Int. J. Peptide Protein Res. 8, 1976, 445-453 Published by Munksgaard, Copenhagen, Denmark No part may be reproduced by any process without written permission from the author(s)
A SQUARE PYRAMIDAL MODEL FOR RIBONUCLEASE ACTION ROBERTR. HOLMES Department of Chemistry, University of Massachusetts Amherst, Massachusetts, US.A. Received 25 August 1975
A mechanism incorporating a square pyramidal model ( S P )is presented for the action
of bovine pancreatic ribonuclease. Its formulation i r based on structural principles governing pentacoordinate behavior. The model is compared with a previous trigonal bipyramidal (TP) representation with regard to the geometry of the active site and enzyme constraints. Of two cariants of the SP model, an adjacent (cis displacement) and in-line (trans displacement)procesy, the in-line mechanism, as with the TP model, fits exivting model studies. Consideration of the energetics ofthe SP us. the TP model leads to an estimated energy difference of’about 1-2 kcallmol. This suggests that the preferred model may be intermediate in geometry between the two idealized representationsfor the enzymatic hydrolysis. Comparisonsare made showing that pseudorotation is an unlikely process in either model.
In order to adequately understand the action of bovine pancreatic ribonuclease (RNase) in catalysing the hydrolysis of the 3’,5’-phosphodiester linkage of RNA (l), it is necessary to acquire an intimate knowledge of the stereochemistry of the active site species. Previous work has accomplished a great deal in achieving this goal. A multitude of chemical studies (1) have shown that the enzymatic hydrolysis is best described as a two-step process, a transphosphorylation in which attack by the 2’-OH group leads to a 2’,3’cyclic phosphate followed by hydrolysis of the latter to yield a terminal 3’-phosphate (2,3). Recent mechanistic proposals (4-6), aided by observations on mechanisms ascribed to non enzymatic hydrolysis of phosphate esters (7,8), have centred on the formation of a trigonal bipyramidal intermediate at the active site (Fig. I). Both Usher et al. (5,6) and Roberts et al. (4) have provided convincing arguments utilizing chemical and structural evidence in outlining the details of RNase action. Their arguments are strengthened by crystal structure determinations
of RNase A (9-11) and of RNase S (11-14) showing the importance of histidine 12, histidine 119, and lysine 41 at the active site (Fig. 2). Both groups of workers (4-6) favor an in-line or linear mechanism, i.e., one in which the entering group does so opposite to the side occupied by the group which leaves in a subsequent stage. This contrasts with the so-called adjacent mechanism (15), wherein the entering group is oriented cis to the potential leaving group. In the latter mechanism, a “pseudorotation” is required to bring the departing group to an apical location in the trigonal bipyramidal intermediate. In so doing, the new trigonal bipyramid is formed by an intramolecular process involving passage through a low lying transition state, presumably having a square pyramidal conformation (16-18). Recently, X-ray determinations (19-21) on oxyphosphoranes containing cyclic substituents have shown that the square pyramid exists as a structurally stable entity. Since the make-up of these derivatives resemble those postulated at the active site in RNase catalysis, it becomes im445
HIROSHI TAKETOMI, YUZO UEDA AND NOBUHIRO G b
transition was discussed. In this paper the specificity is treated, not as a given fact, but as deriving from the heterogeneity of the amino acid sequence of a protein. In order to study the statistical mechanical properties of denaturation and conformational fluctuations of proteins and to understand the mechanism of realization of specificity of native conformations of proteins, we introduce a lattice model of proteins into which heterogeneous aspects in the amino acid sequence are incorporated. In statistical mechanical treatments of polymers in general, inter-unit interactions are conceptually divided into short-range and longrange interactions. Interactions between units that are separated far along the chain are defined to be long-range, even when the units are not widely separated in space. This division of the interactions into short-range and long-range ones is important also in theoretical treatments of protein conformations. Existence of the correlations between short-range amino acid sequence and secondary structures (e.g., a-helix, /?-structure and /?-turn) in globular proteins has been shown recently by many authors. This indicates that the short-range interactions have considerable importance in the process of folding, even though it is apparent that the long-range interactions have their own role in determining and stabilizing the native conformation of proteins. This role of the long-range interaction has not yet bem sufficiently studied theoretically because of the theoretical difficulties intrinsic in the long-range interactions. Differences in amino acids manifest themselves in heterogeneities in both the short-range and long-range interactions. In this paper emphasis is placed on the role of the long-range interactions. The shortrange interactions are not discussed. The heterogeneity of the long range interactions is taken into the lattice model by a very idealized form of the specificity of inter-unit interactions, i.e. attractive interactions are assumed to work for a set of specifically preassigned pairs of units occupying the nearest neighbor lattice points. In section I. the model is described. Even though abstracted and simplified, the model is still far from being amenable to analytical treatment. In this paper the model is studied by the Monte Carlo simulation method of Metropolis et al. (2). The method of Monte Carlo calculation is described in section 11. In section 111, the results of the simulations are
446
presented and discussed. The purpose of abstracting protein molecules as a lattice polymer is to enable otherwise complex phenomena to be studied in as pure and simple a form as possible. Because of the simplification and idealization, some of the conformational properties of proteins become lucidly understandable. However, this is achieved by discarding some of the important aspects of real proteins. The discussion of these aspects is also included in section 111. In section IV brief conclusions and the outlook for further studies are given.
,
I. DESCRIPTION OF THE MODEL
In this paper we are considering a lattice polymer on the two-dimensional square lattice. A study on a lattice polymer on the three-dimensional cubic lattice is now in process and will be presented in a future paper. The polymeric chain on the lattice is assumed to be self-avoiding due to repulsive interactions between two units within the contact distance. The heterogeneity in the amino acid sequence is incorporated as the specificity of inter-unit interactions. Attractive interactions are assumed to work only for a set of preassigned pairs of units occupying the nearest neighbor lattice points. Such pairs of units are defined as being interactable and are indicated by filled (black) squares in Fig. 1. A pair of units remains unfilled (white), i.e. the pair is not interactable, when no interaction works between them even if the pair occupies the nearest neighbor lattice points. Thus Figs. lA, B and C represent the inter-unit interactions with the three different types of specificity, respectively. The distribution of interactable pairs in Fig. 1A is determined as follows. First the native conformation of a protein consisting of 49 units is assumed on the square lattice as in Fig. 2. All pairs of units occupying the nearest neighbor lattice points in the conformation of Fig. 2 are filled black and all other pairs are unfilled in Fig. 1A. The energy of inter-unit interaction is assumed to be identical for all interacting pairs with a value of - 8 in this paper. Therefore the conformational energy of this protein molecule is measured in the unit of ( - 8 ) . Since all neighboring pairs in Fig. 2 are interacting for specificity A specified by Fig. lA, the conformational energy of the conformation of Fig. 2 is - 36.2 for
PROTEIN FOLDING, UNFOLDING AND FLUCTUATIONS
B
1 20 30 40 49 FIGURE 1 Specificity of inter-unit interactions. Attractive interaction is assumed to work when a filled (black or shaded) pair of units occupy the nearest neighbor lattice points, A: strong limit specificity, B: intermediate specificity, C : weak limit specificity.
this specificity of interaction. It is evident that any conformations other than those shown in Fig. 2 (and its mirror image) have conformational energies higher than - 3 6 for this specificity. Therefore for specificity A, the native conformation, i.e. the conformation with the global minimum energy, is that given in Fig. 2. (No distinction will be made between a conformation and its mirror image in this paper.)
