Conformational Study of Two Linear Hexapeptides by Two-Dimensional
NMR and Computer-Simulated Modeling: Implication for Peptide
Cyclization in Solution
Aih-Jing Chiou,* Geok-Toh Ong,† Kung-Tsung Wang,† Shyh-Horng Chiou,*
,† and
Shih-Hsiung Wu*
,†
,1*Institute of Biochemical Sciences, National Taiwan University, Taipei, Taiwan; and †Institute of Biological Chemistry, P.O. Box 23-106, Academia Sinica, Taipei, Taiwan
Received January 3, 1996
Two linear peptides, D-leucyl-L-prolyl-L-isoleucyl-L-valyl-L-alanyl-b-alanine (I) and D-leucyl-L-prolyl-L -isoleucyl-L-valyl-N-methyl-L-alanyl-b-alanine (II), whose sequences were designed from protodestruxin and desmethyldestruxin B by replacingD-leucic acid withD-leucine, two cyclic hexadepsipeptides with insecticidal and immunodepressant activities, have been found to be cyclized in unusually high yields (>85%). In order to gain insight into the conformation and the relative flexibility of different constituent residues in these linear peptides, we have applied various techniques of 2D-NMR spectroscopy coupled with dynamic simulated an-nealing by computer modeling to establish the solution conformations of these two linear peptides. Based on the derived structures, it is found that the distances betweenN- andC-terminal residues of both peptides are short enough to facilitate the cyclization, thus collaborating the observation of favorable cyclization yields for both linear peptides. © 1996 Academic Press, Inc.
Cyclic peptides with ring structure from natural products are well known to play important roles
in their biological activities. A lot of microbial peptides such as peptide antibiotics possess
struc-ture-constrained ring structure [1]. Since the synthesis of gramicidin S was accomplished in 1950s,
many cyclization methods including both the cyclization and cleavage of peptides on the
solid-phase resins have been developed [2,3]. For a successful chemical synthesis of biologically-active
cyclic peptides, the key step is usually the cyclization reaction. In most cyclization reactions of
synthetic peptides, lower yields are obtained [4]. The yield of cyclization for linear peptides
depends on several factors such as sequence and length of linear peptides, concentration of linear
peptides, reagents and conditions of cyclization [2,5], etc.
Destruxins are cyclic hexadepsipeptides first isolated from the culture filtrate of Metarrhizium
anisopliae, an insect-pathogenic fungus [6]. They are composed of 5 amino acids:
L-Pro,
L-Ile,
N
-methyl-
L-Val,
N-methyl-
L-Ala and
b
-alanine plus a
D-
a
-hydroxy acid which differs in five
congeners, destruxins A, B, C and D. and desmethyldestruxin B [6–8]. Destruxin B and
desmeth-yldestruxin B are proposed to be biosynthesized from protodestruxin by
N-methylation at two sites
and subsequent demethylation at one site on the cyclic ring [9].
Our recent interests in these conformation-constrained cyclodepsipeptides lie in their potential
application as useful immunodepressants similar to cyclosporins [10] plus their suppressive effect
on hepatitis B virus surface antigen (HBsAg) production in human hepatoma cells [11]. In an effort
Abbreviations: NMR, nuclear magnetic resonance; BOP, benzotriazolyloxycarbonyl N-oxytridimethylamino-phosphonium hexafluorophosphate; DQF-COSY, double-quantum filtered correlation spectroscopy; HOHAHA, homo-nuclear Hartmann-Hahn spectroscopy; XH-COSY, heterohomo-nuclear shift-correlated spectroscopy; NOESY, homo-nuclear Over-hauser and exchange spectroscopy; ROESY, rotating frame nuclear OverOver-hauser and exchange spectroscopy; NOE, nuclear Overhauser effect; RMSD, root mean square deviation; DMF, dimethylformamide; DMSO, dimethylsulfoxide; HMP resin, p-hydroxymethylphenoxymethylpolystyrene resin; Fmoc-amino acid, N-fluorenylmethoxycarbonyl-amino acid.
