Structural enzymology binding studies of the peptide‐substrate‐binding domain of human collagen prolyl 4‐hydroxylase (type‐II): High affinity peptides have a PxGP sequence motif

Recombinant PSB‐II is an α helical and monomeric protein

PSB‐II was expressed and purified to homogeneity using a construct of residues Met142‐Glu236 of human C‐P4H‐II (corresponding to PSB‐I amino acids Met144‐Glu238, (Fig. 1). A His‐tag precedes the PSB‐II sequence such that the N‐terminal sequence of the construct is Met‐His₆‐Met‐Leu‐Ser‐ (Fig. 1). According to static light scattering (SLS) measurements, PSB‐II is a monodisperse and monomeric protein with an absolute molecular weight of 12 kDa (Supporting Information, Fig. S1). Circular dichroism (CD) spectroscopy studies indicate that this construct is well‐folded with mainly (~75%) α‐helical structure [Supporting Information, Fig. S2(A,B)]. The melting temperature (T_m) is 60.3°C based on its CD melting curve [Supporting Information, Fig. S2(C)].

Calorimetric binding studies with proline‐rich peptides

The binding affinities of PSB‐II for several nine‐residue proline‐rich peptides were studied with isothermal titration calorimetry (ITC), including (PPG)₃ and (P)₉. Of the latter peptides, the structures of the corresponding PSB‐I complexes (PSB‐I being part of DD‐I) are also known.16 In contrast to PSB‐I, having high affinity for (P)₉ (K_d = 9.8 μM) and (PPG)₃ (K_d = 143.5 μM),16 the affinity of (P)₉ and (PPG)₃ for PSB‐II is low and nondetectable, respectively, as measured in these ITC experiments (Table 1, Supporting Information, Fig. S3). In addition, the affinities of the PPG‐PAG‐PPG (PAG), PPG‐PRG‐PPG (PRG), and PPG‐PEG‐PPG (PEG) peptides were determined. The affinities of these three peptides were studied because in the procollagen sequence, after Hyp (34%), the most common residues in the Y‐position are alanine (18%) and arginine (13%).6 Glutamate is rarely found in the Y‐position (2%), but it is a common residue (13%) in the X‐position.6 Interestingly, the ITC experiments showed that the PAG, PRG, and PEG peptides bind with significant affinity to PSB‐II with K_d values of 19.7 μM, 6.8 μM, and 29.2 μM, respectively (Table 1, Supporting Information, Fig. S3). The binding of these peptides to PSB‐II is enthalpy driven (Fig. 3). The much more favorable binding enthalpy term for the PRG and PEG peptides, as compared to the PAG peptide, is to a great extent compensated by a more unfavorable binding entropy term.

Crystal structure of the unliganded PSB‐II

PSB‐II crystallized in the P3₂21 space group with one molecule in the asymmetric unit, and a complete X‐ray data set was collected at 1.87 Å resolution (Table 2). The electron density maps were clearly interpretable from Leu145 to Glu236 (to simplify comparisons the PSB‐I amino acid numbering is used from here onward, the actual numbers in the PSB‐II sequence are Leu143 and Glu234, respectively). The overall TPR fold of the PSB‐II structure is very similar to that of PSB‐I. A root mean square deviation of 1.2 Å is calculated for 92 corresponding Cα atoms when superimposing (using the SSM‐option17 of COOT18) the PSB‐II structure on the PSB‐I structure (PDB code 2V5F [Protein Data Bank, http://www.rcsb.org]). The small rearrangements of the side chains of Tyr196, Arg223 and Asp192 (Fig. 4) correlate with differences in crystal contacts, being that in this structure of the PSB‐I domain, the binding groove is complexed with the C‐terminal His‐tag of a neighboring molecule.

