ADAM19 cleaves the PTH receptor and associates with brachydactyly type E

Introduction

Metacarpal and metatarsal shortening, coupled with cone-shaped epiphyses and short stature, is termed brachydactyly type E (BDE) (Riccardi & Holmes, 1974). When BDE is inherited, an autosomal-dominant pattern is invariably observed. We identified a cis-regulatory site that downregulates PTHLH, the gene encoding parathyroid hormone-related peptide (PTHrP), as responsible for BDE (Maass et al, 2010). Subsequently, PTHLH gene deletions and point mutations were also shown to be responsible (Klopocki et al, 2010). In other BDE families with a balanced translocation, we identified a misplaced lncRNA disrupting the cis-regulatory landscape causing downregulation of PTHLH (Maass et al, 2012). In still a further balanced-translocation BDE family, we encountered a 2q37 deletion that included HDAC4 (Maass et al, 2018). As a result, differential expression of several genes was affected, including PTHLH, SOX9, and the lncRNA we had identified earlier. The above families all exhibited autosomal-dominant BDE with reduced stature as the sole phenotypes. However, we also described kindreds with BDE accompanied by severe arterial hypertension and stroke along with other vascular phenotypes (Schuster et al, 1996). In these families, we showed that mutated overactive phosphodiesterase 3A is responsible (Maass et al, 2015). PTHLH in these families is dysregulated, causing BDE. PTHrP is recognized as a major, locally expressed paracrine regulator of growth-cartilage chondrocyte proliferation, differentiation, synthetic function, and mineralization (Terkeltaub et al, 1998). Here, we investigated another family with solely the inherited BDE phenotype, as described earlier (Cartwright et al, 1980). The BDE could not be explained by any of the hitherto-fore described genetic mechanisms. Our linkage and sequencing data implicate mutated disintegrin and metalloproteinase-19 (ADAM19). The membrane-anchored ADAM enzymes are involved in numerous biological processes involving cell-cell and cell-matrix interactions (Mochizuki & Okada, 2007).

Earlier, our group had shown that the parathyroid hormone (PTH) receptor 1 (PTHR1) can be cleaved by an unidentified metalloprotease (Klenk et al, 2010a). PTHR1 also serves as the receptor for PTHrP (Suva & Friedman, 2020). PTHR1 is a family B, G protein-coupled receptor (GPCR) that regulates skeletal development, bone turnover, and mineral ion homeostasis (Cheloha et al, 2015). Our present work addresses the relationship between ADAM19 and PTHR1. We believe our findings are also relevant to the BDE phenotype and putative PTHrP signaling. Since PTHrP is intimately involved in how cancer metastasizes to bone, we suggest that the relevance of our findings extend far beyond a relatively uncommon genetic syndrome (Ponzetti & Rucci, 2020).

Results

We revisited the BDE family described by Cartwright et al earlier (Cartwright et al, 1980). We examined 10 members, six of whom had the BDE phenotype (Fig 1A). A hand roentgenogram from an affected person shows shortened metacarpals and cone-shaped epiphyses (Fig 1B). In our clinical assessment, we identified no further phenotypes that were not shown earlier; we found no dental phenotypes, and none were previously mentioned. We first performed a linkage analysis with a 900 K single nucleotide polymorphism (SNP) array and detected genomic regions on seven different chromosomes with LOD scores above 1.5. We performed fine mapping with microsatellite markers and identified promising loci (Fig S1A). Indel length and substitution length are given (Fig S1B and C). We then relied on whole-genome sequencing of two affected and one non-affected person from our kindred (Fig S1B). Of the four candidate loci only one, on chromosome 5, contained a gene, ADAM19, exhibiting the same mutation in both affected subjects (Fig 1C). This mutation was then identified in all BDE subjects and in none of the non-affected persons (Fig S2). The mutation features a deletion resulting in an amino acid exchange at 117, followed by a substitution-insertion and a second insertion. As a result, amino acids 117–135 are faulty, compared to the WT amino acid sequence. Finally, a TAG stop codon resides after amino acid 135. Multiple ADAM19 peptide-sequence alignments from five different species near the identified mutation indicate that the region is highly conserved (Fig 1D). We believe that in the PDF, the image is not great. The electropherograms are barely discernible and the text in the image is not readable. We believe that the image we sent should be good enough to make this better.

