If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Biochemistry and Cellular and Molecular Biology and the Center of Excellence in Structural Biology, University of Tennessee, Knoxville, Tennessee 37996
Department of Biochemistry and Cellular and Molecular Biology and the Center of Excellence in Structural Biology, University of Tennessee, Knoxville, Tennessee 37996
Department of Biochemistry and Cellular and Molecular Biology and the Center of Excellence in Structural Biology, University of Tennessee, Knoxville, Tennessee 37996
To whom correspondence should be addressed: Dept. of Biochemistry and Cellular and Molecular Biology, M407 Walters Life Sciences Bldg., University of Tennessee, Knoxville, TN 37996. Tel.: 865-974-5148; Fax: 865-974-6306;
Department of Biochemistry and Cellular and Molecular Biology and the Center of Excellence in Structural Biology, University of Tennessee, Knoxville, Tennessee 37996
* This work was supported in part by NHLBI, National Institutes of Health Grant HL50676 (to C. B. P.) and National Science Foundation Grant MCB 01110741 (to E. H. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The on-line version of this article (available at http://www.jbc.org) contains Supplementary Figs. 1–4 and Supplementary Table I.The atomic coordinates and structure factors (code 1S4G) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ(http://www.rcsb.org/). ‡ Both authors contributed equally and are considered first authors on this work. § Supported by Postdoctoral Fellowship 0120344B from the American Heart Association, Southeast Regional Affiliate.
The three-dimensional structure of an N-terminal fragment comprising the first 51 amino acids from human plasma vitronectin, the somatomedin B (SMB) domain, has been determined by two-dimensional NMR approaches. An average structure was calculated, representing the overall fold from a set of 20 minimized structures. The core residues (18–41) overlay with a root mean square deviation of 2.29 ± 0.62 Å. The N- and C-terminal segments exhibit higher root mean square deviations, reflecting more flexibility in solution and/or fewer long-range NOEs for these regions. Residues 26–30 form a unique single-turn α-helix, the locus where plasminogen activator inhibitor type-1 (PAI-1) is bound. This structure of this helix is highly homologous with that of a recombinant SMB domain solved in a co-crystal with PAI-1 (Zhou, A., Huntington, J. A., Pannu, N. S., Carrell, R. W., and Read, R. J. (2003) Nat. Struct. Biol. 10, 541–544), although the remainder of the structure differs. Significantly, the pattern of disulfide cross-links observed in this material isolated from human plasma is altogether different from the disulfides proposed for recombinant forms. The NMR structure reveals the relative orientation of binding sites for cell surface receptors, including an integrin-binding site at residues 45–47, which was disordered and did not diffract in the co-crystal, and a site for the urokinase receptor, which overlaps with the PAI-1-binding site.
Human vitronectin is a glycoprotein found in the circulation, where it contributes to hemostasis by regulating blood coagulation and fibrinolysis (
). The variety of functions of vitronectin in the two microenvironments is a manifestation of its ability to interact with numerous humoral and cellular proteins. An important binding partner for vitronectin is the serine protease inhibitor PAI-1, which also is found both in circulation and the ECM. Furthermore, it has different activities depending upon this localization; the anti-protease activity of PAI-1 that regulates thrombolysis in the circulation is targeted instead toward pericellular proteolysis when localized to the ECM or cell/matrix boundary. Vitronectin binds to PAI-1 with high affinity and stabilizes the inhibitor in its active conformation (
). Key to the adhesive functions of vitronectin are its interactions with cell-surface receptors including integrins and uPAR.
A widely accepted model suggests that vitronectin is organized into functional domains that provide the broad repertoire necessary for binding to target ligands (
). A threading algorithm gave high confidence predictions for the central domain of ∼200 amino acids and the C-terminal domain of ∼100 amino acids. Both domains exhibit features of a β-propeller fold. The computational approach was less successful for the third domain, corresponding to the N-terminal span of ∼50 amino acids that has been denoted the “somatomedin B” domain of vitronectin (
). An experimental determination of the fold of this domain is thus needed and is relevant to over 100 homologues in the sequence data base. Within this short N-terminal region from vitronectin lie binding sites for three critical ligands: PAI-1, integrins, and uPAR. The SMB domain contains eight cysteine residues that form four disulfide bonds, and recombinant forms of SMB have been used to identify amino acid residues that participate in binding these ligands (
). The structure of a co-crystal of PAI-1 and a recombinant SMB was recently reported, further elucidating the features of the binding interface between vitronectin and PAI-1 (
We have purified the SMB domain from human plasma vitronectin using cyanogen bromide cleavage to release the first 51 amino acids from the bulk of the protein. In this work, two-dimensional NMR methods have been used to determine the solution structure of this domain. A single homogeneous species is observed with a unique disulfide-bonding pattern. Our results complement the x-ray crystallography on the complex and confirm the presence within the SMB domain of a sole α-helix containing known PAI-1-binding residues. However, the remainder of the chain displays an entirely different fold and different pattern of disulfides. Furthermore, the NMR structure defines the position of the integrin-binding tripeptide (RGD) site, which was unstructured in the co-crystal with PAI-1.
