NMR Studies Demonstrate a Unique AAB Composition and Chain Register for a Heterotrimeric Type IV Collagen Model Peptide Containing a Natural Interruption Site*

Background: Heterotrimeric type IV collagen has breaks in the triple-helix repeating Gly-X-Y sequence. Results: NMR studies on a type IV peptide model show heterotrimeric chain selection and register, with new hydrogen bonds formed at an interruption site. Conclusion: Interruption sites may provide a driving force for self-assembly and chain register. Significance: Interruptions can have a positive effect on triple-helix continuity in non-fibrillar heterotrimer collagens. All non-fibrillar collagens contain interruptions in the (Gly-X-Y)n repeating sequence, such as the more than 20 interruptions found in chains of basement membrane type IV collagen. Two selectively doubly labeled peptides are designed to model a site in type IV collagen with a GVG interruption in the α1(IV) and a corresponding GISLK sequence within the α2(IV) chain. CD and NMR studies on a 2:1 mixture of these two peptides support the formation of a single-component heterotrimer that maintains the one-residue staggering in the triple-helix, has a unique chain register, and contains hydrogen bonds at the interruption site. Formation of hydrogen bonds at interruption sites may provide a driving force for self-assembly and chain register in type IV and other non-fibrillar collagens. This study illustrates the potential role of interruptions in the structure, dynamics, and folding of natural collagen heterotrimers and forms a basis for understanding their biological role.

All non-fibrillar collagens contain interruptions in the (Gly-X-Y) n repeating sequence, such as the more than 20 interruptions found in chains of basement membrane type IV collagen. Two selectively doubly labeled peptides are designed to model a site in type IV collagen with a GVG interruption in the ␣1(IV) and a corresponding GISLK sequence within the ␣2(IV) chain. CD and NMR studies on a 2:1 mixture of these two peptides support the formation of a single-component heterotrimer that maintains the one-residue staggering in the triple-helix, has a unique chain register, and contains hydrogen bonds at the interruption site. Formation of hydrogen bonds at interruption sites may provide a driving force for self-assembly and chain register in type IV and other non-fibrillar collagens. This study illustrates the potential role of interruptions in the structure, dynamics, and folding of natural collagen heterotrimers and forms a basis for understanding their biological role.
Collagens present a large family of triple-helical, extracellular matrix molecules that provide structural integrity for the human body (1)(2)(3). Twenty-eight distinct types of collagen have been identified, including those that form fibrils with a 67-nm axial period and others classified as non-fibrillar collagens. Some of the collagens are homotrimers, such as fibrillar type II and type III collagens, whereas others are heterotrimers, such as the fibrillar type I and non-fibrillar basement mem-brane type IV collagens (4,5). Collagen displays a unique (Gly-X-Y) n repetitive sequence pattern, which leads to the formation of the characteristic triple-helix structure (6). The strict requirement of Gly as every third residue is perfectly maintained throughout the triple helical domain of fibrillar collagens, such as type I, II, and III collagens; even the replacement of one Gly by a larger residue leads to disease (7). In contrast, interruptions in the repetitive (Gly-X-Y) n sequences have been observed within the triple helical domain of all non-fibrillar collagens (8). There are more than 350 interruptions found in non-fibrillar collagen chains, such as the ϳ20 breaks in the Gly-X-Y repeating pattern seen in the triple-helix domain of type IV collagen chains in basement membranes.
These natural interruptions in the collagen triple-helix are suggested to play important roles in molecular structure, selfassociation, binding, and degradation (9 -14). A number of flexible regions seen in type IV collagen by rotary shadowing electron microscopy correspond to interruption sites in the sequence and are considered to be important for the networklike structure seen in basement membrane collagen (12). One interruption in type IV collagen appears to play a role in binding to an integrin cell receptor in tumor cell adhesion, whereas two interruptions within type X collagen were shown to serve as specific cleavage sites of matrix metalloproteinases (10,13,14).