The ADAM19 coding sequence contains 918 amino acids (Fig 2A). The gene encodes for a signal peptide, a pro-peptide domain, a metalloproteinase domain, a disintegrin domain, a cysteine-rich domain, an EGF-like domain, a transmembrane domain and a cytoplasmic domain. The mutated protein was predicted to lack the transmembrane-to-cytoplasmic domain region. We next performed immunoprecipitation experiments (Fig 2B). Full-length WT ADAM19-FLAG-tag and mutated ADAM19-FLAG-tag were transiently transfected in HEK 293 cells. A FLAG-tag was fused in frame to the cytoplasmic domain of WT ADAM19 and specific antibodies directed against the pro-domain, the metalloproteinase domain, and the FLAG-tag were employed. The data showed detection of full-length WT ADAM19 and supported premature termination of mutated ADAM19. Mutation of ADAM19 leads to a frame shift resulting in a scrambled sequence after His116 and a new stop codon after amino acid 135. We performed Western blot experiments with antibodies against residues 50–122 of ADAM19 and with antibodies against the Flag epitope, which had been fused to the native C-terminus of ADAM19. The results suggested that neither the N-terminal fragment, nor full-length ADAM19 (e.g., as a result from a read-through event of the new stop codon), were formed and thus defective ADAM19 possibly has been degraded in the cell. However, we cannot fully exclude the possibility that the absence of ADAM19 at protein level may also be due to a pre-translational event.

We also studied 12.5-d-old and 13.5-d-old mouse embryos. These embryos expressed Pthlh as shown earlier (Maass et al, 2015) in the paw developing cartilage and also expressed Adam19 (Figs 2C and S3). We performed immunofluorescence in HEK 239 cells transfected with WT and mutated ADAM19-FLAG-tag. We found that ADAM19 was strongly expressed on the plasma membrane together with PTHR1, while mutated ADAM19, was not detected (Figs 2D and S4). In addition, we assessed ADAM19 mRNA expression in various fetal tissues. We found that ADAM19 is particularly strongly expressed on chondrocytes and placenta. The findings are compared to cardiac expression as reference tissue (Fig S5).

We next studied whether ADAM19 is a PTHR1 cleavage enzyme. PTHR1 is a GPCR with seven transmembrane domains that we outfitted for Western blot analysis with a FLAG-tag at the C terminus and for fluorescence detection with an EYFP tag at the N terminus. We transfected HEK 293 cells with ADAM19-HA-tag and PTHR1-FLAG-tag. PTHR1 was detected using an antibody to the FLAG-tag, the amino acid epitope 52–86, and the amino acid epitope 4–54. ADAM19 was detected using an antibody to the HA-tag. The findings showed that WT ADAM19 cleaved PTHR1 (Figs S6 and 3A). To more precisely discriminate receptor species, proteins were deglycosylated by PNGase F prior to analysis. When co-expressed with catalytically inactive ADAM19 mutant (ADAM19-E384A), PTHR1 mainly appears as a band at ∼60 kD, which is consistent with the molecular weight of the full-length receptor protein. In the presence of WT ADAM19, two additional bands at ∼50 kD became apparent with increasing amounts of protease, while the 60 kD band was reduced. These findings suggest that a 10 kD fragment was lost at the receptor’s N terminus in the presence of catalytically active ADAM19. This result would be in line with our previous findings that PTHR1 is cleaved by metallo-proteases within loop1 of the extracellular domain (ECD) (Klenk et al, 2010a, 2022). In contrast, when cells were cultivated in presence of 10 μM batimastat, a potent metalloprotease inhibitor, only the 60 kD species of PTHR1 were detected. Notably, here we used unmodified receptor which clearly demonstrates the proteolytic activity of ADAM19 on PTHR1 (Fig 3B).

