Proteolysis of fibrillin-2 microfibrils is essential for normal skeletal development

The embryonic extracellular matrix (ECM) undergoes transition to mature ECM as development progresses, yet few mechanisms ensuring ECM proteostasis during this period are known. Fibrillin microfibrils are macromolecular ECM complexes serving structural and regulatory roles. In mice, Fbn1 and Fbn2, encoding the major microfibrillar components, are strongly expressed during embryogenesis, but fibrillin-1 is the major component observed in adult tissue microfibrils. Here, analysis of Adamts6 and Adamts10 mutant mouse embryos, lacking these homologous secreted metalloproteases individually and in combination, along with in vitro analysis of microfibrils, measurement of ADAMTS6-fibrillin affinities and N-terminomics discovery of ADAMTS6-cleaved sites, identifies a proteostatic mechanism contributing to postnatal fibrillin-2 reduction and fibrillin-1 dominance. The lack of ADAMTS6, alone and in combination with ADAMTS10 led to excess fibrillin-2 in perichondrium, with impaired skeletal development defined by a drastic reduction of aggrecan and cartilage link protein, impaired BMP signaling in cartilage, and increased GDF5 sequestration in fibrillin-2-rich tissue. Although ADAMTS6 cleaves fibrillin-1 and fibrillin-2 as well as fibronectin, which provides the initial scaffold for microfibril assembly, primacy of the protease-substrate relationship between ADAMTS6 and fibrillin-2 was unequivocally established by reversal of the defects in Adamts6-/- embryos by genetic reduction of Fbn2, but not Fbn1.


Introduction
In addition to proliferation and differentiation of resident cells, proper tissue and organ development, structure and function require an appropriate extracellular matrix (ECM). How ECM architecture and stoichiometry are maintained and adjusted in coordination with the dynamic tissue changes during morphogenesis and transition to the adult organism is unknown. Ontogenetically, the earliest cell collectives formed sheets and tubes with a well-established basement membrane that provided a substrate for cell migration and adhesive inputs that determined cell polarity. Expansion of ECMencoding genes enabled formation of a complex interstitial matrix that promoted evolution of ever more complex organisms (Huxley-Jones et al., 2007b), but presented a challenge for remodeling of the increasingly diverse ECM repertoire within complex ECM assemblies, which appears to have been met by expansion of secreted and cell-surface protease genes (Huxley-Jones et al., 2005;Huxley-Jones et al., 2007a). For example, there are 19 ADAMTS proteases present in mammals compared to only 2 in the fruitfly and the majority cleave ECM molecules (Dubail and Apte, 2015;Mead and Apte, 2018a). However, which specific ECM molecules and supramolecular assemblies are targets of the individual proteases, the specific contexts they operate in, and relationships between individual proteases are poorly understood. The embryonic interstitial ECM is highly hydrated owing to abundant macromolecular hyaluronan (HA)-proteoglycan aggregates whereas fibrillar components, primarily collagens and elastin, dominate juvenile and adult ECM composition and provide structural resilience needed for the greater mechanical demands imposed by postnatal life. Beyond structural roles, ECM sequesters and regulates the activity of morphogens and growth factors (Ramirez et al., 2018;Thomson et al., 2019), and ECM proteolysis can generate bioactive fragments, termed matrikines (Ricard-Blum and Vallet, 2019).
Fibrillin microfibrils have a crucial role in tissue development and homeostasis by providing mechanical stability and limited elasticity to tissues and/or regulating growth factors of the TGFβ superfamily, including bone morphogenetic proteins (BMPs) and growth/differentiation factors (GDFs) (Ramirez et al., 2018;Thomson et al., 2019), along with a key role in elastic fiber assembly (Shin and Yanagisawa, 2019;Kozel and Mecham, 2019). Fibrillins are large, cysteine-rich glycoproteins containing many epidermal growth factor (EGF)-like repeats. Of the three fibrillin isoforms, fibrillin-2 and fibrillin-3 (in humans) are primarily expressed during embryogenesis (Sabatier et al., 2011;Zhang et al., 1994;Zhang et al., 1995). The gene encoding fibrillin-3 is inactivated in mice (Corson et al., 2004), providing a simpler scenario than in humans for investigating developmental regulation of microfibril composition and the role of proteolytic turnover therein. Among numerous gene mutations affecting the skeleton (Yip et al., 2019), FBN1 and FBN2 mutations cause Marfan syndrome and congenital contractural arachnodactyly, respectively, in humans (Robinson et al., 2006). Despite overlapping features such as skeletal overgrowth and poor muscular development, each disorder mainly has distinct manifestations, indicating that fibrillin isoforms either contribute specific mechanical or regulatory properties to microfibrils, have a tissue-specific function, or form distinct ECM networks. Severe cardiovascular manifestations, especially aortic root and ascending aorta aneurysms, which are potentially lethal, as well as ocular manifestations, occur frequently in Marfan syndrome, but neither is typically associated with FBN2 mutations (Robinson et al., 2006). In mice, Fbn2 deficiency affects myogenesis and distal limb patterning, indicating a key role for fibrillin-2 in BMP/GDF regulation (Arteaga-Solis et al., 2001;Sengle et al., 2015).
Fibrillin microfibrils can be homotypic or heterotypic (Charbonneau et al., 2003;Marson et al., 2005;Lin et al., 2002), but since each fibrillin appears to have distinct roles in vivo as well as in vitro (Nistala et al., 2010), intriguing questions are how the correct stoichiometry of the two fibrillins is maintained, and the impact of excess fibrillin microfibrils or altered fibrillin stoichiometry on biological systems. Here, analysis of mouse mutants of the homologous secreted metalloproteases ADAMTS6 and ADAMTS10 has identified a transcriptionally adapted system for proteostasis of fibrillin microfibrils, defined the impact of ADAMTS6 and ADAMTS10 deficiency on skeletal development, uncovered and demonstrated the importance of a specific protease-substrate relationship between these proteases and fibrillin-2 relevant to skeletal development and unearthed an impact on limb growth factor sequestration.
Since ADAMTS proteases cooperate in several physiological contexts where they are co-expressed (Dubail et al., 2014;Enomoto et al., 2010;McCulloch et al., 2009;Mead et al., 2018b;Nandadasa et al., 2019;Nandadasa et al., 2015) and their pairwise homology suggests the possibility of transcriptional adaptation (El-Brolosy et al., 2019;Sztal and Stainier, 2020), which has not been systematically investigated in this family, we investigated the impact of single or combined inactivation of Adamts6 and Adamts10 on mouse skeletal development, and defined the underlying mechanisms. The findings provide intriguing new insights into fibrillin microfibril proteostasis and mechanisms regulating skeletal development.

