The Peptide-Substrate-binding Domain of Human Collagen Prolyl 4-Hydroxylases BACKBONE ASSIGNMENTS, SECONDARY STRUCTURE, AND BINDING OF PROLINE-RICH PEPTIDES*

The collagen prolyl 4-hydroxylases (C-P4Hs) catalyze the formation of 4-hydroxyproline by the hydroxylation of proline residues in -Xaa-Pro-Gly- sequences. The vertebrate enzymes are (cid:1) 2 (cid:2) 2 tetramers in which protein-disulfide isomerase serves as the (cid:2) subunit. Two isoforms of the catalytic (cid:1) subunit have been identified and shown to form [ (cid:1) (I)] 2 (cid:2) 2 and [ (cid:1) (II)] 2 (cid:2) 2 tetra- mers, the type I and type II C-P4Hs, respectively. The peptide-substrate-binding domain of type I C-P4H has been shown to be located between residues 138 and 244 in the 517-residue (cid:1) (I) subunit and to be distinct from the catalytic domain that is located in the C-terminal region. We report here that a recombinant human C-P4H (cid:1) (I) polypeptide Phe 144 -Ser an Nde a transla- tion codon codon Phe 144 and an Xho I site following the codon for Ser 244 , using Pfu polymerase (Promega). The template was pET15b- (cid:1) (I) Gly 138 -Ser 244 (17) in which the codon for Cys 150 (TGC) was converted to that for serine (TCC). The C150S mutagenesis was performed using a QuikChange™ site-directed mutagenesis kit accord-ing to the manufacturer’s instructions (Stratagene). The recombinant human (cid:1) (II) Met 142 -Ser 242 polypeptide was prepared by amplifying a cDNA fragment encoding residues Met 142 -Ser 242 with an Nde I site preceding the codon for Met 142 and an Xho I site following the codon for Ser 242 , using full-length human C-P4H (cid:1) (II) cDNA (11) as a template. The codon for Cys 148 (TGC) was converted to that for serine (TCC) in the same PCR amplification step. The PCR products were cloned into Nde I- Xho I-digested pET-22b expression vectors (Novagen) in-frame with the sequence encoding a C-terminal histidine tag and transformed into the BL21(DE3) Escherichia coli host strain (Novagen). For expression, 500 ml of Luria-Bertoni medium containing ampicillin (50 (cid:3) g/ml) was inoc- ulated with 5 ml of overnight culture. When A 600 reached 0.8, synthesis of the recombinant proteins was induced with 1 m M isopropyl- (cid:2) - D - thiogalactoside for 4 h at 30 °C,after which the cells were harvested and disrupted by sonication. The soluble proteins were applied to a nickel-nitrilotriacetic acid-chelating Sepharose column (Amersham Biosciences) equilibrated with 20 m M Bis-Tris and 0.1 M glycine, pH 6.8. The bound proteins were eluted with containing 0.25–0.5 m M of the peptide-binding domain in the above buffer at an interval of 4 min with stirring at 300 rpm. The titration data obtained were corrected by subtracting the heat of dilution of the peptides from the raw data and fitted by the non-linear least squares minimization method using one set of sites model (Levenberg-Mar- quardt algorithm) to determine the dissociation constant ( K d ), binding stoichiometry ( n ), and change in enthalpy ( (cid:2) H ) using the Origin soft- ware (Microcal). Other Assays— P4H activity was assayed by a method based on the hydroxylation-coupled decarboxylation of 2-oxo[1- 14 C]glutarate (34), and K m values were determined as described previously (16).