In Fig. 1C all 552 pairs of units separated by (2n + 1) bonds (n 1) are filled black, i.e. attractive interaction works nonspecifically for any pair of units occupying the nearest neighbor lattice points. No other pairs of units can occupy the nearest neighbor lattice points due to geometrical reasons. Because no specificity exists, the case of Fig. 1C is reduced to a homogeneous nonintersecting chain polymer with short-range'
447
ROBERT R . HOLMES
identical equatorial groups in the TP. Similarly, in the SP, more charge resides at basal relative to apical ligands. Due to geometrical constraints, four or five membered rings preferentially span apical-equatorial sites in the T P and cis.basal positions in the SP (22). Pi donor ligands achieve better bonding in equatorial positions of the T P and in the apical position in the SP (27-28). These are the main structural differences derived from work with stable pentacoordinate derivatives and provide a complementary set of principles sufficient to structure the SP substrate model. The substrate structure illustrated in Fig. 3 conforms to these constraints. Moreover, entering and leaving groups are expected to utilize basal positions of the SP rather than the lone apical position because of the reduced bond strength likely to be associated with the basal bonds. Either cis or trans orientations of attacking and displaced groups are possible. Binding of the two nucleoside bases on the enzyme, as suggested from a consideration of the enzyme conformation (4), positions the phosphate unit near the active site cleft and allows hydrogen bonding to occur between the nonesterified phosphate oxygen atoms and probably both histidine 119 and lysine 41. This results in the initial enzyme-substrate conformation suggested by Roberts et al. (4) (Fig. la). In the transesterification step (Fig. 3), a trans displacement is indicated (in-line mechanism) with the SP intermediate structured according to the postulates outlined above. One of the free phosphate oxygens is shown at the apical location rather than the nucleoside positioned at the 5’ ester linkage in line with the preference for strong donor groups to locate in the apical position (cf. the corresponding T P with the donor oxygen atoms occupying equatorial sites and the ester linkage located in one of the apical positions, Fig. 1 b). Hydrogen bonding then completes the formulation. In the second step, essentially the reverse of the first step, attack by a water molecule leads to its entry into a basal site of the second SP intermediate (Fig. 3 with a water molecule in place of the 5’ ester linkage). Upon protonation, hydrolysis proceeds leading to the terminal 3’phosphate with decyclization. The stereochemical consequences are the same in either the SP or T P model for both an adjacent
448
or in-line process. However, unlike the adjacent process for the T P model, pseudorotation is not a necessary process in the adjacent mechanism for the SP model, i.e., cis displacement, to complete the transformation since the apical position of the SP is not involved in the mechanistic sequence. In-line versus adjacent mechanisms Excellent work (5,6) aimed at uncovering the geometry of each step in ribonuclease action centered on the use of the two diastereoisomers of 2’,3’-cyclic phosphorothioate[Up(S)l as model systems.
I
I1
The results are consistent with an in-line process proceeding through a TP intermediate (5,6). The results are also consistent with the SP model proceeding in-line by a trans basal displacement. These parallel mechanisms are compared schematically in Fig. 4 for the enzyme catalyzed synthesis of the dinucleoside uridylthio 3’3’phosphorylcytidine Up(S)C from the crystalline isomer of Ufi(S) and cytidine (6). The latter synthesis was designed to model the first step of ribonuclease action. Similar pathways may be constructed to show the same correspondence between the two models for the RNase catalyzed hydrolysis of the crystalline isomer of Ufi(S) in water enriched in ‘*O,and subsequent recyclization of the resulting monoester with diethyl phosphorochloridate. The latter process serves as a model system for the second step of ribonuclease action ( 5 ) . SP versus TP model At first hand it would appear that the T P model might be preferred over the SP model for ribonuclease action. For isolatable five coordinate phosphorus compounds, including both simple
A SQUARE PYRAMIDAL MODEL FOR RIBONUCLEASE ACTION
0 2'
0
0 [/!
3'
1
I ADJ. ROH
1
-
ROH IN-LINE
S
OR
11
INLINE
1
BASE
OR
J t 3ER
/"
pp\ S
OR FIGURE 4a Schematic diagram of model study (6) of the first step of ribonuclcase action proceeding by both in-line and adjacent mechanisms for (a) a TP intermediate and (b) for a SP intermediate (on next page). The nonenzymatic base catalyzed reaction is known to proceed in-line.
I
BASE
'OR
and monocyclic derivatives, the structure invariably found is a TP (22,25). Only recently have examples of SP structures of phosphorus been uncovered and these contain two small-membered ring systems (19-21). A prime example is the oxyphosphorane (19)
It has been estimated (18,30) that ring strain in C
OR
the latter example is relieved by about 2.0 kcat/ mol per ring compared to the related TP structure. It has also been estimated (18) that a SP structure containing five monocoordinated oxygens PO, is approximately 4 kcal/mol less stable than the isomeric TP.If one adds to this, the effect of hydrogen bonding present in the enzymesubstrate environment (Figs. l b and 3), which tends to favor the SP as a consequence of the resultant charge withdrawal (31), the energy balance may be tipped in favor of the SP model. The ring strain factor stabilizing the SP state (Fig. 3) should be somewhat less than the 2.0 kcal/mol suggested above since the latter value applies to an unsaturated planar system. In the
449
ROBERT R . HOLMES
S
S
S
T
1
It S
I S
Fig. 4b.
