- Email: [email protected]
Conformation of a synthetic hexapeptide substrate of collagen lysyl hydroxylase
ARCHIVES
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 298, No. 1, October, pp. 21-28, 1992
Conformation of a Synthetic Hexapeptide of Collagen Lysyl Hydroxylase Vettai
S. Ananthanarayanan,’
Department
of Biochemistry,
Received November
McMaster
Andre
Saint-Jean,
University,
Substrate
and Ping Jiang
1200 Main Street West, Hamilton,
Ontario L8N 325, Canada
25, 1991, and in revised form April 17, 1992
The hexapeptide Hyp-Gly-Pro-Lys-Gly-Glu was synthesized as a potential substrate for collagen lysyl hydroxylase. Kinetic data on the interaction of this peptide with purified chicken embryo lysyl hydroxylase showed that the hexapeptide is a moderately good substrate having K,, V,,, , and K, JK,,, values comparable to those of synthetic peptide substrates having longer chain lengths. Circular dichroism spectral data suggested a consecutive /3 turn or 310 helical conformation for the peptide in trifluoroethanol. The two-dimensional ‘HTOCSY spectrum of the peptide in dimethylsulfoxide permitted complete assignment of all the protons in the hexapeptide. Through-space connectivities between protons in the peptide molecule were obtained from twodimensional ‘H-NOESY spectral data on the peptide. Using the distances calculated from these data as input constraints, the minimum-energy conformation of the peptide was computed. These calculations and an unconstrained Monte Carlo molecular simulation both led to a folded conformation for the hexapeptide with dihedral angles close to a set of consecutive /3 turns as the lowestenergy conformer. This structure is stabilized further by a salt bridge between the side chains of Lys4 and Glue. Several other conformers energetically close to the minimum-energy conformer exhibited the structural features of the latter except for variations at the N-terminal end and in the side chains. In conjunction with data obtained earlier on lysyl hydroxylase (P. Jiang and V. S. Ananthanarayanan, 1991, J. Biol. Chem. 266,22960-22967) and the functionally related prolyl hydroxylase (P. L. Atreya and V. S. Ananthanarayanan, 1991, J. Biol. Chem. 266,2852-2858), the present results suggest that the folded j3 turn in the respective peptide substrate may be the structural determinant at the catalytic sites of these enzymes. Additional structural features may govern the effective binding of the peptide at the enzymes’ active sites. 0 1992 Academic Press, Inc.
0003-9861/92 $5.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form
One of the early posttranslational modifications that occur during the biosynthesis of collagen is the hydroxylation of selected lysine residues catalyzed by lysyl hydroxylase (EC 1.14.11.4). The resultant hydroxylysine residues serve as sites for carbohydrate attachment and for crosslink formation to stabilize collagen fibrils (1). Although the enzymes involved have different molecular weights and subunit compositions, lysine hydroxylation shares several common features with the hydroxylation of proline residues in collagen. Both processes use molecular oxygen and 2-oxoglutarate as cosubstrates and require ascorbic acid and ferrous iron as cofactors (1). While this implies a common mechanism of action by the two enzymes, the recently available primary structure data on lysyl hydroxylase (2) show that the enzyme has no apparent sequence homology to prolyl hydroxylase. It is, however, possible that the three-dimensional structure at the active sites of the two enzymes may still be similar in order to accomplish similar functions. One way to examine this possibility is to study the structures of the peptide substrates of these enzymes and find out if the substrates bear any conformational similarity. We have recently characterized the conformation of peptide substrates of prolyl hydroxylase and shown that a polyproline II helix followed by a 6 turn is likely to be the minimal structural feature recognized by this enzyme at the site of hydroxylation in the nascent procollagen chain (3). To obtain similar information on lysine hydroxylation, we examined the frequency of occurrence of amino acid residues in the hexapeptide segment Y-GlyX-Hyl-Gly-X’ (where Hyl is 5-hydroxyl-L-lysine)’ at the
1 To whom correspondence should be addressed. a Abbreviations used: Hyl, 5-hydroxyl-L-lysine; Hyp, 4-hydroxy-Lproline; Fmoc, 9-fluorenylmethyloxycarbonyl; TFE, trifluoroethanol; DMSO-ds, dimethyl-&-sulfoxide; rms, root mean square; NOESY, nuclear Overhauser effect spectroscopy; TOCSY, total correlation spectroscopy; TPPI, time-proportional phase increments.
21 Inc. reserved.
22
ANANTHANARAYANAN,
SAINT-JEAN,
AND
JIANG
lysine hydroxylation site in the al(I) collagen sequence (4). Our analysis showed that hydroxyproline (Hyp), Pro, and Glu are, respectively, the most frequent residues in positions Y, X, and X’ of the hexapeptide segment. This led us to synthesize the peptide Hyp-Gly-Pro-Lys-Gly-Glu as the most suitable candidate for a detailed conformational study with respect to lysine hydroxylation. This paper presents kinetic data along with CD, NMR, and conformational energy calculations on the hexapeptide. The structural data show that the preferred conformation of the hexapeptide is a set of two consecutive /3 turns. We suggest that, as with prolyl hydroxylase (3), the /3 turn in the peptide substrate is likely to be the optimal recognition structure at the catalytic site of lysyl hydroxylase. Additional structural determinants might, however, govern the maximal binding of the substrate at the enzyme’s active site. 190
MATERIALS
AND METHODS
WAVELENGTH,
Bovine serum albumin, catalase, dithiothreitol, Trizma base, ferrous sulfate, ascorbic acid, a-ketoglutaric acid, trifluoroacetic acid, ammonium sulfate, and methyl cY-D-glucoside were purchased from Sigma. Solvents and reagent-grade chemicals were obtained from Canlab and Fisher Scientific Company. HypGly-Pro-Lys-Gly-Glu was synthesized using FMoc amino acids on a Milligen Model 9050 continuous-flow peptide synthesizer at the McMaster University Central Facility and desalted on a Sephadex G-10 column. The 11-mer peptide substrate Ala-ArgGly-Ile-Lys-Gly-Ile-Arg-Gly-Phe-Ser-Gly was synthesized for us at the Biotechnology Laboratory in the Institute of Molecular Biology of the University of Oregon at Eugene. The peptides were over 98% pure as judged by C-18 reversed-phase HPLC and quantitative yields of expected amino acids on acid hydrolysis. Protocollagen, used as the standard substrate for lysyl hydroxylase in enzyme assays, was isolated from leg tendons of 17-day-old chick embryos and was labeled with [i4C]lysine according to published procedures (5). Lysyl hydroxylase was isolated from 14-day-old chick embryos and was purified by affinity chromatography on a concanavalin A-Sepharose column according to Turpeenniemi (6). The enzyme frac-
7
-I -6
-6
-4
-2
0
2
14
210
4
6
6
10
12
ml/mg
FIG. 1. Lineweaver-Burk plots for the hydroxylation of the hexapeptide Hyp-Gly-Pro-Lys-Gly-Glu (A) and Ala-Arg-Gly-Ile-Lys-GlyIle-Arg-Gly-Phe-Ser-Gly (+) by lysyl hydroxylase. Temperature 37.0 f 0.5”C, pH 7.8.
nm
FIG. 2. CD spectra of Hyp-Gly-Pro-Lys-Gly-Glu in water (-) and TFE (---). Temperature 22 f 2”C, peptide concentration 0.5 mg/ml.