Basement membranes contain three different forms of type IV collagen, which are all heterotrimers: [␣1(IV)] 2 ␣2(IV), ␣3(IV)␣4(IV)␣5(IV), and [␣5(IV)] 2 ␣6(IV). Interruptions are often found at corresponding sites in the different chains within one heterotrimeric triple-helix, but the lengths of the interruptions usually differ within the two or three different chains at a given site. It is poorly understood how interruptions in three different chains are accommodated collectively in the self-assembly of heterotrimer type IV collagen.
Triple-helical peptides have been extensively utilized as valuable models to investigate the stability, folding, dynamics, and molecular structure of collagen (15)(16)(17)(18)(19). Although the large majority of studies have been done on homotrimer collagen model peptides, a heterotrimer platform is needed to model collagens with more than one type of chain, such as type IV collagen. In some studies, covalent linkages have been employed to force the selection of three different chains and their alignment for heterotrimer peptide models of collagen, although challenges in purity are significant (20 -22). More recently, important progress has been achieved in the design of heterotrimer peptide models using electrostatic interaction strategies (23)(24)(25)(26)(27) to create defined compositions of heterotrimeric triple-helices, introduce mutations in one, two, or three chains, and create sticky ends to promote association. In addition, a computational approach has been used to optimize interactions to produce heterotrimers that are more stable than homotrimeric peptides (28,29).
We have previously reported a study on a heterotrimer composed of two peptides designed to model a natural interruption site in the triple helical domain of the most abundant isoform, [␣1(IV)] 2 ␣2(IV) of type IV collagen (30). One peptide models the GVG interruption at residue 292 of the ␣1(IV) chain, whereas the other peptide contains the corresponding GISLKG interruption at residue 288 of the ␣2(IV) chain at the same site (30). This earlier study indicated the formation of a heterotrimeric triple-helix, but structural details of the interruption and surrounding triple-helix could not be defined because only a single chain was labeled. Here, two 15 N/ 13 C selectively labeled peptides were synthesized with sequences corresponding to the human ␣1(IV) and ␣2(IV) chains at this GVG/GISLKG interruption site. Application of a number of NMR techniques demonstrated that each peptide forms a homotrimer, with an absence of hydrogen bonding throughout the interruption. When the two peptides were mixed in a 2:1 ratio, a singlecomponent heterotrimer was observed with a defined register ␣1␣1␣2(IV) and with interchain hydrogen bonds at the interruption site.

Experimental Procedures
Notation-The number of residues between the repeating Gly-X-Y tripeptides can categorize the interruptions. Interruptions in the form of Gly-X-Y-Gly-(AA) n -Gly-X-Y are denoted as GnG type interruptions (8).
Sample Preparation-Peptides A and B were synthesized by Tufts University Core Facility (Boston, MA) (see Table 1). Peptide B was identical to that previously reported in Ref. 30, synthesized with 15 N/ 13 C-labeled amino acids at positions Ile 13 , Ser 14 , Leu 15 , and Ala 25 . The sequence of peptide A studied here had four minor changes outside of the interruption site when compared with that studied in Ref. 30. These minor changes in flanking residues were designed to try to improve overall triplehelix stability, but as reported below, the A homotrimer was very similar to that found previously. Importantly, the A peptide now contains selectively 15 N/ 13 C doubly labeled residues at four positions, Gly 14 , Val 15 , Gly 16 , and Gly 25 . Samples of homotrimer peptides A and B were prepared in 10% D 2 O, 90% H 2 O with a concentration of 6 mM. Peptides A and B were mixed in solution at a molar ratio of 2A:1B (4 mM:2 mM) to look for formation of heterotrimeric molecules. The mixtures were heated to 50°C, cooled slowly to 0°C, and incubated for 30 h to prepare heterotrimer samples for measurements.
