Search tips
Search criteria 


Logo of jbcThe Journal of Biological Chemistry
J Biol Chem. 2010 October 29; 285(44): 33764–33770.
Published online 2010 August 24. doi:  10.1074/jbc.M110.139394
PMCID: PMC2962475

Implications for Collagen Binding from the Crystallographic Structure of Fibronectin 6FnI1–2FnII7FnIAn external file that holds a picture, illustration, etc.
Object name is sbox.jpg


Collagen and fibronectin (FN) are two abundant and essential components of the vertebrate extracellular matrix; they interact directly with cellular receptors and affect cell adhesion and migration. Past studies identified a FN fragment comprising six modules, 6FnI1–2FnII7–9FnI, and termed the gelatin binding domain (GBD) as responsible for collagen interaction. Recently, we showed that the GBD binds tightly to a specific site within type I collagen and determined the structure of domains 8–9FnI in complex with a peptide from that site. Here, we present the crystallographic structure of domains 6FnI1–2FnII7FnI, which form a compact, globular unit through interdomain interactions. Analysis of NMR titrations with single-stranded collagen peptides reveals a dominant collagen interaction surface on domains 2FnII and 7FnI; a similar surface appears involved in interactions with triple-helical peptides. Models of the complete GBD, based on the new structure and the 8–9FnI·collagen complex show a continuous putative collagen binding surface. We explore the implications of this model using long collagen peptides and discuss our findings in the context of FN interactions with collagen fibrils.

Keywords: Collagen, Extracellular Matrix Proteins, Fibronectin, NMR, Protein Structure, X-ray Crystallography, Gelatin Binding Domain, Protein Interactions


Collagen fibrils are the basis of vertebrate tissue, and their formation is vital for cell differentiation, cell migration, and embryonic development (1). The most abundant form, collagen type I, consists of two chains of α1 and one of α2 that intertwine to create right-handed triple helices, which then assemble to make microfibrils and fibers (2, 3). Fibril formation of type I collagen in vivo requires integrin receptors and fibronectin (FN),5 a large glycoprotein composed of three domain classes, FnI, FnII, and FnIII (4, 5). FN forms fibrils in the extracellular matrix and interacts with denatured collagen (gelatin) or isolated collagen chains. The interaction site on FN is the gelatin binding domain (GBD), consisting of four FnI and two FnII modules (6FnI1–2FnII7–9FnI) (6). The first structure of a FN·collagen complex, a 8–9FnI fragment together with a single-stranded collagen peptide was recently determined (7). This structure and subsequent biophysical analysis suggested that 8–9FnI preferentially binds to unwound collagen. Interestingly, the proposed binding site for FN on collagen I coincides with the cleavage site for metalloproteinase 1, the so-called collagenase site (8,10). This site is hydrophobic and relatively low in proline and hydroxyproline residues, rendering the triple helix unstable at physiological temperature (11). Although these results offered some answers on FN·collagen binding, the role of the remaining four domains in GBD was unclear. The solution NMR structure of 6FnI1–2FnII (12) suggested that 1FnII reorients flexibly with respect to 6FnI-2FnII. However, 6FnI1–2FnII shows substantially decreased binding to gelatin (13) and collagen peptides (14) compared with 6FnI1–2FnII7FnI, implying a significant role for the 7FnI domain in this interaction. Here, we present the crystallographic structure of the 6FnI1–2FnII7FnI fragment at a resolution of 3.0 Å. We compare this new structure with the previous solution structure of 6FnI1–2FnII and show that 1FnII and 2FnII jointly form an interface for 7FnI; the presence of this interface in solution is supported by analysis of the NMR chemical shifts of these fragments. NMR titrations with single-stranded as well as triple-helical peptides from the collagenase site show that a unique binding surface, involving domains 2FnII and 7FnI, is important for collagen binding. Together with previously published data (7), we now offer a model for the complete GBD which suggests that the two GBD subfragments bind collagen in a concerted fashion.


Protein Expression, Purification, and Crystallization

A gene fragment, encoding FN residues 305–515, corresponding to domains 6FnI1–2FnII7FnI, and bearing a single amino acid substitution (H307D) and a C-terminal His6 tag (GTKHHHHHH) was integrated in Pichia pastoris in a manner analogous to that previously described (15). The H307D substitution does not affect the binding of 6FnI1–2FnII7FnI to gelatin-Sepharose columns (data not shown) but increases the solubility of this fragment under physiological pH conditions significantly. Cells were grown under high density fermentation conditions, using a minimal phosphate medium at pH 3.25 and 30 °C; protein expression was induced by the addition of methanol over the course of 5–7 days. 6FnI1–2FnII7FnI was purified from the expression media by metal affinity chromatography, and the N-linked glycan at Asn430 was truncated by endoglycosidase H treatment, leaving a single N-acetylglucosamine at this site. The protein was further purified by reverse phase HPLC, lyophilized, and finally run on a Sephadex G-75 size exclusion column equilibrated in a 100 mm NaCl, 10 mm HEPES, pH 7.2, buffer. 6FnI1–2FnII7FnI was concentrated in this buffer to ~25 mg/ml.

