Crystal structure disposition of thymidylic acid tetramer in complex with ribonuclease A.

The crystal structure of ribonuclease A with bound thymidylic acid tetramer is reported at 2.5-A resolution. The diffusion of the tetramer into native orthorhombic crystals of the ribonuclease allows for the formation of a structurally stable complex where the sin-gle-stranded nucleic acid enters and leaves the enzyme’s catalytic region in a persistent 5’-3’ direc-tion. The binding of the tetramer to the enzyme’s sur- face is facilitated and mediated by electrostatic interactions between basic protein residues and nucleotide phosphates. Two pyrimidine nucleotides are bound to the enzyme’s active site in a manner similar to that observed for other complexes between ribonuclease A and nucleic acid oligomers. The x-ray structure of a complex of ribonuclease A (RNase A)’ with the single-stranded DNA oligomer d(pT)4 is de-scribed in this paper. This study extends the list of crystallographically studied complexes between RNase and nucleic acids by the inclusion of a complex with a pyrimidine oligomer larger than a dinucleotide and further illustrates the role of a pyrimidine nucleotide bound at the active site for RNase catalysis.

the stereochemistry of catalysis (Usher et al., 1970(Usher et al., , 1972 and from complexes of the enzyme with inhibitors (Richards and Wyckoff, 1971;Wodak et al., 1977;Pavlovsky et al., 1978). The side chains of His-12 and His-119 mediate proton transfers with the 2'-oxygen of the pyrimidine ribose to form and resolve a 2',3'-cyclic nucleotide intermediate, and Lys-41 is thought to stabilize a pentacoordinate phosphorus intermediate involving the 2'-, 3'-, and (vicinal) 5"oxygens with the two phosphate oxygens.
The RNase sites for binding of the pyrimidine ribose and base and the positioning of the phosphate (involved in the hydrolysis) 3' to the pyrimidine nucleotide have been well characterized within the enzyme's active site cavity (Richards and Wyckoff, 1971;Borkakoti, 1983). A purine nucleotide, in particular adenosine, preferably contributes its 5"phosphate for hydrolysis. The chemical, physical, and structural properties of RNase A and RNase S (formed by enzymatic cleavage of the peptide bond between residues Ala-20 and Ser-21 of RNase A) have also been extensively reviewed (Anfinsen and White, 1961;Richards and Wyckoff, 1971;Blackburn and Moore, 1982;Wlodawer, 1984).
Numerous studies of three-dimensional structures from xray analyses of crystals of the native protein and of complexes between the protein and single-stranded mono-, di-, and oligonucleotides and various nucleotide inhibitors have been reported (see review by Wlodawer (1985) and references cited therein; McPherson et al., 1986aMcPherson et al., , 1986bCampbell and Petsko, 1987;Wlodawer et al., 1988;Nachman et al., 1990). The enzyme cannot cleave DNA due to the absence of 2'-hydroxyl groups on the nucleotides. However, RNase can bind to DNA with an affinity comparable to that for RNA (Sekine et al., 1969;Walz, 1971), and denatured DNA acts as a competitive inhibitor for the enzyme's hydrolysis of substrate analogues and RNA itself (Sekine et al., 1969). Therefore, it has been concluded that single-stranded DNA forms a stable complex with RNase and probably binds to the enzyme at the active site by mechanisms similar to RNA binding (McPherson et al., 1986b).
It is of interest to determine the structural characteristics of the binding of polymeric and oligomeric single-stranded DNAs to RNase as a means of inferring methods of RNase interactions with, and catalysis on, polymeric RNA substrates. Such studies help to elucidate the nature (efficiency, stability, stereospecificities, etc.) of protein and enzyme interactions with nucleic acids of nonspecific base sequence. The present study involving the association of RNase A with a pyrimidine oligomer offers further insights into nonspecific protein/nucleic acid interactions and enzyme catalysis.

EXPERIMENTAL PROCEDURES
Crystallizations and Complex Formation-Native RNase A crystals were grown in unbuffered polyethylene glycol 4000 (PEG 4000) SO-lutions using vapor diffusion methods  as follows.