Crystallographic binding studies with proline‐rich peptides

PSB‐II cocrystallization experiments were carried out with (PPG)₃ and (P)₉. PSB‐II crystals were obtained (in the same condition as used for the unliganded crystals) in the presence of 5 mM (PPG)₃, but the peptide was not present in the structure. However, the structure of a PSB‐II‐peptide complex, refined at 2.0 Å resolution (Table 2), and referred to as the PSB‐II_(P)₉ structure, was obtained on cocrystallization with (P)₉. The crystal packing of this PSB‐II_(P)₉ complex crystal form is different from the unliganded PSB‐II structure although the space group is the same and the cell dimensions are similar. All nine prolines of (P)₉ could be built in electron density, and the C‐terminal part of the peptide is best defined [Fig. 5, Supporting Information Fig. S4(A)]. The peptide is bound in a PPII‐helix conformation (Supporting Information, Table SI). The C‐terminal region, Pro7–Pro9, interacts tightly with the binding groove via direct hydrogen bonds and via pi‐stacking interactions. Direct hydrogen bonds are formed between the OH groups of Tyr158 and Tyr193 and the peptidyl oxygens of Pro7 and Pro8, respectively. The C‐terminal carboxylate oxygen atoms (of Pro9) are hydrogen‐bonded to the PSB‐II Arg155 and Asn159 side chains (Fig. 5) and this carboxylate group is well‐defined by its corresponding electron density [Supporting Information Fig. S4(A)], indicating that other possible modes of binding (reverse or shifted) occur at most with low occupancy. The N‐terminal prolines (Pro1–Pro3), instead, are interacting only loosely via one water‐mediated hydrogen bond to Gln233 [Fig. 5(B)]. However, this N‐terminal peptide region interacts also with symmetry‐related molecules [Fig. 5(A)]. These additional interactions, caused by the crystal packing, rationalize why it has been possible to capture this mode of binding, despite the low affinity of the peptide, as suggested by the ITC studies. There is also a water‐mediated hydrogen bond interaction between the side chain of Asn227 and the main chain oxygen of Pro5. The Pro8 pocket, occupied by the Pro7 of (P)₉, is lined by PSB‐II Tyr158, Tyr193 and Tyr196, and the Arg223‐Asp192 salt bridge. Only the side chain of Tyr196 and the Arg223‐Asp192 salt bridge have different conformations when comparing the unliganded PSB‐II and PSB‐II_(P)₉ structures.

PSB‐II was cocrystallized with PAG in very similar conditions as used for the unliganded PSB‐II and the PSB‐II_(P)₉ complex. The structure of this complex, refined at 1.48 Å (Table 2), is referred to as the PSB‐II_PAG structure. The packing of the molecules is the same as the crystal packing of the unliganded PSB‐II and different from the PSB‐II_(P)₉ crystal form. In this crystal form, the neighboring molecule is not interacting with the N terminus of the peptide, as seen in the PSB‐II_(P)₉ crystal form. The presence and conformation of the PAG peptide are clearly defined by its electron density [Fig. 6(A) and Supporting Information, Fig. S4(B)] and all nine residues of the PAG peptide were built. The most C‐terminal PPG triplet of the PAG peptide interacts tightly with the peptide‐binding groove in such a way that the X‐position proline of the third triplet, Pro7, binds in the Pro8 pocket (Fig. 6). The carboxylate oxygens of the glycine of this triplet (Gly9, C‐terminus of the PAG peptide) are hydrogen‐bonded to the side chains of Asn159 and Arg155, suggesting, like in the PSB‐II_(P)₉ complex, that other modes of binding (shifted or reversed) occur most at low occupancy. The side chains of Tyr193 and Tyr158 form direct hydrogen bonds with the PAG peptide oxygens like in the PSB‐II_(P)₉ complex. However, in the PAG complex, PSB‐II Tyr193 is interacting with the C‐terminus of the peptide (Fig. 6), whereas in the PSB‐II_(P)₉ complex, Tyr193 is hydrogen‐bonded to Pro8 (Fig. 5). Unlike in the PSB‐II_(P)₉ complex structure, also peptide residues 3–5 interact tightly with the PSB‐II domain. The X‐position proline of the second triplet of the PAG peptide (Pro4) is bound in the Pro5 pocket lined by Tyr196, Phe/Tyr199, Tyr230 and Phe231 (Figs. 2 and 6). The conformation of the Asn227 side chain is well‐defined by the electron density map [Supporting Information Fig. S4(B)] and the mode of binding of the PAG peptide allows for the formation of direct hydrogen bonds between the OD1 and the ND2 side chain atoms of Asn227 and the peptide backbone N atom and carbonyl oxygen, respectively, of Ala5 of the middle triplet of the PAG peptide. Tyr196 is also involved in the hydrogen bond network between the PSB‐II and the PAG by forming one water and one DMSO‐mediated hydrogen bond with the PAG peptide [Fig. 6(B)]. The N‐terminal triplet of the PAG peptide is not interacting with the groove and this triplet has weak electron density in the structure [Fig. 6(A)].

