Coupling Ribose Selection to Fidelity of DNA Synthesis THE ROLE OF Tyr-115 OF HUMAN IMMUNODEFICIENCY VIRUS TYPE 1 REVERSE TRANSCRIPTASE*

The catalytic efficiency of incorporation of deoxyribo-nucleotides by wild-type human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) was around 100-fold higher than for dideoxyribonucleotides, in Mg 2 1 -catalyzed reactions, and more than 10,000-fold higher than for nucleotides having a 2 * -hydroxyl group in Mg 2 1 - and Mn 2 1 -catalyzed reactions. Mutant RTs with nonconservative substitutions affecting Tyr-115 (Y115V, Y115A, and Y115G) showed a dramatic reduction in their ability to discriminate against ribonucleotides in the presence of both cations. However, selectivity of de-oxyribonucleotides versus ribonucleotides was not affected in mutants Y115W and F160W. The substitution of Tyr-115 with Val or Gly had no effect on discrimination against dideoxyribonucleotides, but these mutants were less efficient than the wild-type RT in discriminating against cordycepin 5 * -triphosphate. We also show that Tyr-115 is involved in fidelity of DNA synthesis, but substitutions at this position have different effects depending on the metal cofactor used in the assays. In Mg 2 1 catalyzed reactions, removal of the side chain of Tyr-115 phosphorylated CA) was used to analyze the three-dimensional structures.

The catalytic efficiency of incorporation of deoxyribonucleotides by wild-type human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) was around 100-fold higher than for dideoxyribonucleotides, in Mg 2؉ -catalyzed reactions, and more than 10,000-fold higher than for nucleotides having a 2-hydroxyl group in Mg 2؉ -and Mn 2؉ -catalyzed reactions. Mutant RTs with nonconservative substitutions affecting Tyr-115 (Y115V, Y115A, and Y115G) showed a dramatic reduction in their ability to discriminate against ribonucleotides in the presence of both cations. However, selectivity of deoxyribonucleotides versus ribonucleotides was not affected in mutants Y115W and F160W. The substitution of Tyr-115 with Val or Gly had no effect on discrimination against dideoxyribonucleotides, but these mutants were less efficient than the wild-type RT in discriminating against cordycepin 5-triphosphate. We also show that Tyr-115 is involved in fidelity of DNA synthesis, but substitutions at this position have different effects depending on the metal cofactor used in the assays. In Mg 2؉catalyzed reactions, removal of the side chain of Tyr-115 reduced misinsertion and mispair extension fidelity, while opposite effects were observed in Mn 2؉ -catalyzed reactions. Our results indicate that the aromatic side chain of Tyr-115 plays a role as a "steric gate" preventing the incorporation of nucleotides with a 2-hydroxyl group in a cation-independent manner, while its influence on fidelity could be modulated by Mg 2؉ or Mn 2؉ .
The human immunodeficiency virus type 1 (HIV-1) 1 reverse transcriptase (RT) is a virally encoded enzyme. It converts the viral single-stranded RNA into double-stranded DNA which integrates into the host genome. The enzyme is multifunctional, possessing RNA-and DNA-dependent DNA polymerase, RNase H, strand transfer, and strand displacement activities (1,2). The HIV-1 RT is an error prone enzyme, as manifested by the frequencies of base substitutions, Ϫ1 frameshifts, and complex errors in the polymerization products (3,4). Unlike other DNA polymerases (e.g. Escherichia coli DNA polymerases I and III, T4 DNA polymerase, chicken polymerase ␥, or calf polymerase ␦, among others), retroviral RTs lack a proofreading activity. The low fidelity of HIV-1 RT contributes to retroviral mutagenesis and promotes the emergence of variants escaping the host's immune response, as well as viruses that are resistant to antiretroviral drugs such as RT or protease inhibitors.
The HIV-1 RT is a heterodimer composed of two subunits of 66 and 51 kDa, with subdomains termed fingers, thumb, and palm and connection in both subunits and an RNase H domain in the larger subunit only. The polymerase active site resides within the palm subdomain of the 66-kDa subunit, which bears the catalytic aspartic acid residues 110, 185, and 186. A crystal structure of a covalently trapped catalytic complex of HIV-1 RT containing a DNA template-primer and a deoxyribonucleoside triphosphate (dNTP) has been reported (5). According to this structure, the triphosphate moiety of the dNTP is coordinated by the side chains of Lys-65 and Arg-72, the main chains of Asp-113 and Ala-114, and two magnesium ions. The side chains of Arg-72 and Gln-151 pack against the outer surface of the incoming dNTP, and the ribose moiety of the incoming dNTP binds in a pocket defined by the side chains of Asp-113, Tyr-115, Phe-116, and Gln-151. Non-conservative substitutions at residues involved in dNTP binding are usually detrimental for polymerase activity and viral replication (6 -9).
