Journal of the American Chemical Society
834
1 l01:4 1 February 14, 1979
Crystal Structure and Conformation of the Cyclic Hexapeptide cyclo- (Gly-L-Pro-D-Ala)2 E. C. Kostansek,la W. N. Lipscomb,*la and W. E. Thiessenlb Contributionfrom the Gibbs Chemical Laboratory, Harcard Unicersity, Cambridge, Massachusetts 02138, and the Chemistry Diuision, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830. Receiced July 28, 1978
Abstract: The crystal structure and conformation of cyc/o-(Gly-L-Pro-D-Ala)2, CloH30K606, has been determined by singlecrystal X-ray diffraction analysis. The crystals have the symmetry of the orthorhombic space group P212121 with a = 9.389 (6), 6 = 10.379 (8), and c = 22.006 (16) A. The structure was solved by direct methods and refined to a final R of 0.055 on 1575 unique reflections. The hexapeptide shows significant deviation from internal twofold symmetry. This asymmetric conformer apparently optimized one single 4 1 transannular hydrogen bond at the expense of other possible interactions. Conformation angles are compared with those obtained by C D and N M R techniques, and bond distances are discussed. This hexapeptide contains two type 11 turns. All of the peptide bonds are in the trans conformation and the C, atom in one of the proline rings is disordered. -+
In recent years cyclic peptides have been used with increasing frequency for the study of peptide backbone conformations. Despite a large number of NMR studies of cyclohexapeptide conformations in solution, few X-ray crystal studies have been acc~mplished.*-~ Unhindered cyclopeptides such as cyclohexaglycyl and cyclo- (Gly-Gly-Gly-Gly-~Ala-D-Ala) may assume many low-energy conformations; therefore, conformational analysis is quite difficult. Crystal and solution analyses for this type of compound generally differ with respect to the description of the predominant conformers. The addition of proline residues to the cyclopeptide reduces
IN
c,
1
IC IO 2N 2C, 2CB 2c, 2cs 2c 20 3N 3c 3% 3c 30 4N 4ca
4c
40 5h 5ca 5% 5c,1 a 5C,20 5Ca 5c 50 6N 6Ca 6Co
6C 60 a
X
530 (6) 291 (8) 1371 (7) 2147 (5) I497 (5) 2622 (7) 2290 (9) 1562 (9) 713 (7) 2557 (7) 1496 (5) 3813 (6) 4007 (7) 5530 (8) 3691 (7) 3926 (5) 3181 (6) 2754 (8) 2379 (8) 2776 ( 5 ) 1636 (6) 1369 (7) 672 (9) 967 (13) 192 (14) 1178 (8) 327 (7) -905 (5) 889 (5) 9 (7) 724 (8) -297 (7) - 1265 (5)
Y 7725 (5) 8409 (6) 7952 (6) 7012 (4) 8596 (5) 8255 (6) 9102 (6) 10276 (7) 9769 (7) 6803 (6) 6306 (4) 6188 (5) 4836 (6) 4653 (6) 3893 (6) 27 14 (4) 4292 (5) 3365 (7) 4015 (6) 5132 (4) 3350 (5) 3868 (6) 2748 (7) I582 ( I O ) 1874(12) 1982 (6) 5012 (6) 4914 ( 5 ) 6164 (5) 7324 (6) 8276 (6) 7981 (6) 8769 (4)
Experimental Section Crystals of the title compound were groan by slow evaporation from D 2 0 solution and have the symmetry of the orthorhombic space group
Table 11. Calculated Hydrogen Coordinates ( X IO3)
H bonded to atom
Table I. Fractional Atomic Coordinates ( X IO4) atom
the number of possible conformers and both Blout5 and Kopple6 have synthesized and studied (by NMR) C2-symmetric hexapeptides containing proline. We report here the crystal structure and conformation of cyclo-(Gly-L-Pro-D-Ala)z, which, unlike ~yclo-(D-Ala-L-Pro-D-Phe)z,~ does not retain the twofold symmetry in the solid state.
x
132 -78 41 370 158 3 27 86 23 5 -37 68 463 324 565 565 580 309 183 363 23 3 -47
Z
942 (2) 1503 (3) 1957 (3) 1854 (2) 2482, (2) 2921 (3) 3472 (3) 3201 (3) 2672 (3) 3077 (3) 3291 (2) 2981 (2) 3163 (3) 3408 (3) 2645 (3) 2738 (2) 21 14 (2) 1654 (3) 1068 (3) 968 (2) 664 (2) 55 (3) -284 (3) 50 ( 5 ) 194 (6) 719 (3) 92 (3) 259 (2) -89 (2) -144 (3) -579 (3) 463 (3) 500 ( 2 )
111
144 -23 8 193 -88 19 19 200 63 208 195 -102 5 I70 Occupancy is 0.50.
