Differential effects of Moloney murine leukemia virus reverse transcriptase mutations on RNase H activity in Mg2+ and Mn2+.

We have previously described the in vitro and in vivo characterization of a panel of mutations affecting the RNase H domain of Moloney murine leukemia virus reverse transcriptase (Blain, S. W., and Goff, S. P.(1993) J. Biol. Chem. 268, 23585-23592; Blain, S. W., and Goff, S. P. (1995) J. Virol. 69, 4440-4452). We were intrigued by a discrepancy between in vitro and in vivo RNase H results for two of the mutants. While ΔC and Δ5E appeared to have nearly wild-type RNase H activity in vitro, they were unable to degrade their genomic RNA in vivo and thus were effectively RNase H null mutants in this context. In this present report, we describe the differential effects of these mutations on RNase H activity in vitro in the presence of Mgversus Mn: mutants ΔC and Δ5E were active in the presence of the less biologically relevant Mn and not in the presence of Mg. We also describe three mutants with only partial activity in Mg. The presence of the different cations can also affect DNA polymerization and processivity of an RNase H-deficient mutant.

Thus, although the structure of MMLV RNase H has not been determined, it is likely that the RNase H domain of this enzyme will be similar to those of other RNases H (15).
RNase H activity has been implicated in several steps in reverse transcription: the enzyme is essential for the viral life cycle, and mutant viruses that lack RNase H activity are noninfectious (16). RT initiates DNA synthesis from a tRNA primer bound to a region near the 5Ј-end of the genomic RNA termed the primer-binding site (PBS). Elongation of this tRNA to the 5Ј-end of the genome results in formation of (Ϫ)-strand strong stop DNA, the first DNA intermediate to appear during reverse transcription (17). The newly synthesized (Ϫ)-strand strong stop DNA forms an RNA-DNA hybrid with the (ϩ)strand genomic RNA, which is then degraded to permit translocation to the 3Ј-end of the RNA. Analysis of abortive replication products produced by virions that lack RNase H has shown that the (Ϫ)-strand strong stop DNA remains in hybrid form with the genomic RNA, accounting for the observed reduction in translocation and elongation for these mutants (16). In addition to degradation of the genomic RNA, RNase H performs several specialized functions at later times, including the creation and removal of the polypurine tract primer and the removal of the (Ϫ)-strand tRNA primer (18).
We previously described the in vitro characterization of a panel of mutations made in the RNase H domain of MMLV RT (19). The design of these mutant enzymes was based on sequence alignments and the crystal structures of E. coli and HIV-1 RNases H and the predicted structure of the MMLV RNase H domain (10 -15). Most of the RNase H mutants analyzed retained full or at least partial RNase H activity in vitro as assayed by in situ gel techniques. We additionally characterized these mutants in vivo in the context of the full-length retroviral provirus (20). Two mutants, ⌬5E and ⌬C RTs, which appeared to retain significant RNase H activity in vitro (50 and 100% activity, respectively), were completely noninfectious as virus in vivo. These mutant viruses were further analyzed in the endogenous assay, in which reverse transcription is carried out in vitro in purified virions in the presence of radiolabeled dNTPs, and various radiolabeled DNA products can be detected. This analysis demonstrated that ⌬5E and ⌬C left their (Ϫ)-strand strong stop DNA in hybrid form with the genomic RNA and thus were effectively RNase H null mutants in the context of the endogenous reaction.
We were intrigued by the discrepancy between the presence of RNase H activity in the in situ gel assay and the absence of RNase H activity in the endogenous assay. We hypothesized that the differential activity detected in these two assays might result from the following: 1) a difference between the recombinant and virion-associated RTs analyzed, 2) a difference between the substrates degraded in these two assays (random heteropolymeric radiolabeled RNA-DNA hybrid in the in situ gel assay versus genomic RNA hybridized to the newly synthe-sized (Ϫ)-strand DNA during the endogenous reaction), 3) the presence of other viral proteins that might affect RT activity during reverse transcription, or 4) a difference between the different cations used in these two assays (Mn 2ϩ versus Mg 2ϩ ).
