Unusual life cycle and impact on microfibril assembly of ADAMTS17, a secreted metalloprotease mutated in genetic eye disease

Secreted metalloproteases have diverse roles in the formation, remodeling, and the destruction of extracellular matrix. Recessive mutations in the secreted metalloprotease ADAMTS17 cause ectopia lentis and short stature in humans with Weill-Marchesani-like syndrome and primary open angle glaucoma and ectopia lentis in dogs. Little is known about this protease or its connection to fibrillin microfibrils, whose major component, fibrillin-1, is genetically associated with ectopia lentis and alterations in height. Fibrillin microfibrils form the ocular zonule and are present in the drainage apparatus of the eye. We show that recombinant ADAMTS17 has unique characteristics and an unusual life cycle. It undergoes rapid autocatalytic processing in trans after its secretion from cells. Secretion of ADAMTS17 requires O-fucosylation and its autocatalytic activity does not depend on propeptide processing by furin. ADAMTS17 binds recombinant fibrillin-2 but not fibrillin-1 and does not cleave either. It colocalizes to fibrillin-1 containing microfibrils in cultured fibroblasts and suppresses fibrillin-2 (FBN2) incorporation in microfibrils, in part by transcriptional downregulation of Fbn2 mRNA expression. RNA in situ hybridization detected Adamts17 expression in specific structures in the eye, skeleton and other organs, where it may regulate the fibrillin isoform composition of microfibrils.

. ADAMTS17 undergoes autocatalytic processing. (a) Domain organization of ADAMTS17 and the constructs ADAMTS17-PCD and ADAMTS17-AD. The location of the ADAMTS17 antibody epitopes (black line) and the sites of site-directed mutagenesis (furin-site: R 223 A, active site: E 390 A) are indicated above the ADAMTS17 domains. The predicted furin/PACE cleavage sites are shown below the ADAMTS17-PCD construct. Cleavage at site R223 results in mature ("M") ADAMTS17. (b) Western blot analysis of conditioned medium (Med) and cell lysate (Lys) from full-length ADAMTS17 (17), the active site mutant ADAMTS17 EA , and empty vector (v) expressing HEK293F cells. Western blots were probed with the indicated antibodies and detected with enhanced chemiluminescence. (c) Western blot analysis of conditioned medium (Med) and cell lysate (Lys) from ADAMTS17-PCD (PCD), ADAMTS17-PCD EA (PCD-EA), and empty anomalies 13,14 . ADAMTS17 was also linked to primary open angle glaucoma and to variant height in humans and dogs [15][16][17][18][19] . These genetic studies strongly linked ADAMTS17 to FBN1 and implicated it in the formation of the ciliary zonule in the eye, the fibrillin-based structure defective in ectopia lentis, and in the regulation of skeletal growth.
Fibrillin-1 represents one of three fibrillin isoforms in humans. Fibrillins are the core constituents of tissue microfibrils, an ECM scaffold that can include microfibril-associated glycoproteins, TGFβ superfamily proteins, latent TGFβ -binding proteins, fibulins, tropoelastin, and ADAMTS proteins, such as ADAMTS10 [20][21][22][23] . These additional components may regulate microfibril assembly or contribute specialized functions to the microfibril scaffold. During the embryonic period, FBN2 mRNA expression predominates, whereas FBN1 is the dominant fibrillin mRNA isoform in adult tissues 24,25 . However, fibrillin-2 may persist in the core of adult microfibril bundles 26 . The transition from Fbn2 mRNA prevalence to Fbn1 mRNA dominance in the mouse occurs during the late embryonic to early juvenile period 25 . Because the proportion of fibrillin-1 or -2 within microfibrils may be determined by fibrillin isoform availability without a bias toward inclusion of either, the expression levels of the respective genes are likely to be important in determining microfibril composition. However, other mechanisms regulating fibrillin isoform selection may be active. For example, genetic ablation of ADAMTS-like 2 (ADAMTSL2) resulted in the accumulation of fibrillin-2 microfibrils around bronchial smooth muscle cells 27 . In addition, ADAMTS10, ADAMTSL4, and ADAMTSL6 accelerated formation of fibrillin-1 microfibrils by cultured cells, or upon overexpression in mice [28][29][30] .
Here we report the first molecular and functional characterization of ADAMTS17, which reveals important characteristics of this protease, such as autocatalytic generation of several proteoforms and identifies post-translational modifications required for ADAMTS17 secretion. In addition, we demonstrate that ADAMTS17 may have a role in regulating the fibrillin isoform composition of microfibrils in specific tissues during development.

Results
ADAMTS17 is autoproteolytically processed upon secretion. Recombinant full-length human ADAMTS17 (Fig. 1a) was detected as two molecular species in western blots of cell lysates from transiently transfected HEK293F cells, using anti-myc monoclonal antibody, as well as monoclonal and polyclonal antibodies directed to epitopes located in the ADAMTS17 ancillary domain or propeptide, respectively ( Fig. 1a,b, left-hand panel). The dominant species in cell lysates corresponded to the ADAMTS17 zymogen (Z, observed MW: ~160 kDa, predicted MW: 125.3 kDa) and the minor species to mature ADAMTS17 (observed MW: ~130 kDa, predicted MW: 102.9 kDa), consistent with constitutive proprotein convertase (e.g., furin)-mediated cleavage of the propeptide (predicted MW: 22.4 kDa) (Fig. 1b, left-hand panel). The greater than predicted MW of ADAMTS17 is ascribed to glycosylation (see below). In conditioned medium of transfected cells, secreted ADAMTS17 was poorly detected with anti-myc antibody (Fig. 1b, left-hand panel). In contrast, ADAMTS17 harboring a Glu 390 Ala mutation within the active site (ADAMTS17 EA ) was readily detected in the conditioned medium by the anti-myc antibody. The Glu 390 Ala mutation targets the catalytic Glu residue and is traditionally employed to inactivate metalloproteases 31,32 . Hence, the apparent absence of ADAMTS17 in conditioned medium is likely due to autoproteolysis (Fig. 1b, left-hand panel). As expected, most of the secreted ADAMTS17 EA was processed to the mature form. ADAMTS17 autoproteolysis was also evident from detection of ADAMTS17 peptides using the monoclonal and polyclonal antibodies against the ancillary domain and the propeptide, respectively, which were not detected in the conditioned medium of the ADAMTS17 EA mutant (Fig. 1b, center and right-hand panels). Absence of specific ADAMTS17 cleavage products in cell lysates (other than the furin-processed form), suggested that ADAMTS17 underwent autoproteolysis after secretion. Since the C-terminal antimyc reactivity was lost due to autoproteolysis of secreted wild-type ADAMTS17, and a cleavage product smaller than the zymogen but larger then mature ADAMTS17 (~150 kDa) was observed with the antibody against the propeptide, one of the autoproteolytic cleavage sites must be located close to the ADAMTS17 C-terminus. However, this putative C-terminal fragment was neither detected in western blots nor by mass spectrometry (see below). The specificity of ADAMTS17 antibodies was demonstrated by the lack of reactive bands in medium and cell lysate from cells transfected with the empty vector (Fig. 1b, lanes marked by v). A similar contrasting fragmentation of ADAMTS17 and ADAMTS17 EA was observed after transfection of ADAMTS17 in COS-1 cells (data not shown), supporting the observed autoproteolysis of ADAMTS17.
