Supramolecular Organization of the α121-α565 Collagen IV Network*

Background: The supramolecular organization of the collagen IV network containing α5 and α6 chains is unknown. Results: Sulfilimine-cross-linked α1-α5 and α2-α6 NC1 dimers compose the α121–565 hexamer. Conclusion: The α121-α565 network is formed by a heterotypic interaction between α121 and α565 protomers. Significance: Sulfilimine bonds cross-link all three collagen IV networks: α121, α345, and α121-α565 networks. Collagen IV is a family of 6 chains (α1-α6), that form triple-helical protomers that assemble into supramolecular networks. Two distinct networks with chain compositions of α121 and α345 have been established. These oligomerize into separate α121 and α345 networks by a homotypic interaction through their trimeric noncollagenous (NC1) domains, forming α121 and α345 NC1 hexamers, respectively. These are stabilized by novel sulfilimine (SN) cross-links, a covalent cross-link that forms between Met93 and Hyl211 at the trimer-trimer interface. A third network with a composition of α1256 has been proposed, but its supramolecular organization has not been established. In this study we investigated the supramolecular organization of this network by determining the chain identity of sulfilimine-cross-linked NC1 domains derived from the α1256 NC1 hexamer. High resolution mass spectrometry analyses of peptides revealed that sulfilimine bonds specifically cross-link α1 to α5 and α2 to α6 NC1 domains, thus providing the spatial orientation between interacting α121 and α565 trimers. Using this information, we constructed a three-dimensional homology model in which the α565 trimer shows a good chemical and structural complementarity to the α121 trimer. Our studies provide the first chemical evidence for an α565 protomer and its heterotypic interaction with the α121 protomer. Moreover, our findings, in conjunction with our previous studies, establish that the six collagen IV chains are organized into three canonical protomers α121, α345, and α565 forming three distinct networks: α121, α345, and α121-α565, each of which is stabilized by sulfilimine bonds between their C-terminal NC1 domains.

Collagen IV is an evolutionarily conserved protein of basement membranes (BMs), 3 a specialized layer of extracellular matrix underlying all epithelia (1,2). As an important structural component of BMs, one essential function of collagen IV is to maintain tissue architecture during embryonic development, remodeling, and repair (3). Collagen IV also serves as a ligand for integrin cell surface receptors influencing cell behavior including adhesion, migration, and differentiation (4). Importantly, several genetic and acquired diseases affect type IV collagen and thus impair BM function leading to deadly diseases, including Alport syndrome (5,6), cerebral hemorrhages (7), porencephaly (8), and Goodpasture disease (6).
Collagen IV molecules form triple-helical protomers that further assemble into supramolecular networks stabilized by novel sulfilimine (OSANO) bonds (9). Sulfilimine bonds cross-link the C-terminal non-collagenous (NC1) domains of interacting protomers, forming a globular structure known as the NC1 hexamer. Catalysis of these cross-links in vivo is promoted by the BM-embedded enzyme peroxidasin via a novel hypobromous acid-mediated mechanism (10). Further studies of sulfilimine bond formation revealed that bromine, a required cofactor of peroxidasin, is an essential trace element for successful embryogenesis and development in Drosophila (11). This unique cross-link is known to be an evolutionarily conserved feature of collagen IV networks, arising at the divergence of Porifera and Cnidaria over 500 Mya (1,9). Thus, the machinery that assembles sulfilimine bonds is a primordial innovation of collagen IV networks essential for organogenesis and tissue evolution.