Enzymatic characterization of recombinant HIV-1 RT variants led to the identification of mutations affecting Tyr-115 and other residues in its vicinity (e.g. Gln-151, Phe-160, Tyr-183, or Met-184) that influenced dNTP binding (8, 10 -15). In the case of Tyr-115, its replacement with Phe rendered RT fully active, although other amino acid changes such as Y115W, Y115V, Y115A, or Y115G diminished the polymerase activity of the enzyme, by increasing the K m values for the incorporation of dNTPs (11,13). Based on the crystallographic data, it has been suggested that the side chain of Tyr-115 is important for modifications at the 2Ј and 3Ј positions of the dNTP. In support of this proposal, the substitution of Val for the equivalent residue of Moloney murine leukemia virus (Mo-MLV) RT (Phe-155) rendered an enzyme with a dramatically increased affinity for ribonucleotides, compared with the wild-type (WT) RT (16). Unlike in the case of HIV-1 RT, the introduced mutation did not alter the affinity for dTTP. However, the kinetic parameters reported for HIV-1 RT were determined in the presence of magnesium cations (Mg 2ϩ ), while in the case of Mo-MLV, manganese cations (Mn 2ϩ ) were used as cofactors.
The consequence of replacing Mg 2ϩ with Mn 2ϩ in DNA polymerization was originally documented by Hall and Lehman (17), who showed that Mn 2ϩ caused the phage T4 DNA polymerase to be error prone. Evidence of increased error frequency in the presence of Mn 2ϩ has been observed in vitro with Escherichia coli DNA polymerase I (18 -22), T4 DNA polymerase (23), DNA polymerases ␣ and ␤ (24 -26), and avian myeloblastosis virus RT (27). In addition, Mn 2ϩ has been shown to induce preferential incorporation of dideoxy-versus deoxyribonucleotides in T7 DNA polymerase, Taq polymerase, and E. coli DNA polymerase I (28,29). In this paper, we have studied the effects of Mg 2ϩ and Mn 2ϩ on the nucleotide specificity of HIV-1 RT, by focusing on the role of Tyr-115 in nucleotide recognition at the 2Ј and 3Ј positions of the ribose ring and fidelity of DNA synthesis, in Mg 2ϩ -and Mn 2ϩ -catalyzed reactions. The reported data indicate that the mutagenic effect of Mn 2ϩ on DNA polymerization by HIV-1 RT operates mainly at the level of mispair extension. Non-conservative substitutions of Tyr-115 decrease fidelity of DNA synthesis only in Mg 2ϩ -catalyzed reactions. An aromatic amino acid residue is required at position 115 to discriminate against nucleotides having a 2Ј-OH group. The proposed role of Tyr-115 as a "steric gate" in HIV-1 RT does not depend on the cations used in the DNA polymerization reactions.
Enzymes-WT RT and mutants Y115W, Y115V, Y115A, Y115G, F160W, and G541* were constructed and purified as described previously (8,11,13,30). All RTs were purified as p66/p51 heterodimers. In this study, mutations were introduced in both subunits of the RT. The 51-kDa polypeptide was obtained with an extension of 14 amino acid residues at its N-terminal end, which includes six consecutive histidine residues to facilitate its purification by metal chelate affinity chromatography.
Single Nucleotide Extension Assays-Nucleotide incorporation assays were performed in 10 l of 50 mM Hepes (pH 7.0) buffer, containing 15 mM MgCl 2 or MnCl 2 , 15 mM NaCl, 130 mM KCH 3 COOH, 1 mM dithiothreitol, and 5% polyethylene glycol 6000. The template-primer concentration was 30 nM for D2-47/PG5-25 and 15 nM for M13 singlestranded DNA/pT. Both concentrations were saturating for WT HIV-1 RT in Mg 2ϩ -catalyzed reactions. Primers PG5-25 and pT were 32 Plabeled at their 5Ј end as described above. The active enzyme concentration in these assays was around 10 nM. Reactions were initiated by incubating the enzyme with the corresponding annealed templateprimer in the presence of Mg 2ϩ or Mn 2ϩ , but in the absence of nucleotide triphosphates (10 min at 37°C), followed by the addition of appropriate nucleotides at various concentrations. The reaction mixtures were incubated for 20 s in the case of D2-47/PG5-25, and 30 s in the case of M13mp2 single-stranded DNA/pT, and then the reactions were stopped by adding 6 l of the EDTA-formamide stop solution described above. The extension products resulting from the incorporation of 1 or 2 nucleotides at the 3Ј end of the primer were resolved by electrophoresis in 20% polyacrylamide-urea gels and primer extension was quantitated using a BAS 1500 scanner. Elongation measurements were fitted to the Michaelis-Menten equation and the k cat and K m values were determined as described previously (11).