In contrast to full-length ADAMTS17, a C-terminally truncated construct, ADAMTS17-PCD (Fig. 1a), was readily detectable in the medium of transfected HEK293F cells using anti-myc antibody (Fig. 1c, red/yellow). The observed MW of the major species corresponded to the zymogen (Z, observed MW: ~65 kDa, predicted MW: 61.3 kDa, yellow) and the mature form (M, observed MW: ~40 kDa, predicted MW: 38.9 kDa, red only). The ADAMTS17-PCD zymogen, but not its mature form, was detected with the polyclonal propeptide antibody vector (v) expressing HEK293F cells. The asterisk indicates a band which is reactive with anti-propeptide antibody (anti-PP, green), but not anti-myc (red), and which is absent in ADAMTS17-PCD EA . (d) Western blot analysis of conditioned medium (Med) and cell lysate (Lys) from ADAMTS17-AD (AD) and empty vector (v) expressing HEK293F cells. The asterisk indicates intact ADAMTS17-AD (yellow), reactive with anti-myc (red) and anti-ancillary domain antibody (anti-AD, green). Arrowheads indicate species reactive with only anti-myc, but not anti-AD (red). (e) Semi-tryptic peptides resulting from ADAMTS17 autoproteolysis and identified by LC-MS/MS (blue: enriched in ADAMTS17, black: enriched in ADAMTS17 EA , red: present only in wild-type ADAMTS17) (f) Location of semi-tryptic peptides in ADAMTS17. Numbers and color coding correspond to the list in e. mAb, monoclonal antibody; M, mature enzyme; pAb, polyclonal antibody; v, vector; Z, zymogen. (green/yellow). Appearance of an anti-propeptide antibody reactive band (~55 kDa) in ADAMTS17-PCD which was unreactive with anti-myc and absent in the ADAMTS17-PCD EA construct suggested limited autoproteolysis of ADAMTS17-PCD (Fig. 1c, asterisk, green only). The fluorescent western blotting technique used here and in the following experiments allows for the simultaneous detection of two antibody epitopes on the same membrane. Therefore, protein species specifically reacting with either one of the antibodies appear red or green. Protein species reacting with both antibodies appear yellow.
A construct expressing the ADAMTS17 ancillary domain (ADAMTS17-AD, Fig. 1a) was also readily detected on western blots of medium with anti-myc and with the ancillary domain antibody (observed MW: ~85 kDa, predicted MW: 67.2 kDa) (Fig. 1d, yellow). Additional bands in ADAMTS17-AD containing medium observed with anti-myc (~73 kDa, 65 kDa, Fig. 1d, red), but unreactive with the ADAMTS17 monoclonal antibody indicated fragmentation near the N-terminus of ADAMTS17-AD by an ambient protease, with loss of the epitope for the monoclonal antibody. In both empty vector and ADAMTS17-AD transfected cell lysates but not the medium, multiple bands reactive with the monoclonal antibody were observed, indicating cross reactivity with unknown proteins (Fig. 1d, green).
To identify ADAMTS17 autocatalytic processing sites, we compared liquid chromatography tandem mass spectra (LC-MS/MS) of conditioned medium from cells expressing ADAMTS17 or ADAMTS17 EA . We detected twelve semi-tryptic peptides (i.e. peptide sequences with one tryptic cleavage site and one non-tryptic cleavage site arising from ADAMTS17 autoproteolysis) (Fig. 1e,f, Supplemental Table 1 and Supplemental Fig. 1). One peptide was uniquely present in ADAMTS17 medium, and eight were enriched in ADAMTS17 medium, suggesting autoproteolysis, whereas only three peptides were enriched in ADAMTS17 EA medium. Taken together, western blotting results comparing ADAMTS17 with ADAMTS17 EA and LC-MS/MS data unequivocally indicate that wild type ADAMTS17 is secreted and autocatalytically processed at multiple sites (Fig. 1f). Secondary processing by ambient proteases, or proteases activated by ADAMTS17 may also contribute to the fragmentation pattern, but remain to be characterized. ADAMTS17 autoproteolysis occurs in trans and does not require the ancillary domain or furin processing. To determine whether ADAMTS17 autoproteolysis occurred in cis or in trans, we co-transfected HEK293F cells with plasmids expressing ADAMTS17 EA plus ADAMTS17, ADAMTS17 EA , ADAMTS17-PCD, or empty vector control, respectively (Fig. 2a). ADAMTS17 EA was detected at reduced levels in conditioned medium upon co-expression with ADAMTS17 or ADAMTS17-PCD. Since the Glu 390 Ala mutation abolishes the proteolytic activity of ADAMTS17, the findings indicate that proteolysis of ADAMTS17 EA was mediated by co-transfected ADAMTS17 or ADAMTS17-PCD, which could be described as autoproteolysis in trans (Fig. 2a). Proteolysis of ADAMTS17 EA by ADAMTS17-PCD suggested that the ancillary domain was not required for ADAMTS17 EA proteolysis.
To determine whether furin processing of ADAMTS17 was required for autoproteolysis, we introduced an Arg223Ala mutation at the P1 residue within the furin recognition sequence 220 RERR 223 in ADAMTS17 and ADAMTS17-PCD (Fig. 1a). The resulting retarded in-gel migration (Fig. 2b, middle panel, green, Δ 1 and Δ 2, respectively) reflects retention of the propeptide in the furin-resistant ADAMTS17 RA mutant constructs. Furin-resistance of ADAMTS17-PCD RA was demonstrated by the absence of mature ADAMTS17-PCD using the anti-catalytic domain antibody (Fig. 2c, middle panel, green).