Transcriptional adaptation of Adamts6 and Adamts10
Previous work had suggested that knockdown of ADAMTS10 in cultured human ARPE-19A cells increased ADAMTS6 expression (Cain et al., 2016). To investigate the possibility that germline mutations of either mouse gene affected expression of the other, Adamts6 and Adamts10 mRNA levels were measured in limbs, heart and lungs of Adamts6 -/and Adamts10 -/mice. qRT-PCR analysis showed that Adamts6 mRNA was reduced in limb and lung with respect to wild type levels in Adamts6-mutant mice [which have a missense Ser 149 Arg mutation that abrogates protein secretion, essentially comprising a knockout (Prins et al., 2018)], but was consistently increased in Adamts10 -/tissues ( Figure 1A). Adamts10 mRNA was undetectable in Adamts10 -/mice (which have an intragenic IRES-lacZ insertion that disrupts mRNA continuity), and increased in Adamts6 -/tissues ( Figure 1A). Thus, each gene shows reciprocal increased expression when either is inactivated, suggestive of transcriptional adaptation. In contrast, mRNA for Adamts17, also linked to WMS (Morales et al., 2009), and its homolog Adamts19 were unchanged in Adamts6-or Adamts10-deficient tissues ( Figure 1A). We also asked whether ADAMTS6 and ADAMTS10 cleaved each other. Neither ADAMTS10 nor furinoptimized ADAMTS10 cleaved ADAMTS6 and ADAMTS6 did not cleave ADAMTS10 (Figure 1-figure  supplement 1). These finding raised the possibility that Adamts10 -/and Adamts6 -/phenotypes could have been buffered by a compensating increase in the homologous mRNA and resulting protease activity. Furthermore, these genes could have cooperative functions, such as were previously identified in combined mutants of other homologous ADAMTS proteases (Mead et al., 2018b;Nandadasa et al., 2019). Both possibilities were addressed by analyzing the role of ADAMTS6 in skeletal development, characterization of the skeletal defects in Adamts6 -/-;Adamts10 -/mice and determination of ADAMTS6 activity in regard to fibrillins, none of which were previously undertaken. Combinations of the two mutant alleles were obtained by interbreeding, since each gene is on a distinct mouse chromosome.
Histologic comparison of alcian blue-stained hindlimb long bone sections at daily intervals from E14.5-E18.5 showed delayed appearance of femoral and fibular primary ossific centers in Adamts6 -/embyros, with persistence of hypertrophic chondrocytes and delayed vascular invasion; in the tibia, the primary ossific center appeared on schedule, but was persistently smaller (Figure 1-figure supplements 6-8). The distal tibia and fibula showed pronounced angulation as early as E14.5, persisting through E18.5 (Figure 1-figure supplements 7-8). Mineral deposition, visualized by von Kossa staining at E14.5, was reduced in the perichondrium/periosteum as well as within the primary center hypertrophic zones in Adamts6 -/femur and tibia and undetectable in the fibula as compared to control (Figure 1-figure supplement 9A, B), whereas by E18.5, mineralization was seen in each primary center (Figure 1-figure supplement 9C, D).
The axial and craniofacial skeleton were also abnormal in E18.5 Adamts6 -/embryos, with shortened, tubular ribs, absence of sternal segmentation and under-ossified xiphoid processes (Figure 1-figure  supplement 10). Additionally, their vertebral bodies were smaller in size with corresponding size reduction of all vertebral ossification centers. Adamts6 -/craniofacial skeletons had reduced anteriorposterior and nasal-occipital dimensions, a corresponding reduction in size of individual elements, delayed ossification of parietal bones and wider anterior fontanelles (Figure 1-figure supplement  10).
Adamts6 -/-;Adamts10 -/embryos demonstrated markedly more severe external anomalies than Adamts6 -/mutants including subcutaneous hemorrhage, micrognathia and an omphalocele, along with more severe forelimb and hindlimb dysmorphology ( Figure 1D). Skeletal preparations showed more severe hindlimb anomalies than in Adamts6 -/mutants and appearance of forelimb abnormalities similar in degree of severity to Adamts6 -/hindlimbs with externally evident shortening, the greatest reduction of nose-rump length among all genotypes and the shortest long bones among the generated genotypes ( Figure 1D-F, Figure 1-figure supplement 4). Fibular ossification was minimal, and the zeugopods showed pronounced tibio-fibular angulation ( Figure 1F). The shortened and tubular ribs, vertebral bodies with smaller ossification centers, lack of xiphoid process ossification and poor           . Whereas inactivation of one Adamts6 allele in Adamts10 -/mice did not significantly worsen the observed dysmorphology and skeletal phenotype, Adamts10 haploinsufficiency further reduced Adamts6 -/crown to rump length and led to shorter femur, tibia, fibula, humerus, radius and ulna ( Figure 1-figure supplement 4B).
Adamts6 and Adamts10 have overlapping expression in limb skeletal development To follow up on prior publications showing strong expression of Adamts6 in the heart and of Adamts10 in multiple embryonic and adult tissues (Wang et al., 2019a;Prins et al., 2018;Somerville et al., 2004) as well as immunohistochemical localization of ADAMTS10 in limb growth cartilage, perichondrium and muscle, we compared the spatial and temporal localization of their mRNAs during hindlimb development using RNAscope in situ hybridization. Adamts6 and Adamts10 exhibited overlapping expression in E13.5 resting and proliferating zone chondrocytes and perichondrium (Figure 2A). At E14.5, when the first primary ossific centers are formed, both genes maintained strong expression in the perichondrium ( Figure 2B). They are expressed in the the knee joint mesenchyme, and Adamts6 was strongly expressed in the patellar tendon ( Figure 2C). Both genes were expressed in muscles and tendons around the hip, elbow, and ankle and in vertebral endplates ( Figure 2C and D).