1Corresponding address: Institute of Biological Chemistry, Academia, P.O. Box 23-106, Taipei, Taiwan. Fax: (886)-2-3635038. E-mail: shwu@gate.sinica.edu.TW.
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS219, 572–579 (1996)
ARTICLE NO. 0275
572 0006-291X/96 $18.00
to prepare useful analogues of these biologically active cyclic peptides, we have synthesized two
linear peptides,
D-leucyl-
L-prolyl-
L-isoleucyl-
L-valyl-
L-alanyl-
b
-alanine (I) and
D-leucyl-
L-prolyl-L
-isoleucyl-
L-valyl-
N-methyl-
L-alanyl-
b
-alanine (II), whose sequences were designed from
pro-todestruxin and desmethyldestruxin B by replacing
D-leucic acid with
D-leucine. We have analyzed
these flexible short polypeptides using 2D-NMR spectroscopy coupled with dynamic simulated
annealing to derive the solution conformations for these two linear peptides. Based on the derived
structures, it is found that the separation between
N- and
C-terminal residues of both linear peptides
are in a close distance favorable for the cyclization, in agreement with a high yield obtained in the
final cyclization reaction for these synthetic analogues of destruxins.
MATERIALS AND METHODS
Preparation of linear peptides. The peptides used for NMR determination in this work was synthesized by solid-phase peptide synthesis in HMP resin using an ABI automatic peptide synthesizer (Model 431A) essentially according to the previous report [12]. The Fmoc-amino acids were introduced using the manufacturer’s prepacked cartridges (1 mmol each) with a stepwise FastMoc protocol. The synthesized peptide was further purified by HPLC on a Vydac C18semi-preparation column. The purified peptides were characterized by mass spectroscopy (JEOL JMS-HX 110 Mass Spectrometer) and amino acid analysis (Beckman 6300 Amino Acid Analyzer).
Cyclization of linear peptides. To a solution of the linear peptide (50 mg; 0.085 mmol) in 120 ml of DMF at 0°C was added BOP (442 mg; 0.1 mmol) and solid sodium bicarbonate (30 mg; 0.357 mmol) [13,14]. The mixture was stirred at 0°C for 1 h and then at room temperature for further reaction. The progress of reaction was monitored by HPLC. The reaction was stopped by quick freezing and lyophilization and the cyclic peptide was isolated by HPLC. The purified cyclic peptide was also characterized by mass spectroscopy and amino acid analysis.
NMR spectroscopy. The NMR samples were made of ca. 15 mg in 0.5 ml DMSO-d6. All NMR experiments were performed on either Bruker AM-400, or AXL-400 or ARX-500 MHz spectrometer. Chemical shifts are referred to the solvent signal at 2.49 ppm. The assignments of the proton spectra of the two hexapeptides were made via the use of 1D proton and carbon spectra, DQF-COSY [15,16], HOHAHA [17], XH-COSY [18], NOESY [19] and ROESY [20] 2-D spectra. Spectra were recorded in the phase-sensitive mode with the time-proportion phase increments (TPPI) [19] method except XH-COSY, which was recorded in the magnitude mode. Typical two-dimensional spectra were recorded at 306 K with 512 t1increments, 2048 complex points (256 t1, 1024 for ROESY and NOESY, 80 t1, 2048 for XH-COSY). For the HOHAHA experiments, mixing time 80–200 ms was used. A mixing time of 150–500 ms was used for the NOESY and ROESY experiments. The ROESY experiment was performed by using a small flip angle (17°) in the pulsed spin-locked experiment to suppress chemical exchange [20]. After zero filling to a 1K × 1K or 2K × 2K matrix,p/2-shifted squared sine-bell functions were applied in both dimensions prior to Fourier transformation. The ROESY and NOESY spectra were processed by Felix program (version 2.05, Hare Research, Inc.). The temperature coefficients were determined over a temperature range of 27 to 67 °C at 5°C interval.