tions were stored in small aliquots at -20°C in a buffer (0.2 M NaCl, 0.1 M glycine, 10 fiM dithiothreitol, 20 mM Tris-HCl, pH 7.5) containing 60% ethylene glycol and were dialyzed before use against the above buffer without the glycol. Enzyme activity was determined at 37°C either by using the synthetic peptide substrate Ala-Arg-Gly-Ile-Lys-Gly-IleArg-Gly-Phe-Ser-Gly (7) and measuring the “CO, released stoichiometrically from 2-oxo[l-i4C]glutarate, or by using [“Cllysine-labeled protocollagen substrate and estimating the hydroxylysine produced (5). The reaction mixture for enzyme assay and for kinetic study of peptide substrates contained bovine serum albumin, catalase, dithiothreitol, ascorbic acid, ferrous sulfate, and P-oxo[l-‘“Clglutarate in Tris-HCl at pH 7.2, along with the enzyme (50-80 pg of total protein) and varying amount of peptide or protocollagen substrate. The details of measurements are as described in (5). CD spectra were recorded using a Jasco-J6OOA spectropolarimeter equipped with a microprocessor for spectral accumulation and data manipulation. A l.O-mm-pathlength quartz cell was used. Peptide concentrations ranged from 0.2 to 2 mg/ml and were determinedusing A:Fo nm = 13.6 in water and 14.8 in TFE. The spectra were measured at room temperature (22 f 1°C). The pH of the unbuffered aqueous solution was near neutral. The mean residue molar ellipticity [B], expressed in deg cm’ dmol-‘, was used to normalize the observed ellipticity for peptide chain length. Proton NMR spectra were recorded on a Bruker WM-250 for temperature variation studies, on a Bruker AC-200 for TOCSY experiments, and on a Bruker AM-500 for NOESY experiments. All experiments were performed at room temperature (25 f 1°C). The peptide concentration was 6.5 mg ml-’ in DMSO-$. The resonance of residual proton of DMSO-d, was used as an internal reference (a quintuplet set at 2.49 ppm). The TOCSY spectrum provided a straightforward way to proton resonance assignments despite the overlap that made the one-dimensional analysis laborious. The MLEV-17 mixing pulse sequence was used to set up Hartmann-Hahn transfer of magnetization throughout the spin system (8,9). The spectrum was acquired in the phase-sensitive mode using the TPPI procedure (9). The two-dimensional plot is a 1K X 1K matrix presented in positive value mode resulting from 256 increments, zero-filled twice in the fl dimension and multiplied by a sinebell squared function shifted by a factor of r/2. There were 2 dummy scans prior to 32 scans for each of the experiments with a relaxation delay of 2.0 s between each scan: the evolution time had an initial value
CONFORMATION
OF PEPTIDE
SUBSTRATE
OF LYSYL
HYDROXYLASE
23
b Gil Ky
c -
.s
-
1.0
-
1.5
-
2.0
-
2.5
-
3.0
-
3.5
_
4.0
-
4.5
-
5.0
0
lx &3 CA K6 Et? CH HPB
Kt?
0
3
PE
CHHpa Hpy
r PPM
,‘,,‘I,.“,....,...‘,...
9.0
8.5
6.0
7.5
7.0
PPU
FIG. 3. Two-dimensional ‘H-TOCSY spectrum of the hexapeptide HypGly-Pro-Lys-Gly-Glu in DMSO-ds showing different regions of proton resonances (see Materials and Methods for details). Labeling of peaks in (a) and (b) is based on one-letter codes for amino acid residues in the numbered positions of the hexapeptide sequence, except for Hyp which is labeled as Hp. Greek letters denote positions in the peptide backbone and side chains.
of 1 ps with an increment of 250 ~.ls. The length of the spin lock was 100 ms. The sweep width was 2000 Hz. In NOESY runs, a mixing time of 300 ms was found to be optimum for the crosspeak intensity; with mixing times longer than 500 ms, the effect of spin diffusion was not negligible. Two hundred fifty-six tl increments were used with 128 scans (plus 2 dummy scans) per tl value. The initial value of the evolution time was 3 ps, incremented by 250 PCS.The relaxation delay was 3.0 s. A random variation of 20 ms was introduced to minimize contributions from coherent magnetization transfer between scalar coupled nuclei (10). The resulting 256 X 1024 data matrix was zero-filled twice in the jl dimension and multiplied by a sine-bell squared function shifted by an offset of r/2 in both dimensions. The two-dimensional plots based on a 1024 X 1024 matrix are shown in the absolute value mode. Distances were evaluated from the volume integrals of the NOESY crosspeaks using the following relationship between the extent of transfer of magnetization Eab, during the NOE mixing period, between spins a and b separated by a distance r,b (11): Eab = E,(ref) {r.b/‘;u(ref)}“s. Here, E,(ref) and r? (ref) represent magnetization transfer and corresponding interproton distance in a reference spin pair 1: and y. Molecular modeling was performed using a Biograf version 2.2 software (BioDesign, USA) on a Personal Iris computer. Molecular simulation using the Monte Carlo method (12) and energy minimization with and without incorporating the NOESY-derived distance constraints were
done. The generic force field used for molecular simulations was DREIDING (13). Random generation of conformers in the Monte Carlo simulation was achieved by applying random rotations to the rotatable bonds in the peptide molecule. The peptide bond was fixed in the truns orientation (w = 180’). The conformational energy was treated as a sum of bonded and nonbonded (including hydrogen-bonding) interactions. A distance cutoff of 9 A was set as a limit for nonbonded interactions to reduce computational time. The energy was minimized using a conjugate gradient method until convergence to 0.1 kcal/mol/A was achieved or until a cutoff of 25,000 steps was reached. In the constrained Monte Carlo protocol, a range of NMR (NOESY)-derived distances were input in the harmonic force field with a force constant of 25 kcal/mol. The resultant conformational energies were minimized using a dielectric constant of 45 to correspond to DMSO as solvent. The unrestrained Monte Carlo simulations and energy minimization were also carried out with the same dielectric constant.
RESULTS Kinetic
AND
DISCUSSION
Studies
The interaction of the hexapeptide Hyp-Gly-Pro-LysGly-Glu with purified lysyl hydroxylase was studied by
24
ANANTHANARAYANAN, TABLE
Assignment
I
of ‘H Chemical Shifts in Hyp-Gly-Pro-Lys-GlyGlu” from TOCSY Data in DMSO-ds Chemical shift (pm)
Group Hyp’ Hyp’ Hyp’ HYP’ HYP’ Hyp’ Gly’ Gly* Pro3 Pro3 Pro3 Pro3 Lys’ Lys4 Lys4 Lys4 Lys4 Lys4 Lyd Gly5 Gly5 Glu6 Glue Gld Glu6 Glu6
yOH NH olC!H PC& $Hz KH, NH &H c&H /3CH, yCHz dCH2 NH cNH3+ aCH fiCHz -&H, 6CHz &HP NH aCH NH aCH /3CH, yCH, COOH
’ The assignments proton resonances.
7.53 8.05 3.95 1.78, 1.98 3.92 3.48, 3.52 8.22 3.58, 3.87 4.50 1.97, 2.18 1.76, 1.87 3.45 8.68 6.90 4.33 1.90, 2.08 0.90, 1.24 1.46, 1.55 2.73 8.35 3.65, 3.80 8.57 4.20 1.70, 1.95 2.83 8.89, 9.18
were made through homonuclear
shift-correlated
SAINT-JEAN.