We transfected the furin-resistant constructs into the ADAMTS17-AD stable cell line. Anti-myc reactivity of ADAMTS17-AD was reduced in the presence of ADAMTS17 and ADAMTS17 RA , but not ADAMTS17 EA (Fig. 2b, left-hand panel, red). Using the monoclonal ADAMTS17 antibody, we observed a band at ~75 kDa, which was only present upon proteolytic processing of ADAMTS17-AD by ADAMTS17 and ADAMTS17 RA (Fig. 2b, right-hand panel, red, arrow), suggesting proteolytic activity despite the presence of the propeptide. Similar results were obtained upon transfection of ADAMTS17-PCD, ADAMTS17-PCD EA , or ADAMTS17-PCD RA into stable ADAMTS17-AD expressing cells (Fig. 2c). ADAMTS17-PCD and ADAMTS17-PCD RA , but not the active site mutant ADAMTS17-PCD EA , cleaved ADAMTS17-AD, demonstrated by reduced reactivity with anti-myc (Fig. 2c, left-hand panel, red) and appearance of the ~75 kDa band detected with the monoclonal ADAMTS17 antibody (Fig. 2c, right-hand panel, arrow, red). Thus, furin processing of the ADAMTS17 propeptide is not required for ADAMTS17 activation. We conclude that the domains contained in ADAMTS17-PCD are necessary and sufficient for the intermolecular interaction required for full-length ADAMTS17 autoproteolysis in trans and medium (Med) from HEK293F cells, stably expressing ADAMTS17-AD (AD) and transiently expressing ADAMTS17 (17), ADAMTS17 EA (17-EA), ADAMTS17 RA (17-RA) or empty vector (v). Note reduced levels of ADAMTS17-AD in the presence of ADAMTS17 or ADAMTS17 RA (left-hand panel, red) and mass shift of furinresistant species compared to wild type (brackets indicated by Δ 1 and Δ 2) (middle panel, green). The proteolytic product detected with anti-AD is indicated with an arrow (right-hand panel, red) and only appears upon transfection of ADAMTS17-AD expressing cells with ADAMTS17 or ADAMTS17 RA . (c) Western blot analysis of conditioned medium (Med) from HEK293F cells, stably expressing ADAMTS17-AD (AD) and transiently expressing ADAMTS17-PCD (PCD), ADAMTS17-PCD EA (PCD-EA), ADAMTS17-PCD RA (PCD-RA) or empty vector (v). The mature form of ADAMTS17-PCD (PCD-M) was not generated in ADAMTS17-PCD RA , indicating lack of furin processing (left-hand panel, red and middle panel, green). Furin processing did not abolish proteolytic activity as observed by reduction in the intensity of anti-myc reactive ADAMTS17-AD (lefthand panel, top band) or the appearance of an additional band reactive with the anti-AD antibody (arrow, right panel, red). Note that the α -Myc antibody (red) detects both ADAMTS17-PCD and ADAMTS17-AD. M, mature enzyme; Z, zymogen.
Scientific RepoRts | 7:41871 | DOI: 10.1038/srep41871  ADAMTS17 is N-glycosylated and its secretion requires O-fucosylation of thrombospondin type-1 repeats. ADAMTS17 contains seven predicted N-glycosylation sites, of which five are located in the ancillary domain (Fig. 3a). Digestion of ADAMTS17 EA with PNGase F resulted in more rapid migration of the zymogen and mature forms under reducing conditions, indicating N-glycosylation (Fig. 3b). In contrast, secreted ADAMTS17 EA was insensitive to endoglycosidase H, suggesting the N-glycans are of the complex or hybrid type (Fig. 3b).
The thrombospondin type 1 repeats (TSRs) of ADAMTS proteins are potential targets for post-translational modification by O-fucosylation, which adds a Glucose-β 1-3Fucose disaccharide to the consensus sequence Cys-Xaa-Xaa-Ser/Thr-Cys (Xaa is any amino acid, the modified residue is underlined) (Fig. 3c) 33 . Four of five TSRs in ADAMTS17 carry the required consensus sequence (Fig. 3a). Recombinant ADAMTS17 constructs 1C and 25P ( Fig. 3a) were purified to homogeneity from transiently transfected HEK293F cells for analysis by LC-MS/MS. We detected the Glucose-β 1-3Fucose disaccharide on TSR1, 3, and 5, but not TSR4 ( Table 2). O-Fucosylation is implicated in quality control of proteins containing TSRs 34,35 . To determine whether lack of or incomplete O-fucosylation affected the secretion of ADAMTS17, we transfected the ADAMTS17 constructs 1C, 25P, and ADAMTS17 EA into HEK293F cells with inactivated POFUT2 or B3GLCT, the enzymes attaching the fucose and glucose moiety, respectively, to the serine/threonine residue. In the absence of POFUT2, secretion of 1C into the medium was significantly reduced and secretion of 25P and ADAMTS17 EA was nearly abolished (Fig. 3e,f, red channel). B3GLCT knockout had no effect on the secretion of 1C, but significantly reduced secretion of 25P and ADAMTS17 EA correlating with the presence of 1, 2, or 3 O-fucosylation sites, respectively. The levels of the ADAMTS17 constructs in the cell lysates were not affected by the absence of POFUT2 or B3GLCT. Together, these results show extensive glycosylation of ADAMTS17 and its functional impact on ADAMTS17 secretion.

ADAMTS17 forms multimers via intermolecular disulfide bonds. Electrophoresis of purified
ADAMTS17-PCD under reducing and non-reducing conditions followed by Coomassie blue staining demonstrated that mature ADAMTS17-PCD migrated more rapidly under non-reducing conditions (Fig. 4a), consistent with a globular conformation in the presence of disulfide bonds. Under non-reducing conditions, however, the ADAMTS17-PCD zymogen band was considerably weaker than the mature form and high MW bands were evident, indicative of disulfide-bonded multimers (Fig. 4a). The disproportionate shift of the zymogen, but not the processed catalytic domain under non-reducing conditions suggested that the propeptide may mediate the formation of ADAMTS17-PCD multimers.
The formation of multimers by ADAMTS7-PCD was further investigated by incubating purified ADAMTS17-PCD with bis(sulfosuccinimidyl)suberate (BS 3 ), a non-reducible bifunctional amine-reactive cross-linker with a 11.4 Å spacer in the presence or absence of DTT (Fig. 4b), followed by electrophoresis under reducing conditions and Coomassie blue staining (Fig. 4c). In the absence of DTT, we observed a BS 3 concentration-dependent electrophoretic shift of the zymogen and mature forms of ADAMTS17-PCD to the upper regions of the gel (Fig. 4c, left-hand panel). In the presence of DTT, which reduces disulfide bonds, these high MW bands were absent. Incubation of bovine serum albumin (BSA) with BS 3 under the same experimental conditions did not alter its electrophoretic mobility (data not shown). This experiment, together with the differential migration of the zymogen and the mature ADAMTS17-PCD under reducing versus non-reducing conditions suggests that the oligomers arise by disulfide bond formation, but other residues of ADAMTS17-PCD are in sufficient proximity for experimental BS 3 -mediated cross-linking.
