patient phenotypic data, we show that haploinsufficiency of the V1H subunit of V-ATPases can result to osteoporosis.
Through the NIH Undiagnosed Diseases Program (UDP) [5][6][7], we identified a 48-year old woman and her 24-year old son who exhibited markedly low bone density. This phenotype segregated with a monoallelic mutation in ATP6V1H. Since osteoporosis is the opposite of the osteopetrosis seen in Atp6i-deficient mice and expected for reduced V-ATPase activity, we studied the phenotype of zebrafish harboring a null mutation in atp6v1h. Indeed, this animal model displayed reduced bone mass, and we demonstrated that the bone defect was mediated by increased activity of mmp9 and mmp13. Small molecule inhibitors of both proteins notably restored bone volume, suggesting that MMP9 and MMP13 could be therapeutic targets for treatment.

Patients
Patient 1 (P1; II.1 in Fig 1A) is a 48-year old woman and Patient 2 (P2; III.1 in Fig 1A) is her 24-year old son. Both exhibited short stature, bone pain and decreased bone density (Fig 1C, 1-5 and S1A Fig). The father of P1 (I.2 in Fig 1C) reportedly had the same features, with dilated frontal sinuses and decreased bone density (S1B and S1C Fig). Exome sequencing The family comes from a three-generation pedigree; affected members are shaded in black (A); Chromatograms showing that the ATP6V1H mutation is present in affected family members, (II.1 and III.1 (B); DNA from I.1 is not available; (C) Radiographs (C) from Patient 1 (P1) (1, 2, and 3) and Patient 2 (P2) (4, and 5) reveal that P1 has scoliosis (C1) and increased radiolucency in tibia (C2). Bone scan shows multiple sites of increased bone mineral density uptake-tip of right scapula and multiple areas of thoracic and lumbar spine, sacrum, both sacroiliac joints, left humeral head, shaft of left femur, right tibia and right ribs (C3). Some increase in radiolucency especially in the lumbar spine (C4), is seen in P2.
doi:10.1371/journal.pgen.1006481.g001 revealed NM_015941.2:c.1158_1159delinsTT (p.Lys386_Asn387delinsAsnTyr, named K386N/ N387Y hereafter) in ATP6V1H that segregated with the osteoporotic phenotype ( Fig 1B). In silico analysis for protein stability predicted that both amino acid changes due to the mutation (K386N and N387Y) can alter protein stability (I-Mutant2.0). To confirm this finding, we generated a stable cell line expressing K386N/N387Y and measured protein half-life. After protein synthesis inhibition by cycloheximide treatment, the mutant protein decayed rapidly compared to the wild-type protein, suggesting that the mutation produced an unstable protein (S2 Fig). Bone but not cartilage is defective in atp6v1h-deficient zebrafish Based on the finding that K386N/N387Y mutant protein was unstable, we reasoned that loss of function mutation in an animal model could address the function of ATP6V1H. We obtained a null mutation of atp6v1h in zebrafish through CRISPR/Cas9 mediated gene knockout (S3 Fig and supplementary information, designated as atp6v1h la439 ). Homozygous mutants died at 10-11 days post fertilization (dpf). Up to dpf 6, the mutant embryos did not appear to have major abnormalities in morphology, but they failed to inflate their swim bladders (Fig 2). Although a retrovirus-induced insertional mutation of atp6v1h has been previously reported for zebrafish, no bone related phenotypes were analyzed [8]. In this study we showed that atp6v1h expression was largely confined to the head region, where cartilage and bone cells first develop, and this expression was completely lost in our mutant (S3 Fig), suggesting true loss of function. Analysis by calcein staining indicated that atp6v1h -/-embryos had greatly reduced or absent calcification of bone cells (Fig 2). Co-staining of bone and cartilage cells showed that cartilage cells were largely unaffected, while mineralized bone was nearly absent (Fig 2). We did not observe any obvious skeletal defects in wild type (+/+) or heterozygous (+/-) embryos, suggesting that complete loss of function for atp6v1h is required to produce the observed embryonic phenotype.