Tests revealed that the basis for the difference in the activity of these mutant enzymes in vitro and in vivo resulted from differential RNase H activity when assayed in the presence of Mg 2ϩ or Mn 2ϩ . In particular, we describe two mutants that appear to be active only in the presence of the less biologically relevant Mn 2ϩ . The presence of the different cations also appears to affect DNA polymerization and processivity.

EXPERIMENTAL PROCEDURES
Description of Mutants-MMLV RTs were constructed by oligonucleotide-mediated site-directed mutagenesis as described previously (19). R657S, Y598V, S526A, Y586F, and D524N contain point mutations in the RNase H domain of MMLV RT; the mutant enzymes were named by appending the amino acid present in wild-type RT, the residue number, and the amino acid present in the mutant at that position. ⌬5E has a 5-amino acid deletion from Ser-643 to Arg-647. ⌬C has an 11-amino acid deletion from Ile-593 through Leu-603 in the RNase H domain (19,21). H7 is a linker insertion mutant, containing a frameshift at the start of the RNase H domain, and thus is effectively an RNase H null form of RT (9,16).
pRT30-2 (the wild-type enzyme), D524N, ⌬5E, and ⌬C were analyzed as E. coli-expressed proteins, purified as described previously (19,21). Mutants R657S, Y598V, S526A, Y586F, ⌬5E, and H7 were analyzed as virion-associated RTs. To prepare the mutant RTs, these mutations were moved into the context of the full-length provirus pNCA (16,20,22). Stable producer lines were generated for these mutant proviruses by the calcium phosphate-mediated cotransformation method as described previously (16,20). Maintenance of cells in Dulbecco's modified Eagle's medium supplemented with 10% calf serum was as described previously (16,20). To prepare virions, producer cells were fed Dulbecco's modified Eagle's medium supplemented with 10% NuSerum (Collaborative Biomedical Products, Bedford, MA) for 12 h prior to harvest. The virions were pelleted for 3 h at 25,000 rpm, resuspended, layered over a 25/45% sucrose step gradient, and sedimented to the interface. The viral band was collected and repelleted for 2 h at 25,000 rpm following dilution in TNE buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA). Viral RTs were assayed without further purification; virions were lysed in Nonidet P-40 present in the reaction mixture as described below. The enzyme diluent was TNE buffer for all of the dilutions.
Both bacterially expressed and virion-associated RTs were quantitated to determine the units/milliliter of DNA polymerase activity by oligo(dT)⅐poly(rA) primer template assays in Mn 2ϩ (23), and enzymes to be compared were normalized by this assay. Since all of the mutant RTs had previously been shown to have wild-type DNA polymerase activity in this assay (19), normalization by the measure of RT activity was appropriate. Protein levels were also compared by Western blot analysis with anti-30-2, a polyclonal rabbit anti-RT antibody (19,20). In all cases, the protein levels and thus the polymerase-specific activity of the mutants were the same as those of wild-type RT.
RNase H Defined Substrate Assay-An RNA-DNA hybrid spanning the long terminal repeat region was used to assay RNase H activity in vitro. Radiolabeled, single-stranded (Ϫ)-strand DNA was made by asymmetric polymerase chain reaction using 60 pmol of T7 primer, 1 pmol of RS primer (Stratagene, La Jolla, CA), and a Bluescript template (Stratagene) containing pNCA sequences from the MscI (position 662) to the SacI (position 414) sites. The radiolabeled, single-stranded product was gel-purified and electroeluted in 0.25 ϫ Tris borate/EDTA. Nonradiolabeled complementary (ϩ)-strand RNA was made by standard transcription methods from the same template, using T3 RNA polymerase. The RNA and DNA were mixed in buffer containing 40 mM PIPES, pH 6.4, 1 mM EDTA, 400 mM NaCl, and 80% formamide; heated to 85°C for 10 min; and annealed at 45°C for 3 h. To remove unannealed RNA or DNA and to blunt the ends of the hybrid, the reaction was treated with 50 units of S1 nuclease in 1 mM ZnCl 2 , 300 mM NaCl, 25 mM NaOAc, pH 4.5, for 1 h at 45°C. The reaction was stopped by the addition of EDTA, and the product was phenol-extracted and ethanolprecipitated in the presence of glycogen carrier. The final product was a 245-base pair hybrid spanning the complete (Ϫ)-strand strong stop DNA from the PBS and extending to the SacI site in U3. The hybrid also included 55 bases 3Ј of the PBS to the MscI site.