formation of the 2',3'-cyclic phosphate here, ring puckering provides a means of alleviating ring strain in the T P relative to the SP. While no reliable estimate of this reduction has as yet been made, it is noted that the PO apical bond distance involved in cyclic phosphoranes having TP structures drops from about an average of 1.75 8, for unsaturated ring systems (22) to about 1.71 8, for related saturated five-membered ring systems (32,33). These values compare to 1.67 A for an apical PO bond distance in a TP assumed (18) as a normal value in the absence of ring constraints and a 1.65 A value for the PO bond distance for a five-membered ring containing oxyphosphorane oriented in cis basal positions of a SP (18,19). 450
Thus, the ring strain expected in a 2',3'-cyclic phosphate spanning an eq. apical pair in a TP conformation does indeed appear to be less than that in a related unsaturated ring structure. Again, no quantitative assessment of the stabilization of a SP relative to a TP as a result of hydrogen bonding is at hand. We note, however, that theoretical calculations of DelEene & Pople (34) on the formation of water dimers suggest a shift of about 0.05 electrons from the OH bonds of the donor water molecule. If we use this as an approximation of the reduction in charge density for each of the P-0 bonds involved in hydrogen bonding with the histidine and lysine residues, the phosphorus atom experiences an increased
A SQUARE PYRAMIDAL MODEL FOR RIBONUCLEASE ACTION
positive charge of 0.15. This amounts to a change in electronegativity of approximately 0.1 unit on the Pauling scale (35), about that experienced by inserting an arsenic atom in place of a phosphorus atom. It has been estimated that the SP-TP energy difference drops from -4 to 3 kcal/mol for AsF, compared with PF, (36). This difference should not change appreciably for comparisons between AsO, and PO, moieties. Consideration of the various energy terms given here then suggests that the TP transitional state (Fig. lb) is more stable than the corresponding SP transitional state (Fig. 3) by only 1-2 kcal/ mol. Although extensive refinement is needed, it is felt that the qualitative arguments demonstrate the closeness in energy of the two idealized pentacoordinate structural forms of phosphorus as models for RNase activity. Because of the magnitude of energy barriers for intramolecular ligand exchange deduced from n.m.r. analysis of solutions of pentacoordinate ‘phosphorus derivatives containing either bulky groups (37-39) or substituents undergoing hindered rotation (27,28,40-42) ranging from 9 to 23 kcal/mol, it seems unlikely that the more severe constraints present in the enzyme environment will allow pseudorotation to be a feasible process. For pseudorotation to occur in the enzyme-substrate intermediates shown in Figs. 1b and 3, visualizing a modification of these diagrams to give an adjacent location of the 5‘ ester linkage in the former and a cis orientation of this same linkage in the latter, higher barriers for intra ligand reorientation are expected in these complexes compared with a similar nonenzymatic rearrangement because of the specific binding sites provided by the enzyme. The latter interactions should hinder the pseudorotational process which has been found to be concentration independent (25b,c) and, hence, unassisted by solvent molecules. In fact, interaction of basic solvent systems (4344) with pseudorotating molecules appears to hinder the ligand exchange process (45). Therefore, pseudorotation in a pentacoordinated intermediate involved in RNase action would serve to lower rather than increase the hydrolytic breakdown of nucleic acids. On this basis, an adjacent process for the TP model does not seem to warrant serious consideration. From the foregoing analysis, what appears more likely is the presence of either a five-
coordinate transition state (concerted process) or intermediate formation which bears some distortion between the two idealized representations. Considering the closeness in energy, estimated for the two models, the extent of distortion may be appreciable. In this context, Wood (46) and Muetterties & Guggenberger (47) have shown that distortion between the two idealized states, determined from X-ray studies on a variety of five coordinate species of DBhsymmetry, follows a well-defined path, that of the C2” cons,raint associated with the Berry reaction coordinate for intramolecular exchange (1 6). However, the distortion would not follow the same pathway for the intermediates (or transition states) in ribonuclease action since the initial symmetry of the proposed structure is lower than Dab. More likely, the transitional state distortion will be dictated not only by the closeness in energy of the TP and SP, as argued above, but also by the non covalent interactions provided by the enzyme environment. Furthermore, the phosphate substrate is no doubt distorted to some degree from the tetrahedral stale toward the transitional state as a result of enzyme constraints. This seems to be apparent, at least, for PO bond lengths. Phosphoester PO bond lengths in uridylyl3’,3’-adenosine phosphate (48), for example, average 1.61 A while the phosphoryl PO bond lengths are over 0.1 A shorter, averaging 1.48 A. An overall expansion of bond lengths by approximately 0.2 A is needed to form the T P in Fig. 1b and about 0.25 A to form the SP depicted in Fig. 3 (18). Both the non covalent bonding and hydrogen bonding, via electron withdrawal, are expected to provide some of the required energy to loosen the bonds in the phosphate bound structure in RNase. In view of the above considerations, the preferred model for enzymatic hydrolysis is expected to involve a transition state intermediate in geometry between the two extreme pentacoordinate forms. However, for non enzymatic hydrolysis involving acyclic phosphorus esters or phosphorus esters containing one small-membered ring system, distortion from the TP model should be considerably less. ACKNOWLEDGMENTS
This investigation was supported by grants from the National Science Foundation (MPS 74-1 1496) and 45 1