EXPERIMENTAL PROCEDURES
Materials—The C-18 reverse phase HPLC column was obtained from Vydac, Hesperia, CA. HPLC grade acetonitrile was from Fisher. Cyanogen bromide and trifluoroacetic acid were from Pierce. D2O was from Cambridge Isotope Laboratories (Andover, MA). All other chemicals were of the highest purity available.
Cyanogen Bromide Digestion of Vitronectin—Human plasma vitronectin was isolated as described in Zhuang et al. (
). Purified vitronectin in an ammonium sulfate suspension was centrifuged at 4000 rpm for 20–30 min at 4 °C. The pellet was resuspended in 0.005% trifluoroacetic acid and dialyzed against 0.005% trifluoroacetic acid for 4 h at 4 °C. The solvent was removed by lyophilization, and cyanogen bromide in 70% trifluoroacetic acid containing 200 mm iodoacetamide was added to the lyophilized vitronectin. The reaction mixture was flushed with N2 and incubated at room temperature in the dark for 22 h. After 22 h the reaction mixture was diluted with twice the volume of water and lyophilized. The lyophilized product was resuspended in 1% trifluoroacetic acid and centrifuged. The supernatant was separated on a 1 × 25-cm reverse phase C-18 HPLC column by gradient elution with 17–32% acetonitrile in water containing 0.1% trifluoroacetic acid to isolate the desired N-terminal fragment representing the SMB domain. Fractions were analyzed for purity by re-injecting an aliquot on the reverse phase C-18 column, and the desired fractions were combined and stored at –20 °C. The size (m/z 5762.3) of the fragment was confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry, and its monomeric state was confirmed by sedimentation equilibrium analysis using an Optima XL-I ultracentrifuge from Beckman. The typical yield of vitronectin was 7–8 mg/liter of human plasma. The yield of SMB domain was ∼5%. Four preparations of vitronectin, each from 2 liters of plasma, were combined to accumulate the required amount of SMB domain for the NMR experiments.
NMR Spectroscopy—NMR studies were performed with ∼90 μm SMB in a 500-μl sample volume. The lyophilized sample was dissolved in D2O or H2O containing 10% D2O at the appropriate pH. The sample pH was adjusted with either 100 mm trifluoroacetic acid or 100 mm NaOH. All NMR experiments were performed on a 600 MHz Varian Inova instrument equipped with a single gradient axis and a triple resonance probe for the observation of proton, carbon, and nitrogen nuclei. To assign overlapping resonances in two-dimensional NMR spectra, experiments were recorded at 288, 298, and 308 K and pH values ranging from 1.4 to 7.3. Best resolution was obtained at pH 4.4, and all the assignments were reported with respect to the spectra obtained at pH 4.4 and 298 K. Two-dimensional NMR data were acquired in phase-sensitive mode using the States-Haberkorn (
) were recorded using DIPSI spin-lock sequence with a 8 kHz radio frequency field and mixing times of 15, 30, 60, and 80 ms. Typically, spectra were acquired with 256 t1 increments, 2048 data points, a relaxation delay of 1 s, and a spectral width of 8500 Hz. For DQF-COSY spectra (
), 512 t1 increments were acquired. Spectra were recorded with 64–96 scans per increment for NOESY, 24–80 scans per increment for TOCSY, and 56 scans per increment for DQF-COSY. In all NOESY and TOCSY spectra, the data were multiplied by a 60–90° phase shifted sin2 window function in both dimensions before Fourier transformation.