The prolyl 4-hydroxylases (P4Hs) 1 catalyze the formation of 4-hydroxyproline by the hydroxylation of proline residues in peptide linkages. Two P4H families are known today. The collagen P4Hs (C-P4Hs), enzymes residing within the lumen of the endoplasmic reticulum, have a central role in the synthesis of all collagens, the resulting 4-hydroxyproline residues being essential for assembly of the triple-helical molecules (1)(2)(3). The hypoxia-inducible factor (HIF) P4Hs, a family of cytoplasmic enzymes (4 -6), play a key role in the response of cells to hypoxia by catalyzing hydroxylation of the ␣ subunit of HIF. This subunit is synthesized continuously, and at least one of two critical proline residues becomes hydroxylated under normoxic conditions, the resulting 4-hydroxyproline being essential for rapid degradation of HIF-␣ (7)(8)(9). In hypoxia this hydroxylation ceases, HIF-␣ forms a dimer with HIF-␤, and the dimer then becomes bound to the HIF-responsive elements in a number of hypoxia-inducible genes.
All vertebrate C-P4Hs are ␣ 2 ␤ 2 tetramers in which the ␤ subunit is identical to the enzyme and chaperone protein-disulfide isomerase (1)(2)(3). Two isoforms of the catalytic ␣ subunit have been characterized from human and mouse tissues and shown to form [␣(I)] 2 ␤ 2 and [␣(II)] 2 ␤ 2 tetramers, called the type I and type II C-P4Hs, respectively (10,11). The type I enzyme is the most abundant form in most cells, but type II is the main form in chondrocytes, endothelial cells, and some other cell types (12,13). The HIF-P4Hs appear to consist of only one type of monomer, the size of which ranges from 239 to 426 residues in the three human isoenzymes (4 -6).
The C-P4Hs act on -Xaa-Pro-Gly-triplets in collagens and more than 15 other proteins with collagen-like sequences (1)(2)(3)14), whereas the HIF-P4Hs hydroxylate -Leu-Xaa-Xaa-Leu-Ala-Pro-Tyr-and -Leu-Xaa-Xaa-Leu-Ala-Pro-Ala-sequences (7)(8)(9). All the P4Hs require Fe 2ϩ , 2-oxoglutarate, O 2 , and ascorbate, the 2-oxoglutarate being stoichiometrically decarboxylated during hydroxylation (1)(2)(3). The C-terminal regions of the C-P4H ␣ subunits and the HIF-P4H monomers contain four conserved residues, two histidines, and one aspartate that bind the Fe 2ϩ atom and a basic residue that binds the C-5 carboxyl group of the 2-oxoglutarate (15,16). This basic residue is a lysine in all C-P4Hs, whereas it is an arginine in the HIF-P4Hs (4,5,16).  The peptide-substrate-binding domain of the C-P4Hs has recently been shown to be separate from the catalytic domain and to be located between residues 138 and 244 in the 517residue human ␣(I) subunit (17). Its sequence is distinct from those of other proline-rich peptide-binding modules such as the Src homology 3, WW, enabled/vasodilator-stimulated phosphoprotein, and GYF domains (18 -21) and from that of the profilins (22,23). Poly(L-proline) is an effective competitive inhibitor of the type I C-P4H but only a very weak inhibitor of the type II C-P4H, there also being smaller differences between the two isoenzymes in their binding properties with respect to peptide substrates (10,11,17). The presence of a glutamate and glutamine in the peptide-substrate-binding domain of the ␣(II) subunit in the positions corresponding to Ile 182 and Tyr 233 in the ␣(I) subunit was found to explain most of these differences (17).