ROBERT R. HOLMES
20. EISENHUT, M., SCHMUTZLER, R. & SHELDRICK, W. S . (1973) J. Chem. SOC., Chem. Commun. 144-145, and personal communication. 21. HOWARD, J. A., RUSSELL,D. R. & TRIPPETT, S. REFERENCES (1973) J. Chem. SOC.,Chem. Commun. 856-857. R. R. (1974) J. Am. Chem. SOC.96,41431. RICHARDS,F. M. & WYCKOFF,H. W. (1971) in 22. HOLMES, 4149 and references cited therein. The Enzymes (Boyer, P. D., ed.), 3rd edn., vol. 23. BEAUCHAMP, A. L., BENNETT, M. J. & COTTON, IV, pp. 647-806, Academic Press, New York. 2. BROWN,D. M. & TODD,A. R. (1952) J. Chem. F. A. (1968) J. Am. Chem. Sor. 90, 6675-6680. F. W. B. & TUCK, D. G. SOC. 52-58, ibid. (1953) 2040-2049; BROWN, 24. BROWN,D. S., EINSTEIN, (1969) Inorg. Chem. 8, 14-18. D. M., DEKKER, C. A. &TODD,A. R. (1952) J. 25. MUETTERTIES, E. L., MAHLER, W. & SCHMUTZLER, Chem. SOC.27 15-272 1. R. (1963) Inorg. Chem. 2,613-618; MUETTERTIES, 3. MARKHAM, R. & SMITH, J. D. (1952) Biochem. J. E. L., MAHLER, W., PACKER, K. J. & SCHMUTZLER, 52, 558-565. R. (1964) ibid. 3, 1298-1303; HOLMES,R. R., 4. ROBERTS,G. C. K., DENNIS,E. A., MEADOWS, CARTER, R. P., JR & PETERSON, G . E. (1964) ibid. D. H., COHEN,J. S. & JARDETZKY, 0. (1969) 3, 1748-1754; GRIFFITHS, J. E., CARTER, R. P., JR. Proc. U S , Natl. Acad. Sci. 62, 1151-1158. & HOLMES,R. R. (1964) J. Chem. Phys. 41,8635. USHER,D. A., RICHARDSON, D. I., JR & ECKSTEIN, 876; HANSEN,K. W. & BARTELL, F. (1970) Nature (Lond.) 228,663-665. L. S. (1965) Inorg. Chem. 4, 1775-1777; BARTELL,L. S. & 6. USHER,D. A.. ERENRICH, E. S. & ECKSTEIN,F. HANSEN,K. W. (1965) ibid., 4, 1777-1782; (1972) Proc. U.S. Natl. Acad. Sci. 69, 115-118. PIERCE,S. B. & CORNWELL, C. D. (1968) J. Chem. 7. DENNIS, E. A. & WESTHEIMER, F. H. (1966) J. Am. Phys. 48, 2118-2120; ADAMS,W. J. & BARTELL, Chem. SOC.88, 3431-3432, 3432-3433. L. S. (1971) J. Mol. Struct. 8, 23-30; YOW,H. & 8. WESTHEIMER, F. H. (1968) Accounts Chem. Res. 1, BARTELL, L. S . (1973) ibid. 15, 209-215. 70-78 and references cited therein. 9. AVEY,H. P., BOLES,M. O., CARLISLE,C. H., 26. RAUK,A., ALLEN,L. C. & MISLOW,K. (1972) J. Am. Chem. SOC.94,3035-3040. EVANS, S. A., MORRIS,S. J., PALMER,R. A., R., HOWELL,J. M. & MUETTERTIES, WOOLHOUSE, B. A. & SHALL,S . (1967) Nature 27. HOFFMANN, E. L. (1972) J. Am. Chem. Sor. 94, 3047-3058. (Lond.)213,557-563. 10. KARTHA,G., BELLO,J. & HARKER,D. (1967) 28. STRICH,A. & VEILLARD, A. (1973) J. Am. Cheni. SOC.95, 5574-5581. Nature (Lond.) 213,862-865. 11. See also DICKERSON, R. E. & GEIS,I. (1969) The 29. VAN DER VOORN,P. C. & DRAGO,R. S. (1966) J . Am. Chem. SOC.88, 3255-3260. Structure and Action of Proteins, pp. 79-81, 30. HOLMFS, R. R. (1974) 168th Meeting American Harper and Row, New York. Chemical Society, Sept. 9-13 INOR Abstracts 65. 12. WYCKOFF, H. W., HARDMAN, K. D., ALLEWELL, N. M., INAGAMI, T., TSERNOGLOW, D., JOHNSON, 31. In terms of electron pair repulsion theory (GILL. N. & RICHARDS, LESPIE, R. (1972) Molecular Geometry, Van F. M. (1967) J. Biol. Chem. Nostrand Reinhold, New York), reduction in 242,3749-3753. electron density of the phosphorus oxygen bonds K. D., ALLEWELL, 13. WYCKOFF, H. W., HARDMAN, caused by hydrogen bond formation should lead N. M., INGAMI, T., JOHNSON, L. N. & RICHARDS, to decreased repulsion between bond pairs and F. M. (1967) J. Biol. Chem. 242,39863988. reduce the slight preference in stability accorded 14. WYCKOFF,H. W., TSERNOGLOW, D., HANSON, the TP relative to the SP. A. W., KNOX,J. R., LEE,B. & RICHARDS, F. M. 32. NEWTON, M. G., COLLIER, J. E. &WOLF,R. (1974) (1970) J. Biol. Chem. 245,305-328. J. Am. Chem. SOC.96,6888-6892. 15. USHER,D. A. (1969) Proc. U S . Natl. Acad. Sci. 33. HOLMES,R. & MEUNIER,P., completed X-ray 62,661-667. analysis of a dioxadiazaspirophosphorane. 16. BERRY,R. S. (1960) J. Chem. Phys. 32,933-938. 17. HOLMES,R. R. (1972) Accounts Chem. Res. 5, 34. DELBENE, J. & POPLE,J. A. (1970) J. Chem. Phys. 4858-4866. 296-303. 35. PAULING,L. (1960) The Nature of the Chemical 18. HOLMES, R. R. (1975) J. Am. Chem. SOC.97,5379Bond, 3rd edn., p. 98, Cornell University Press, 5385. New York. In an ab-initio calculation, Strich & R. & 19. WUNDERLICH, H., MOOTZ,D., SCHMUTZLER, WIEBER,M. (1974) 2. Naturforsch. 29b, 32-34; Veillard (28) calculate a positive charge residing at the phosphorus atom of 2.1 in either the T P or WUNDERLICH, H. (1974) Acta Cryst. B30, 939945; WUNDERLICH, H. & MOOTZ,D. (1974) Acta SP configuration. This gives an average charge oh each fluorine atom of about 0.5 electron suggestCrysr. B30, 935-939.
from the National Institutes of Health (GM 21466), and are gratefully acknowledged.
452
A SQUARE PYRAMIDAL MODEL FOR RIEONUCLEASE ACTION
ing a bond ionicity near 50%. An increase of ionicity of 4% at this level amounts to a change in electronegativity of 0.1 unit according to Pauling’s scale. 36. HOLMES, R. R., COUCH,L. S. & HORA,C. J., JR. (1974) J. Chem. SOC.,Chem. Commun. 175-177. 37. WHITESIDES, G. M. & BUNTING, W. M. (1967) J. Am. Chem. SOC.89,6801-6802. 38. HELLWINKEL, D. (1968) Chemia (Aarau) 22, 488-
44. GIBSON,J. A., IBBOTT,D. G. & JANZEN, A. F. (1973) Can. J. Chem. 51, 3203-3210. 45. HOLMFS,R. R. Spectroscopy and Structure of
Pentacoordinated Phosphorus Compounds with Applications to Reaction Mechanisms, ACS Monograph Series, accepted for publication. 46. WOOD,J. S. (1972) Progr. Inorg. Chem. 16, 227486. E. L. & GUGGENBERGER, L. J. 47. MUETTERTIES, (1974) J. Am. Chem. SOC.96,1748-1756. 48. SUSSMAN,J. L., SEEMAN, N. C., KIM, S. H. & BERMAN, H. M. (1972) J. Mol. Biology 66,403421.
491. D. & WILFINGER,H. J. (1969) 39. HELLWINKEL, TetrahedronLetters 3423-3426. G. M. & MITCHELL, H. L. (1969) J. 40. WHITESIDES, Am. Chem. SOC.91,5384-5386. E. L., MEAKIN, P. & HOFFMAN, R. 41. MUETTERTIES, Address : (1972) J. Am. Chem. SOC.94,5674-5676. S. C. & SCHMUTZLER, R. (1970) J. Chem. R. R. Holmes 42. PEAKE, SOC.(A) 1049-1054; ibid. (1968) Chem. Commun. Department of Chemistry 1662-1 663 ; ibid. (1 968) Chem. Commun. 665-666 University of Massachusetts 43. MUETTERTIES, E. L., BITHER,T. A., FARLOW,Arnherst M. W. & COFFMAN, D. D. (1960) J. Inorg. Nucl. Massachusetts 01002 Chem. 16,52-59. U.S.A.