Data Analysis and Structure Determination—NMR data were processed with Felix 2000 and SPARKY (T. D. Goddard and D. G. Kneller, The University of California, San Diego) software operating on a Silicon Graphics Indigo 2 work station or on a Silicon Graphics Origin 300 server. Cross-peak intensities observed in NOE experiments were divided into four categories as strong, medium, weak, and very weak. These intensities were converted into distance restraints as follows: strong, 1.8–2.7 Å; medium, 1.8–3.4 Å; weak, 1.8–4.5 Å, and very weak, 1.8–6.0 Å. An additional 1.0 Å was added to upper limits involving methyl protons. Similarly, an additional 0.5 Å for methylene protons and 2.3 Å for degenerate Hδ and Hϵ protons of tyrosines and phenylalanines were added to the upper limits. Also, 0.2 Å was added to the upper limits of NOEs involving amide protons. Coupling constants were extracted from the DQF-COSY spectra. Backbone Φ angles were restrained to –120 ± 50° for 3JHNHα = 8–9 Hz, and –120 ± 40° for 3JHNHα > 9 Hz. A restraint of –100 ± 80° was also applied to the Φ angle for residues that show stronger NHi-Hαi-1 NOE than the intraresidue NH-Hα NOE (
). A total of 329 NOE restraints and 18 Φ restraints were used in structure determination.
All calculations were carried out using the AMBER force field interfaced with DISCOVER (Accelerys, San Diego, CA) on an Origin 300 work station. Random structures were generated by subjecting the peptide to an initial 10,000-step minimization at 298 K. The temperature was then raised gradually to 1000 K during a 1000-step dynamics simulation. The peptide was subjected to minimization and a 10-ps dynamics run at 1000 K. The NMR-derived restraints were then imposed on the peptide and the peptide was slowly annealed to 298 K in a 100-ps trajectory. Finally, the structures were subjected to further minimization at 298 K. The force constant for the distance restraints was 100 kcal/mol Å–2 and the dielectric constant was 4. Using this simulated annealing protocol, an ensemble of 20 starting structures was obtained. These structures were then used to make further resonance assignments and increase the number of unambiguous NOE restraints. In the final step of the refinement, 60 structures were obtained and analyzed by MOLMOL (
This number represents the number of violations among all 20 structures. Significantly, no single restraint was violated by the majority of structures.
This number represents the number of violations among all 20 structures. Significantly, no single restraint was violated by the majority of structures.
Residues in additional allowed regions (a, b, l, p)
339 (37.7%)
15 (34.9%)
Residues in generously allowed regions (∼a, ∼b, ∼l, ∼p)
51 (5.7%)
2 (4.7%)
Residues in disallowed regions
58 (6.4%)
4 (9.3%)
Total number of non-glycine and non-proline residues
900 (100.0%)
43 (100.0%)
Other statistics (standard deviation in degrees)
H-bond energy
0.6
0.7
Chi-1 “pooled”
12.8
15.8
Chi-2 trans
17.0
18.6
aThe NMR_refine module within InsightII 2002 (Accelrys, San Diego, CA) was used to determine the NOE and dihedral angle violations.
bThis number represents the number of violations among all 20 structures. Significantly, no single restraint was violated by the majority of structures.
Resonance Assignments—The material used in this study was the SMB domain isolated from plasma vitronectin, so 1H NMR was used exclusively because the sample could not be isotopically labeled. Sequential resonance assignments were made by the analysis of fingerprint regions of TOCSY and NOESY spectra using standard procedures (
). Excellent dispersion was observed in the spectra. Examples that demonstrate the quality of the data and the dispersion observed in a one-dimensional spectrum and TOCSY and NOESY spectra are given in Supplemental Materials Figs. 1–3. There were numerous overlaps, presumably because of the compact and globular nature of the polypeptide, causing some residues to be in similar magnetic environments. DQF-COSY was used to determine 13 unambiguous NH-Hα coupling constants. Cross-peaks representing smaller coupling constants were absent from the DQF-COSY spectrum because of limitations in concentration of the biological material. All the determined coupling constants were >8 Hz. There were no patterns of cross-peaks to indicate the presence of regular secondary structure elements in SMB. Rapid exchange of virtually all amide protons also supported this observation. Thus, the backbone assignment strategy relied heavily on the sequential NOESY-TOCSY cross-peak analyses of the NH-Hα regions of the spectra.