AND
JIANG
though the latter is a better substrate (i.e., higher Kc,/ K,) for lysyl hydroxylase than the hexapeptide. It may, however, be noted that the K,,, and V,,, values of much larger synthetic peptide substrates such as (Pro-ProGly),-Ala-Arg-Gly-Met-Lys-Gly-His-Arg-Gly-(ProPro-Gly)4 used by Kivirikko et al. (7) are, respectively, 0.2 mM and 23.0 nmol/h/mg (7) and are not vastly better than those obtained for the hexapeptide used in our study. Moreover, a heptapeptide Leu-Hyp-Gly-Ala-Lys-GlyGlu that has a sequence close to that of our hexapeptide was found to competitively inhibit the hydroxylation of protocollagen by lysyl hydroxylase (14). These observations would imply that the hexapeptide Hyp-Gly-ProLys-Gly-Glu has the essential conformational information required at the catalytic site of lysyl hydroxylase. A detailed conformational analysis of this peptide was therefore undertaken. Conformational
Studies
As in our earlier study (3), we used water and organic solvents such as TFE and DMSO to obtain information about the possible secondary structure that a peptide substrate is capable of assuming in the generally hydrophobic microenvironment of the active site of many enzymes (15). This is particularly relevant for short peptides which display an ensemble of conformations in polar aqueous solvents so that the specific conformation recognized by the enzyme is hard to be gleaned from structural studies in water alone. CD Data
estimating its kinetic parameters. The data obtained are presented in Fig. 1 in the form of a Lineweaver-Burk plot. Also shown in Fig. 1 are the data for the 11-mer peptide Ala-Arg-Gly-Ile-Lys-Gly-Ile-Arg-Gly-PheSer-Gly which is used as a standard peptide substrate for lysyl hydroxylase (7). The relative binding affinities of the peptides for the enzyme may, to a first approximation, be obtained from their K, values, while the V,,, values provide a means to compare the relative conformational suitability of the peptide substrates at the catalytic site of the enzyme (14). The relative efficacies of the peptides as substrates may be judged from their I&/K, values (14). The values for the kinetic parameters of the 11-mer peptide derived from the Lineweaver-Burk plots in Fig. 1 are K, = 0.36 f 0.04, V,,,,, = 24.2 f 0.06, and K,,/K, = 13.5 + 1.37 (the units for K,,,, I’,,,,,, and K,,/K, are mM, nmol/h/mg enzyme, and h-l mM-‘, respectively). Corresponding values for the hexapeptide are K,,, = 0.69 t 0.04, v,,, = 9.3 -+ 0.04, and KJK,,, = 2.69 f 0.10. Using the criteria mentioned above, we observe that the hexapeptide has comparable affinity and conformational features obtained in the standard ll-mer peptide, al-
The CD spectra of the hexapeptide in water and in TFE are shown in Fig. 2. The spectra were independent of peptide concentration over a tenfold range (0.2-2.0 mg/ ml), indicating absence of significant aggregation. The spectrum in TFE shows a negative CD band around 203 nm and a weak shoulder in the region 215-225 nm. The latter is absent in the CD spectrum of the peptide in water where only a relatively strong negative band is seen around 200 nm. The spectrum in TFE is immediately comparable to those reported by Hollosi et al. (16) for Cbz-Gly-Gly-Pro-Gly-0-stearate in methanol. This tetrapeptide and our hexapeptide exhibit the class C CD spectrum of Woody (17) and, as deduced for the tetrapeptide by Hollosi et al. (16), may contain a type III p turn or a 310helix. The main difference between the TFE and water spectra of the hexapeptide is the enhancement of the negative band near 200 nm in water which, together with the absence of indication for a negative band around 220 nm, would suggest the onset of a polyproline II extended structure. The latter, which does not involve any intramolecular hydrogen bonding, is usually populated in aqueous solvents (3, 16) in contrast to nonpolar solvents
CONFORMATION
OF PEPTIDE
a
-
1.5
-
2.0
-
2.5
:
3.0
1
3.5
-
4.0
-
4.3
-
5.0
SUBSTRATE
OF LYSYL
25
HYDROXYLASE
_
1.6
-
2.0
-
2.5
-
3.0
_
3.5
_
4.0
_
4.5
PPH
.
.A
1
7.0 .-,--,
e
(1
7.5 PPli
8.0
I'.'.I,.'.I",'I'.',I".'1"..I...'I 3.5 4.0
3.5
3.0 PPH
2.5
2.0
1.5
1.0
8.5
'I4 I 0.5
’
“.
I, 0.0
1. PPH
7,. 7.5
.
.
.
) 7.0
.
.
FIG. 4. Two-dimensional NOESY spectrum of Hyp-Gly-Pro-Lys-Gly-Glu (b) region containing backbone and amide protons, (c) region containing is as in Fig. 3.
that promote the intramolecularly structures such as the p turn.
hydrogen-bonded
NMR Data To obtain more detailed information on the conformation of the hexapeptide, we carried out one- and twodimensional NMR measurements of the peptide dissolved in DMSO-d6. The resonances in the one-dimensional ‘HNMR spectrum of the peptide were not fully assignable due to overlaps caused by the fi, y, and 6 groups of the side chains having their proton resonances in the narrow region of 1.1 to 2.4 ppm and the cyprotons in the region 3.5 to 4.3. The amide resonances are broad, four of them being compressed between 8.4 and 8.7 ppm. No concen-
in DMSO-d,: (a) region containing backbone and side-chain protons, the amide protons (see Materials and Methods for details). Peak labeling
tration dependence of linewidths in the range l-7 mg/ml was observed, precluding aggregation as a cause for line broadening. Solvent viscosity, quadrapolar coupling between 14N and ‘H and hydrogen bonding are other factors that could cause line broadening (10, 11). Coupling constants between amide protons and the information derived from them about the dihedral angle 4 could not therefore be determined from the one-dimensional spectrum of the peptide. On the other hand, the two-dimensional TOCSY spectrum, shown in Fig. 3, provided a straightforward means of assigning all the necessary resonances through analysis of the crosspeaks. For instance, the NH, CaH, CyH, and OyH of Hyp’ were assigned through their connectivities with the CPH of this residue. Similarly, the connectivities exhibited by Gly5 protons with the spin
ANANTHANARAYANAN,
26
SAINT-JEAN, TABLE
Interproton Amino acid protons
Connectivities
in the Hexapeptide
AND
II
Derived
from
NOESY
NOE”
Hyp’ NH
Lys4 NH Gly5 CaH Gly5 NH
+ + +
Gly’ CaH
Hyp’ CoH
Gly’ Gly* Pro3 Gly’ Glys Glue Pro3 Gly’ Gly*
t t t t t+ t t tt +
Pro3 CrH
Hyp’ CPH Hyp’ C-yH
Hyp’ C6H
Gly* NH
Hyp’ OyH
Lys5 Gly5 Pro3 Pro3 Glys Glua
Gly’ NH
NH CaH CyH C6H CaH NH
’ Qualitative representation based on the magnitudes (3.0-4.0 A); t, weak (3.5-5.0 A).
Data
Amino acid protons
Connection
ColH NH C6H CcvH NH NH CrH NH ColH
JIANG
Pro3 C6H Lys4 NH
+
Lys4 Lys’ Lys4 Gly5
t + ++ +
CaH C6H CcH NH
Gly5 ColH
t
in DMSO-ds
Connection
at 25°C
NOE
Pro3 Pro3 Pro3 Lys4 Gly5 Glu6 Glu6 Lys”
CaH C-rH C6H NH CoH CPH NH C/3H
+ + ++ ++ ++ t +t t
Lys” Pro3 Gly5 Gly5 Glye Glue Glue Glu6 Glue Glu6 Glue Glu6
NH CotH NH CoH NH CaH CaH NH CaH NH ColH CPH
t +t tt t tt + t tt t +t
t t
++ of volume integrals
systems of Lys4 and Glu6 protons were observable in the TOCSY spectrum. Table I lists the assignments made through the homonuclear shift-correlated proton resonances. Proline-containing peptides are known to display cis-tram isomerization around the X-Pro bond (18). Resonances due to the cis conformer were, however, insignificant in the one- and two-dimensional NMR spectra of the hexapeptide. Attempts were made at identifying the amide protons that may be involved in intramolecular hydrogen bonding by measuring the magnitudes of the temperature coefficients of individual NH chemical shifts in the peptide (18). A relatively high value for this coefficient (expressed in ppm/“C) was obtained for Gly’NH (-8.4 X 10p3), precluding its participation in hydrogen bonding, whereas a relatively low value was obtained for the NH proton of Glu6 (-4.0 X 10e3). The values obtained for the NH protons of other NH protons were intermediate (-4.6 to -5.9 X 10-3) and did not permit unambiguous assignments regarding their participation in hydrogen bonding. Molecular modeling studies (see below) on the hexapeptide suggest a salt bridge formation between the side-chain carboxyl group of Glu6 and the side-chain
of crosspeaks; connectivities:
+t+,
strong (2.5-3.5 A); ++, medium
amine protons of Lys3 (temperature coefficient -4.6 X 10m3), thus accounting for the lower value observed for the Glu6 NH protons. Two-dimensional NOESY spectra of the hexapeptide were collected so as to obtain further details of its molecular geometry. Expanded regions of the two-dimensional plot are shown in Fig. 4. Besides the strong intraresidual ones, several interresidue NOE connectivities linking residue i with residues i + 1 to i + 5 are visible in these spectra. This would suggest a relatively folded structure for the hexapeptide rather than an extended conformation. Using the interproton distances in the proline ring as the standard, distances between different parts of the peptide molecule were derived from the volume integrals of NOESY peaks. Table II lists the through-space connectivities between the peptide units in the hexapeptide derived from these data. It may be mentioned here that, similar to the case of the hexapeptide, the presence of a significant proportion of the folded (4 --* 1 p turn) structure in organic solvents was deduced for [Leu5]enkephalin by Picone et al. (19) from NOE connectivities in spite of the relatively high
CONFORMATION
OF PEPTIDE
SUBSTRATE
FIG. 5. Molecular model of the peptide obtained by using the NOESYderived constraints as input in an energy minimization program (see Materials and Methods and text for details). The helical structure of the backbone is shown by the ribbon.