Bovine RNase A (Sigma) was dissolved in distilled water to a concentration of 60 mg/ml with no further purification. Crystallizations were carried out using siliconized 9-well depression plates sealed in plastic boxes and stored for equilibration at 4 "C. Reservoir solutions (25 ml) consisted of 28% PEG 4000 and 0.02% azide. Each well sample contained 7 pl of RNase A solution combined with 3 pl of 5% 8-octyl glucoside (Sigma), a nonionic detergent, and 7 pl of reservoir solution. Native protein crystals appeared within 2 weeks and were allowed to grow for about 4 months. They were judged to be typical of the orthorhombic prismatic habit seen for co-crystals of the RNase A.d(pA)Q complex . X-ray precession photographs determined that they belonged to the space group P2]2]2]. Reflections from three native crystals were used to solve the protein structure. The crystal yielding the most extensive set of reflec$ions had unit cell lengths of a = 44.9 A, b = 74.9 A, and c = 44.0 A with 1 molecule/asymmetric unit.
Particular care was made during nucleic acid diffusion attempts so as not to perturb the physical structure of the native protein crystals. A low mole ratio of d(pT)4 tetramer to RNase (2:9) was eventually employed. To form the RNase A.d(pT)4 complex within the crystals the tetramer (McPherson et al., 1986~) was dissolved in 27% PEG 4000 and adjusted to a pH of 7.0 at a concentration of 1.0 pg/pl. The d(pT)4 solution was added to wells containing native crystals in four steps of 2 pl each over 5 days. The complex was allowed to form for over 1 month at 4 "C before crystals were selected tor data collection. A single crystal with unit cell lengths of a = 44.9 A, b = 75.0 A, and c = 43.6 A yielded the most extensive data set and was used for the structure determination and refinement of the RNase A.d(pT)4 complex.
Data Collection-An Enraf Nonius CAD-4 automated diffractometer with an extended counter arm and helium-filled path was used for all data collection. The x-ray generator was operated at 40 kV and 40 mA with a fine focus x-ray tube and omega scan rate of 1.33"/ min. Data were collected at 16 "C. For each data set 32 phi independent standard reflections were used to generate an empirical absorption correction curve applied to all reflections. Two standard reflections, measured at intervals of about 2 h of x-ray exposure, were used to generate a time-dependent crystal decay curve which was applied to all data. Net intensities were determined by subtracting from the measured peak intensity twice the sum of the 16 left and 16 right nearest neighbors in reciprocal space.
Crystals of the native protein yielded a collective data set of 4466 independent reflections to 2.5-A resolution at better than 2u. A crystal for the RNase A. d(pT 4 complex was used to collect 4347 independent reflections to 2.5 1 at better than 2u. The integrated intensities for both data sets were corrected for Lorentz-polarization effects and absorption; backgrounds were subtracted. Reflection intensities were converted to structure factors for use in refinement programs. Structure Solutions and Refinements-A model for the RNase A structure, based on a combination of x-ray diffraction and neutron diffraction data (Wlodawer and Hendrickson, 1982;) from crystals not grown in the presence of p-octyl glucoside, had previously been used as a molecular replacement model to position the RNase A molecule in the P212121 unit cells of native protein crystals with the typical prismatic habit seen of RNase A crystals grown in PEG 4000 . This model, with its positioning in the orthorhombic unit cell, was initially used to solve the structure of the enzyme using reflection data from native crystals of the present study. A refinement summary is given in Table I. The general strategy involved initial least-squares rigid body and atomic refinements under CORELS (Sussman, 1985) folunder X-PLOR (Brunger, 1988). X-PLOR refinements for the com-lowed by energy minimizations and simulated annealing refinements plex typically resulted in a loss of perfect thymine base planarity, so they were followed by replacement of bases with planar ones and further CORELS positional refinements leading to the final structure. Refinements were interspersed with calculations of difference Fourier maps using the FFT algorithm (TenEyck et al., 1976), and model fitting under the molecular modeling programs Frodo (Jones, 1982) and Quanta (Polygen Corp., Waltham, MA), as implemented on Evans and Sutherland PS330 and PS390 and Silicon Graphics Iris graphics stations. A refined structure of the native protein (in p-octyl glucoside solvent) was then used to determine the location of nucleic acid for the RNase A.d(pT)Q complex, followed by tetramer fitting in difference Fourier maps and complex refinement as outlined above. Rotation function and translation function searches under X-PLOR (Brunger, 1990), using the refined structure for RNase A (in P-octyl   3 Refinement of RNase A. d(pT)4 complex and generation of Fo-Fc omit map using protein phases only, followed by refitting of the DNA and further refinement. The Fo-Fc omit map yielded the first clear evidence of active site density from the tetramer. R factor is 0.258 at 3.0-A resolution.