We report here on NMR data indicating that a recombinant human C-P4H ␣(I) subunit polypeptide Phe 144 -Ser 244 forms a folded domain consisting of five ␣ helices and one short ␤ strand, its secondary structure being distinct from those of other proline-rich peptide-binding modules. Binding of a peptide substrate to this domain was found to cause significant shifts in many backbone amide resonances. The K d values of this domain and the corresponding ␣(II) subunit domain for peptide substrates and inhibitors were similar to the K m and K i values of these peptides for the type I and type II enzyme tetramers, suggesting that this domain plays a critical role in the binding of peptide substrates to the enzyme.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The recombinant human ␣(I) Phe 144 -Ser 244 polypeptide was prepared by amplifying a cDNA fragment encoding residues Phe 144 -Ser 244 with an NdeI site and a translation start codon preceding the codon for Phe 144 and an XhoI site following the codon for Ser 244 , using Pfu polymerase (Promega). The template was pET15b-␣(I) Gly 138 -Ser 244 (17) in which the codon for Cys 150 (TGC) was converted to that for serine (TCC). The C150S mutagenesis was performed using a QuikChange™ site-directed mutagenesis kit according to the manufacturer's instructions (Stratagene). The recombinant human ␣(II) Met 142 -Ser 242 polypeptide was prepared by amplifying a cDNA fragment encoding residues Met 142 -Ser 242 with an NdeI site preceding the codon for Met 142 and an XhoI site following the codon for Ser 242 , using full-length human C-P4H ␣(II) cDNA (11) as a template. The codon for Cys 148 (TGC) was converted to that for serine (TCC) in the same PCR amplification step. The PCR products were cloned into NdeI-XhoI-digested pET-22b expression vectors (Novagen) in-frame with the sequence encoding a C-terminal histidine tag and transformed into the BL21(DE3) Escherichia coli host strain (Novagen). For expression, 500 ml of Luria-Bertoni medium containing ampicillin (50 g/ml) was inoculated with 5 ml of overnight culture. When A 600 reached 0.8, synthesis of the recombinant proteins was induced with 1 mM isopropyl-␤-Dthiogalactoside for 4 h at 30°C, after which the cells were harvested and disrupted by sonication. The soluble proteins were applied to a nickel-nitrilotriacetic acid-chelating Sepharose column (Amersham Biosciences) equilibrated with 20 mM Bis-Tris and 0.1 M glycine, pH 6.8. The bound proteins were eluted with a 200-ml linear imidazole gradient (0 -0.5 M), and the fractions were analyzed by 16.5% Tris-tricine PAGE. The fractions containing the recombinant polypeptide were pooled, concentrated, and loaded onto a SuperDex 75 HR column (Amersham Biosciences) equilibrated with 10 mM sodium phosphate buffer, pH 6.8, and fractions were analyzed by 16.5% Tris-tricine PAGE. Those containing the pure 13-kDa recombinant protein were pooled, and all experiments were performed with polypeptides containing their histidine tag. Poly(L-proline) binding of the recombinant domains was tested as described previously (17). The 15 N-and 15 N, 13 C-labeled proteins were expressed in the Bio-Express cell growth media U-15 N and U-13 C, 15 N (Cambridge Isotope Laboratories), respectively. Limited proteolysis of the purified recombinant ␣(I) domain was performed with trypsin, and the N termini were sequenced as described (17). The M r values of the trypsin resistant fragments were determined by matrixassisted laser desorption ionization time-of-flight mass spectrometer (Voyager DE-STR, Applied Biosystems).
Peptides-The peptides (Pro-Pro-Gly) 2 and (Gly-Pro-4Hyp) 5 were purchased from Innovagen, (Pro-Pro-Gly) 5 and (Pro-Pro-Gly) 10 from the Peptide Institute, and poly(L-proline) from Sigma. Circular Dichroism Spectroscopy (CD)-CD spectra were recorded with a JASCO J-715 CD-spectropolarimeter, the sample cell temperature being controlled by JASCO PFD-350S. Far-UV spectra were measured using a 1-mm path-length cell for the 185-240 nm region. The step size was 0.2 nm, scan speed 20 nm/min, response 1 s, bandwidth 1 nm, and number of accumulations 4/spectrum. Thermal denaturation of the peptide-binding domain was followed by continuous monitoring of ellipticity changes at a fixed wavelength of 222 nm while the sample was heated at a constant rate of 30°C/h and renaturation was followed by cooling at the same rate. The protein concentration in the samples was 0.1 mg/ml in 10 mM sodium phosphate buffer, pH 6.8, or in 10 mM sodium phosphate buffer and 100 mM urea.