We next determined if the ATP6V1H K386N/N387Y mutation had a dominant effect on zebrafish development. The affected region of the protein is fully conserved in zebrafish atp6v1h (S3 Fig). We generated cDNA constructs containing the same mutation or the wild type sequence as a control to evaluate the effects of the mutation. We next restored bone mineralization in the mutant zebrafish by injecting with wild type mRNA; injecting K386N/ N387Y mutant mRNA did not restore mineralization (S5 Fig), suggesting that the change leads to a loss-of-function mutation. Additionally, injection of wild-type embryos with K3866N/N387Y mRNA had no obvious effect on embryonic development, suggesting that this mutation did not result in gain-of-function S5 Fig).
We then analyzed the phenotype of adult fishes that lacked atp6V1h in one allele (heterozygous, +/-). By eight months after birth, over 50% of the heterozygous adults showed a curved body (Fig 3). X-ray and micro-CT studies revealed that bone mineral density, bone volume and bone surface were all reduced but the later two reductions were more dramatic (S6 Fig). The most remarkable discover is that the vertebrae of heterozygous fish completely lacked calcified structure in the centrum cavity (Fig 3). Additionally, overall skeleton structures appeared smaller and had subtle but notable abnormalities. Furthermore, this adult phenotype was observed in the previously reported retroviral insertion mutant allele of zebrafish atp6v1h gene (S6 Fig) [8].
mmp9 and mmp13 are highly induced by atp6v1h-deficiency To determine the molecular mechanism underlying the bone abnormalities associated with atp6v1h deficiency, we analyzed lineage-specific markers involved in different stages of osteoblast and osteoclast development. First, we determined if atp6v1h was expressed in osteoblast or osteoclast cells of zebrafish by performing high resolution double fluorescent in situ hybridization using atp6v1h probe with the osteoblast marker runx2a or osteoclast marker rank. This analysis showed that atp6v1h had broader co-expression with osteoclast marker rank but less with osteoblast marker runx2a (S4 Fig), which is consistent with the previous report that V-ATPase is enriched in osteoclast cells [1]. Quantitative PCR (qPCR) analysis was further performed to determine mRNA expression of major bone-specific marker genes, including runx2, osx, col1a2, osc, osn and alp for osteoblasts and rank, cstk and mmp9 for osteoclasts between homozygous mutant and sibling embryos. The expression of intermediate and late osteoblast genes in the atp6v1h-deficient zebrafish was marginally reduced (Fig 4A). RNA in situ hybridization (S7 Fig), showing a slight reduction in the expression of the osteoblast genes, confirmed the qPCR data. The expression of the osteoclast genes rank and cstk did not obviously differ from wild type, but mmp9 mRNA levels were vastly elevated in the atp6v1hdeficient zebrafish ( Fig 4A). In situ hybridization demonstrated that the pattern and level of expression of the osteoclast marker rank were similar in the mutant and wild type zebrafish, and that the increased mmp9 mRNA expression occurred primarily in the head region of the zebrafish (Fig 4B), where mmp9 was previously identified to be expressed as an osteoclast marker [9,10]. Considering that we mainly detected mmp9 up-regulation in ATP6V1H-defficient mammalian osteoclast but not in osteoblast and chondrocytes ( Fig 4C and S8 Fig), we hypothesize that mmp9 is mainly up-regulated in zebrafish osteoclasts. It remains to be determined if other cell types are also induced to express excess amount of mmp9.
The finding that mmp9 was up-regulated prompted us to analyze mmp2 and mmp13, previously reported to be involved in mmp9 related bone homeostasis [9,10]. Expression of mmp13a mRNA was significantly increased in atp6v1h-deficient zebrafish embryos, but that of mmp2 was not ( Fig 4A). Zebrafish also have a gene named mmp13b that appears to be a pseudogene since there was no detectable expression by qPCR either in the wild type or atp6v1hdeficient embryos. Further analysis of mmp13a expression by RNA in situ hybridization revealed a pattern and intensity that was very similar to that of mmp9 ( Fig 4B).
To determine if Mmp9 and Mmp13 expression was similarly induced in mammalian cells deficient of Atp6v1h, we performed siRNA knockdown experiments in mouse osteoclast, osteoblast and chondrocyte cell types. We found that both Mmp9 and Mmp13 were induced when Atp6v1h was knocked down in mouse osteoclast cells, but reduced in osteoblast and  Table). Consistent to this finding, we found adult zebrafish +/-for atp6v1h appeared to have more TRAP staining in bone sections (S10 Fig). Overall, these data suggest that a higher osteoclast activity is induced by atp6v1h deficiency.