Sheddases are membrane-bound enzymes that cleave extracellular portions of transmembrane proteins, releasing the soluble ectodomains from the cell surface. To independently confirm our findings and to test whether ADAM19 acts as a sheddase for PTHR1 on living cells, we next used an experimental approach based on fluorescence. To do so, HEK-293T cells were transiently co-transfected with EYFP-PTHR1 together with WT ADAM19, ADAM19-MT, or with CD4 as a control. By using a control that inherently does not exhibit shedding activity, we aimed to delineate the absence of shedding, thereby validating the specificity of our assay for detecting protease-mediated shedding of PTH1R, with particular focus on ADAM19. Shedding was measured by comparing the ratio of EYFP fluorescence in the cell supernatants to that of EYFP at the cell. When the EYFP-labeled PTHR1 was co-expressed with ADAM19, the majority of the fluorescence was found in the supernatant, indicating cleavage and shedding of the N-terminus of the receptor. In contrast, ADAM19-MT did not increase soluble YFP when compared to the CD4 control (Fig 4A). Notably, N-terminal receptor cleavage was dependent on the ratio between ADAM19 and PTHR1 (Fig 4B). We also examined the effect of a metalloprotease inhibitor on the cleavage behavior of ADAM19. We next incubated 100 μM Ilomastat (GM6001) in the medium for 48 h. We observed that the PTHR1 cleavage by ADAM19 was inhibited by the addition of Ilomastat (Fig 4C). A bar graph representation of the total fluorescent sum of supernatant and cell values is shown for the fluorescence assay, different expression ratio of ADAM19 and cleaved EYFP-PTHR1 ratio, and the influence of a metalloprotease inhibitor (Fig S7A–C respectively). Collectively, these data show the PTHR1 cleavage-dependency of ADAM19, and we conclude that the enzyme is a sheddase for PTHR1. We believe that the figure 4 is too large when compared to the other figures.

From the Western blot experiments, we observed that a cleavage of the PTHR1 receptor is carried out by ADAM19 and that the site is located after the amino acid position 54 (Fig 3A), which is in line with our previous findings that PTHR1 can be cleaved by metalloproteases at position Ser 61. To precisely map the cleavage site of ADAM19, we introduced various alanine substitutions at the N-terminus of EYFP-tagged receptor and measured N-terminal shedding by release of EYFP from the cell surface. To complete this task, EYFP-PTHR1 was manipulated between the amino acids 48 through 79 (referring to the native PTHR1 sequence) in segments of eight amino acid positions as shown (Fig 5A). We observed a significant decrease in soluble EYFP fluorescence between PTHR1 residues 56–71, which was most prominent for mutant 56–63A suggesting the ADAM19 cleavage site in this region of PTHR1 (Fig 5B). PTHR1 localization and internalization (trafficking) were monitored in HEK 293 cells by confocal microscopy imaging. We next determined whether or not the alanine substitutions made the receptor behave differently in terms of internalization compared to the unmodified WT receptor. Cellmask was used as the membrane marker. We observed that the alanine substitutions had no visible influence on the internalization of the receptor. All manipulated amino acids between 48 and 79 revealed the same membrane localization as the WT receptor (Fig S8). To solidify these findings, we then performed mass spectrometry and could identify both c-terminal and n-terminal peptides specific for either the n-terminal or the c-terminal cleavage product of PTHR1. Based on these identifications, we can pinpoint the cleavage site to reside between amino acid position 64 and 65 (Fig 5C). This figure also seems too small.

To assess the functional implications of PTHR1 cleavage on downstream signaling, we first aimed to compare G-protein activation by PTHR1 versus the partially cleavage-protected PTHR1-56-63A mutant in the presence of ADAM19 in HEK-293T cells (Fig 6A). For this purpose, we used bioluminescence resonance energy transfer (BRET)-sensors measuring dissociation of Gα and Gβγ subunits upon receptor-mediated G-protein activation (Schihada et al, 2021). In line with our previous findings (Klenk et al, 2022), after stimulation with PTH we observed increased G_s-activation and decreased G_q-activation for WT PTHR1 when compared to the cleavage-protected PTHR1 mutant (Fig 6B and C). After stimulation with PTHrP, likewise G_q-activation was decreased for WT PTHR1 compared to the PTHR1-56-63A mutant in the presence of ADAM19. However, in contrast to PTH, no differences in G_s-activation between cleaved and uncleaved PTHR1 were observed after PTHrP stimulation (Fig 6D and E). To corroborate these findings in a more physiological cell model, we used human osteosarcoma U2OS cells in which we measured PTHR1-mediated cAMP accumulation with the fluorescence resonance energy transfer (FRET)-based cAMP biosensor Epac-S-H187 (Klarenbeek et al, 2015), in the absence or presence of ADAM19. The U2OS is a cell line with epithelial morphology derived from an osteosarcoma. We reasoned that the U2OS cell line would exhibit a molecular background of relevant interacting proteins closely resembling those of chondrocytes, making it the most suitable model cell system available for our investigation. We recognize associated limitations of this cell line.