Concentrated virions were incubated for 30 min on ice in the pres-ence of 0.3% Nonidet P-40, 50 mM NaCl, and 1 mM dithiothreitol in a 15-l volume. Reaction mixtures were then diluted to a total of 85 l with buffer lacking Nonidet P-40 to give final concentrations of 0.05% Nonidet P-40, 50 mM NaCl, 1 mM dithiothreitol, and either 6 mM MgCl 2 with 50 mM Tris-HCl, pH 8.3, or 2 mM MnCl 2 with 50 mM Tris-HCl, pH 7.5. The substrate was added to a final concentration of ϳ1 pM. The reaction was allowed to proceed at 37°C for 1 h, followed by SDS and proteinase K treatment for 30 min at 37°C; and reaction products were extracted with phenol and ethanol-precipitated in the presence of 1 g of tRNA. The products were analyzed by electrophoresis on 8% nondenaturing polyacrylamide gels. Purified recombinant enzymes were incubated in a 50-l reaction containing 0.05% Nonidet P-40, 50 mM NaCl, 1 mM dithiothreitol, 3 pM substrate, and either 6 mM MgCl 2 with 50 mM Tris-HCl, pH 8.3, or 2 mM MnCl 2 with 50 mM Tris-HCl, pH 7.5, for 1 h at 37°C. Following incubation, the reactions were treated as described above for the virion reactions.
Endogenous RT Reactions-In the endogenous reaction, reverse transcription is carried out in vitro in purified virions on the endogenous viral RNA template in the presence of radiolabeled dNTPs. Endogenous reactions using concentrated virions were performed to detect (Ϫ)-strand strong stop products. Reactions were incubated for various times (Ͻ30 min) at 37°C with 1 mM each dATP, dGTP, and dCTP and 2.5 M [␣-32 P]TTP at 400 Ci/mmol as described previously (16,20). Reactions contained either 6 mM MgCl 2 with 50 mM Tris-HCl, pH 8.3, or 2 mM MnCl 2 with 50 mM Tris-HCl, pH 7.5. Reaction products were treated with SDS and proteinase K for 15 min at 37°C, phenol-extracted, and ethanol-precipitated in the presence of carrier tRNA. These products were then treated with RNase A (2 g in 50 l) in either a low salt (TE) or high salt (TE containing 0.5 M NaCl) buffer for 30 min at 37°C as described previously (16,20). The enzyme was inactivated by diethyl pyrocarbonate treatment, and the products were phenol-extracted and ethanol-precipitated in the presence of carrier tRNA. The samples were then either diluted into electrophoresis sample buffer and disaggregated without denaturation by mild heating (50°C for 5 min) or diluted into buffer containing formamide and fully denatured (90°C for 10 min). The samples were analyzed by electrophoresis on 8% nondenaturing polyacrylamide gels.

