Crystal Structure of NC1 domains: Structural Basis for Type IV collagen
Assembly in Basement Membranes
Munirathinam Sundaramoorthy*, Muthuraman Meiyappan§, Parvin Todd, and Billy G. Hudson.
Department of Biochemistry and Molecular Biology, University of Kansas Medical Center, 3901
Rainbow Boulevard, Kansas City, KS 66160-7421.
* Corresponding author
Phone: (913) 588-6574
Fax: (913) 588-7440
Email: [email protected]
§Present address:
Department of Chemistry and Chemical Engineering, Cornell University, Ithaca, NY 14853-
1301
Keywords: 3D domain swapping, Br-MAD, type IV collagen, NC1 domain, network assembly,
disulfide crosslinks
Running title: Structure of Type IV Collagen NC1 Domain
1
Copyright 2002 by The American Society for Biochemistry and Molecular Biology, Inc.
JBC Papers in Press. Published on April 22, 2002 as Manuscript M201740200 by guest on A
pril 5, 2018http://w
ww
.jbc.org/D
ownloaded from
SUMMARY
Type IV collagen, which is present in all metazoan, exists as a family of six homologous α(IV)
chains, α1-α6, in mammals. The six chains assemble into three different triple helical protomers
and self-associate as three distinct networks. The network underlies all epithelia as a component
of basement membranes (BM), which play important roles in cell adhesion, growth,
differentiation, tissue repair and molecular ultrafiltration. The specificity for both protomer and
network assembly is governed by amino acid sequences of the C-terminal noncollagenous (NC1)
domain of each chain. In this study, the structural basis for protomer and network assembly was
investigated by determining the crystal structure of the ubiquitous [(α1)2.α2]2 NC1 hexamer of
bovine lens capsule basement membrane (LBM) at 2.0 Å resolution. The NC1 monomer folds
into a novel tertiary structure. The (α1)2.α2 trimer is organized through the unique 3D domain
swapping interactions. The differences in the primary sequences of the hypervariable region
manifest in different secondary structures, which determine the chain specificity at the monomer-
monomer interfaces. The trimer-trimer interface is stabilized by the extensive hydrophobic and
hydrophilic interactions without a need for disulfide cross-linking.
2
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
INTRODUCTION
Basement membranes (BM)1 or basal laminae are a heterogeneous and highly specialized form
of extracellular matrix found in all animal phyla (1). These thin amorphous sheet-like structures
underlie epithelia and endothelia, and surround muscle, peripheral nerve, and adipose cells. BMs
have several important functions such as: (a) providing the structural support for cells and
compartmentalizing tissues, (b) acting as the selective filtration barriers for macromolecules, and
(c) providing an anchorage for the neighboring cells as well as stimuli for cell growth, migration,
and differentiation. The morphology, structure, and molecular composition of BMs vary with the
origin of tissue. Nevertheless, all BMs are known to contain four major components: type IV
collagen, laminin, entactin/nidogen, and sulfated proteoglycan (2-4). The intermolecular
interactions between various components determine the ultrastructure and function of BMs.
Type IV collagen, which provides scaffold for the binding of other protein components in BMs,
belongs to the large family of collagens comprised of more than 20 subtypes that form highly
organized supramolecular assemblies (5,6). All collagen molecules consist of three
polypeptides, called α chains, each containing a large central triple helix-forming domain of Gly-
Xaa-Yaa repeats, flanked by small N- and C-terminal globular domains. Some types of triple
helical protomers contain genetically identical α chains forming homotrimers, while others
contain two or three different α chains forming heterotrimers (7). The chain composition of a
given collagen molecule is dictated by two factors: (1) expression of specific chains in a given
tissue and (2) specific association of chains. The chains first associate through a series of non-
covalent interactions between the C-terminal noncollagenous domains that provide correct
alignment and registration for the nucleation of triple-helix formation (7-11). This early
3
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
molecular recognition event is common for all types of collagen, which are assembled from
highly homologous, but genetically distinct procollagen chains (12). Even though the exact
mechanism by which the C-propeptides initially associate is not fully understood, it is widely
accepted that they contain molecular recognition sequences that select specific chains for the
protomer assembly (7,13). For example, type I and III collagens are assembled in a type-
specific manner despite both being synthesized in skin fibroblasts and having high level of
identity in their procollagen sequences. While type I collagen exists as a heterotrimer of two
proα1(I) and one proα2(I) chains, type III collagen is an obligate homotrimer comprising of
three proα1(III) chains, further underscoring the level of discrimination in the assembly of
specific types of collagens (14-16). Similarly, distinct protomers of type IV collagen are
assembled through molecular recognition of the noncollagenous (NC1) domains (17,18). In
contrast to the fibril-forming collagens, in which the C-terminal noncollagenous domain
(propeptides) is processed to permit fibril-formation, the NC1 domains of type IV collagen are
retained associating tail-to-tail to form hexamer in the network assembly.
Mammalian type IV collagen is a family of six homologous α chains, designated α1-α6 (19).
Each chain is characterized by a long collagenous domain of ~1400 residues of Gly-Xaa-Yaa
repeats, interrupted by ~20 short noncollagenous sequences, an NC1 domain of ~230 residues at
the C-terminus, and a small noncollagenous sequence at the N-terminus (Fig. 1a) (20). The six
chains assemble into three distinct protomers, differing in chain composition, that are assembled
by the association of the NC1 domains, followed by triple-helix formation of the collagenous
domains (Fig. 1b). These protomers self-associate to form three distinct networks, α1.α2, (as
illustrated in Fig. 1c) α3.α4.α5 and α1.α2-α5.α6 networks (18). The networks are assembled by 4
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
the dimerization at the C-terminus through NC1 domain interactions and by the tetramerization
of the 7S domain at the N-terminus (Fig. 1c). The specificity for both the protomer and the
network assembly is governed by molecular recognition sequences encoded within the NC1
domains (17,18). Thus, the NC1 domains play a fundamental molecular function in the assembly
of chain-specific networks by (a) the initial alignment and selection of chains for protomer
assembly and (b) the connection and selection of protomers for network assembly. The NC1
domains of the six α chains can be divided as αl-like (α1, α3, & α5) and α2-like (α2, α4 & α6)
subfamilies based on the sequence identity (21,22). Further, each NC1 domain consists of two
homologous subdomains with ~35% sequence identity.
The collagen IV network is present in all metazoan (Fig. 2a&b). Collagen type IV is an ancient
(>500 million years old) scaffold with evolutionarily conserved structural features including
overall protomer dimensions, multiple noncollagenous interruptions of the triple-helix, and the
NC1 hexamer structure at the protomer-protomer interface. The protomers of Hydra vulgaris
(23) appears to contain only one type of α chain, but a second type has not been ruled out. The
protomers of Pseudocorticium jarrei (24), Caeneorhabditis elegans (25,26), Drosophila
melanogaster (27,28), and sea urchin (29,30) contain two chain types, forming an α1.α2
network. In mammals, the α1.α2 network is distributed in all tissues whereas the α3.α4.α5- and
the α1.α2-α5.α6-networks have a restricted distribution (17,18,31).
The type IV collagen network is essential for the tissue development and maintenance of
function. In hydra, exogenously added NC1 domains of collagen IV cause a perturbation of ECM
formation and subsequent blockage of morphogenesis (32) and, functional antisense studies 5
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
showed inhibition of collagen IV translation causing subsequent blockage of head regeneration.
In C. elegans, mutations in either of the two α chains cause embryonic lethality (33,34). In
humans beings, mutations in the α3, α4, and α5 chains cause the loss of the α3.α4.α5 network
(17) and leads to Alport syndrome, characterized by progressive loss of sensorial hearing and
renal function (35,36). In mice, gene knockout of the α3 chain causes the loss of the same
network and leads to auditory dysfunction and renal failure (37,38). Thus, the α1.α2 network
appears to play an essential role in the development of all tissues whereas, the α3.α4.α5- and
α1.α2.α5.α6-networks confer specialized functions and stability.
In the present study, the structural basis for collagen IV network assembly, with respect to
generic interactions of the NC1 domain that govern the network assembly and specific
interactions that govern the selection of chains, was investigated. This was accomplished by
crystallization of the ubiquitous [(α1)2.α2]2 NC1 hexamer, isolated from bovine lens capsule
basement membrane (LBM), and determination of its three dimensional structure at 2.0 Å
resolution.
6
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
EXPERIMENTAL PROCEDURES
Protein Purification and Crystallization. The [(α1)2.α2]2 NC1 hexamer was isolated from
bovine eye lenses purchased from Pel-Freeze Biologicals (Rogers, AR) following the previously
published procedure (39). Briefly, LBM was prepared by sonication of the lenses in the
presence of 1 M NaCl and protease inhibitors (40). To cleave the NC1 domain from the full-
length type IV collagen, the LBM preparation was digested with bacterial collagenase at 37° C.
The NC1 hexamer is purified using DE-52 and S-300 column chromatography.
