Electrostatic field of the large fragment of Escherichia coli DNA polymerase I

J. Mol. Biol. (1985)

186,

645-649

Electrostatic Field of the Large Fragment of Escherichia coli DNA Polymerase I J. Wax-wicker, D. Ollis, F. M. Richards and T. A. Steitz Department of Molecular Biophysics and Biochemistry Yale University, New Haven, Conn. 06511, U.S.A. (Received 17 May 1985) The electrostatic field of the large fragment of Escherichia coli DNA polymerase I (Klenow fragment) has been calculated by the finite difference procedure on a 2 L%grid. The potential field is substantially negative at physiological pH (reflecting the net negative charge at this pH). The largest regions of positive potential are in the deep crevice of the C-terminal domain, which is the proposed binding site for the DNA subst,rate. Within the crevice, the electrostatic potential has a partly helical form. If the DNA is positioned t,o fulfil stereochemical requirements, then the positive potential generally follows the major groove and (to a lesserextent) the negative potential is in the minor groove. Such an arrangement could stabilize DNA configurations related by screw symmetry. The histidine residues of the Klenow fragment give the positive field of the groove a sensitivity to relatively small pH changes around neutrality. We suggest that the histidine residues could change their ionization states in response to DNA binding, and that this effect could contribute to the protein-DNA binding energy. -

1. Introduction

and Methods

The structure of the large (Klenow) fragment of DNA polymerase I (Pol I) from Escherichia coli has recently been solved at 3.3 b resolution (Ollis et al., 1985a). Pol I is involved in the repair of damaged duplex DNA and in the processing of Okazaki fragments (Kornberg, 1980). The Klenow fragment of Pol I has both DNA polymerase and 3’ to 5’ exonuclease activities, and a molecular weight of 68,000. Within the Klenow fragment the protein folds into an N-terminal domain of about 200 residues, and a C-terminal domain with about 400 residues. Substrate binding studies with the crystals suggest that the 3’ to 5’ exonuclease site is on the N-terminal domain, whilst the shape of the C-terminal domain appears to be suitable for binding DNA. It has been found that a DNA duplex can be model-built into the crevice of the C-terminal domain so that it lies along the base of the groove, whilst packing neatly against the protein that forms the sides of the groove (Ollis et al., 1985a). Figure 1 shows two views of the alphacarbon backbone of the Klenow fragment and the model-built DNA. Translation of the DNA (along its helix axis) in this conformation is restricted by the intrusion of a two-helix element into the major groove. This location of the DNA on the protein is supported by the existence of Pol I mutations in the cleft region that reduce DNA binding (Kelly & 0022-2836/S5/230645-05 22

l

$03.00/O

645

Grindley, 1976) and by sequence homology around the cleft to phage T7 DNA polymerase (Ollis et al., 19858). Electrostatic calculations were performed on the partially refined Klenow structure, which included data to 2.8 a resolution (Ollis, unpublished results). There are two regions of polypeptide missing due to disorder, with about 50 residues in all. One of these regions carries a net charge of +2, whilst the other has a net charge of -2. These missing polypeptide segments and charges may play a role in DNA binding. Since disorder will smear out the field and since the sum of missing charges is zero, we do not expect the omission of these charges from the calculation to affect the general form of the electrostatic field of the native Klenow fragment. Calculations were made with the finite difference method (Warwicker & Watson, 1982; Warwicker, unpublished results). In this method, the irregular boundary between low dielectric protein and high dielectric solvent is constructed on a Cartesian grid, and the potential field is calculated on a related grid. The boundary treatment of this method makes it suitable for studying a molecule such as the Klenow fragment, which has a large crevice. Dielectric values of 3 for protein and 80 for solvent were incorporated onto the grid. An ionic strength of 0.1 M was used, and the pH value was varied between 6.5 and 7.5. The finite difference method gives long-range electrostatic fields but does not 0

1985

Academic

Press

Inc.

(London)

Ltd.

646

J. Warwicker et’ al.

Figure 1. Stereoscopic views of the Klenow fragment alpha-carbon backbone and the model-built DNA. (a) The view is along the proposed DNA binding groove, which is formed by the C-terminal domain in the upper part of the Figure: with the N-terminal domain in the lower part. (b) The view is onto the length of the groove, with the C-terminal domain forming most of the upper part of the diagram, and the N-terminal domain below this. The C-C distance in the backbone of 4 A gives the scale of the diagram. This Figure was produced by the program ARPLOT (Lesk & Hardman, 1982). have sufficient accuracy at short range to give the deviation from standard pK values that the protein charges

effect

on each other.