temperature protons. Molecular
coefficients
exhibited
OF LYSYL
HYDROXYLASE
27
FIG. 6. Low-energy conformers of HypGly-Pro-Lys-Gly-Glu generated by using the Monte Carlo technique and subsequent energy minimization (see Materials and Methods and text for details). Shown is a superimposition of 10 of the 54 conformers that lie within a rms deviation of 1 A (for the whole peptide) from the lowest-energy conformer.
by most of the NH
Modeling
A molecular model of the peptide was obtained using the above NOE constraints as input in an energy minimization program. The model, shown in Fig. 5, is a helical structure with two consecutive ,8 turns in the hexapeptide stabilized by 4 + 1 intramolecular hydrogen bonds between Gly’ CO and Gly5 NH and between Pro3 CO and Glu6 NH. Further stabilization of this structure is provided by a 3 + 1 hydrogen bonding (in a y turn) between Pro3 CO and Gly5 NH, by 5 + 1 hydrogen bonding between Gly2 CO and Glu6 NH, and by a salt bridge between the NH of Lys* and the OyH of Glu6. Hydrogen bonding was deduced by using the criterion of N + 0 distance of ~3.5 A (20). We also attempted to arrive at the lowestenergy conformer by means of unconstrained molecular simulation using the Monte Carlo technique (11, 12) at 300°K. This was accomplished by initially starting with a fully extended structure and, subsequently, allowing random variations for all torsional angles in the hexapeptide and minimizing the energy of the conformers generated until convergence was attained (see Materials and Methods). About 27% of the lower-energy conformers obtained had a right-handed helical conformation with a salt bridge between Lys* and Glu6. Further Monte Carlo search with energy minimization led to the lowest-energy conformer which was about 60 kcal/mol less in energy compared to the starting extended structure. Fourteen
percent of the 1000 conformers thus generated had a rootmean-square (rms) deviation of 2 A and 5.4% had a rms deviation of 1 A, for the entire peptide molecule. Figure 6 shows a superimposition of some of the latter set of conformers. The average backbone conformation in these conformers is seen to be in good agreement with the model derived from the NMR data (Fig. 5). Relatively large variations are, however, observed in the N-terminal part of the peptide and in the side chains, indicating confor-
TABLE
III
Conformational Angles of the Peptide Backbone of Hyp-Gly-Pro-Lys-Gly-Glu” From model incorporating NOESY constraints
Amino acid
WP’ Gly’ Pro3 Lys4 Gly5 Glu6 ’ Estimated
From computergenerated model
4 (deg)
ti (ded
-75 118 -53 -83 -116 -116
118 -74 -47 13 -61
average variability
in dihedral
+ (deg) -75 110 -64 -73 -112 -70 angles f5O.
70 -88 -51 19 -58
28
ANANTHANARAYANAN,
mational flexibility (Fig. 6). The models shown in Fig. 5 and 6 are consistent with the conclusion derived from the CD data in the nonpolar solvent. The dihedral angles of the peptide helix (derived for the model in Fig. 5) are listed in Table III. These are close to those of a type IV /3turn followed by a type I 0 turn. The p turn conformation around the Lys residue was also indicated by spectral data on lysine-containing tripeptides (21) and by energy computations on Lys-containing tetrapeptides (22). These peptides were also substrates for lysyl hydroxylase but had very high K,,, values (14). Summing up, kinetic data on the interaction of the hexapeptide Hyp-Gly-Pro-Lys-Gly-Glu with purified lysyl hydroxylase indicate that this peptide is a moderately good substrate for the enzyme so that the study of its conformation may be useful for understanding the substrate conformation recognized by the enzyme at its active site. The structural data obtained for this hexapeptide show a folded, consecutive P-turn conformation. A related study of several peptide substrates of prolyl hydroxylase (3) showed that a combination of two different conformations, namely an extended polyproline II-type helix followed by a 6 turn, is the optimal structure in these peptides. The polyproline II and p turn, are respectively, required at the binding and catalytic sites of prolyl hydroxylase. Such a situation is also likely to prevail in the functionally analogous lysyl hydroxylase so that the fl turn found in the hexapeptide substrate examined in this study may fulfill the requirement at the catalytic site of this enzyme rather than at the binding site. The latter may, however, require additional structural elements (e.g., polyproline II helix) that would help lower the K,,, values in longer peptide substrates such as the 11-mer peptide. [The extremely low K,,, of protocollagen (23) may be accounted for by assuming that the enzyme acts by a processive mechanism proposed for prolyl hydroxylase (24) whereby the substrate’s effective encounter with the enzyme is greatly enhanced.] Structural studies on larger synthetic peptides should prove useful in delineating the criteria for effective binding of the substrate of lysyl hydroxylase.
SAINT-JEAN.
AND JIANG
REFERENCES K. I., and Myllylii, R. (1980) in The Enzymology of Post1. Kivirikko, translational Modification of Proteins (Freedman, R. B., and Hawkins, H. C., Eds.), pp. 53-104, Academic Press, New York.
2. Myllyll,
R., Pihlajaniemi, P., Pajunen, L., Turpeenniemi-Hujanen, T., and Kivirikko, K. I. (1991) J. Bid. Chem. 266, 2805-2810.
V. S. (1991) J. Biol. Chem. 266, 3. Atreya, P., and Ananthanarayanan, 2852-2858. 4. Bornstein, P., and Traub, W. (1979) in The Proteins (Neurath, H.,
5.
6. 7. 8.
and Hill, R. L., Eds.), Vol. 4, pp. 470-483, Academic Press, New York. Kivirikko, K. I., and Myllyll, R. (1982) in Methods in Enzymology (Cunningham, L. W., and Frederiksen, D. W., Eds.), Vol. 82, p. 256, Academic Press, New York. Turpeenniemi, T. M., Puistola, U., Anttinen, H., and Kivirikko, K. I. (1977) Biochim. Biophys. Acta 438, 215-219. Kivirikko, K. I., Shudo, K., Sakakibara, S., and Prockop, D. J. (1972) Biochemistry 11, 122-129. Bax, A., and Davis, D. G. (1985) J. Magn. Resort. 65, 355-360.
9. Marion, D., and Wuthrich, mun. 113,967-974.
K. (1983) Biochem. Biophys. Res. Com-
10. Sanders, J. K. M., and Hunter, B. K. (1988) Modern NMR Spectroscopy. A Guide for Chemists, Oxford University Press, Toronto. 11. Noggle, J. H., and Schirmer, R. E. (1971) The Nuclear Overhauser Effect. Chemical Applications, Academic Press, New York. 12. Finney, J. L. (1982) &o&em. Sot. Trans. 10, 305-307. 13. Mayo, S. L., Olafson, B. D., and Goddard, W. A. III (1990) J. Phys. Chem. 94,8897-8909. V. S. (1991) J. Biol. Chem. 266, 14. Jiang, P., and Ananthanarayanan,
22960-22967. 15. Fersht, A. (1985) Enzyme Structure
and Mechanism,
pp. 85-101,
W. H. Freeman, New York.
16. Hollosi, M., Kawai, M., and Fasman, G. D. (1985) Biopolymers 24, 211-242. 17. Woody, R. W. (1974) in Peptides, Polypeptides and Proteins (Blout, E. R., Bovey, F. A., Goodman, M., and Lotan, N., Eds.), pp. 339349, Wiley, New York.
18. Brahmachari, thanarayanan,
S. K., Rapaka, R. S., Bhatnagar, R. S., and AnanV. S. (1982) Biopolymers 21, 1107-1125.
19. Picone, D., D’Ursi, A., Motta, A., Tancredi, (1990) Eur. J. Biochem. 192, 433-439.
T., and Temussi, P. A.
20. Chou, P. Y., and Fasman, G. D. (1977) J. Mol. Biol. 115,135-175. 21. Zijenah, L. S. (1987) M.Sc. thesis, Memorial University of Newfoundland,
St. John’s, Newfoundland.
23. Takahashi, ACKNOWLEDGMENT This work was supported by a grant from the Canadian Heart Foundation.
K., Nemethy, G., Kang, K., Scheraga, H. A., and Ananthanarayanan, V. S. (1987) Biopolymers 26,1781-1788.
23. Ryhanen, L., and Kivirikko, K. I. (1974) Biochim. Biophys. Acta 343,129-137. 24. de Waal, A., and de Jong, L. (1988) Biochemistry 27, 150-155.