FRET measurements were performed on a single-cell basis using epifluorescence microscopy, and normalized FRET ratios were calculated as CFP/FRET (%). An increase in delta FRET, therefore, signifies an increase in cAMP throughout the cell. In response to PTH1R stimulation with both PTH1-34 and PTHrP, an increase in cAMP was detected in cells transfected with PTH1R and Epac-S-H187 alone, as well as in cells which were co-transfected with Adam19. FRET ratios were increased by 70–80% on average in both cases, visualized by representative traces (Fig 7C–F). Taken together in the responses measured in all experiments (Fig 7A and B), there were no significant differences between cells transfected with PTH1R and Epac-S-H187 and those co-transfected with ADAM19. Comprehensive control experiments indicated that YFP direct excitation values with PTHR1, ADAM19, the sensor H187 were constant across a wide range of YFP excitation values (Fig S9A–D).

We then focused on PTHR1-mediated β-arrestin2 recruitment in human osteosarcoma U2OS and SaOS cells. Interestingly, in both cell lines PTH significantly increased β-arrestin translocation to the PTHR1 when ADAM19 was present (Fig 8A), similar to the results obtained for G_s-activation and cAMP accumulation. In contrast, after PTHrP stimulation only subtle differences in β-arrestin2 recruitment to PTHR1 in absence or presence of ADAM19 were observed (Fig 8B). Control expressions for cleavage and β-arrestin2 recruitment experiments (Fig 5B) are presented with total fluorescence (Fig S10A). The control Saos2 cell experiments (Fig S10B) with control luminescence and fluorescence data are shown (Fig S10C and D). Nonetheless, β-arrestin2 recruitment to PTHR1 in U2OS cells in absence or presence of ADAM19 was different (Fig 8). β-arrestin recruitment was augmented in the presence of ADAM19 for PTH, while with PTHrP, β-arrestin recruitment was slightly decreased. Control expressions for cleavage and β-arrestin2 recruitment experiments for PTHrP and PTH are shown (Fig S11A and B, as well as Fig S11C and D respectively). Together, these findings suggest that PTHR1 cleavage through ADAM19 modulates differential effects on PTH-mediated receptor signaling for G_s, G_q and β-arrestin, whereas in the case of PTHrP mainly the G_q pathway is affected. Possibly, differential effects of PTH1R cleavage on PTH and PTHrP signaling could be related mechanistically. We envision altered interaction of the C-termini of PTH and PTHrP with the ECD of PTH1R (Fig S12A and B). Differences in amino acid sequences in relevant PTH and PTHrP regions could result in altered interactions within the tethered helix1 for PTH and PTHrP as shown (Fig S12C).

Finally, to bring our findings to a phenotypically more definitive (albeit speculative and preliminary) conclusion, we performed mass-spectrometry proteomics in U2OS cells. A volcano plot and heat map are presented (Fig S13A and B). We conducted a comprehensive proteomic analysis encompassing endogenous cells, cells expressing ADAM19, cells over-expressing PTHR1, and cells expressing both PTHR1 and ADAM19 concurrently. Our primary focus was to discern the distinctions between cells expressing PTHR1 (representing the disease cell model) and cells expressing both PTHR1 and ADAM19 (representing the WT model). Through application of a t test, we identified several upregulated proteins in both sets of cells. Noteworthy attention was given to genes upregulated in PTHR1-expressing cells (the disease cell model) that might hold phenotypic significance. For instance, the sugar phosphate exchanger protein 3 (SLC37A3), collagen Type I Alpha 1 Chain (COL1A1), collagen Type II Alpha 1 Chain (COL2A1), osteoclastogenesis-associated transmembrane protein 1 (OSTM1), G protein subunit gamma 7 (GNG7), and leucine rich repeat containing G protein-coupled receptor 6 (LGR6) displayed differential regulation.