Initial crystallization screening with commercial sparse matrix kits (Hampton Research, Laguna
Niguel, CA) was carried out using concentrated protein (10 mg/ml) and hanging drop vapor
diffusion method. The LBM NC1 crystals grow as small clusters overnight in 10% (w/v) PEG
20K, 0.1 M bicine buffer (pH 9.0) at room temperature. Diffraction quality crystals were grown
using microseeding procedures under similar conditions with lower protein concentration. The
crystals belong to monoclinic P21 space group with unit cell dimensions a = 129.41 Å, b =
143.87 Å, c = 162.92 Å, and β = 91.3º at room temperature and four hexamers in the asymmetric
unit. Cryocooling of the crystals in 25% 2,4-methyl pentanediol (MPD) or glycerol results in the
shrinkage of the unit cell (a = 127.16 Å, b = 139.57Å; c = 160.20 Å; β = 91.3º).
Structure Determination and Refinement. Initial heavy atom soaks were carried out at the
crystallization pH and later switched to neutral pH with phosphate buffer. The NC1 crystals
soaked in a synthetic mother liquor containing 2mM LuCl3 or K2PtCl6 transform the lattice to a
smaller unit cell of dimensions a = 80.07 Å, b = 137.96 Å, c = 127.13 Å, β = 90.3° and two
hexamers in the asymmetric unit. The crystals were routinely transformed to the new form by
7
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
soaking in 2 mM LuCl3 overnight and used for further heavy atom soakings. Multiwavelength
anomalous diffraction (MAD) data sets were collected at peak, inflection and two remote
wavelengths using a single crystal soaked in 0.5 M KBr for 1 min and flash-frozen in a cold N2
stream (Table 1). The heavy atom soak screens were carried out at beamlines 1-5 and 9-2 of
Stanford Syncrhortron Radiation Laboratory (SSRL) and beamline X8C of National Synchrotron
Light Source (NSLS) at Brookhaven National Laboratory. The Br-MAD data sets used in this
study were collected at SSRL and processed using DENZO and SCALEPACK programs of HKL
suite (41). The Br- sites were located using SOLVE program (42) and 33 highest peaks (> 6σ)
were used for phasing the reflections at 2.2 Å resolution. The resulting phases were improved by
solvent flattening using RESOLVE (43) with 45% solvent content and the electron density map
was calculated using FFT program of CCP4 suite (44). Polypeptides of two α1 chains and one
α2 chain (chains A-C) were traced using the TOM FRODO graphics program (45). The
complete asymmetric unit was generated using NCS relations obtained from Br- sites—first the
second trimer (chains D-F) was generated to complete one hexamer and then the second hexamer
(chains G-L) was generated from the first hexamer.
The 2.0 Å data set collected at 0.8856 Å (λ4) was used for model refinement using CNS program
(46) and 5% of the data were set aside for monitoring Rfree. The initial model was subjected to
rigid body refinement using reflections in 30.0-3.0 Å resolution range (Rcryst = 0.361 and Rfree =
0.364) followed by simulated annealing refinement in 10.0-2.5 Å resolution range (Rcryst = 0.287
and Rfree = 0.326). The resolution was slowly extended to 2.0 Å in several iterative cycles of
model building and refinement of the positional and thermal parameters. During the final rounds
of refinement, solvent molecules (water and glycerol) were added in steps using 2Fo-Fc and Fo-
8
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Fc maps and hydrogen bonding criteria. Multiple conformers of few side chains were modeled
and, Br- and Lu3+ ions were included at a later stage of the refinement. The structure was
analyzed using SETOR(47) and GRASP(48) graphics software packages and various utility
programs in CCP4 suite. The hexamer interface was analyzed using HBPLUS(49) and protein-
protein interaction web server (http://www.biochem.ucl.ac.uk/bsm/PP/server/).
Sequences of α1 and α2 Chains. The map was originally fitted using human α1 and α2 NC1
sequences as the guide. The amino acid sequences deduced from the electron density maps were
later compared with those predicted from cDNA sequences. The complete primary structure of
bovine α1 NC1 was obtained from the NCBI’s Expressed Sequence Tags (EST) databank. The
bovine N-terminal sequence previously reported (50) was used as the query sequence in the
search, which returned a 454-bp cDNA clone from Bos taurus (BE589226). This clone encodes
the query sequence and in-frame N- and C-terminal amino acid sequences that are highly
homologous to human α1 NC1 sequence. The in-frame C-terminal amino acid sequence was
used in a second search, which identified a 444-bp overlapping cDNA clone (BE846087)
encoding additional C-terminal sequence. This sequence, when used in the final search,
identified another 544-bp overlapping cDNA clone (BE845675) encoding the remaining C-
terminal sequence. A similar search using the N-terminal sequences of human and mouse α2
NC1 failed to find any homologous bovine cDNA clone. However, when C-terminal sequences
of human and mouse α2 NC1 domains were used a 461-bp cDNA clone (BF039765) was found.
This clone contains the C-terminal sequence and the stop codon, which are more than 95 %
sequence identical to the human sequence. Using a similar strategy described for α1 NC1, two
more overlapping clones BM956598 and AV613094 were found. The three overlapping cDNA 9
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
sequences encode the bovine α2 NC1 sequence, except for the 36 amino acid residues at the N-
terminus. To determine the missing sequence, poly-A mRNA from bovine kidney was subjected
to one step reverse transcription coupled to PCR (Life Technologies) using ON-A2-1c [5’-
AAGCACCAAAGTGGCCCCGGACTG-3’] representing 3’ untranslatable sequence present in
BF039765 as the reverse primer, and ON-A2-1m [5’-AGAGGTGGCGTGTCTGCTGTTCC-3’],
an oligonucleotide representing a highly conserved region in human and mouse α2 mRNA as the
forward primer. The cDNA thus obtained was characterized by direct nucleotide sequencing.
RESULTS AND DISCUSSION Structure Determination and Overview. The bovine LBM NC1 hexamer, composed of α1 and
α2 chains, crystallizes in a monoclinic space group P21 (A-form) with four hexamers per
asymmetric unit. This is different from the crystal forms reported for the mouse EHS tumor NC1
(51) and human placenta NC1 hexamers (52), which crystallized with two hexamers and one
hexamer in the asymmetric unit, respectively. The intensity statistics of the preliminary
diffraction data suggested the presence of pseudo-translation symmetry along the c axis in the
LBM NC1 crystals. Extensive search for heavy atom derivatives using soaking experiments was
not successful. However, crystals soaked in LuCl3 at pH 7.0 transformed the lattice to a smaller
unit cell as a result of pseudo-translation symmetry becoming crystallographic translation in the
same space group with only two hexamers in the asymmetric unit (B-form). MAD data of the
crystals soaked in LuCl3 did not provide useful phase information, probably due to weak binding
as suggested by the high B-factors for potential metal sites. However, we took advantage of the
smaller unit cell for further heavy atom screening, including the newly suggested short-soaking
strategy with halides (53,54). The LuCl3-soaked B-form crystal structure was determined at 2.0
10
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Å resolution by the MAD method using Br- as the anomalous scatterer combined with solvent
flattening. The data collection, phasing and refinement statistics are shown in Table 1.
The map was initially fitted using human α1 and α2 NC1 sequences (Fig. 2) and more than 95%
of the residues fit the experimental electron density map (Fig. 3). With the presence of eight
copies of α1 chain and four copies of α2 chain in the crystallographic asymmetric unit,
sequencing using the 2.0 Å resolution electron density was unambiguous. These sequences were
later found to be identical to those predicted from cDNA sequences. The sequences are
numbered so that the residue after the last Gly-Xaa-Yaa repeat of the collagenous region is
counted as the first residue in both α chains. The 12 chains in two hexamers have been assigned
chain IDs A-L in the order of α1, α1 and α2 in each trimer. The map shows disorder for 5-6
residues at N- and two residues at C-termini of all the chains. The final model includes two
hexamers, four Lu3+ ions, 36 Br- ions, eight glycerol molecules and 1139 water oxygens. The
final Rcryst and Rfree of the refinement are 0.168 and 0.197 respectively. More than 90% of the
residues are within the favorable regions in the Ramachandran map. Arg76 and Ser148 of the
first α1 chain, Ser148 of the second α1 chain and Arg75, Glu95 and Ala145 of the α2 chain in
each trimer lie outside the allowed region. All these residues are at the interfaces of the chains in
the trimer or hexamer. Ser178/Ala145 introduce a bulge in a β-turn and Gly97/Glu95 are at the
end of a β-strand where a 310 helix begins, explaining the strained conformation. The
significance of Arg76/75 is discussed in a later section. Only a handful of residues are in
multiple conformations. The two hexamers in the asymmetric unit are similar with no apparent
differences due to crystal contacts. The hexamer comprising chains A-F is used to describe the
model. 11
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
The overall structures of the trimer and hexamer are illustrated in Fig. 4. The monomers within a
trimer are related by a pseudo 3-fold symmetry coinciding with the triple helix axis (“polar
axis”) and the two trimers are related by a 2-fold non-crystallographic symmetry (NCS) axis at
the interface (“equatorial plane”).
Monomer Topology: The NC1 monomer folds into a novel tertiary structure with predominantly
β−strands as predicted by our earlier study using multiple sequence alignment (22) (Fig. 2 and
4). The two α1 chains in the trimer are identical and the α2 chain has a similar overall structure.