Occupancy is 0.50.
0002-7863/79/1501-0834$0 I . O O / O
G 1979 American Chemical Society
Y
Z
705 82 I 944 842 860 938 1073 I098 953 I047 667 462 360 5 20 500 525 283 270 422 290 268 227 309 91 113 214 89 190 140
91 167 143 275 379 37 1 353 305 28 1 230 278 352 340 380 298 203 182 157 -17 -3 1 -74 -58 -55 I -12 34 2 97 94
180
115
I32 622 70 I 900 860
68 -19 -32 -70 -40
Kostansek, Lipscomb, Thiessen Table 111. Bond Lengths bonds Ni-Cai cai-Ci ci-0, Ci-N,+
I
i = l
2
1.441 1.500 1.237 1.341
1.474 1.546 1.217 1.358 1.53 1 1.519 1.507 1.482
Cyi-CGi CGi-Ni
_ _ _ _ _ ~ - ~
Ci-iNiCai NiCaiCi CaiCiNi+ 1 caicioi Nj+lCiOi
cicaicpi
Gly
835
(A) and Angles (deg)
coli-cpi cpi-cri angles
1 cyclo- ( G l y - ~ - P r o - ~ - A l a ) z
L-Pro
NiCaiCPi cai epicyi CpiCyicGi CyiCGiNi CaiN,CGi Ci- 1NjCGi
4
1.471 1.532 1.259 1.330 1.540
1.452 1.498 1.238 1.323
D-Ala
Gly
L-Pro
D-Ala
120.4 111.5 117.9 120.2 121.9
120.9 110.0 114.4 123.7 121.9 109.9 104.0 107.8 (104.6)" 106.6 (108.8)" 102.5 (100.9)" I 1 1.9 126.6
121.7 113.6 119.5 119.8 120.7 110.0 109.6
120.9 112.5 121.7 117.8 120.5 111.2 109.2
120.5 110.6 112.9 122.1 124.9 112.1 103.6 104.4 105.2 104.0 111.7 127.5
119.2 108.0 118.4 121.7 119.9
5
3
Values for alternate disordered position.
6
1.465 1.541 1.218 1.367 1.529 1.443 (1.460)' 1.541 (1.484)" 1.489
X I
av for polypeptidesb
1.461 1.524 1.232 1.343
1.455 1.51 1.24 1.325
120.6 111.0 117.5 120.9 121.6
122 I11 116 120.5 123.5
1.465 1.526 1.226 1.338 1.531
See R. Marsh and J. Donohue. Adc. Protein C h e w , 22, 235 (1967). 60
Table IV. Proline Dihedral Anglesa (den)
Pro2 Pro5 Pro5 disordered
av for cyclo-(Gly-Pro-D-Ala)*
XI
x2
x3
-28.5 -17.6 17.7
34.8 26.9 -31.0
-27.0 -25.1 30.6
x4
9.1 14.3 -18.9
6W
I ,5@ 5c
= N-Ca-CP-CG. x 2 = Ca-CP-Cy-CG, x3 = CP-Cy-CG-N, x4
= Cy-CG-N-Ca.
P212121.The unit cell hasdimensionsa = 9.389 (6), b = 10.379 (8). and c = 22.006 (16) 8, (A = 1.541 78 A) and contains four molecules and no solvent. Calculated and observed7 densities are 1.395 and 1.39 f 0.01 g/cm3, respectively. The crystal used for data collection had dimensions 0.07 X 0.23 X 0.40 mm. Integrated intensities of I705 independent reflections (28,,, = 115') were measured with Ni-filtered Cu K a radiation on a Syntex P21 four-circle diffractometer. The data were collected using w scans ( 1 '/min) with IO-s background counts a t each end of the scan. Three reference reflections were measured every 50 reflections and these remained constant throughout data collection. Reflection intensities were corrected for background, polarization, and Lorentz effects. Those I575 reflections (92%) which had values of F , 1 30(F,) were included in the refinement. Structure Determination and Refinement. The structure was solved using the weighted multiple solution tangent formula approach of direct methods.8 A total of 64 phase sets each consisting of those 200 normalized structure factors, I E I 's, having 1 E I 2 1.50 was generated. An E map calculated from the phase set with the highest absolute figure of merit revealed 30 of the 32 atoms in the structure. The remaining two atoms were located from a n F , map based on this model. Block-diagonal matrix least-squares refinement utilizing individual isotropic temperature factors and a fractional weighting scheme9 reduced the R value to 0.1 23, where R = 81 IFo[ - 1F.l / / Z l F , , . At this point we noticed an unusually large temperature factor associated with the 5C, atom of a proline ring. Also, the 5Cp-SC,-5Ca angle was abnormally large ( 1 1.5'). These conditions are indicative of a disordered C, carbon atom. Careful inspection of a difference Fourier map generated without the 5Cp, 5C,, and 5Ca carbons revealed two peaks of equal magnitude near the 5C, site (one above and one below the proline ring plane). The 5Cp and 5Cfiatoms appeared as single, though elongated, peaks. The structural model was altered to include two positions of one-half occupancy for the 5C, carbon. After several cycles of full-matrix least-squares refinemcnt employing individual anisotropic temperature factors, a difference
I
I
10 20
I'