Crystallization drops were formed by 1:1 mixtures of protein in the final purification buffer at 17.5 mg/ml concentration, and a 0.2 m (NH4)H2PO4, 0.1 m Tris-Cl, pH 8.5, 50% v/v 2-methyl-2,4-pentanediol solution. Sitting-drop vapor diffusion was employed in 96-well plates, with drop volumes varying between 200 and 400 nl. 6FnI1–2FnII7FnI crystals of lens-like appearance and ~200 μm in the longest axis developed after 6–8 weeks at 20 °C. Crystals were cryoprotected by brief immersion in crystallization mother liquor and frozen in a nitrogen cryostream at 100 K.

X-ray Data Collection and Processing, Structure Determination

Data were collected under cryogenic conditions at the PXIII macromolecular crystallography beamline at the Swiss Light Source (Villigen, Switzerland). Reflection data were indexed by LABELIT (16), refined and integrated in XDS (17), and merged by SCALA (18). The Laue group and space group were suggested by POINTLESS (18) from the unmerged data, and data quality was assessed by PHENIX.xtriage (19) (supplemental Table 1).

Initial structure determination was performed by molecular replacement using PHASER (20) and superimposed ensembles of FnI- and FnII-type modules of known crystallographic structures. Iterative cycles of model building in COOT (21) and refinement using PHENIX.refine (19) with TLS restraints (one chain) resulted in a final model with satisfactory Rwork/Rfree and MolProbity (22) statistics (supplemental Table 1).

NMR Data Collection and Assignments

Details of the 6FnI1–2FnII7FnI sample preparation for NMR experiments as well as the titration with collagen peptides were described earlier (7). The triple-helical α1(II) Gly775-Ser801 peptide was synthesized and verified as described (23, 24) and was then allowed to anneal for 24 h at 4 °C at high concentrations (>1 mm) prior to use. Sequence-specific chemical shift assignments were performed using a uniformly 13C15N-enriched 6FnI1–2FnII7FnI sample of ~1 mm concentration in a 20 mm Na2HPO4, pH 7.2, buffer at 37 °C. Home-built or Bruker Avance II spectrometers were used for a combination of standard through-bond triple-resonance experiments (25) supplemented by through-bond experiments for assignments of aromatic side chains (26).

Circular Dichroism (CD) Spectroscopy

CD spectra were collected using Jasco J-720 or AppliedPhotphysics Chirascan spectropolarimeters with a 0.1-cm path length. Peptide samples of 50 μm1(II) Gly775-Ser801) or 300 μm1(I) Gly778-Arg816) concentration in an aqueous buffer containing 150 mm NaCl, 20 mm Na2HPO4, pH 7.4, were used.

Isothermal Titration Calorimetry (ITC)

Recombinant 8–9FnI and GBD were prepared as described earlier (7). The α1(I) Gly778-Arg816 peptide was purchased from GL Biochem (Shanghai) as HPLC-purified, lyophilized powder. Protein concentration was established by UV absorbance at 280 nm, and peptide concentration was initially estimated from dry weight. Protein solutions were dialyzed overnight against 150 mm NaCl, 20 mm Na2HPO4, pH 7.2. The pH of peptide solutions was adjusted with 1 m NaOH to match this buffer. ITC experiments (VP-ITC; MicroCal) were performed as follows: one injection of 2 μl followed by 44 injections of 5 μl at 0.5 μl/s. The stirring speed was 307 rpm; the delay between the injections was 210 s. To take into account heats of dilution, blank titrations were performed by injecting peptide solution into buffer, and the averaged heat of dilution was subtracted from the main experiment. Raw data were processed and fitted to a one-site model using MicroCal Origin software.

Data Deposition

Amino acid composition and numbering for FN fragments correspond to UniProt entry B7ZLF0. α1(I) and α1(II) numbering (accession numbers P02452 and P02458, respectively) begins at the estimated start of the helical region. O in peptide sequences denotes 4-hydroxyproline. Structural analysis was performed, and figures were prepared using PyMOL (27). Interdomain interactions were analyzed with the PISA service from the European Bioinformatics Institute (28). Structural data have been deposited in the Protein Data Bank under accession number 3MQL, and NMR chemical shift assignments are available in the BioMagResBank under accession number 16841.