NMR Spectroscopy-The samples used for the NMR measurements were prepared at concentrations of 0.35-1 mM in 10 mM sodium phosphate buffer, dissolved in 90/10% H 2 O/D 2 O at pH 6.8, and urea was added at a final concentration of 100 mM to 15 N, 13 C-labeled samples. To establish the sequential assignment, gradient-selected sensitivity-enhanced HNCA, HN(CO)CA, CBCANH, and CBCA(CO)NH (24 -29) spectra were collected at 15°C on Varian Unity INOVA 600 and 800 MHz NMR spectrometers. All the spectra were processed using the VNMR6.1 software package (Vnmr 6.1b, Varian Inc., Palo Alto, CA). The number of data points on the 13 C and 15 N dimensions was doubled by forward linear prediction before a Fourier transform. Sparky 3.100 (30) was used for spectral assignment and analysis.
Titration of the peptide-binding domain with a synthetic (Pro-Pro-Gly) 2 peptide was performed by adding known amounts of concentrated peptide stock solutions to the 15 N-labeled peptide-binding domain (using peptide to domain concentration ratios of 0:1, 0.5:1, 1:1, 2:1, 5:1, 10:1, and 20:1) and recording a two-dimensional, gradient-selected sensitivity-enhanced 15 N-1 H HSQC (31) spectrum after each addition. The perturbations in the chemical shifts were calculated using the following equation.
Dissociation constants (K d ) for binding were determined, assuming 1:1 binding stoichiometry, from the concentration dependence of the weighted chemical shift changes in which the observed chemical shift change (⌬␦ obs ) is where ⌬␦ max is the chemical shift difference between the bound and free peptide-binding domains, C pbd is the total concentration of the peptidebinding domain, and C is the concentration of bound peptide-binding domain, which is given by the equation, where C pep is the total concentration of the (Pro-Pro-Gly) 2 peptide. The chemical shift changes in the seven most affected residues were used to calculate K d by minimization of the goodness-of-fit parameter (32), as shown in the equation, when both K d and ⌬␦ max were varied (see Fig. 5B) Surface Plasmon Resonance (SPR)-Surface plasmon resonance assays were performed with a BIAcore® 3000 instrument (Biacore AB, Uppsala, Sweden). The peptide-binding domains were immobilized covalently on a hydrophilic carboxymethylated dextran matrix (CM5 sensor chip, Biacore AB) with standard amine coupling chemistry (33) at pH 5.0 at a level of about 1500 resonance units. Binding analyses were carried out in 10 mM sodium phosphate, pH 6.8, containing 0.05% P-20 surfactant (Biacore AB) at a flow rate of 10 l/min. Injection times were 4 min followed by 8 min of dissociation. The bulk effects were subtracted using reference surfaces. The chips were regenerated with 10-s injections of 10 mM glycine, pH 2.0. The steady state binding value, R eq , for each peptide analyte concentration was plotted as a function of total analyte concentration. To derive binding constants, data from these plots were analyzed by means of the steady state affinity-binding model provided in the BIAevaluation software version 3.1.
Isothermal Titration Calorimetry (ITC)-ITC experiments were performed using a VP-ITC titration microcalorimeter (Microcal, Northampton, MA). In the individual titrations, 5-l doses of peptide solutions in 2.5-10 mM concentrations (in 10 mM sodium phosphate buffer, pH 6.8) were injected into a sample cell (volume 1.4476 ml) containing 0.25-0.5 mM of the peptide-binding domain in the above buffer at an interval of 4 min with stirring at 300 rpm. The titration data obtained were corrected by subtracting the heat of dilution of the peptides from the raw data and fitted by the non-linear least squares minimization method using one set of sites model (Levenberg-Marquardt algorithm) to determine the dissociation constant (K d ), binding stoichiometry (n), and change in enthalpy (⌬H) using the Origin software (Microcal).
Other Assays-P4H activity was assayed by a method based on the hydroxylation-coupled decarboxylation of 2-oxo[1-14 C]glutarate (34), and K m values were determined as described previously (16).