MMP9 and MMP13 inhibitors rescue bone abnormalities in atp6v1h mutants
Because abnormal expression of mmp9 and mmp13 occurs in human diseases such as cancer, arthritis and cardiovascular disorders, small molecule inhibitors specific to MMP9 and MMP13, respectively, or to both MMP9 and MMP13 were available [11,12]. We treated atp6v1h mutant zebrafish embryos with these inhibitors and found that an MMP13 inhibitor restored bone formation nearly toward normal while an MMP9 inhibitor had a reduced rescue effect (Fig 4D). Treatment with an inhibitor that has dual activity on both MMP9 and MMP13 rescued the bone phenotype to an extent similar to the inhibitor of MMP13 (Fig 4D). These findings suggest that increased mmp9 and mmp13 expression contribute significantly to the bone defect, and targeting MMP13 would likely alleviate this problem.

Discussion
We describe the consequences of atp6v1h mutation in zebrafish, based upon the ascertainment of a family with 3 generations of osteoporosis segregating with a K386N/N387Y mutation in ATP6V1H. We discovered a profound reduction in the amount of mature bone in homozygous atp6v1h zebrafish mutants; cartilage formation was largely unaffected. Since injection of mRNA carrying the human point mutation failed to induce any abnormality in wild type embryos, we propose that K386N/N387Y is a loss-of-function mutation. Given that adult zebrafish heterozygous for atp6v1h deficiency has bone defects and the human mutation was transmitted in a heterozygous manner, this mutation can lead to haploinsufficiency of ATP6V1H protein in humans.
The cellular targets of ATP6V1H function remain to be determined. Osteoblast markers were only marginally affected, suggesting that ATP6V1H does not exert its primary effects on osteoblast formation. The skeletal abnormalities in atp6v1h mutant zebrafish were associated with dramatically increased expression of mmp9 and mmp13, and inhibition of MMP9 and MMP13 significantly restored bone volume. MMP9 and MMP13 are members of the matrix metalloproteinase family, which are important for remodeling of the extracellular matrix in diverse biological processes. The functions of MMP2, MMP9 and MMP13 in skeletal remodeling have been extensively investigated through genetic studies. While Mmp2-null mice have osteoporosis [13], Mmp9-null and Mmp13-null mice exhibit enlarged hypertrophic zones and osteopetrosis [14,15]. The double knockout of Mmp9 and Mmp13 leads to a more dramatic expansion in the length of the hypertrophic zone than in single knockouts, suggesting a synergistic relationship between Mmp9 and Mmp13 [16]. This is consistent with the findings in our atp6v1h zebrafish mutant, in which increased mmp9 and mmp13 expression combined to induce the bone defect.
MMP13 acts as a collagenase to degrade type 1 collagen, the main component of bone extracellular matrix. Collagenases degrade the osteoid layer covering the surfaces of bone to expose calcified bone matrix to osteoclasts [17][18][19][20]. Significant induction of mmp13 in atp6v1h zebrafish mutant might, therefore, accelerate the degradation of type 1 collagen and provide more mineral contact for osteoclasts, without increasing osteoclast numbers (Fig 4B). Moreover, extracellular matrix has been demonstrated to play a role in osteoblast growth and differentiation [21]. Type I collagen influences osteoblast development and maintenance by modifying the expression of genes supporting cell growth, adhesion and extracellular matrix mineralization [21]. Thus, enhanced degradation of matrix collagen leads not only to increased bone resorption but also to suppression of bone formation [22]. This is consistent with our findings that expression of genes that label intermediate and mature stage osteoblasts is reduced relative to early markers of osteoblast formation (Fig 4).