2c ZN
30 Figure 1. The chemical structure and numbering scheme of c,vc'/o-(glycql-L-proiyi-D-alanyl)Z.
Fourier map was calculated. Because of the disorder some of the hydrogens could not be located. Consequently, the calculated positions of the hydrogen atoms (including alternate positions necessitated by the disordered C, carbon) were included as constants in the final least-squares cycles. An isotropic thermal parameter, B , equal to 5.0 A2 was assigned to each hydrogen. The approximate positions of the six methyl hydrogens used in the calculations were derived from the difference map. The refinement converged at a final residual, R.of 0.055.
Results Figure I is a drawing of t h e molecule indicating t h e labeling scheme. The positional parameters for t h e nonhydrogen atoms, including estimated s t a n d a r d deviations, are listed in Table I. T a b l e I I contains t h e calculated coordinates of t h e hydrogen a t o m s a n d Table 111 lists t h e hexapeptide bond lengths a n d angles. T h e largest estimated s t a n d a r d deviations a r e less t h a n 0.03 8, for t h e lengths a n d 1.7' for t h e angles. A comparison of t h e average backbone distances a n d angles for this structure with s t a n d a r d values shows t h a t t h e lengths a n d angles a r e close t o expected values. Figure 2 is a stereoscopic view l oof c);clo-(Gly-L.-Pro-D-Ala)2 p e r p e n d i c u l a r t o t h e p e p t i d e ring.
836
Journal of the American Chemical Society
/
101:4
/ February
14, 1979
L
Figure 2. A stereoscopic view of eyelo-(Gly-L-Pro-D-Ala)z perpendicular to the peptide ring. The cy carbons are numbered. Note the disordered atoms in Pro5 at lower left. The 4-1 transannular 14 bond I S indicated by the t h i n line.
Figure 3. A stereoscopic view of the unit cell contents of cyclo-(Gly-L-Pro-D-Aia)z parallel to h.
The anomalously short ~ C B - ~ Cbonds , most probably result from the disordered 5C, atom and the large thermal motions associated with the adjacent atoms. Similar behavior of a C, pyrrolidine atom has been observed in the analysis or L-LeuL-Pro-Gly." The alternate 5C, positions are separated by 0.85 A. The y carbon in pyrrolidine rings usually exhibits a large thermal motion. It is interesting that just one of the carbons in the structure displays the disorder. Table IV lists the proline dihedral angles for both ordered and disordered residues. Pro2 assumes a Cz-C,-exo conformation,12 while the Pro5 disorder creates intermediate conformations. The peptide bonds are all close to the ideal trans conformation, which indicates a rclatively unstrained molecuie. The average deviation from trans is 5.7', whereas the maximum is 7'. The structure contains two type I1 p turnsI3 and a single 4-1 transannular hydrogen bond. This bond is characterized by a 4N to 10 distance of 3.041 A and a 156' 4N-H 10 angle. By contrast, the 1N to 4 0 distance, a t 3.418 /i,is too long to be considered a hydrogen bond. The 1 0 - 4 0 carbonyl oxygens are separated by a 2.835-A distance across the peptide ring. Each hexapeptide molecule participates in two intermolecular hydrogen bonds. The first is a relatively strong bond of 2.805 /i between atom 6 N and atom 6 0 ' of an adjacent molecule. The 6N-H 6 0 ' atoms form a 177' angle. The second hydrogen bond, which forms with a different adjacent molecule, is a weaker one with a 3N to 3 0 ' distance of 3.096 A and a 172" 3N-H e 3 0 ' angle. Thus, the carbonyls parallel to the peptide rings participate in a hydrogen-bonding network with the amide hydrogens perpendicular to the rings. Figure 3 shows a stereoscopic view of the unit cell contents.