Side chain assignments were made on the basis of the scalar connectivities obtained through a series of TOCSY spectra using different mixing times at different pH values and temperatures. Aromatic parts of tyrosine and phenylalanine residues were assigned on the basis of NOE cross-peaks. No stereospecific assignments were made for methylene protons. Because of ring opening of homoserine lactone at higher pH values, a few residues near the C terminus showed two sets of resonances with slight differences in chemical shifts that did not hinder the resonance assignments. A sequential connectivity for the last three residues was not observed, thus these residues were tentatively assigned. Residues 9 and 10 were also tentatively assigned because of resonance overlap. Overall, backbone, and side chain assignments of 46 residues were completed, sufficient for structural calculations to be performed on this polypeptide. The single proline residue (Pro41) of SMB was identified to be in a trans conformation based on the observation of NOEs between lysine 40 NH/Hα and proline 41 Hδ. The complete resonance assignment of SMB is given in Supplemental Materials Table I.
Assignment of Disulfides—Attempts to identify the disulfide bridges from NOE cross-peaks observed between CHαi and CHβj and/or between CHβi and CHβj of cysteines using published methods (
) were not successful. Spectral overlap between intra- and inter-residue NOE cross-peaks involving cysteines, combined with the spatial proximity of the six cysteines in the core region of SMB, precluded unequivocal identification of the disulfide bridges solely by NMR. Similar problems have been encountered with other cysteine-rich proteins (
In early stages of simulated annealing calculations, “floating” cysteines were used without any disulfide assignments. These early structures consistently showed that cysteines 5 and 9 were near each other, and four cysteines (Cys19, Cys21, Cys31, and Cys32) formed a cluster in the core of SMB. Either cysteine 25 or cysteine 39 was also near this cluster in most structures. In a concurrent study from this laboratory, two of the disulfide bonds of SMB were determined by biochemical methods and mass spectrometry.
Horn, N. A., Hurst, G. B., Mayasundari, A., Whittemore, N. A., Serpersu, E. H., and Peterson, C. B. (2004) J. Biol. Chem., in press.
This work revealed one disulfide bridge as 5–9 and another as 25–39. These findings were in agreement with the initial calculations, so the two experimentally determined disulfides were imposed to refine the structure of the SMB domain, and two possible alternatives were then pursued separately in the structure determination. These alternatives were: SMB with Cys19–Cys31 and Cys21–Cys32 disulfides or SMB with Cys19–Cys32 and Cys21–Cys31 disulfides. A third alternative pairing among the four cysteines is not feasible and was not considered because it would involve a disulfide bridge between adjacent cysteines 31 and 32. The disulfide bridge restraints were set for the two alternatives (
), and both were subjected to simulated annealing separately. Calculations with the second alternative involving disulfides 19–32 and 21–31 did not yield acceptable converged structures. Thus, the structure of SMB containing disulfides 5–9, 19–31, 21–32, and 25–39 was determined and is described in this work.
Structure Refinement—The three-dimensional structure of SMB was determined by simulated annealing calculations using a total of 347 NMR-determined restraints (Table I). A large fraction of intra- and intermolecular NOEs were observed in the region spanning residues 18–41 of SMB. It should be noted that NOEs involving any tentatively assigned residues (Supplemental Materials Table I) were not included in structure calculations. As described above, the best structures from initial calculations were selected and used as starting structures with imposed disulfide restraints (Cys5–Cys9 and Cys25–Cys39) for the final stages of the structure calculation. The 20 lowest energy structures were chosen from among an ensemble of 60 minimized structures. The Ramachandran plot showed that >90% of Φ and ψ angles were within the most favored and allowed regions, and less than 7% of the Φ and ψ angles from all 20 structures were within the disallowed region. Cluster analysis resulted in a r.m.s. deviation value for the backbone atoms of the 20 structures of 2.29 ± 0.62 Å when the region limited to residues 18–41 was considered (Fig. 1a). The low r.m.s. deviation for this portion of the structure resulted because the majority of the observed NOEs were from the residues in this region, which also contains three of the four disulfide bridges that stabilize the compact structural core of the domain.