In mammals, collagen IV is present as a family of six distinct genes encoding the ␣1-␣6 chains. Of all possible combinations, only three collagen IV networks of defined ␣-chain compositions have been observed: the ubiquitous ␣121 (12) and the tissue-restricted ␣345 (13,14) and ␣1256 networks (15,16). The crystal structure of the ␣121 NC1 hexamer demonstrated that the ␣121 network results from a homotypic interaction between ␣121 trimeric protomers (17). Because sulfilimine cross-links bridge the trimer-trimer interface within the NC1 hexamer, forming NC1 dimers, the location of these cross-links also confirmed the relative orientation of interacting protomers. A similar approach using sulfilimine-cross-linked NC1 dimers was used to elucidate the quaternary structure of the ␣345 hexamer (18), which established that the ␣345 network is formed by a homotypic interaction between ␣345 protomers. In contrast, although analyses of BM from X-linked Alport patients, who suffer from an inherited form of kidney disease caused by mutations in the COL4A5 gene, provided clues linking ␣5 to ␣6 chain assembly (19), the organization of the ␣1256 network is not fully understood. Initial characterization of ␣1256 NC1 hexamers from aorta smooth muscle BM suggested a heterotypic interaction between a putative ␣565 protomer and a classical ␣121 protomer. However, inherent limitations of the immunoblotting techniques used to characterize NC1 dimers could not unambiguously establish the organization of the network (16).
The new discovery of sulfilimine bonds, its mechanism of assembly, and the availability of the ␣121 hexamer crystal structure open new possibilities to gain further insight into the supramolecular organization of the ␣1256 network. In this study we aimed to determine if sulfilimine cross-linking occurs in aorta smooth muscle BMs and use this chemical information to elucidate the quaternary structure of the ␣1256 NC1 hexamer. We used high-resolution mass spectrometry (MS) to determine the identity of sulfilimine-cross-linked NC1 dimers resulting from the interaction between protomers. Our results indicate that only ␣1-␣5 and ␣2-␣6 sulfilimine-cross-linked NC1 dimers are present in the ␣1256 hexamer. These findings unambiguously establish the quaternary organization of the NC1 hexamer revealing a heterotypic interaction between ␣121 and ␣565 protomers to form the ␣1256 network.

EXPERIMENTAL PROCEDURES
Isolation of ␣1256 NC1 Hexamer from Aorta-Aorta NC1 hexamers were prepared as described previously by Kahsai et al. (20). Briefly, NC1 hexamers were solubilized from bovine aorta tissue by the collagenase digestion method and purified on DE-52 cellulose and Sephacryl S-300 columns.
Mass Spectrometry Analysis of NC1 Dimer Peptides Derived from ␣1256 NC1 Hexamer-The fraction containing the ␣1256 hexamer (Mono S peak 2) was denatured and reduced by mixing equal volumes of protein solution with a 0.4 M Tris-HCl, pH 7.5, buffer containing 8 M guanidine-HCl and 50 mM dithiothreitol. After heating samples in a boiling water bath for 10 min, the proteins were alkylated with 50 mM iodoacetamide in the dark for 30 min at room temperature. After ethanol precipitation at Ϫ20°C for 2 h, NC1 subunits were resuspended in 0.1 M ammonium bicarbonate, pH 7.5, and digested with trypsin overnight. Subsequently, tryptic peptides were fractionated on a Superdex Peptide column (GE Healthcare), and polypeptides from 3000 to 6500 Da were collected, lyophilized, and stored at Ϫ20°C before MS analysis.
For mass spectrometric analysis of sulfilimine-cross-linked peptides, dry samples were reconstituted in 0.1% formic acid and loaded onto a capillary reverse phase analytical column (360 m outer diameter ϫ 100 m inner diameter) using an Eksigent NanoLC Ultra HPLC and autosampler. The analytical column was packed with 20 cm of C18 reverse phase material (Jupiter, 3 m beads, 300 Å, Phenomenex) equipped with a laser-pulled emitter tip. Peptides were gradient-eluted at a flow rate of 500 nanoliters/min, and the mobile phase solvents consisted of 0.1% formic acid, 99.9% water (solvent A) and 0.1% formic acid, 99.9% acetonitrile (solvent B). A 90-min gradient was performed and consisted of the following: 0 -10 min, 2% B; 10 -50 min, 2-35% B; 50 -60 min, 35-90% B; 60 -65 min, 90% B; 65-70 min, 90 -2% B; 70 -90 min, 2% B. Upon gradientelution, peptides were mass analyzed on a LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific, San Jose, CA), equipped with a nanoelectrospray ionization source. The instrument was operated using a data-dependent method where full scan (m/z 400 -2000) spectra were acquired with the Orbitrap as the mass analyzer (resolution 60,000), and the three most abundant ions in each MS scan were selected for collisioninduced dissociation (CID) fragmentation in the LTQ Velos ion trap, which were followed by data-dependent MS 3 analysis of the three most abundant ions in each MS 2 scan. An isolation width of 2 m/z, activation time of 30 ms, and 35% normalized collision energy were used to generate MS 2 and MS 3 spectra. For MS 2 scans, a minimum threshold of 500 was used to trigger data-dependent spectra, whereas a threshold of 500 or 100 was used to trigger MS 3 spectra. Dynamic exclusion was enabled, with an exclusion duration time of 20 ms.