RESULTS
Optimization of the Recombinant ␣(I) Polypeptide-Initial experiments were performed using a purified recombinant human C-P4H ␣(I) subunit polypeptide Gly 138 -Ser 244 (17) in which the only cysteine (Cys 150 ) was mutated to serine. The quality of the NMR spectra obtained with this 15 N-labeled polypeptide was not satisfactory, however, and therefore further experiments were performed with recombinant polypeptides lacking a few residues from the N-or C-terminal end (details not shown). A Phe 144 -Ser 244 polypeptide was found to give the best spectra and was therefore used for all the experiments reported here. It was purified to homogeneity based on its histidine tag. Limited proteolysis experiments of this purified domain with trypsin (17) demonstrated that it is proteaseresistant, except that a minor cleavage product was identified by SDS-PAGE. N-terminal sequencing showed that the major and minor bands had intact N termini. Mass spectrometry analysis indicated that the minor product corresponded to a polypeptide that had lost its C-terminal serine and the histidine tag (details not shown).
CD Analysis of the ␣(I) Peptide-binding Domain-Far-UV CD experiments sensitive to protein secondary structure were performed to study the fold and thermal stability of the recombinant ␣(I) polypeptide. The CD spectra showed a typical pattern of ␣-helical proteins, with negative minima at 208 and 222 nm and a positive maximum at 193 nm (Fig. 1A). A thermal scan from 5 to 95°C demonstrated that the recombinant polypeptide displays a continuous, gradual decrease in helical content (Fig. 1B) that is almost completely reversible (Fig. 1, A  and B). The denaturation data indicate that the recombinant domain does not have a very compact tertiary structure and behaves more like proteins in a molten globule state. CD spectra were also recorded for a sample containing urea at the same concentration as used in NMR measurements, 100 mM. The shape of the far-UV curve remained unchanged, but a small increase in ellipticity was detected, indicating a slight increase in the compactness of the fold (Fig. 1C). The difference in the ellipticity between the native and thermally denatured proteins was likewise 20% greater in the urea sample than in the sample without urea (Fig. 1, B and D).
NMR Assignments of the ␣(I) Peptide-binding Domain-Uniformly 15 N-and 15 N/ 13 C-labeled samples of the ␣(I) peptidebinding domain were prepared for NMR studies, the 1 H-15 N HSQC spectrum of the 15 N-labeled domain being shown in Fig.  2. The spectra of the 15 N/ 13 C double-labeled samples had broader lines and poorer dispersion than those of the 15 Nlabeled samples, but the quality of the spectra could be improved by adding 100 mM urea to the samples. All the spectra were acquired at 15°C because of the low thermal stability of the domain revealed by the CD and preliminary NMR studies. Backbone assignments were obtained based on the intraresidual and sequential C␣/C␤ connectivities observed in the three-dimensional NMR experiments HNCA, HN(CO)CA, CB-CANH, and CBCA(CO)NH with the exception of the first few N-terminal residues, part of the histidine tag in the C terminus, and the region between the residues Pro 219 and Glu 232 , where assignments could not be obtained because of spectral overlapping.
The C␣ and C␤ chemical shift indices (35) indicate that the ␣(I) peptide-binding domain is composed of five ␣ helices and a short ␤ strand, Ile 182 -Asp 186 , between the second and third helices (Fig. 3). The first four ␣ helices consist of residues Glu 148 -Thr 159 , Tyr 163 -Asp 178 , Lys 187 -Gln 200 , and Leu 204 -Leu 217 (Fig. 3), whereas the fifth is located in the C terminus of the domain, but its length could not be determined because of the missing assignments. Predictions based on the amino acid sequence (36) suggest, however, that this helix may begin at His 221 . As the predicted locations and sizes of the other four ␣ helices are in good agreement with the data obtained here, the N terminus of helix 5 is shown in Fig. 3 based on such predic-tions. All these data are in good agreement with the measured dipolar couplings (data not shown).