Fig. 1Solution structure of the SMB domain determined by NMR. An overlay of the backbone in the core region (residues 18–41) from the 20 lowest energy structures calculated from the two-dimensional 1H NMR data is shown in panel a. A backbone overlay of these 20 structures for the entire length of the SMB domain is shown in panel b, with the core region in blue, the N-terminal residues (1–17) in red, and the C-terminal residues (42–51) in green. The r.m.s. deviation for backbone residues in the core region (panel A) is 2.29 ± 0.62 Å, and the r.m.s. deviation for all the residues (1–51, panel B) is higher (6.72 ± 1.38 Å), reflecting the mobility of the N- and C-terminal regions in solution. The average structure calculated from the 20 lowest energy structures is shown in the ribbon structure for the main chain in panel c. The four disulfide connections are shown in yellow, and the single-turn α-helix is highlighted in green. The side chains for the integrin-binding tripeptide (RGD) sequence are shown, along with the van der Waals surface associated with this binding site.
Fig. 1b shows the entire chain trace for all 20 resulting structures superimposed in this region spanning residues 18–41. These data indicate that, in solution, the core region of SMB yields converged structures, whereas the structures at both the N- and C-terminal ends of the molecule are less well defined. The greater flexibility of the C-terminal region (residues 42–51) compared with the core is apparent from the NMR data, which showed sharper resonances and chemical shifts similar to random coil values for these residues (Fig. 2). Another interesting feature of the structure is the lower refinement of the N-terminal section, which contains the fourth disulfide bridge between cysteines 5 and 9. Despite all efforts, no unambiguous NOEs could be identified between this section and the core region to define its relative orientation. This is perhaps not surprising because global bends can be difficult to detect by NMR, as long-range NOEs do not exist. Also, the higher r.m.s. deviation for this region of the SMB domain may result from inherent limits in sensitivity that arise from the low concentration of this sample isolated from human plasma, precluding observation of weaker NOEs. Also, there are relatively few unambiguous intra-domain NOEs. On the other hand, the less refined structure may indicate that this region is more flexible than the core region in solution. The lack of interaction of the 4Hβ protons of cysteines 5 and 9, combined with the ambiguity of the resonance assignments of cysteine 9, is consistent with these observations and suggests that the corresponding disulfide bridge is either somewhat flexible or located in a more mobile part of the molecule.
Fig. 2Differences in chemical shifts for α-protons in SMB relative to standard values for free amino acids. Up-field shifts are shown below the line, and downfield shifts above the line. The two clusters of up-field shifts correspond to the two secondary structural elements, the α-helix from 26 to 30 and the 310 helix from 36 to 38.
Solution Structure of the Somatomedin B Domain of Vitronectin—The average structure, shown in Fig. 1c, yields an r.m.s. deviation value of 2.29 Å for the core backbone atoms of the ensemble. As shown in this figure, the core region of SMB is held by three disulfide bridges, which form the disulfide knot. The only regular secondary structure is a single-turn α-helix spanning residues 26–30. Observation of successive dNN(i,i+1) NOEs (see Supplemental Materials Fig. 4), a single dNN(i,i+3) NOE, and up-field shifted Hα resonances of residues 26–30 (Fig. 2) also support the presence of a short α-helix in this region. A second region that has a loosely defined 310 helix spans residues 36–38. The remainder of the molecule is comprised of loops, some of which appear to be flexible in solution. Rapidly exchanging amide protons and the lack of strong dβN(i,i+1) NOEs are consistent with the absence of β-sheet in the rest of the molecule. Hydrogen-deuterium exchange studies of the SMB domain showed that virtually all of the backbone amide and other exchangeable protons of the molecule were exchanged within a few minutes of exposure to D2O, indicating that the majority of the residues of SMB are exposed to solvent and signifying the absence of a hydrogen-bonded network of protons in the molecule. This observation is in accord with the determined solution structure of SMB, in which the N- and C-terminal segments are flexible with a small, solvent-accessible core that is stabilized by three disulfide bonds.