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 298, No. 1, October, pp. 21-28, 1992
Conformation of a Synthetic Hexapeptide of Collagen Lysyl Hydroxylase Vettai
S. Ananthanarayanan,’
Department
of Biochemistry,
Received November
McMaster
Andre
Saint-Jean,
University,
Substrate
and Ping Jiang
1200 Main Street West, Hamilton,
Ontario L8N 325, Canada
25, 1991, and in revised form April 17, 1992
The hexapeptide Hyp-Gly-Pro-Lys-Gly-Glu was synthesized as a potential substrate for collagen lysyl hydroxylase. Kinetic data on the interaction of this peptide with purified chicken embryo lysyl hydroxylase showed that the hexapeptide is a moderately good substrate having K,, V,,, , and K, JK,,, values comparable to those of synthetic peptide substrates having longer chain lengths. Circular dichroism spectral data suggested a consecutive /3 turn or 310 helical conformation for the peptide in trifluoroethanol. The two-dimensional ‘HTOCSY spectrum of the peptide in dimethylsulfoxide permitted complete assignment of all the protons in the hexapeptide. Through-space connectivities between protons in the peptide molecule were obtained from twodimensional ‘H-NOESY spectral data on the peptide. Using the distances calculated from these data as input constraints, the minimum-energy conformation of the peptide was computed. These calculations and an unconstrained Monte Carlo molecular simulation both led to a folded conformation for the hexapeptide with dihedral angles close to a set of consecutive /3 turns as the lowestenergy conformer. This structure is stabilized further by a salt bridge between the side chains of Lys4 and Glue. Several other conformers energetically close to the minimum-energy conformer exhibited the structural features of the latter except for variations at the N-terminal end and in the side chains. In conjunction with data obtained earlier on lysyl hydroxylase (P. Jiang and V. S. Ananthanarayanan, 1991, J. Biol. Chem. 266,22960-22967) and the functionally related prolyl hydroxylase (P. L. Atreya and V. S. Ananthanarayanan, 1991, J. Biol. Chem. 266,2852-2858), the present results suggest that the folded j3 turn in the respective peptide substrate may be the structural determinant at the catalytic sites of these enzymes. Additional structural features may govern the effective binding of the peptide at the enzymes’ active sites. 0 1992 Academic Press, Inc.
0003-9861/92 $5.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form
One of the early posttranslational modifications that occur during the biosynthesis of collagen is the hydroxylation of selected lysine residues catalyzed by lysyl hydroxylase (EC 1.14.11.4). The resultant hydroxylysine residues serve as sites for carbohydrate attachment and for crosslink formation to stabilize collagen fibrils (1). Although the enzymes involved have different molecular weights and subunit compositions, lysine hydroxylation shares several common features with the hydroxylation of proline residues in collagen. Both processes use molecular oxygen and 2-oxoglutarate as cosubstrates and require ascorbic acid and ferrous iron as cofactors (1). While this implies a common mechanism of action by the two enzymes, the recently available primary structure data on lysyl hydroxylase (2) show that the enzyme has no apparent sequence homology to prolyl hydroxylase. It is, however, possible that the three-dimensional structure at the active sites of the two enzymes may still be similar in order to accomplish similar functions. One way to examine this possibility is to study the structures of the peptide substrates of these enzymes and find out if the substrates bear any conformational similarity. We have recently characterized the conformation of peptide substrates of prolyl hydroxylase and shown that a polyproline II helix followed by a 6 turn is likely to be the minimal structural feature recognized by this enzyme at the site of hydroxylation in the nascent procollagen chain (3). To obtain similar information on lysine hydroxylation, we examined the frequency of occurrence of amino acid residues in the hexapeptide segment Y-GlyX-Hyl-Gly-X’ (where Hyl is 5-hydroxyl-L-lysine)’ at the
1 To whom correspondence should be addressed. a Abbreviations used: Hyl, 5-hydroxyl-L-lysine; Hyp, 4-hydroxy-Lproline; Fmoc, 9-fluorenylmethyloxycarbonyl; TFE, trifluoroethanol; DMSO-ds, dimethyl-&-sulfoxide; rms, root mean square; NOESY, nuclear Overhauser effect spectroscopy; TOCSY, total correlation spectroscopy; TPPI, time-proportional phase increments.
21 Inc. reserved.
22
ANANTHANARAYANAN,
SAINT-JEAN,
AND
JIANG
lysine hydroxylation site in the al(I) collagen sequence (4). Our analysis showed that hydroxyproline (Hyp), Pro, and Glu are, respectively, the most frequent residues in positions Y, X, and X’ of the hexapeptide segment. This led us to synthesize the peptide Hyp-Gly-Pro-Lys-Gly-Glu as the most suitable candidate for a detailed conformational study with respect to lysine hydroxylation. This paper presents kinetic data along with CD, NMR, and conformational energy calculations on the hexapeptide. The structural data show that the preferred conformation of the hexapeptide is a set of two consecutive /3 turns. We suggest that, as with prolyl hydroxylase (3), the /3 turn in the peptide substrate is likely to be the optimal recognition structure at the catalytic site of lysyl hydroxylase. Additional structural determinants might, however, govern the maximal binding of the substrate at the enzyme’s active site. 190
MATERIALS
AND METHODS
WAVELENGTH,
Bovine serum albumin, catalase, dithiothreitol, Trizma base, ferrous sulfate, ascorbic acid, a-ketoglutaric acid, trifluoroacetic acid, ammonium sulfate, and methyl cY-D-glucoside were purchased from Sigma. Solvents and reagent-grade chemicals were obtained from Canlab and Fisher Scientific Company. HypGly-Pro-Lys-Gly-Glu was synthesized using FMoc amino acids on a Milligen Model 9050 continuous-flow peptide synthesizer at the McMaster University Central Facility and desalted on a Sephadex G-10 column. The 11-mer peptide substrate Ala-ArgGly-Ile-Lys-Gly-Ile-Arg-Gly-Phe-Ser-Gly was synthesized for us at the Biotechnology Laboratory in the Institute of Molecular Biology of the University of Oregon at Eugene. The peptides were over 98% pure as judged by C-18 reversed-phase HPLC and quantitative yields of expected amino acids on acid hydrolysis. Protocollagen, used as the standard substrate for lysyl hydroxylase in enzyme assays, was isolated from leg tendons of 17-day-old chick embryos and was labeled with [i4C]lysine according to published procedures (5). Lysyl hydroxylase was isolated from 14-day-old chick embryos and was purified by affinity chromatography on a concanavalin A-Sepharose column according to Turpeenniemi (6). The enzyme frac-
7
-I -6
-6
-4
-2
0
2
14
210
4
6
6
10
12
ml/mg
FIG. 1. Lineweaver-Burk plots for the hydroxylation of the hexapeptide Hyp-Gly-Pro-Lys-Gly-Glu (A) and Ala-Arg-Gly-Ile-Lys-GlyIle-Arg-Gly-Phe-Ser-Gly (+) by lysyl hydroxylase. Temperature 37.0 f 0.5”C, pH 7.8.
nm
FIG. 2. CD spectra of Hyp-Gly-Pro-Lys-Gly-Glu in water (-) and TFE (---). Temperature 22 f 2”C, peptide concentration 0.5 mg/ml.