NMR Studies of Peptide Binding-To obtain information on the interaction between the ␣(I) peptide-binding domain and peptide substrates, NMR was used to study binding of the synthetic peptide (Pro-Pro-Gly) 2 (Fig. 5). 6 of the 15 residues with the largest chemical shift changes are located in ␣ helix 3, 3 in helix 4, 2 in helix 1, 1 in helix 2, 1 in the loop between helices 1 and 2, and 1 in the short ␤ strand. None of the 15 residues are located in helix 5, but even this helix had residues showing chemical shift changes exceeding 0.1 ppm, its 3 residues showing the highest chemical shifts being Ile 234 , Met 235 , and Tyr 233 (Fig. 5A). The residues with chemical shift changes exceeding 0.25 ppm were selected for determination of the dissociation constant, K d , for (Pro-Pro-Gly) 2 . The chemical shift changes of these residues were plotted against peptide concentration, and the binding curves were then fitted individually by optimizing the values of the dissociation constant and the maximum chemical shift change. The dissociation constant for (Pro-Pro-Gly) 2 obtained with this method was 2600 Ϯ 500 M (Fig. 5B, Table I).
SPR Studies of Peptide Binding-The interaction of the ␣(I) and the corresponding ␣(II) peptide-binding domains with synthetic peptides of different lengths was studied by real-time biomolecular interaction analysis using the BIAcore biosensor system, which detects changes in the refractive index at the surface of a sensor chip (37). The peptides became bound to the immobilized domain on the chip with very high on and off rates, giving a square-wave type of binding curve indicative of a low binding affinity (Fig. 6A). Reliable determination of the association and dissociation kinetic rate constants (k on and k off ) was thus not possible. Equilibrium dissociation constants (K d ) could nevertheless be determined for the peptides by plotting the steady state values for each peptide analyte concentration as a function of total analyte concentration (Fig. 6B, Table I).
The K d of the ␣(I) peptide-binding domain for the peptide (Pro-Pro-Gly) 2 was about 1 mM, this value being in relatively good agreement with the K d of 2.6 mM measured by NMR and the K m of about 2.6 mM measured for this peptide as a substrate of the type I C-P4H tetramer ( Table I). The K d values measured by SPR decreased with peptide chain length, with the changes being similar to those seen in the K m values for the same peptides as substrates for the type I C-P4H (Table I). The K d of the ␣(I) domain and K m of the type I enzyme for (Pro-Pro-Gly) 10 were highly similar, 30 and 20 M, respectively (Table I). The SPR measurements further indicated that the K d of the ␣(I) domain for (Gly-Pro-4Hyp) 5 was more than one order of magnitude higher than that for (Pro-Pro-Gly) 5 (Table I). The K d of 4 M for poly(L-proline), M r ϭ 5000, was identical to the K i measured for the enzyme tetramer (Table I).
The K d of the ␣(II) peptide-binding domain for the peptide (Pro-Pro-Gly) 2 was much higher than that of the ␣(I) domain and was too high to be determined accurately (Table I). The K d values of the ␣(II) domain for (Pro-Pro-Gly) 5 and (Pro-Pro-Gly) 10 were likewise much higher than those of the ␣(I) domain, with these values being roughly similar to the K m values of the type II C-P4H tetramer for the same peptides (Table I).
The K d of the ␣(II) domain for (Gly-Pro-4Hyp) 5 was again distinctly higher than that for (Pro-Pro-Gly) 5 (Table I). The K d of the ␣(II) domain for poly(L-proline), M r ϭ 5000, was much higher than that of the ␣(I) domain (Table I).
ITC Studies of Peptide Binding-Peptide binding to the re-combinant ␣(I) domain was also investigated using isothermal titration calorimetry. Fig. 7 shows the result of an ITC experiment in which the ␣(I) peptide-binding domain was titrated with the peptide (Pro-Pro-Gly) 10 . The K d values obtained with this technique were in relatively good agreement with those measured by SPR (Table II). The changes in enthalpy observed upon binding of different peptides were negative indicating exothermic interaction, the change correlating with the increase in peptide length and affinity (Table II).