Computer-simulated modeling. The interproton distance information was derived by integrating the volume of the ROESY spectra’s cross peak similar to that described in the previous report [12]. The ROESY cross peak intensities of linear peptides I and II were integrated and used for the computation of the tertiary structure. The dihedral angles of Ile3 and Val4residues in both peptides were constrained within the limits [−180°, −80°] based on the coupling constants3J
HNa. All calculations were carried out on a Silicon Graphics, Iris 4D/35 and Iris Indigo Elan 4000 workstation. Structures were examined by the macromodel computer-modeling system “Insight II” from Biosym (version 2.2). Structures were calculated in two steps using NMRchitech program: DGII (distance geometry) and SA (simulated annealing) methodology. Both were executed with ROE constraints and dihedral angles. DGII includes smoothing, embedding and optimizing steps. Twenty structures obtained from DGII calculation were then used as starting points for the second-stage modeling based on the simulated annealing. From 20 random structures, 10 structures were selected based on the maximum distance violation of less than 0.5 Å, low total energy and the smallest root-mean-square deviation (RMSD) value.
RESULTS AND DISCUSSION
The characterization of conformational and dynamic properties of linear peptides in solution is
a challenging and important focus in current structural biology. NMR spectroscopy remains the
method of choice in this realm of research. During the process of the synthesis of destruxin
analogues, it was found that in contrast to most synthetic peptides a higher than 85% yield of
cyclization was obtained in this series of cyclodepsipeptides. The high yield of the ring closure was
probably due to a favorable conformation of the linear chain precursor. Therefore a systematic
analysis of two linear peptides,
D-leucyl-
L-prolyl-
L-isoleucyl-
L-valyl-
L-alanyl-
b
-alanine (I) and
Vol. 219, No. 2, 1996 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONSD
-leucyl-
L-prolyl-
L-isoleucyl-
L-valyl-
N-methyl-
L-alanyl-
b
-alanine (II), was carried out by
2D-NMR coupled with computer modeling to derive their solution conformation.
Preparation and Cyclization of Linear Peptides
The two linear peptides (I) and (II) were synthesized by automatic solid-phase peptide
synthe-sizer. The linear peptides were purified by HPLC and characterized by mass spectroscopy and
amino acid analysis. The cyclization of the two linear peptides was carried out in DMF solution
with low concentration to avoid dimerization or polymerization. The final reaction solution
ana-lyzed by HPLC showed almost only one product in the cyclization reaction; I
→
III, cyclo-(
D-leucyl-
L-prolyl-
L-isoleucyl-
N-methyl-
L-valyl-
N-methyl-
L-alanyl-
b
-alanyl) and II
→
IV, cyclo-(
D-leucyl-
L-prolyl-
L-isoleucyl-
L-valyl-
N-methyl-
L-alanyl-
b
-alanyl). Both of these two cyclic peptides
showing longer retention times than their corresponding linear peptides in reverse-phase HPLC
were isolated by semi-preparative HPLC with over 85% yield, and their structures were confirmed
by mass spectroscopy and amino acid analysis. It is noteworthy that Lee and Izumiya [21] failed
to cyclize
D-leucic acid-
L-prolyl-
L-isoleucyl-
L-valyl-
L-alanyl-
b
-alanine for the synthesis of
pro-todestruxin. Therefore the cyclization of linear peptides by amide-bond formation should be easier
than by ester-bond formation.
NMR Chemical-Shift Assignments for Linear Peptides
DMSO instead of DMF was used as solvent in NMR study for two reasons: (1) the solvent peaks
of DMF had more interference for the assignment of NMR spectra. (2) DMSO which possesses
similar polar property to DMF seems more suitable as a solvent for peptides.