Structure of 6FnI1–2FnII7FnI

A FN fragment spanning domains 6FnI1–2FnII7FnI crystallized using the sitting-drop vapor diffusion method as described under “Experimental Procedures.” Lens-like birefringent objects grew slowly out of initial light precipitate over the course of ~2 months at 20 °C. Despite the lack of well defined edges, these objects were crystalline in nature and diffracted to ~3.5 Å resolution in synchrotron x-ray sources. Screening around the initial crystallization conditions or use of additives did not improve the morphology of these crystals; however, it was possible to collect a complete 3.0 Å dataset (supplemental Table 1) by screening different crystals from the original conditions. The Matthews coefficient strongly suggested a single 6FnI1–2FnII7FnI molecule/asymmetric unit, and the structure was solved by molecular replacement using two copies of FnI- and FnII-type structural ensembles for a total of four search objects. The initial molecular replacement map was of sufficient quality to allow tracing of interdomain linkers and to establish domain identity and connectivity.

All individual 6FnI1–2FnII7FnI domains adopt canonical structures (12, 29), a result anticipated based on chemical shift index analysis (30) of this fragment in solution. FnI-type domains form a β-sandwich with antiparallel two-stranded and three-stranded β-sheets. FnII-type domains are characterized by extensive loop segments and only two short antiparallel two-stranded sheets. Both FnI- and FnII-type domains feature two disulfide bridges that contribute substantially to these domain folds.

In contrast to common beads-on-string models of multidomain proteins (31), the crystallographic structure of 6FnI1–2FnII7FnI adopts a pyramidal shape. As shown in Fig. 1A 6FnI1–2FnII forms an approximately equilateral triangle with sides of ~ 35 Å. From this base, 7FnI projects out by ~38 Å (Fig. 1B) and forms extensive hydrogen-bonding interactions with a 7FnI domain from a crystallographic 2-fold symmetry-related molecule. Other crystal contacts include further β-sheet extensions through 7FnI-7FnI and 6FnI-6FnI interactions; however, it should be noted that similar extensions are common among crystallographic structures of FnI-type domains (29, 32). No evidence of protein oligomerization was apparent in solution NMR experiments even at sample concentrations of >1 mm (data not shown).

A, crystal structure of 6FnI1–2FnII7FnI shows domain 6FnI packing against 1–2FnII to form a compact triagonal shape. B, 7FnI, in contrast, protrudes at an approximately 90° angle, presenting a large interface for potential interaction ...

The compact 6FnI1–2FnII7FnI conformation is maintained through interdomain interactions that show a remarkable degree of conservation (supplemental Fig. 1). 6FnI interacts with 2FnII in a manner essentially identical to that observed in the solution 6FnI1–2FnII structure (12), burying ~395 Å2 of solvent-accessible surface area (Fig. 2A). The Cα root mean square deviation for 6FnI·2FnII between the solution structure and our crystallographic model is only 1.7 Å, whereas the individual domains differ by 1.0 and 1.1 Å for 6FnI and 2FnII, respectively. Residues 314–323 of 6FnI, and 414–421 and 448–449 of 2FnII are primarily involved in forming the interface, with significant contributions from Met320, Ser415, Ala418, Leu419, Thr448, and Thr449 (Fig. 2A).

Prominent domain-domain contacts involving 6FnI-2FnII (A), 1FnII-2FnII (B), and 1–2FnII-7FnI (C). Individual domains are colored as in Fig. 1, and specific residues are shown as sticks colored similar to their respective domains. Compared with ...

The solution structure of 6FnI1–2FnII featured a well defined 1FnII domain which was, however, mobile in relation to the 6FnI·2FnII complex (12). In contrast, our crystallographic model shows the formation of a 1FnII-2FnII interface (Fig. 2B) burying ~330 Å2. 1FnII rotates and translates toward 2FnII and 7FnI (Fig. 3, A and B, and below); this motion places 1FnII outside the ensemble of conformations shown in the solution structure of 6FnI1–2FnII. The 1FnII-2FnII interface involves primarily Tyr372, Val408, and Pro462 as well as long range hydrogen bonds between Val315 O′-Asn416 Nδ2, Gln321 O′-Leu419 N, and Tyr316 Oη-Asn416 O′. In addition, 1FnII helps to structure the relatively long linker (residues 461–468) connecting 2FnII and 7FnI, which in turn stabilizes the 7FnI conformation relative to the remaining domains. 7FnI interacts with both 1FnII and 2FnII across an interface burying ~390 Å2 of solvent-accessible surface area (Fig. 2C). Residues Thr365, Ser390, Asn391, Met463, Ala464, His466, Ile469, and Gly502 are involved in hydrophobic burial and hydrogen bonding interactions, whereas Arg479 and Ile480 help anchor 7FnI to the structured 2FnII-7FnI linker (Fig. 2C).