4 Generation of Fo-Fc omit map and tetramer refitting, followed by further refinement at an extended resolution range. R factor is 0.249 at 2.7-A resolution.
5 After more tetramer refitting, the resolution range of CORELS refinements is extended to 2.5 A. The refined model is used to initiate rounds of X-PLOR energy minimization and simulated annealing refinements followed by replacement of bases for perfectly planar ones and further CORELS refinements, resulting in much cleaner difference density maps. R factor is 0.239 at 2.5-A resolution.
6 Generation of Fo-Fc omit map, tetramer refitting, and further CORELS refinements to the final structure. The final Fo-Fc map is very sparse and clean with tetramer density clearly seen. 2Fo-Fc maps for the final model reveal density from a P-octyl glucoside molecule associated with the protein. R factor is 0.235 at 2.5-A resolution.
glucoside solvent) as the molecular replacement model, were done to confirm the correct positioning of the protein in the native and complex unit cells (Table I). Table I1 gives a summary of the refinement stages for the RNase A.d(pT)4 complex. Throughout the text the individual nucleotides of the d(pT)4 tetramer are referred to by using the following notation convention: 5'-Tl-T2-T3-T4-3'.

RESULTS AND DISCUSSION
Electron Density Fitting-The final Fo-Fc difference Fourier "omit maps" (based on phases from the protein refined alone) for the RNase A. d(pT)4 complex were very clean and sparse and showed densities which could be associated with only 1 molecule of d(pT)4 (Fig. 1). No attempts were made to fit nucleic acid moieties from possible oligomers with low occupancies. Electron densities were particularly clear near and within the enzyme's active site cavity and near catalytically important protein residues. The nucleic acid moieties of the 3'-nucleotide, including the phosphate at the P1 site, were well fitted to difference densities for the final structure.
Positive electron densities from F,-F, maps for most of the d(pT)4 tetramer were persistent features. The T2 nucleotide may be partially disordered since some deoxyribose and base atoms show unclear or no 2F0-Fc density. The T3 phosphate is associated with weak F,-F, "omit map" density and a high temperature factor (Table 111), but its position is supported by the stereochemistry of the tetramer and its density is evident in 2F,,-Fc maps from phases for the final model (Fig.  2 ) .
2F,,-Fc maps based on phases from the refined complex reveal density resulting from a p-octyl glucoside molecule within the asymmetric unit near the side chains of protein residues Gln-69, Asn-71, and Glu-111. Although the detergent molecule was not used in refinements nor for R factor and electron density map calculations, density for the D-glucopyranoside moiety clearly shows not only the general contours of the pyranoside but also the location of the terminal C6-06 group. No density, however, can be discerned for the anomeric linkage and the alkyl group, and the density for the sugar is much attenuated in the more stringent Fo-Fc omit maps. A glucopyranoside molecule, taken from the x-ray crystal structure of 1-decyl-a-D-glucopyranoside (Moews and Knox, 1976) with the 01 (anomeric) oxygen placed in a p orientation, can be convincingly fitted (as a rigid body) into the density (Fig.  3). The original a linkage would direct the alkyl group away from the protein; however, a / 3 linkage directs the alkyl group along the side of the protein surface to facilitate intermolecular hydrophobic interactions. The proximity of the detergent molecule may affect the orientations of nearby protein residues and short contacts seen between these residues and the 3"terminal thymine base, since a manual fi!ting of the glucopyranoside places its 0 6 atom around 3 A from both the NE2 nitrogen of Gln-69 and the OD1 oxygen of Asn-71. Similar difference density is not seen for the native RNase A structure (in p-octyl glucoside solvent), suggesting that the diffusion of d(pT)4 into native crystals may have aided the trapping of the detergent molecule in the crystalline complex.