DISCUSSION
A crystal structure of a complex between PAI-1 and a recombinant 51-residue SMB domain has been published recently (
), with an ordered structure observed only for residues 3 to 39 of the domain. The x-ray structure does not superimpose well with the solution structure calculated from the NMR data, either as the entire structure from residues 3 to 39 (Fig. 3A, r.m.s. deviation = 8.3 ± 0.6 Å for backbone atoms) or within the more highly refined core region defined by residues 18 to 39 (r.m.s. deviation = 6.3 ± 0.4 Å for backbone atoms). Furthermore, the assignments of disulfide bridges are altogether different in the two structures. The disulfide pattern of the domain used in crystallography also differs from a third disulfide pattern determined with another recombinant form of the SMB domain (
). The inconsistencies may stem from difficulties involved in expressing fragments of proteins with multiple disulfide bonds by recombinant technology, amounting to over 100 possibilities for the set of 8 cysteine residues. Significantly, there is no evidence for disulfide “scrambling” in the SMB domain derived from plasma vitronectin, as a homogeneous species with a unique disulfide-bonded structure was obtained from multiple preparations using cyanogen bromide digestion. This contrasts with the production of a mixture of folds from expression of the recombinant SMB protein, with only a fraction that is active in binding PAI-1 (
Fig. 3Comparison of SMB structures from 1H NMR and x-ray crystallography.Panel A shows an overlay of the solution structure (blue) and crystal structure (red) for residues 3–39, the ordered region that gave a diffraction pattern in the crystallography work (Ref.
, Protein Data Bank accession code 1OC0). The structures were aligned by overlaying the single α-helix, shown in yellow for both structures, with an r.m.s. deviation of 1.09 Å. The 310 helix is shown in green for both structures. Panel B shows the relative orientation of PAI-1-binding residues in the solution structure (blue) and x-ray structure (red). Residues identified to participate in PAI-1 binding by site-directed mutagenesis (
) are shown by side chains for Asp22, Leu24, Tyr27, Tyr28, and Asp34, and by highlighting the backbone region for Gly12. Three additional residues, Thr10, Phe13, and Glu23, were identified at the PAI-1-binding interface from the crystallography work (
). Labels are shown in violet for residues that occupy overlapping positions, and labels are shown in black for residues with different orientations in the two structures. Hydrogen positions were not determined in the co-crystal (
Because of the inherent difficulties associated with expression of cystine-rich sequences in prokaryotes, which lack the necessary post-translational machinery to incorporate disulfide bonds in an analogous way to higher eukaryotes, the strategy for this work mandated purification of this domain from full-length vitronectin that, in turn, was isolated from natural sources. It was reasoned that an unambiguous assignment of the disulfides in the SMB domain from human circulatory vitronectin was required as a “standard” by which the recombinant surrogates must be measured. Neither of the recombinant SMB domains exhibits the same pattern of the 4 cystines observed for the human plasma counterpart; the only common disulfide comparing the SMB structure in this work with the two recombinant forms is the bond between cysteines 5 and 9, which is also observed in the form of SMB characterized by the Loskutoff laboratory (
). A puzzling result is that both recombinant forms of the SMB domain bind to PAI-1. Could these results imply that the particular disulfide order is not critical for the function of this domain, as long as a compact structure with properly oriented residues for ligand binding is manifested? Inspection of the structural features of this domain, as follows, is needed to further consider this question.
Although the assignments of the disulfide bridges are completely different between the recombinant SMB domain used in crystallography and the SMB domain prepared from plasma vitronectin, there are notable similarities between the two structures. Both structures show a single-turn α-helix involving residues 26–30 (Fig. 3A). Superpositioning of the crystal structure and NMR structure in this region yields 1.09 Å r.m.s. deviation. Also, a single-turn 310 helix identified in the crystal structure is also observed in the solution structure, and superpositioning of this region (residues 36–38) yields a r.m.s. deviation of 0.88 Å. Indeed, these are the only secondary structural elements in either of the structures, and the rest of the SMB domain shows no regular secondary structure elements from NMR or x-ray crystallography analysis. Although these two secondary structural features are nearly identical, their relative orientation in the two structures differs (Fig. 3A).
The single α-helix is found within a fairly large loop defined by the disulfide pair, Cys25–Cys39. This structure is in many ways reminiscent of a subclass of cystine knot structures classified as cystine-stabilized α-helices (CSH motifs (
)), although there are differences in the sequence positions of individual cysteines in the SMB domain relative to the helix compared with prototypical CSH motifs. Such structures are known to function in binding extracellular ligands (
). Despite different disulfide bonding in the core, the plasma-derived and recombinant versions of SMB fold with a single helix analogous to the CSH motif. In the large family of CSH domains, it is observed that a variety of disulfide-bonding patterns can be tolerated within the core of the domain, providing a structural scaffold that orients the α-helix on the surface (
). Thus, the curious retention of PAI-1 binding among this set of clearly different folds for the SMB domain is consistent with the demonstrated malleability of cystine knot structures to accommodate cysteine substitutions, yet retain activity (
). Even though the recombinant forms of the SMB domain do not have the correct cystine bonds found in circulating vitronectin, PAI-1 binding appears to be maintained because the various disulfide cross-linked frameworks support the formation of the sole α-helix in this domain.