A and B, comparison of 6FnI1–2FnII7FnI (green) with the solution structure of 6FnI1–2FnII (12) (blue) in two orientations. 1FnII is mobile with respect to 6FnI·2FnII in the structure of 6FnI1–2FnII; only the best model ...

Crystallographic versus Solution 6FnI1–2FnII7FnI Conformation

To evaluate whether the 6FnI1–2FnII7FnI conformation observed is present in solution, we compared the NMR chemical shifts of 6FnI1–2FnII with those of 6FnI1–2FnII7FnI under the same experimental conditions. Addition of 7FnI to the fragment causes chemical shift differences that extend further than the immediate attachment point (i.e. the C terminus of 2FnII; see supplemental Fig. 2). Fig. 3, C and D, shows residues whose NMR resonances differ by more than 1 S.D. compared with the average; these include: residues at the 6FnI1–2FnII interdomain linkers (Ala346, Thr348, Gln349, Thr402, Asp403, and Leu407); residues at the new 1FnII-2FnII interface (Tyr372, Cys434) and numerous residues on both 1FnII and 2FnII that circumscribe the 7FnI anchoring point (Fig. 3D). These differences correlate well with differences between the solution 6FnI1–2FnII structure and our model and are likely to report on the structural changes induced by the addition of 7FnI to the construct. Comparison of heteronuclear {1H}-15N NOE data, which report on fast time scale NMR dynamics, between 6FnI1–2FnII (12) at 25 °C and 6FnI1–2FnII7FnI at 37 °C (supplemental Fig. 3) shows moderate stabilization of both 6FnI-1FnII and 1FnII-2FnII loops in the larger construct. Specifically, the lowest {1H}-15N NOE values observed in these loops are 0.4 and 0.5, respectively, for 6FnI1–2FnII, whereas the equivalent values for 6FnI1–2FnII7FnI are 0.52 and 0.68. For reference, {1H}-NOE values below 0.6–0.65 typically denote flexible residues (33). Thus, although a measure of flexibility remains in 6FnI1–2FnII7FnI we argue that the compact conformation seen in the crystal structure is also present in solution.

6FnI1–2FnII7FnI Interactions with Collagen Peptides

Previously, we reported that a single-stranded peptide derived from the collagen α1(I) chain, spanning residues Gly778-Gly799 (GQRGVVGLOGQRGERGFOGLOG), interacts with 6FnI1–2FnII7FnI with a Kd ~ 60 μm at 37 °C (7). NMR spectra under near physiological conditions showed significant chemical shift perturbations over a large number of residues, indicating an extensive binding surface. The interaction time scale was estimated to be in the microsecond range based on spectral properties. However, the lack of sequential assignments and a structural model for 6FnI1–2FnII7FnI prevented further analysis of these data at the time.

As shown in supplemental Fig. 2, chemical shift perturbations larger than 2 S.D. compared with the average (red bars) localize exclusively on residues of 2FnII7FnI, and smaller but significant perturbations (1–2 S.D., yellow bars) largely follow the same pattern. Mapping these perturbations on the 6FnI1–2FnII7FnI structure shows a dominant interaction surface (Fig. 4) spanning domains 2FnII7FnI and primarily involving residues Asn427, Tyr452, Phe458, Gly500, Arg503, Gly504, Trp506, and Thr507. Although 1FnII does not participate in this interaction interface it stabilizes the 7FnI orientation thereby contributing to the formation of a continuous 2FnII7FnI binding surface.

6FnI1–2FnII7FnI residues perturbed by single-stranded α1(I) Gly778-Gly799 binding. Residues with chemical shift perturbations larger than 2 S.D. compared with the average are shown in red and are labeled. Residues perturbed between 1 and ...

Our previous study of collagen peptides binding to GBD fragments (7) and work since then have identified α1(I) residues Gly778-Gly799 as the single-stranded peptide with the highest known affinity for 6FnI1–2FnII7FnI in type I collagen. This peptide is adjacent to the matrix metalloproteinase 1 cleavage site and coincides with the collagen fragment implicated in FN binding using fluorescent probes (6) as well as competition assays between serum FN and collagen (34). Several other peptides are known to bind with low affinity, with Kd values of well over 1 mm (7), but these tend to cause few chemical shift perturbations. Analysis of these weak interactions showed effects primarily on Trp385 (1FnII), Trp445 (2FnII), and residues in their vicinity. These residues belong to a hydrophobic pocket on the surface of FnII-type modules that is known to interact weakly with collagen-like peptides (14, 35); in our crystallographic model the same sites are occupied by two 2-methyl-2,4-pentanediol molecules from the crystallization solution. We believe that these lower affinity interactions correspond to nonspecific binding events, possibly related to the generic gelatin affinity displayed by many FnII-type modules (36,38). In contrast, our higher affinity α1(I) Gly778-Gly799 peptide interacts with a unique interface involving only one of these two hydrophobic pockets; the peptide is likely to be a specific ligand.