Sulfilimine Cross-linking Analysis of NC1 Peptide Mass Spectra Data-Mass spectra analyses were performed both manually with Xcalibur 2.1 and computationally with the proteomic analysis software SEQUEST (22,23) and/or MyriMatch (24). For manual searches, theoretical peptide masses for NC1 peptides containing either Met 93 or Lys 211 were generated using GPMAW software (Lighthouse Data). Sulfilimine-cross-linked peptide masses were calculated by subtracting 2 hydrogen atoms from the sum of the masses of Met 93 -and Lys 211 -containing peptides. Although we hypothesized that the ␣1256 hexamer results from the interaction of a ␣121 trimer with a ␣565 trimer, forming ␣1-␣5 and ␣2-␣6 NC1 dimers, we did not limit our analyses to these combinations and also considered other possible combinatorial arrangements. Sulfilimine-crosslinked peptide complexes were manually searched for using the QualBrowser of Xcalibur 2.1 software.
For sulfilimine-cross-linked peptides not identified manually, we took advantage of the characteristic CID fragmentation of sulfilimine-cross-linked peptides (9), which generates modified Met and Lys residues, to design an alternative unbiased bioinformatics approach to analyze mass spectra. Proteomic analysis software Scaffold 3 (Proteome Software, Portland, OR) was used to compile SEQUEST and/or MyriMatch searches of the bovine subset of the Uniref100 database for sulfilimine cross-link-derived peptides. The MyriMatch algorithm (24) was run on a two six-core Intel Xeon processor HP Proliant DL160 G6 server running Windows server 2008 R2 enterprise using the Bumbershoot suite (25). The search was adapted for identifying peptides with the following amino acid modifications: variable, Ϫ48.0034 on Met, ϩ45.9877 or ϩ61.9826 on Lys, and oxidation of methionine (ϩ15.9949) as well as fixed carbamidomethylation of cysteine (ϩ57.021). The proteomic analysis software ScanRanker was used to assess the MS 3 spectral quality of peptide sequences containing Met 93 or Lys 211 (26). Spectra were annotated and validated with the utility program IonMatcher (24).

Isolation and Characterization of ␣1256 Hexamer from
Bovine Aorta-The ␣1256 hexamer was isolated from bovine aorta tissue to provide NC1 peptides amenable to analysis by mass spectrometry. Digestion of the aorta tissue with bacterial collagenase routinely yielded about 1.5 mg of NC1 hexamers per 50 g of wet tissue (Fig. 1). In accord with previous results (16), immunoblot analysis using chain-specific monoclonal antibodies confirmed the presence of the ␣1, ␣2, ␣5, and ␣6 NC1 domains in the purified aorta NC1 hexamer (Fig. 1B). Notably, ␣3 and ␣4 NC1 domains were not detected. Monoclonal antibodies detected NC1 dimeric subunits, suggesting the presence of covalent cross-linking between any combination of ␣1, ␣2, ␣5, and ␣6 NC1 monomeric subunits.