The Cα atoms of 214 matching residues in one of the α1 chains and the α2 chain superimpose
with an RMS deviation of 0.9 Å (Fig. 5a). Each chain can be divided into two homologous
subdomains, N- and C-subdomains. The two subdomains fold in a similar topology and Cα atoms
of 96 matching residues of two subdomains of the α1 chain superimpose with an RMS deviation
of 1.0 Å (Fig. 5b). The 12 invariant cysteine residues form six disulfides, three in each
subdomain, at conserved positions (Fig. 2 and 5). The major difference between the two
subdomains occurs at the regions encompassing Pro86-Pro95 in the N-subdomain and Ile196-
Thr209 in the C-subdomain, which are least conserved within the family of six human
sequences. Each subdomain has two β-sheets—a three-strand anti-parallel sheet (I & I’) close to
the triple helical junction and a six-strand anti-parallel sheet (II & II’) close to the hexamer
interface (Fig. 6). There are three short 310 helices in the α1 chains and two in the α2 chain. The
β-sheet I is formed by the three non-contiguous strands (β1, β10 and β2) of the sequence
belonging to the first half of the polypeptide. However, in the β-sheet II, only four strands (β4,
12
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
β3, β8, and β9) belong to the first half of the sequence and the remaining two strands (β6’ and
β7’) form a part of the second half of the sequence. Thus, a β-hairpin structure from the second
half of the sequence swaps into the N-subdomain to form a six-strand β-sheet. The two halves of
the polypeptide being topologically similar, the region in the C-subdomain corresponding to the
six-strand β-sheet in the N-subdomain lacks two strands to form a similar β-sheet in the isolated
monomer structure. Similarly, β6-β7 hairpin in the N-terminal sequence corresponding to β6’-
β7’ hairpin in the C-terminal sequence that is involved in the domain swapping interaction
extends out in the monomer structure. These two features form the basis for the trimer
organization described in the next section.
13
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Trimer Organization: Two chains of α1 and one chain of α2 form the trimer structure with a
pseudo 3-fold molecular symmetry (Fig. 4a). Since each chain is made up of topologically
similar subdomains there is even a pseudo 6-fold symmetry. The trimer structure is
approximately cone-shaped with a base diameter of about 65 Å and a hollow core of about 12-
14.0 Å inner diameter. This is about the same of as the diameter of the collagen triple helix, with
N-termini of all three chains coming together at the vertex of the cone where the triple helical
collagenous domain links with the NC1 domain. The trimer is tightly packed through several
interchain hydrophobic and hydrogen bonding interactions (Table 2). Residues of five segments
in the N-subdomain of one chain make contact with those of seven segments in the C-subdomain
of the second chain. The most important interactions are confined to one N-subdomain segment
and two C-subdomain segments (Fig. 6). There are two levels of interactions between the
monomers, one essential for the “generic trimer” assembly and the other dictating the chain
specificity.
Generic Trimer: At the first level, the monomers intertwine with each other to form the trimer
through 3D domain swapping interaction (Fig. 4a, 6 and 7a) (55). A six-strand β−sheet (II’) is
formed in the C-subdomain from strands of two different α chains similar to the β−sheet II in the
N-subdomain formed from the strands in two halves of the same chain. These β−sheets are
indistinguishable in the α1 and α2 chains. Thus, there are six β-sheets (II/II’), one in each of the
six subdomains, forming the close-ended 3D domain swapping interactions in the NC1 trimer
structure. Each of these six-strand β-sheets is formed by four strands (β4/4’, β3/3’, β8/8’, β9/9’)
in one half of the sequence and the remaining two strands (β6’/6, β7’/7) are contributed by the
other half of the same chain or adjacent chain. The amino acid sequences of all the strands with 14
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
an exception of β9 are highly conserved in α chains within and across the species (Fig. 2). The
six topologically similar β-sheets formed in cyclical fashion give the pseudo 6-fold symmetry
appearance for the trimer (Fig. 7a). In each of the β-sheets, the outermost strand (β9/β9’) lies
on the surface parallel to the equatorial plane of the hexamer interface forming a part of the outer
ring and the innermost strand (β4/β4’) runs nearly parallel to the polar axis or pseudo 3-fold axis
in the core. The angle between these two strands within each sheet is about 75° giving it a right-
handed twist. The β4/β4’ strands from all the six β-sheets form a parallel β barrel-like core of
about 14 Å diameter even though there are no backbone hydrogen bonds between them (Fig. 7a).
However, these core strands are stabilized by backbone-side chain hydrogen bonds either
directly or mediated through solvent molecules. The β4/4’-strands have a mixture of
hydrophobic and hydrophilic residues with the former pointing to the core and the latter pointing
towards the adjacent strand. Interestingly, the β4-strands contain long chain hydrophilic amino
acids so that they form more direct hydrogen bonds with the backbone atoms of the β4’ strand of
the neighboring chain indicating stronger interchain interactions. The interactions between β4’
and β4 within a chain are mainly mediated through solvent molecules. Thus, the six-strand β-
sheets are essential structural components in the organization of the generic trimer structure
through 3D domain swapping interactions and compact β barrel-like core structure. However,
they may play only a limited role in the chain specific assembly of the trimer.
Chain Specificity in the Trimer Structure: The sequence of the loop connecting β8’- and β9’-
strands is the most variable region in all the six human α chains. This hypervariability in the
primary sequences manifests into different secondary structures in the α1 and α2 chains in the
15
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
crystal structure. Whereas it forms a short 310 helix (g2’) in all the α1 chains (Glu200-Lys204),
the corresponding region in the α2 chain (Ser198-Gln200) adopts an extended conformation
(βp’) and pairs with the extended structure (βp, Phe57-Thr59) in the adjacent α1B chain to form
a short parallel β−sheet (Fig. 7b). This is the only parallel β-sheet in the entire structure, which is
predominantly made up of β−strands. The sequence of the βp is highly conserved in all the six α
chains and forms the same extended structure in the α2 chain also, even though it doesn’t have a
partner in α1A chain to form the parallel β-sheet.
These additional main chain hydrogen bond interactions between the two chains are found only
at the α1Β−α2 interface, but not in the α2-α1A and α1A-α1B interfaces due to the presence of
the 310 helical structure in the α1 chains rather than the extended structure present in the α2
chain. Besides this difference in the secondary structural elements in the three interfaces, there
are also differences in the main chain – side chain and side chain – side chain interactions (Fig.
7b). This is also reflected in different ratios of polar to non-polar atoms at the three interfaces
(Table2).
The side chain of Lys56(α1B) is sandwiched between the backbone of the loop preceding the
parallel β-sheet in α2 chain and the contiguous bonds of backbone and side chain of Gln120(α2).
In this tightly locked position Lys56(α1B) assumes a linear conformation to form two strong
hydrogen bonds with carbonyl of Ile194(α2) and carboxyl of Asp121(α2) and two more weak
interactions with carbonyls of Gln120(α2) and Glu196(α2). The region corresponding to the
parallel β-sheet of α2 chain is a 310 helix in the α1 chains spanning longer sequence. Hence, in
16
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
the α1A-α1B interface, Lys56(α1A) is not quite parallel to the backbone bonds, which provides
more room for this lysine to adopt a different rotamer conformation forming only a weak
hydrogen bond with the carbonyl oxygen of Ile196(α1B). This may also be influenced by the
presence of hydrophobic Thr124 in the α1 chains in place of hydrophilic Asp121 in the α2. At
the α2-α1A interface Arg55(α2) is docked in similar position as Lys56 of α1 chains in other two
interfaces with one strong hydrogen bond interaction with carbonyl of Ile196(α1A). Other
differences in amino acid sequences including Arg55/Ala54 and Gly98/Glu95 make differences
in hydrogen bonding patterns at the interfaces.
The α1B-α2 interface has a maximum number of contact residues and a highest proportion of
hydrophilic atoms and contain more hydrogen bonds (Table 2). On the other hand, the buried
surface area is largest for the α1A-α1B interface. From these observations, it is evident that
α1B-α2 interface is formed predominantly through hydrogen bonding interactions and the α1A-
α1B interface is stabilized by more hydrophobic forces.
In addition to the specific interactions at the interfaces, packing considerations may also play an
important role in determining the chain stoichiometry. Even though the α1 and α2 chains fold in
a similar tertiary structure with a low RMS deviation, the relative orientation of the two
subdomains in each chain is different near the triple helical junction (Fig. 5a). The region
encompassing Thr13-Tyr30 of N-subdomain in α2 chain is farther from its equivalent region
Asp121-Tyr138 of C-subdomain compared to the relative orientations of similar regions in the
α1 structure. The larger width of the α2 structure near the triple helical junction results in
17
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
serious steric clashes when packed into a hypothetical α2-homotrimer. However, it is possible to
accommodate three α1 chains in a hypothetical homotrimer albeit with weaker interactions.
Hexamer Assembly: The type IV collagen triple helix, once formed in the endoplasmic lumen, is
secreted into the extracellular space where it assembles into supramolecular network through N-
and C-terminal associations. The NC1 domains play a dominant role in this assembly
determining the C-terminal dimeric association. In this section we describe the forces that
influence such an assembly as observed in the crystal structure and provide a rationale for the
specificity of protomers in the network assembly.