DISCUSSION
The peptide-substrate-binding domain studied here was initially identified as a fragment of the ␣(I) subunit that became bound to poly(L-proline) after limited proteolysis of the human type I C-P4H tetramer by proteinase K, thermolysin, or trypsin (17). Further experiments demonstrated that a recombinant polypeptide corresponding to the ␣(I) domain became effectively bound to poly(L-proline) agarose and could be eluted with (Pro-Pro-Gly) 10 (17). This domain could not be identified simply by sequence criteria, but this region is followed by a region of about 20 residues with a particularly low degree of identity between the ␣(I) and ␣(II) subunits suggesting the presence of a variable interdomain region between this domain and the subsequent domain (17).
The data reported here indicate that the recombinant human ␣(I) polypeptide Phe 144 -Ser 244 forms a folded domain consisting of five ␣ helices and one short ␤ strand. This structure is quite different from those of the other proline-rich peptide-binding domains SH3, WW, EVH1, and GYF (18 -21) and from that of the proline-rich peptide-binding protein profilin (22,23), as they all consist mainly of ␤ strands. The other proline-rich peptide-binding modules typically have a hydrophobic path, with critical aromatic residues that are required for the binding of their ligands (18 -23). The peptide-binding domain of the C-P4Hs also appears to have these characteristic features, as the 15 residues showing the largest chemical shift changes in NMR analysis upon interaction with (Pro-Pro-Gly) 2 were mainly hydrophobic, including 3 tyrosines. The residues showing the highest chemical shift changes were found especially in ␣ helices 3 and 4, all three most affected tyrosines being located in helix 3. 11 of the 15 residues with the highest chemical shift changes, including all 3 tyrosines, were found to be conserved or replaced with similar residues in the C-P4H isoenzymes I and II and C-P4Hs from various species (Fig. 8).
Site-directed mutagenesis has demonstrated that most of the

and K i values of the human type I and type II C-P4H tetramers
The peptide-binding domains were immobilized on CM5 sensor chips. The peptides were tested at various concentrations, and the equilibrium responses were plotted against concentration (steady-state analysis). K d values were calculated by nonlinear least squares analysis of the data points from at least three independent determinations. marked difference in poly(L-proline) binding between the type I and type II C-P4H tetramers can be explained by the presence of a glutamate and glutamine in the ␣(II) subunit in the positions corresponding to those of Ile 182 and Tyr 233 in the ␣(I) subunit (17). The same amino acid substitutions also explain most of the difference in the binding of peptide substrates, but the difference in the K m values for (Pro-Pro-Gly) 10 between the type I and type II C-P4Hs is only 5-fold, whereas that in the K i values for poly(L-proline), M r ϭ 44,000, is more than 1000-fold (11). The present NMR analysis indicates that Ile 182 showed a negligible chemical shift change of about 0.05 ppm upon interaction with (Pro-Pro-Gly) 2 , whereas Tyr 233 showed a modest change of about 0.1 ppm. It therefore seems possible that these two residues may not be directly involved in the binding of (Pro-Pro-Gly) 2 , and the less effective binding to the ␣(II) domain may primarily be because of indirect inhibitory effects of the glutamate and glutamine residues. Nevertheless, it is also possible that (Pro-Pro-Gly) 2 was too short to interact with Tyr 233 and thus a direct contribution of this residue to the binding of longer peptides and especially of poly(L-proline) cannot be excluded. The ␣(I) domain contains several other conserved aromatic residues in addition to Tyr 193 , Tyr 196 , Tyr 199 , and Tyr 233 (Fig. 8), but all except two of them showed only very small chemical shift changes (less than 0.03 ppm) upon the binding of (Pro-Pro-Gly) 2 , the two exceptions being Phe 151 and Trp 169 , with changes of about 0.11 ppm in both cases (Fig. 5).