The catalytic and disintegrin-like domains of ADAMTS proteases contain eight cysteine residues each, shown to form four disulfide bonds within each domain by high-resolution structures of ADAMTS1, 4, and 5 [36][37][38] . In contrast, neither structural information nor disulfide bond mapping is available for ADAMTS propeptides. Most ADAMTS propeptides, including ADAMTS17, contain three cysteine residues, suggesting that one unpaired cysteine is available for intermolecular disulfide-bond formation. We probed for free cysteines in purified ADAMTS17-PCD, using an LC-MS/MS approach. Free and solvent-accessible cysteines in ADAMTS17-PCD were labeled with N-ethylmaleimide (NEM) prior to reduction, alkylation with iodoacetamide (IAA), and proteolytic digestion. LC-MS/MS identified peptides containing six cysteine residues that were NEM-modified, indicating at least partial availability for disulfide-bonding (Fig. 4d, Supplemental Fig. 3, Supplemental Table 3). Five of the NEM-modified cysteine residues were located in the catalytic domain and Cys 201 was located in the propeptide. The two other cysteine residues present in the propeptide (Cys 125 and Cys 144 ) were detected only in the IAA-modified form, indicating that they were not accessible to NEM or were oxidized, i.e., engaged in disulfide conditions (no DTT). Areas of the gel in the high MW range (> 100 kDa) reacting with anti-myc in the absence of DTT in medium and lysate are outlined with a bar. The asterisk indicates prominent anti-myc reactive ADAMTS17-PCD band in the lysate under non-reducing conditions. (f) Surface plasmon resonance indicates interaction of ADAMTS17-PCD to surface immobilized ADAMTS17-PCD in a dose dependent manner. The dissociation constant (K D ) was calculated assuming 1:1 binding. (g) Schematic of ADAMTS17-PCD truncation mutants (top panel). Western blot analysis of conditioned medium and cell lysate from HEK293 cells transiently transfected with ADAMTS17-PCD truncation mutants indicates anti-myc and anti-catalytic domain antibody reactive bands (yellow) of the predicted molecular weight in cell lysates for each mutant (bottom, right-hand panel). However, the entire propeptide is required for secretion (bottom, left-hand panel). bonds. To exclude the possibility that ADAMTS17-PCD oligomerization was an artifact arising from purification and concentration of the protein, we analyzed whether oligomerization was present in conditioned medium of cells expressing ADAMTS17 EA or ADAMTS17-PCD (Fig. 4e). Our data suggest that in the absence of DTT, high MW ADAMTS17 EA species reacting with anti-myc are present in the medium.
We used ADAMTS17-PCD as both the analyte and immobilized ligand in surface plasmon resonance (SPR) and found that it could bind to itself with a characteristic SPR binding curve and a dissociation constant (K D ) of 1.4 μ M (Fig. 4f). Taken together, the findings suggested that secreted PCD formed high MW multimers through a To elucidate the role of the ADAMTS17 propeptide in ADAMTS17 secretion, we sequentially truncated the propeptide at the second and third furin processing site respectively (Fig. 4g). When these N-terminally truncated ADAMTS17 constructs encoding either the catalytic domain alone or the catalytic domain and the disintegrin-like domain were transfected into HEK cells, they were detected in cell lysate as bands of the predicted molecular mass, but were not detected in the conditioned medium (Fig. 4g, yellow due to simultaneous reactivity with anti-myc, red, and anti-Cat, green). This suggests that inclusion of the propeptide is crucial for effective secretion of ADAMTS17. Tissue-specific Adamts17 mRNA expression during mouse embryonic development. The distribution of Adamts17 mRNA was analyzed by in-situ hybridization in sections from embryonic day (E) 16.5 old mouse embryos and from neonates (Fig. 5). In the eye at E16.5 (Fig. 5a), the strongest Adamts17 expression was detected in lens fiber cells at the lens equator, the site of the insertion of the ciliary zonule which is defective in ectopia lentis, in the non-pigmented epithelium of the ciliary body, which is the other attachment site of the zonule, and in the trabecular meshwork, which if blocked, can lead to glaucoma. We also detected expression in the blood vessels of the tunica vasculosa lentis and in the corneal stroma. In neonatal eyes (Fig. 5b), Adamts17 expression continued strongly in lens fiber cells, but was reduced in other ocular components expressing Adamts17 at E16.5. In long bones, Adamts17 expression was seen in the perichondrium surrounding the cartilage of long bone primordia, such as the metacarpals, metatarsals (Fig. 5c), and the femur (Fig. 5d), but not in the growth plate cartilage (Fig. 5d). In addition, we detected Adamts17 expression in developing tendons (Fig. 5c). In the intervertebral disc we observed Adamts17 expression in the nucleus pulposus (Fig. 5e). The strongest Adamts17 mRNA expression was observed diffusely in lung parenchyma, but was absent from the bronchial epithelium (Fig. 5f), and was shown to be specific using a control RNA probe (data not shown). In the skin, Adamts17 was expressed in the epidermal basal cell layer, the underlying dermal mesenchyme and intensely in developing hair follicles (Fig. 5g). In blood vessels, Adamts17 mRNA was restricted to smooth muscle cells of the tunica media, but was absent from the endothelium (Fig. 5h). Notably, fibrillins are expressed at each of these sites 25,27,39 . ADAMTS17 interacts with fibrillin-2 but not fibrillin-1 or fibronectin, and does not cleave fibrillin-1 or fibrillin-2. Because of the genetic association of ADAMTS17 with FBN1 revealed by WMS spectrum-causing mutations, we asked whether ADAMTS17 bound to or cleaved recombinant fibrillin-1 and -2 (Fig. 6a,b) 7,11,12,40 . Unexpectedly, analysis of ADAMTS17-PCD binding to recombinant fibrillin fragments using SPR showed that ADAMTS17-PCD bound to the N-and C-terminal halves of FBN2 (Fig. 6c, left-hand panels), but not to FBN1 or full-length cellular fibronectin (Fig. 6d, top panel). The calculated affinities for the ADAMTS17-PCD/FBN2 interactions were in the low μ M range, assuming a 1:1 stoichiometry, and the interaction was Ca 2+ -dependent (Fig. 6c, bottom left panel). We observed a similar selectivity and Ca 2+ -dependency for ADAMTS17-AD binding to FBN2 (Fig. 6c, right-hand panels). However, the calculated affinities for the ADAMTS17-AD/FBN2 interactions were in the low nM range indicating much stronger binding of ADAMTS17-AD to FBN2 compared to the ADAMTS17-PCD. ADAMTS17-AD did not bind to FBN1 or fibronectin, and did not self-interact nor bind to the ADAMTS17-PCD peptide (Fig. 6d, bottom panel). These findings suggest at least two FBN2 binding sites within ADAMTS17, with the ancillary domain conferring the strongest binding to FBN2.