The foot-ball shaped hexamer is made up of two identical trimers, each containing two α1 chains
and one α2 chain as described in the previous section (Fig. 4b). Each protomer formed by the
tightly intertwined trimer is considered as a single entity so that the dimeric protomer complex
can be analyzed similar to several homodimeric protein complexes (45). We have determined
several parameters defining the interface to evaluate the strength of interactions between the two
trimers to understand the hexamer assembly in the type IV collagen network (Table 3). Like
most homodimers, the two NC1 trimers are related by a 2-fold NCS axis lying in the equatorial
plane and perpendicular to the pseduo 3-fold axis (Fig. 4). This symmetry constraint may be
partly influenced by a few differences in the interface residues of α1-like and α2-like sequences
in addition to more efficient packing. The interface is formed by the nearly flat surfaces of the
two trimers with an RMS deviation of 1.9 Å for all the interface atoms from the mean plane. This
is significantly lower than the average planarity value of 3.5 Å for the 32 homodimers (45). The
interface formed by six segments each of the three monomers with a total of 109 residues per
18
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
trimer is nearly circular with the major and minor axial lengths of the mean plane measuring
approximately 69 and 61 Å, respectively. This flat circular interface covers about 4400 Å2 of
solvent accessible area per trimer, which correlates with the observation of larger molecules
having larger interfaces (56). Such a large interface facilitates strong interaction between the
trimers involving both hydrophobic and hydrophilic residues. The polar (45.5%) and non-polar
atoms (54.5%) in the interface are nearly in equal proportions underscoring the importance of
both types of interactions in the hexamer stabilization.
The discussion thus far focused on the overall nature of trimer-trimer interface. Next, the
interactions between the individual chains are analyzed in more detail. Each monomer of one
trimer makes contact with two monomers of the other trimer, designated as the “major” and
“minor” contacts based on the extent of the contact area and the number of hydrogen bonds. The
two monomers making the major contact are referred to as a “dimer” in a similar sense as is used
in the denaturation experiments of hexamers (57). The 2-fold NCS between the two trimers
results in only one “homodimer” formed by the two α1 chains and remaining two are
“heterodimers” formed by the α1 and α2 chains. A 120° rotation of one trimer with respect to
the other about the pseudo 3-fold axis will result in “all homodimers” structure. Why such
arrangement is not possible can be explained mainly on symmetry consideration. In such a
scenario, the 2-fold symmetry will no longer be perfect, which might result in less efficient
packing with possibly fewer interactions and some unfavorable contacts. In order to understand
the complex hydrogen bonding interactions at the interface, it is essential to look into the
interactions of each monomer with its “major” and “minor” interacting partners. The complexity
presented even at this level may be simplified further by breaking down the interactions to three
19
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
regions in the structure: “core” and “outer” regions of “major” contact and the major-minor
junction.
The two 6-strand β-sheets, II and II’, formed by the 3D domain swapping interactions play as
crucial role in the formation of hexamer assembly as in the case of trimer organization. The
interface is populated with β-turns connecting β3/β3’-β4/β4’ in the core. These turns along with
the remaining strands of the β-sheets II/II’ position a large number of conserved residues for
extensive hydrogen bonding interactions between the two trimers. The core β-turns (two per
monomer contributed by the two equivalent subdomains) in the two trimers pack in staggered
configuration such that each turn in one trimer contacts with two turns in the other trimer. The
turns in the N-subdomains are of type I’/III’ containing hydrophilic amino acids in the second
(Asn39/Gln38) and third positions (Glu40/39). The C-subdomain turns are of type II in the α1
chains and type II’ in the α2 chains with small hydrophobic amino acids, Ala149/146-
Gly150/147-Ala151/Asp148, with Ala149 α1 or Asp148 of α2 introducing a β-bulge. Thus, the
hydrophilic side chains of the turns in the N-subdomain participate in hydrogen bonds and the
hydrophobic residues of turns in C-subdomain pack through hydrophobic interactions as well as
stacking interaction of peptide planes (Fig. 8a). Whereas Asn39(Gln38) side chain in the N-
subdomain makes hydrogen bond with backbone amide in C-subdomain turn, the conserved
Glu40(39) penetrates between the N- and C-subdomains of a monomer chain in the other trimer
to make a hydrogen bond with the side chain of conserved Gln37(36). The Glu40 residues in the
α1-α1 dimer make a strong hydrogen bond with each other that is missing in α1-α2 dimers. The
packing of the turns and the side chains appear to be tight at the core interface in CPK models
20
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
indicating strong van der Waals interactions in additions to the obvious hydrogen bonding
interactions.
The sequence variability preceding Arg179(177) influences the number of potential H bonds in
the α1-α2 interface. The interactions in the outer region involve the highly conserved loop
connecting β7 and β8, and β7’-β8’ sheet. In the α1-α1 major interface, five contiguous carbonyl
oxygens of highly conserved Ala74-Asp78 in one chain form hydrogen bonds with side chains
Asn77, Arg179, and Tyr185 of the other chain in symmetrical sets (Fig. 8b). These side chains
are also conserved in both α1 and α2 chains. Moreover, Arg76/75 of α1A/α2 chains are outliers
in the Ramachandran map, whose backbone carbonyls participate in the hydrogen bonding.
However, insertion of Ala176 and substitution of Asn174 in the α2 sequence alters the
orientation of the conserved Asn78 and Arg177 residues, which results in a fewer hydrogen
bonds in the α1-α2 interface. Arg76 of the α1B chain need not adopt a strained conformation
due to the lack of possibility for its carbonyl to participate in hydrogen bonding.
21
The major-minor junction is where two chains from one trimer contact two chains of the other
trimer. There are two types of junctions, one involving three α1 chains and one α2 chain, and
the other involving two each of α1 and α2 chains. The hydrogen bonding pattern in the two
junctions is highly conserved (Figure 8c). Both α1-α1 and α2-α2 form a Asn187(185)-
Tyr189(188) hydrogen bond pairs in the interface. In addition to this, Asn187(185) forms a pair
of hydrogen bonds with Arg76(75) of another chain from the opposite trimer. The multiple
hydrogen bonds formed by Asn187(185) involving residues from two different chains are
probably one of the major factors stabilizing the trimer-trimer interface.
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Disulfide bonds: Interchain or Intrachain?
Disulfide cross-linking is a recurring theme in the assembly of collagens and is believed to play
an important role in the stabilization of the trimeric structure (11). Fibrillar procollagens are
believed to form interchain disulfide bonds catalyzed by protein disulfide isomerase in either the
C-telopeptide or C-propeptide (58,59). Interchain disulfides have been proposed to form both in
the collagenous and NC1 domains of type IV collagen. Whereas the interchain disulfides in the
collagenous domains are formed within a protomer to stabilize the collagen triple helix, those in
the NC1 domains are believed to occur between the protomers to stabilize the network at the C-
terminus. Disulfide exchange between the NC1 domains of similar α chains from two different
protomers was proposed as one of the major stabilizing forces in the hexamer assembly (60).
Under denaturing conditions, the human placenta derived NC1 hexamer dissociated as dimers
and monomers. The dimers were shown to be crosslinked predominantly by disulfide bridges.
However, a later study by Langeveld et al (57) comparing the NC1 hexamers isolated from
several BMs revealed rather complex results. Whereas the results of the placenta BM and kidney
glomerular BM NC1 hexamers agreed with the previous observations, dissociating as dimers
upon denaturation, the LBM NC1 hexamer dissociated predominantly as monomers implying the
absence of disulfide crosslinks. The crystal structure of LBM NC1 hexamer reveals just that—all
the cysteines are involved in intrachain disulfides.
Siebold et al (60) proposed disulfide exchanges involving Cys20(20’)- Cys111’(111) and
Cys53(53’)-Cys108’(108) pairs in N-subdomain ( and those in similar positions in C-subdomain)
in the α1 chain resulting in a total of four disulfide cross-links in each subdomain based on the
22
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
cynogen bromide cleavage studies. The topological arrangement of disulfides observed in the
crystal structure suggests that the possibility for such a rearrangement is extremely remote
(Figure 9). The disulfides in the NC1 monomer are arranged in three tiers with Cys20-Cys111
and Cys130-Cys225 are close to the triple helical junction, Cys65-Cys71 and Cys176-Cys182
are close to the interface and Cys53-Cys108 and Cys164-Cys222 lies in between. The disulfide
pairs Cys20-Cys111 and Cys53-Cys108 in the monomers of α1A-α1D dimer are about 70 Å and
50 Å apart, respectively. Thus, the possibility for disulfide exchange, if any, exists only for the
Cys65-Cys71 and Cys176-Cys182 pairs. However, the staggered arrangement of the two trimers
brings Cys65-Cys71 pair of α1A closer to its C-subdomain equivalent Cys176’-Cys182’ pair of
α1D chain rather than its counterpart Cys65’-Cys71’ in the N-subdomain. These two closest
disulfide pairs in α1A-α1D dimer are about 16 Å from each other. Even more importantly, these
intrachain disulfides are located in the 3D domain-swapped β-hairpin regions. If the disulfide
exchanges were indeed possible between these pairs it would involve major conformational
alterations. Such a movement of the β-hairpins containing the “exchangeable” cysteine residues
would break both the interchain and intrachain 3D domain swapping interactions, thus
destabilizing the trimer structure. From these arguments, it is difficult to envisage a disulfide
cross-linking between the monomers belonging to two separate protomers in the present
structure. We also examined the possibility of intraprotomer disulfides, which would also
require major conformational changes and potentially move the N-termini of the three chains
severely affecting collagen-NC1 linkage. An alternative conformation must exist for the NC1
domains in other BMs to account for the interprotomer disulfide cross-links.