The K m values of the C-P4Hs for peptide substrates and the K i values for poly(L-proline) decrease markedly with increasing chain length of the substrate or inhibitor (1). These effects are not because of the charges in the N and C termini of the peptides as such charge effects are negligible for substrates longer than hexapeptides and as the length dependence is also seen with peptides having protected N and C termini (38). The length dependence cannot be explained by structural properties such as a higher tendency of longer peptides to form triple helices either, as the C-P4Hs act only on non-triple-helical peptides (1)(2)(3). To explain the length dependence mechanisms have been proposed in which the two peptide-binding sites in the enzyme tetramer act processively and bind to the same peptide, which would prevent dissociation of the enzyme-substrate complex between successive interactions with a long peptide having multiple substrate sites (1,39). The present data indicate, however, that the effects of peptide chain length are seen even with the individual peptide-binding domains; the K d values determined for various peptides being very similar to their K m and K i values for the enzyme tetramers. It is thus obvious that longer peptides become bound to the peptidebinding domain more effectively than shorter ones even when the domain is not present in the enzyme tetramer and in the absence of any catalysis. The great similarities between the K d values measured for the recombinant domains and the K m and K i values determined for the enzyme tetramers indicate that it is the peptide-binding domain rather than the catalytic domain that appears to be critical for most, if not all, binding properties of various peptide substrates and inhibitors with respect to the enzymes.
The catalytic cycle of P4Hs consist of two half-reactions (1-3), the first half-reaction coinciding with the formation of a ferryl ion bound to His 412 , Asp 414 , and His 483 (in the case of the human ␣(I) subunit sequence), while the ferryl ion then hydroxylates a proline residue during the second (1-3). The C-4 of the proline to be hydroxylated must thus point toward the ferryl ion at the catalytic site. The present data indicate that formation of the hydroxy group markedly reduces the affinity of the substrate for its binding domain, as demonstrated by the differences in the K d values for (Pro-Pro-Gly) 5 and (Gly-Pro-4Hyp) 5 . This suggests that a change in the affinity of the hydroxylated substrate to the peptide-binding domain contributes to the release of the product and prevents the binding of hydroxylated sequences to the enzyme.
The present data thus indicate that the sequence of about 100 residues located between residues 140 and 245 in the C-P4H ␣ subunits forms a folded domain consisting mainly of ␣ helices, possessing many hydrophobic and aromatic residues that show significant changes in chemical shifts in NMR analysis upon the binding of peptide substrates. Many of the characteristic properties found in the binding of peptide substrates and inhibitors to the type I and type II C-P4Hs can be explained by the properties of this domain rather than by those of the catalytic domain. We have recently obtained crystals of this domain that diffract to at least 3 Å resolution (40), and attempts to determine the complete crystal-based structure are now in progress.  (n), and change in enthalpy (⌬H) of the ␣(I) peptide binding domain, determined by isothermal titration calorimetry The data were measured using 50-500 M of the peptide binding domain and 10-20-fold molar excesses of the different peptides at 25°C and fitted using a nonlinear least squares algorithm. The values in parentheses are the errors associated with each fitted parameter for a given experiment. (Pro-Pro-Gly) 2 5000 (Ϯ400) 1.5 (Ϯ0.84) Ϫ4348 (Ϯ2647) (Pro-Pro-Gly) 5 100 (Ϯ15) 0.9 (Ϯ0.17) Ϫ10019 (Ϯ842) (Gly-Pro-4Hyp) 5 1200 (Ϯ100) 0.8 (Ϯ0. 21) Ϫ6548 (Ϯ1436) (Pro-Pro-Gly) 10 50 (Ϯ2) 0.6 (Ϯ0.008) Ϫ12880 (Ϯ222) Poly(L-Pro), M r ϭ 5000 3 (Ϯ0.3) 0.4 (Ϯ0.007) Ϫ20010 (Ϯ440)