In an effort to ascertain whether or not these observed protein upregulations were attributed to the overexpression of either ADAM19 or PTHR1, we compared the protein intensities across all cell lines. Our findings indicated that the noted proteins that are particularly intriguing, exhibited diminished abundance in cells solely expressing ADAM19. This observation suggests that the effect is more likely attributed to the presence of PTHR1 rather than to ADAM19.

Discussion

We showed that ADAM19 is a metalloproteinase that cleaves the N-terminus of the PTH receptor. Earlier work from our group demonstrated that the ECD of PTHR1 is subject to metalloproteinase cleavage (Klenk et al, 2010a). Recently, our group localized the metalloproteinase cleavage site to the first loop of the PTHR1 ECD (Klenk et al, 2022). The cleaved receptor exhibited enhanced signaling to G_s and decreased activation of G_q compared to WT PTHR1. The findings reported here verified ADAM19 as a responsible metalloprotease capable of such cleavage. We identified the ADAM19 locus by means of genetic linkage in a small pedigree, solely exhibiting autosomal-dominant BDE. Genome sequencing subsequently revealed a truncated ADAM19 allele that was identified in all affected family members and in none of nonaffected persons. Since our earlier studies had shown that PTHrP is responsible for BDE, we tested the hypothesis that ADAM19 interacts with the PTHrP-PTHR1 signaling pathway. We found that ADAM19 is capable of PTHR1 cleavage and showed that the truncated ADAM19 was unable to cleave PTHR1. We used alanine substitutions to map the N-terminal cleavage site of PTHR1. Our mass-spectrometry data on the cleavage site provide convincing evidence that ADAM19 is indeed a sheddase for PTHR1. Finally, we examined the effects of PTHR1 cleavage on downstream signaling and compared G-protein activation by the cleaved versus the noncleaved PTHR1. We found that protection from the ADAM19 cleavage of its N-terminal ECD shifts signaling of the PTHR1 between the major downstream pathways. Noncleaved PTH1R exhibited increased activation of G_q and decreased activation of G_s in response to PTH stimulation when compared to cleaved PTH1R. Likewise, PTH-mediated β-arrestin recruitment to PTH1R was reduced in the absence of ADAM19. In contrast, cleavage-mediated effects on receptor signaling were more directed after stimulation with PTHrP. Cleavage protection of PTH1R resulted in increased G_q signaling whereas only subtle changes in G_s activation and β-arrestin recruitment were found. Although PTH and PTHrP are both full agonists for PTH1R, their binding kinetics and their selectivity for distinct receptor conformations differ (Hattersley et al, 2016). While the N-termini of both peptides have high sequence similarity and largely make comparable molecular contacts to the transmembrane domain of PTH1R, the C-terminal regions of the peptide ligands are less conserved resulting in marked differences in their interaction with the ECD (Pioszak et al, 2009; Ehrenmann et al, 2018; Kobayashi et al, 2022) (Fig S12). Importantly, several of these non-conserved residues make contacts with the N-terminal α1 helix of the ECD and have been shown to strongly influence the differences in binding kinetics and signaling properties between both ligands (Mannstadt et al, 1998; Mann et al, 2008; Pioszak et al, 2009). Proteolytic cleavage of the ECD at position 64 would likely increase the flexibility of the N-terminal α1 helix, which is located proximal to the cleavage site, but remains tethered to the distal part of the ECD through a disulfide bond between C64 and C117 (Klenk et al, 2010a) (Fig S12). Given the ligand-specific contacts at the proximal end of α1 helix, it is well conceivable that increased flexibility of the binding pocket after receptor cleavage may result in the observed ligand-dependent changes in receptor signaling. This hypothesis is further corroborated by the notion that in other congenital growth disorders, underlying mutations in the ECD of PTH1R are found. Homozygous mutation of E35, which is located at the N-terminal end of α1 helix and which makes contacts to the variant residues 16 and 19 in PTH/PTHrP, is found in the rare Eiken syndrome characterized by delayed bone mineralization and epiphyseal dysplasia (Moirangthem et al, 2018). Notably, this mutation leads to distinct changes in signaling response to PTH and PTHrP similar to what has been observed in the present study (Moirangthem et al, 2018). Altered Signaling and Desensitization Responses in PTH1R Mutants Associated with Eiken Syndrome have been described (Portales-Castillo et al, 2023). Together, these findings underscore the importance of extracellular regions of GPCR for fine-tuning and adjusting receptor activity and signaling selectivity.