ADAMTS6 binds to fibrillin-2 microfibrils formed by cultured fibroblasts
To determine the functional relationship of ADAMTS6 to the two fibrillins, we investigated ADAMTS6 binding to homotypic fibrillin-2 or fibrillin-1 microfibrils assembled by Fbn1 -/and Fbn2 -/mouse embryo fibroblasts (MEFs), respectively. This specific approach allowed categoric examination of binding to each individual fibrillin, since when co-expressed, the fibrillins form heterotypic fibrils and thus isoform-specific binding cannot be otherwise discerned. Co-cultures of Fbn1 -/-MEFs with HEK293T cells stably expressing catalytically inactive ADAMTS6 (ADAMTS6 Glu 404 Ala, referred to hereafter as ADAMTS6 EA) for 6 days illustrated specific co-localization of ADAMTS6 EA with fibrillin-2 microfibrils ( Figure 5A, top row). When Fbn1 -/-MEFs were co-cultured with HEK293F cells expressing active ADAMTS6 ( Figure 5A, 2nd row), no fibrillin-2 staining was seen, suggestive of proteolytic degradation of fibrillin-2 microfibrils or interference with their assembly. Similarly, 4-day-old cultures of Fbn2 -/mouse skin fibroblasts (MSF) revealed co-localization of ADAMTS6 EA with fibrillin-1 and demonstrated absence of fibrillin-1 microfibrils in the presence of ADAMTS6 ( Figure 5A, center panels). Since a fibronectin fibrillar matrix is formed soon after plating of cells and acts as a template for fibrillin microfibril assembly (Beene et al., 2013;Sabatier et al., 2009), we similarly analyzed fibronectin staining, observing that ADAMTS6 EA co-localized with fibronectin and that fibronectin fibrils were absent in the presence of ADAMTS6, suggesting also that over-expressed ADAMTS6 may cleave fibronectin ( Figure 5A, bottom panels). In agreement, MEFs isolated from Adamts6 -/embryos showed greater fibrillin-2, fibrillin-1, and fibronectin microfibril staining intensity than wild type MEFs (Figure 4-figure supplement 1A, Figure 5-figure supplement 1). Together, these findings suggested that ADAMTS6 binds to and proteolytically degrades fibronectin and fibrillin microfibrils or alternatively, unassembled fibrillin monomers prior to their assembly.
Since fibrillin microfibrils contain additional components beside fibrillins (Cain et al., 2006;De Maria et al., 2017;Fujikawa et al., 2017;Mecham and Gibson, 2015), which could bind to ADAMTS6, we next asked whether purified ADAMTS6 constructs bound directly to purified fibrillin-2 and fibrillin-1 in a sensitive binary interaction assay. Since ADAMTS proteases typically bind to their substrates via interactions of their C-terminal ancillary domains, and because of the difficulty in purifying full-length ADAMTS6, we used C-terminal ADAMTS6 constructs for these studies. Biacore analysis showed that C-terminal ADAMTS6 constructs (ADAMTS6-Ct, ADAMTS6-S4TSR and ADAMTS6-4TSR) bound the C-terminal half of fibrillin-2 (fibrillin-2-Ct) ( Figure 6A-C, Table 1). Since all 4TSR-array-containing constructs bound to fibrillin-2-Ct we concluded that the fibrillin-2 binding region of ADAMTS6 was located in the C-terminal TSR array. In reciprocal Biacore analysis using ADAMTS6-Ct as the immobilized ligand, its binding to both the N-and C-terminal halves of fibrillin-2 was observed ( Figure 6D, Table 2). Comparable K D values of 43 nM for fibrillin-2-Nt and 80 nM for fibrillin-2-Ct suggested two or more binding sites in each half of fibrillin-2 with similar affinities for ADAMTS6. Alternatively, the ADAMTS6 binding site on fibrillin-2 may lie in the overlapping region of the two fragments between cbEGF22 and cbEGF24. Consistent with prior work showing ADAMTS6 binding to fibrillin-1 (Cain et al., 2016), ADAMTS6-Ct bound to fibrillin-1 N-and C-terminal halves ( Figure 6-figure supplement 1A, B), although stronger binding was observed to the N-terminal half ( Figure 6-figure supplement 1C). In summary, these high-affinity binary interactions suggested that ADAMTS6 binds directly to both fibrillin-1 and fibrillin-2 to cleave them. was strongly expressed in the patellar tendon (C), arrowhead, PT. Both genes were expressed in the perichondrium, muscles and tendons around the hip, elbow tendons (arrow; H, humerus, R, radius) and muscle (arrowhead), and in the shoulder rotator cuff tendons (arrowhead, RC). (D) Adamts6 and Adamts10 were expressed in the vertebral endplates (arrows), rib intercostal muscles (arrows), Achilles tendon in the ankle (arrow; C, calcaneus), in perichondrium and tendons in the autopod. Direct binding of ADAMTS6 constructs to recombinant fibrillin-2, ADAMTS6 EA co-localization with fibrillin-2 microfibrils, loss of fibrillin-2 microfibrils in vitro in the presence of wild-type ADAMTS6, and increased fibrillin-2 staining in Adamts6-deficient hindlimbs suggested that fibrillin-2 was likely to be an ADAMTS6 substrate. To define the cleavage site(s), HEK293F cells stably expressing the N-or C-terminal halves of fibrillin-2 were transfected with either ADAMTS6 or ADAMTS6 EA and the serum-free conditioned medium was collected for Terminal Amine Isotopic Labeling of Substrates (TAILS) ( Figure 7A), an N-terminomics strategy for identifying protease substrates and cleavage sites (Kleifeld et al., 2010;Grant and Li, 2016). This strategy was previously applied for identification of substrates of other ADAMTS proteases (Bekhouche et al., 2016). Proteins were labeled with stable isotopes of formaldehyde (natural (CH 2 O)/light isotope applied to the ADAMTS6-containing medium or isotopically heavy ( 13 CD 2 O), applied to the ADAMTS6 EA-containing medium). The ensuing reductive dimethylation labels and blocks free protein N-termini as well as lysine side-chains. Labeled proteins from each pair of ADAMTS6/ADAMTS6 EA digests were combined, digested with trypsin to obtain peptides and mixed with hyperbranched polyglycerol aldehyde polymer which binds to free N-termini of tryptic peptides to exclude them ( Figure 7A), enriching the peptides having blocked and labeled N-termini for liquid chromatography tandem mass spectrometry (LC-MS/MS). Following LC-MS/MS, we undertook a targeted search for fibrillin-2 peptides with isotopically labeled N-termini having a statistically significant light:heavy ratio. This analysis revealed a putative cleavage site at the Gly 2158 -His 2159 peptide bond in a linker between TGFβ binding-like domain 6 and calcium-binding epidermal growth factor (cbEGF) repeat 32 in the C-terminal half of fibrillin-2 ( Figure 7B-F). The sequence of the indicator peptide, which was semi-tryptic, and therefore unlikely to result from trypsin digestion during sample preparation and the presence of an N-terminal label was determined with high confidence by the MS 2 spectrum ( Figure 7C). Quantification of the relative ion abundance of isotoptically labeled peptides showed a considerable excess of this peptide in the presence of active ADAMTS6 ( Figure 7D-E). Western blot of the medium from these experiments for orthogonal validation of the cleavage showed that ADAMTS6 cleaved the C-terminal half of fibrillin-2 ( Figure 7G). The cleavage products of 100 kDa and 75 kDa matched the predicted cleavage fragments in size and their sum matched the expected mass of the FBN2-Ct construct (175 kDa). Importantly, the Gly 2158 -His 2159 cleavage site is located between two folded, disulfide-bonded domains, predicting separation of the resulting fragments (as opposed to cleavage sites within disulfide-bonded fibrillin-2 domains that would remain linked). The cleavage can potentially interfere with microfibril assembly, which is reliant on multimerization via C-terminal interactions downstream of the cleavage site (Hubmacher et al., 2008).

