Collagen prolyl 4-hydroxylase tetramers and dimers show identical decreases in Km values for peptide substrates with increasing chain length: mutation of one of the two catalytic sites in the tetramer inactivates the enzyme by more than half.

The collagen prolyl 4-hydroxylases (collagen P4Hs, EC 1.14.11.2) play a key role in the synthesis of the extracellular matrix. The vertebrate enzymes are alpha(2)beta(2) tetramers, the beta subunit being identical to protein disulfide isomerase (PDI). The main Caenorhabditis elegans collagen P4H form is an unusual PHY-1/PHY-2/(PDI)(2) mixed tetramer consisting of two types of catalytic alpha subunit, but the PHY-1 and PHY-2 polypeptides also form active PHY/PDI dimers. The lengths of peptide substrates have a major effect on their interaction with the P4H tetramers, the K(m) values decreasing markedly with increasing chain length. This phenomenon has been explained in terms of processive binding of the two catalytic subunits to long peptides. We determined here the K(m) values of a collagen P4H having two catalytic sites, the C. elegans mixed tetramer, and a form having only one such site, the PHY-1/PDI dimer, for peptides of varying lengths. All the K(m) values of the PHY-1/PDI dimer were found to be about 1.5-2.5 times those of the tetramer, but increasing peptide length led to identical decreases in the values of both enzyme forms. The K(m) for a nonhydroxylated collagen fragment with 33 -X-Y-Gly-triplets but only 11 -X-Pro-Gly-triplets was found to correspond to the number of the former rather than the latter. To study the individual roles of the two catalytic sites in a tetramer, we produced mutant PHY-1/PHY-2/(PDI)(2) tetramers in which binding of the Fe(2+) ion or 2-oxoglutarate to one of the two catalytic sites was prevented. The activities of the mutant tetramers decreased to markedly less than 50% of that of the wild type, being about 5-10% and 20-30% with the enzymes having one of the two Fe(2+)-binding sites or 2-oxoglutarate-binding sites inactivated, respectively, while the K(m) values for these cosubstrates or peptide substrates were not affected. Our data thus indicate that although collagen P4Hs do not act on peptide substrates by a processive mechanism, prevention of hydroxylation at one of the two catalytic sites in the tetramer impairs the function of the other catalytic site.

The hydroxylation reaction catalyzed by P4Hs requires FeP 2+ 4 long peptide substrates (2,21,22). According to this model the enzyme-substrate complex is formed at one catalytic site upon the first encounter, and before the first hydroxylation is complete, the other catalytic site has a high chance of encountering another region of the peptide substrate (21,22). Such a mechanism leads to a low KB m B for a long peptide by overcoming diffusional constraints on the association between the enzyme and the various substrate sites present in the peptide. The processive mechanism requires the presence of two active catalytic sites. The catalytic sites are located in the C-terminal regions of the collagen P4H α subunits and contain four conserved catalytically critical residues, two histidines and one aspartate that bind the FeP 2+ P atom and a lysine that binds the C-5 carboxyl group of 2oxoglutarate (23). The peptide-substrate-binding domain of the collagen P4Hs is separate from the catalytic domain and is located between residues 138 and 244 in the 517-residue human α(I) subunit (24).
To study the proposed processive reaction mechanism further, we determined the KB m B values of the PHY-1/PHY-2/(PDI-2)B 2 B tetramer and the PHY-1/human PDI dimer for peptide substrates of different lengths, the PHY-1/human PDI dimer being chosen because it is formed much more effectively than any PHY/PDI-2 dimer. Lower KB m B values with increasing substrate chain length were observed in the case of both the enzyme tetramer and the dimer, the results suggesting that these occur on account of the higher affinity of the peptidesubstrate-binding domain for long peptides rather than through processive binding. To study the separate roles of the two catalytic sites in one collagen P4H tetramer, we generated mutant recombinant C. elegans PHY-1/PHY-2/(PDI-2)B 2 B tetramers in which the catalytic site of one α subunit was inactivated by mutation of either the FeP 2+ P-binding aspartate or the 2oxoglutarate-binding lysine residue. The P4H activities of the mutant tetramers were markedly less than 50% of that of the wild-type, indicating that the remaining single wildtype catalytic site in the mutants does not function entirely independently.  (Invitrogen) were cultured on plates in TNM-FH medium (Sigma) supplemented with 10% fetal bovine serum (Euroclone). The cells, seeded at a density of 0.6 x 10P 6 P/ml, were coinfected at a multiplicity of 5 with viruses codingP Pfor the C. elegans PHY-1 (15), PHY-2 (20) and the C. elegans PDI-2 polypeptides (25), or the C. elegans PHY-1 and human PDI polypeptides (26). The cells were harvested 72 h after infection, washed with a solution of 0.15M NaCl and 0.02M phosphate, pH 7.4, homogenized in a 0.1M NaCl, 0.1 M glycine, 0.1% Triton X-100, 10 µM dithiothreitol and 0.01 M Tris buffer, pH 7.8, and centrifuged at 10,000 x g for 20 min.