Extension of Primers with Four dNTPs or Four rNTPs-Primer oligonucleotide PG5-25 was 5Ј end labeled and annealed to template oligonucleotide D2-47 as described above. The primer (30 nM) was extended by WT or mutant enzymes at a concentration of 15-30 nM in the buffer conditions described for single nucleotide extension assays, using either four dNTPs (at 1 mM each) or four rNTPs (at 1 mM each) as substrates. Incubations were carried out for 15, 30, 60, and 120 min. Products were processed and analyzed as indicated above.
Fidelity Assays-Misinsertion and mispair extension fidelity assays were performed essentially as described previously (31,32), using a standing-start protocol. End labeling of primers, template-primer annealing, and polymerization reactions were done in the conditions described for single nucleotide extension assays with template-primer D2-47/PG5-25. For mispair extension fidelity assays, three additional primers were used: PG5-25C, PG5-25G, and PG5-25A. All of them are identical to PG5-25, but have C, G, or A, respectively, at their 3Ј end. Template-primer concentrations were kept at 30 nM in all assays. The relative binding affinity of WT RT for matched and mismatched template-primer ends was determined using the equilibrium competition method (33).
Computer Analysis of Crystal Structures-Coordinates of crystal structures used in this study were taken from the Brookhaven Protein Data Bank (Upton, NY). The viewing program Insight II version 98.0 (Molecular Simulations Inc., San Diego, CA) was used to analyze the three-dimensional structures.

Discrimination between dNTPs and rNTPs by the HIV-1 RT Mutant Y115V-In
Mo-MLV RT, the replacement of Phe-155 by Val rendered an enzyme which incorporated rNTPs more efficiently than the WT RT (16). Examination of the active site of a high-resolution structure of HIV-1 RT complexed with template-primer and a dNTP substrate showed that the 2Ј-OH of an incoming ribonucleotide would overlap with the side chain of Tyr-115 (the equivalent residue of Phe-155 of Mo-MLV RT). The ability of WT HIV-1 RT and mutant derivative Y115V to discriminate between rNTPs and dNTPs was first assessed qualitatively. Primer extension was carried out with a DNA/ DNA duplex formed by M13 single-stranded DNA and a 20-mer oligonucleotide primer (pT). Assays were done with dNTP and rNTP competing as substrates for the polymerase. Shorter oligonucleotide products indicate that misincorporation of rNTPs instead of dNTPs occurs less frequently. As shown in Fig. 1, WT RT was able to discriminate very efficiently dNTPs versus rNTPs in the presence of Mg 2ϩ and Mn 2ϩ , although incorporation of rNMP was somewhat more efficient in the presence of Mn 2ϩ . Significant incorporation of rCMP was observed at a ratio of rNTP to dNTP of over 100:1. In contrast to WT RT, band patterns obtained with mutant Y115V revealed significant incorporation in the presence of low concentrations of competing rNTPs. For example, large products (Ͼ50 nucleotides long) were obtained in assays carried out in the presence of rGTP and dATP, dCTP, and dTTP, at a 1:1 ratio of rNTP to dNTP. As in the case of WT RT, misincorporation of rUTP was rather inefficient with mutant Y115V. Similar results in terms of ribonucleotide preferences were obtained either by using Mg 2ϩ or using Mn 2ϩ as metal ion cofactors. The results obtained with mutant G541* were almost identical to those obtained with the WT RT. G541* is a mutant enzyme lacking the last 20 amino acids of the 66-kDa subunit (30). This mutant is devoid of RNase H activity, and therefore cannot cleave potential rNMP:dNMP pairs which may be formed during primer extension assays.
Kinetic Analysis of Nucleotide Incorporation in the Presence of Mg 2ϩ or Mn 2ϩ -Steady-state kinetic parameters for the incorporation of nucleotides at the 3Ј end of the primer were obtained with two different DNA/DNA template-primer complexes (M13 single-stranded DNA/pT and D2-47/PG5-25). In experiments carried out with duplex M13 single-stranded DNA/pT, we compared the selectivity for the incorporation of rATP, ddATP, and cordycepin 5Ј-triphosphate (3Ј-dATP), instead of dATP for the WT RT and several mutants having substitutions at positions 115 and 160 (Table I). WT RT was able to discriminate very efficiently between dATP and rATP, and between dATP and 3Ј-dATP, with selectivity values around 10 Ϫ6 and 10 Ϫ5 . These effects were due to the large differences in the apparent K m values obtained with those nucleotides. The presence of Mg 2ϩ or Mn 2ϩ did not have an important effect in rATP/dATP selectivity, but discrimination against ddATP and 3Ј-dATP was 35-100 times more efficient in the presence of Mg 2ϩ . Removal of the side chain at position 115 (as in mutants Y115V, Y115A, and Y115G) produced an increase in the K m for dATP, in experiments carried out in the presence of Mg 2ϩ , but the corresponding K m values did not change significantly in the presence of Mn 2ϩ .