To test whether native-like, triple-helical peptides interact with 6FnI1–2FnII7FnI in a similar fashion, we performed NMR titrations using a synthetic peptide based on the homotrimeric collagen type II sequence. This fragment, spanning residues Gly775-Ser801 of the α1(II) chain (GPC(GPP)5GLAGQRGIVGLOGQRGERGFOGLOGPS(GPP)5GPC), is highly homologous to our high affinity α1(I) peptide and includes the matrix metalloproteinase 1 cleavage site. Previous work (7, 23), and CD data shown here (supplemental Fig. 4) confirm that this peptide adopts a stable triple-helical conformation at room temperature. NMR spectra of 6FnI1–2FnII7FnI with this peptide at 25 °C and 950 MHz 1H frequency (supplemental Fig. 5) show extensive line broadening for many residues. Varying the temperature between 15 and 37 °C did not produce a marked difference in spectral appearance. Control titrations using (GPP)10 triple-helical peptides did not show broadening of 6FnI1–2FnII7FnI residues under the same conditions (data not shown). Mapping the most affected resonances on the primary sequence (supplemental Fig. 1) produces an image that is qualitatively similar to that of the interaction with the single-stranded peptide, with the vast majority of perturbations seen on domains 2FnII and 7FnI. Compared with the single-stranded peptide, the line broadening observed here indicates that this interaction probably occurs in a slower time scale (microseconds to milliseconds). Spectral broadening prevented us from measuring an affinity value for this peptide by NMR.

GBD Modeling and Interactions with Long Collagen Peptides

Previously, we showed that NMR spectra of the GBD overlay well with spectra of the individual 6FnI1–2FnII7FnI and 8–9FnI subfragments, especially in the presence of collagen peptides (7). This overlay indicates that the GBD structure does not feature radical rearrangement of domains compared with the subfragments. Rather, the GBD is likely to comprise a linear combination of 6FnI1–2FnII7FnI and 8–9FnI connected through the 7–8FnI interface. We reconstructed this missing interface using the relative FnI domain orientation observed in almost all structures of FnI-domain pairs (7, 29, 32, 39) as guidance; the resulting GBD model is presented in Fig. 5. We predict that the GBD will adopt an elongated conformation of ~100 Å across the longest axis and ~35 Å maximal width at the 6FnI1–2FnII base. The overall chain direction between 6FnI and 9FnI differs by ~90°, a significant change compared with the GBD hairpin models previously considered (12). As seen in Fig. 5, the 8–9FnI-bound collagen peptide (7) is co-linear with the collagen binding interface in 6FnI1–2FnII7FnI, indicating the presence of a single continuous binding surface. 8–9FnI binds collagen in an antiparallel orientation; the last collagen residue interacting with 8–9FnI is Arg792 of α1(I), which suggests that residues C-terminal to Arg972 would bind 6FnI1–2FnII7FnI. However, isolated peptides spanning α1(I) residues Gly796-Arg816 interact with 6FnI1–2FnII7FnI only very weakly as judged by NMR (7).

Model of the fibronectin GBD. 6FnI1–2FnII7FnI is shown with collagen-binding residues marked as in Fig. 4, 8–9FnI is shown in complex with a single-stranded collagen peptide (magenta) (7), and the 7FnI-8FnI conformation is based on available ...

To test whether 6FnI1–2FnII7FnI and 8–9FnI can bind collagen in a concerted fashion in the context of the GBD, we performed ITC experiments using GBD, 8–9FnI, and a long peptide, α1(I) residues Gly778-Arg816. This peptide includes the high affinity 8–9FnI epitope and extends to potentially cover the 6FnI1–2FnII7FnI collagen binding surface observed here (Fig. 4). Although this peptide was not designed to form triple helices, a degree of helical structure was detected by CD at conditions similar to those used for ITC (supplemental Fig. 4). The partial helical structure of α1(I) Gly778-Arg816 may impede 8–9FnI binding as it will necessitate peptide unwinding and adoption of an extended conformation prior to interaction (7). As seen in supplemental Fig. 6, 8–9FnI binds α1(I) Gly778-Arg816 with relatively weak affinity (Kd ~115 μm). However, under the same conditions, GBD bound α1(I) Gly778-Arg816 ~3 times more tightly. Although the increase in affinity is moderate, it is in agreement with our model for concerted binding of the two GBD subfragments to collagen.