RNase H Activity in Mg 2ϩ and Mn 2ϩ : Defined Substrate
Assay-To test for the activity of mutant RTs in vitro in a setting resembling the natural one in the virion, we assayed the mutants using a defined RNA-DNA hybrid as substrate. This substrate consists of a heteropolymeric hybrid with unlabeled (ϩ)-strand RNA spanning the 5Ј-end of the MMLV genome annealed to radiolabeled complementary (Ϫ)-strand DNA. This substrate contains the PBS, U5, R, and part of the U3 region and corresponds to the authentic sequence that MMLV RNase H encounters during reverse transcription. This assay differs from more conventional solution assays in that we radiolabeled the DNA strand and analyzed the products on an 8% nondenaturing polyacrylamide gel (Fig. 1). The doublestranded hybrid migrates at ϳ240 base pairs (Fig. 1, lane 19), while the single-stranded DNA species produced upon denaturation of the substrate or RNase H degradation of the RNA migrates more slowly (lane 20). With this substrate, we were able to test both detergent-permeabilized virion-associated RNases H and purified recombinant enzymes. Several of the mutants were analyzed as virion-associated enzymes to avoid possible E. coli nuclease contamination.
Purified recombinant RTs were added to this substrate and incubated at 37°C for 1 h (Fig. 1). Purified E. coli RNase H was able to degrade this substrate efficiently, producing the singlestranded slower species, which corresponded to full RNA degradation and/or release (Fig. 1, lane 2). Purified wild-type RT (pRT30-2) was able to degrade the substrate efficiently in both Mn 2ϩ and Mg 2ϩ (Fig. 1, lanes 3-8; and Fig. 2, lanes 3-6). However, wild-type RT was ϳ8 -16-fold more active in the presence of Mn 2ϩ compared with Mg 2ϩ (Fig. 1, compare lanes 6 and 7 in Mn 2ϩ to lanes 3 and 4 in Mg 2ϩ ).
Virion-associated wild-type RT was compared with the recombinant enzyme. Wild-type RT from virions was able to efficiently degrade the substrate in both Mn 2ϩ and Mg 2ϩ , and a 16-fold increase in activity in the presence of Mn 2ϩ was observed (Fig. 3, lanes 2-4; and Fig. 4, lanes 3-6), as with the recombinant enzyme. This result suggests that MMLV RNase H was indeed significantly more active in the presence of Mn 2ϩ compared with Mg 2ϩ and that similar results could be obtained with the recombinant and virion-associated enzymes analyzed in this assay.
Analysis of Mutant RTs-Several mutant RTs were assayed in the defined substrate assay to determine if their alterations could differentially affect the cation preferences of the enzymes. Mutants ⌬5E and ⌬C prepared as recombinant proteins were able to degrade the hybrid efficiently in Mn 2ϩ (Fig. 1, lanes 9 -14) almost as well as the wild type (compare lanes 6, 9, and 12). This result is consistent with our previous results in Mn 2ϩ in the in situ gel assay (Table I) (19). In Mg 2ϩ , however, these two mutants showed dramatically reduced activity and were unable to completely degrade all of the RNA to release the single-stranded DNA (Fig. 2, lanes 7-14). Some activity was apparent at the highest concentrations of ⌬5E (Fig. 2, lane 7); the hybrid form disappeared, but only a slower migrating smear appeared for ⌬5E at the 1:1 concentration. This concen-   Fig. 3. ds, double-stranded; ss, single-stranded. tration was ϳ10 times more than the concentration of wild-type RT needed to degrade the substrate to completion. This smear presumably corresponded to a heterogeneous population of DNAs with various sized RNA species still annealed.
Since these enzymes were purified from E. coli, we could not rule out the possibility that the residual activity of ⌬5E and ⌬C at higher enzyme concentrations was due to contaminating E. coli RNase H activities, although by in situ gel techniques, these enzyme preparations did not contain any other detectable RNase H activities (19,21). To address this concern, ⌬5E was analyzed as a virion-associated enzyme. ⌬5E exhibited wildtype activity in Mn 2ϩ , but little or no activity in Mg 2ϩ (Fig. 3,  lanes 14 -16; and Fig. 5, lanes 1-4). As a virion-associated enzyme, the slower migrating smear was not detected, even at high enzyme concentrations (Fig. 5, lanes 1-4). Thus, ⌬5E behaved effectively as an RNase H null mutant in Mg 2ϩ . This result might account for the discrepancy in the activities seen in the endogenous reaction and the in situ gel assay for mutants ⌬5E and ⌬C (Table I). While ⌬5E and ⌬C do retain nearly wild-type RNA-DNA nuclease activity in Mn 2ϩ , they are essentially inactive in the presence of Mg 2ϩ .