Y115V, Y115A, and Y115G showed a large reduction in their ability to discriminate between rATP and dATP, in the presence of both cations. This effect was more pronounced with mutants having a smaller side chain at position 115. As shown in Table I, the mutant Y115G showed similar kinetic parameters (k cat and K m ) for the incorporation of dATP and for the incorporation of rATP, in Mg 2ϩ -and Mn 2ϩ -catalyzed reactions. Discrimination between rATP and dATP was around 10 3 , 10 4 , and 10 5 times less efficient for mutants Y115V, Y115A, and Y115G, respectively, than for the WT RT. On the other hand, differences in selectivity were very small in the case of mutants Y115W and F160W compared with the WT RT. Discrimination of 2Ј-H versus 2Ј-OH was also significantly affected by nonconservative substitutions at position 115, in the absence of a 3Ј-OH group. Mutations at position 115 were also found to be critical for discrimination between ddATP and 3Ј-dATP. The catalytic efficiency (k cat /K m ) of incorporation of ddATP by the WT RT is around 10 3 times more efficient than the incorpora- tion of 3Ј-dATP in the presence of Mg 2ϩ or Mn 2ϩ as cofactors. These differences in efficiency are strongly reduced in mutants Y115V and Y115G. Thus, Y115G shows a mere 10-fold preference for ddATP over 3Ј-dATP. These results indicate that the aromatic side chain of Tyr-115 is critical to prevent the incorporation of nucleotides with a ribose moiety having a 2Ј-OH group, either in the presence or absence of a 3Ј-OH group.
Interestingly, in the presence of a 2Ј-OH, discrimination of 3Ј-OH versus 3Ј-H was affected by mutations at position 115. Thus, the incorporation of rATP or 3Ј-dATP by the WT RT was very inefficient, with selectivity values around 10 Ϫ6 and 10 Ϫ5 in Mg 2ϩ and Mn 2ϩ -catalyzed reactions, respectively (Table I). Mutants Y115V and Y115G showed higher catalytic efficiency for the incorporation of rATP compared with 3Ј-dATP, with selectivity values which are around 2 orders of magnitude higher for rATP than for 3Ј-dATP. However, discrimination of 3Ј-OH versus 3Ј-H was not significantly affected by substitutions at position 115, in the absence of a 2Ј-OH group. Selectivity values for the incorporation of ddATP instead of dATP ranged from 6.2 ϫ 10 Ϫ3 to 3.1 ϫ 10 Ϫ2 and from 2.9 ϫ 10 Ϫ2 to 0.48, in Mg 2ϩ -and Mn 2ϩ -catalyzed reactions, respectively. A clear effect of the substitutions at position 115 was not observed, although misincorporation of ddATP was somewhat higher for WT RT than for mutants Y115V and Y115G in assays carried out with Mn 2ϩ . Our results indicate that the 3Ј-OH might be important for nucleotide binding, but the side chain of Tyr-115 is not critical for recognition of the 3Ј-OH of the ribose.
The template-primer D2-47/PG5-25 was used to study the incorporation of pyrimidine nucleotide derivatives, and results obtained with this substrate were broadly in agreement with those obtained with the complex M13 single-stranded DNA/pT. Table II, WT RT showed a similar rUTP/dTTP discrimination in the presence of Mg 2ϩ and Mn 2ϩ , with selectivity values around 10 Ϫ5 . As shown with rATP/dATP determinations, the partial or total elimination of the side chain of Tyr-115 led to enzymes which poorly discriminate between ribonucleotides and deoxyribonucleotides, in Mg 2ϩ -and Mn 2ϩcatalyzed reactions. The effects of substituting Tyr-115 were similar in magnitude to those observed with template-primer M13 single-stranded DNA/pT. As observed in the case of dATP and ddATP, ddTTP/dTTP selectivity was more sensitive to mutations in the presence of Mn 2ϩ , with WT RT being the enzyme with less ability to discriminate against ddNTPs.