8–9FnI was previously shown to interact tightly (Kd ~5 μm) with a specific collagen peptide derived from the α1(I) chain (7). Crystallographic analysis of this complex also allowed us to identify a number of putative sites for 8–9FnI on that chain. The role of 6FnI1–2FnII7FnI in the GBD, however, remained unknown. Here, we presented the crystal structure of 6FnI1–2FnII7FnI, which together with the structure of 8–9FnI (7) and extensive NMR titration analysis allowed us to model the structure of the full GBD of FN. Our results offer new insights into how FN may interact with collagen.

GBD adopts an elongated structure in our model in which the two spatially distinct subfragments show good binding to the same collagen type I sequence, residues Gly788-Gly799. A question thus emerges as to how both 6FnI1–2FnII7FnI and 8–9FnI can associate with the same epitope. Previous hypotheses included GBD binding to two of the three strands in a single triple helix, or GBD associating with multiple strands across the collagen microfibril (7). However, neither of these hypotheses is plausible as they would necessitate collagen chain displacements beyond the local fluctuations believed to be present under physiological conditions (40).

An alternative hypothesis emerges from our GBD model and the ITC data: a continuous collagen binding site on GBD can interact with a single long collagen epitope in a concerted fashion. Although the available evidence is tentative, the interplay between the two GBD subfragments in binding collagen chains could facilitate tight attachment. It is intriguing to speculate on the relative contributions of the GBD subfragments to such a binding event. 6FnI1–2FnII7FnI was shown here, and earlier (7), to form relatively weak, transient complexes with single-stranded and triple-helical forms of collagen peptides; in contrast 8–9FnI binds more tightly to unwound collagen strands. There is a growing body of evidence that weak, transient encounter complexes precede formation of tight intermolecular complexes in biological systems (41, 42). Thus, it is possible that 6FnI1–2FnII7FnI binds adjacent to the collagenase site and shifts the local fluctuations of the collagen triple helix toward the unwound state (40, 43,45); this state is then bound tightly by 8–9FnI. Indeed, our ITC data suggest that the complete GBD can bind partly helical collagen peptides better than 8–9FnI.

Recently, Graille et al. determined the crystallographic structure of the GBD in the presence of millimolar level concentrations of Zn2+ (46). Under these conditions GBD forms a compact homodimer in which domain 8FnI loses the canonical FnI fold; instead 8FnI forms β-strand extension-type interactions with domains 7FnI and 9FnI. Experiments with fluorescently labeled collagen peptides showed that Zn2+ at concentrations of hundreds of micromolar interferes with collagen binding by the GBD. These concentrations are much higher than the physiological levels of Zn2+ in blood, 10–15 μm (47,49), although Zn2+ levels may be higher locally in specific tissues. Thus, Zn2+ could play a regulatory role in the FN-collagen interaction through conformational changes; in this case the GBD model we propose here would correspond to the collagen binding form of FN.

The role of FN-collagen binding in vivo has been assessed over many years; putative FN roles include scavenging of collagen fragments (50,52), stabilization and protection of collagen fibrils (4), and acting as a molecular tag for collagen proteolysis. FN has also been reported to be an opsonic protein (53, 54), enhancing phagocytosis when bound to a target. Numerous examples of FN localization to damaged tissue are known, including localization to burned skin as well as injured liver (50, 55,57). Given the ubiquitous distribution of FN in the extracellular matrix as well as in plasma, we believe that the FN-collagen interaction may serve more than one purpose. Considering the widespread interest in extracellular matrix components and their interactions in the field of biomaterials and prosthetics (58, 59), our structural data may have significant influence on future biotechnological as well as medical studies.

Supplementary Material

Supplemental Data:


We thank Drs. Vincent Olieric and Meitian Wang for help with at the SLS crystallographic beamline and Nick Soffe and Dr. Jonathan Boyd for upkeep of the Oxford Biochemistry NMR instrumentation.

An external file that holds a picture, illustration, etc.
Object name is sbox.jpgThe on-line version of this article (available at contains supplemental Figs. 1–6 and Table 1.

The atomic coordinates and structure factors (code 3MQL) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (

The NMR chemical shifts have been deposited in the BioMagResBank, (accession no. 16841).

5The abbreviations used are:

collagen type I/II α1 chain
FN type I/II/III domains
gelatin binding domain
isothermal titration calorimetry.