23
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Biological Significance. Until recently, only monomeric structures of homologous fragments of
noncollagneous domains from types XVIII (61) and XV (62) collagens, known as endostatins,
were available. Recently, Bogin et al (63) reported the crystal structure of human type X
collagen NC1 trimer. This trimer is mainly stabilized by a cluster of calcium ions and three strips
of exposed aromatic amino acids were suggested to be important for the supramolecular
assembly of type X collagen. In the present study we have determined a NC1 hexamer structure,
providing the structural basis for supramolecular assembly in type IV collagen. The structure
unambiguously confirms the chain stoichiometry of α1.α2 network and explains a basis for the
chain specific assembly of type IV collagen. The NC1 monomer folds into a novel tertiary
structure and the close-ended trimer of (α1)2.α2 is organized through unique 3D domain
swapping interactions. These features must be conserved in type IV collagen networks from all
the species due to overall sequence similarity and very high sequence identity of the regions
participating in 3D domain swapping. The chain specificity is determined by the differences in
the primary sequences of the hypervariable regions of the NC1 domains of the constituent
chains, which manifest as different secondary structures at the monomer-monomer interfaces.
The hexamer structure is stabilized by the extensive hydrophobic and hydrophilic interactions at
the trimer-trimer interface without a need for disulfide cross-linking. The crystal structure of
LBM NC1 hexamer and the denaturation studies of NC1 hexamers from several BMs suggest an
alternative conformation must exist in hexamers that are cross-linked by interchain disulfides.
Some hitherto unknown enzymatic process might be responsible for folding the same amino acid
sequences into different conformations in different tissues. There may be other BMs, in addition
to LBM, where the NC1 domains are not crosslinked by disulfide bonds. The absence of cross-
24
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
links at sites such as the BMs of venules might allow for the penetration of leukocytes and
macrophages through the collagen IV network without requiring proteolysis.
The importance of NC1 domains in the chain specific assembly of type IV collagen network has
been established by our previous studies (17,18). The role of α3.α4.α5 network is well known
in the pathogenesis of renal diseases, Alport and Goodpasture syndromes. The NC1 domains are
emerging as a new class of angiogenesis inhibitors through an integrin-dependent mechanism
(64). Preliminary analysis of the NC1 hexamer structure throws some light on the effects of
Alport mutations and cryptic nature of the Goodpasture epitopes.
The Alport mutations data complied from the Human Gene Mutation Database
(http://www.uwcm.ac.uk/uwcm/mg/search/120596.html) suggest that there are at least 18 α5
phenotypes with mutations localized in the NC1 domain including the 12 missense mutations
(Table 4). All these mutation sites are conserved in all six human α chains as well as across the
species. All these sites occur in β strands and none of those side chains interfaces with another
chain or lies on the surface. In three phenotypes, cysteines are mutated to non-cysteine residues
and most of the other changes increase the size of the side chains. Loss of a cysteine that is
participating in a disulfide bond or replacement of a smaller side chain by a bulky one in a rigid
β-sheet structure in the interior of the protein might make it difficult to accommodate the
changes in a native-like fold. It is very likely that the Alport mutations in α5 NC1 domain affect
the folding of the monomer and prevent its participation in the trimer assembly. Further
structural and biochemical studies are required to confirm this prediction.
25
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
The cryptic Goodpasture epitopes have been localized to two regions encompassing residues 17-
31(EA) and 127-141(EB) in the α3 chain (corresponding to residues 15-29 and 125-139 in the α1
chain) (65). These are homologous regions in the two subdomains close to the triple helical
junction containing β2/β2’ strands and the loops connecting them with β1/β1’ strands. They
interface with the other two chains in the trimer structure making them inaccessible to
autoantibodies in the native trimeric structure. A manuscript describing the detailed analysis of
the structural and experimental data is in preparation.
In view of these, our present work establishes a foundation for future studies to define 1) the
code for chain-specific collagen assembly, 2) pathogenic mechanisms that underlie mutations in
the NC1 domains, 3) cryptic nature of Goodpasture epitopes, 4) integrin binding sites on NC1
domains, and 5) role in angiogenesis.
26
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
References
1. Timpl, R., and Brown, J. C. (1996) Bioessays 18(2), 123-131
2. Weber, M. (1992) Kidney International 41, 620-628
3. Pihlajaniemi, T. (1996) in Molecular Pathology and Genetics of Alport Syndrome
(Trygvasson, K., ed) Vol. 117, pp. 46-79, Karger, Basel
4. Miner, J. (1999) Kidney International 56, 2016-2024
5. Prockop, D. J., and Kivirikko, K. I. (1995) Ann. Rev. Biochem. 64, 403-34
6. Myllyharju, J., and Kivirikko, K. I. (2001) Ann Med 33, 7-21
7. Kadler, K. (1994)
8. Bachinger, H.-P., Bruckner, P., Timpl, R., Prockop, D. J., and Engel, J. (1980) Eur. J.
Biochem. 106, 619-632
9. Bachinger, H.-P., Fessler, L. I., Timpl, R., and Fessler, J. H. (1981) J. Biol. Chem. 256,
13193-13199
10. Dolz, R., Engel, J., and Kuhn, K. (1988) Eur J Biochem 178(2), 357-66
11. McLaughlin, S. H., and Bulleid, N. J. (1998) Matrix Biology 16, 369-377
12. Lees, J. F., Tasab, M., and Bulleid, N. J. (1997) EMBO J. 16(5), 908-916
13. Dion, A. S., and Myers, J. C. (1987) J Mol Biol 193(1), 127-43
14. Rosenbloom, J., Endo, R., and Harsch, M. (1976) J. Biol. Chem. 251, 2070-2076
15. Schofield, D. J., Uitto, J., and Prockop, D. J. (1974) Biochemistry 13, 1801-1806
16. Uitto, V., Uitto, J., and Prockop, D. J. (1981) Arch. Biochem. Biophys. 210, 445-454
17. Boutaud, A., Borza, D.-B., Bondar, O., Gunwar, S., Netzer, K.-O., Singh, N., Ninomiya,
Y., Sado, Y., Noelken, M. E., and Hudson, B. G. (2000) J. Biol. Chem. 275, 30716-30724
27
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
18. Borza, D. B., Bondar, O., Ninomiya, Y., Sado, Y., Naito, I., Todd, P., and Hudson, B. G.
(2001) J Biol Chem 276(30), 28532-40.
19. Hudson, B. G., Reeders, S. T., and Tryggvason, K. (1993) J Biol Chem 268(35), 26033-6
20. Timpl, R., Wiedemann, H., van Delden, V., Furthmayr, H., and Kuhn, K. (1981) Eur J
Biochem 120(2), 203-11
21. Zhou, J., Ding, M., Zhao, Z., and Reeders, S. T. (1994) J. Biol. Chem. 269, 13193-13199
22. Netzer, K. O., Suzuki, K., Itoh, Y., Hudson, B. G., and Khalifah, R. G. (1998) Protein Sci
7(6), 1340-51
23. Fowler, S. J., Jose, S., Zhang, X., Deutzmann, R., Sarras, M. P., Jr., and Boot-Handford,
R. P. (2000) J Biol Chem 275(50), 39589-99.
24. Boute, N., Exposito, J. Y., Boury-Esnault, N., Vacelet, J., Noro, N., Miyazaki, K.,
Yoshizato, K., and Garrone, R. (1996) Biol Cell 88(1-2), 37-44
25. Guo, X. D., and Kramer, J. M. (1989) J Biol Chem 264(29), 17574-82.
26. Sibley, M. H., Johnson, J. J., Mello, C. C., and Kramer, J. M. (1993) J Cell Biol 123(1),
255-64.
27. Blumberg, B., MacKrell, A. J., and Fessler, J. H. (1988) J Biol Chem 263(34), 18328-37.
28. Yasothornsrikul, S., Davis, W. J., Cramer, G., Kimbrell, D. A., and Dearolf, C. R. (1996)
GenBank , AAB64082
29. Exposito, J. Y., D'Alessio, M., Di Liberto, M., and Ramirez, F. (1993) J Biol Chem
268(7), 5249-54.
30. Exposito, J. Y., Suzuki, H., Geourjon, C., Garrone, R., Solursh, M., and Ramirez, F.
(1994) J Biol Chem 269(18), 13167-71.
28
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
31. Gunwar, S., Ballester, F., Noelken, M. E., Sado, Y., Ninomiya, Y., and Hudson, B. G.
(1998) J Biol Chem 273(15), 8767-75
32. Zhang, X., Hudson, B. G., and Sarras, M. P., Jr. (1994) Dev Biol 164(1), 10-23
33. Guo, X., Johnson, J. J., and Kramer, J. M. (1991) Nature 349, 707-709
34. Sibley, M. H., Graham, P. L., von Mende, N., and Kramer, J. M. (1994) EMBO J. 13,
3278-3285
35. Kashtan, C. E., and Michael, A. F. (1993) Am. J. Kid. Dis. 22, 627-640
36. Kashtan, C. E., and Michael, A. F. (1996) Kidney Int 50, 1445-1463
37. Cosgrove, D., Meehan, D. T., Grunkemeyer, J. A., Kornak, J. M., Sayers, R., Hunter, W.
J., and Samuelson, G. C. (1996) Genes Dev 10(23), 2981-92.