PHY-1/Human PDI Dimer in Insect Cells -High Five insect cells
Samples of the Triton X-100 soluble supernatants were analyzed by 8% nondenaturing PAGE followed by Coomassie Blue staining.

Production of Recombinant Non-hydroxylated Collagen Polypeptides in the Yeast Pichia
pastoris -To generate a Pichia strain expressing recombinant nonhydroxylated type I collagen α1 chain fragments of 510 and 100 amino acids, cDNA products originating from the codons for amino acids 683 and 1093 of the proα1(I) chain, respectively, and extending to the codon for the last amino acid of the collagenous domain were generated by PCR. These fragments had an artificial EcoRI site at their 5' ends and a stop codon followed by a NotI site at their 3' ends. After digestion with the corresponding restriction enzymes, the cDNA fragments were ligated to an EcoRI-NotI-digested pPIC9K expression vector (Invitrogen) in frame with the yeast α-mating factor secretory signal. The constructs were linearized with SalI and electroporated into a GS115 Pichia strain (Invitrogen) according to the manufacturer's instructions (27). The strains were cultured in 100-ml shaker flasks in a buffered glycerol complex medium, pH 6.0, with 1 g/l yeast extract and 2 g/l peptone.
Expression was induced in a buffered minimal methanol medium, pH 6.0, and methanol was added every 12 h to a final concentration of 0.5%. The culture medium was collected 60 h after induction, concentrated in an Ultrafree 0.5 centrifugal filter unit with a Biomax-5 membrane (Millipore) and the recombinant type I collagen polypeptide fragments were analyzed by 12% SDS-PAGE followed by Coomassie Blue staining.
The mutagenesis steps were performed in a pVL1392 vector (Invitrogen) containing the fulllength C. elegans phy-1 (15) and phy-2 (20) cDNAs using the QuikChangeP TM elegans P4H tetramers, the cells were coinfected with the viruses coding for the mutant PHY-7 nondenaturating PAGE gels. In the case of the mutant tetramers produced with the wild-type PHY-1 and mutant PHY-2 subunits, the amount of activity generated by the PHY-1/PDI-2 dimer was subtracted from the activity values. The activity of the PHY-1/PDI-2 dimer was determined based on the amount generated in insect cells expressing only the PHY-1/PDI-2 dimer and densitometry comparisons of the amount of this dimer and the dimer formed in addition to the mutant tetramer. Since assembly of the recombinant PHY-2/PDI-2 dimers in insect cells is very inefficient, lying below the detection limit of Western blot analysis (20), the activity generated by the recombinant PHY-2/PDI-2 dimer was regarded as nonsignificant and was not subtracted. KB m B values for the peptide substrates of different lengths and for the cosubstrates FeP 2+ P and 2-oxoglutarate were determined as described previously (29). In some experiments the amount of 4-hydroxyproline formed was determined by a colorimetric method in samples hydrolyzed with 6 M HCl at 120°C overnight (30).