Circular Dichroism Spectroscopy-CD spectra were recorded on an Aviv model 62DS spectrophotometer with a Peltier thermoelectric temperature controller (Aviv Biomedical Inc., Lakewood, NJ). Samples were prepared at a concentration of 0.28 mM in 20 mM PBS (10 mM NaH 2 PO 4 , 10 mM Na 2 HPO 4 , 150 mM NaCl) at pH 7. Peptide concentrations were determined by tyrosine absorbance at 275 nm using ⑀ 275 ϭ 1400 M Ϫ1 cm Ϫ1 . Wavelength scans of the peptides were carried out by collecting the signal from 260 to 215 nm at 0.5-nm intervals at 0°C.
CD was applied to determine the thermal stability by monitoring the amplitude of the peak at 225 nm as a function of increasing temperature with an average heating rate of 0.1°C/ min (31). The peptides or mixtures were equilibrated for at least 40 h at 0°C prior to the melting experiments. The melting temperature (T m ) is defined as the temperature at which the fraction folded is equal to 0.5 in the curve fitted to the trimer to monomer transition.

Results
Peptide Design-Peptides were designed to model one natural interruption in the [␣1(IV)] 2 ␣2(IV) isoform of type IV collagen ( Fig. 1). Peptide A includes residues 289 -299 from the sequence of the human ␣1(IV) chain with a G1G interruption (underlined) ( Table 1). Residues GVG at the interruption site were selectively 15 N/ 13 C-labeled together with a labeled Gly 25 within a Gly-Pro-Hyp triplet at the C terminus. Peptide B includes residues 288 -295 from the human ␣2(IV) chain at the corresponding GISLKG interruption (underlined) with selectively 15 N/ 13 C-labeled residues ISL at the interruption site and a labeled Ala 25 near the C terminus (Table 1). All collagen sequences are flanked by stabilizing triplets and a terminal Tyr for concentration determination.

CD Characterization of Peptides A and B and the Mixture of 2A:
1B-CD studies were performed on peptides A and B separately, as homotrimers. Dissolving peptide A in solution leads to formation of a stable triple-helix with a characteristic positive CD peak at 225 nm (MRE 225 ϭ 2611 deg cm 2 /dmol Ϫ1 ) and a melting temperature of 14.9°C (Table 1, Fig. 2). Dissolving peptide B in solution only leads to partial formation of homotrimers, with a lower MRE 225 value (1150 deg cm 2 / dmol Ϫ1 ) and a partial thermal transition (Fig. 2). Mixing peptides A and B at a molar ratio of 2A:1B to model the type IV [␣1(IV)] 2 ␣2(IV) composition leads to the formation of triplehelix structure with a characteristic CD peak at 225 nm (MRE 225 ϭ 2218 deg cm 2 /dmol Ϫ1 ) and a single-mode transition with a melting temperature of 14.4°C (Fig. 2). The mixture shows a stability comparable with homotrimer A, but is much more stable than homotrimer B (Table 1).
NMR Shows That the 2A:1B Peptide Mixture Forms a Singlecomponent Heterotrimer-NMR studies were performed on the 2A:1B peptide mixture, as well as A homotrimers and B homotrimers, to evaluate whether heterotrimers had formed (Fig. 3). The 1 H-15 N HSQC spectra of peptides A and B separately and a 2A:1B peptide mixture were obtained, and the chain-specific sequential assignments were based on three-dimensional HNCA experiments. The peaks corresponding to monomer states are denoted with superscript M, whereas the trimer resonances are denoted with superscript A and B, indicating the chain composition of the trimer species (Fig. 3). Minor resonances in the HSQC spectrum arise due to cis-trans isomerization of Gly-Pro and Pro-Hyp bonds in the peptide in the unfolded states (40). The HSQC spectrum of homotrimer peptide A (green) shows three trimer peaks for labeled residues Val 15 , Gly 16 , and Gly 25 , indicating distinct chemical environ-ments in three different chains (Fig. 3A). The one-residue stagger of the three chains within the triple-helix leads to a different environment for residues in each chain. Only two trimer peaks are seen for Gly 14 ; the expected third peak may be overlapped with other resonances (Fig. 3A). The HSQC spectrum of the homotrimer peptide B (blue) shows three trimer peaks for labeled residue Leu 15 ( B Leu 15 , superscript B for trimer resonances), indicating different environments for the Leu residues located in the interruption region in each of the three chains of the homotrimer (Fig. 3B). The presence of only one or two trimer resonances for Ile ( B Ile 13 ), Ser ( B Ser 14 ), and Ala ( B Ala 25 ) is likely due to overlapping trimer resonances (Fig. 3B).