ADAMTS6 cleaves fibrillin-1 and fibronectin
Because ADAMTS6 also bound directly to fibrillin-1, TAILS was applied in a similar approach as above to determine if ADAMTS6 cleaved fibrillin-1 (Figure 7-figure supplement 1A). A targeted search for fibrillin-1 peptides with labeled N-termini revealed a putative cleavage site at the R 516 -A 517 peptide bond in a linker between EGF4 and cbEGF3 in the N-terminal half (Figure 7-figure supplement  1B). The peptide sequence and presence of an N-terminal label was determined with high confidence by the MS 2 spectrum and quantification of the ion abundance showed a considerable excess of this peptide in the presence of ADAMTS6 (Figure 7-figure supplement 1C-E).
The online version of this article includes the following figure supplement(s) for figure 3:        (MEFs), which can produce fibrillin-2 but not fibrillin-1, were cocultured with human embryonic kidney (HEK) cells overexpressing ADAMTS6 or ADAMTS6 EA (inactive ADAMTS6). Fibrillin-2 (green) co-localized with histidine (His)-tagged ADAMTS6 EA (red), but no microfibrils were seen in the presence of To determine whether ADAMTS6 cleaved fibronectin, ADAMTS6 or ADAMTS6 EA-expressing stable cell lines were plated on human fibronectin-coated plates and the serum-free medium from these cultures was collected 24 hr later. By comparing the ratio of light:heavy dimethyl-labeled peptides, TAILS revealed several potential cleavage sites in the medium from the ADAMTS6 sample ( Fibrillin-2 reduction, but not fibrillin-1 reduction specifically reverses morphogenetic and molecular defects in Adamts6 -/mice Identification of fibrillin-1, fibrillin-2 and fibronectin as ADAMTS6 substrates, yet observation of only fibrillin-2 accumulation in the hindlimbs suggested that while all three ECM components could be cleaved by ADAMTS6, fibrillin-2 could either be a preferred substrate, or not as extensively cleaved by other proteases active during limb development, which may compensate for the absence of ADAMTS6. We therefore undertook a genetic approach to determine the importance of fibrillin-2 and fibrillin-1 with regards to ADAMTS6 activity. To test whether fibrillin-2 had a significant role in the observed anomalies, fibrillin-2 was reduced genetically in Adamts6 -/mice. Fbn2 haploinsufficiency dramatically restored the external craniofacial and hindlimb morphology of Adamts6 mutants as well as the maturity, length and shape of skeletal components of the hindlimb and axial skeleton ( Figure 8A  In addition, cleft secondary palate which occurred with a high incidence in Adamts6 -/embryos, was not observed in Adamts6 -/-;Fbn2 +/embryos ( Figure 8-figure supplement 2B). Adamts6 -/-;Fbn2 +/hindlimb fibrillin-2 staining showed reduced intensity comparable to wild type hindlimb ( Figure 8C, Figure 8-figure supplement 3). Aggrecan, cartilage link protein and Sox9 staining intensity in Adamts6 -/-;Fbn2 +/femur growth plates were restored to comparable levels as in wild type limbs, indicating restoration of the appropriate ECM of cartilage ( Figure 8D, Figure 8-figure supplement 3). Fbn2 +/embryos had normal hindlimb development and growth plate zone organization as wild type controls (Figure 8figure supplement 3, Figure 8-figure supplement 4C-E). In contrast to Adamts6 -/-;Fbn2 +/embryos, Adamts6 -/embryos with ~80% reduction in fibrillin-1 levels (Adamts6 -/-;Fbn1 mgR/mgR embryos) had no amelioration of limb defects (Figure 8-figure supplement 5A, B). Taken together, these genetic analyses using well-characterized Fbn1 and Fbn2 mutants, suggested a specific role for ADAMTS6 in modulating fibrillin-2 abundance during limb microfibril proteostasis.
Several BMPs and GDFs bind fibrillin-2 directly through their propeptides (Furlan et al., 2021;Wohl et al., 2016;Gregory et al., 2005;Sengle et al., 2008;Sengle et al., 2011). We therefore considered the possibility that increased fibrillin-2 in Adamts6-deficent hindlimbs resulted in sequestration of BMPs/GDFs. Antibodies applicable for immunostaining of mouse tissues are available for very few of these; accordingly, we co-immunostained sections with a fibrillin-2 antibody and a GDF5 antibody. We consistently observed stronger co-localized fibrillin-2 and GDF5 signal in Adamts6 mutant knee joint mesenchyme, as well as in the perichondrium and interjoint regions in distal limbs, which are at less mature stages of development and have abundant undifferentiated limb mesenchyme (illustrated in knee joint interzone in Figure 10A-C, Figure 10-figure supplement 1A). Thus, increased fibrillin-2 was associated with increased GDF5 sequestration in Adamts6-deficent limbs.