The single conserved α-helix houses two of the residues, Tyr27 and Tyr28, which were identified from site-directed mutagenesis studies (
) to participate in binding to PAI-1, with two others (Asp22 and Leu24) present in a turn that immediately precedes this helix. Fig. 3B shows the relative positioning in the NMR and crystal structures of these and other side chains that have been implicated as important for PAI-1 binding to vitronectin. As shown in the figure, five residues previously identified to participate in PAI-1 binding (Thr10, Glu23, Leu24, Tyr27, and Tyr28) occupy the same positions. Differences are observed for the relative orientation of Gly12, Phe13, Asp22, and Asp34 in the two structures. As shown in more detail in the space-filling representation of the SMB domain presented in Fig. 4A, these residues appear to form a PAI-1-binding surface that encompasses two contiguous regions; the first region includes Gly12, Phe13, and Thr10, and the second contains Asp22, Glu23, Leu24, Tyr27, and Tyr28. The first set of residues lies within the N-terminal region containing the Cys5–Cys9 disulfide, whereas the second group is found in the vicinity of the α-helix. As discussed above, this helix lies within the more structured SMB core that is stabilized by the other three disulfides (Cys19–Cys31, Cys21–Cys32, and Cys25–Cys39). Thr10 and Phe13 are positioned further apart in the x-ray structure (Fig. 3B), and Gly12 is not found within the binding surface identified in the co-crystal. Asp34 is remote from the PAI-1-binding interface in both the NMR and crystal structures.
Fig. 4Features of the SMB domain involved in PAI-1 binding.Panel A shows a ribbon trace of the backbone of the SMB domain with CPK surfaces for residues that are proposed to interact with PAI-1 from prior mutagenesis work (
). These include Thr10, Gly12, Phe13, Asp22, Glu23, Leu24, Tyr27, and Tyr28, shown by labels in the figure. Disulfide bonds within the SMB domain are shown in yellow. Panel B presents the structure for the SMB domain determined by NMR superimposed on the structure from the co-crystal of recombinant SMB and PAI-1 (
). The SMB domain was positioned in the structure by overlaying the α-helix (residues 26–30) that is coincident in the NMR structure and the crystal structure; only the NMR structure is shown for the SMB domain to depict the positions of the proposed residues at the PAI-1/SMB interface. The SMB domain is presented in the turquoise ribbon structure with turquoise CPK surfaces for the residues (highlighted in panel A) that have been implicated as important for PAI-1 binding from previous work on the SMB domain (
). The RGD integrin-binding motif is highlighted in green CPK structures. PAI-1 is shown by the gray strand, depicting β-sheets with yellow ribbons, and α-helices as red cylinders. The central β-sheet (A sheet) and reactive center loop (RCL) of PAI-1 are labeled, as well as the F helix that appears to form numerous contacts with the SMB domain. Residues that have been identified as important for vitronectin binding in site-directed mutagenesis studies on PAI-1 ((
An overlay was constructed with the SMB domain determined by NMR in place of the SMB domain observed in the co-crystal to test whether this putative binding surface comprises the functional PAI-1-binding site in this region of vitronectin. These results are shown in Fig. 4B. The intimate association of the SMB domain with PAI-1 is apparent from this figure, which highlights PAI-1-binding residues within the SMB domain in turquoise, as well as residues in PAI-1 that are known to be important for vitronectin binding (
) shown in blue. Clearly, this surface containing Thr10, Glu12, Phe13, Asp22, Glu23, Leu24, Tyr27, and Tyr28 is the binding site for PAI-1. Measured distances between backbone atoms (all less than 8 Å between α-carbons) from the SMB domain to PAI-1 in the overlaid structure suggest that the region from Thr10 to Phe13 in the SMB domain contacts Phe109, Met110, and Leu116 in PAI-1, three residues proposed to interact with vitronectin (
). Likewise, the region in the vicinity of the α-helix encompassing Asp22 to Tyr28 interacts with Gln125 in PAI-1, another residue that is critical for binding vitronectin (
). Extensive interactions from the second group of residues (near the helix within the stable core of the SMB domain) also are observed with helix F in PAI-1. Taken together, these results agree with mutagenesis work, combined with protection of PAI-1 against chemical inactivation, that has localized the vitronectin-binding site in PAI-1 to the region spanning helix F, β-strand 2A, and helix E in PAI-1 (
From this analysis, it appears that vitronectin stabilizes PAI-1 via this interface in the SMB domain, which binds across the E and F helices of PAI-1, preventing the movement of strands 1 and 2 of the main β-sheet in PAI-1 that occurs in the relaxation of PAI-1 to a latent form (
). As described above, this α-helix in the SMB domain is highly homologous between the two structures (Fig. 3A), although the remainder of the structure, including the 310 helix region, does not overlap. Theoretically, it is possible that changes in the fold of the domain occur upon binding to PAI-1, and these could account for some of the differences in the structure of the domain comparing results from NMR and crystallography. However, the overlaid structures do not support such changes because the interface for PAI-1 binding appears to be fully formed and extensive in the NMR structure without PAI-1 present.