The HSQC spectrum of the 2A:1B peptide mixture is well resolved (Fig. 4a). In the spectrum for the mixture, all labeled residues in peptide A have two trimer resonances (superscript A), whereas the labeled residues in peptide B all show a single trimer resonance (superscript B), which is consistent with the composition of the heterotrimer as two A chains and one B chain (Fig. 4a). The difference in the chemical shifts of trimer peaks in the homotrimers when compared with the 2A:1B mixture indicates that the residues in polypeptide chains A and B have different chemical environments in homotrimers versus   heterotrimers (Fig. 3, A and B). These differences are clearly illustrated by the trimer resonances of Gly 25 ( A Gly 25 ) and Ile 13 ( B Ile 13 ) in Fig. 3, C and D. Three trimer resonances of Gly 25 ( A Gly-25) in homotrimer peptide A become two peaks in the 2A:1B mixture (Fig. 3C), whereas the trimer resonance of Ile 13 ( B Ile 13 ) in homotrimer peptide B is significantly shifted in the 2A:1B mixture (Fig. 3D). One monomer resonance is also observed for each residue (indicated by superscript M), supporting an equilibrium of trimers and monomers in the 2A:1B peptide mixture, which is also seen for homotrimers (Fig. 3, A  and B). No trimer resonance corresponding to the chemical shifts seen for homotrimers is found for the labeled residues in the peptide mixture, indicating that no homotrimer is formed. These results indicate that the 2A:1B peptide mixture forms a single-component heterotrimer without any observable homotrimers.
Chain Register of Heterotrimer 2A:1B-The heterotrimer formed by the peptide mixture 2A:1B has three possible registers, AAB, ABA, and BAA, with B as the lagging, middle, or leading chain, respectively. Chain assignments for 2A:1B were based on interchain NOEs, which are summarized in the experimental contact map (Fig. 4b) (41). Violet and gray colors are used to distinguish two different A chains in the heterotrimer, and chain B is colored in azure. The colored circles were utilized to represent experimentally observed NOEs (HN-HN (green circles); HN-HA (red circles); and HN-side chain protons (blue circles)). The shaded squares in the background denote intrachain NOEs. Only a few interchain NOEs could be observed including A Gly 14 NH-B Ile 13 (Fig. 4a). All other interchain NOEs supported these chain assignments. Thus, the peptide 2A:1B mixture forms a single-component heterotrimer with chain register AAB.
Hydrogen Bonding of Heterotrimer Versus Homotrimer-Amide proton temperature gradients were measured for homotrimers of A 3 and B 3 and the heterotrimer AAB (Fig. 5). Amide protons showing the temperature gradients with a more positive value than a Ϫ4.6 threshold are considered to be involved in hydrogen bonding (42,43), and this value is used as the cut-off in our analysis. For homotrimer A, Gly 25 in all three chains of triple-helix has amide proton temperature gradients ϳϪ2, indicating the formation of hydrogen bonds in this stable C-terminal Gly-Pro-Hyp region. Residues Gly 14 -Val 15 -Gly 16 at the interruption site all show temperature gradients more negative than Ϫ6, indicating that the hydrogen bonds of Gly, which are preserved for continuous Gly-X-Y region, have been broken (Fig. 5a). For homotrimer B, all labeled residues including Ile 13 , Ser 14 , and Leu 15 at the interruption sites and Ala 25 at the Tyr position in the Gly-X-Y repeating triplets show temperature gradients more negative than Ϫ6, indicating the absence of hydrogen bonds (Fig. 5b).