tions were stored in small aliquots at -20°C in a buffer (0.2 M NaCl, 0.1 M glycine, 10 fiM dithiothreitol, 20 mM Tris-HCl, pH 7.5) containing 60% ethylene glycol and were dialyzed before use against the above buffer without the glycol. Enzyme activity was determined at 37°C either by using the synthetic peptide substrate Ala-Arg-Gly-Ile-Lys-Gly-IleArg-Gly-Phe-Ser-Gly (7) and measuring the “CO, released stoichiometrically from 2-oxo[l-i4C]glutarate, or by using [“Cllysine-labeled protocollagen substrate and estimating the hydroxylysine produced (5). The reaction mixture for enzyme assay and for kinetic study of peptide substrates contained bovine serum albumin, catalase, dithiothreitol, ascorbic acid, ferrous sulfate, and P-oxo[l-‘“Clglutarate in Tris-HCl at pH 7.2, along with the enzyme (50-80 pg of total protein) and varying amount of peptide or protocollagen substrate. The details of measurements are as described in (5). CD spectra were recorded using a Jasco-J6OOA spectropolarimeter equipped with a microprocessor for spectral accumulation and data manipulation. A l.O-mm-pathlength quartz cell was used. Peptide concentrations ranged from 0.2 to 2 mg/ml and were determinedusing A:Fo nm = 13.6 in water and 14.8 in TFE. The spectra were measured at room temperature (22 f 1°C). The pH of the unbuffered aqueous solution was near neutral. The mean residue molar ellipticity [B], expressed in deg cm’ dmol-‘, was used to normalize the observed ellipticity for peptide chain length. Proton NMR spectra were recorded on a Bruker WM-250 for temperature variation studies, on a Bruker AC-200 for TOCSY experiments, and on a Bruker AM-500 for NOESY experiments. All experiments were performed at room temperature (25 f 1°C). The peptide concentration was 6.5 mg ml-’ in DMSO-$. The resonance of residual proton of DMSO-d, was used as an internal reference (a quintuplet set at 2.49 ppm). The TOCSY spectrum provided a straightforward way to proton resonance assignments despite the overlap that made the one-dimensional analysis laborious. The MLEV-17 mixing pulse sequence was used to set up Hartmann-Hahn transfer of magnetization throughout the spin system (8,9). The spectrum was acquired in the phase-sensitive mode using the TPPI procedure (9). The two-dimensional plot is a 1K X 1K matrix presented in positive value mode resulting from 256 increments, zero-filled twice in the fl dimension and multiplied by a sinebell squared function shifted by a factor of r/2. There were 2 dummy scans prior to 32 scans for each of the experiments with a relaxation delay of 2.0 s between each scan: the evolution time had an initial value
CONFORMATION
OF PEPTIDE
SUBSTRATE
OF LYSYL
HYDROXYLASE
23
b Gil Ky
c -
.s
-
1.0
-
1.5
-
2.0
-
2.5
-
3.0
-
3.5
_
4.0
-
4.5
-
5.0
0
lx &3 CA K6 Et? CH HPB
Kt?
0
3
PE
CHHpa Hpy
r PPM
,‘,,‘I,.“,....,...‘,...
9.0
8.5
6.0
7.5
7.0
PPU
FIG. 3. Two-dimensional ‘H-TOCSY spectrum of the hexapeptide HypGly-Pro-Lys-Gly-Glu in DMSO-ds showing different regions of proton resonances (see Materials and Methods for details). Labeling of peaks in (a) and (b) is based on one-letter codes for amino acid residues in the numbered positions of the hexapeptide sequence, except for Hyp which is labeled as Hp. Greek letters denote positions in the peptide backbone and side chains.
of 1 ps with an increment of 250 ~.ls. The length of the spin lock was 100 ms. The sweep width was 2000 Hz. In NOESY runs, a mixing time of 300 ms was found to be optimum for the crosspeak intensity; with mixing times longer than 500 ms, the effect of spin diffusion was not negligible. Two hundred fifty-six tl increments were used with 128 scans (plus 2 dummy scans) per tl value. The initial value of the evolution time was 3 ps, incremented by 250 PCS.The relaxation delay was 3.0 s. A random variation of 20 ms was introduced to minimize contributions from coherent magnetization transfer between scalar coupled nuclei (10). The resulting 256 X 1024 data matrix was zero-filled twice in the jl dimension and multiplied by a sine-bell squared function shifted by an offset of r/2 in both dimensions. The two-dimensional plots based on a 1024 X 1024 matrix are shown in the absolute value mode. Distances were evaluated from the volume integrals of the NOESY crosspeaks using the following relationship between the extent of transfer of magnetization Eab, during the NOE mixing period, between spins a and b separated by a distance r,b (11): Eab = E,(ref) {r.b/‘;u(ref)}“s. Here, E,(ref) and r? (ref) represent magnetization transfer and corresponding interproton distance in a reference spin pair 1: and y. Molecular modeling was performed using a Biograf version 2.2 software (BioDesign, USA) on a Personal Iris computer. Molecular simulation using the Monte Carlo method (12) and energy minimization with and without incorporating the NOESY-derived distance constraints were
done. The generic force field used for molecular simulations was DREIDING (13). Random generation of conformers in the Monte Carlo simulation was achieved by applying random rotations to the rotatable bonds in the peptide molecule. The peptide bond was fixed in the truns orientation (w = 180’). The conformational energy was treated as a sum of bonded and nonbonded (including hydrogen-bonding) interactions. A distance cutoff of 9 A was set as a limit for nonbonded interactions to reduce computational time. The energy was minimized using a conjugate gradient method until convergence to 0.1 kcal/mol/A was achieved or until a cutoff of 25,000 steps was reached. In the constrained Monte Carlo protocol, a range of NMR (NOESY)-derived distances were input in the harmonic force field with a force constant of 25 kcal/mol. The resultant conformational energies were minimized using a dielectric constant of 45 to correspond to DMSO as solvent. The unrestrained Monte Carlo simulations and energy minimization were also carried out with the same dielectric constant.
RESULTS Kinetic
AND
DISCUSSION
Studies
The interaction of the hexapeptide Hyp-Gly-Pro-LysGly-Glu with purified lysyl hydroxylase was studied by
24
ANANTHANARAYANAN, TABLE
Assignment
I
of ‘H Chemical Shifts in Hyp-Gly-Pro-Lys-GlyGlu” from TOCSY Data in DMSO-ds Chemical shift (pm)
Group Hyp’ Hyp’ Hyp’ HYP’ HYP’ Hyp’ Gly’ Gly* Pro3 Pro3 Pro3 Pro3 Lys’ Lys4 Lys4 Lys4 Lys4 Lys4 Lyd Gly5 Gly5 Glu6 Glue Gld Glu6 Glu6
yOH NH olC!H PC& $Hz KH, NH &H c&H /3CH, yCHz dCH2 NH cNH3+ aCH fiCHz -&H, 6CHz &HP NH aCH NH aCH /3CH, yCH, COOH
’ The assignments proton resonances.
7.53 8.05 3.95 1.78, 1.98 3.92 3.48, 3.52 8.22 3.58, 3.87 4.50 1.97, 2.18 1.76, 1.87 3.45 8.68 6.90 4.33 1.90, 2.08 0.90, 1.24 1.46, 1.55 2.73 8.35 3.65, 3.80 8.57 4.20 1.70, 1.95 2.83 8.89, 9.18
were made through homonuclear
shift-correlated
SAINT-JEAN.