As shown in
Addition of Successive rNTPs-To further evaluate the ability of mutants Y115V and Y115G to incorporate ribonucleotides, we carried out assays using D2-47/PG5-25 as templateprimer and either dNTPs or rNTPs as nucleotide substrates. In the presence of Mg 2ϩ or Mn 2ϩ , the incorporation of ribonucleotides at position ϩ1 was barely detectable in the case of WT RT (Fig. 2). However, there was a substantial accumulation of bands representing the addition of 2 to 4 ribonucleotides at the 3Ј end of the primer, in experiments performed with mutants Y115V and Y115G. The use of rNTPs relative to dNTPs was more efficient in the case of Y115G, particularly in the presence of Mg 2ϩ . These results are consistent with the kinetic data obtained in single nucleotide extension assays, since the incorporation of dNTPs is strongly impaired in the case of mutant Y115G, which displays a higher K m value than the WT RT for the incorporation of dNTPs, in the presence of Mg 2ϩ . Despite the increased efficiency of incorporation of rNTPs shown by the Tyr-115 mutants, these enzymes were far from being true RNA polymerases, failing to render long RNA products.
Fidelity of DNA Synthesis in the Presence of Mg 2ϩ or Mn 2ϩ - M13mp2 single-stranded DNA/pT After formation of the RT⅐DNA/DNA complex, M13mp2 single-stranded DNA/pT elongation reactions with all enzymes and nucleotides were incubated at 37°C for 30 s, except for the incorporation of 3Ј-dATP in the presence of Mg 2ϩ , and the incorporation of rATP by mutants Y115W and F160W which were incubated at 37°C for 10 -30 min. Incorporation of rATP at position ϩ2 was observed in the case of Y115V, Y115A, and Y115G, and reported k cat values assume in these cases the incorporation of two nucleotides at the 3Ј end of the primer. Data shown are the mean values Ϯ S.D., obtained from a nonlinear least-squares fit of the kinetics data to the Michaelis-Menten equation. Each of the experiments was performed independently at least three times. Misinsertion and mispair extension fidelity assays were used to estimate the fidelity of WT and mutant RTs, using D2-47/ PG5-25 as the template-primer. Misinsertion fidelity assays involved kinetic measurements for the incorporation of a correct (T) or an incorrect (A, C, or G) nucleotide at the 3Ј end of the primer. The misinsertion ratios obtained for the WT RT ranged from less than 5 ϫ 10 Ϫ6 to 3.9 ϫ 10 Ϫ5 in the presence of Mg 2ϩ , and from 1.6 ϫ 10 Ϫ5 to 3.3 ϫ 10 Ϫ4 in the presence of Mn 2ϩ (Table III). Misinsertion ratios for the WT RT were higher in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . In contrast, Y115V was slightly more faithful in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . Interestingly, misinsertion ratios were 2-5-fold higher for the Y115V mutant than for the WT RT in the presence of Mg 2ϩ . However, in the presence of Mn 2ϩ , the effect was the opposite, with Y115V being the more faithful enzyme. Misinsertion fidelity assays were not done with Y115G due to the very low incorporation rate of incorrect nucleotides at the 3Ј end of the primer. In the case of Y115A, misinsertion ratios were 4 -10-fold higher than for the WT RT in the presence of Mg 2ϩ , in consistency with the observed trend toward reduced fidelity as the volume of the side chain at position 115 decreases (13). The kinetics of mispair extension were studied for correctly matched base pairs (A:T) and for mismatches A:C, A:G, and A:A. In all cases, we measured the incorporation of a correct T opposite of A at the 3Ј end of the primer. Our analysis assumes that RTs bind with roughly equal affinity to the matched and mismatched template-primer ends. In the case of WT HIV-1 RT, we found that the K D values for A:C, A:G, and A:A mispairs were less than 4-fold higher than for A:T, in our assay conditions. These data are in agreement with previous measurements using other polymerases, including avian myeloblastosis virus RT (33), and reporting no significant differences in the K D values for binding matched versus mismatched templateprimer ends. Mispair extension efficiencies for WT RT, Y115V, and Y115G are shown in Table IV. Mispair extension ratios for  the WT RT were about 100-fold higher in the presence of Mn 2ϩ than in the presence of Mg 2ϩ , for all three mispairs tested. Transversion mispairs were less efficiently extended by the WT RT than A:C mispairs. In the presence of Mg 2ϩ , the removal of the side chain of Tyr-115 rendered enzymes which were more efficient in extending mispairs. Thus, A:C mispair extension efficiencies relative to the WT RT were 3.9 and 50.2 times higher for mutants Y115V and Y115G, respectively. The effects observed with these substitutions were more pronounced in the case of transversion mispairs (e.g. A:A or A:G), whose extension efficiencies were at least 2,500-fold higher for Y115V and Y115G, than for the WT RT. In the presence of Mn 2ϩ , mispair extension ratios are not largely affected by substitutions at position 115, although mutants Y115V and Y115G appear to be more faithful than the WT RT.  (34). DNA polymerase activity requires Mg 2ϩ , although the enzyme retains significant activity in the presence of Mn 2ϩ . However, it has been reported that when Mn 2ϩ is used as cofactor, the HIV-1 RT can produce long repetitive products due to extensive primer slippage during DNA synthesis (35), and could also incorporate rGTP when poly(rC)⅐oligo(dG) 10 is used as template-primer (36). Moreover, treatment of cells with Mn 2ϩ and subsequent HIV-1 infection resulted in at least 10fold increases in the observed mutation frequency (37), in agreement with previous reports showing that avian myeloblastosis virus RT has a decreased fidelity in the presence of Mn 2ϩ (27). Our results are broadly consistent with these observations, and reveal that the substitution of Mg 2ϩ with Mn 2ϩ alters the substrate specificity of HIV-1 RT. In our assay conditions, the catalytic efficiency of WT HIV-1 RT was higher in the presence of Mg 2ϩ than in the presence of Mn 2ϩ . However,  After formation of the RT⅐DNA/DNA complex, mispair extension reactions were incubated at 37°C for 20 s for the elongation of A:T and A:C mispairs, and 5-10 min for A:G and A:A mispairs. In all cases, we measured the incorporation of T opposite to A at position ϩ1. Data shown are the mean values Ϯ S.D., obtained from a nonlinear least-squares fit of the kinetics data to the Michaelis-Menten equation. Each of the experiments was performed independently at least three times. discrimination between ddNTPs and dNTPs, and between 3Ј-dATP and dATP was largely affected by the presence of Mg 2ϩ or Mn 2ϩ . In the case of ddNTPs and dNTPs, apparent K m values were similar for both nucleotides in Mn 2ϩ -catalyzed reactions, while in the presence of Mg 2ϩ , the WT RT showed a 100-fold lower K m for dNTP. These results suggest that the environment of the 3Ј-ribose group could be largely affected by Mn 2ϩ . On the other hand, nucleotide discrimination between rNTPs and dNTPs was similar with both metal ion cofactors. Misinsertion and mispair extension fidelity assays carried out with the WT RT also showed a loss of nucleotide specificity in the presence of Mn 2ϩ , in agreement with the higher mutagenic potential of this cation. However, the largest effects were observed in determinations of mispair extension efficiencies that were about 2 orders of magnitude higher in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . Although reported evidence on the effects of cations on misinsertion and mispair extension fidelity is limited, our results suggest that error discrimination in HIV-1 RT operates at a different level than with other DNA polymerases such as E. coli DNA polymerase I or human DNA polymerase ␣ (18,25), where the effect of Mn 2ϩ on fidelity of DNA synthesis is more pronounced at the insertion step.

DISCUSSION
The crystal structure of a ternary complex of HIV-1 RT, a DNA template-primer, and dTTP, has revealed that in the catalytic subunit, Tyr-115 is located in the nucleotide-binding site of the enzyme, below the ribose ring of the incoming dNTP (5). The 3Ј-hydroxyl group of the sugar moiety points toward the side chain of Tyr-115, making a hydrogen bond with the amido group of the peptide bond between Ala-114 and Tyr-115 (Fig. 3). Phe is the only residue that can replace Tyr-115 of HIV-1 RT without causing a detrimental effect in its DNA polymerase function (6,11,13,14). Enzymological characterization of mutant RTs with non-conservative substitutions at position 115 resulted in enzymes displaying lower affinity for dNTPs than the WT RT, in DNA polymerase assays carried out in the presence of Mg 2ϩ (11,13). Therefore, viruses having non-conservative substitutions at this position (e.g. Y115L, Y115A, Y115N, or Y115D) are not viable (6,9). In Mo-MLV RT, an aromatic residue at the equivalent position of Tyr-115 is required for infectivity, since only Phe-155 or Tyr can support virus replication (38). Enzymatic characterization of a mutant having Val instead of Phe-155 revealed that this substitution had no effect on the K m for the correct nucleotide (dTTP) in assays done with poly(rA)⅐oligo(dT) 12 (16). These results sug-gested that Tyr-115 of HIV-1 RT and Phe-155 of Mo-MLV RT could play a different role depending on the sequence context (38). However, our results show that this difference may be attributed to the metal ion cofactor used in the RT assays. In the presence of Mg 2ϩ , removal of the side chain of Tyr-115 lowers the dNTP binding affinity of the enzyme, but in the presence of Mn 2ϩ which is the cation used in Mo-MLV RT assays, substitutions at position 115 have a minor effect on the catalytic efficiency of the enzyme (measured as k cat /K m ).