In the heterotrimer AAB, the amide proton temperature gradient values indicate the formation of new hydrogen bonds at the interruption site (Fig. 5c). Leu 15 in the B chain now has a temperature gradient with a value of Ϫ2.8, indicating involvement of its NH in hydrogen bonding. In addition, Gly 14 residues in the two A chains have temperature gradients with values of Ϫ4.4 and Ϫ4.6, near the cut-off value, indicating the potential hydrogen bonds. When compared with the A chain Gly 14 residues in heterotrimer AAB, the A chain Gly 16 residues have temperature gradients more negative than Ϫ6, indicating the absence of hydrogen bonding. Gly 25 in the Gly-X-Y C terminus of the heterotrimer AAB forms hydrogen bonds, as indicated by its temperature gradients with values more positive than Ϫ3.
Dynamics of Heterotrimer AAB on the Fast, Picosecond Timescale-The flexibility of residues within the heterotrimer AAB was estimated by measuring the 1 H-15 N NOE values (Fig.  6). Heteronuclear 1 H-15 N NOE measurements provide useful information on the fast, picosecond timescale motion, such as bond vibration and side-chain rotation. Smaller NOE values indicate higher flexibility on this time scale (44). Among the labeled residues of the A chains, Gly 14 -Val 15 -Gly 16 , Gly 14 is the most rigid, whereas Gly 16 is the most flexible, in both chains. Gly 25 residues in the C-terminal region of both chains are very rigid. For the B chain, the labeled residues Ile 13 -Ser 14 -Leu 15 at the interruptions sites as well as Ala 25 in the non-disrupted C-terminal region all have NOE values larger than 0.6, indicating that all of the labeled residues in the B chain are rigid. Overall, most residues at the interruption sites are rigid except for the two Gly 16 at the C terminus of the GVG interruption sites in the A chains.

Discussion
Although collagen is distinguished by its repeating (Gly-X-Y) n amino acid sequence, the triple-helix domains of non-fibrillar collagens all include a number of sites where this repeating tripeptide sequence is interrupted. For the heterotrimeric type IV collagen, ϳ20 of these sequence discontinuities are distributed throughout the triple-helix, and they may play significant roles in the organization and mechanical integrity of basement membranes. Knowledge about the effect of heterotrimeric interruptions on the triple-helix is important for understanding their biological role. NMR studies are reported here on a heterotrimeric peptide that models an interruption in both chains at residues ϳ289 -299 of the most widespread type IV collagen isoform [␣1(IV)] 2 ␣2(IV), and the results provide direct experimental evidence on chain composition, hydrogen bonding, register, and interactions between disparate length interruptions.
The 1450-residue-long triple-helix of type IV collagen has a C-terminal globular domain (NC1) responsible for chain selection, which likely also promotes triple-helix nucleation and the register in which the three chains are staggered by one residue (e.g. AAB, ABA, or BAA). In type IX collagen, the non-collagenous NC2 domain has been shown to be crucial to control of the chain register (45). Some non-fibrillar collagens, such as type X collagen, are homotrimers, with the same interruptions in all three chains, so that chain register can be preserved on both sides. However, in heterotrimeric type IV collagens, there are almost always disparate lengths of interruptions in different chains at a given site (Fig. 7a). These disparate lengths can be either commensurate, maintaining the three chains in the same register at either side of the interruption, or non-commensurate, where the chain register is not maintained (46). The interruption in our peptide model, with a G1G interrup-

tion (GVG) in two chains and a G4G interruption (GISLKG)
in the third chain, would be categorized as commensurate, because Gly residues could be maintained in the same register on both sides (46).