453
A SQUARE PYRAMIDAL MODEL FOR RIBONUCLEASE ACTION ROBERTR. HOLMES Department of Chemistry, University of Massachusetts Amherst, Massachusetts, US.A. Received 25 August 1975
A mechanism incorporating a square pyramidal model ( S P )is presented for the action
of bovine pancreatic ribonuclease. Its formulation i r based on structural principles governing pentacoordinate behavior. The model is compared with a previous trigonal bipyramidal (TP) representation with regard to the geometry of the active site and enzyme constraints. Of two cariants of the SP model, an adjacent (cis displacement) and in-line (trans displacement)procesy, the in-line mechanism, as with the TP model, fits exivting model studies. Consideration of the energetics ofthe SP us. the TP model leads to an estimated energy difference of’about 1-2 kcallmol. This suggests that the preferred model may be intermediate in geometry between the two idealized representationsfor the enzymatic hydrolysis. Comparisonsare made showing that pseudorotation is an unlikely process in either model.
In order to adequately understand the action of bovine pancreatic ribonuclease (RNase) in catalysing the hydrolysis of the 3’,5’-phosphodiester linkage of RNA (l), it is necessary to acquire an intimate knowledge of the stereochemistry of the active site species. Previous work has accomplished a great deal in achieving this goal. A multitude of chemical studies (1) have shown that the enzymatic hydrolysis is best described as a two-step process, a transphosphorylation in which attack by the 2’-OH group leads to a 2’,3’cyclic phosphate followed by hydrolysis of the latter to yield a terminal 3’-phosphate (2,3). Recent mechanistic proposals (4-6), aided by observations on mechanisms ascribed to non enzymatic hydrolysis of phosphate esters (7,8), have centred on the formation of a trigonal bipyramidal intermediate at the active site (Fig. I). Both Usher et al. (5,6) and Roberts et al. (4) have provided convincing arguments utilizing chemical and structural evidence in outlining the details of RNase action. Their arguments are strengthened by crystal structure determinations
of RNase A (9-11) and of RNase S (11-14) showing the importance of histidine 12, histidine 119, and lysine 41 at the active site (Fig. 2). Both groups of workers (4-6) favor an in-line or linear mechanism, i.e., one in which the entering group does so opposite to the side occupied by the group which leaves in a subsequent stage. This contrasts with the so-called adjacent mechanism (15), wherein the entering group is oriented cis to the potential leaving group. In the latter mechanism, a “pseudorotation” is required to bring the departing group to an apical location in the trigonal bipyramidal intermediate. In so doing, the new trigonal bipyramid is formed by an intramolecular process involving passage through a low lying transition state, presumably having a square pyramidal conformation (16-18). Recently, X-ray determinations (19-21) on oxyphosphoranes containing cyclic substituents have shown that the square pyramid exists as a structurally stable entity. Since the make-up of these derivatives resemble those postulated at the active site in RNase catalysis, it becomes im445
HIROSHI TAKETOMI, YUZO UEDA AND NOBUHIRO G b
transition was discussed. In this paper the specificity is treated, not as a given fact, but as deriving from the heterogeneity of the amino acid sequence of a protein. In order to study the statistical mechanical properties of denaturation and conformational fluctuations of proteins and to understand the mechanism of realization of specificity of native conformations of proteins, we introduce a lattice model of proteins into which heterogeneous aspects in the amino acid sequence are incorporated. In statistical mechanical treatments of polymers in general, inter-unit interactions are conceptually divided into short-range and longrange interactions. Interactions between units that are separated far along the chain are defined to be long-range, even when the units are not widely separated in space. This division of the interactions into short-range and long-range ones is important also in theoretical treatments of protein conformations. Existence of the correlations between short-range amino acid sequence and secondary structures (e.g., a-helix, /?-structure and /?-turn) in globular proteins has been shown recently by many authors. This indicates that the short-range interactions have considerable importance in the process of folding, even though it is apparent that the long-range interactions have their own role in determining and stabilizing the native conformation of proteins. This role of the long-range interaction has not yet bem sufficiently studied theoretically because of the theoretical difficulties intrinsic in the long-range interactions. Differences in amino acids manifest themselves in heterogeneities in both the short-range and long-range interactions. In this paper emphasis is placed on the role of the long-range interactions. The shortrange interactions are not discussed. The heterogeneity of the long range interactions is taken into the lattice model by a very idealized form of the specificity of inter-unit interactions, i.e. attractive interactions are assumed to work for a set of specifically preassigned pairs of units occupying the nearest neighbor lattice points. In section I. the model is described. Even though abstracted and simplified, the model is still far from being amenable to analytical treatment. In this paper the model is studied by the Monte Carlo simulation method of Metropolis et al. (2). The method of Monte Carlo calculation is described in section 11. In section 111, the results of the simulations are
446
presented and discussed. The purpose of abstracting protein molecules as a lattice polymer is to enable otherwise complex phenomena to be studied in as pure and simple a form as possible. Because of the simplification and idealization, some of the conformational properties of proteins become lucidly understandable. However, this is achieved by discarding some of the important aspects of real proteins. The discussion of these aspects is also included in section 111. In section IV brief conclusions and the outlook for further studies are given.
,
I. DESCRIPTION OF THE MODEL
In this paper we are considering a lattice polymer on the two-dimensional square lattice. A study on a lattice polymer on the three-dimensional cubic lattice is now in process and will be presented in a future paper. The polymeric chain on the lattice is assumed to be self-avoiding due to repulsive interactions between two units within the contact distance. The heterogeneity in the amino acid sequence is incorporated as the specificity of inter-unit interactions. Attractive interactions are assumed to work only for a set of preassigned pairs of units occupying the nearest neighbor lattice points. Such pairs of units are defined as being interactable and are indicated by filled (black) squares in Fig. 1. A pair of units remains unfilled (white), i.e. the pair is not interactable, when no interaction works between them even if the pair occupies the nearest neighbor lattice points. Thus Figs. lA, B and C represent the inter-unit interactions with the three different types of specificity, respectively. The distribution of interactable pairs in Fig. 1A is determined as follows. First the native conformation of a protein consisting of 49 units is assumed on the square lattice as in Fig. 2. All pairs of units occupying the nearest neighbor lattice points in the conformation of Fig. 2 are filled black and all other pairs are unfilled in Fig. 1A. The energy of inter-unit interaction is assumed to be identical for all interacting pairs with a value of - 8 in this paper. Therefore the conformational energy of this protein molecule is measured in the unit of ( - 8 ) . Since all neighboring pairs in Fig. 2 are interacting for specificity A specified by Fig. lA, the conformational energy of the conformation of Fig. 2 is - 36.2 for
PROTEIN FOLDING, UNFOLDING AND FLUCTUATIONS
B
1 20 30 40 49 FIGURE 1 Specificity of inter-unit interactions. Attractive interaction is assumed to work when a filled (black or shaded) pair of units occupy the nearest neighbor lattice points, A: strong limit specificity, B: intermediate specificity, C : weak limit specificity.
this specificity of interaction. It is evident that any conformations other than those shown in Fig. 2 (and its mirror image) have conformational energies higher than - 3 6 for this specificity. Therefore for specificity A, the native conformation, i.e. the conformation with the global minimum energy, is that given in Fig. 2. (No distinction will be made between a conformation and its mirror image in this paper.)