The

sets of charges

used for the field calculations were therefore derived from a separate procedure that was able to give pK deviation effects. In this procedure charge-charge with an empirical interactions were iterated formula, which used an effective dielectric value of 50, and a Debye-Huckel

counterion

screening

term

at an ionic strength of O-1 M. It has been observed (Warwicker, unpublished results) that this form of interaction gives as good agreement to the

experimental inhibitor

pK data for bovine pancreatic trypsin

as

does

the

frequently

used

solvent

accessibility-modified Tanford-Kirkwood model (see e.g. Matthew & Richards, 1982). An Evans and Sutherland PS300 colour graphics system and the computer program PROD0 (Jones, 1978) were used to visualize the fields. 2. Results and Discussion Figure 2 shows the overall electrostatic potential field at pH 7.0. The net charge at this pH is around

Electrostatic

Field of the Klenow

Fragment

647

Figure 2. A stereoscopic view of the Klenow fragment and its electrostatic potential field at pH 7.0, in the same as in Fig. l(a). The alpha-carbon backbone is yellow, the potential contoured at +kT/e is blue, and at orientation -kT/e is red. DNA has been omitted from this Figure. The field calculation was performed on a 2 A grid, at an ionic strength of 0.1 M. -8 e, and this is reflected in the dominance of the negative potential contours, particularly in the N-terminal domain. The view along the proposed DNA binding groove clearly shows the predominance of positive potential in this region. It spreads across between the protein projections that

form the sides of the groove. Large areas on the outside of the projections are covered with negative potential. The presence of the most extensive positive potential region in the C-terminal groove is strong corroboration that this is the DNA binding location. The lack of complementarity to DNA over

Figure 3. A stereoscopic view in the same orientation as Fig. l(b), showing the potential in the vicinity of the modelbuilt DKA. The DNA is green, the potential at + kT/2e is blue, and at - kT/Be is red. The protein has been omitted from the Figure.

648

J. Warwicker

much of the surface away from the interior of the groove could act to inhibit the formation of nonproductive protein-DNA complexes. In Figure 3 the model-built duplex DNA is shown with the protein potential, but without the protein backbone, in the orientation of Figure l(b). Only the protein potential field in the immediate vicinity of the DNA is shown. It can be seen that the positive potential has three large envelopes, all in the major groove. These envelopes are connected to form a spiral along the major groove. The negative potential shows four relatively small incursions into the DNA duplex. The larger two of these occur in the minor groove between the regions of positive potential that spiral in the major groove. There are two small negative potential incursions into the major groove. Three sections of the phosphate backbone lie between the positive potential (major groove) and the negative potential (minor groove). If the DNA were translated along its helix axis, then some regions of phosphate backbone would enter into the negative potential of the protein. There is an additional patch of negative potential through which phosphates would move if the DNA were translated away from the N-terminal domain. The gradient of the protein potential will be large at boundaries between positive and negative patches. Those parts of the phosphate backbone that lie in such regions will be subject to large forces from the protein field. Since translation either way will take one of the DNA strands into the minor groove negative potential, we can surmise that the modelbuilt DNA is in an equilibrium position with respect to this translation. A shift from this position would give rise to strong restoring forces. Ollis et al. (1985a) suggested that the incursion of the J/K helix pair into the major groove could cause the DNA to move in a restricted fashion, in much the same way as a screw on a thread. The evidence reported here supports this idea, since proteinDNA conformations with electrostatic complementarity will be related to screw symmetry. The helical form of the Klenow fragment potential contrasts with the more general positive field of catabolite activator protein in the proposed DNA binding region (Weber & Steitz, 1984; Warwicker, unpublished results). Although these two proteins have very different functions, it is likely that they both have the ability to move along a DNA molecule (Berg et al., 1981). A comparison of the electrostatics of catabolite activator protein, which searches for its operator sequence, and Klenow , which processes successive base-pair sites, will be instructive. The variation of the electrostatic field with pH is shown in Figure 4 on two-dimensional contour plots. In Figure 4(a) the field was calculated at a pH of 7.5, whilst in Figure 4(b) a pH of 6.5 was used. A section through the C-terminal groove is shown, and the generally negative surface of the Klenow fragment is immediately obvious. The major change between the two pH plots is the substantial increase of positive potential in the groove at the lower pH.

et al.