--
e
..
Discussion Crystal structure data are now available for two cyclic hexapeptides containing proline residues. The crystal conformation of cyclo-(Gly-~-Pro-D-Ala)2 differs from that of cyclo- ( ~ - A l a - ~ . - P r o - ~ - P hinasmuch c)2 as the latter structure preserves the C? symmetry observed in the solution studies,6
Table V. Comparison of Crystal and S o l ~ t i o n(Approximate) '~ Peptide Backbone Conformational Angles'] (deg) i = 2.5-~-Pro i = l,4-Gly crvstsoh crwt soln 6, -179, -173 -160 -54, -70 -70 4, 170.-163 I60 125, 116 90 ( L ' ~ -175,-173 180' 174. 174 180'
i = 3,6-D-Ala
crvst
soh
94,79 -5, 19 -174, 176
90 0 180'
r7 Using the conventions set forth by ILPAC-IUB Commission on Biochemical Nomenclature, Biochemistry, 9, 347 1 ( I 970). Ideal trans conformation assumed in N M R studies. Solution Conformation is C: symmetric.
whereas the former does not. Both cyclo-(Gly-~-Pro-~-Ala)2 and its analogue, cyclo-(Gly-~-Pro-D-Phe)2, have been examined in solution by ' H and I3C nuclear magnetic resonance and by circular dichroism (CD).I4These data indicate that the preferred conformation of both molecules consists of all-trans peptide bonds and is stabilized by 4-1 intramolecular hydrogen bonds in type I1 0 turns. These cyclopeptides are C2 symmetric on the % M K time scale and coupling constants, together with model building and examination of theoretical CD spectra, support the following approximate CP, \k angles: Gly (-160, 160°), Pro (-70,90°), and D-Pheor D-Ala (90, 0'). It waf concluded that averaging around this structure in solution was probable on the basis of the difficulty in fitting all of the data by one unique structure. Solution and X-ray data are compared in Table V. The crystal and N M R data for c~clo-(Gly-L-Pro-D-Ala)z agree qualitatively. The largest discrepancies occur in the angles and the loss of twofold rotation symmetry. The observation of any asymmetry in solution is precluded by the dynamics of the experiment.I5 The presence of C2 symmetry i n the cyclo- (L-Ala-L-ProD-Phe)z crystal structure allows no transannular hydrogen bond formation because of close contact distances. Apparently,
+
Maeda, Ingold / EPR of Diazirinyl Radicals
837
an asymmetric conformer can optimize one hydrogen bond in References and Notes a fl turn a t the expense of the other and still maintain undis(1) (a) Harvard University; (b) Oak Ridge National Laboratory. torted backbone angles. This is the case for c y c l ~ - ( G l y - ~ - (2) (a) I. Karle and J. Karle, Acta Crystallogr., 16, 969 (1963); (b) I. Karle, J. Gibson, and J. Karie, J. Am. Chem. SOC.,92, 3755 (1970). Pro-D-Ala)z. The intermolecular hydrogen bonds seem to (3) J. Brown and R. Teller, J. Am. Chem. Soc., 98, 7565 (1976). support the effect. The stronger one reinforces the peptide twist (4) M. 6.Houssain and D. van der Helm, J. Am. Chem. SOC., 100, 5191 (1978). away from a possibly favorable I N - H 4 0 interaction and (5) E. Blout, C. Deber, and L. Pease, in "Peptides, Polypeptides and Proteins", the weaker one reinforces the twist toward the existing favorProceedings of the Rehovot Symposium, 1974, Wiley-lnterscience,New able 4N-H . * 10 hydrogen bond. York, N.Y., 1974. (6) K. Kopple, T. Schamper, and A. Go, J. Am. Chem. SOC., 96, 2597 We take this opportunity to correct 41from 109' to -109' (1974). in the studyt6 of the naturally occurring cyclic peptide, p(7) The density was determined experimentally by flotation of several crys!als in n-heptane-carbon tetrachloride mixtures. amanitin. (6) G. Germain, P. Main, and M. M. Woolfson, Acta Crystallogr,, Sect. A, 27, -B 6-R- ,(1971) .- . Acknowledgment. W e wish to thank Drs. Lila Pease of (9) E. Hughes,"J. Am. Chem. Soc., 63,1737 (1941). (IO) C. Johnson, "ORTEP, a Fortran Thermal-Ellipsoid Plot Program", USAEC Amherst College, C.-H. Niu of the National Institutes of Renort ORNL-3794. 1965. - Health, and Elkan Blout of the Harvard Medical School for (11) Y. Leungand R. Marsh, Acta Crystallogr., 11, 17 (1956). providing the title compound and making the NMR data (12) T. Ashida and M. Kakudo, Bull. Chem. SOC. Jpn., 47, 1129 (1974). (13) C. M. Venkatachalam, Biopolymers, 6, 1425 (1966). available to us. W e acknowledge support of this research by (14) (a)L. Pease, Ph.D. Thesis, Harvard University, 1975; (b) C . 4 Niu, unpubthe National Institutes of Health (GM 06920) and the Nalished results. (15) In one case, that for cyclo-(a-Phe-L-Pro-GIy)p, the molecule appeared to tional Science Foundation (CHE-7719899). This research was be CZsymmetric from NMR spectra, but displayed four amide I stretches sponsored in part by the Energy Research and Development in an infrared spectrum (in CHCIS).'~ This result suggests that some conAdministration under contract with Union Carbide Corp. formational asymmetty may occur in this type of peptide, yet go unobserved
-.