Several other mutants were analyzed to determine whether the differential ability to degrade RNA depending on the divalent cation used was a common feature of many RNase H mutants. Mutant D524N, containing a change in a residue implicated in catalytic activity, was unable to degrade the defined substrate efficiently in either Mg 2ϩ or Mn 2ϩ (Fig. 1,  lanes 15-18; and Fig. 2, lane 15). Even when almost 16-fold more D524N enzyme was analyzed, no activity on the defined substrate was observed in the presence of Mn 2ϩ (Fig. 1, compare lane 16 for D524N with lane 7 for the wild type). Likewise, mutant Y586F, assayed as a virion-associated enzyme, was almost completely inactive in Mg 2ϩ and Mn 2ϩ (Fig. 5, lanes  5-7, 10, and 11). At a 1:128 dilution, where the wild type was able to degrade the substrate efficiently, Y586F was completely inactive (Fig. 5, lane 11); at a 1:32 dilution, the mutant was slightly active in Mn 2ϩ (lane 10). As mutant Y586F has a tendency to revert to a more active RNase H form during cell culture (24), this slight activity may be the result of trace amounts of this reverted enzyme contaminating this viral preparation. It is clear that D524N and Y586F were essentially inactive RNases H in the presence of both cations. These results suggest that the differential activity of ⌬5E and ⌬C in the presence of Mg 2ϩ is a defect specific to these two, and not a phenotype intrinsic to all RNase H mutants.
Mutants S526A, R657S, and Y598V were previously shown to be active in the in situ assay and are able to degrade their genomic RNA as well as the wild type in the endogenous reaction (Table I) (19,20). When these mutants were analyzed as virion-associated RTs on the defined substrate, all had wildtype activity in Mn 2ϩ (Fig. 3, lanes 5-13). In the presence of Mg 2ϩ , however, only partial activity was observed, and a novel intermediate degradation species was detected (Fig. 4, lanes  7-18). The double-stranded substrate migrating at 240 base pairs disappeared when the mutants were assayed at the 1:8 enzyme concentration, but instead of releasing the singlestranded product corresponding to complete RNA digestion, these mutants appeared to chase much of the hybrid to a form that migrated only slightly more slowly (Fig. 4, compare lane 4 for the wild type to lanes 9, 13, and 17 for the mutants). Mutant ⌬5E left the substrate completely double-stranded at this 1:8 concentration and at higher concentrations, ruling out the possibility that this was due to a contaminating background activity (Fig. 5, lanes 1-4). We suspect that this form corresponds to a discrete DNA-RNA species, but cannot estimate how much RNA remains associated with the DNA. S526A did degrade the substrate to the fully single-stranded form when increasing amounts of enzyme were assayed (Fig. 4, lane 7). The intermediate form can also be seen for the wild type at very low enzyme concentrations (Fig. 4, lane 6). These results suggest that while these mutants are active RNases H, they may be less effective than wild-type RT at removing all the RNA from these substrates in Mg 2ϩ .