(b) Figure 4. The potential on

a

2-dimensionalsection

through the grooved region of the Klenow fragment is shown. (a) The potential field was calculated at a pH of 7.5; (b) calculated at a pH of 65. The molecular boundary is drawn with a heavy continuous line, positive potential contours with a thin continuous line, and negative potential contours with broken lines. Contour levels are drawn from -ZkT/e to +2kT/e, at increments of kTj2e. The groove opens out toward the top right-hand corner of the section.

The spread of the potential between the two sides of the groove is similar to the form of potential contours observed in other enzyme cleft systems treated by the finite difference method (Warwicker

Electrostatic

Field of the Klenow

& Watson, 1982; Warwicker, unpublished results). If the field is calculated using the same empirical formula that is used for the charge-charge interactions (and pK deviations), then the characteristic potential form across the groove is not so clear. The potential contours from the empirical formula retain a spherical form around the charge centres. The finite difference type of field, in which relatively high potentials are maintained in cleft solvent regions, is the result of the dielectric boundary conditions acting on the geometry of the cleft systems. Since many catalytic sites have this geometry, it is clear that the effect could have a general physiological significance. The groups responsible for the potential change with pH illustrated in Figure 4 are the 16 histidine residues of the Klenow fragment. These histidines are distributed throughout the molecule, and they pick up about six protons as the pH changes from 7.5 to 6.5. Since they are able to exert a relatively large effect on the groove region, whilst not clustering there, the calculations lead us to suggest that the potential in the groove results from a delicate charge balance over the whole molecule. The glutamic and aspartic acid residues and the lysines and arginines contribute to this field balance, but not to its pH sensitivity around neutrality, leaving the net field extremely sensitive to the smaller number of histidines. The binding of a DNA molecule to the Klenow fragment will almost certainly represent a large pertubation to the electrostatics of this protein, in the same way that the groove is sensitive to the charge balance of the native protein. Preliminary calculations of Klenow and Klenow-DNA electrostatic energies, using the empirical interaction formula, indicate that binding DNA will cause an uptake of protons by the histidine residues, as expected from the addition of substantial negative charge. This gives rise to an additional attractive term in the complexation energy. Dissociation is rendered more difficult, while movement along the DNA, within the helical potential constraints, is iso-energetic. These calculations are made in the presence of a counterion cloud, and it is the macro-ion nature of the DNA that causes the change in the histidine ionization state. A complexation energy component

649

Fragment

of this type could aid general recognition of DNA over smaller anionic species. We are currently investigating the detailed structure of the electrostatic field of the Klenow fragment, in conjunction with biochemical data that indicate the regions of interaction between protein and DNA (C. Joyce, unpublished results). Calculations of electrostatic energies are only possible with empirical formulae at present, due to the limited resolution of the finite difference grid. The presence of the groove in the Klenow fragment makes it highly desirable to have energies derived from the actual protein (and DNA) geometries. We are working on a finite element method of field calculation that uses a grid matched precisely to the solvent boundaries. This method should allow a more accurate calculation of complex energies, and enable comparison with experimental binding data for the protein-DNA complexes. We thank M. Bannon and J. L. Mouning for their help in the preparation of the manuscript, and A. Perlo for system programming on the VAX 11/750 and PS300. This work was supported by grant POl-GM-22778 from the National Institute of General Medical Sciences. and grant NP-421 from the American Cancer Society.

References Berg, 0. G., Winter, R. B. & von Hippel, P. H. (1981). Biochemistry, 20, 6929-6948. Jones, A. T. (1978). J. Appl. Crystallogr. 11, 268-272. Kelly, W. S., & Grindley, N. D. F. (1976). ArucZ. Acids Res. 3, 2971-2984. Kornberg, A. (1980). DNA Replication, W. H. Freeman and Co., San Francisco. Lesk, A. M. BEHardman, K. D. (1982). Science, 216, 539540. Matthew, J. B. & Richards, F. M. (1982). Biochemistry, 21) 4989-4998. Ollis, D., Brick, P., Hamlin, R., Xuong, N. G. & Steitz, T. A. (1985a). Nature (London), 313, 762-766. Ollis, D., Kline, C. & Steitz, T. A. (19853). Nature (London), 313, 818-819. Warwicker, J. & Watson, H. C. (1982). J. Mol. Biol. 157, 671-679. Weber, I. T. & Steitz, T. A. (1984). Proc. Nat. Acud. Sci., U.S.A. 81, 3973-3977.

Edited by A. Klug