To ask whether ADAMTS17 cleaved fibrillin-1 or fibrillin-2, we incubated recombinant fibrillin polypeptides (Fig. 6a,b) with purified ADAMTS17-PCD and visualized the proteins by Coomassie staining after SDS-PAGE (Fig. 7a). No fragmentation of any of the fibrillin peptides was evident. In this assay, we used purified ADAMTS17-PCD, which showed catalytic activity (Fig. 2), because full-length ADAMTS17 could not be purified owing to its autoproteolysis. However, because the greater binding affinity of the ancillary domain suggested that ADAMTS17 ancillary domain was required for the recognition and cleavage of the fibrillin substrates, we transfected cell lines stably expressing full-length fibrillin-1 (FBN1), FBN2-N, or FBN2-C with ADAMTS17, ADAMTS17 EA or ADAMTS17-PCD, respectively. Conditioned medium and cell lysate were analyzed by western blotting using antibodies detecting the fibrillins (green) and recombinant ADAMTS17 proteins (red) (Fig. 7b,c). We found no evidence for cleavage of fibrillin-1 or -2 by ADAMTS17, either as reduction of the signal intensity for the intact fibrillin bands (Fig. 7b,c, asterisks) or by appearance of fibrillin fragments. These results indicate that ADAMTS17 binds selectively to fibrillin-2, but does not cleave either fibrillin isoform.
Recombinant ADAMTS17 binds to microfibrils formed by cultured fibroblasts, with differential effects on fibrillin-1 and fibrillin-2 assembly into microfibrils. We investigated the ECM localization of ADAMTS17 in relation to fibrillin microfibrils assembled by cultured human dermal fibroblasts (HDFs) 28,29 . The cell number was reduced upon addition of ADAMTS17-PCD and ADAMTS17-AD to HDFs. Despite this, microfibrils were formed in these cultures, and we found that ADAMTS17-PCD (Fig. 8a) and ADAMTS17-AD (Fig. 8b) co-localized with FBN1 and FBN2 positive microfibrils. This was a surprising finding given that both ADAMTS17 peptides selectively bound to recombinant fibrillin-2, but not fibrillin-1 (Fig. 6c,d), but could be explained by the binding to heteropolymeric microfibrils containing both fibrillin-1 and fibrillin-2 41,42 . Since both fibrillins co-localize with fibronectin and require fibronectin for microfibril assembly in the ECM 41,43 , co-localization of the ADAMTS17 peptides with fibronectin was also seen (Fig. 8a,b).
We used mouse embryo fibroblasts (MEFs) as a second cell type from a different species to confirm the ADAMTS17 localization results obtained in HDFs, as well as to investigate potential colocalization with fibrillin-2 microfibrils, which are made reliably by MEFs 41 . After culturing MEFs in the presence of purified ADAMTS17-PCD, immunostaining detected recombinant ADAMTS17-PCD associated with fibrillin-1 (Fig. 9a,  top row). Surprisingly, few fibrillin-2 microfibrils were seen in ADAMTS17-PCD treated MEF cultures (Fig. 9a, middle row), but ADAMTS17-PCD was still localized as fibrils in the ECM. The reduction in fibrillin-2 could be explained by the stronger reduction of Fbn2 mRNA upon addition of ADAMTS17-PCD compared to Fbn1 mRNA (Fig. 9b). As with HDFs, recombinant ADAMTS17-PCD co-localized with fibronectin in MEF cultures, consistent with a demonstrated overlap between fibrillin-1 and fibronectin staining in cultured fibroblasts 43 . Considerable reduction of the number of adherent MEFs was observed after addition of ADAMTS17-AD (Fig. 9c), precluding microfibril staining.
To determine whether the apparent co-localization of ADAMTS17 with fibronectin represented localization to overlying microfibrils, we performed similar co-staining in rat aortic smooth muscle cells, which do not form detectable fibrillin-1 or fibrillin-2 microfibrils. ADAMTS17-PCD and ADAMTS17-AD showed punctate deposition in the ECM with only minimal co-localization of ADAMTS17-PCD or ADAMTS17-AD with fibronectin (Fig. 9d), suggesting, together with protein-protein interaction data, that ADAMTS17 did not bind to FN or FN fibrils.

Discussion
Determining the molecular mechanisms and substrates of orphan proteases, such as ADAMTS17, is crucial for understanding the pathogenesis of their associated genetic disorders. We have shown here that, i) ADAMTS17 undergoes extensive autoproteolysis, which occurs in trans, i.e. ADAMTS17 is its own substrate; ii) furin/PACE processing of ADAMTS17 is not required for autoproteolysis; iii) O-fucosylation of ADAMTS17 is required for its secretion; iv) ADAMTS17-PCD self-interacts via disulfide bond formation likely involving Cys201 in the propeptide; and, v) ADAMTS17 binds to fibrillin microfibrils, but does not cleave either of the fibrillin isoforms. These properties of ADAMTS17 identify it as a novel fibrillin binding protein and allow comparison of its properties with other ADAMTS proteins. ADAMTS1, 4, and 5 require furin-mediated excision of their propeptides for proteolytic activity toward versican and aggrecan, but ADAMTS9 does not 32,[44][45][46][47][48] . The ADAMTS13 propeptide is unusually short and structurally distinct harboring a single cysteine and the propeptide does not appear to regulate the enzymatic activity of ADAMTS13 49,50 . ADAMTS2, 3, and 14 have propeptides containing only two Cys residues 1 . Thus, ADAMTS17, along with ADAMTS9 and 13, is an unusual protease whose propeptide apparently lacks a role in zymogen latency. Rather, like ADAMTS9, the ADAMTS17 propeptide functions as an intramolecular chaperone, since its truncation prevents secretion 44 . Activation of matrix metalloproteases (MMPs) can be achieved by a cysteine switch 51 . Latency of MMPs is mediated by a free cysteine residue within the MMP propeptide that coordinates the zinc ion in the active site and blocks access to the substrate. MMP activation can be mediated by reducing agents or alkylation in vitro. A similar mechanism was also described for ADAM12, where either processing by furin or mutation of the free cysteine to alanine resulted in activation 52 . We identified a free cysteine residue (Cys 201 ) in the ADAMTS17 propeptide. However, furin processing was not required for ADAMTS17 activation and the amino acid sequence context of Cys 201 bears little resemblance to the consensus sequence described for MMPs 51 . Therefore, a cysteine switch mechanism similar to that in MMPs or some ADAMs is unlikely to mediate ADAMTS17 activation.