Increasing Substrate Chain Length Reduces the KB m B Values of the PHY-1/PHY-2/(PDI-2)B 2 B
Tetramer and PHY-1/Human PDI Dimer Equally -According to the hypothesis of processive action of the two peptide-binding sites in the collagen P4H tetramer, an enzyme with two such sites should be much more efficient in hydroxylating long peptide substrates than an enzyme with only one site, whereas the synergistic relation between the two sites should not be evident in the hydroxylation of short peptides (21,22).
To study the suggested co-operation between the two peptide-binding sites in more detail, a recombinant C. elegans PHY-1/PHY-2/(PDI-2)B 2 B tetramer, i.e. a P4H with two peptide-binding and catalytic sites, and the PHY-1/human PDI dimer, i.e. a P4H with one peptide-binding and one catalytic site, were produced in insect cells (  (Table I), but the chain length of the peptide substrate was found to have an identical effect on both enzyme forms, as their KB m B values for the 510-aminoacid fragment were 0.03% of those for (Pro-Pro-Gly)B 5 B , and 0.8-1.4% of those for (Pro-Pro-Gly)B 10 B (Fig. 3 and Table I). The effect of increasing peptide chain length on the KB m B values was also seen when expressed in terms of the molar concentration of the hydroxylatable -X-Pro-Gly-triplets (Table I). Interestingly, the KB m B values of the enzyme tetramer and dimer for the 100-amino-acid fragment, when expressed per peptide, were 24 and 28% of those for (Pro-Pro-Gly)B 10 B , respectively, although these two peptides differ by only one in terms of the number of possible hydroxylation sites. The KB m B values for this collagen fragment thus clearly corresponded to the number of all -X-Y-Gly-triplets in it rather than the number of -X-Pro-Gly-triplets (Table I).  (Fig. 4). The expression levels of the mutant PHY-1 and PHY-2 polypeptides were found to be comparable to those of the wild-type polypeptides (Fig. 4). and PDI-2 (Fig. 5). The cells were harvested 72 h after infection, homogenized in a Triton X-100 buffer, and the soluble fractions analyzed by nondenaturing PAGE followed by Western blotting with a PHY-1 antibody (Fig. 6). All the mutant PHY polypeptides became assembled into mixed tetramers with the wild-type polypeptides, a small amount of wild-type or mutant PHY-1/PDI-2 dimer also being detected (Fig. 6).

Inactivation of One Catalytic Site in the Collagen
The P4H activity of the mutant enzyme tetramers generated in the insect cell homogenates was analyzed using (Pro-Pro-Gly)B 10 B as a substrate (Table II). The amounts of wild-type and mutant enzyme tetramers in the activity assay reactions were adjusted to be equal on the basis of densitometry of the tetramer bands on the ECL Western blots of nondenaturating PAGE gels (Fig. 6). Furthermore, the activity generated by the wild-type PHY-1/PDI-2 dimer (Fig. 6, lanes 1, 3 and 5), 4-6% (data not shown), was subtracted from the activity values as described in Materials and Methods. Inactivation of the iron-binding aspartate in either the PHY-1 or the PHY-2 polypeptide led to a marked reduction in enzyme activity, the values being 5-8% of that of the wild-type tetramer (Table II) (Table II). The activities of all the mutant tetramers were thus reduced by more than 50%, inactivation of one of the iron-binding sites having a more severe effect, however, than inactivation of one of the 2-oxoglutarate-binding sites, the former resulting in about a 3-5-fold lower residual activity (Table II). The catalytic sites in the PHY-1 and PHY-2 polypeptides seem to be equally important, as no difference was observed between the activity values obtained with the enzymes having a mutation in either the PHY-1 or PHY-2 polypeptide (Table II). None of the mutations affected the peptide substrate binding of the enzymes, as the KB m B values for (Pro-Pro-Gly)B 10 B were, within the accuracy of the assays, identical between the wild-type and mutant enzymes (Table II). Mutation of the iron-binding or 2-oxoglutarate-binding site in one of the two catalytic subunits had no effect on the binding of these cosubstrates to the remaining wild-type subunit, either, as the KB m B values of the mutant enzymes for these cosubstrates were identical to those of the wild-type enzyme (Table III).
The P4H activity of the mutant enzyme tetramers was also determined with saturating concentrations of the recombinant 100 and 510-amino-acid nonhydroxylated α1(I) collagen fragments as substrates (Table IV). The results were very similar to those obtained with (Pro-Pro-Gly)B 10 B above. Mutation of the iron-binding aspartate or the 2-oxoglutarate-binding lysine in either of the PHY polypeptides reduced the enzyme activity to 5-8% and 18-33% of that of the wild-type enzyme, respectively (Table IV).

The Less Severe Effect of Inactivation of One of the Two 2-Oxoglutarate-binding Sites Is
Not Due to an Increased Uncoupled 2-Oxoglutarate Decarboxylation -A surprising finding was that mutation of an iron-binding residue in one of the two catalytic sites had a 3-5-fold more severe effect on the enzyme activity than the abolition of 2-oxoglutarate-binding to one of the catalytic subunits. One possible explanation for this difference is that binding of the  (Table V) to those measured with the 2-oxoglutarate decarboxylation assay (Table II).
Essentially identical values were obtained in both assays indicating that inactivation of one of the two 2-oxoglutarate-binding sites in a tetramer indeed had a less severe effect on the rate of the hydroxylation reaction than mutation of one of the two iron-binding sites.