38. Miner, J. H., and Sanes, J. R. (1996) J Cell Biol 135(5), 1403-13.
39. Gunwar, S., Noelken, M. E., and Hudson, B. G. (1991) J Biol Chem 266(21), 14088-94
40. Peczon, B. D., McCarthy, C. A., and Merrit, R. B. (1982) Exp. Eye. Res. 35, 643-651
41. Otwinowski, Z., and Minor, W. (1997) Methods in Enzymology 276, 307-326
42. Terwilliger, T. C., and Berendzen, J. (1997) Acta Crystallogr. D55, 849-861
43. Terwilliger, T. C. (2000) Acta Crystallogr D 56(Pt 8), 965-72.
44. Dodson, E. J., Winn, M., and Ralph, A. (1997) Methods in Enzymology 277, 620-633
45. Jones, S., and Thornton, J. M. (1996) Proceedings of The National
Academy of Science (U.S.A) 93, 13-20
46. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve,
R. W., Jiang, J. S., Kuszewski, J., Pannu, N. S., and al., e. (1998) Acta Crystallogr. D54,
905-921
47. Evans, S. V. (1993) J. Mol.Graphics 11, 134-138
29
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
48. Nicholls, A., Sharp, K. A., and Honig, B. (1991) Proteins 11, 281-296
49. McDonald, I. K., and Thornton, J. M. (1994) J. Mol.Biol. 238, 777-793.
50. Iwata, M., Sado, Y., Sasaki, T., Imamura, Y., Ninomiya, Y., and Hayashi, T. (1996) J
Biochem (Tokyo) 120(1), 133-7.
51. Timpl, R., Oberbaumer, I., von der Mark, H., Bode, W., Wick, G., Weber, S., and Engel,
J. (1985) Ann NY Acad Sci 460, 58-72
52. Stubbs, M., Summers, L., Mayr, I., Schneider, M., Bode, W., Huber, R., Ries, A., and
Kuhn, K. (1990) J Mol Biol 211, 683-684
53. Dauter, Z., and Dauter, M. (1999) J. Mol. Biol. 289, 93-101
54. Dauter, Z., Dauter, M., and Rajashankar, K. R. (2000) Acta Crystallogr. D56, 232-237
55. Schlunegger, M. P., Bennett, M. J., and Eisenberg, D. (1997) Advances in Protein
Science 50, 61-132
56. Jones, T. A. (1978) J. Appl. Crystallogr. 11, 268-272
57. Langeveld, J. P., Wieslander, J., Timoneda, J., McKinney, P., Butkowski, R. J., Wisdom,
B. J., Jr., and Hudson, B. G. (1988) J Biol Chem 263(21), 10481-8
58. Uitto, J., and Prockop, D. J. (1973) Biochem. Biophys. Res. Commun. 55, 904-911
59. Koivu, J., and Myllyla, R. (1987) J. Biol. Chem. 262, 6159-6164
60. Siebold, B., Deutzmann, R., and Kuhn, K. (1988) Eur J Biochem 176(3), 617-24
61. Hohenester, E., Sasaki, T., Olsen, B. R., and Timpl, R. (1998) Embo J. 17, 1656-1664
62. Sasaki, T., Larsson, H., Tisi, D., Claesson-Welsh, L., Hohenester, E., and Timpl, R.
(2000) J. Mol. Biol. 301, 1179-1190
63. Bogin, O., Kvansakul, M., Rom, E., Singer, J., Yayon, A., and Hohenester, E. (2002)
Structure (Camb) 10(2), 165-73.
30
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
64. Petitclerc, E., Boutaud, A., Prestayko, A., Xu, J., Sado, Y., Ninomiya, Y., Sarras, M. P.,
Jr., Hudson, B. G., and Brooks, P. C. (2000) J Biol Chem 275(11), 8051-61
65. Netzer, K. O., Leinonen, A., Boutaud, A., Borza, D. B., Todd, P., Gunwar, S., Langeveld,
J. P., and Hudson, B. G. (1999) J Biol Chem 274(16), 11267-74
66. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J. M. (1993) J. Appl.
Cryst.. 26, 283-291
67. Barton, G. J. (1993) Prot. Eng. 6, 37-40
31
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Legends for Figures
Figure 1. Schematic illustration of type IV collagen supramolecular network assembly. (a) Six
genetically distinct α-chains (α1-α6) assemble into three distinct protomers. The protomers are
characterized by a long central collagen triple helix, 7 S domain at the N-terminus and globular
NC1 trimer at the C-terminus. (b) NC1 domains provide chain specificity for chain association,
alignment, registration and propagation from C- to N-terminal direction. This sequence of events
shown for α1.α2 protomer is true for other protomers also. (c) Supramolecular network is
assembled through N-terminal tetrameric association and C-terminal dimeric association. NC1
hexamer is formed at the C-terminus.
Figure 2. Alignment bovine (a) α1 and (b) α2 NC1 domain sequences with other known
sequences. Residues that are conserved in only mammalian sequences (red), five mammalian and
at least one invertebrate sequences (green) and all the sequences (blue) in each family are
highlighted. The cysteine pairs forming intrachain disulfides are identified at the bottom. The six
segments forming the hexamer interface are shown in boxes. Three-state secondary structural
elements are assigned based on the crystal structure. Both α1 and α2 structures contain β-strands
β1-β10 and β1’-β10’ and 310 helices g1 and g1’. The difference in secondary structures of two
chains occur as a 310 helix in α1 and β-strand βp’ in α2 at the equivalent regions. The partner of
βp’ strand of α2 chain is in one of the two α1 chains. The corresponding region in α2 and the
other α1 chains are extended structures. These regions marked by boxes. The secondary
structures were from PROCHECK(66). These figures were prepared using AMPS and
ALSCRIPT (67).
32
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Figure 3. Stereo view of a section of 2Fo-Fc map (contoured at 1 σ), where differences between
human and bovine sequences occur (α2: Asp96Glu, Glu97Asp) (also refer Fig 2).
Figure 4. Structures of (a) trimer down the pseudo 3-fold axis and (b) hexamer down 2-fold
NCS axis. Different chains within a trimer are shown in different colors: α1A, red; α1B, green;
α2, blue. The trimer-trimer interface (“Equatorial Plane”), collagen triple helical junction, and
pseudo 3-fold axis or triple helix axis (“Polar Axis”) are identified. This figure and Figs. 5, 8, 9
and 10b were made using SETOR (47).
Figure 5. (a) Superposition of Cα traces of α1 (red) and α2 (blue) chains represented in stereo.
The residue numbers are shown only for α1 chain. (b). Superposition N- (red) and C-subdomains
(green) in α1 chain. All the β-strands and disulfides superimpose with low RMS deviation and
the difference occurs only in a small region connecting β8(8’) and β9(9’) strands, both essential
for domain swapping interactions. The extra 310 helix in the C-subdomain interfaces with another
monomer and the corresponding region in the α2 chain is extended conformation forming the
parallel β-sheet with the neighboring chain.
Figure 6. Topology diagram of NC1 trimer (α1A, red; α1B, green; α2, blue) depicting interchain
and intrachain 3D domain swapping interactions (generic assembly) and chain interfaces with
different secondary structural elements (specific assembly). The secondary structural elements
are labeled only for α1B chain. The β-sheets, I & II in the N-subdomain and I’ & II’ in the C-
subdomain are identified. Each subdomain has 10 β-strands (β1-β10 and β1’-β10’) and two short
33
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
310 (g1 and g2’) helices. Additionally there are distinct secondary structures at the three
interfaces—a parallel β-sheet (βp-βp’) at α1B-α2 interface and a 310 helix (g1’) and extended
structure at α1A-α1B and α2-α1A interfaces.
Figure 7. a) Generic interactions in the trimer. Six-strand β-sheets formed by interchain and
intrachain 3D domain swapping interactions form the major force in the trimer organization. The
sheets belonging to subdomains are shown in boxes to highlight such interactions. Central β
barrel-like core, shown inside the circle, also plays a role in packing and stabilizing this scaffold.
(b) Unique secondary structures and prominent side chain interactions at the three interfaces are
shown. The α1B-α2 interface has more number of hydrogen bonds than the other interfaces.
Fig. 8. Comparison of essential hydrogen bonding interactions in the interface at “core”, “outer”
and major-minor junction for α1-α1 and α1-α2 dimers at the trimer-trimer interface (see text for
details).
Fig. 9. Disulfide bonds in LBM NC1 domains. The topological distribution of intrachain
disulfides observed in the crystal structure suggests that the proposed disulfide exchanges
involving C20-C111/C20’-C111’ and C53-C108/C53’-C71 pairs in chains of two protomers are
unlikely. However, it is possible for the C65-C71/C20’-C111’ disulfide pairs to exchange
between the trimers provided there is a large conformational change breaking the 3D domain
swapping interaction within a trimer.