1. Leitinger B., Hohenester E. (2007) Matrix Biol. 26, 146–155 [PubMed]
2. Orgel J. P., Irving T. C., Miller A., Wess T. J. (2006) Proc. Natl. Acad. Sci. U.S.A. 103, 9001–9005 [PubMed]
3. Holmes D. F., Kadler K. E. (2006) Proc. Natl. Acad. Sci. U.S.A. 103, 17249–17254 [PubMed]
4. Kadler K. E., Hill A., Canty-Laird E. G. (2008) Curr. Opin. Cell Biol. 20, 495–501 [PMC free article] [PubMed]
5. Vakonakis I., Campbell I. D. (2007) Curr. Opin. Cell Biol. 19, 578–583 [PMC free article] [PubMed]
6. Ingham K. C., Brew S. A., Isaacs B. S. (1988) J. Biol. Chem. 263, 4624–4628 [PubMed]
7. Erat M. C., Slatter D. A., Lowe E. D., Millard C. J., Farndale R. W., Campbell I. D., Vakonakis I. (2009) Proc. Natl. Acad. Sci. U.S.A. 106, 4195–4200 [PubMed]
8. Fields G. B. (1991) J. Theor. Biol. 153, 585–602 [PubMed]
9. Ingham K. C., Landwehr R., Engel J. (1985) Eur. J. Biochem. 148, 219–224 [PubMed]
10. Nagase H., Fields G. B. (1996) Biopolymers 40, 399–416 [PubMed]
11. Leikina E., Mertts M. V., Kuznetsova N., Leikin S. (2002) Proc. Natl. Acad. Sci. U.S.A. 99, 1314–1318 [PubMed]
12. Pickford A. R., Smith S. P., Staunton D., Boyd J., Campbell I. D. (2001) EMBO J. 20, 1519–1529 [PubMed]
13. Katagiri Y., Brew S. A., Ingham K. C. (2003) J. Biol. Chem. 278, 11897–11902 [PubMed]
14. Pagett A., Campbell I. D., Pickford A. R. (2005) Biochemistry 44, 14682–14687 [PubMed]
15. Vakonakis I., Langenhan T., Prömel S., Russ A., Campbell I. D. (2008) Structure 16, 944–953 [PMC free article] [PubMed]
16. Sauter N. K., Grosse-Kunstleve R. W., Adams P. D. (2004) J. Appl. Crystallogr. 37, 399–409 [PMC free article] [PubMed]
17. Kabsch W. (2010) Acta Crystallogr. D. Biol. Crystallogr. 66, 125–132 [PMC free article] [PubMed]
18. Evans P. (2006) Acta Crystallogr. D. Biol. Crystallogr. 62, 72–82 [PubMed]
19. Adams P. D., Grosse-Kunstleve R. W., Hung L. W., Ioerger T. R., McCoy A. J., Moriarty N. W., Read R. J., Sacchettini J. C., Sauter N. K., Terwilliger T. C. (2002) Acta Crystallogr. D. Biol. Crystallogr. 58, 1948–1954 [PubMed]
20. Zwart P. H., Afonine P. V., Grosse-Kunstleve R. W., Hung L. W., Ioerger T. R., McCoy A. J., McKee E., Moriarty N. W., Read R. J., Sacchettini J. C., Sauter N. K., Storoni L. C., Terwilliger T. C., Adams P. D. (2008) Methods Mol. Biol. 426, 419–435 [PubMed]
21. Emsley P., Cowtan K. (2004) Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 [PubMed]
22. Davis I. W., Leaver-Fay A., Chen V. B., Block J. N., Kapral G. J., Wang X., Murray L. W., Arendall W. B., 3rd, Snoeyink J., Richardson J. S., Richardson D. C. (2007) Nucleic Acids Res. 35, W375–W383 [PMC free article] [PubMed]
23. Konitsiotis A. D., Raynal N., Bihan D., Hohenester E., Farndale R. W., Leitinger B. (2008) J. Biol. Chem. 283, 6861–6868 [PubMed]
24. Raynal N., Hamaia S. W., Siljander P. R., Maddox B., Peachey A. R., Fernandez R., Foley L. J., Slatter D. A., Jarvis G. E., Farndale R. W. (2006) J. Biol. Chem. 281, 3821–3831 [PubMed]
25. Vakonakis I., Staunton D., Rooney L. M., Campbell I. D. (2007) EMBO J. 26, 2575–2583 [PubMed]
26. Yamazaki T., Forman-Kay J. D., Kay L. E. (1993) J. Am. Chem. Soc. 115, 11054–11055
27. DeLano W. L. (2002) The PyMOL Molecular Graphics System, DeLano Scientific LLC, San Carlos, CA
28. Krissinel E., Henrick K. (2007) J. Mol. Biol. 372, 774–797 [PubMed]
29. Bingham R. J., Rudiño-Piñera E., Meenan N. A., Schwarz-Linek U., Turkenburg J. P., Höök M., Garman E. F., Potts J. R. (2008) Proc. Natl. Acad. Sci. U.S.A. 105, 12254–12258 [PubMed]