Discussion
Regulated ECM proteostasis is required for ensuring proper adult tissue function (mitigating the risk of organ fibrosis, for example), but presumably also occurs during transition from the early to late embryo ECM, and may be a major determinant of adaptation of embryonic tissues to the dramatically different postnatal mechanical and regulatory landscapes. Because of the diversity of molecules and networks that comprise the ECM and existence of numerous secreted proteases, we hypothesized that selective proteolysis by specific proteases having preferred ECM substrates, may have a crucial role in ECM proteostasis in specific contexts of embryonic development. One example of a specific ECM proteostatic phenomenon mediated by ADAMTS proteases that influences embryonic cell behavior and is strongly supported by mouse genetic models is their role in versican turnover during interdigital web regression (Dubail et al., 2014;McCulloch et al., 2009;Nandadasa et al., 2021). Other than this example, little is known about the fate of macromolecular ECM complexes, especially those formed by homologous proteins with different stoichiometry in the embryo versus adult, and their proteostatic mechanisms.
Here, using a combination of genetic, biochemical, cell culture and proteomics approaches, we have identified ADAMTS6 and ADAMTS10 as major participants in fine-tuning of fibrillin microfibrils during limb development. Although prior work linked ADAMTS10 to genetic conditions that also resulted from FBN1 mutations, the existence of a closely related homolog, ADAMTS6, raised questions about the role of transcriptional adaptation in this protease pair, their overlapping functions on an organ, tissue and molecular scale, the identity of their specific targets and relevant function in regard to fibrillin-1 and fibrillin-2. Resolution of these questions provides a conceptual precedent and strategic framework for elucidation of embryonic other than ADAMTS6-TCS, which did not bind. (D) Reciprocal Biacore analysis using immobilized ADAMTS6-Ct shows that fibrillin-2-Nt and fibrillin-2-Ct used as the analyte each bound strongly to ADAMTS6-Ct.
The online version of this article includes the following figure supplement(s) for figure 6: Figure supplement 1. ADAMTS6 binds directly to fibrillin-1.  ECM proteostatic mechanisms, about which little is known relative to cellular homeostatic mechanisms in the embryo. Prior work showed that transcriptional adaptation is operational in genes encoding paired homologous proteases ADAMTS7 and ADAMTS12, and is potentially a broad phenomenon in this family, whose mammalian members have arisen largely from gene duplication (Huxley-Jones et al., 2005). Although the present study illustrates application of the phenomenon to the homologous pair of Adamts6 and Adamts10, the two proteases are clearly distinct in their biological functions as outlined in prior and present work. While the combined knockouts clearly had a more severe phenotype (indicating cooperative function), it is presently not possible to demarcate the relative contributions of transcriptional buffering to mitigation of the individual phenotypes.

Figure 6 continued
In the context of mechanisms that would ensure fibrillin-1 predominance postnatally, prior RNA in situ hybridization analysis had demonstrated dramatic reduction of Fbn2 mRNA expression after birth, with Fbn1 expression continuing (Zhang et al., 1995) and corresponding to this, little fibrillin-2 is detected immunohistochemically or by proteomics in adult mouse tissues (De Maria et al., 2017;Hubmacher et al., 2008;Cain et al., 2005;Dallas et al., 2000;Kettle et al., 1999;Kinsey et al., from cleavage by ADAMTS6 (black) and the same peptides sought in ADAMTS6 EA-containing medium (red) that were used for measuring abundance. (C) Annotated MS 2 spectrum of the light (ADAMTS6-generated) dimethyl peptide, showing b-(N-terminus preserved) and y-type (C-terminus preserved) ions generated by amide bond cleavage during collisional-induced dissociation that were used to derive the fibrillin-2 peptide sequence indicated at the top right. (D) Retention time-aligned extracted ion chromatograms (EICs) comparing abundance of the light dimethyl-labeled HGTVPSLHDTR peptide (blue) in ADAMTS6 medium and isotopically heavy dimethyl-labeled peptide (orange) in ADAMTS6 EA medium from three replicate TAILS experiments. (E) The area under the EIC curves was quantified and comparison of ion abundance is shown in a dumb-bell plot (from 3 TAILS and 3 pre-TAILS samples). Significance was determined using a two-tailed, paired Student t-test, * indicates p-value < .05. (F) Domain structure of fibrillin-2 and the C-terminal construct FBN2-Ct showing the location of the cleaved peptide bond Gly 2158 -His 2159 in the linker between TB6 and cbEGF32. (G) Orthogonal validation of fibrillin-2 cleavage by ADAMTS6 using western blot analysis of the conditioned medium from A, shows distinct molecular species (100 kDa and 75 kDa) reactive with anti-fibrillin-2-Ct antibody (green, N-terminal fragment of fibrillin-2-Ct) and C-terminal anti-His 6 antibody (red, C-terminal fragment of fibrillin-2-Ct), respectively, obtained in the presence of ADAMTS6 WT, but not ADAMTS6 EA, indicative of fibrillin-2-Ct cleavage. The green band of ~350 kDa is endogenous fibrillin-2 produced by HEK293F cells. The yellow band at ~175 kDa indicates overlapping anti-His 6 and anti-fibrillin-2-Ct antibody staining of the fibrillin-2-Ct construct. Cells transfected with empty vector were also used to obtain control medium.
The online version of this article includes the following figure supplement(s) for figure 7:     2008). Thus, differential transcription of fibrillin genes favors reduced fibrillin-2 synthesis postnatally, and together with dominance of Fbn1 expression, was thought to underlie the prevalence of fibrillin-1 microfibrils in juvenile and adult mice. Fibrillin-2 was identified postnatally in only a few locations, and in some tissues, microfibril bundles had a core of fibrillin-2 microfibrils surrounded by abundant fibrillin-1, suggesting that postnatal microfibrils (comprising fibrillin-1) were added on pre-existing fibrillin-2-rich microfibrils (De Maria et al., 2017;Charbonneau et al., 2010). Previous work also suggested that fibrillin-2 epitope availability was masked postnatally by such overlay of fibrillin-1. Specifically, it was shown that fibrillin-2 staining in postnatal tissues such as perichondrium was enhanced by digestion with collagenase, that microfibrils treated with chaotropic agent had enhanced reactivity to fibrillin-2 antibodies, and that Fbn1 null juvenile mice, which die by 2 weeks of age, had robust fibrillin-2 staining (Charbonneau et al., 2010). Thus, it was thought that the fibrillin isoform content of microfibrils is substantially determined by the level of transcription of the respective genes and that fibrillin-2 fibrils are mostly masked postnatally, such that they may not have much of a regulatory role. The fate of fibrillin-2 microfibrils produced in the embryonic period, the possibility of specific proteolytic mechanisms to reduce their abundance, and a deleterious impact of fibrillin-2 overabundance on morphogenesis had not been previously considered and no animal models with excess fibrillin-2 are available.
The present work suggests that ADAMTS6 and ADAMTS10 work collaboratively, but in distinct ways, to support prevalence of fibrillin-1 in microfibrils after birth. ADAMTS10 is innately resistant to activation by furin, with only a small proportion converted to an active protease. ADAMTS10 promotes fibrillin-1 microfibril assembly (Kutz et al., 2011), consistent with the observation that recessive ADAMTS10 mutations and dominantly inherited FBN1 mutations each lead to Weill-Marchesani syndrome (Dagoneau et al., 2004;Faivre et al., 2003a). ADAMTS10 was therefore thought to function akin to several ADAMTS-like proteins, which lack catalytic activity, and accelerate biogenesis of fibrillin-1 containing microfibrils (Hubmacher and Apte, 2015;Bader et al., 2010;Bader et al., 2012;Gabriel et al., 2012;Le Goff et al., 2008;Saito et al., 2011;Tsutsui et al., 2010). ADAMTS10, if furin-activated by introduction of a furin cleavage site, was recently demonstrated to be proteolytically active against fibrillin-2 (Wang et al., 2019a), suggesting that the small fraction of ADAMTS10 that underwent activation could cleave fibrillin-2, and indeed, persistent fibrillin-2 fibrils were seen in Adamts10-deficient mice, which are viable to adulthood. The present work shows that ADAMTS6, whose zymogen is efficiently activated by furin (Cain et al., 2016), cleaves both monomeric fibrillin-1 and fibrillin-2, and has identified specific sites of cleavage of these substrates using TAILS and validated proteolysis biochemically in the case of the fibrillin-2 cleavage site. Fibrillin microfibril proteolysis was supported by loss of fibrillin-2 and fibrillin-1 microfibrils assembled by Fbn1 -/and Fbn2 -/fibroblasts, respectively, in the presence of active ADAMTS6. Although we cannot exclude the possibility that ADAMTS6 acts via activation of another protease in the cell culture system that was used, the direct binding of ADAMTS6 to the fibrillins in a binary interaction assay strongly supports direct cleavage. The significance of reduced fibrillin-2 proteolysis and fibrillin-2 accumulation after fibular (arrows) length, shape and alignment (B). The lower panels in (B) illustrate the tibia and fibula (arrowhead) at higher magnification. (C) Fibrillin-2 staining is reduced to wild type levels in Adamts6 -/-;Fbn2 +/distal femoral perichondrium. T, tibia, F, femur. (D) Restitution of aggrecan, cartilage link protein (CLP) and Sox9 staining in Adamts6 -/-;Fbn2 +/distal femoral cartilage. T, tibia, F, femur.
The online version of this article includes the following figure supplement(s) for figure 8:      Adamts6 inactivation was demonstrated by dramatic reversal of the observed skeletal and palate defects after introducing Fbn2 haploinsufficiency in Adamts6 mutants. The lack of a comparable effect after reduction of fibrillin-1 suggested a primacy of the ADAMTS6-fibrillin-2 relationship. Although ADAMTS6 cleaves fibrillin-1, it is not as essential for proteolysis of fibrillin-1, presumably because other proteases potentially compensate. While we cannot exclude the likelihood of accumulation of additional ADAMTS6 proteolytic targets in the skeleton, such as LTBP-1, which it was previously shown to cleave (Cain et al., 2016), the present work strongly suggests that fibrillin-2 proteolysis by ADAMTS6 is necessary for proper skeletogenesis. pSMAD5 levels from the western blot in (C) was quantified after normalization to GAPDH loading control. n = 3. *p ≤ 0.05. (E) No change in pSMAD2 staining in Adamts6 -/femur as compared to wild-type control. The results are quantified in (F). (G) Western blot analysis shows no change in anti-pSMAD2 (green; 52 kDa) in Adamts6-deficient E18.5 hindlimb lysates. Anti-GAPDH (red, 37 kDa) was used as a loading control. (H) Quantification of anti-pSMAD2 signal in (G) using GAPDH signal intensity as the control. n = 3.