Rescue of DNA Polymerase Activity of Mutant H7 by Mn 2ϩ -RNase H mutants of RT analyzed in the endogenous reverse transcription assay frequently show defects in DNA elongation (16,20,21). The endogenous assay, typically performed in Mg 2ϩ , permits the detection of various DNA intermediates using, as a template, the endogenous viral RNA in permeabilized virion particles. (Ϫ)-Strand strong stop DNA is readily detected in reactions with wild-type virions, and some RNase H mutants cannot even complete synthesis of this DNA (20).  (19). The values for RNase H activity in the defined substrate assay are as follows: wt, wild-type degradation of substrate, 5%, low but detectable activity, -, no detectable activity, *, intermediate degradation species detected. The nature of the (Ϫ)-strand strong stop DNA (Ϫsss) in the endogenous reverse transcription reaction was determined previously (20). "Hybrid" corresponds to (Ϫ)strand strong stop DNA in hybrid form with the genomic RNA, and "ss" corresponds to single-stranded (Ϫ)-strand strong stop DNA. Treatment of the endogenous reaction products with RNase A under either low or high NaCl conditions allows further differentiation of whether the (Ϫ)-strand strong stop DNA is in single-or double-stranded form. Treatment with RNase A in high salt will only degrade single-stranded RNA, leaving intact any RNA remaining in hybrid form with the (Ϫ)-strand strong stop DNA; treatment in low salt removes all the associated RNAs (16,20).

In vitro
As described previously, mutant ⌬5E synthesizes full-length (Ϫ)-strand strong stop DNA, but this species remains in hybrid form with the genomic RNA (Table I) (20). Treatment of this species with RNase A under high salt concentrations produced a perfect duplex, termed the FF (fast form) DNA, which migrated at a characteristic position on a nondenaturing polyacrylamide gel (Fig. 6, lane 7). When this FF species was denatured, a 163-nucleotide form was detected, corresponding to the 145-nucleotide (Ϫ)-strand strong stop DNA plus 18 nucleotides of the tRNA primer that were resistant to RNase A treatment in high salt due to base pairing with the genomic PBS sequences (Fig. 6, lane 9). A small amount of a smaller species was also visible. Treatment with RNase A in low salt degraded all of the genomic RNA, allowing detection of the 145-nucleotide (Ϫ)-strand strong stop DNA species free of any RNA (Fig. 6, lane 8) as well as a small amount of a shorter product. These experiments show that the assays are functioning as expected and provide marker DNAs for the various products.
RNase H null mutants do not make full-length strong stop DNA because of defects in the processivity of RT (20,21). Mutant H7 has a frameshift between the DNA polymerase and RNase H domains and is thus effectively a single domain DNA polymerase, i.e. an RNase H null mutant (9,16). This mutant produced predominantly a truncated form of (Ϫ)-strand strong stop DNA, termed VFF for "very fast form," in the endogenous reaction in Mg 2ϩ (Fig. 6) (20,21,25). Little or no completed full-length DNA was detected after RNase A treatment in high salt (Fig. 6, lane 1). When the VFF species was denatured, very little of the 163-nucleotide product was detected, confirming the tendency of this mutant to pause prematurely (Fig. 6, lane  3). However, assays using the other cation showed that mutant H7 was indeed able to synthesize full-length (Ϫ)-strand strong stop DNA in Mn 2ϩ . Treatment of the Mn 2ϩ reaction products with RNase A in high salt produced the FF species (Fig. 6, lane  4); after denaturation, the 163-nucleotide form was detected (lane 6). Treatment with RNase A in low salt produced the 145-nucleotide form (Fig. 6, lane 5). Thus, the DNA polymerase activity of the mutant was enhanced in Mn 2ϩ such that efficient formation of the (Ϫ)-strand strong stop DNA was induced. The signature of this DNA, the FF species, was in doublestranded form. Thus, although Mn 2ϩ improved the ability of the DNA polymerase to synthesize the full-length strong stop DNA, it could not restore RNase H activity to this null mutant. DISCUSSION Previous analyses indicated that the mutant RTs ⌬5E and ⌬C had considerable RNase H activity in vitro as assayed by in situ gel techniques, but left the (Ϫ)-strand strong stop DNA in hybrid form in the endogenous reaction. The experiments described here show that ⌬C and ⌬5E could efficiently degrade an RNA-DNA substrate with Mn 2ϩ but not with Mg 2ϩ as divalent cation (Table I). As Mg 2ϩ is probably the biologically relevant divalent cation, these two mutants are effectively RNase H null mutants in vivo, consistent with the observed loss of infectivity and reduction in strand translocation in the mutant viruses (20).