O-Fucosylation is a unique modification of TSR-containing proteins such as the ADAMTS proteins, all of which carry the required consensus sequence in one or more TSRs 53 . Suppression of either of the glycosyltransferases (POFUT2 or B3GLCT) acting sequentially in the O-fucosylation pathway resulted in impaired ADAMTS17 secretion, supporting its role in quality control during folding and secretion of TSR-containing proteins. Previously, it was shown that ADAMTS9, ADAMTS13, ADAMTSL1 and ADAMTSL2 were subjected to this quality control mechanism 34,35,54,55 . Interestingly, Peters Plus syndrome (PPS) (MIM #261540), which results from B3GLCT mutations is characterized by ocular anterior segment dysgenesis combined with short stature and brachydactyly, which are features of the WMS spectrum resulting from mutations in ADAMTS10, ADAMTS17,or FBN1 7,8,56,57 . PPS is thought to be a composite of phenotypes arising from impaired function of several B3GLCT target proteins as a result of their reduced secretion. On the basis of reduced ADAMTS17 secretion in the absence of B3GLCT and its ocular and skeletal expression and functions, ADAMTS17 could be one of the B3GLCT target proteins whose impaired secretion contributes to the PPS phenotype.
ADAMTS10, ADAMTS17, ADAMTSL2 and ADAMTSL4 have a robust genetic relationship with fibrillins [58][59][60] . ADAMTS10, whose mutations cause WMS, does not have an optimal furin processing consensus sequence and its zymogen is the predominantly secreted form 28,61 . Upon restoration of an optimal furin processing sequence by mutagenesis, ADAMTS10 cleaved recombinant fibrillin-1 in vitro 28 . In contrast, ADAMTS17 did not cleave fibrillin-1 or fibrillin-2, despite efficient processing by furin and autoproteolytic activity. Thus, ADAMTS10 and ADAMTS17 share the inability to cleave fibrillins, and ADAMTS10 behaves like ADAMTSL proteins, which constitutively lack proteolytic activity 62,63 . Unlike ADAMTS10, there was no evidence for ADAMTS17-mediated acceleration of microfibril formation. These distinct properties, as well as distinct expression profiles and intermolecular interactions may underlie some of the phenotypic differences observed between patients with ADAMTS10 or ADAMTS17 mutations, i.e. WMS and WMS-like syndrome, respectively 7,8 . ADAMTSL4, mutated in isolated ectopia lentis and ectopia lentis et pupillae, enhances fibrillin-1 assembly in cultured cells 29 . Strong expression of Adamtsl4 by the embryonic and juvenile mouse lens epithelium and loss of the lens-zonule connection in Adamtsl4 mutant mice suggests that ADAMTSL4 ensures anchorage of fibrillin microfibrils to the lens capsule 64 . ADAMTSL5 promoted assembly of fibrillin-1 microfibrils in vitro 65 , and ADAMTSL6 enhanced microfibril assembly upon transgenic overexpression in mice 30 . Neither is associated with a genetic condition in humans.
Despite the absence of binding to fibrillin-1 peptides, ADAMTS17 bound to fibrillin-1 containing microfibrils assembled by fibroblasts. One explanation could be that the ADAMTS17 binding site on fibrillin-1 microfibrils is contributed by two or more fibrillin-1 monomers and/or requires a specific spatial arrangement of fibrillin-1 within microfibrils rather than monomers in solution. In contrast, the apparent interference of ADAMTS17 with fibrillin-2 inclusion in microfibrils reflected suppression of Fbn2 transcription.
The outcome of ADAMTS17 transcriptional suppression on fibrillin-2 assembly was akin to that ascribed to ADAMTSL2, which is mutated in geleophysic dysplasia, another short-stature-brachydactyly condition. ADAMTSL2 binds both fibrillin-1 and -2, and its inactivation in mice results in increased fibrillin-2 microfibrils in the bronchial wall in the lung without altering Fbn2 gene expression, suggesting that ADAMTSL2 may participate in preferential inclusion of fibrillin-2 during microfibril assembly 27,66 . In contrast to ADAMTS17, ADAMTSL2 suppression of fibrillin-2 assembly did not involve transcriptional regulation. The mechanisms of transcriptional suppression of Fbn2 by ADAMTS17 are unclear, but could involve signaling from disrupted ECM via integrins, direct binding of ADAMTS17 to a cell surface receptor, or disturbance of a feedback loop for fibrillin-2 expression. The profound effect of the ADAMTS17-AD ancillary domain construct on cell viability could be explained by similar mechanisms, such as outside-in signaling. However, since HEK293 cells stably express recombinant ADAMTS17-AD without apparent ill effect, a general cytotoxicity of ADAMTS17-AD is unlikely. ADAMTS17-AD could also interfere with the assembly or function of ECM networks other than fibrillin microfibrils and promote detachment and cell death (anoikis) of MEFs.
The expression of ADAMTS17 mRNA was previously reported in the fetal brain, heart, lung, kidney and eye and in the adult brain, eye, and ovaries 7,67 . ADAMTS17 mRNA was detected in the synovium and is expressed in normal human breast tissue and invasive ductal breast carcinoma [68][69][70] . Using a sensitive in-situ hybridization technique, we showed that Adamts17 was expressed during mouse development in specific cell populations in a variety of tissues and was reduced after birth. This mimics the dynamic expression of Fbn2, which is predominantly expressed during embryonic development and downregulated after birth 24,25 . The data support a potential role for ADAMTS17 in the regulation of fibrillin-2 microfibril assembly or their functionalization. For example, during eye development, Fbn2 is strongly expressed in the non-pigmented epithelium of the ciliary body, but not in the differentiating lens epithelium 39 . Adamts17, however, is expressed in the lens epithelium, but not the non-pigmented epithelium of the developing ciliary body. We speculate that ADAMTS17 could prevent the premature assembly of microfibrils in the vicinity of the lens and may play a role in the proper development of a ciliary zonule. An Adamts17-deficient mouse model is currently not available to test these possibilities in vivo.