Figure 9 continued
In addition to mechanical effects of excess perichondrial fibrillin-2 on cartilage that could affect skeletal development, reduced fibrillin-2 proteolysis could lead to dysregulated sequestration and release of growth factors of the TGFβ superfamily, since fibrillin-2 can bind to numerous superfamily members including several BMPs and GDFs (Sengle et al., 2015;Gregory et al., 2005;Sengle et al., 2011). Fibrillin microfibrils appear to have little role in activation of the BMPs, and primarily affect their local concentrations and release, for example, by BMP cleavage, which was recently attributed to MMPs (Furlan et al., 2021;Wohl et al., 2016). In contrast, proteolysis of fibrillins was not previously implicated in control of local BMP/GDF activity. In the present work, the microfibril changes in the perichondrium were associated with changes in the underlying cartilage, in further support of phenotype modulation across the perichondrium-cartilage boundary (Colnot et al., 2005;Maes, 2017;Vortkamp et al., 1996). Prior work has shown that combinations of TGFβ superfamily members strongly promote expression of typical cartilage genes (Murphy et al., 2015;Hatakeyama et al., 2004). This previous work together with the observed reduction in pSmad5 levels (indicative of reduced BMP signaling) may explain reduced Sox9, aggrecan and link protein that we observed. The resulting reduced cartilage proteoglycan potentially results in softer cartilage that could be susceptible to angulation forces as limb muscles develop.
Consistent with the conclusion that Adamts6 deficiency reduces limb cartilage BMP signaling as a result of excess fibrillin-2, Fbn2-deficent mice have increased BMP signaling (Sengle et al., 2015). Thus, accumulation of fibrillin-2 microfibrils resulting from combined ADAMTS6 and ADAMTS10 deficiency may excessively sequester BMPs and GDFs in the perichondrium, leading to impaired BMP signaling in the underlying cartilage ( Figure 11). Fibrillins interact with the related latent TGFβbinding proteins and furthermore, MAGP1, a fibrillin-binding partner shown to accumulate in the Adamts6 mutant mice, also regulates the cellular microenvironment via growth factor binding (Miyamoto et al., 2006;Nehring et al., 2005). Moreover, ADAMTS6-mediated proteolysis of cell-surface heparan-sulfate proteoglycans (Cain et al., 2016) could also influence the skeletal microenvironment. However, although such complex effects on diverse soluble morphogens and growth factors may ensue downstream of the observed ECM changes, the genetic evidence from our studies unequivocally indicates that fibrillin-2 accumulation is a central mechanism underlying the skeletal defects observed in the Adamts6 mutant limbs. Thus, the present work highlights the significance of specific proteolytic mechanisms in fibrillin microfibril proteostasis, their importance for morphogen/growth factor regulation by ECM and ultimately, for cellular regulation by ECM.
Alizarin red-alcian blue stained skeleton preparations were performed as described (Mead, 2020a). Briefly, skinned, eviscerated mice were fixed in 80% ethanol for 24 hr, dehydrated in 95% ethanol for 24 hr and acetone for 48 hr, stained (0.1% alizarin red S, 0.3% alcian blue, 1% glacial acetic acid in 95% ethanol) for 48 hr, cleared in 95% ethanol for 1 hr, muscle tissue was gently removed with forceps, and the preparations were cleared in a series of increasing (20-80%) glycerol/1% KOH ratio until storage in 100% glycerin and photography (Leica MZ6; Insight Spot software camera and software). Crown to rump or individual bone length measurements were masked during data collection utilizing NIH Fiji software (Schindelin et al., 2012).