In Fig. 1C all 552 pairs of units separated by (2n + 1) bonds (n 1) are filled black, i.e. attractive interaction works nonspecifically for any pair of units occupying the nearest neighbor lattice points. No other pairs of units can occupy the nearest neighbor lattice points due to geometrical reasons. Because no specificity exists, the case of Fig. 1C is reduced to a homogeneous nonintersecting chain polymer with short-range'
447
ROBERT R . HOLMES
identical equatorial groups in the TP. Similarly, in the SP, more charge resides at basal relative to apical ligands. Due to geometrical constraints, four or five membered rings preferentially span apical-equatorial sites in the T P and cis.basal positions in the SP (22). Pi donor ligands achieve better bonding in equatorial positions of the T P and in the apical position in the SP (27-28). These are the main structural differences derived from work with stable pentacoordinate derivatives and provide a complementary set of principles sufficient to structure the SP substrate model. The substrate structure illustrated in Fig. 3 conforms to these constraints. Moreover, entering and leaving groups are expected to utilize basal positions of the SP rather than the lone apical position because of the reduced bond strength likely to be associated with the basal bonds. Either cis or trans orientations of attacking and displaced groups are possible. Binding of the two nucleoside bases on the enzyme, as suggested from a consideration of the enzyme conformation (4), positions the phosphate unit near the active site cleft and allows hydrogen bonding to occur between the nonesterified phosphate oxygen atoms and probably both histidine 119 and lysine 41. This results in the initial enzyme-substrate conformation suggested by Roberts et al. (4) (Fig. la). In the transesterification step (Fig. 3), a trans displacement is indicated (in-line mechanism) with the SP intermediate structured according to the postulates outlined above. One of the free phosphate oxygens is shown at the apical location rather than the nucleoside positioned at the 5’ ester linkage in line with the preference for strong donor groups to locate in the apical position (cf. the corresponding T P with the donor oxygen atoms occupying equatorial sites and the ester linkage located in one of the apical positions, Fig. 1 b). Hydrogen bonding then completes the formulation. In the second step, essentially the reverse of the first step, attack by a water molecule leads to its entry into a basal site of the second SP intermediate (Fig. 3 with a water molecule in place of the 5’ ester linkage). Upon protonation, hydrolysis proceeds leading to the terminal 3’phosphate with decyclization. The stereochemical consequences are the same in either the SP or T P model for both an adjacent
448
or in-line process. However, unlike the adjacent process for the T P model, pseudorotation is not a necessary process in the adjacent mechanism for the SP model, i.e., cis displacement, to complete the transformation since the apical position of the SP is not involved in the mechanistic sequence. In-line versus adjacent mechanisms Excellent work (5,6) aimed at uncovering the geometry of each step in ribonuclease action centered on the use of the two diastereoisomers of 2’,3’-cyclic phosphorothioate[Up(S)l as model systems.
I
I1
The results are consistent with an in-line process proceeding through a TP intermediate (5,6). The results are also consistent with the SP model proceeding in-line by a trans basal displacement. These parallel mechanisms are compared schematically in Fig. 4 for the enzyme catalyzed synthesis of the dinucleoside uridylthio 3’3’phosphorylcytidine Up(S)C from the crystalline isomer of Ufi(S) and cytidine (6). The latter synthesis was designed to model the first step of ribonuclease action. Similar pathways may be constructed to show the same correspondence between the two models for the RNase catalyzed hydrolysis of the crystalline isomer of Ufi(S) in water enriched in ‘*O,and subsequent recyclization of the resulting monoester with diethyl phosphorochloridate. The latter process serves as a model system for the second step of ribonuclease action ( 5 ) . SP versus TP model At first hand it would appear that the T P model might be preferred over the SP model for ribonuclease action. For isolatable five coordinate phosphorus compounds, including both simple
A SQUARE PYRAMIDAL MODEL FOR RIBONUCLEASE ACTION
0 2'
0
0 [/!
3'
1
I ADJ. ROH
1
-
ROH IN-LINE
S
OR
11
INLINE
1
BASE
OR
J t 3ER
/"
pp\ S
OR FIGURE 4a Schematic diagram of model study (6) of the first step of ribonuclcase action proceeding by both in-line and adjacent mechanisms for (a) a TP intermediate and (b) for a SP intermediate (on next page). The nonenzymatic base catalyzed reaction is known to proceed in-line.
I
BASE
'OR
and monocyclic derivatives, the structure invariably found is a TP (22,25). Only recently have examples of SP structures of phosphorus been uncovered and these contain two small-membered ring systems (19-21). A prime example is the oxyphosphorane (19)
It has been estimated (18,30) that ring strain in C
OR
the latter example is relieved by about 2.0 kcat/ mol per ring compared to the related TP structure. It has also been estimated (18) that a SP structure containing five monocoordinated oxygens PO, is approximately 4 kcal/mol less stable than the isomeric TP.If one adds to this, the effect of hydrogen bonding present in the enzymesubstrate environment (Figs. l b and 3), which tends to favor the SP as a consequence of the resultant charge withdrawal (31), the energy balance may be tipped in favor of the SP model. The ring strain factor stabilizing the SP state (Fig. 3) should be somewhat less than the 2.0 kcal/mol suggested above since the latter value applies to an unsaturated planar system. In the
449
ROBERT R . HOLMES
S
S
S
T
1
It S
I S
Fig. 4b.