Protein Structure-The inclusion of 8-octyl glucoside in the crystallization medium for RNase A allows for the diffusion of single-stranded DNA oligomers larger than dinucleotides into native crystals without conformational disruption or denaturation of the protein molecule. Major conformational changes involving the polypeptide backbone for the RNase A molecule in the protein-nucleic acid complex were avoided and none were necessary during the refinements. Backbone conformations for the refined native and complex protein structures were virtually identical and only slight changes in the positioning of side chain atoms were found. The average rms differences between the proteins of the refined native and complex structuTes were for the backbone and side chain atoms 0.48 and 1.2 A, respectively. Noticeable changes in the orientations of some amino acid side chains, especially those involved in tetramer binding, were observed. Very few protein-protein intermolecular contacts were found for the native and complex crystal structures.
The fractionai unit cell volume occupied by protein in the native RNase A crystals of the present study, 0.433, results in an opera11 volume to mass ratio, V, (Matthews, 1968 (Matthews, 1968). Therefore, p-octyl glucoside allows for the growth of RNase A crystals with a greater fractional volume of solvent than otherwise. Protein residues Lys-41, His-12, and His-119 appear most crucial for the configuration of the nucleotides within the active site region of the complex. The side chains of Lys-41 and His-12 moved little during refinement of their positions in the native structure, a lack of repositioning also reported for the RNase A-d(pA)4 complex (McPherson et d., 198613). A noticeable torsional adjustment of the imidazole ring of residue His-119 from the present study was observed. Its side chain, however, remains in the minor position for His-119 (Borkakoti, 1983).
Nucleic Acid Structure-There are some structural similarities between the nucleic acid of the present RNase A. d(pT)4 complex and that for the co-crystalline complex of RNase A with four d(pA)4 oligomers (McPherson et al., 1986a). Each of the four deoxyribose sugars refine to within the C3"endo conformational family of sugar puckers (see Table 111) ac-

FIG. 1. Stereoview of the Fo-Fc difference electron density map at the d(pT)4 region of the complex.
Bias in the map was minimized by 50 cycles of positional refinement for the protein without phases from the tetranucleotide. The omit *map is contoured to the 3a level at 3.0 A. Some amino acid residues involved with nucleic acid binding are also shown. They are (running clockwise from the top) Arg-39, Lys-66, His-119,  cording to calculated pseudorotation angles. The C2'-exo sugar pucker observed for two of the nucleotides is not unexpected. A C2'-exo deoxyribose conformation was found in the crystal structure of a deoxythymidine covalently bound to RNase A at residue His-12 (Nachman et al., 1990).
Group temperature factors for the DNA moieties (phosphate, deoxyribose, base) of the d(pT)4 tetramer compare favorably to those of the most firmly bound tetramers in the co-crystalline RNase A.d(pA)4 complex (McPherson et al.,  1986b). In particular, the thermal parameters of the 3'-D1-C4-C3-C2-5'-phosphates of the co-crystalline complex are similar to the B factors for the 3'-T4-T3-T2-T1-5'-phosphates, with the T3 phosphate B factor (like that for the C4 phosphate) being the highest of the four. As suggested for the RNase A. d(pA)4 complex, the thermal parameters associated with the d(pT)4 tetramer probably reflect statistical disorder for the nucleotides. Although the crystalline complex was formed under physiological conditions, many of the proteinnucleic acid interactions appear electrostatic in nature and are also similar to those observed in crystals of low ionic strength for complexes between RNase and single-stranded nucleic acid oligomers (Wodak et al., 1977;McPherson et al., 1986b). Each of the thymine bases of the d(pT)4 tetramer have different plane normals and there is no evidence of base stacking, in contrast to the adenosine bases of a d(pA)4 oligomer bound at the active site for the RNase A.d(pA)4 complex (McPherson et al., 1986b). The thymidine tetramer is associated with the RNase A molecule in a partially extended manner as similarly defined by Webb et al. (1973). Nearest neighbor distances between C2 base atoms range from hDetermined from torsion angle about glycosyl bond, defined as ~:01"Cl'-Nl-C2.