AND
JIANG
though the latter is a better substrate (i.e., higher Kc,/ K,) for lysyl hydroxylase than the hexapeptide. It may, however, be noted that the K,,, and V,,, values of much larger synthetic peptide substrates such as (Pro-ProGly),-Ala-Arg-Gly-Met-Lys-Gly-His-Arg-Gly-(ProPro-Gly)4 used by Kivirikko et al. (7) are, respectively, 0.2 mM and 23.0 nmol/h/mg (7) and are not vastly better than those obtained for the hexapeptide used in our study. Moreover, a heptapeptide Leu-Hyp-Gly-Ala-Lys-GlyGlu that has a sequence close to that of our hexapeptide was found to competitively inhibit the hydroxylation of protocollagen by lysyl hydroxylase (14). These observations would imply that the hexapeptide Hyp-Gly-ProLys-Gly-Glu has the essential conformational information required at the catalytic site of lysyl hydroxylase. A detailed conformational analysis of this peptide was therefore undertaken. Conformational
Studies
As in our earlier study (3), we used water and organic solvents such as TFE and DMSO to obtain information about the possible secondary structure that a peptide substrate is capable of assuming in the generally hydrophobic microenvironment of the active site of many enzymes (15). This is particularly relevant for short peptides which display an ensemble of conformations in polar aqueous solvents so that the specific conformation recognized by the enzyme is hard to be gleaned from structural studies in water alone. CD Data
estimating its kinetic parameters. The data obtained are presented in Fig. 1 in the form of a Lineweaver-Burk plot. Also shown in Fig. 1 are the data for the 11-mer peptide Ala-Arg-Gly-Ile-Lys-Gly-Ile-Arg-Gly-PheSer-Gly which is used as a standard peptide substrate for lysyl hydroxylase (7). The relative binding affinities of the peptides for the enzyme may, to a first approximation, be obtained from their K, values, while the V,,, values provide a means to compare the relative conformational suitability of the peptide substrates at the catalytic site of the enzyme (14). The relative efficacies of the peptides as substrates may be judged from their I&/K, values (14). The values for the kinetic parameters of the 11-mer peptide derived from the Lineweaver-Burk plots in Fig. 1 are K, = 0.36 f 0.04, V,,,,, = 24.2 f 0.06, and K,,/K, = 13.5 + 1.37 (the units for K,,,, I’,,,,,, and K,,/K, are mM, nmol/h/mg enzyme, and h-l mM-‘, respectively). Corresponding values for the hexapeptide are K,,, = 0.69 t 0.04, v,,, = 9.3 -+ 0.04, and KJK,,, = 2.69 f 0.10. Using the criteria mentioned above, we observe that the hexapeptide has comparable affinity and conformational features obtained in the standard ll-mer peptide, al-
The CD spectra of the hexapeptide in water and in TFE are shown in Fig. 2. The spectra were independent of peptide concentration over a tenfold range (0.2-2.0 mg/ ml), indicating absence of significant aggregation. The spectrum in TFE shows a negative CD band around 203 nm and a weak shoulder in the region 215-225 nm. The latter is absent in the CD spectrum of the peptide in water where only a relatively strong negative band is seen around 200 nm. The spectrum in TFE is immediately comparable to those reported by Hollosi et al. (16) for Cbz-Gly-Gly-Pro-Gly-0-stearate in methanol. This tetrapeptide and our hexapeptide exhibit the class C CD spectrum of Woody (17) and, as deduced for the tetrapeptide by Hollosi et al. (16), may contain a type III p turn or a 310helix. The main difference between the TFE and water spectra of the hexapeptide is the enhancement of the negative band near 200 nm in water which, together with the absence of indication for a negative band around 220 nm, would suggest the onset of a polyproline II extended structure. The latter, which does not involve any intramolecular hydrogen bonding, is usually populated in aqueous solvents (3, 16) in contrast to nonpolar solvents
CONFORMATION
OF PEPTIDE
a
-
1.5
-
2.0
-
2.5
:
3.0
1
3.5
-
4.0
-
4.3
-
5.0
SUBSTRATE
OF LYSYL
25
HYDROXYLASE
_
1.6
-
2.0
-
2.5
-
3.0
_
3.5
_
4.0
_
4.5
PPH
.
.A
1
7.0 .-,--,
e
(1
7.5 PPli
8.0
I'.'.I,.'.I",'I'.',I".'1"..I...'I 3.5 4.0
3.5
3.0 PPH
2.5
2.0
1.5
1.0
8.5
'I4 I 0.5
’
“.
I, 0.0
1. PPH
7,. 7.5
.
.
.
) 7.0
.
.
FIG. 4. Two-dimensional NOESY spectrum of Hyp-Gly-Pro-Lys-Gly-Glu (b) region containing backbone and amide protons, (c) region containing is as in Fig. 3.
that promote the intramolecularly structures such as the p turn.
hydrogen-bonded
NMR Data To obtain more detailed information on the conformation of the hexapeptide, we carried out one- and twodimensional NMR measurements of the peptide dissolved in DMSO-d6. The resonances in the one-dimensional ‘HNMR spectrum of the peptide were not fully assignable due to overlaps caused by the fi, y, and 6 groups of the side chains having their proton resonances in the narrow region of 1.1 to 2.4 ppm and the cyprotons in the region 3.5 to 4.3. The amide resonances are broad, four of them being compressed between 8.4 and 8.7 ppm. No concen-
in DMSO-d,: (a) region containing backbone and side-chain protons, the amide protons (see Materials and Methods for details). Peak labeling
tration dependence of linewidths in the range l-7 mg/ml was observed, precluding aggregation as a cause for line broadening. Solvent viscosity, quadrapolar coupling between 14N and ‘H and hydrogen bonding are other factors that could cause line broadening (10, 11). Coupling constants between amide protons and the information derived from them about the dihedral angle 4 could not therefore be determined from the one-dimensional spectrum of the peptide. On the other hand, the two-dimensional TOCSY spectrum, shown in Fig. 3, provided a straightforward means of assigning all the necessary resonances through analysis of the crosspeaks. For instance, the NH, CaH, CyH, and OyH of Hyp’ were assigned through their connectivities with the CPH of this residue. Similarly, the connectivities exhibited by Gly5 protons with the spin
ANANTHANARAYANAN,
26
SAINT-JEAN, TABLE
Interproton Amino acid protons
Connectivities
in the Hexapeptide
AND
II
Derived
from
NOESY
NOE”
Hyp’ NH
Lys4 NH Gly5 CaH Gly5 NH
+ + +
Gly’ CaH
Hyp’ CoH
Gly’ Gly* Pro3 Gly’ Glys Glue Pro3 Gly’ Gly*
t t t t t+ t t tt +
Pro3 CrH
Hyp’ CPH Hyp’ C-yH
Hyp’ C6H
Gly* NH
Hyp’ OyH
Lys5 Gly5 Pro3 Pro3 Glys Glua
Gly’ NH
NH CaH CyH C6H CaH NH
’ Qualitative representation based on the magnitudes (3.0-4.0 A); t, weak (3.5-5.0 A).
Data
Amino acid protons
Connection
ColH NH C6H CcvH NH NH CrH NH ColH
JIANG
Pro3 C6H Lys4 NH
+
Lys4 Lys’ Lys4 Gly5
t + ++ +
CaH C6H CcH NH
Gly5 ColH
t
in DMSO-ds
Connection
at 25°C
NOE
Pro3 Pro3 Pro3 Lys4 Gly5 Glu6 Glu6 Lys”
CaH C-rH C6H NH CoH CPH NH C/3H
+ + ++ ++ ++ t +t t
Lys” Pro3 Gly5 Gly5 Glye Glue Glue Glu6 Glue Glu6 Glue Glu6
NH CotH NH CoH NH CaH CaH NH CaH NH ColH CPH
t +t tt t tt + t tt t +t
t t
++ of volume integrals
systems of Lys4 and Glu6 protons were observable in the TOCSY spectrum. Table I lists the assignments made through the homonuclear shift-correlated proton resonances. Proline-containing peptides are known to display cis-tram isomerization around the X-Pro bond (18). Resonances due to the cis conformer were, however, insignificant in the one- and two-dimensional NMR spectra of the hexapeptide. Attempts were made at identifying the amide protons that may be involved in intramolecular hydrogen bonding by measuring the magnitudes of the temperature coefficients of individual NH chemical shifts in the peptide (18). A relatively high value for this coefficient (expressed in ppm/“C) was obtained for Gly’NH (-8.4 X 10p3), precluding its participation in hydrogen bonding, whereas a relatively low value was obtained for the NH proton of Glu6 (-4.0 X 10e3). The values obtained for the NH protons of other NH protons were intermediate (-4.6 to -5.9 X 10-3) and did not permit unambiguous assignments regarding their participation in hydrogen bonding. Molecular modeling studies (see below) on the hexapeptide suggest a salt bridge formation between the side-chain carboxyl group of Glu6 and the side-chain
of crosspeaks; connectivities:
+t+,
strong (2.5-3.5 A); ++, medium
amine protons of Lys3 (temperature coefficient -4.6 X 10m3), thus accounting for the lower value observed for the Glu6 NH protons. Two-dimensional NOESY spectra of the hexapeptide were collected so as to obtain further details of its molecular geometry. Expanded regions of the two-dimensional plot are shown in Fig. 4. Besides the strong intraresidual ones, several interresidue NOE connectivities linking residue i with residues i + 1 to i + 5 are visible in these spectra. This would suggest a relatively folded structure for the hexapeptide rather than an extended conformation. Using the interproton distances in the proline ring as the standard, distances between different parts of the peptide molecule were derived from the volume integrals of NOESY peaks. Table II lists the through-space connectivities between the peptide units in the hexapeptide derived from these data. It may be mentioned here that, similar to the case of the hexapeptide, the presence of a significant proportion of the folded (4 --* 1 p turn) structure in organic solvents was deduced for [Leu5]enkephalin by Picone et al. (19) from NOE connectivities in spite of the relatively high
CONFORMATION
OF PEPTIDE
SUBSTRATE
FIG. 5. Molecular model of the peptide obtained by using the NOESYderived constraints as input in an energy minimization program (see Materials and Methods and text for details). The helical structure of the backbone is shown by the ribbon.