formation of the 2',3'-cyclic phosphate here, ring puckering provides a means of alleviating ring strain in the T P relative to the SP. While no reliable estimate of this reduction has as yet been made, it is noted that the PO apical bond distance involved in cyclic phosphoranes having TP structures drops from about an average of 1.75 8, for unsaturated ring systems (22) to about 1.71 8, for related saturated five-membered ring systems (32,33). These values compare to 1.67 A for an apical PO bond distance in a TP assumed (18) as a normal value in the absence of ring constraints and a 1.65 A value for the PO bond distance for a five-membered ring containing oxyphosphorane oriented in cis basal positions of a SP (18,19). 450
Thus, the ring strain expected in a 2',3'-cyclic phosphate spanning an eq. apical pair in a TP conformation does indeed appear to be less than that in a related unsaturated ring structure. Again, no quantitative assessment of the stabilization of a SP relative to a TP as a result of hydrogen bonding is at hand. We note, however, that theoretical calculations of DelEene & Pople (34) on the formation of water dimers suggest a shift of about 0.05 electrons from the OH bonds of the donor water molecule. If we use this as an approximation of the reduction in charge density for each of the P-0 bonds involved in hydrogen bonding with the histidine and lysine residues, the phosphorus atom experiences an increased
A SQUARE PYRAMIDAL MODEL FOR RIBONUCLEASE ACTION
positive charge of 0.15. This amounts to a change in electronegativity of approximately 0.1 unit on the Pauling scale (35), about that experienced by inserting an arsenic atom in place of a phosphorus atom. It has been estimated that the SP-TP energy difference drops from -4 to 3 kcal/mol for AsF, compared with PF, (36). This difference should not change appreciably for comparisons between AsO, and PO, moieties. Consideration of the various energy terms given here then suggests that the TP transitional state (Fig. lb) is more stable than the corresponding SP transitional state (Fig. 3) by only 1-2 kcal/ mol. Although extensive refinement is needed, it is felt that the qualitative arguments demonstrate the closeness in energy of the two idealized pentacoordinate structural forms of phosphorus as models for RNase activity. Because of the magnitude of energy barriers for intramolecular ligand exchange deduced from n.m.r. analysis of solutions of pentacoordinate ‘phosphorus derivatives containing either bulky groups (37-39) or substituents undergoing hindered rotation (27,28,40-42) ranging from 9 to 23 kcal/mol, it seems unlikely that the more severe constraints present in the enzyme environment will allow pseudorotation to be a feasible process. For pseudorotation to occur in the enzyme-substrate intermediates shown in Figs. 1b and 3, visualizing a modification of these diagrams to give an adjacent location of the 5‘ ester linkage in the former and a cis orientation of this same linkage in the latter, higher barriers for intra ligand reorientation are expected in these complexes compared with a similar nonenzymatic rearrangement because of the specific binding sites provided by the enzyme. The latter interactions should hinder the pseudorotational process which has been found to be concentration independent (25b,c) and, hence, unassisted by solvent molecules. In fact, interaction of basic solvent systems (4344) with pseudorotating molecules appears to hinder the ligand exchange process (45). Therefore, pseudorotation in a pentacoordinated intermediate involved in RNase action would serve to lower rather than increase the hydrolytic breakdown of nucleic acids. On this basis, an adjacent process for the TP model does not seem to warrant serious consideration. From the foregoing analysis, what appears more likely is the presence of either a five-
coordinate transition state (concerted process) or intermediate formation which bears some distortion between the two idealized representations. Considering the closeness in energy, estimated for the two models, the extent of distortion may be appreciable. In this context, Wood (46) and Muetterties & Guggenberger (47) have shown that distortion between the two idealized states, determined from X-ray studies on a variety of five coordinate species of DBhsymmetry, follows a well-defined path, that of the C2” cons,raint associated with the Berry reaction coordinate for intramolecular exchange (1 6). However, the distortion would not follow the same pathway for the intermediates (or transition states) in ribonuclease action since the initial symmetry of the proposed structure is lower than Dab. More likely, the transitional state distortion will be dictated not only by the closeness in energy of the TP and SP, as argued above, but also by the non covalent interactions provided by the enzyme environment. Furthermore, the phosphate substrate is no doubt distorted to some degree from the tetrahedral stale toward the transitional state as a result of enzyme constraints. This seems to be apparent, at least, for PO bond lengths. Phosphoester PO bond lengths in uridylyl3’,3’-adenosine phosphate (48), for example, average 1.61 A while the phosphoryl PO bond lengths are over 0.1 A shorter, averaging 1.48 A. An overall expansion of bond lengths by approximately 0.2 A is needed to form the T P in Fig. 1b and about 0.25 A to form the SP depicted in Fig. 3 (18). Both the non covalent bonding and hydrogen bonding, via electron withdrawal, are expected to provide some of the required energy to loosen the bonds in the phosphate bound structure in RNase. In view of the above considerations, the preferred model for enzymatic hydrolysis is expected to involve a transition state intermediate in geometry between the two extreme pentacoordinate forms. However, for non enzymatic hydrolysis involving acyclic phosphorus esters or phosphorus esters containing one small-membered ring system, distortion from the TP model should be considerably less. ACKNOWLEDGMENTS
This investigation was supported by grants from the National Science Foundation (MPS 74-1 1496) and 45 1