In summary, we present the first biochemical and functional characterization of ADAMTS17 and expand the repertoire of ADAMTS/ADAMTSL proteins modulating the formation and function of fibrillin microfibrils. The findings expand the growing body of work suggesting that ADAMTS proteases constitute an important class of microfibril-associated regulatory proteins that are directly involved in microfibril formation or that confer tissue-specific functions to microfibrils in diverse ways 62 .
Cell Culture and Transfection. Human embryonic kidney cells (HEK293F, HEK293T) and monkey kidney cells (COS-1) were purchased from ATCC (Manassas, VA) and maintained in DMEM supplemented with 10% FBS, 100 units/ml penicillin, 100 μ g/ml streptomycin in a 5% CO 2 atmosphere in a humidified incubator at 37 °C. HEK293F cells were seeded in 12-well plates and transfected with 1-2 μ g plasmid DNA using Lipofectamine 3000 according to the manufacturer's protocol. Stable clones were selected with geneticin (G418) for pcDNA based plasmids or zeocin (Invitrogen, San Diego, CA) for pSecTag 2B based plasmids and were analyzed for protein expression and secretion by western blotting. For O-fucosylation analysis, HEK293T cells were cultured Scientific RepoRts | 7:41871 | DOI: 10.1038/srep41871 in DMEM supplemented with 10% bovine calf serum in 6 cm cell culture dishes. 0.5-1 μ g plasmid DNA was transfected using polyethylenimine (PEI).

Protein Expression and Western Blotting. Stable expressing cells or transiently transfected cells were
analyzed for the presence of recombinant proteins in conditioned medium and the cell layer. Confluent cells were cultured for 48-72 h after transfection in serum-free DMEM medium in a 12-well tissue culture plate (BD Bioscience, San Jose, CA). For O-fucosylation analysis, transiently transfected HEK293T cells were cultured for 48 h in reduced serum medium (OptiMEM). Conditioned medium was harvested, cell debris removed by centrifugation, and equal volumes subjected to SDS-PAGE under reducing conditions. The cell layer was washed once with 1 ml PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 KPO 4 , 1.8 mM KH 2 PO 4 ) and lysed in 0.1% NP40, 0.01% SDS, 0.05% Na-deoxycholate in PBS (RIPA buffer). Cell lysates were incubated for 5 min at 4 °C with rotation end-over-end. The lysate was cleared by centrifugation (5 min, > 20,000 g, 4 °C) and equal volumes were subjected to SDS-PAGE under reducing conditions. Proteins were transferred onto PVDF membranes (Immobilon P for enhanced chemiluminescence, Immobilon F for infrared fluorescent detection, EMD Millipore, Billerica, MA) for 1.5 h at 70 V at 4 °C in 25 mM Tris, 192 mM glycine, 20% methanol buffer. The membrane was blocked with 5% (w/v) milk in TBS (10 mM Tris-HCl, pH 7.2, 0.15 mM NaCl) for 1 h at room temperature and primary antibodies were diluted in 5% (w/v) milk in TBS + 0.1% Tween 20 (TBST) and incubated overnight at 4 °C. The following antibodies and dilutions were used: anti-Myc (9E10, Invitrogen, 1:500-1:1,000), mouse monoclonal antibody to human ADAMTS17 (Abcam, Cambridge, MA; ab58099, 1:500), rabbit polyclonal antibody to human ADAMTS17 (N-term) (Abgent, San Diego, CA; 1:500), rabbit polyclonal antibody to human ADAMTS17 (catalytic domain) (Abcam, ab59828, 1:750). The polyclonal antibodies against human fibrillin-1 (C-terminus) (1:1,000) and fibrillin-2 (N-and C-terminus) (1:500) were described previously 42,71,72 . Membranes were washed three times with TBST for 5 min at RT and incubated with the corresponding secondary horseradish peroxidase coupled goat-anti-mouse or goat-anti-rabbit antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA; 1:2,500 in blocking buffer for enhanced chemiluminescence) in 5% milk in TBST or fluorophore-labeled IRDye goat-anti-mouse or goat-anti-rabbit secondary antibodies (LI-COR Biosciences, Lincoln, NE) in 5% milk in TBST + 0.01% SDS for 1 h at room temperature. For enhanced chemiluminescence, membranes were washed three times with TBST for 5 min at room temperature and the bound antibodies were visualized using enhanced chemiluminescence (ECL Prime, GE Healthcare, Piscataway, NJ, USA). For infrared detection, blots were washed three times with TBST for 5 min, one time with TBS for 5 min at room temperature, and were scanned wet on an Odyssey CLx scanner (LI-COR Biosciences, Lincoln, NE).
Identification of semi-tryptic peptides. Proteins were acetone-precipitated from conditioned medium containing recombinant ADAMTS17 or ADAMTS17 EA , solubilized in 6 M urea, disulfide-bonds were reduced with 200 mM dithiothreitol (DTT) and free cysteines alkylated with 200 mM iodoacetamide. The urea concentration was diluted to below 2 M and samples were trypsin digested overnight. The trypsin digests were subjected to a C18 cleanup step, evaporated to < 10 μ L in a Speedvac and resuspended in 1% acetic acid in a final volume of ~50 μ L for liquid-chromatography coupled mass-spectrometry (LC-MS/MS) analysis, using a Finnigan LTQ-Orbitrap Elite hybrid mass spectrometer system equipped with a Dionex 15 cm x 75 μ m Acclaim Pepmap C18 reversed phase capillary chromatography column (Thermo Scientific, Sunnyvale, CA). The identification of semi-tryptic ADAMTS17 peptides was performed by searching the MS/MS spectra specifically against the sequence of ADAMTS17 using the program Sequest (bundled into Proteome Discoverer v. 1.4.0) considering the formation of semi-tryptic peptides. Additional search criteria include a 10 ppm and 0.6 Da mass error for MS and MS/MS scans respectively, oxidation of methionine as a dynamic modification, and carbamidomethylation of cysteines as a static modification. All semi-tryptic ADAMTS17 peptides were high confidence identifications and were subjected to manual validation. The relative abundance of these semi-tryptic peptides was determined by plotting chromatograms for both the semi-tryptic and fully tryptic forms of each peptide, integrating these chromatographic peaks, and using the resulting peak areas to determine the peak area ratio: semi-tryptic/fully tryptic.