The cocrystallization experiments with the PRG and PEG peptides produced the same crystal form as with the PAG peptide and as with the unliganded PSB‐II crystal form. The structures of these complexes, refined at 1.55 Å and 1.68 Å (Table 2) and referred to as the PSB‐II_PRG and PSB‐II_PEG structures, respectively, are very similar to the PSB‐II_PAG structure and the mode of peptide binding is identical with the PAG complex, such that the C‐terminus of the peptide has well‐defined electron density whereas the N terminus is weakly bound [Supporting Information, Fig. S5(A‐D)]. In this mode of binding, the side chain of the Y‐position residue of the second triplet (being either Ala, Arg, or Glu in the PAG, PRG, and PEG structures, respectively) is pointing to the bulk solvent. In the PRG and PEG structures, the side chains of Arg5 and Glu5, respectively, are disordered and have only weak density in the refined electron density maps [Supporting Information, Fig. S5(B,D)].

Cloning, expression and purification of human PSB‐II

The PSB‐II construct was generated using standard molecular biology protocols using a codon optimized gene for the full‐length CP4H α(II) subunit (GenScript). The construct Met142‐Glu236 [corresponding to Phe144‐Glu238 of human PSB‐I (Fig. 1) was amplified from this template using a forward primer containing the NdeI site (underlined) (5′‐GAAGGAGATATACATATGCACCACCACCACCACCACATGCTGAGCGTG‐3′) and a reverse primer containing the BamHI site (underlined) followed by a STOP codon (TAA) (5′‐GAGCTCGAATTCGGATCCTTATTCTTCTTCCAGCAGCTGTTCAAAATAGCG‐3′). The amplified product was digested using NdeI and BamHI. This was then ligated into a linear (double digested) pET22b(+) plasmid (Novagen) and its sequence was verified by DNA sequencing. This N terminally His‐tagged PSB domain was expressed using Rosetta™(DE3)pLysS cells in LB medium containing 100 μg/mL ampicillin and 34 μg/mL chloramphenicol. The cells were grown at 37°C until the OD₆₀₀ reached 0.6–0.8 after which they were induced for protein expression using 0.5 mM isopropyl β‐D‐1‐thiogalactopyranoside and kept at 20°C overnight. The cells were harvested and resuspended in lysis buffer composed of 50 mM tris, 100 mM NaCl, and 100 mM glycine at pH 8.0. After sonication, the supernatant was loaded on a Talon® metal affinity resin column (Clontech Laboratories Inc.), washed with lysis buffer and eluted in the same buffer containing 250 mM imidazole. The fractions were pooled, concentrated, and loaded on a Superdex75 16/600 preparative grade size‐exclusion chromatography (SEC) column (GE Healthcare), preequilibrated with SEC buffer composed of 20 mM tris, 50 mM NaCl, and 50 mM glycine at pH 8.0. The final protein solution was concentrated to about 6 mg/mL (concentration measured with a NanoDrop 1000, Thermo Scientific), and stored in SEC buffer in 100 μL aliquot volumes at −70°C.

Static light scattering

SLS analysis of PSB‐II was carried out using the miniDAWN™ TREOS multiangle SLS detector (Wyatt Technologies) in on‐line mode. For these experiments, 100 μL of the purified PSB‐II sample was filtered (with 0.1 μm pore size) and loaded onto a Superdex200 10/300GL column (GE Healthcare) preequilibrated with SEC buffer. Before entering the SLS detector, the samples flow through a RI‐101 refractive index (RI) detector (Shodex). The RI‐signal is used to measure the concentration of the protein sample. Molecular mass, polydispersity, and other analyses were carried out using the ASTRA software (Wyatt Technologies).

Circular dichroism

The CD spectra of purified PSB‐II were obtained using a Chirascan™ CD spectrophotometer (Applied Photophysics Ltd.) in a quartz cuvette with 1 mm path length. PSB‐II was diluted to 0.1 mg/mL using filtered milli‐Q water, the final buffer conditions being 2 mM tris, pH 7.8, 5 mM glycine, and 5 mM NaCl. A wavelength scan from 180 nm–260 nm was used to analyze the secondary structure of the protein. Thermal denaturation was recorded by measuring the CD spectrum between 190 nm and 260 nm using a QUANTUM temperature controller with a 2°C step size at 1°C/min ramp rate with ±0.2°C tolerance. Data analysis was carried out using Pro‐Data Viewer (Applied Photophysics Ltd.), CDNN (http://bioinformatik.biochemtech.uni-halle.de/cdnn), and Global3 (Applied Photophysics Ltd.).