8.9 A to 10.1 A. There is no stacking of the T3 base to the phenyl moiety of the Phe-120 side chain. The tetramer can be divided into essentially two regions, an extended dinucleotide bound at the active site (T3-T4) and a more contracted dinucleotide (Tl-T2).
General Protein-DNA Interactions-The path of the d(pT)4 tetramer across the protein surface is roughly similar to that followed by nucleotides C2 to C4 and continued by the 5'phosphate of tetramer D in the RNase A d(pA)4 complex. In both cases the majority of intermolecular contacts involve binding of single-stranded DNA oligomers to a single protein molecule in the asymmetric unit. The d(pT)4 tetramer runs from the 5' terminus into the RNase A active site cleft with the next to last nucleotide (T3) and the 3'-terminal phosphate adjacent to catalytically important amino acid residues (Fig.  4). The 3'-end of the d(pT)4 tetramer is anchored to the protein with the T4 phosphate occupying the P1-binding site (Borkakoti, 1983). The T4 thymine base and deoxyribose occupy the respective B2 and R2 sites of the enzyme (Wyckoff et al., 1970). The ribose and base of the T3 nucleotide are found at the corresponding pyrimidine R1-and B1-binding sites (Borkakoti, 1983). Both nucleotides from the 3'-end are involved in intermolecular contacts with side chain from amino acid residues at the enzyme's active site.
The disposition of the d(pT)4 tetramer probably reflects the complementarity of positive charges and nucleoside-binding sites on the surface of the protein with negatively charged phosphate groups and bases, as was concluded for the RNase A.d(pA)4 complex (McPherson et al., 1986a). These electrostatic interactions are of primary importance for complex formation since similar complexes form between RNase and heteronucleotides devoid of heterocyclic bases (Karpel et al., 1981). The persistent 5'-3' orientation of the tetramer is common for many dinucleoside monophosphate crystalline complexes involving RNase A and RNase S molecules (Richards and Wyckoff, 1973;Wodak et al., 1977;Wlodawer, 1984), and would be as required for catalysis on RNA substrates (Richards and Wyckoff, 1971;Blackburn and Moore, 1982;Wlodawer, 1984).
Many direct interactions between d(pT)4 and RNase A involve the nucleic acid backbone. The location of the d(pT)4 tetramer supports, in general, the positioning of a virtually continuous strand of eight nucleotides suggested by two d(pA)4 oligomers bound at the active site cleft and emerging from near the pyrimidine-binding site, as postulated by McPherson et al. (1986a). In this way a "virtual DNA strand" can be intramolecularly associated with the protein as the general nature of single-stranded nucleic acid binding to RNase A. The influence of the enzyme in the binding of the nucleic acid extends throughout the length of the tetramer, particularly at the active site region, and directly for at least three of the four nucleotides of the tetramer. The T2 phosphate is most distant from basic protein residues within the complex, and the T2 nucleotide is the most awkwardly situated unit of the tetramer with its thymine base adopting a slightly syn conformation (x = -35") to its deoxyribose.
Specific Protein-DNA Interactions-The extension within the d(pT)4 tetramer allows the 5"phosphate to form a salt bridge with the guanidinium group of Arg-39, similar to that formed by the C1 nucleotide's phosphate in the RNase A. d(pA)4 complex (McPherson et al., 1986b), even though the d(pT)4 tetramer runs into and out of the active site. The C5 methyl group of the 5' base lies less than 3.5 A from the NH2 nitrogen of Arg-39. All 3 enzyme residues associated with catalytic activity (Richards and Wyckoff, 1971) are involved in the binding of the thymidine tetramer ( Table IV). The oxygens of the phosphate bound at the P1 site have intermolecular contacts to the NE2 nitrogen of protein residue His-12 and the NZ nitrogen of residue Lys-41. The His-119 residue influences the positioning of the P1 site phosphate by forming electrostatic contacts with the 03'-and 05'-oxygens of nucleotides T3 and T4, respectively.
The 01'-atom of the deoxyribose bound at the R1 site (T3) is near the NZ nitrogen of Lys-41, and the 05'-oxygen from this nucleotide may also be involved in a contact with the Lys-41 side chain. The 0 2 atom from the thymine base bound at the B1 site (T3) is close to the YE2 nitrogen of His-12.