The
1H-NMR spectra of I and II were shown in Fig. 1. The
1H chemical shift assignments for
I and II are relatively straightforward from HOHAHA, COSY and the conventional procedures of
sequence assignment. Judging from the spectra, there are two forms existed in these linear peptides
due to the presence of proline residue. The ratio of major and minor forms is about 3:1 on the basis
of integrated peak areas. From the ROESY spectra, it was found that strong correlation was
observed between
D-Leu
1a
H and
L-Pro
2d
H in the major form, indicating clearly the major form
as the trans Leu
1-Pro
2conformer. In contrast, cross-peak could be observed between
D-Leu
1a
H
and
L-Pro
2a
H in the minor form which should be the cis Leu
1-Pro
2conformer (Fig. 2). [22]. The
two conformers in the two linear peptides were also demonstrated in the two-dimensional
XH-COSY experiment. All
1H and
13C chemical shifts for two forms of both linear peptides are listed
in Table 1 and Table 2. There are two sets of
13C signals of Pro
2a
, Pro
2b
, Pro
2g
, Pro
2d
and Leu
1a
for I and II.
The
1H-NMR spectra of cyclic peptides III and IV (data not shown) revealed only a single form
in these cyclic peptides. Strong correlation between
D-Leu
1a
H and
L-Pro
2d
H indicated that the
cyclic peptides were the trans Leu
1-Pro
2conformers.
Structure Calculation by Distance Geometry and Refinement by Simulated Annealing
The final structures of I and II were calculated from distance geometry and simulated annealing
based on ROESY distance constraints and dihedral angles. The minor forms of these linear peptides
had too few ROE distance constraints to be calculated. The superpositions of 10 structures of I and
II major forms are shown in Fig. 3. The conformational ensembles of I and II major forms show
good convergence and indeed provide qualitative evidence for folding. The root mean square
deviation (RMSD) values between all atom pairs among the ten structures of I and II major forms
are as follows: the average pairwise RMSD for all atoms is 1.661 ± 0.552 and 1.558 ± 0.645 Å for
I and II respectively whereas that for backbone atoms is 0.448 ± 0.368 and 0.296 ± 0.300 Å for
I and II respectively. The distances between
N- and
C-terminal residues in the trans forms of I and
Vol. 219, No. 2, 1996 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONSFIG. 1. H-NMR spectra of linear hexapeptides (A) peptide I and (B) peptide II. The NMR peaks corresponding to the minor forms of two peptides are labeled with prime (9 ) marks.
II are 5.26 and 4.30 Å, respectively, which are short enough to facilitate the cyclization of I and
II in high yields.
Conformation of the Linear Hexapeptides in Relation to Cyclization
Small linear peptides are usually found to exist in random-coil structure and they exhibit a high
degree of conformational flexibility. However the high yields of cyclization obtained for the linear
peptides studied here would imply that the conformation of these linear peptides in organic solvent
does not exist in completely random-coil like states and possesses somewhat stable folded
struc-tures in order to facilitate the cyclization reaction. Based on the solution strucstruc-tures generated from
NMR data and computer-modeling, the major form of linear peptides with trans Leu
1-Pro
2con-former possesses a distinct
b
-turn in DMSO solution. However, the structure of minor species
could not be calculated owing to weaker NMR signals with overlapping peaks from those of major
species. Major species of linear peptides and cyclic peptides exist as conformers of trans Leu
1-Pro
2.
It is believed that cyclic peptides are produced directly from the cyclization of major trans forms
of linear peptides and the yield of cyclization increases by shifting gradually cis isomer (minor) to
trans (major) isomer with the equilibrium and subsequent conversion of both forms of linear
peptides to a cyclic peptide of trans form.