Isothermal titration calorimetry

The ITC experiments were performed using a MicroCal iTC‐200 microcalorimeter (Malvern). All the used peptides were ordered from TAG Copenhagen A/S (Denmark). In the ITC experiments, 280 μL of PSB‐II protein (25 μM) was loaded into the sample cell and 40 μL of each peptide (1 mM) was loaded into the injection syringe. PSB‐II and the peptides were dissolved in the SEC buffer. The experiments were performed according to the instructions provided by the manufacturer. Titration measurements of peptides into the buffer (without protein) were used for the baseline correction calculations. The titrations were carried out at 25°C with a stirring rate of 750 rpm. Titrations for binding were initiated by one injection of 0.4 μL followed by 3.6 μL injections every 150 s. In total 11 injections were monitored. ITC data analyses were done using the ORIGIN software (OriginLab). The first injection was excluded from integration according to the manufacturer's instructions. In particular, the affinities of the PAG and PEG peptides are relatively weak, therefore the integrated data were analyzed using the one set of sites fitting model with a fixed stoichiometry (n = 1),20 ensuring that the calculated affinities of the PAG, PRG and PEG peptides can be compared. All measurements were repeated three times and the standard deviations are shown in Table 1 and Fig. 3.

Crystallization, data collection, and structure refinement

Crystallization trials were carried out using in‐house screens21 and the experiments were done in IQ 96‐well sitting drop plates (TTP Labtech) using the mosquito LCP nanodispenser (TTP Labtech). Initial hits were obtained only at 4°C with PSB‐II in the absence of peptides using well solutions containing 100 mM MOPS, pH 7.5, 2.45 M ammonium sulfate and 10% DMSO. The protein was then subjected to an additive screen (Hampton Research) crystallization experiment using the above solution as the reference condition. Several additives improved the crystal shape, and the unliganded PSB‐II crystal (~200 μm in size) obtained with 4% (v/v) pentaerythritol ethoxylate (Hampton Research) was used for further crystallographic studies.

Cocrystallization experiments were performed with (PPG)₃, (P)₉, PPG‐PAG‐PPG, PPG‐PRG‐PPG, and PPG‐PEG‐PPG and at 4°C. The peptides were added to the protein solution at a final concentration of 5 mM [except (P)₉, where 2.0 mM final concentration was used] and incubated for 45 min–60 min on ice before setting up the crystallization experiments, always in the cold room at 4°C. PSB‐II crystallizes in the presence of (PPG)₃, (P)₉, PAG, PRG, and PEG peptides in the same conditions as used for the experiments with the unliganded PSB‐II. These crystals were grown in the presence of the following additives: 10 mM copper(II)chloride dihydrate, 4% hexanediol, 3% d‐galactose, 5% PEG 400, and 10 mM EDTA respectively, of the crystallization additive screen (Hampton Research).

For the X‐ray experiments, crystals were frozen in the cold room by transfer into liquid nitrogen. No cryoprotectant was used for the freezing. From all PSB‐II crystals, data sets with a resolution of approximately 1.9 Å–2.1 Å were obtained, using the in‐house Microstar X8 Proteum Cu‐rotating anode X‐ray generator (Bruker) of the Biocenter Oulu protein X‐ray crystallography core facility. A synchrotron source was also used for data collection and for the PSB‐II_PAG, PSB‐II_PRG, and PSB‐II_PEG crystals higher resolution data sets were obtained at the European radiation synchrotron facility, Grenoble (ESRF), France (Table 2). The home source data were integrated and processed using the Proteum2 software package (Bruker), whereas the synchrotron data were integrated with XDS22 by using the autoPROC data processing pipeline23 at the ESRF, and then scaled with AIMLESS.24 Structure determination, structure refinement, and model validation were done using the Phenix package25 and the model building was done with COOT.18 The PSB‐I structure (PDB code 2V5F) was used as an initial search model in the molecular replacement calculations of unliganded PSB‐II by PHASER.26 The bound peptides were built in their corresponding density once the structure of the protein part had been well‐refined. A summary of the refinement and validation results is presented in Table 2. The coordinates and structure factors of the refined structures have been deposited in the Protein Data Bank with accession codes 6EVL (unliganded PSB‐II), 6EVM [PSB‐II_(P)₉], 6EVN (PSB‐II_PAG), 6EVO (PSB‐II_PRG), and 6EVP (PSB‐II_PEG).