The NZ nitrogen of Lys-66 lies 5.4 A from th? 0 4 oxygen of the T3 nucleotide's base, and less than 4 A from the C5 methyl group of the same base. The 03"oxygen of nucleotide T3 is associated with the (T4) phosphate bound at the P1 site and is in a position similar to the 08'-oxygen from the covalently bound thymidine nucleotide in the T-H12 + RNase A crystalline complex (NachTan et al., 1990). This 03'-atom (present complex) is under 3 A from the ND1 nitrogen of His-119 within the complex. Apparently, the T3 nucleoside has sufficient mobility to allow positioning of the (T3) thymine base into the B1 site, as opposed to the T-Hl2. RNase A complex.
The NE2 nitrogen of Glu-111 is located within hydrogen bonding distance from one of the phosphate oxygens of the T 4 nucleotide, similar to that observed for some RNase Anucleotide inhibitor complexes (Borkakoti, 1983). Contacts of under 3.6 A between the phosphate oxygens of the 3"nucleotide, the OEl atom from the side chain of Gln-11, and the backbone nitrogen of residue Phe-120 are observed, similar to the crystalline complex of RNase S with cytidilyl (2', 5') adenosine (Wodak et al., 1977).
The side chain of protein residue Asn-67 is involved in contacts with the 3"deoxythymidine (T4) similar to those found for the covalently attached thymidine of the T-Hl2. RNase A complex. The distances between the 0 4 oxygen of the T4 thymine base to both the ND2 nitroJen of Asn-67 and the NE3 nitrogen of Gln-69 are about 5 A. The C5 methyl group of the T4 base isovery close to the ND2 atom of Asn-67 and is less than 4.5 A from the OE2 oxygen of the same protein residue. Also similar to the T-Hl2. RNase A complex, the (T4) 04 oxygen may be hydrogen bonded to the side chain of residue Asn-71. The side chain of His-119 may stabilize the location of the T4 base, the 0 2 oxygen of the this base lying relatively close to the ND1 and NE2 nitrogens of His-119.
The syn conformation for a pyrimidine nucleotide is not commonly observed (Saenger, 1984); the crystallization of 4thiouridine from aqueous solution has lead to a reported syn conformation of the base due to the inclusion of water of hydration in the crystal lattice (Saenger and Scheit, 1970;Saenger, 1984), but most pyrimidine nucleotides are found in the anti conformation. For the present RNase Aed(pT)4 complex the T2 nucleotide is found within the syn conformational range (Table 111). This may have been influenced by a number of interactions of the nucleotide with nearby charged amino acid side chains within the compltx. The NZ nitrogen of Lys-66 may form a contact (under 5 A) with the 01'-atom from the T2 deozyribose. The 0 2 oxygen of the T2 thymine base lies under 6 A from the NZ nitrogen of Lys-66. T h e 0 4 oxygen and C5 methyl group of this base are under 7 A from the ND1 and NE2 nitrogens of the His-105 residue from a neighboring protein molecule. The 0 4 atom also lies less than 8 A from the OD1 oxygen of the Asn-113 residue from a different symmetry related protein molecule and is about 9 A from the ND2 nitrogen of the same protein residue. A noticeable peak of difference Fourier electron density, which may be attributable to a solvent molecule, lies between the Lys-66 and Asn-67 side chains and the T2 deoxyribose within the complex, possibly influencing the syn conformation. Attempts to reposition the T2 and T3 nucleotides so that the T3 phosphate could occupy this density resulted in structurally inferior models with higher R factors. The syn orientation of the T2 base was a persistent feature of the model during refinements and reoccurred even after adjusting the glycosyl bond's torsion angle (x) to well within the anti range.
Comparisons to Other Complexes-No association of nucleic acid to the region of a cluster of basic amino acids, the "deep anion trap" (McPherson et al., 1986b), was detected in contrast to the RNase A.d(pA)4 complex. This may be due to the relatively neutral pH of the d(pT)4 solution used for diffusion into native crystals. Such a pH may inhibit electrostatic interactions with residues from this electropositive cavity (Lys-31, Lys-37, Lys-91, Arg-33).