Identification of solvent-accessible cysteine residues. 30 μ g purified ADAMTS17-PCD in Tris-HCl buffer containing 6 M urea was treated with 200 mM N-ethylmaleimide (NEM) to label solvent accessible free cysteinyl residues, then reduced with 200 mM DTT, and alkylated with 200 mM iodoacetamide prior to protease digestion. Mass-spectrometry grade trypsin, chymotrypsin, and GluC (Roche Diagnostics, Indianapolis, IN) were added at a 1:20 (protease to protein) ratio, and the digest was incubated overnight at room temperature for completion. The protein digest was desalted using C18 SPE method (PepClean C-18 spin column) and reconstituted in 30 μ L 1% acetic acid for LC-MS/MS analysis as described above. The identification of cysteine-containing ADAMTS17 peptides was performed by searching the MS/MS spectra specifically against the sequence of ADAMTS17 using the program Sequest (bundled into Proteome Discoverer v. 1.4.0) considering the formation of both carbamidomethylated and NEM modified cysteine residues. All cysteine-containing peptides were high confidant identifications and were subjected to manual validation.
Cross-linking experiments. 10 μ g purified non-reduced or reduced and alkylated ADAMTS17-PCD was incubated with different amounts of bis(sulfosuccinimidyl)suberate (BS 3 ), an amine-to-amine cross linker with a 11.4 Å spacer for 30 min at RT. The reaction was terminated by quenching with 100 mM Tris-HCl, pH 7.5 and samples were separated on a 7.5% reducing SDS-PAGE followed by Coomassie Brilliant Blue staining.
RNA In-situ Hybridization. 16.5 day old mouse embryos were fixed in 4% paraformaldehyde overnight at 4 °C, processed and paraffin-embedded. Fresh 6 μ m sections were used for in-situ hybridization using RNAscope (Advanced Cell Diagnostics, Newark, CA) following the manufacturer's protocol. All steps requiring incubation at 40 °C were performed in the HybEZ Oven (Advanced Cell Diagnostics, Newark, CA). Tissue localization of the specific probe against mouse Adamts17 mRNA (#316441) was detected with the RNAscope 2.0 HD detection kit "RED". Probes against peptidylprolyl isomerase B (Cyclophylin B, Ppib) or bacterial dihydrodipicolinate reductase (DapB) mRNA (#313911, #310043) were used as positive and negative control, respectively. Sections were counterstained with hematoxylin prior to coverslipping with Cytoseal 60 (Electron Microscopy Science, Hatfield, PA, USA).

Protein-protein interaction analysis by surface plasmon resonance (SPR). Purified ADAMTS17-
PCD and ADAMTS17-AD were diluted in 10 mM sodium acetate pH 4.5 and immobilized on a BIAcore CM5 sensor chip (GE Healthcare, Piscataway, NJ, USA) with the amine coupling kit according to the manufacturer's instructions. For analysis in a BIAcore 3000 instrument (GE Healthcare, Piscataway, NJ, USA), 3300 resonance units for ADAMTS17-PCD and 2530 resonance units for ADAMTS17-AD were coupled to the chip. Kinetic analyses were performed at 25 °C in TBS including 2 mM CaCl 2 and 0.005% SP20 (running buffer) at a flow rate of 20 μ l/min. Purified recombinant fibrillin-1 (FBN1-N, FBN1-C) and fibrillin-2 (FBN2-N, FBN2-C) peptides were described previously 42,73 and were diluted in running buffer at the respective concentrations and injected in series through an uncoupled control flow cell followed by the flow cell with immobilized ADAMTS17-PCD or ADAMTS17-AD. Association was allowed for 3 min followed by a 5 min dissociation phase. To analyze calcium dependency of fibrillin binding to ADAMTS17-PCD or ADAMTS17-AD, FBN2-N and FBN2-C were diluted in TBS including 10 mM EDTA and 0.005% SP20 and the same buffer was used as running buffer. The CM5 chip was regenerated by injecting 60 μ l of 500 mM EDTA, pH 8.6 before analyzing a different peptide. The kinetic data were calculated using BIAevaluation 4.0.1 software (GE Healthcare, Piscataway, NJ, USA) assuming a 1:1 stoichiometry.
Digestion of FBN1 and FBN2 with ADAMTS17-PCD. 5 μ g of purified FBN1 or FBN2 peptides were incubated with purified ADAMTS17-PCD in PBS for 5 h at 37 °C. Samples were separated by 7.5% SDS-PAGE under reducing conditions and the gel was stained with Coomassie Brilliant Blue. ADAMTS17 microfibril binding using cultured fibroblasts. Wild-type (WT) mouse embryonic fibroblasts (MEF, 50000-75000 cells/well), human dermal fibroblasts (HDF) derived from foreskin explants, or rat aortic smooth muscle cells A7r5 (ATCC, Manassas, VA) were cultured in the presence of 50 μ g purified Scientific RepoRts | 7:41871 | DOI: 10.1038/srep41871 ADAMTS17-PCD, ADAMTS17-AD, or PBS buffer control in 8-well chamber slides (BD Bioscience, San Jose, CA). Recombinant ADAMTS17-PCD or ADAMTS17-AD was added 24 h after seeding the cells and cells were cultured for up to five additional days. Cells were fixed in ice-cold 70% methanol/30% acetone for 5 min at room temperature. After blocking with 10% NGS in PBS (blocking buffer) for 1 h at room temperature, cells were incubated with the following antibody combinations, diluted in blocking buffer, for 2 h at room temperature: anti-Myc (1:300) + anti-fibrillin-1-C (1:500), anti-Myc (1:300) + anti-fibrillin-2-N (1:300), or anti-Myc (1:300) + anti-fibronectin (pAB 2033, 1:500, EMD Millipore, Billerica, MA) 71,72 . Cells were washed three times with PBS for 5 min at room temperature and incubated with the corresponding Alexa-Fluor labeled secondary goat-anti-mouse or goat-anti-rabbit antibodies (diluted 1:350 in blocking buffer, Jackson ImmunoResearch Laboratories, West Grove, PA) for 1.5 h at room temperature. Cells were washed three times with PBS for 5 min at room temperature and mounted with ProLong Gold Antifade Reagent with DAPI. Slides were imaged using an Olympus BX51 upright fluorescent microscope equipped with a CCD camera (Leica Microsystems). Post-acquisition image analysis and counting of cell nuclei was performed with Corel PHOTO-PAINT or ImageJ (National Institutes of Health, Bethesda, MD) 74 .