In smooth muscle tissues two distinct subpopulations of NC1 hexamers of different ␣-chain compositions were observed: the ␣121 and ␣1256 hexamers (16). To isolate the ␣1256 hexamer for MS analysis, the preparation of aorta NC1 hexamers was fractionated on a cation-exchange Mono S chromatography column ( Fig. 2A). Elution of bound proteins with increasing concentrations of sodium chloride yielded three distinct chromatographic peaks. Immunoblot analysis of each peak using chain-specific monoclonal antibodies showed the ␣5 and ␣6 NC1 subunits eluting most abundantly in peaks 1 and 2, with the ␣1 and ␣2 NC1 subunits also found in these peaks (Fig. 2B). Peak 3 is predominantly composed of ␣1 and ␣2 NC1 subunits, whereas the staining for ␣5 and ␣6 NC1 domains was minimal. These data indicate that the ␣121 hexamer elutes in peak 3, whereas the ␣1256 hexamer elutes in peaks 1 and 2. Peak 1 and 2 were considered indistinguishable by immunoblot analyses, but because of its greater abundance, peak 2 was used for MS analyses.
Mass Spectrometry Analysis of Sulfilimine-Cross-linked NC1 Dimer Peptides-NC1 hexamers are formed by the interaction of the C-terminal NC1 domain trimers of two protomers (17). The large, planar interface formed by the interacting trimers is stabilized by sulfilimine bonds (OSANO) in which the sulfur atom of Met 93 from an NC1 monomer forms a covalent bond with the ⑀-nitrogen from Lys 211 (or hydroxylysine) of an adjacent NC1 monomer (9,29). Because of the spatial alignment of NC1 domains across the trimer-trimer interface, the occur-FIGURE 3. A, experimental strategy for cross-linking analysis of NC1 dimers of the ␣1256 hexamer. The schematic shows that the ␣12156 hexamer putatively contains two different types of NC1 dimers, one composed of ␣1 (yellow) and ␣5 (green) NC1 domains and another composed of ␣2 (purple) and ␣6 (orange) NC1 domains. The NC1 monomers are presumed to be cross-linked by sulfilimine bonds (black bars), which were previously found in NC1 dimers of the ␣121 hexamer (9). The sulfilimine cross-link is formed between Met 93 from an NC1 domain of one trimer and a Hyl 211 from an NC1 domain of the opposite trimer (Lys 211 is post-translationally modified to Hyl 211 , noted as K OH ). A total of six sulfilimine cross-links (two per NC1 dimer) can be formed at the trimer-trimer interface. Dissociation of ␣1256 hexamer yields ␣1-␣5 and ␣2-␣6 NC1 dimers. Trypsin digestion of ␣1-␣5 NC1 dimers releases two peptide complexes (T-3958 and T-4943), each containing the sulfilimine cross-link. Likewise, trypsin digestion of ␣2-␣6 NC1 dimers releases two peptide complexes (T-5307 and T-4393). B, fractionation of ␣1256 hexamer tryptic peptides. Tryptic peptides derived from Mono S peak 2 ( Fig. 2A) were fractionated on a Superdex Peptide column (GE Healthcare). The shaded area contained peptides ranging from 3000 to 6500 Da and was collected for mass spectrometry analysis. Absorbance (milliabsorbance units (mAU) was monitored at 214 nm.
rence of sulfilimine bonds results in the formation of three covalently linked NC1 dimers (Fig. 3). Therefore, the composition of such sulfilimine-cross-linked NC1 dimers reflects the way in which triple-helical protomers interact to form collagen IV networks.