CONCLUSION
In most cyclization reactions of linear peptides, lower yields were observed [4]. In the synthesis
of two destruxin analogues reported here we have obtained high yields of cyclization for both linear
peptides I and II. The high yields of the ring closure in these two linear peptides probably arise
from a favorable conformation of the linear chain precursor [23]. By 2-D NMR coupled with
computer simulated annealing we have demonstrated qualitative evidence for the existence of
well-defined and folded conformation in the solution structures of these two supposedly flexible
and random-coil like short hexapeptides. The distances between
N- and
C-terminal residues in the
FIG. 2. Part of ROESY spectra (tm4 300 ms) for Peptide I which shows major (trans) and minor (cis) forms. Vol. 219, No. 2, 1996 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
TABLE 1
The Chemical Shifts (ppm),3JNH-CH(Hz) and the Temperature Coefficients ofd(NH)(10−3ppm/K) of the Major Form and Minor Form£of Peptide I in DMSO-d6at 306K
1H chemical shifts Residue NH a-CH b-CH g-CH d-CH Others 3J NH-CH Dd(NH)/DT D-Leu1 7.78(−) 4.13(3.82) 1.55(1.41) 1.75(1.40) 0.88(0.79) 8.77 0.92(0.91) Pro2 − 4.37(4.74) 2.02(2.09) 1.89(1.67) 3.73 1.81 3.38(3.42) Ile3 8.06(8.13) 4.08(4.10) 1.73(−) 1.08(−) 0.80(−) g-CH30.81(0.81) 8.68(8.09) 4.53 1.48(−) Val4 7.60(−) 4.17(−) 1.95(1.95) 0.88(0.88) 8.86 4.14 Ala5 7.99(7.87) 4.20(−) 1.15(1.14) 7.36 4.94 b-Ala6 7.85(−) 2.34(−) 3.23(−) 13C chemical shifts Residue a-C b-C g-C d-C Others D-Leu1 49.71(49.36) 38.88(38.75) 23.40(−) 21.10, 23.02(21.05,23.02) Pro2 59.69(59.44) 29.42(29.30) 23.95(23.10) 46.59(47.05) Ile3 57.62(57.59) 36.05(−) 24.58(−) 11.00(−) g-CH315.45(15.32) Val4 57.21(−) 30.65(−) 17.84,19.11(17.93,−) Ala5 48.13(−) 18.22(18.27) b-Ala6 33.73(−) 34.72(−)
£chemical shifts of minor form in parentheses. Some assignment of the minor species are not listed owing to weak signal or overlap.
TABLE 2 The Chemical Shifts (ppm),3J
NH-CH(Hz) of Major Form and Minor Form
£of Peptide II in DMSO-d 6at 306K 1H chemical shifts Residue NH a-CH b-CH g-CH d-CH Others 3J NH-CH D-Leu1 8.12(8.20) 4.15(3.81) 1.49(1.58) 1.75(1.73) 0.89(0.89) 0.92(0.92) Pro2 − 4.38(4.74) 1.81 1.89(1.69) 3.40(3.42) 2.05(2.10) 3.73(−) Ile3 8.00(8.07) 4.10(−) 1.71(1.72) 1.08(−) 0.82(0.82) g-CH 30.79(0.79) 8.70 1.54(−) Val4 7.78(8.06) 4.52(4.54) 1.98(1.98) 0.84(0.85) 8.74 0.87(−) Me-Ala5 4.91(4.88) 1.16(1.23) NCH32.63(2.62) b-Ala6 7.67(7.87) 2.35(2.39) 3.22(3.32) 13C chemical shifts Residue a-C b-C g-C d-C Others D-Leu1 49.69(49.31) 38.88(38.75) 23.39(23.08) 21.08, 23.00(21.02, 23.10) Pro2 59.67(59.35) 29.41(29.35) 23.95(23.95) 46.59(46.67) Ile3 57.30(−) 36.31(36.10) 24.52(−) 11.02(11.10) g-CH315.34(15.21) Val4 53.56(53.78) 30.11(29.98) 17.97(18.21), 19.15(−) Me-Ala5 51.77(51.64) 14.21(15.11) NCH327.11(26.85) b-Ala6 33.72(33.64) 34.86(35.07)
£chemical shifts of minor form in parentheses. Some assignment of the minor species are not listed owing to weak signal or overlap.
trans forms of I and II are found to be 5.26 and 4.30 Å, respectively, which are indeed close
enough to facilitate the cyclization of I and II in high yields.