In zebrafish and mammals, mmp9 is highly expressed in skeletal cells, especially in osteoclasts [9,[23][24][25]. Mmp9 can degrade denatured collagens. Therefore, increased mmp9 expression in atp6v1h mutant zebrafish would be predicted to contribute to ECM degradation. Since the expression level of other osteoclast genes, rank and ctsk, remained largely unchanged, we speculate that increased osteoclast numbers do not play a significant role in reducing bone density in atp6v1h mutants. However, increased maturation and activity of osteoclasts might be more relevant, as demonstrated by the increased differentiation revealed by TRAP staining in mouse ATP6V1H deficient pre-osteoclast cells and adult heterozygous atp6v1h zebrafish.
Apart from matrix degradation, Mmp9 also promotes the apoptosis of hypertrophic chondrocytes in the growth plate [14]. Hence, increased MMP9 activity in atp6v1h mutants may also be associated with accelerated loss of hypertrophic chondrocytes or abnormal apoptosis, contributing to bone loss. Interestingly, Mmp13 and Mmp9 induction are also found in patients with arthritis [26]; these individuals also have abnormal matrix, and could share the same pathogenetic mechanism. Inhibition of MMP9 and MMP13 might be considered as a therapeutic approach for treating these diseases.
Vacuolar ATPases have complicated structures and functions, which may explain why mutations in some subunits cause osteopetrosis while mutations in others cause osteoporosis. V-ATPase is composed of 14 different subunits arranged in two structural domains, V1 and V0. ATP hydrolysis occurs in the cytoplasmic V1 domain whereas proton translocation occurs through the membrane V0 domain [1]. Knockout of the a3 subunit of V0 directly blocks its function in proton translocation, thus causing an acidification defect; Atp6i-deficient mice exhibit severe osteopetrosis [2]. Subunit H of V1 activates ATP-driven proton pumping in the intact V-ATPase complex but, when V1 and V0 are disassociated, the H subunit inhibits the Mg-ATPase activity of the cytosolic V1 domain to prevent unnecessary ATP hydrolysis [27,28]. Isolated V1 complexes without subunit H show significantly higher levels of ATPase activity compared to the wild-type V1 complexes. Therefore, in addition to the acidification defect caused by loss of the H subunit, ATP hydrolysis is increased [29]. In yeast strains lacking subunit H, V1-V0 complexes are inactive and unstable, suggesting that the H subunit serves as an activator of the intact V-ATPase [30].
We propose a model of action in which mutation of the V0 subunit a3 causes reduced acidification, increasing bone density whereas mutation of the V1 subunit H causes excessive ATP hydrolysis, decreasing bone density through increased MMP9/MMP13 activity (Fig 5). Supporting this hypothesis is that knockdown of Atp6i in B16-F10 cells causes decreased expression of Mmp9 [31]. In our model, although acidification by V-ATPase is marginalized when ATP6V1H is mutated, the increase in MMP9 /MMP13 activity related to the constitutive Mg-ATPase activity of the cytosolic V1 sector, dominates, resulting in a net reduction in bone density. Enhanced ATP hydrolysis by cytosolic Mg-ATPase activity might activate signaling pathways to stimulate mmp9 and mmp13 expression, a topic for future research. In any event, the bone defect caused by atp6v1h deficiency through activation of mmp9/13 represents a novel finding that offers potential therapeutic entry points for related diseases, since MMP9 levels have been correlated to severity of osteoporosis [32].