The SMB domain is key to many of the varied functions of vitronectin because it contains binding sites not only for PAI-1, but also for cell surface receptors, uPAR, and integrins. The orchestration of vitronectin binding to PAI-1 and these receptors modulates fibrinolysis and cell migration. The binding of PAI-1 appears to block interaction with uPAR (
) has shown that residues involved in uPAR binding virtually overlap with important PAI-1-binding residues in the region of the α-helix: Asp22, Glu23, Leu24, Tyr27, and Tyr28 (Fig. 3B). The binding site for integrins is the RGD sequence comprising residues 45–47 in the SMB domain, shown in the van der Waals surface in Fig. 1C. This sequence lies within the C-terminal portion of the SMB domain, in a region that exhibits more flexibility than the core containing residues 18–41 (Fig. 1). The RGD sequence was disordered in the co-crystal structure with PAI-1, so insight into the relative orientation of the RGD sequence and the PAI-1-binding site is provided for the first time from the solution structure. As shown in the space-filling model in Fig. 4B, the RGD sequence is clearly separate from the PAI-1-binding surfaces in this small domain.
The demonstrated flexibility of this region may be relevant to ongoing work needed to clarify conflicting results regarding PAI-1 effects on vitronectin activity in vitro and in vivo (
). It is usually accepted that PAI-1 and integrin binding are mutually exclusive because of the proximity of the two binding sites within the compact SMB domain (
). To the extent that these two ligand-binding regions are located close to each other on the SMB domain, the solution structure supports this idea. However, the binding sites are clearly separate and will not necessarily be mutually exclusive in their binding functions. In this regard, we previously have observed that PAI-1 binds to vitronectin and promotes the assembly of higher-order complexes that become preferentially associated with the ECM, are internalized and degraded together in cell culture, and actually bind integrins more avidly (
). The relative disorder within this C-terminal region of the SMB domain housing the RGD sequence, combined with a closely spaced, but separate binding site for PAI-1, suggests that there may be scenarios that can accommodate simultaneous binding of PAI-1 and integrins. Indeed, the flexibility of this region may contribute to modulating affinity of PAI-1 for vitronectin, ultimately influencing the role of PAI-1 in regulating cellular adhesion and migration.
The role in adhesion and matrix organization is one of the most compelling areas of research on vitronectin. There is a close interplay between uPAR, integrins, vitronectin, and PAI-1 in regulating cell invasiveness. Multiple lines of evidence exist for association of uPAR with integrins to coordinate the effects of the two classes of receptors in mediating signaling and promoting cellular rearrangement and/or migration (
). This new information defining the relative location of binding sites for PAI-1 and the two receptors within the SMB domain will be invaluable for considering mechanisms by which these receptors and the ECM communicate in a reciprocal manner, ultimately with the goal of developing new anti-tumor and anti-thrombotic agents.
Acknowledgments
We are indebted Dr. Greg Hurst in the Division of Chemical Sciences at Oak Ridge National Laboratory for assistance with the mass spectrometry. We also thank Dr. Jane Dyson, Scripps Research Institute, La Jolla, CA, for productive discussions on the NMR analysis.
71186-7&id=gr1.jpg){kind=link}
71186-7&id=gr2.jpg){kind=link}
71186-7&id=gr3.jpg){kind=link}
71186-7&id=gr4.jpg){kind=link}