34
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
1The abbreviations used are: BM, basement membrane; LBM, lens capsule basement membrane;
NC1, noncollagenous domain of type IV collagen, MPD, 2,4-methyl pentanediol; NCS, non-
crystallographic symmetry; MAD, multiwavelength anomalous diffraction; ECM, extracellular
matrix
Acknowledgments: We would like to thank Drs. Aina Cohen and Paul Ellis at Stanford
Synchrotron Radiation Laboratory for technical support with MAD data collection, Dr. Shanthi
Govindaraj for help in initial crystallization trials, Dr. Dorin-Bogdon Borza for critical comments
and Richard Grabbe for technical help. We gratefully acknowledge Drs. Jaun Saus, Fernando
Revert, and Pilar Martinez-Martinez (Instituto Investigaciones Citologicas, Valencia, Spain) for
providing the bovine α1 and α2 NC1 sequences. The initial characterization of the crystals and
heavy atom screening were carried out at the University of Missouri-Kansas City X-ray
Diffraction Facility.
This work was supported by Grants RI 80086 from the Ernst F. Lied Basic Science Program
University of Kansas Medical Center and DK63925 from the US National Institutes of Health (to
MS); by Grants DK18381 and DK53763 from the US National Institutes of Health (to B.G.H.);
by Grant SAF200-0047 of Plan Nacional I+D, Spain (to Juan Saus).
35
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Tables
Table 1. Summary of Crystallographic Analysis Data Collection Dataset Peak Inflection Remote1 Remote2
Wavelength (Å) 0.9195 0.9197 0.9537 0.8856
Resolution (Å) 2.1 2.1 2.15 2.0
Measured reflections 602,172 603,309 568,640 686,286
Unique reflections 159,617 159,667 149,817 184,445
Completeness (%)* 98.3 (90.9) 98.2 (90.5) 98.7 (95.1) 97.9 (87.8)
Rsym (%)† 4.0 (7.7) 3.0 (6.7) 2.4 (4.9) 3.4 (8.6)
I/σ(I) 29.2(15.0) 33.0 (18.2) 37.6 (26) 30.5 (13.1)
Phasing Statistics Resolution range (Å) 50.0 – 2.2
Number of Br sites 33
Overall Z-score 127
Figure of Merit SOLVE / RESOLVE
0.67 / 0.76
Refinement Statistics Resolution range (Å) 8.0 – 2.0
Number of reflections (σ>2) working / test
166,448 / 8,789
Rcryst / Rfree (%)‡ 16.8 / 19.7
Average B-factor (Å2)
Protein atoms 15.7
Solvent & ions 19.8
36
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
All atoms 15.9
RMS deviation
Bond lengths (Å) 0.0051
Bond angles (°) 1.29
* The overall completeness is given, with the completeness in the highest resolution shell shown in the parentheses. Similar convention is followed for Rsym and I/σ(I) also. †Rsym = ∑h∑i |<I(h)> - I(h)i| / ∑h∑i|I(h)i|. ‡ 5% of the data were excluded from refinement and were used to determine the Rfree. The Rcryst does not include these reflections. In both cases R=∑( |Fo| - k|Fc|)/∑|Fo|, with an appropriate choice of reflections for the summation.
Table 2. Comparison of monomer-monomer interfaces in the trimer. α1A-α1B α1b-α2 α2-α1A Interface Parameter α1A α1B α1B α2 α2 α1A Number of segments 5 7 5 7 5 8
Number of residues 49 60 51 65 49 59
∆ ASA (Å2) 2137 2182 2087 2066 1985 2044
Polar/non-polar atoms (%) 40.1/59.9 24.5/75.5 44.3/55.7 32.5/67.5 39.9/60.1 24.8/75.3
Hydrogen bonds M-M/M-S/S-S
9/8/5
11/8/12
9/9/3
∆ ASA, interface solvent accessible area; M, main chain; S, side chain
37
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
38
Table 3. Comparison of interface parameters defining the trimer-trimer interaction in the NC1 hexamer and observed mean for 32 homodimer complexes (45). Interface Parameter NC1 Hexamer Observed Mean
(32 Homodimers) 399∆ASA (Å2) 4173.1 1685.03
Planarity 1.91 3.46
Circularity 0.87 0.71
Segmentation 18 5.22
Hydrogen bonds per 100 Å2 1.2 0.70
Gap Index 1.24 2.2
Percentage of polar and non-polar atoms are 45.5 and 54.5 respectively.
Table 4. Alport mutations in α5 NC1 domain compiled from Human Gene Mutation Database. The residue numbers are identical in α1 chain.
Mutation
Gly30 → Ala Ser32 → Phe Ala42 → Asp Pro61 → Thr
Trp82 → Arg/Ser Cys108 → Ser Cys111→ Arg Gly140 → Asp Leu193 → Arg Arg2221 → Gln Cys222 → Trp
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
A
1
2
4
5
6
3
Type IV Collagen Chains
NC1 TrimerNC1 Trimer
{
Protomers
7S7S
{
1
1265
5
33
554
1
12
ProtomerProtomer
{NC1 NC1
HexamerHexamer
DimerDimerTetramerTetramer
7S7S
Network AssemblyC
{
NC1 TrimerNC1 Trimer7S7S Protomer Protomer
Protomer AssemblyB
α1.α2 α1.α2 NetworkNetwork
α1.α2.α1α1.α2.α1
α3.α4.α5α3.α4.α5
α5.α6.α5α5.α6.α5
α1.α2.α1α1.α2.α1 {
1
1
21
1
21
1
2
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
6161616161616359
127127127127127127127123
192192192192192192193188
229229229229219229231225
hydra drosophila c.elegans dog rabbit mouse human bovine structure Secondary
1 2D F G F Y L V K H S Q S I K V P S C P A G M Q T M W E G Y S F L Y A Q G N E R A F G Q D L G Q P G S C L K R F S T M P
A P A A L D Y L T G I L I T R H S Q S E T V P A C S A G H T E L W T G Y S L L Y V D G N D Y A H N Q D L G - - - S C V P R F S T L P- - - A P S R - - G F T F A K H S Q T T A V P Q C P P G A S Q L W E G Y S L L Y V Q G N G R A S G Q D L G Q P G S C L S K F N T M P- - - S V V H - - G F L V T R H S Q T T D D P Q C P P G T K I L Y H G Y S L L Y V Q G N E R A H G Q D L G T A G S C L R K F S T M P- - - S V D H - - G F L V T R H S Q T I H D P Q C P P G T K I L Y H G Y S L L Y V Q G N E R A H G Q D L G T A G S C L R K F S T M P- - - S V D H - - G F L V T R H S Q T T D D P L C P P G T K I L Y H G Y S L L Y V Q G N E R A H G Q D L G T A G S C L R K F S T M P- - - S V D H - - G F L V T R H S Q T I D D P Q C P S G T K I L Y H G Y S L L Y V Q G N E R A H G Q D L G T A G S C L R K F S T M P- - - S V D H - - G F L V T R H S Q T T D D P Q C P P G T K I L Y H G Y S L L Y V Q G N E R A H G Q D L G T A G S C L R K F S T M P
disorderedβ1 β2 β3 β4 β5
1 10 20 30 40 50 60
hydra drosophila c.elegans dog rabbit mouse human bovine structure Secondary
3 3 2 1F L F C D I Q N K C V V A S R N D Y S F W L S T A E K P K E A P S S G A D L E N Y I S R C I V C E A P S H V L A V H S Q S E L DV L S C G Q N N V C N Y A S R N D K T F W L T T N A A I P - - M M P V E N I E I R Q Y I S R C V V C E A P A N V I A V H S Q T I E VF M F C N M N S V C H V S S R N D Y S F W L S T D E P M T P M M N P V T G T A I R P Y I S R C A V C E V P T Q I I A V H S Q D T S VF L F C N I N N V C N F A S R N D Y S Y W L S T P E P M P M S M A P I A G D N I R P F I S R C A V C E A P A M V M A V H S Q T I Q IF L F C N I N N V C N F A S R N D Y S Y W L S T P E P M P M S M A P M W G D N I R P F I T R C A V C E A P A M V M A V H S Q T I Q IF L F C N I N N V C N F A S R N D Y S Y W L S T P E P M P M S M A P I S G D N I R P F I S R C A V C E A P A M V M A V H S Q T I Q IF L F C N I N N V C N F A S R N D Y S Y W L S T P E P M P M S M A P I T G E N I R P F I S R C A V C E A P A M V M A V H S Q T I Q IF L F C N I N N V C N F A S R N D Y S Y W L S T P E P M P M S M A P I T G E N I R P F I S R C A V C E A P A M V M A V H S Q T I Q I
β6 β7 β8 β9 g1 β10 β1’70 80 90 100 110 120 130
hydra drosophila c.elegans dog rabbit mouse human bovine structure Secondary