Our study potentially adds human phenotypes to metalloprotease cleavage of PTHR1. In earlier studies, we identified PTHrP as responsible for BDE (Maass et al, 2010, 2012, 2015, 2018). Mutated PTHLH encoding PTHrP has also been shown to produce the BDE phenotype (Klopocki et al, 2010). Since PTHrP signals by means of PTHR1, we were interested in what interactions ADAM19 might have with bone and cartilage development (Suva & Friedman, 2020). Chesneau and colleagues investigated the catalytic properties of ADAM19 (Chesneau et al, 2003). They found that the enzyme splits TNF-α cleavage sites, osteoprotegerin ligand, and Kit ligand. The authors identified a role for ADAM19 in regulation of the shedding process. The ADAM19 targets were generously expressed in osteoblasts. Amongst other metalloproteinases, ADAM19 participated in cartilage development. A role for the enzyme was identified in studies of microenvironmental changes and mesenchymal stem-cell differentiation towards chondrocytes (Djouad et al, 2007).

In higher vertebrates, cephalic neural crest cells form the craniofacial skeleton by differentiating into chondrocytes and osteoblasts. Arai et al recently identified a ADAM19 as a position-specific fate regulator of neural crest cells (Arai et al, 2019). ADAM19 mediated the cleavage of bone morphogenic protein (BMP) receptor type I (Alk2). The group also presented evidence that the sex determining region Y-box 9 (Sox9) cascade was suppressed presumably related to altered Alk2 processing. ADAM19 has also recently been identified as a microRNA target. Kong et al found that synovial mesenchymal stem cell-derived miR-320c targeted ADAM19 directly, thereby enhancing chondrogenesis (Kong et al, 2022).

In BDE, metacarpals and metatarsals are selectively shortened. Chondrocyte hypertrophy is accompanied by increased organelle synthesis and rapid intracellular water uptake, which serve as the major drivers of longitudinal bone growth. This process is delicately regulated by major signaling pathways and their target genes, including growth hormone, insulin growth factor-1, Indian hedgehog (IHH), PTHrP, BMPs, Sox9, runt-related transcription factors (Runx) and fibroblast growth factor receptors (Hallett et al, 2021).

PTHrP was identified as causing the hypercalcemic paraneoplastic syndrome. Its receptor and the encoding gene were subsequently identified (Clemens et al, 2001). PTHLH is expressed in the distal ends of cartilage anlagen and in the perichondrium of developing bones. PTHrP diffuses away from its site of production and binds to PTHR1 located on prehypertrophic chondrocytes. PTHrP binding results in PTHR1 and later Gα_s activation, which in turn keeps chondrocytes proliferating through suppression of the cyclin-cdk inhibitor p57, thereby increasing the pool of proliferating nondifferentiated chondrocytes (Kronenberg, 2006). PTHrP synthesis is controlled by IHH (St-Jacques et al, 1999). The pathologic, molecular, and clinical correlates, including signaling involved in disorders of G protein-mediated cAMP activation have been reviewed (Cohen, 2006).

Presumably, heterozygosity for ADAM19 in our affected BDE patients reduced their ADAM19 activity and promoted a relative increased activation of G_q and decreased activation of G_s, perhaps at a critical stage in cartilage development. PTHR1 mutations resulting in receptor loss of function have been associated with endochondral developmental defects earlier (Couvineau et al, 2008). The receptor’s signaling responses to PTH and PTHrP is currently being elucidated with high resolution near-atomic structural techniques (Sutkeviciute et al, 2019). Such efforts could contribute to further mechanistic explanations of our BDE syndrome. We investigated plasma membrane-related G_s and G_q effects; however, endosomal cAMP responses possibly also play a role. For instance, PTHR1-related G_s and G_q/11 activation has been shown to mediate endosomal cAMP responses (Ferrandon et al, 2009; Jean-Alphonse et al, 2017; White et al, 2020).