In this study we used a MS-based strategy to determine the occurrence of sulfilimine bonds by identifying cross-linked peptides derived from NC1 dimeric subunits. Thus, as previously suggested (16), we hypothesized that this hexamer is formed from a ␣121 trimer and a ␣565 trimer. This arrangement may generate up to four distinct sulfilimine cross-linking sites: two sites between ␣1 and ␣5 NC1 domains and two sites between ␣2 and ␣6 NC1 domains (Fig. 3A). Theoretical masses for ␣1-␣5 and ␣2-␣6 sulfilimine-cross-linked NC1 peptides as well as other possible combinations were calculated as indicated under "Experimental Procedures." To identify these sul-filimine-cross-linked peptides, Mono S peak 2, containing the ␣1256 NC1 hexamer (Fig. 2), was digested with trypsin, and the resulting peptides were fractionated by gel filtration (Fig. 3B). Peptides ranging from 3000 to 6500 Da in size were collected and analyzed in a high resolution LTQ Orbitrap Velos in a datadependent manner as previously described (9,18). For convenience, sulfilimine-cross-linked peptides identified by manual and bioinformatics analyses of MS data described under "Experimental Procedures" are summarized and presented by the type of NC1 dimer from which they originate.
Mass Spectrometry Analyses of ␣1-␣5 NC1 Sulfilimine-Crosslinked Peptides-The ␣1-Hyl 211 Ϫ␣5-Met 93 complex was found in the full scan mass spectrum containing the multiply protonated ions (ϩ4, ϩ5, ϩ6) that represent the T-3958 complex (Fig. 4A). The difference between observed and calculated masses for this peptide complex reflects the loss of 2 hydrogen and m/z 730.7 (ϩ2) represent the ␣5-Met 93 -and ␣1-Hyl 211 -containing peptides, which deviate from the expected theoretical peptide masses by Ϫ48 and ϩ46 atomic mass units, respectively. These modifications correspond to the olefin fragment derived from Met 93 and a methylsulfenamide fragment derived from Hyl 211 that result from the characteristic gas-phase Cope elimination previously described for sulfilimine-cross-linked peptides (9). B, full-scan MS 1 and MS 2 spectra for tryptic peptide complex T-4943 composed of a ␣1 NC1 peptide covalently cross-linked to a ␣5 NC1 peptide by a sulfilimine bond between ␣1-Met 93 and ␣5-Hyl 211 . Ions with m/z 989.6782 and 824.9004 correspond to the multiply protonated (ϩ5) and (ϩ6) ions of the T-4943 peptide complex, respectively. CID fragmentation of m/z 824.9004 (ϩ6) yielded a MS 2 profile with two predominant peptide fragments: m/z 893.6 (ϩ4) and m/z 689.1 (ϩ2) represent the ␣1-Met 93 -and ␣5-Hyl 211 -containing peptides, which deviate from the expected theoretical peptide masses by Ϫ48 and ϩ46 atomic mass units, respectively, as a result of sulfilimine bond fragmentation (9). atoms upon sulfilimine bond formation (Table 1). To confirm the presence of a sulfilimine bond connecting the two peptides, the m/z 792.5850 (ϩ5) ion was selected for MS 2 fragmentation by CID. As observed in the MS 2 spectrum (Fig. 4A), CID fragmentation yielded two predominant ions: m/z 834.3 (ϩ3) and m/z 730.7 (ϩ2), which represent the ␣5-Met 93 -and ␣1-Hyl 211containing peptides but deviate from the expected theoretical peptide masses by Ϫ48 and ϩ46 atomic mass units, respectively. These peptide modifications correspond to the olefin fragment derived from Met 93 and the methylsulfenamide fragment derived from Hyl 211 that result from the characteristic gas-phase Cope elimination previously described for sulfilimine-cross-linked peptides (supplemental Figs. S1 and S2) (9). The T-3958 peptide complex demonstrates hydroxylation of ␣1-Lys 211 and establishes the presence of the sulfilimine bond in one of the two cross-linking sites within the ␣1-␣5 NC1 dimer.