The CD spectrum and thermal transition indicate that the 2A:1B peptide mixture is forming a largely triple-helical structure, consistent with triple-helix forming on both sides of the interruption. NMR studies indicated, surprisingly, that the 2A:1B peptide mixture formed only one type of triple-helical molecule, with composition A 2 B. No A 3 homotrimers were observed, although they have a slightly higher stability than the 2A:1B mixture, nor were B 3 homotrimers, a much less stable species, or AB 2 molecules observed. Within the triple-helix structure, the three chains are staggered by one residue with respect to each other, and the NOEs showed the register for this heterotrimeric peptide to be AAB, or ␣1␣1␣2(IV) (leading, middle, trailing). Interestingly, the register for the type IV [␣1(IV)] 2 ␣2(IV) molecule was experimentally determined for one other site in type IV collagen (47). Golbik et al. (47) used FRET on a tryptic fragment to define the stagger of type IV (Gly-X-Y) n residues 460 -480 (which includes the ␣1␤1 integrin binding site) and found it to be ␣1␣1␣2(IV), the same observed in our heterotrimeric peptide. It was unexpected to find the same register for these two different sites separated by ϳ200 residues and by four additional interruptions, including three that are non-commensurate. The observation of the same register at two distinct and separate locations raises the possibility that the ␣1␣1␣2(IV) register is maintained throughout the entire type IV triple-helix, despite the many non-commensurate interruption sites. The ability of the small A and B peptides studied here to self-assemble into a heterotrimer triplehelix with a unique chain composition and register suggests  that this particular G1G/G1G/G4G (␣1/␣1/␣2) interruption plays a role in preserving the type IV collagen register and may facilitate triple-helix propagation.
The current study suggests that hydrogen bonding may play a key role in defining chain composition and the precise chain register within this heterotrimer peptide. Interchain hydrogen bonding involving Gly NH of one chain and OϭC in the X residue on another chain is known to be a main driving force for triple-helix stabilization (48). The homotrimer A 3 , which forms a stable triple-helix structure (T m ϳ15°C), has amide proton gradient values reflecting hydrogen bonding for the Gly NH in the C-terminal Gly-Pro-Hyp region, but no hydrogen bonding for any residues in the Gly-Val-Gly interruption. This is surprising because there is at least one Gly residue at every axial level for the A 3 homotrimer, and direct H-bonds might be expected (Fig. 7b) (46). For the weakly stable homotrimer B 3 peptide, no residues in the ISLK interruption site form hydrogen bonds. This may be due to the lack of Gly residues at two axial levels for the B 3 homotrimer (see boxed axial triplets for B 3 in Fig. 7b). In contrast, within the heterotrimer context of AAB, Gly 14 (A) in the leading and middle chains form hydrogen bonds, and Leu 15 (B) in the lagging chain does as well. A recent analysis by Bella (46) proposed that the most favorable register for a heterotrimeric interruption consisting of two G1G and one G4G would be AAB (␣1␣1␣2) because the presence of a Gly at every axial level would allow maximum hydrogen bonding. This predicted favorable register is observed for the AAB peptide (Fig. 7b), but the expected H-bonds are not observed for Gly 16 .
An unexpected hydrogen bond was observed between the amide group of Leu 15 in chain B and the carbonyl in an adjacent chain A (Fig. 7c). The Leu within the ␣2(IV) interruption sequence GISLKG occupies the position normally occupied by a Gly residue, and this residue is predominantly hydrophobic within all such G4G interruptions (11,41). Previous NMR studies showed that a typical G4G interruption GAAVMG within a homotrimer context allowed the preservation of the one-residue stagger and close packing of the three chains, with a very local alteration of typical triple-helix dihedral angles leading to a small hydrophobic cluster in the center of the triple-helix (41). In the heterotrimer, the Leu from the ␣2(IV) chain may have such hydrophobic interactions with the Val chains from ␣1(IV) chains, with the buried Leu forming a backbone hydrogen bond to another chain as well.