The observed differences in PTH and PTHrP effects on PTH1R-related G_s activation could potentially stem from variations in the execution-time points of the assays (different days and different passage of the cells). Our observations suggest that the dynamic range of the G_s BRET assay exhibits greater variability compared to the G_q BRET assay that was more stable and had lower day-to-day variability. Alternatively, the observed differences could arise from the specific timepoint chosen for generating the concentration-response curve, as PTHrP-mediated G_s activation reaches saturation at a distinct point compared to PTH-mediated activation.

The FRET experiments, which measure cAMP accumulation, inherently evaluate processes further downstream in the signaling cascade, introducing the possibility of other factors masking the observed effects closer to the receptor, such as enhanced β-arrestin interactions (Fig 8). Additionally, the transient transfection of ADAM19 and thus incomplete cleavage (Fig 5B), different cell line used (HEK293 versus U2OS), and different heterologous expression protocol (transfection versus transduction) could potentially obscure downstream effects of enhanced G_s signaling. With these caveats in mind, we nonetheless suggest that our genetic findings support the idea that mutated ADAM19 is associated with our clinical phenotypes.

Together, IHH and PTHrP form a feedback loop regulating chondrocyte hypertrophic differentiation and thereby endochondral bone development (Vortkamp et al, 1996). The effects of PTHrP on chondrocyte differentiation are also mediated by means of the Sox9 phosphorylation, a transcription factor important for chondrocyte differentiation, and by suppressing Runx2. We identified Sox9 participation in BDE earlier (Maass et al, 2018). Runx2 is a transcription factor essential for osteoblast differentiation. Independent of our findings here, ADAM19 is implicated in bone development, BMP, and Sox9 signaling.

Our findings that ADAM19 cleaves PTHR1 underscore the relevance of these observations. The BDE syndromes we encountered earlier and the family we describe here had no abnormalities in calcium or phosphorus homeostasis. Furthermore, their PTH levels, when measured were not perturbed. We have insufficient data on PTHrP concentrations in our patients; however, PTHrP exerts paracrine functions and is not a circulating hormone aside from overproduction by tumors in paraneoplastic syndromes.

We found an effect of ADAM19 cleavage on β-arrestin recruitment. The finding was driven by a report on PTH1R-modification by such enzymes (Klenk et al, 2010a). GPCRs activate heterotrimeric G proteins. In order to turn off a response, or to adapt to a persistent stimulus, active receptors must be desensitized. The first step in desensitization is phosphorylation of the receptor by G protein-coupled receptor kinase. Arrestin binding to the receptor blocks further G protein-mediated signaling and targets receptors for internalization, and redirects signaling to alternative G protein-independent pathways, such as β-arrestin signaling. The importance of arrestins in general and in PTHR1-related signaling in particular is well established (Lohse et al, 1990; Gesty-Palmer et al, 2006; Ferrari & Bouxsein, 2009; Klenk et al, 2010b). β-arrestin is also important in modulating opposing effects of PTHR1 in calcium-sensing receptor signaling (Gorvin et al, 2018). Moreover, receptor activity-modifying protein 2 was shown to modulate PTHR1-stimulated amounts of ß-arrestin2 recruitment and downstream signaling (Nemec et al, 2022). GPCR signaling by PTHrP and fragments has been investigated, including effects on β-arrestin2 (Pena et al, 2022).

The lack of ADAM19 cleavage effects on cAMP responses to PTH in U2OS cells warrants explanations. Increased β-arrestin coupling could counteract the observed rise in G_s activation and mask increased cAMP accumulation. Additionally, the choice of cell type (HEK293 versus U2OS) and protocols for heterologous expression could impact the outcomes, as the stoichiometry and expression levels of interacting proteins can vary. The cAMP accumulation was evaluated in U2OS cells, where we opted for transduction over transfection to enhance assay performance.