Ethics statement
Patients were enrolled in the NIH Undiagnosed Diseases Program [5][6][7] and in protocol 76-HG-0238, "Diagnosis and Treatment of Patients with Inborn Errors of Metabolism and Other Genetic Disorders", approved by the NHGRI Institutional Review Board. All patients involved in this study gave written, informed consent. All zebrafish experiments were approved by the UCLA Animal Care and Use Committee and the NHGRI ACUC Committee.

Zebrafish
To establish a mutant ATP6V1H zebrafish line, gRNA targeting the fourth exon of ATP6V1H was designed and shown to induce indels by injection with Cas9 mRNA in zebrafish. Several mutant lines were established using gRNA with the sequence 5'-GTGTGTCATCAATCAGGG TCAGG; the line used in this study carried a 17 bp insertion and produced a truncated peptide of 94aa, whereas the wild type ATP6V1H encodes a 463aa protein. Stable heterozygous carriers of the mutation were mated to produce homozygous mutants. Details of mutation generation and characterization are presented in the supplementary section and S3 Fig. Staining for bone and cartilage cells was performed as described [33,34].

Quantitative PCR
Total RNA was extracted in Trizol Reagent (Life Technologies) according to the manufacturer's protocol. cDNA was synthesized using Superscript III (Life Technologies) and oligodT (Life Technologies) following the manufacturer's protocol. Quantitative real-time PCR was performed using FastStart SYBR Green Master (Roche). Primer sequences are listed in S1 Table. Values of three independent samples (n = 50 each sample) are shown.

siRNA knockdown of Atp6v1h in mouse cells
See supplementary materials. The V1H subunit activates ATP-driven proton pumping in the intact V-ATPase complex, but when V1 and V0 are disassociated, inhibition of the Mg-ATPase activity of the V1 domain occurs to prevent unnecessary ATP hydrolysis. When V1H is mutated, significantly higher levels of ATPase activity compared to the wild-type V1 complexes is expected, causing increased matrix degradation, leading to decreased bone density via increased MMP9/MMP13 activity.

Whole-mount in situ hybridization
The probes listed in the figures were generated by PCR using zebrafish embryonic cDNA (Primers in S1 Table). Regular and double fluorescence whole-mount in situ hybridization was, respectively, performed as previously described [35,36].

Micro-computed tomography analysis and TRAP staining in adult zebrafish
For X-ray imaging, adult zebrafish were fixed in 10% neutral-buffered formalin for 24 hours and X-ray images were taken using Faxitron MX-20 System Digital Radiography system (Faxitron Bioptics, Tucson, Arizona). For micro-computed tomography (μ-CT), adult zebrafish were fixed in 10% neutral-buffered formalin for 24hours and mounted in 1% low-melt agarose (Sigma) in a plastic vial. Specimens were scanned using SkyScan1172 Micro-CT scanner (Bruker MicroCT, Kontich, Belgium) with the X-ray power at 55kVp and 185μA, and specimens were scanned at a 10 μm voxel resolution. A three-dimensional reconstruction was generated with NRecon software from the set of scans. The images were then analyzed using CTAn (v.1.14.4) and CTVol (v.2.2) softwares. The first caudal bones were assessed for bone mineral density (BMD), bone volume (BV), and bone surface (BS). For TRAP staining, fixed fish were decalcified and sectioned (5 um) after paraffin embedding. TRAP staining was performed using a kit (Sigma 387A-KT) according to the manufacturer's instructions. 2) and zebrafish (NP_775377.1) shows high homology (~85%). More importantly, the region where the mutation is located is highly conserved (A). Using CRISPR/Cas9, guide RNA (gRNA) targeting exon 4 of zebrafish atp6v1h was co-injected with Cas9 mRNA into zebrafish embryos and detected by T7 endonuclease digestion. Sequence of gRNA and target is shown (B). Several founders with indels were screened for germline transmission, and the one with 17bp insertion was used throughout the study; The 17bp insertion is predicted to produce a termination codon 94 amino acids after initiation of start codon (C). Genotyping done by polyacrylamide gel electrophoresis (PAGE) analysis demonstrating a 168 bp band for wild type allele, and 185 bp band for mutant allele; representative results for wild type (+/+), heterozygous (+/-), and homozygous (-/-) embryos are shown (D). Using probes specific for atp6v1h, RNA in situ hybridization analysis of atp6v1h showed expression in the head region in wild type embryos, while the expression is nearly absent in homozygous atp6v1h embryos, suggesting that the 17bp insertion may lead to nonsense-mediated decay (NMD) of mRNA and result in loss-of-function of this gene (E). Differentiation of Raw264.7 (pre-osteoclast) cell was induced by RANKL (100ng/ml). Raw264.7 was transfected by siRNA anti-ATP6V1H 2 times, and osteoclast cells were identified by TRAP staining (red color). (A) The ATP6V1H siRNA groups have significantly more mature osteoclasts (red staining of large multi-nuclei cells) than the scrambled controls. Included here are sets of five randomly selected fields of images of ATP6V1H siRNA groups and scrambled controls. (B) The ratio of TRAP positive cells and total cells is significantly higher in ATP6V1H siRNA group. Specific data is listed in S2 Table. (TIF) S10 Fig. Analysis of osteoclast by TRAP in adult zebrafish. Adult zebrafish (10 month old) of +/-for atp6v1h (line atp6v1h la439 ) and wild type sibling were fixed, sectioned and stained by TRAP. Atp6v1h la439+/fish appear to have more TRAP staining compared to their sibling (n = 3, two images are shown here for wild type and +/-mutant). Scale bar: 50 um. (TIF) S1 Table. Table lists