temperature protons. Molecular
coefficients
exhibited
OF LYSYL
HYDROXYLASE
27
FIG. 6. Low-energy conformers of HypGly-Pro-Lys-Gly-Glu generated by using the Monte Carlo technique and subsequent energy minimization (see Materials and Methods and text for details). Shown is a superimposition of 10 of the 54 conformers that lie within a rms deviation of 1 A (for the whole peptide) from the lowest-energy conformer.
by most of the NH
Modeling
A molecular model of the peptide was obtained using the above NOE constraints as input in an energy minimization program. The model, shown in Fig. 5, is a helical structure with two consecutive ,8 turns in the hexapeptide stabilized by 4 + 1 intramolecular hydrogen bonds between Gly’ CO and Gly5 NH and between Pro3 CO and Glu6 NH. Further stabilization of this structure is provided by a 3 + 1 hydrogen bonding (in a y turn) between Pro3 CO and Gly5 NH, by 5 + 1 hydrogen bonding between Gly2 CO and Glu6 NH, and by a salt bridge between the NH of Lys* and the OyH of Glu6. Hydrogen bonding was deduced by using the criterion of N + 0 distance of ~3.5 A (20). We also attempted to arrive at the lowestenergy conformer by means of unconstrained molecular simulation using the Monte Carlo technique (11, 12) at 300°K. This was accomplished by initially starting with a fully extended structure and, subsequently, allowing random variations for all torsional angles in the hexapeptide and minimizing the energy of the conformers generated until convergence was attained (see Materials and Methods). About 27% of the lower-energy conformers obtained had a right-handed helical conformation with a salt bridge between Lys* and Glu6. Further Monte Carlo search with energy minimization led to the lowest-energy conformer which was about 60 kcal/mol less in energy compared to the starting extended structure. Fourteen
percent of the 1000 conformers thus generated had a rootmean-square (rms) deviation of 2 A and 5.4% had a rms deviation of 1 A, for the entire peptide molecule. Figure 6 shows a superimposition of some of the latter set of conformers. The average backbone conformation in these conformers is seen to be in good agreement with the model derived from the NMR data (Fig. 5). Relatively large variations are, however, observed in the N-terminal part of the peptide and in the side chains, indicating confor-
TABLE
III
Conformational Angles of the Peptide Backbone of Hyp-Gly-Pro-Lys-Gly-Glu” From model incorporating NOESY constraints
Amino acid
WP’ Gly’ Pro3 Lys4 Gly5 Glu6 ’ Estimated
From computergenerated model
4 (deg)
ti (ded
-75 118 -53 -83 -116 -116
118 -74 -47 13 -61
average variability
in dihedral
+ (deg) -75 110 -64 -73 -112 -70 angles f5O.
70 -88 -51 19 -58
28
ANANTHANARAYANAN,
mational flexibility (Fig. 6). The models shown in Fig. 5 and 6 are consistent with the conclusion derived from the CD data in the nonpolar solvent. The dihedral angles of the peptide helix (derived for the model in Fig. 5) are listed in Table III. These are close to those of a type IV /3turn followed by a type I 0 turn. The p turn conformation around the Lys residue was also indicated by spectral data on lysine-containing tripeptides (21) and by energy computations on Lys-containing tetrapeptides (22). These peptides were also substrates for lysyl hydroxylase but had very high K,,, values (14). Summing up, kinetic data on the interaction of the hexapeptide Hyp-Gly-Pro-Lys-Gly-Glu with purified lysyl hydroxylase indicate that this peptide is a moderately good substrate for the enzyme so that the study of its conformation may be useful for understanding the substrate conformation recognized by the enzyme at its active site. The structural data obtained for this hexapeptide show a folded, consecutive P-turn conformation. A related study of several peptide substrates of prolyl hydroxylase (3) showed that a combination of two different conformations, namely an extended polyproline II-type helix followed by a 6 turn, is the optimal structure in these peptides. The polyproline II and p turn, are respectively, required at the binding and catalytic sites of prolyl hydroxylase. Such a situation is also likely to prevail in the functionally analogous lysyl hydroxylase so that the fl turn found in the hexapeptide substrate examined in this study may fulfill the requirement at the catalytic site of this enzyme rather than at the binding site. The latter may, however, require additional structural elements (e.g., polyproline II helix) that would help lower the K,,, values in longer peptide substrates such as the 11-mer peptide. [The extremely low K,,, of protocollagen (23) may be accounted for by assuming that the enzyme acts by a processive mechanism proposed for prolyl hydroxylase (24) whereby the substrate’s effective encounter with the enzyme is greatly enhanced.] Structural studies on larger synthetic peptides should prove useful in delineating the criteria for effective binding of the substrate of lysyl hydroxylase.
SAINT-JEAN.
AND JIANG
REFERENCES K. I., and Myllylii, R. (1980) in The Enzymology of Post1. Kivirikko, translational Modification of Proteins (Freedman, R. B., and Hawkins, H. C., Eds.), pp. 53-104, Academic Press, New York.
2. Myllyll,
R., Pihlajaniemi, P., Pajunen, L., Turpeenniemi-Hujanen, T., and Kivirikko, K. I. (1991) J. Bid. Chem. 266, 2805-2810.
V. S. (1991) J. Biol. Chem. 266, 3. Atreya, P., and Ananthanarayanan, 2852-2858. 4. Bornstein, P., and Traub, W. (1979) in The Proteins (Neurath, H.,
5.
6. 7. 8.
and Hill, R. L., Eds.), Vol. 4, pp. 470-483, Academic Press, New York. Kivirikko, K. I., and Myllyll, R. (1982) in Methods in Enzymology (Cunningham, L. W., and Frederiksen, D. W., Eds.), Vol. 82, p. 256, Academic Press, New York. Turpeenniemi, T. M., Puistola, U., Anttinen, H., and Kivirikko, K. I. (1977) Biochim. Biophys. Acta 438, 215-219. Kivirikko, K. I., Shudo, K., Sakakibara, S., and Prockop, D. J. (1972) Biochemistry 11, 122-129. Bax, A., and Davis, D. G. (1985) J. Magn. Resort. 65, 355-360.
9. Marion, D., and Wuthrich, mun. 113,967-974.
K. (1983) Biochem. Biophys. Res. Com-
10. Sanders, J. K. M., and Hunter, B. K. (1988) Modern NMR Spectroscopy. A Guide for Chemists, Oxford University Press, Toronto. 11. Noggle, J. H., and Schirmer, R. E. (1971) The Nuclear Overhauser Effect. Chemical Applications, Academic Press, New York. 12. Finney, J. L. (1982) &o&em. Sot. Trans. 10, 305-307. 13. Mayo, S. L., Olafson, B. D., and Goddard, W. A. III (1990) J. Phys. Chem. 94,8897-8909. V. S. (1991) J. Biol. Chem. 266, 14. Jiang, P., and Ananthanarayanan,
22960-22967. 15. Fersht, A. (1985) Enzyme Structure
and Mechanism,
pp. 85-101,
W. H. Freeman, New York.
16. Hollosi, M., Kawai, M., and Fasman, G. D. (1985) Biopolymers 24, 211-242. 17. Woody, R. W. (1974) in Peptides, Polypeptides and Proteins (Blout, E. R., Bovey, F. A., Goodman, M., and Lotan, N., Eds.), pp. 339349, Wiley, New York.
18. Brahmachari, thanarayanan,
S. K., Rapaka, R. S., Bhatnagar, R. S., and AnanV. S. (1982) Biopolymers 21, 1107-1125.
19. Picone, D., D’Ursi, A., Motta, A., Tancredi, (1990) Eur. J. Biochem. 192, 433-439.
T., and Temussi, P. A.
20. Chou, P. Y., and Fasman, G. D. (1977) J. Mol. Biol. 115,135-175. 21. Zijenah, L. S. (1987) M.Sc. thesis, Memorial University of Newfoundland,
St. John’s, Newfoundland.
23. Takahashi, ACKNOWLEDGMENT This work was supported by a grant from the Canadian Heart Foundation.
K., Nemethy, G., Kang, K., Scheraga, H. A., and Ananthanarayanan, V. S. (1987) Biopolymers 26,1781-1788.
23. Ryhanen, L., and Kivirikko, K. I. (1974) Biochim. Biophys. Acta 343,129-137. 24. de Waal, A., and de Jong, L. (1988) Biochemistry 27, 150-155.