Recently, Chu and colleagues reported that matrix metalloprotein-14 (MMP14) cleaves PTH1R in the chondrocyte-derived osteoblast lineage, curbing signaling intensity for proper bone anabolism (Chu et al, 2023). Their study identifies a novel paradigm of MMP14 activity-mediated modulation of PTH signaling in the osteoblast lineage, contributing new insights into bone metabolism. Their findings suggested that MMP14 dampens PTH signaling. Our investigation was rather focused on PTHrP. Nonetheless, the findings underscore the role of regulatory enzymes in PTHR1 signaling. Conceivably, other such enzymes influence PTHR1 activity.

Modulation of PTH1R signaling by a large ECD binding antibody results in inhibition of β-arrestin 2 coupling (Sarkar et al, 2019). Others (Chu et al, 2023) and we (Klenk et al, 2010a) have shown earlier that PTH1R-ECD can also be subject to cleavage by proteases other than ADAM19. Indeed, we observe a minor fraction of processed PTH1R that may be cleaved by other metalloproteases endogenously expressed in HEK293 cells, the cell line that was employed for the experiments.

The “warts, hypogammaglobulinemia, infections, and myelokathexis” syndrome is an inherited immune disorder caused by an autosomal dominant heterozygous mutation in CXCR4. Kumar et al (2023) recently found reduced GPCR signaling despite impaired internalization and β-arrestin recruitment in patients carrying a CXCR4Leu317fsX3 mutation causing warts, hypogammaglobulinemia, infections, and myelokathexis syndrome. Our data show that in the presence of ADAM19, lesser amount of β-arrestin2 are recruited with PTHrP, compared to the effects of PTH. We suggest that differential β-arrestin2-related effects in response to ligands could have contributed to our findings. LGR6 is a Wnt-associated adult stem-cell marker and is required for osteogenesis and fracture healing (Doherty et al, 2023).

PTHrP signaling could also be of relevance above-and-beyond development. PTHrP is abundant in osteoarthritic cartilage and a PTHrP isoform has been implicated in faulty mineralization of osteoarthritic cartilage (Terkeltaub et al, 1998). PTHrP signaling has also been identified in tumor metastases (Ponzetti & Rucci, 2020). PTHrP expression in breast cancer is enriched in bone metastases compared to primary tumors. The protein exerts a cytokine-like function in tumor microdomains, where it serves a paracrine function to promote breast-cancer metastases (Johnson et al, 2018). Independent of PTHrP, ADAM19 is also implicated in the oncogenic properties of breast cancer. The oncohistone, H2BE76K upregulated ADAM19, which in turn participated in enhanced tumor colony formation that could be inhibited by knock-down (Kang et al, 2021).

Finally, we attempted to identify possible molecular candidate genes that could mechanistically contribute to our findings. SLC37A3 is a transporter complex responsible for the cytosolic entry of nitrogen-containing bisphosphonates, important to bone metabolism (Yu et al, 2018). The collagen-encoding genes are obvious fits. The osteopetrosis-associated transmembrane protein 1 (OSTM1) causes craniofacial and dental abnormalities (Ma et al, 2023). These genes could serve as guidance parameters for future studies.

Our data have limitations. The evidence that our ADAM19 mutation causes BDE is not circumstantial; we can offer no direct proof. Zhou et al generated ADAM19 gene-deleted mice (Zhou et al, 2004). Homozygous mice exhibited severe cardiac defects. Oddly, although the mice were smaller, ADAM19 did not appear essential for bone growth and skeletal defects. ADAM19 was expressed in proliferating chondrocytes distal to IHH expression. Whole mounts of paws exhibited similar bone development of ADAM19−/−, ADAM19−/+, and ADAM19+/+ mice. Our patients are heterozygous and exhibited no cardiac defects.

Nevertheless, “mouse is not man;” for instance, the impressive effect of ADAM−/− on cardiac defects has not translated into documenting the effects of ADAM19 on human congenital heart disease. Furthermore, the facts that ADAM19 is expressed adjacent to IHH in the paws of mice during development and that ADAM19 cleaves the signaling receptor of PTHrP, the known driver of BDE, is compelling. We suggest that ADAM19 is a sheddase for PTHR1 and is implicated in regulating the diverse functions of this important GPCR.

Materials and Methods