-
. - r -
Supplementary Material Available: Temperature factors ( i page). Ordering informaCion is given on any current masthead page.
(16)
-
I
by NMR. Unfortunately,the solubility of cyclo-(Gly-L-Pro-o-Aia)2in CHCI:, was too low to examine it in the same manner. E. C. Kostansek, W. N. Lipscomb, R. R . Yocum, and W. E. Thiessen, Bo-
chemistry, 17, 3790 (1978).
Kinetic Applications of Electron Baramagnetic Resonance Spectroscopy. 33. Diazirinyl Radicals' Y. Maeda2 and K. U. Ingold* Contributionfrom the Dicision of C'hernistry, National Research Council of Canada, Ottawa K1 A OR6, Ontario, Canada. Receiced J u l y 28, I978
___I
Abstract: Some 3-substituted diazirinyl radicals, KC=NN-, have been generated by photolysis of the parent bromides in the presence of hexa-n-butylditin. The principal EPR parameters for 3-alkyldiazirinyl and 3-phenyldiazirinyl are similar: a N ( 2 N ) = 7.8 Ci, g = 2.0042. I N D O calculations give I4N and I3C hyperfine splittings in good agreement with experiment. Diazirinyls are IT radicals, the two nitrogens' 2p, atomic orbitals making the major contribution to the semioccupied orbital. Diazirinyls decay with second-order kinetics to yield the corresponding nitrile. Like other N-centered three-membered ring radicals, they do not form nitroxides. Studies on the products of reaction of aziridinyl, CH2CH2N., with tert-butylperoxy have revealed that a nitroxide is probably formed, but it decomposes (to ethylene and NO) too rapidly for i t to be detected. It is suggested that analogous processes occur with diazirinyls and other N-centered three-membered ring radicals.
'The t h e r m ~ i l ~and - ~ photolytic8--l3 decomposition of 3alkyl-3-halodiazirines and 3-aryl-3-halodiazirines,I,l4 have been studied as sources of "free" halocarbenes, 2. The possiR
A
hV
I
'
'C:
x
/
+
N,
2
bility that free radicals are involved in some of the systems investigated does not appear to have been explored, nor even suggested. W e have discovered that 3-organodiazirinyl radicals, 3, can be derived from a variety of 1. These species rep-
resent a hitherto unidentified class of nitrogen-containing radicals. In this paper we report on their generation, identifi0002-78631791 I50l-O837$Ol.OO/O
cation by EPR spectroscopy, decay kinetics, and decay products.
Experimental Section Materials. Bromodiazirines. The 3-alkyl-3-bromodiazirines and 3-phenyl-3-bromodiazirine were prepared from the corresponding amidine hydrochloride5 by oxidation with freshly prepared aqueous sodium hypobromite in MezSO according to the general method described by Graham.IJ The volatile alkylbromodiazirines ( R = CH, and CH3CH2) were collected continuously by means of a vacuum
pump which pulled them through a train of four U-tubes held at -35, -80, and -80 OC with the gases bubbling through n-pentane, and - 196 OC. These diazirines were retained in the pentane-filled U-tube. The less volatile organobromodiazirines ( R = (CH3)3C, ChHj, and C ~ H S C Hwere ~ ) extracted continuously into n-pentane and were then purified by column chromatography through silica gel. The infrared @ I979 American Chemical Socicty