4 5 6 6P K C P D G W E N L W T G F S F L M Y N S A G A Q G S G Q L L S S S G S C L E D F R V N P Y I E C H G R G T C W Y Y G P T L S F WP D C P N G W E G L W I G Y S F L M H T A V G N G G G G Q A L Q S P G S C L E D F R A T P F I E C N G A K G T C H F Y E T M T S F WP Q C P Q G W S G M W T G Y S F V M H T A A G A E G T G Q S L Q S P G S C L E E F R A V P F I E C H G - R G T C N Y Y A T N H G F WP Q C P S G W S S L W I G Y S F V M H T S A G A E G S G Q A L A S P G S C L E E F R S A P F I E C H G - R G T C N Y Y A N A Y S F WP P C P N G W S S L W I G I S F V M H T S A G A E G S G Q A L A S P G S C L E E F R S A P F I E C H G - R G T C N Y Y A N A Y S F WP Q C P N G W S S L W I G Y S F V M H T S A G A E G S G Q A L A S P G S C L E E F R S A P F I E C H G - R G T C N Y Y A N A Y S F WP P C P S G W S S L W I G Y S F V M H T S A G A E G S G Q A L A S P G S C L E E F R S A P F I E C H G - R G T C N Y Y A N A Y S F WP Q C P T G W S S L W I G Y S F V M H T S A G A E G S G Q A L A S P G S C L E E F R S A P F I E C H G - R G T C N Y Y A N A Y S F W
β2’ β3’ β4’ β5’ β6’ β7’ β8’140 150 160 170 180 190
hydra drosophila c.elegans dog rabbit mouse human bovine structure Secondary
5 4L S T I G E S N M F Q V P K F E I L E R N L K A R V S R C A V C M K S V PM Y N L E S S Q P F E R P Q Q Q T I K A G E R Q S H V S R C Q V C M K N S SL S I V D Q D K Q F R K P M S Q T L K A G G L K D R V S R C Q V C L K N R -L A T I E R S E M F K K P T P S T L E A G E L R T H V - - - - - - - - - - -L A T I E R S E M F K K P T P S T L K A G E L H T H V S R C Q V C M R R T -L A T I E R S E M F K K P T P S T L K A G E L R T H V S R C Q V C M R R T -L A T I E R S E M F K K P T P S T L K A G E L R T H V S R C Q V C M R R T -L A T I E R S E M F K K P T P S T L K A G E L R T H V S R C Q V C M R R T -
g2’ β9’ g1’ β10 ’200 210 220 230
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
59595959505966
123123123123114123132
189189189189191189197
227227227227232231229
drosophila c.elegans dog rabbit mouse human bovine structure Secondary
1 2A P A P P A P K S R G F I F A R H S Q S V H V P Q C P A N T N L L W E G Y S L S G N V A A S R A V G Q D L G Q S G S C M M R F T T M- - - - - - - Y R D G F V L V K H S Q T T E V P R C P E G Q T K L W D G Y S L L Y I E G N E K S H N Q D L G H A G S C L Q R F S T M- - - - - - - - - - - - - - - - H S Q T D Q E P M C P V G M N K L W S G Y S L L Y F E G Q E K A H N Q D L G L A G S C L A R F S T M- - - - - - - V S I G Y L L V K H S Q T E Q E P M C P L G M N K L W S G C S L L Y F E G P E K A H N Q E P G L A G S C L A R F S T M- - - - - - - V S I G Y L L V K H S Q T D Q E P M C P V G M N K L W S G Y S L L Y F E G Q E K A H N Q D L G L A G S C L A R F S T M- - - - - - - V S I G Y L L V K H S Q T D Q E P M C P V G M N K L W S G Y S L L Y F E G Q E K A H N Q D L G L A G S C L A R F S T M- - - - - - - I S I G Y L L V K H S Q T D Q E P M C P V G M N K L W S G Y S L L Y F E G Q E K A H N Q D L G L A G S C L A R F S T M
disorderedβ1 β2 β3 β4 β5
1 10 20 30 40 50 60
drosophila c.elegans dog rabbit mouse human bovine structure Secondary
3 3 2 1P Y M L C D I T N V C H F A Q N N D D S L W L S T A E P M P M T M T P I Q G R D L M K Y I S R C V V C E T T T R I I A L H S Q S M SP F L F C D F N N V C N Y A S R N E K S Y W L S T S E A I P M M - - P V N E R E I E P Y I S R C A V C E A P A N T I A V H S Q T I QP F L Y C N P G D V C Y Y A S R N D K S Y W L S T T A P L P M M - - P V A E D D I K P Y I S R C S V C E A P A V A I A V H S Q D V SP F L Y C N P G D V C Y Y A S R N D K S Y W L S T T A P L P M M - - P V A E D E I K P Y I S R C S V C E A P P V A I A V H S Q D V SP F L Y C N P G D V C Y Y A S R N D K S Y W L S T T A P L P M M - - P V A E E E I K P Y I S R C S V C E A P A V A I A V H S Q D T SP F L Y C N P G D V C Y Y A S R N D K S Y W L S T T A P L P M M - - P V A E D E I K P Y I S R C S V C E A P A I A I A V H S Q D V SP F L Y C N P G D V C Y Y A S R N D K S Y W L S T T A P L P M M - - P V A E E D I R P Y I S R C S V C E A P A V A I A V H S Q D V S
β6 β7 β8 β9 g1 β10 β1’70 80 90 100 110 120 130
drosophila c.elegans dog rabbit mouse human bovine structure Secondary
4 5 6 6I P D C P G G W E E M W T G Y S Y F M S T L D N V G G V G Q N L V S P G S C L E E F R A Q P V I E C H G H G R C N Y Y D A L A S FI P N C P A G W S S L W I G Y S F A M H T G A G A E G G G Q S P S S P G S C L E D F R A T P F I E C N G A R G S C H Y F A N K F S FI P H C P A G W R S L W I G Y S F L M H T A A G D E G G G Q S L V S P G S C L E D F R A T P F I E C S G G R G T C H Y Y A N K Y S FI P H C P A G W R S L W I G Y S F L M H T A A G D E G G G Q S L V S P G S C L E D F R A T P F I E C N G G R G T C H Y Y A N K Y S FI P H C P A G W R S L W I G Y S F L M Y T A A G D E G G G Q S L V S P G S C L E D F R A T P F I E C N G G R G T C H Y F A N K Y S FI P H C P A G W R S L W I G Y S F L M H T A A G D E G G G Q S L V S P G S C L E D F R A T P F I E C N G G R G T C H Y Y A N K Y S FI P H C P A G W R S L W I G Y S F L M H T A A G D E G G G Q S L V S P G S C L E D F R A T P F I E C N G A R G T C H Y Y A N K Y S F
β2’ β3’ β4’ β5’ β6’ β7’ β8’140 150 160 170 180 190
drosophila c.elegans dog rabbit mouse human bovine structure Secondary
5 4W L T V I E E Q D Q F V Q P R Q Q T L K A D F T S K I S R C T V C R R R G N S F V A R T A K S R A D A S S G V H R W F C L EW L T T I D N D S E F K V P E S Q T L K S G N L R T R V S R C Q V C V K S T D G R H - - - - - - - - - - - - - - - - - - - - -W L T T I P E Q S F Q G S P S A D T L K A G - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -W L T T I P E R S F Q G S P S A D T L K A G L I R T H I S R C Q V C M K N L - - - - - - - - - - - - - - - - - - - - - - - - -W L T T I P E Q N F Q S T P S A D T L K A G L I R T H I S R C Q V C M K N L - - - - - - - - - - - - - - - - - - - - - - - - -W L T T I P E Q S F Q G S P S A D T L K A G L I R T H I S R C Q V C M K N L - - - - - - - - - - - - - - - - - - - - - - - - -W L T T I P E Q S F Q G T P S A D T L K A G L I R T H I S R C Q V C M K N L - - - - - - - - - - - - - - - - - - - - - - - - -
β p’ β9’ g1’ β10 ’200 210 220 230 240 250 260
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
N(5,115) C(114,227)
C-subdomain
N-subdomain
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
NC
β1
β4
β5
β7
β6
’
’
’β4
β3 β8
’
’’
g1
g1’
β6
β7 β5’
I
II
I’
II’
β1
β4
β5
β7
β6
β4
β3 β8
g1
g1’
β6
β7 β5
’β6 ’β7 β9
β2 β10β2β10β1
’β3β8
β9
α1B
α2
α1A
βββββββppppppp’’’ pppp
gggggg2222’
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
N-subdomain
C-subdomain
β barrel-likecore
β9
β9'β4
β3'
β8
β7'
β6
β7
β8'
β4'
β3
β6'
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
F57
Y31
R55
D121 T124 T124
R55A54K56 R55 K56
S198
F199
F202 F202
Y31Y30
F56 F57
M201 M201
R102 R102K102
α1B-α2 α2-α1A α1A-α1B
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
70 Å
50 Å
16 Å 16 Å
C20-C111
C20'-C111'
C53-C108
C53'-C108'
C65-C71
C65'-C71'
C176-C182
C176'-C182'
α1A
α1D
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from
HudsonMunirathinam Sundaramoorthy, Muthuraman Meiyappan, Parvin Todd and Billy G.
basement membranesCrystal structure of NC1 domains: Structural basis for type IV collagen assembly in
published online April 22, 2002J. Biol. Chem.
10.1074/jbc.M201740200Access the most updated version of this article at doi:
Alerts:
When a correction for this article is posted•
When this article is cited•
to choose from all of JBC's e-mail alertsClick here
by guest on April 5, 2018
http://ww
w.jbc.org/
Dow
nloaded from