The effects seen in these two mutants are more extreme versions of effects seen with other enzymes and mutants. In the defined substrate assay, wild-type MMLV RT RNase H was ϳ16-fold more active in the presence of Mn 2ϩ compared with Mg 2ϩ , assayed both as a purified recombinant protein and as a detergent-permeabilized virion-associated enzyme. Differential activities with the two cations have been observed for other RNases H in other assays as well. The HIV-1 RNase H single domain, expressed as a hexahistidine-tagged fusion protein independently of the DNA polymerase domain, similarly exhibits Mn 2ϩ -but not Mg 2ϩ -dependent RNase H activity (26). Furthermore, the addition of an appropriate C-helix into an inactive single domain version of the HIV-1 RNase H, which normally lacks this helix, restores Mn 2ϩ -but not Mg 2ϩ -dependent activity (27,28). However, it should be noted that E. coli RNase H and HIV-1 RNase H as assayed in the intact RT prefer Mg 2ϩ for optimal catalysis.
Analogous divalent-dependent behavior has been reported for many nucleic acid-binding enzymes (e.g. Refs. 29 and 30; very recently, Ref. 31). The basis for the differential activity of these enzymes with the two divalent cations is unclear, but could be due to aspects of the cation binding with the enzyme or with the substrate. Determination of cation requirements is FIG. 6. RNase A treatment in high and low salt of endogenous reaction products synthesized in Mn 2؉ . The endogenous reaction was performed as described under "Experimental Procedures." Lanes 1-6, H7; lanes 7-9, ⌬5E. Lanes 1-3 and 7-9 were performed in the presence of 6 mM MgCl 2 . Lanes 4 -6 were performed in the presence of 2 mM MnCl 2 . The endogenous reaction products were treated with RNase A in high (H; lanes 1, 3, 4, 6, 7, and 9) or low (L; lanes 2, 5, and 8) salt buffer prior to electrophoresis on 8% nondenaturing polyacrylamide gels. Lanes 2,3,5,6,8,and 9 were denatured prior to loading by suspension in dye-containing formamide. Lanes 1, 4, and 7 were analyzed without denaturation. The VFF and FF forms are labeled. nt, nucleotide. complicated by the fact that they bind not only to enzymes, but also to nucleic acids and dNTPs, leading to different template and substrate complexes depending on the cation used (32). There is some evidence that these mutant enzymes, assayed in Mg 2ϩ , may be affected in substrate binding. It is interesting to note that the ⌬C mutation removes a basic handle region that has been implicated in preferential substrate binding (21,33). Mutation and substitution of the lysines in this region in the E. coli RNase H enzyme raise the K m without changing the V max (33). However, the ⌬5E mutation removes a loop between the fifth ␤-sheet and the last ␣-helix, near a conserved histidine residue that has been implicated in catalysis rather than substrate binding (34,35).
Mn 2ϩ and Mg 2ϩ may bind to RNase H in two different positions. While the crystal structures of several RNases H are known, it remains unclear whether one or two divalent cations are involved in catalysis. A two-cation mechanism, similar to that for the 3Ј 3 5Ј exonuclease of DNA polymerase I, has been suggested for E. coli RNase H (10,36). Additionally, crystallographic analysis of the HIV-1 RNase H single domain revealed two divalent cations (Mn 2ϩ ) bound at the enzyme's active site (12). But in other studies, a single Mg 2ϩ was observed bound at the active site of E. coli RNase H (11,37). As all of these structures were determined in the absence of substrate, it is difficult to ascertain whether the observed cation binding is productive or rather inappropriate binding that would be altered when the substrate was present. Co-crystallization of RNase H with an RNA-DNA substrate would help address this issue.