RNA in situ hybridization (ISH)
Adamts6 and Adamts10 ISH was performed using RNAscope (Advanced Cell Diagnostics, Newark, CA) as described (Mead and Apte, 2020b). Briefly, 6 μm sections were deparaffinized and hybridized to mouse Adamts6 and Adamts10 probe sets (428301, 585161, respectively; Advanced Cell Diagnostics) using a HybEZ oven (Advanced Cell Diagnostics) and the RNAScope 2.5 HD Detection Reagent Kit (322360; Advanced Cell Diagnostics) and counterstained with eosin. Figure 11. Schematic illustrating the observed anomalies in the absence of ADAMTS6 and proposed roles and mechanisms of ADAMTS6 in skeletal development. (A) Adamts6 expression (pink dots) is shown in the perichondrium, peripheral chondrocytes, and in mesenchyme surrounding the knee joint. BMP signaling (blue circles) is evident in growth plate cartilage while fibrillin-2 (purple rods) and BMP/GDF (orange dots) are co-localized in the mesenchyme surrounding the knee joint and long bones. Adamts6-deficient limbs have a reduction of BMP signaling in the growth plate, retention of fibrillin-2 microfibrils and sequestration of GDF5, and presumably, other known fibrillin-2 binding BMPs/GDFs. (B) The schematic shows that in wild type limbs, ADAMTS6 cleaves fibrillin-2, allowing release of growth factors and contributing to proper cartilage growth and development. In the absence of ADAMTS6, there is an over-abundance of fibrillin-2 microfibrils, sequestration of GDF5/BMPs and as a consequence, reduced Sox9, aggrecan (blue shading) and cartilage link protein, impairing cartilage differentiation and structural integrity. Mechanistically, we conclude from the data that ADAMTS6 acts principally via fibrillin-2 cleavage and release of sequestered BMP/GDF. RZ, resting zone; CZ, columnar zone; HZ, hypertrophic zone; N, N-terminus; C, C-terminus of fibrillin-2.
Cell culture HEK293F cells were purchased and authenticated from ATCC and maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 μg/ml streptomycin at 37 °C in 5% CO 2 using a humidified chamber. The cells were transiently transfected with ADAMTS6 WT or ADAMTS6 EA expression plasmids using PEI MAX (24765; Polysciences) and were co-cultured with Fbn1 -/mouse embryo fibroblasts MEFs, prepared from Fbn1 tm3Rmz homozygous mutants (Carta et al., 2006;Beene et al., 2013) or Fbn2 -/mouse skin fibroblasts (MSFs) (Beene et al., 2013) in a 1:1 ratio on 8-well culture slides (354118; Falcon). Similarily, Adamts6 -/and wild type MEFs were plated on 8-well culture slides for immunofluorescent staining. The cells were cultured for 6 days, fixed in ice-cold 70% methanol/30% acetone for 5 min at room temperature, blocked with 5% normal goat serum in PBS for 1 hr at room temperature and incubated with primary antibody (Supplementary file 1) overnight at 4 °C as described (Hubmacher et al., 2017). The cells were washed three times with PBS for 5 min at room temperature and incubated with Alexa-Fluor labeled secondary antibodies (goat anti-mouse 568 or goat anti-rabbit 488; Invitrogen A11004, A11008, respectively, 1:400).

RNA isolation and quantitative real-time PCR (qRT-PCR)
Mouse hindlimbs, hearts and lungs were snap-frozen and stored at -80 °C until use. Total RNA was isolated using TRIzol (15596018, Invitrogen), and 2 μg of RNA was reverse transcribed into cDNA using a High-Capacity cDNA reverse transcription kit following the manufacturer's instructions (4368814; Applied Biosystems, Foster City, CA). qRT-PCR was performed with Bullseye EvaGreen qPCR MasterMix (BEQPCR-S; MIDSCI) using an Applied Biosystems 7,500 instrument. The experiments were performed with three independent biological samples and reproducibility was confirmed with two to three technical replicates. Gapdh was used as a control for mRNA quantity. The ∆∆Ct method was used to calculate relative mRNA expression levels of target genes and shown as standard error of the mean (SEM). See Supplementary file 2 for primer sequences.

Surface plasmon resonance analysis
The human fibrillin-2 recombinant halves (rFBN2-N and rFBN2-C, termed FBN2-Nt and FBN2-Ct in this manuscript, respectively) were purified to homogeneity (> 90% purity) as described previously (Lin et al., 2002). Recombinant human ADAMTS6-Ct (residues 558-1117), ADAMTS6-S4TSR (residues 717-1117), ADAMTS6-4TSR (residues 840-1117) and ADAMTS6-TCS (residues 558-839) were expressed using the mammalian expression vector pCEP-pu/AC7 modified with a C-terminal V5-His6 tag, in 293-EBNA cells as described previously (Cain et al., 2016). Purified FBN2-Ct or ADAMTS6-Ct in 10 mM acetate, pH 4.0 were immobilized on a Biacore CM5 sensor chip (research grade) with the amine coupling kit following the manufacturer's instructions (GE Healthcare). A total of 1700 resonance units of FBN2-Ct or ADAMTS6-Ct was coupled to the chip for analysis in a Biacore 3000 instrument (GE Healthcare). The kinetics analysis was performed at 25 °C in 10 mM Hepes buffer, pH 7.4 with 0.15 M NaCl, 2 mM CaCl 2 , and 0.005% or 0.05% (v/v) surfactant P20 at a flow rate of 30 μl/ min. All the analytes were diluted in the above buffer at different concentrations and injected through an uncoupled control flow cell in series with the flow cell coupled with FBN2-Ct or ADAMTS6-Ct constructs. The sample injection time was 2 min for ADAMTS6 and 3 min for FBN2 analytes. The dissociation time was 10 min. 1 M ethanolamine, pH 8.5 was used for chip surface regeneration at a flow rate of 30 μl/min for 30-60 s followed by 2 min stabilization time. All data were corrected with reference to the background binding in the control flow cell. The association and disassociation curves were generated with the BIAevaluation software (version 4.0.1; GE Healthcare). The kinetic constants were calculated using the steady state affinity method. Similar experiments were performed with the human fibrillin-1 recombinant halves (rFBN1-N and rFBN1-C, termed FBN1-Nt and FBN1-Ct in this manuscript, respectively) (Lin et al., 2002) with ADAMTS6-Ct.