Similarly for the ␣1-Met 93 Ϫ␣5-Hyl 211 NC1 peptide complex, the full scan mass spectrum contained multiply protonated ions (ϩ5 and ϩ6) representing the T-4943 complex (Fig.  4B). The difference between the observed and calculated masses for this peptide complex reflects the loss of 2 hydrogen atoms upon sulfilimine bond formation (Table 1). To confirm the presence of a sulfilimine bond connecting the two peptides, the m/z 824.9004 (ϩ6) ion was selected for MS 2 fragmentation by CID. As observed in the MS 2 spectrum (Fig. 4B), CID fragmentation yielded two predominant ions; m/z 689.1 (ϩ2) and m/z 893.6 (ϩ4) ions represent the ␣5-Hyl 211 -and ␣1-Met 93containing peptides but deviate from the expected theoretical peptide masses by ϩ46 and Ϫ48 atomic mass units and, respectively, which is consistent with a MS 2 fragmentation profile characteristic of peptides cross-linked by a sulfilimine bond (9). Furthermore, MS 3 fragmentation confirmed the NC1 peptide sequences and localized the peptide modifications to Met 93 and Hyl 211 (supplemental Figs. S3 and S4). Mass spectrometric analyses of the T-4943 peptide complex demonstrates hydroxylation of ␣5-Lys 211 and establishes the presence of the sulfilimine bond in the second of the two ␣1-␣5 NC1 dimer crosslinking sites.
Mass Spectrometry Analyses of ␣2-␣6 NC1 Sulfilimine-Crosslinked Peptides-Although the hydroxylation of ␣6-Lys 211 had never been reported, we included this modification in our mass spectra searches because this residue lies embedded within a conserved hydroxylation motif (30). The ␣2-Met 93 Ϫ␣6-Hyl 211 NC1 peptide complex was found in the full scan mass spectrum containing the multiply protonated ions (ϩ5, ϩ6, ϩ7) that represent the T-5307 complex (Fig. 5A). The difference between the observed and calculated masses reflects the loss of two hydrogen atoms upon sulfilimine bond formation ( Table 1) Similarly for the ␣6-Met 93 Ϫ␣2-Hyl 211 NC1 peptide complex, the full scan MS1 mass spectrum contained multiply protonated ions (ϩ3, ϩ4, ϩ5) representing the T-4393 complex (Fig. 5B). The difference between the observed and calculated mass for this peptide complex reflects the loss of 2 hydrogen atoms upon sulfilimine bond formation (Table 1). To confirm the presence of a sulfilimine cross-link, the m/z 1099.3181 (ϩ4) ion was selected for MS 2 fragmentation by CID. As observed in the MS 2 spectrum (Fig. 5B), CID fragmentation yielded two predominant ions; m/z 731.7 (ϩ2) and m/z 1468.4 (ϩ2) ions represent the ␣2-Hyl 211 -and ␣6-Met 93 -containing peptides but deviate from the expected theoretical peptide masses by ϩ46 and Ϫ48 atomic mass units, respectively, which is consistent with a MS 2 fragmentation profile characteristic of peptides cross-linked by a sulfilimine bond (9). Furthermore, MS 3 fragmentation confirmed the NC1 peptide sequences and localized the modifications to Met 93 and Hyl 211 (supplemental Figs. S7 and S8). Collectively, these results demonstrate hydroxylation of ␣2-Lys 211 and establish the presence of the sulfilimine bond in the second of two cross-linking sites in the ␣2-␣6 NC1 dimer,

Mass spectrometry identified sulfilimine-cross-linked NC1 peptide complexes
Tryptic peptide complexes (T-XXXX) containing Met 93 and Hyl 211 residues were identified with mass spectrometry and compared to theoretical peptide masses.
* The theoretical peptide masses were calculated from the sum of the masses of the respective NC1 tryptic peptides containing Met 93 and Hyl 211 . The observed peptide masses were calculated from the monoisotopic ions observed for each multiply protonated precursor ion detected. Calculated peptide masses were averaged and compared to the theoretical mass. A mass difference (theoretical Ϫ observed) equivalent to two hydrogen atoms was found in each case.
which along with the four ␣1-␣5 sulfilimine bonds establishes the presence of a total of six sulfilimine bonds per ␣121-␣565 NC1 hexamer. Taken together, our MS studies of NC1 heterodimers support our hypothesis that the ␣1256 hexamer is formed by the covalent interaction of ␣121 trimer and the ␣565 trimer.