ROBERT R. HOLMES
20. EISENHUT, M., SCHMUTZLER, R. & SHELDRICK, W. S . (1973) J. Chem. SOC., Chem. Commun. 144-145, and personal communication. 21. HOWARD, J. A., RUSSELL,D. R. & TRIPPETT, S. REFERENCES (1973) J. Chem. SOC.,Chem. Commun. 856-857. R. R. (1974) J. Am. Chem. SOC.96,41431. RICHARDS,F. M. & WYCKOFF,H. W. (1971) in 22. HOLMES, 4149 and references cited therein. The Enzymes (Boyer, P. D., ed.), 3rd edn., vol. 23. BEAUCHAMP, A. L., BENNETT, M. J. & COTTON, IV, pp. 647-806, Academic Press, New York. 2. BROWN,D. M. & TODD,A. R. (1952) J. Chem. F. A. (1968) J. Am. Chem. Sor. 90, 6675-6680. F. W. B. & TUCK, D. G. SOC. 52-58, ibid. (1953) 2040-2049; BROWN, 24. BROWN,D. S., EINSTEIN, (1969) Inorg. Chem. 8, 14-18. D. M., DEKKER, C. A. &TODD,A. R. (1952) J. 25. MUETTERTIES, E. L., MAHLER, W. & SCHMUTZLER, Chem. SOC.27 15-272 1. R. (1963) Inorg. Chem. 2,613-618; MUETTERTIES, 3. MARKHAM, R. & SMITH, J. D. (1952) Biochem. J. E. L., MAHLER, W., PACKER, K. J. & SCHMUTZLER, 52, 558-565. R. (1964) ibid. 3, 1298-1303; HOLMES,R. R., 4. ROBERTS,G. C. K., DENNIS,E. A., MEADOWS, CARTER, R. P., JR & PETERSON, G . E. (1964) ibid. D. H., COHEN,J. S. & JARDETZKY, 0. (1969) 3, 1748-1754; GRIFFITHS, J. E., CARTER, R. P., JR. Proc. U S , Natl. Acad. Sci. 62, 1151-1158. & HOLMES,R. R. (1964) J. Chem. Phys. 41,8635. USHER,D. A., RICHARDSON, D. I., JR & ECKSTEIN, 876; HANSEN,K. W. & BARTELL, F. (1970) Nature (Lond.) 228,663-665. L. S. (1965) Inorg. Chem. 4, 1775-1777; BARTELL,L. S. & 6. USHER,D. A.. ERENRICH, E. S. & ECKSTEIN,F. HANSEN,K. W. (1965) ibid., 4, 1777-1782; (1972) Proc. U.S. Natl. Acad. Sci. 69, 115-118. PIERCE,S. B. & CORNWELL, C. D. (1968) J. Chem. 7. DENNIS, E. A. & WESTHEIMER, F. H. (1966) J. Am. Phys. 48, 2118-2120; ADAMS,W. J. & BARTELL, Chem. SOC.88, 3431-3432, 3432-3433. L. S. (1971) J. Mol. Struct. 8, 23-30; YOW,H. & 8. WESTHEIMER, F. H. (1968) Accounts Chem. Res. 1, BARTELL, L. S . (1973) ibid. 15, 209-215. 70-78 and references cited therein. 9. AVEY,H. P., BOLES,M. O., CARLISLE,C. H., 26. RAUK,A., ALLEN,L. C. & MISLOW,K. (1972) J. Am. Chem. SOC.94,3035-3040. EVANS, S. A., MORRIS,S. J., PALMER,R. A., R., HOWELL,J. M. & MUETTERTIES, WOOLHOUSE, B. A. & SHALL,S . (1967) Nature 27. HOFFMANN, E. L. (1972) J. Am. Chem. Sor. 94, 3047-3058. (Lond.)213,557-563. 10. KARTHA,G., BELLO,J. & HARKER,D. (1967) 28. STRICH,A. & VEILLARD, A. (1973) J. Am. Cheni. SOC.95, 5574-5581. Nature (Lond.) 213,862-865. 11. See also DICKERSON, R. E. & GEIS,I. (1969) The 29. VAN DER VOORN,P. C. & DRAGO,R. S. (1966) J . Am. Chem. SOC.88, 3255-3260. Structure and Action of Proteins, pp. 79-81, 30. HOLMFS, R. R. (1974) 168th Meeting American Harper and Row, New York. Chemical Society, Sept. 9-13 INOR Abstracts 65. 12. WYCKOFF, H. W., HARDMAN, K. D., ALLEWELL, N. M., INAGAMI, T., TSERNOGLOW, D., JOHNSON, 31. In terms of electron pair repulsion theory (GILL. N. & RICHARDS, LESPIE, R. (1972) Molecular Geometry, Van F. M. (1967) J. Biol. Chem. Nostrand Reinhold, New York), reduction in 242,3749-3753. electron density of the phosphorus oxygen bonds K. D., ALLEWELL, 13. WYCKOFF, H. W., HARDMAN, caused by hydrogen bond formation should lead N. M., INGAMI, T., JOHNSON, L. N. & RICHARDS, to decreased repulsion between bond pairs and F. M. (1967) J. Biol. Chem. 242,39863988. reduce the slight preference in stability accorded 14. WYCKOFF,H. W., TSERNOGLOW, D., HANSON, the TP relative to the SP. A. W., KNOX,J. R., LEE,B. & RICHARDS, F. M. 32. NEWTON, M. G., COLLIER, J. E. &WOLF,R. (1974) (1970) J. Biol. Chem. 245,305-328. J. Am. Chem. SOC.96,6888-6892. 15. USHER,D. A. (1969) Proc. U S . Natl. Acad. Sci. 33. HOLMES,R. & MEUNIER,P., completed X-ray 62,661-667. analysis of a dioxadiazaspirophosphorane. 16. BERRY,R. S. (1960) J. Chem. Phys. 32,933-938. 17. HOLMES,R. R. (1972) Accounts Chem. Res. 5, 34. DELBENE, J. & POPLE,J. A. (1970) J. Chem. Phys. 4858-4866. 296-303. 35. PAULING,L. (1960) The Nature of the Chemical 18. HOLMES, R. R. (1975) J. Am. Chem. SOC.97,5379Bond, 3rd edn., p. 98, Cornell University Press, 5385. New York. In an ab-initio calculation, Strich & R. & 19. WUNDERLICH, H., MOOTZ,D., SCHMUTZLER, WIEBER,M. (1974) 2. Naturforsch. 29b, 32-34; Veillard (28) calculate a positive charge residing at the phosphorus atom of 2.1 in either the T P or WUNDERLICH, H. (1974) Acta Cryst. B30, 939945; WUNDERLICH, H. & MOOTZ,D. (1974) Acta SP configuration. This gives an average charge oh each fluorine atom of about 0.5 electron suggestCrysr. B30, 935-939.
from the National Institutes of Health (GM 21466), and are gratefully acknowledged.
452
A SQUARE PYRAMIDAL MODEL FOR RIEONUCLEASE ACTION
ing a bond ionicity near 50%. An increase of ionicity of 4% at this level amounts to a change in electronegativity of 0.1 unit according to Pauling’s scale. 36. HOLMES, R. R., COUCH,L. S. & HORA,C. J., JR. (1974) J. Chem. SOC.,Chem. Commun. 175-177. 37. WHITESIDES, G. M. & BUNTING, W. M. (1967) J. Am. Chem. SOC.89,6801-6802. 38. HELLWINKEL, D. (1968) Chemia (Aarau) 22, 488-
44. GIBSON,J. A., IBBOTT,D. G. & JANZEN, A. F. (1973) Can. J. Chem. 51, 3203-3210. 45. HOLMFS,R. R. Spectroscopy and Structure of
Pentacoordinated Phosphorus Compounds with Applications to Reaction Mechanisms, ACS Monograph Series, accepted for publication. 46. WOOD,J. S. (1972) Progr. Inorg. Chem. 16, 227486. E. L. & GUGGENBERGER, L. J. 47. MUETTERTIES, (1974) J. Am. Chem. SOC.96,1748-1756. 48. SUSSMAN,J. L., SEEMAN, N. C., KIM, S. H. & BERMAN, H. M. (1972) J. Mol. Biology 66,403421.
491. D. & WILFINGER,H. J. (1969) 39. HELLWINKEL, TetrahedronLetters 3423-3426. G. M. & MITCHELL, H. L. (1969) J. 40. WHITESIDES, Am. Chem. SOC.91,5384-5386. E. L., MEAKIN, P. & HOFFMAN, R. 41. MUETTERTIES, Address : (1972) J. Am. Chem. SOC.94,5674-5676. S. C. & SCHMUTZLER, R. (1970) J. Chem. R. R. Holmes 42. PEAKE, SOC.(A) 1049-1054; ibid. (1968) Chem. Commun. Department of Chemistry 1662-1 663 ; ibid. (1 968) Chem. Commun. 665-666 University of Massachusetts 43. MUETTERTIES, E. L., BITHER,T. A., FARLOW,Arnherst M. W. & COFFMAN, D. D. (1960) J. Inorg. Nucl. Massachusetts 01002 Chem. 16,52-59. U.S.A.
453