Cell co-culture for degradomics
Transiently transfected HEK293F cells expressing wild type ADAMTS6 or ADAMTS6 EA were cocultured with HEK293F cells stably expressing either fibrillin-2 Ct or full-length fibrillin-1. To determine whether fibronectin was an ADAMTS6 substrate, HEK 293F cells were transiently transfected with either ADAMTS6 or ADAMTS6 EA plasmids and seeded onto wells coated with human fibronectin as previously described (Wang et al., 2019b). These experiments used serum-free and phenol red-free DMEM.

TAILS sample workflow
The conditioned medium from the above cell cultures were centrifuged at 4,000 rcf for 20 min at 4 °C and the supernatant was filtered through a 0.22 µM filter. The medium was concentrated 20-fold using a 3 kDa (Amicon) stirring filter. Proteins were isolated using chloroform/methanol precipitation and resuspended in 2.5 M GuHCl and 250 mM HEPES pH 7.8. Protein concentration was measured using the Bradford assay (Pierce, Thermo) to determine the volume needed for 500 µg of protein from each condition. Proteins were reduced with 10 mM dithiothreitol (DTT) for 30 min at 37 °C followed by alkylation with 20 mM N-ethylmaleimide in the dark for 20 min. The reaction was quenched by adding DTT to a final concentration of 20 mM. Proteins were labeled overnight with 40 mM light or heavy formaldehyde, which binds specifically to free N-termini and lysine residues (α and ε amines, respectively) in the presence of 20 mM sodium cyanoborohydride at 37 °C as described (Martin et al., 2020). They were treated with an additional fresh 20 mM formaldehyde and 10 mM sodium cyanoborohydride for 2 hr the following day at 37 °C and the reaction was quenched with 100 mM Tris for 1 hr at 37 °C. 500 μg of each sample was combined for buffer exchange on a 3 kDa molecular weight cut-off column (EMD Millipore) into 100 mM ammonium bicarbonate and digested overnight at 37 °C with mass spectrometry grade trypsin gold (Promega) at a 1:50 trypsin:protein ratio. Peptides were eluted via centrifugation and 30 µg of this digest was reserved for pre-TAILS analysis. The remaining peptides underwent enrichment using hyperbranched polyglycerol-aldehyde polymers (HPG-ALD, Flintbox, https://www.flintbox.com/public/project/1948/) at a 5:1 polymer:protein ratio. HPG-ALD binds to unblocked (trypsin-generated) amino acid termini, excluding them from the sample and thus enriches for blocked/labeled N-termini (Kockmann et al., 2016). The peptides were filtered through a 10 kDa MWCO column (EMD Millipore) to remove the polymer and obtain the TAILS fraction. TAILS and pre-TAILS fractions were desalted on a C18 Sep-Pak column (Waters) and eluted in 60:40 ACN: 1% trifluoroacetic acid. Samples were vacuum-centrifuged until dry and resuspended in 1% acetic acid for mass spectrometry.

Mass spectrometry
Samples were analyzed on a ThermoFisher Scientific Fusion Lumos tribrid mass spectrometer system with a Thermo Ultimate 3,000 UHPLC interface. The HPLC column was a Dionex 15 cm x 75 µm internal diameter Acclaim Pepmap C18, 2 μm, 100 Å reversed-phase capillary chromatography column. Five μL volumes of the samples were injected and the peptides eluted from the column by an acetonitrile/0.1% formic acid gradient at a flow rate of 0.3 μL/min were introduced into the source of the mass spectrometer on-line over a 120 min gradient. The nanospray ion source was operated at 1.9 kV. The digest was analyzed using a data-dependent method with 35% collision-induced dissociation fragmentation of the most abundant peptides every 3 s and an isolation window of 0.7 m/z for iontrap MS/MS. Scans were conducted at a maximum resolution of 120,000 for full MS. Dynamic exclusion was enabled with a repeat count of 1 and ions within 10 ppm of the fragmented mass were excluded for 60 s.

Proteomics data analysis
Peptides were identified using a precursor mass tolerance of 10 ppm, and fragment mass tolerance of 0.6 Da. The only static modification was carbamidomethyl (C), whereas dynamic modifications included the light (28.03 Da) dimethyl formaldehyde (N-terminal, K), heavy (34.06) dimethyl formaldehyde (N-terminal, K), oxidation (M, P), deamidation (N), acetylation (N-terminal), and Gln to pyro-Glu N-terminal cyclization. Peptides were validated using a false discovery rate (FDR) of 1% against a decoy database. Only high-confidence proteins (containing peptides at a 99% confidence level or higher) were recorded from each sample for data analysis. Internal peptides were identified based on the criteria of having an N-terminal modification, and a sequence that does not begin prior to the third amino acid in the protein or immediately following a known signal, transit, or propeptide sequence. Peptides that met these criteria were further analyzed based on the average fold-change ratio (ADAMTS6 WT/ ADAMTS6 EA sample abundance) between the three technical replicate pairs. The internal peptide abundance was divided by the total protein abundance fold-change to account for differences in protein levels between groups. Peptides that met these criteria and contained a weighted ratio (ADAMTS6 WT/ ADAMTS6 EA) greater than 1 underwent a t-test for significance. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD027096 and 10.6019/PXD027096.

Statistics
Representative data of three independent, biological replicates are reported unless otherwise indicated. The two-tailed, unpaired Student t-test was used to obtain p values. Asterisks indicate differences with statistical significance as follows: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. In the dimethyl-TAILS experiments a two-tailed, paired Student t-test was used to obtain p values. Asterisks indicate differences with statistical significance as follows: *p ≤ 0.01, # ≤ 0.05.