30. Wishart D. S., Sykes B. D., Richards F. M. (1992) Biochemistry 31, 1647–1651 [PubMed]
31. Leahy D. J., Aukhil I., Erickson H. P. (1996) Cell 84, 155–164 [PubMed]
32. Rudiño-Piñera E., Ravelli R. B., Sheldrick G. M., Nanao M. H., Korostelev V. V., Werner J. M., Schwarz-Linek U., Potts J. R., Garman E. F. (2007) J. Mol. Biol. 368, 833–844 [PubMed]
33. LiWang A. C., Cao J. J., Zheng H., Lu Z., Peiper S. C., LiWang P. J. (1999) Biochemistry 38, 442–453 [PubMed]
34. Kleinman H. K., Wilkes C. M., Martin G. R. (1981) Biochemistry 20, 2325–2330 [PubMed]
35. Xu X., Mikhailova M., Ilangovan U., Chen Z., Yu A., Pal S., Hinck A. P., Steffensen B. (2009) Biochemistry 48, 5822–5831 [PMC free article] [PubMed]
36. Gehrmann M. L., Douglas J. T., Bányai L., Tordai H., Patthy L., Llinás M. (2004) J. Biol. Chem. 279, 46921–46929 [PubMed]
37. Trexler M., Briknarová K., Gehrmann M., Llinás M., Patthy L. (2003) J. Biol. Chem. 278, 12241–12246 [PubMed]
38. Hornebeck W., Bellon G., Emonard H. (2005) Pathol. Biol. 53, 405–410 [PubMed]
39. Schwarz-Linek U., Werner J. M., Pickford A. R., Gurusiddappa S., Kim J. H., Pilka E. S., Briggs J. A., Gough T. S., Höök M., Campbell I. D., Potts J. R. (2003) Nature 423, 177–181 [PubMed]
40. Perumal S., Antipova O., Orgel J. P. (2008) Proc. Natl. Acad. Sci. U.S.A. 105, 2824–2829 [PubMed]
41. Fawzi N. L., Doucleff M., Suh J. Y., Clore G. M. (2010) Proc. Natl. Acad. Sci. U.S.A. 107, 1379–1384 [PubMed]
42. Tang C., Ghirlando R., Clore G. M. (2008) J. Am. Chem. Soc. 130, 4048–4056 [PubMed]
43. Chung L., Dinakarpandian D., Yoshida N., Lauer-Fields J. L., Fields G. B., Visse R., Nagase H. (2004) EMBO J. 23, 3020–3030 [PubMed]
44. Fiori S., Saccà B., Moroder L. (2002) J. Mol. Biol. 319, 1235–1242 [PubMed]
45. Stultz C. M. (2002) J. Mol. Biol. 319, 997–1003 [PubMed]
46. Graille M., Pagano M., Rose T., Ravaux M. R., van Tilbeurgh H. (2010) Structure 18, 710–718 [PubMed]
47. Fallah S., Sani F. V., Firoozrai M. (2009) Contraception 80, 40–43 [PubMed]
48. Iyengar V., Woittiez J. (1988) Clin. Chem. 34, 474–481 [PubMed]
49. Versieck J. (1985) Crit. Rev. Clin. Lab. Sci. 22, 97–184 [PubMed]
50. La Celle P., Blumenstock F. A., McKinley C., Saba T. M., Vincent P. A., Gray V. (1990) Blood 75, 470–478 [PubMed]
51. Blumenstock F. A., La Celle P., Herrmannsdoerfer A., Giunta C., Minnear F. L., Cho E., Saba T. M. (1993) J. Leukocyte Biol. 54, 56–64 [PubMed]
52. Di Lullo G. A., Sweeney S. M., Korkko J., Ala-Kokko L., San Antonio J. D. (2002) J. Biol. Chem. 277, 4223–4231 [PubMed]
53. Blumenstock F. A., Saba T. M., Weber P., Laffin R. (1978) J. Biol. Chem. 253, 4287–4291 [PubMed]
54. Kaplan J. E., Molnar J., Saba T. M., Allen C. (1976) J. Reticuloendothel. Soc. 20, 375–384 [PubMed]
55. Deno D. C., McCafferty M. H., Saba T. M., Blumenstock F. A. (1984) J. Clin. Invest. 73, 20–34 [PMC free article] [PubMed]
56. Eriksen H. O., Kalaja E., Jensen B. A., Clemmensen I. (1984) Burns Incl. Therm. Inj. 10, 422–426 [PubMed]
57. Lanser M. E., Saba T. M., Scovill W. A. (1980) Ann. Surg. 192, 776–782 [PubMed]
58. Metcalfe A. D., Ferguson M. W. (2007) J. R. Soc. Interface 4, 413–437 [PMC free article] [PubMed]
59. van den Dolder J., Bancroft G. N., Sikavitsas V. I., Spauwen P. H., Mikos A. G., Jansen J. A. (2003) Tissue Eng. 9, 505–515 [PubMed]

Articles from The Journal of Biological Chemistry are provided here courtesy of American Society for Biochemistry and Molecular Biology