Although the appearance of hydrogen bonds in the heterotrimer that are absent in homotrimers is striking, there may well be contributions from Gly-X-Y sequences surrounding the interruption determining heterotrimer chain composition and stagger. It has been reported that atypical sequences are found The presence of a one-residue GVG interruption in two chains and a four-residue GISLKG interruption in the other two chains allows maintenance of the stagger on both sides of the interruption (com-mensurate). Gly residues are shown in red to visualize the stagger, and interruptions are underlined. Black boxes are used to designate an axial level where there is no Gly and the typical stabilizing interchain Gly NH . . . CO bonds cannot be formed. Green boxes are used to exemplify a regular axial level that contains one and only one Gly with the expected formation of a hydrogen bond. Only AAB allows hydrogen bonding at all axial positions, as proposed by Bella (46). c, hydrogen bonding at the interruption site GVG/GVG/ISLK in heterotrimer AAB deduced from amide temperature-dependent NMR studies, with hydrogen bonds formed between the amide group of Gly 14 /Leu 16 and the carbonyl group of nearby amino acids indicated as dashed lines. adjacent to interruptions in type IV collagen (47), which could play a role. In addition, electrostatic interactions have the capacity to control chain selection (48), and the presence of oppositely charged residues at two sites in A versus B chains may make some contribution to stability.
The dynamics at interruption sites is of interest because interruptions are thought to be associated with looseness or flexibility based in part on rotary shadowing electron microscopy and on the known cleavage of two G4G sites in type X collagen by MMP-9 (12,14). All labeled residues within the G1G/G1G/G4G interruption studied here appear to be rigid on the fast timescale, reflecting bond vibration and side-chain rotation, except for Gly 16 in the two A chains (Fig. 6). The Gly 16 residues do not show hydrogen bonding and are the most flexible residues when compared with the more rigid Gly 14 on the N-terminal side of the Gly 14 -Val 15 -Gly 16 sequence, which does form hydrogen bonds. This is consistent with the high resolution x-ray structure of a homotrimer peptide with a GPG interruption, where the Gly N-terminal to the Pro shows standard hydrogen bonding, whereas the Gly C-terminal to the Pro does not (49). It is possible that the mobility seen for Gly 16 could have some functional significance in binding or cleavage.
Although type IV collagen contains more than 20 interruptions in each chain, it has a similar melting temperature as type I collagen without any breaks. Peptides, such as those studied in this report, magnify the destabilizing effect of such interruptions and allow biophysical studies to characterize distinctive molecular features. Our studies suggest that complementary interruptions in different chains may provide specificity and selectivity within the triple-helix as well as serving as external recognition sites for MMPs or receptors. MMP-9 cleaves a Gly-Leu or Gly-Ile in a G4G interruption in type X collagen (14). Type X is a homotrimer with G4G interruptions in all three chains, and it cannot have H-bonds at three axial levels. The NMR studies reported here suggest that the structure in the homotrimer is looser and therefore may be more exposed than the G1G/G4G case in heterotrimer.
Jalan and Hartgerink (1) raised the possibility that the triplehelix itself has the capacity for chain selection and chain registration, and have employed an electrostatic strategy to promote formation of heterotrimers in (Gly-X-Y) n peptides. Here, we have shown that heterotrimers can be produced by peptide models with natural interruptions in collagen and that hydrogen bonding, as well as electrostatic interactions, may provide a driving force for heterotrimer formation. We suggest that the data presented here reframe our understanding of at least some interruptions in non-fibrillar collagens. Previously, interruptions have been viewed as "obstacles" to folding, which may either passively allow the chain register to continue (commensurate) or alter the register (non-commensurate). We propose that the interactions between interruptions in different chains, together with atypical surrounding sequences, may play a role in facilitating folding around these interruptions. The heterotrimeric interruption studied here does not represent an "obstacle" to folding and a "disruption" of register, but rather a stable, relatively rigid site whose interactions favor this register and chain interaction. We hypothesize that heterotrimeric interruptions in type IV collagen contain important informa-tion that supports the establishment and maintenance of chain register. Definition of the basement membrane collagen structure is important for understanding the normal functioning of basement membranes and their role in kidney diseases and tumor metastases, as well as forming substrates for stem cell growth in tissue engineering.
Author Contributions-J. B. and B. B. conceived and coordinated the study and wrote the paper. J. X. designed, performed, and analyzed the experiments. X. S. contributed to the analysis of data and the preparation of the figures. M. B. performed and analyzed the experiments shown in Table 1. All authors reviewed the results and approved the final version of the manuscript.