Role of Tyr-115 Substitutions in Nucleotide Sugar Discrimination and Fidelity of DNA Synthesis-Studies with Mo-MLV
RT showed that WT RT exhibited a 15,000-fold preference for dNTPs compared with rNTPs, in the presence of Mn 2ϩ . However, this selectivity value was reduced to about 25-fold for mutant F155V (16). Our data are consistent with these observations and indicate that Tyr-115 of HIV-1 RT is critical to discriminate between dNTPs and rNTPs in the presence of Mn 2ϩ and also in the presence of Mg 2ϩ . The equivalent mutant (Y115V) showed a preference for dNTP over rNTP that was about 10 2 to 3 ϫ 10 3 times lower than the WT RT, in the presence of both cations. Moreover, discrimination against rNTPs was further decreased by introducing mutations encoding for residues with smaller side chains at position 115, such as Y115A or Y115G. In the case of Y115G, the selectivity of rNTPs over dNTPs was close to 1 (around 100,000-fold less efficient than the WT), thereby indicating that this variant was unable to distinguish between dNTPs and rNTPs. Interestingly, increasing the size of the side chain at position 115 rendered an enzyme (Y115W) which displayed a decreased affinity for dNTP in the presence of Mg 2ϩ , but displayed similar rNTP/dNTP discrimination than the WT in both Mg 2ϩ -or Mn 2ϩ -catalyzed reactions. It has been reported that substituting Trp for Phe-155 of Mo-MLV RT or Tyr-115 of HIV-1 RT renders a virus which either cannot replicate or replicates at a very low rate (9,38). Our data suggest that the poor viability of these mutant viruses is probably due to the lower affinity for dNTPs of their RTs in the presence of the more physiological cation Mg 2ϩ , rather than to inefficient discrimination between dNTPs and rNTPs.
Mutational analysis of Phe-160 of HIV-1 RT has shown that this residue could also affect dNTP binding through its hydrophobic interaction with Tyr-115 (8). Although substituting Trp for Phe-160 produced a 10 -30-fold increase in the K m for the incorporation of a correct dNTP, this mutant displayed a similar rNTP/dNTP discrimination compared with the WT RT. It can be concluded that discrimination against the 2Ј-hydroxyl group of a ribonucleotide substrate can be maintained if an aromatic ring is present at position 115. Tyr-115 appears to function as a steric gate that prevents the incorporation of rNTPs by interfering with the 2Ј-OH of the ribose moiety. In agreement with this view, discrimination between dATP and 3Ј-dATP was very efficient for WT RT in the presence of Mg 2ϩ and Mn 2ϩ , as expected from the potential steric interference between the 2Ј-OH of 3Ј-dATP and the side chain of Tyr-115. The elimination of the side chain of position 115 reduced the ability of the RT to discriminate between 3Ј-dATP and dATP. In this case the effects were not as pronounced as with rNTP/ dNTP discrimination, due to the poor binding of 3Ј-dATP to mutant RTs, and therefore, suggesting that the 3Ј-OH is critical for the interaction between the enzyme and the nucleotide substrate. Interestingly, ddNTP/dNTP discrimination by the WT HIV-1 RT was less efficient than rNTP/dNTP and 3Ј-dATP/ dATP discrimination, and was not largely affected by amino acid changes at position 115. The absence of a hydroxyl group at the 3Ј position of the sugar eliminates the specificity for dNTP of the nucleotide binding pocket. Several nucleoside analog inhibitors of RT used in the therapeutic treatment of HIV infection are dideoxyribonucleosides. In vitro assays have shown that mutant Y115N was resistant to azidothymidinetriphosphate and ddTTP, although viruses carrying this mutation were not viable (39). In addition, substitutions affecting Tyr-115 are not frequently identified during drug therapy, and only Y115F has been found to confer low level resistance to abacavir, a dideoxyguanosine derivative (40).
Previously, we showed that in the presence of Mg 2ϩ , misinsertion of G instead of T was higher for mutants lacking an aromatic side chain at position 115 (11), while mispair extension efficiencies for A:C mispairs increased from a few fold higher than the WT RT with Y115W or Y115V, to about 35-fold as observed with mutant Y115G (13). The results reported in this paper for Mg 2ϩ -catalyzed reactions are in agreement with those findings. Thus, mutants Y115V and Y115G showed a significant decrease in misinsertion and mispair extension fidelity in the presence of Mg 2ϩ , while Y115V and Y115G are slightly more faithful than the WT RT in Mn 2ϩ -catalyzed reactions. Taken together, our data suggest that the dNTP, the template-primer or both could adopt a different conformation within the catalytic site of HIV-1 RT in Mg 2ϩ -or Mn 2ϩ -catalyzed reactions.