Homology Modeling of the ␣121-␣565 Hexamer-Identification of sulfilimine-cross-linked NC1 dimers establishes the spatial arrangement and connectivity of the NC1 domains at the trimer-trimer interface and provides the framework to construct a three-dimensional homology model of the ␣121-␣565 hexamer quaternary structure. The amino acid sequences of ␣1, ␣2, ␣5, and ␣6 NC1 domains were threaded onto the ␣121 NC1 hexamer crystal structure template, as described under "Experimental Procedures." As shown in Fig. 6, 4 the overall structure of the homology model for ␣1256 hexamer is very similar to the already established ␣121 and ␣345 hexamer. The model illustrates the interaction of a ␣121 trimer with a ␣565 trimer in which the orientation of one with respect to the other is determined by the alignment of ␣1-␣5 and ␣2-␣6 NC1 dimers revealed by the MS analysis of tryptic peptides identified in this study. Henceforth, we propose to change the nomenclature of the ␣1256 hexamer to ␣121-␣565 hexamer to better represent the heterotypic interaction of ␣121 and ␣565 protomers.

DISCUSSION
Sulfilimine-cross-linked NC1 hexamers provide insightful information regarding the composition and organization of the different collagen IV networks. Previous research described the composition and suggested a quaternary structure for the ␣121-␣565 hexamer, but it could not unequivocally establish covalent linkages and the organization of NC1 domains (16). In this study, we overcame such limitations by using high resolution MS to identify sulfilimine-cross-linked ␣1-␣5 and ␣2-␣6 NC1 dimers. Construction of a three-dimensional model by homology modeling using this chemical information provides the framework for understanding the spatial organization of NC1 domains within the hexamer. Thus, the quaternary structure of the ␣121-␣565 hexamer shown in Fig. 6 defines the supramolecular organization of this collagen IV network.
Initially, the identification of ␣5 and ␣6 peptides in the MS data of unfractionated aorta NC1 hexamers was confounded by overcrowding of the spectra by peptides derived from the more prevalent ␣1 and ␣2 NC1 domains (supplemental Figs. S9 and S10), as expected based upon previous studies determining the aorta BM abundance of the ␣121 hexamer as 88% and the ␣121-␣565 hexamer as 12% (16). Therefore, use of the Mono S cationexchange column to separate the different types of NC1 hexamers was essential for isolating ␣121-␣565 hexamers, from which trypsin digestion produced peptides derived from NC1 dimers identifiable by MS analyses (Fig. 2).
MS analyses identified sulfilimine cross-links at the trimertrimer interface of the ␣121-␣565 hexamer. Sulfilimine bonds have previously been shown between Met 93 and Hyl 211 of NC1 dimers in the ␣121 and ␣345 collagen IV networks, with the exception of the sulfilimine bond between ␣4-␣4 NC1 dimer forming between Met 93 and Lys 211 (9). In this study MS analyses show that sulfilimine bonds form between Met 93 and Hyl 211 of ␣1-␣5 and ␣2-␣6 dimers. Notably, the hydroxylation state of the ␣6-Lys 211 was confirmed by MS fragmentation of the T-5307 complex, which revealed an extra 16 atomic mass units localized on the Lys 211 of the ␣6 NC1 peptide (Fig. 5A and supplemental Fig. S5). Hydroxylation of ␣6-Lys 211 is consistent with the ␣6 NC1 domain containing the XK(A/S/G) sequence motif that triggers hydroxylation of lysyl residues by lysyl hydroxylase (30). These findings together with previous studies in the ␣121 and ␣345 hexamers (18,29) establish that Lys 211 of all six NC1 domains form sulfilimine cross-links, and all except ␣4-Lys 211 appear predominantly as Hyl 211 in BMs.