ACKNOWLEDGMENTS
This work was supported by Academia Sinica and the National Science Council, Taipei, Taiwan. We also thank Mr. Fong-Ku Shi in M & Vactek Corporation, Taiwan for measuring 2-D ROESY spectra.
REFERENCES
1. Kessler, H. (1982) Angew. Chem. Int. Ed. Engl. 21, 512–523. 2. Kopple, K. D. (1972) J. Pharmaceutical Sci. 61, 1345–1356.
FIG. 3. Superposition of ten energy-minimized structures for the major (trans) forms of (A) peptide I and (B) peptide II. The dashed lines with numerical values indicate the distances between theN- andC-terminal residues of two linear
peptides.
3. Nishino, N., Xu, M., Ueno, Y., Kumagai, H., Mihara, H., and Fujimoto, T. (1992) in Peptide Chemistry 1991 (A. Suzuki, Ed.), pp. 135–140, Protein Research Foundation, Osaka, Japan.
4. Bodanszky, M., and Bodanszky, A. (1984) in The Practice of Peptide Synthesis, pp. 207–210, Springer-Verlag, Berlin. 5. Bodanszky, M. (1984) in Principles of Peptide Synthesis, pp. 217–225, Springer-Verlag, Berlin.
6. Kodaira, Y. (1961) Agr. Biol. Chem. 25, 261–262.
7. Tamura, S., Kiyama, S., Kodaira, Y., and Higashikawa, S. (1964) Agr. Biol. Chem. 28, 137–138. 8. Suzuki, A., Kawakami, K., and Tamura, S. (1971) Agr. Biol. Chem. 35, 1641–1643.
9. Naganawa, H., Takita, T., Suzuki, A., Tamura, S., Lee, S., and Izumiya, N. (1976) Agr. Biol. Chem. 40, 2223–2229. 10. Vey, A., Quiot, J. -M., Vago, C., and Fargues, J. (1985) C.R. Acad. Sci. Paris 300, series III, 647–651.
11. Chen, H. C., Yeh, S. F., Ong, G. -T., Wu, S. -H., Sun, C. -M., and Chou, C. -K. (1995) J. Nat. Prod. 58, 527–531. 12. Liao, S. -Y., Ong, G. -T., Wang, K. -T., and Wu, S. -H. (1995) Biochim. Biophys. Acta 1252, 312–320.
13. Spatola, A. F., Anwer, M. K., and Rao, M. N. (1992) Int. J. Peptide Protein Res. 40, 322–332. 14. Fournier, A., Wang, C. -T., and Felix, A. M. (1988) Int. J. Peptide Protein Res. 31, 86–97. 15. Marion, D., and Wüthrich, K. (1983) Biochem. Biophys. Res. Commun. 113, 967–974.
16. Rance, M., Sørensen, O. W., Bodenhausen, G., Wagner, G., Ernst, R. R., and Wüthrich, K. (1983) Biochem. Biophys. Res. Commun. 117, 479–485.
17. Bax, A., and Davies, D. G. (1985) J. Magn. Reson. 65, 355–360. 18. Bax, A., and Morris, G. A. (1981) J. Magn. Reson. 42, 501–505.
19. Bodenhausen, G., Kogler, H., and Ernst, R. R. (1984) J. Magn. Reson. 58, 370–388.
20. Kessler, H., Griesinger, C., Kerssebaum, R., Wagner, K., and Ernst, R. R. (1987) J. Am. Chem. Soc. 109, 607–609. 21. Lee, S., and Izumiya, N. (1977) Int. J. Peptide Protein Res. 10, 206–218.
22. Wüthrich, K. (1986) in NMR of Protein and Nucleic Acids, pp. 123–125, John Wiley & Sons, New York. 23. Bodanszky, M., and Henes, J. B. (1975) Bioorg. Chem. 4, 212–218.