DISCUSSION
Vertebrate collagen P4Hs are tetrameric enzymes containing two catalytic α subunits (1-3). It has previously been shown that the chain length of the peptide substrate has a major effect on the KB m B values of the vertebrate enzymes, which decrease markedly with increasing chain length, while the maximal reaction velocity is not affected (1)(2)(3). This phenomenon has been explained by a processive mechanism of binding of the two peptide-substrate-binding sites of the collagen P4H tetramers, leading to much faster binding by overcoming the diffusional constraints on the rate of association between the enzyme and the individual hydroxylatable sites (21,22). Here we studied the KB m B values of a C. elegans PHY-1/PHY-  (Table I), possibly due to small conformational differences between the enzyme tetramer and dimer.

12
The peptide-substrate-binding domain of human type I collagen P4H has recently been located between residues 138-244 in the α(I) subunit and shown to be separate from the catalytic C-terminal region (24). Determination of the KB d B values of the recombinant human type I and type II collagen P4H peptide-substrate-binding domains for the binding of synthetic peptides varying in lengths from 2 to 10 -Pro-Pro-Gly-triplets showed that the length dependence is observed even with the individual domains, the KB d B values measured with the domains being very similar to the KB m B values for the enzyme tetramers (31). The results obtained here provide further evidence for the hypothesis that the peptide length dependence most probably results from a more effective binding of longer peptides to the peptide-substrate-binding domain rather than from the processive action of two binding sites.
The KB m B values of the collagen P4H dimer and tetramer for the 100-amino-acid type I collagen α1 chain fragment were 24-28% of those for (Pro-Pro-Gly)B 10 B , although the fragment has only one additional possible hydroxylation site relative to the synthetic peptide (Table I).
The interaction of peptide substrates with collagen P4Hs has been shown to be affected by the amino acid present in the X position of the -X-Pro-Gly-triplets to be hydroxylated, proline in the X position probably giving the highest maximal reaction velocity, and by amino acids in other parts of the peptide (2). The present data clearly indicate that interaction of the 100-residue collagen fragment with the P4H tetramer and dimer was influenced by all its -X-Y-Gly-triplets rather than only the -X-Pro-Gly-triplets.
Surprisingly, mutation of the iron-binding aspartate or the lysine that binds the C-5 carboxyl group of the 2-oxoglutarate in either of the PHY polypeptides was found to inactivate the C. elegans PHY-1/PHY-2/(PDI-2)B 2 B tetramer by more than half, indicating that the remaining wild-type catalytic site is not capable of functioning entirely independently. This cannot be due to impaired binding of FeP 2+  (1)(2)(3)23). Binding studies with the peptides (Gly-Pro-4Hyp) 5 and (Gly-Pro-Pro) 5 have shown that the K d of the peptide-substrate-binding domain of the human type I collagen P4H for the hydroxylated peptide is more than one order of magnitude higher than that for the nonhydroxylated peptide, so that the peptide-substrate-binding domain most probably contributes to the release of the hydroxylated product (31). When the hydroxylation reaction in one of the two catalytic sites of a collagen P4H tetramer is inhibited by a mutation, it is highly likely that the long peptide substrate will remain bound at the mutant subunit, thus interfering with the "free search" performed by the functional catalytic subunit and the peptide substrate for new hydroxylation events.
The two catalytic sites in the C. elegans PHY-1/PHY-2/(PDI-2) 2 tetramer seem to have equally important roles in the hydroxylation of a peptide substrate, as essentially identical activity values were obtained with enzymes having a catalytic site mutation in either the PHY-1 or PHY-2 polypeptide. Although the activity of the mutant enzymes was reduced by more than half with respect to that of the wild-type enzyme, our results clearly show that the two catalytic sites are both functional in the enzyme tetramer. An unexpected finding was, however, that inactivation of one of the two iron-binding sites had a more severe effect on the enzyme activity than that of one of the two 2-oxoglutarate-binding sites. Mutation of either the iron-binding aspartate or the positively charged residue that binds the C-5 carboxyl group of the 2-oxoglutarate at both catalytic sites of the human type I collagen P4H tetramer and the lysyl hydroxylase homodimer and at the single catalytic site of the monomeric A. thaliana P4H has been shown to inactivate these enzymes completely (23,(32)(33)(34), and the present data indicate a complete inactivation of the PHY-1/PDI dimer by these mutations (Table II). A possible explanation for the marked difference between the two types of mutant tetramer studied here is that binding of the peptide substrate to the mutant subunit caused a conformational change that was more severe in the case of the mutants involving one of the two iron-binding sites than those involving one of the 2-oxoglutarate-binding sites.
The present data show that efficient hydroxylation of long peptide substrates by collagen