The binding of the d(pA)4 oligomers to this region may also have been enhanced by the co-crystallization of the complex. Binding to the deep anion trap region of RNase A has not been reported for other complexes formed by diffusion of single-stranded nucleic acid monomers and oligomers into native crystals (Borkakoti, 1983;, even when the pH of the crystallization yedium is 5.5 (Wodak et al., 1977).
The 3.1-A structure of a complex formed by the diffusion of d(pT)4 into native crystals of the Klenow fragment of DNA polymerase I suggests that the pyrimidine oligomer binds to the 3'-5'-exonuclease site with splayed thymine bases, anticonformations, and C3'-endo sugars (Beese and Steitz, 1991). Unlike the polymerase I complex, where binding of the d(pT)4 tetramer to the protein seems uninfluenced by the base type, for the RNase A.d(pT)4 complex there are specific interactions between protein residues and base atoms. For example, probable van der Waals interactions involving the 5 ' -and 3'-C5 methyl groups allow for short contacts which may stabilize tetramer binding. The 3'-thymine base of the RNase A. d(pT)4 complex lies within the B2 or purine base-binding site. Its C5 methyl group appears to function in a role similar to that of the N1 and N6 atoms of an adenine base bound at the B2 site. The crystal structure of cytidilyl (a', 5') adenosine complexed with RNase S, for example, suggests that the N1 atom is protonated and forms a hydrogen bond to the side chain of Glu-111 and the N6 atom may hydrogen bond to the side chain of Asn-71 (Wodak et al., 1977). Short contacts which cannot be relieved by further refinements without loss of model integrity occur between the C5 methyl group of the 3'-nucleotide (T4) and both the OEl oxygen of Glu-111 and the ND2 nitrogen of Asn-71 in the RNase A. d(pT)4 complex. While interactions involving the thymine C5 methyl group are generally expected to be hydrophobic in nature, Kissinger et al. (1990) report an important sequence-specific van de: Waals contact within the major groove of DNA in the 2.8-A crystal structure of an engrailed homeodomain-DNA complex. Here the interaction is between the side chain of residue Gln-50 and the C5 methyl group from the thymine of base pair 14. It is formed in preference to making a hydrogen bond with the 0 4 oxygen of the thymine base. The theoretical calculation of a partial atomic charge of -0.42 for the C5 methyl carbon of deoxythymidine (Pearlman and Kim, 1990) implies that under some circumstances the C5 methyl group may be available for such vap der Waals interactions.
Low resolution (4 A) x-ray diffraction studies on crystalline complexes between RNase S with thymidine 3',5'-diphosphate and adenylyl-3',5'-cytidine suggest the presence of a nonspecific phosphate-binding site near protein residue Lys-66 (Sawada and hie, 1969;Mitsui et al., 1978;Iwahashi et al., 1981). A phosphate for the nucleotide 5' to the p l site could occupy this position, termed site PO. The location of the PO site was determined in part from coordinates of the Lys-66 residue elucidated crystallographically for uncomplexed RNase S (Wyckoff et al., 1970;Iwahashi et al., 1981). It was concluded that one of the phosphate oxygens at the PO site may form a hydrogen bond with the NZ nitrogen of Lys-66 (Iwahashi et al., 1981). In the present RNase A.d(pT)4 complex the T3 phosphate oxygens are not close enough to form such a hydrogen bond. The T3 phosphate lies closer to the side chain of residue Lys-41, with one of its oxygens being under 5 A from the NZ nitrogen of Lys-41. The PO site appears to be occupied by the most 3"phosphate of one of the d(pA)4 oligomers bound near the active site in the RNase A.d(pA)4 co-crystallization complex (McPherson et al., 198613). However, for that complex the PO and p l stites are not occupied by phosphates from the same tetramer. The PO site itself cannot fix the target phosphate group and does not attract free anions (such as sulfate ions) for the uncomplexed RNase molecule (Iwahashi et al., 1981). Therefore, occupation of the PO site by nucleic acid moieties may be limited to phosphates from the 5' or 3' terminus of single-stranded DNA/RNA oligomers, where there is greater conformational flexibility.