Comparison with Other Polymerases-Sequence alignments of RNA-dependent DNA polymerases, including RTs revealed that Tyr-115 is part of a highly conserved motif which contains the catalytic residue Asp-110 (Fig. 4), known as motif A (41). Tyr-115 of HIV-1 RT is conserved in DNA polymerases ␣ (42), where it has been shown to participate in rNTP/dNTP discrimination, as demonstrated for Thermococcus litoralis (Vent TM ) DNA polymerase (43) and bacteriophage 29 DNA polymerase (44), by analyzing the effects of substituting the equivalent Tyr residues with Val. These effects were observed with Mg 2ϩ as cofactor in the case of Vent TM DNA polymerase, and with Mn 2ϩ in the case of 29 DNA polymerase. If Tyr-115 of HIV-1 RT and the equivalent residues of Mo-MLV RT and Vent TM and 29 DNA polymerases appear to be critical to discriminate against rNTPs, a similar role has been proposed for Glu-710 of E. coli DNA polymerase I (Klenow fragment). Replacement of Glu-710 by Ala decreases discrimination against rNTPs by 1,000-fold (45). In addition, this substitution leads to a 12-20-fold decrease in the enzyme's ability to discriminate against ddNTPs, in experiments carried out in the presence of Mg 2ϩ (46). These data strongly suggest that Glu-710 of Klenow polymerase acts as a steric gate, exerting a higher selectivity on ribose 2Ј-group discrimination, as observed for Tyr-115 of HIV-1 RT. The participation of the side chain of Glu-710 in hydrogen bonds with the substrate could be implicated in the discrimination mechanism (47). Examination of the available three-dimensional structures of ternary complexes of T7 DNA polymerase and Taq DNA polymerase bound to template-primer and a nucleotide indicates that Glu-480 and Glu-615, respectively, which are the equivalent residues of Glu-710 of E. coli DNA polymerase I, are located under the sugar moiety of the incoming nucleotide, in a structural position that is equivalent to that of Tyr-115 of HIV-1 RT (48,49). As shown for Tyr-115, substitutions affecting Glu-710 of E. coli DNA polymerase I (e.g. E710A) or Glu-480 of T7 DNA polymerase (e.g. E480D) may also affect fidelity of DNA synthesis (47,50).
In T7 RNA polymerase, replacement of Tyr-639 by Phe leads to a gross deficit in discrimination between rNTPs and dNTPs, due to an enhanced ability of the enzyme to utilize dNTPs (51). In this case, the mechanism of discrimination between deoxyribonucleotides and ribonucleotides differs from the one suggested for HIV-1 RT. The release of a hydrogen-bonded water molecule upon binding of rNTP to the enzyme versus mutant Y639F has been invoked to explain 2Ј-group discrimination by this polymerase (52). The crystal structure of T7 RNA polymerase has revealed that the equivalent position of Tyr-115 of HIV-1 RT or Glu-710 of Klenow polymerase is occupied by Gly-542, a residue that could also play a role in discriminating against dNTPs (53). In addition to HIV-1 RT, DNA polymerases of type I, and T7 RNA polymerase, the nucleotide binding pocket of DNA polymerase ␤ has also been studied in detail (54 -56). Close van der Waals contacts have been observed between the protein backbone atoms of Tyr-271, Phe-272, and Gly-274 and the ribose ring carbons C2Ј and C3Ј of ddCTP. It has been suggested that the protein backbone segment spanning Tyr-271 to Gly-274 participates in nucleotide selectivity of deoxyribose over ribose (54). The structural disposition of Tyr-271 and Phe-272 of DNA polymerase ␤ resembles that observed with Phe-160 and Tyr-115 of HIV-1 RT. Interestingly, non-conservative substitutions in DNA polymerase ␤, such as Y271A, Y271S, or F272L, decreased fidelity of DNA synthesis as compared with the WT RT (57,58).
Conclusions-Tyr-115 of HIV-1 RT plays a pivotal role in the discrimination of dNTPs versus nucleotide derivatives having a 2Ј-OH group (rNTPs and 3Ј-dATP), by acting as a steric gate that impedes the correct positioning of the rNTP (or 3Ј-dNTP). This effect does not depend on using Mg 2ϩ or Mn 2ϩ in the polymerization reaction. Tyr-115 is also involved in fidelity of DNA synthesis, but substitutions have different effects depending on the metal ion cofactor used. While in Mg 2ϩ -catalyzed reactions removal of the side chain of Tyr-115 decreases mis- insertion and mispair extension fidelity, the effects are the opposite in the presence of Mn 2ϩ , suggesting that interactions at the nucleotide-binding site are significantly different with both cations. The mutagenic effect of Mn 2ϩ on DNA polymerization by HIV-1 RT is mediated by its higher efficiency of mispair extension in the presence of this cation, and could be mediated by alterations in the positioning of the templateprimer. Our results together with data reported by other groups suggest that fidelity of HIV-1 RT (and other polymerases) results from multiple interactions involving substrates and RT interacting sites, which can be modulated by external factors including metal ion cofactors. A detailed analysis of the role played by amino acid residues contributing to dNTP binding will be necessary to address new ways to inhibit or impair virus replication, including lethal mutagenesis with mutagenic deoxynucleoside analogs (59).