Collagen IV plays a role in maintaining the vascular integrity of the aorta, a smooth muscle tissue, which relies on the mechanical properties of the extracellular matrix for proper pulsatile blood flow (31). Indeed, the distinct expression of the ␣5 and ␣6 chains has been correlated with BMs of tubular structures that are constantly exposed to mechanical stress or tensile strength (15,32). Two collagen IV networks (␣121 and ␣121-␣565) coexist in the smooth muscle BM of the aorta medial layer. Interestingly, it has been reported that some Alport patients not only endure nephropathy (6) but also have an increased susceptibility to aortic aneurysms (33,34), possibly due to mutations in the COL4A5 gene leading to loss of the ␣121-␣565 network of the BM surrounding aortic smooth muscle cells. It is conceivable that the absence of this network may weaken the aortic wall by either reducing its mechanical strength and/or by interrupting cell-matrix interactions important for normal smooth muscle cell development. Thus, Alport syndrome, which previously emphasized the important role of the ␣345 network in molecular ultrafiltration in the kidney, may now also help establish the role of the ␣121-␣565 network in the vascular integrity of the aorta.
The discovery of sulfilimine bonds in the ␣121-␣565 network inspires further exploration to gain a better understanding of the function and regulation of these cross-links in vascular tissues. As we have recently shown in Drosophila (11), the concerted activity of peroxidasin, bromide, and oxidant is an absolute requirement for the formation of both sulfilimine bonds as well as functional collagen IV networks capable of providing mechanical stability to BM and tissues. For instance, the occurrence of sulfilimine bonds demonstrated in this study suggests that all of these elements converge in the aorta BM. In fact, recent reports have shown that peroxidasin, also known as vascular peroxidase (VPO1), is expressed in the vascular wall of arterioles both in the endothelium and medial layer (35). Conditions that may decrease peroxidasin sulfilimine cross-linking activity may result in weakened collagen IV networks, which may lead to abnormalities of the vascular wall such as aortic aneurysm. On the other hand, it is also plausible, for example, that deregulation of peroxidasin could lead to excessive crosslinking of collagen IV network leading to its accumulation in the smooth muscle BM, which may contribute to pathological states such as arteriosclerosis.
In conclusion, the quaternary structures of the three NC1 hexamers representing the three known collagen IV networks are depicted in Fig. 6. Although the ␣121 and ␣345 hexamers are formed by a homotypic interaction between protomers, the ␣121-␣565 hexamer is formed by a heterotypic interaction of FIGURE 6. NC1 hexamer structures of the three known mammalian collagen IV networks. The illustration shows that the overall structure of the ␣121-␣565 NC1 hexamer is similar to those of the ␣121 and ␣345 hexamer structures determined in previous studies. The sulfilimine bond formed between methionine 93 and lysine (or hydroxylysine) 211 of adjacent NC1 trimers is shown in white for all the ␣121, the ␣345, and the ␣121-␣565 NC1 hexamers. Unlike the ␣121 and ␣345 NC1 hexamers, the ␣121-␣565 hexameric quaternary structure is formed by a heterotypic interaction between a ␣121 trimer and a ␣565 trimer. The alignment of one trimer with respect to the other is defined by the cross-linked ␣1-␣5 and ␣2-␣6 NC1 dimers. Molecular surfaces images were produced using UCSF Chimera (27).
␣121 and ␣565 protomers. In addition, this study provides chemical evidence supporting the quaternary structure of the ␣565 protomer illustrated by homology modeling, where the ␣5 and ␣6 NC1 domains assemble into a heterotrimer with a structural organization analogous to that of the ␣121 and ␣345 heterotrimers. All three triple-helical protomers are the fundamental building blocks of collagen IV networks, and the elucidation of their structures has provided insight into BM function and explained the molecular basis for human diseases including Alport syndrome (6). Taken together, our findings in conjunction with our previous studies establish that the six collagen IV chains are organized into three canonical protomers ␣121, ␣345, and ␣565 forming three distinct networks: ␣121, ␣345, and ␣121-␣565, each of which is stabilized by sulfilimine bonds between their C-terminal NC1 domains.