We also detected significant differences between RNase H activities measured in the in situ gel assay and the defined substrate assay even when both assays were carried out with Mn 2ϩ . D524N had ϳ10% activity in the in situ gel assay, but was inactive in the defined substrate assay, and conversely, S526A was fully active in the defined substrate assay, but had only 25% activity in the in situ assay. During the in situ assay, the gel is typically allowed to renature for several days, during which time the enzyme is constantly surrounded by substrate. Thus, this assay may not be sensitive to subtle defects in affinity. An additional difference that may account for the discrepancies between the two assays is the fact that proteins must renature in the in situ gel assay, and mutations may specifically affect this step.
With amounts of enzyme sufficient for wild-type RT to fully degrade the RNA from the radiolabeled DNA substrate in the RNase H defined substrate reaction, mutants S526A, R657S, and Y598V showed some activity in Mg 2ϩ , but were unable to degrade all of the RNA from the substrate. The full-length double-stranded species completely disappeared, and a new discrete species was detected. While we cannot tell how much RNA was removed by these mutants, the major product was the same for all three, and at low concentrations, this product was detected with wild-type RT. This result suggests that these mutants may be less processive RNases H than the wild type or that they only cleaved efficiently up to a specific site in the RNA. The structural features of the substrate that might determine the accumulation of this intermediate are unknown. It should be noted that this cleavage is not due to background activity seen for all the mutants: similar products were not seen with mutant ⌬5E, H7, or Y586F. The defects seen in this assay may be among those responsible for the delayed replication of viruses carrying these mutations (20).
In the endogenous reaction, mutants S526A, R657S, and Y598V were able to degrade their genomic RNA completely, like wild-type RT. What then is the difference between the endogenous reaction and the RNase H defined substrate reac-tion? One main difference is the presence of a more intact capsid in the endogenous reaction and the resulting increased effective concentration of other viral proteins. Virions were permeabilized with 0.01% Nonidet P-40 in the endogenous reaction and were more fully lysed with 0.3% Nonidet P-40 in the in vitro defined substrate reactions. Thus, processive RNase H activity may be effected by the presence of other retroviral proteins (most notably NC, the basic single-strand nucleic acid-binding protein), which are diluted out during the defined substrate reaction. Although we do not favor the idea, we cannot rule out the possibility that the mutant enzymes are simply more sensitive to high Nonidet P-40 concentrations.
A second significant difference between the two assays is the fact that DNA polymerization and RNA degradation may occur simultaneously during the endogenous reaction. It is possible that these mutants are more active in degradation when it is coupled to polymerization, a feature that we are not testing during the defined substrate reaction. There is evidence that RNase H may behave differently when coupled with polymerization (38,39). Further analysis of these mutations may help us understand their differential effects in the endogenous and defined substrate assays, in an attempt to understand polymerization-dependent and -independent RNase H activity.
Previous work has shown that the DNA polymerase activity of most RTs, including HIV-1, Rous sarcoma virus, and MMLV RTs, prefers Mg 2ϩ for full activity. The DNA polymerase activity of MMLV RT is unusual in preferring Mn 2ϩ over Mg 2ϩ as a divalent cation (40 -43). MMLV RT shows a reduced rate of synthesis in Mg 2ϩ , and the DNA products are also generally shorter, suggesting that the enzyme may be less processive (Ref. 40 and data not shown). Mutations in the RNase H domain can also affect the DNA polymerase activity and can particularly reduce its ability to form long products (25). In these studies, we found that Mn 2ϩ is able to qualitatively influence DNA polymerase processivity during the endogenous reaction. In the presence of MnCl 2 , the normally nonprocessive mutant H7 was converted to a more processive form, efficiently synthesizing full-length (Ϫ)-strand strong stop DNA. Thus, assays of MMLV RT in Mn 2ϩ show alterations in both the DNA polymerase and RNase H activities, generally showing enhanced activity and masking significant defects. These results suggest that RT should be assayed in Mg 2ϩ to detect biologically significant effects with the greatest sensitivity.