Localization of the Active Site of HIV-1 Reverse Transcriptase-associated RNase H Domain on a DNA Template Using Site-specific Generated Hydroxyl Radicals*

Reverse transcriptase (RT)-associated ribonuclease H (RNase H) can cleave both the RNA template of DNA/RNA hybrids as well as double-stranded (ds) RNA. This report shows that human immunodeficiency virus (HIV)-RT can also cleave the template strand of dsDNA when Mg2+ is replaced by Fe2+ in the RNase H active site of HIV-RT. The cleavage mechanisms as well as the positions of the cut vary depending on whether RNA or DNA is used. While DNA is cleaved 17 base positions upstream of the primer 3′-end, RNA is cleaved 18 base positions upstream. Competition experiments show that Fe2+ replaces the catalytically active Mg2+ of RT-associated RNase H. The bound Fe2+ is the source of locally generated OH-radicals that cleave the most proximate base in the DNA. Electrophoretic mobility studies of the cleaved fragments suggest that DNA is cleaved by an oxidative mechanism, while RNA is cleaved by an enzymatic mechanism which is indistinguishable from the Mg2+-dependent cleavage. The Fe2+-dependent cuts can be used to trace the active site of RT-associated RNase H on dsDNA as well as on dsRNA and DNA/RNA hybrids. The observed 1 base difference in the cleavage positions on DNA and RNA templates can be attributed to conformational differences of the bound nucleic acids. We suggest that the lower pitch of dsRNA and DNA/RNA hybrids compared with dsDNA permits accommodation of an additional base pair in the region between the primer 3′-end and the Fe2+-dependent cleavage position at the RNase H active site.

Retroviral reverse transcriptases (RTs) 1 are multifunctional enzymes having an RNA-dependent and a DNA-dependent polymerase activity and a ribonuclease H (RNase H) activity, which degrades the RNA strand of RNA/DNA hybrids (1). These activities facilitate the conversion of single-stranded genomic RNA to the double-stranded proviral DNA which is integrated into the host genome. Synthesis of the first DNA strand, the minus strand DNA, is initiated from a cellular tRNA which binds with its 3Ј terminus to the complementary primer-binding site near the 5Ј-end of the viral RNA (2)(3)(4). The newly synthesized minus-strand DNA serves in turn as template for the synthesis of the second DNA strand or plusstrand. Prerequisite for the plus-strand synthesis is removal of the viral RNA from the minus-strand DNA. Both activities act simultaneously, presupposing a spatial and temporal interdependence of the active sites.
The interplay of the two activities has been studied extensively in RT of human immunodeficiency virus type 1 (HIV-1) (5)(6)(7)(8)(9)(10)(11)(12)(13). It has been demonstrated that the viral RNA template is cleaved 18 nucleotides upstream with respect to the 3Ј-end of the nascent primer terminus (7) during minus-strand synthesis. But cleavage does not occur after each nucleotide incorporation step, indicating that the two activities, the polymerase and RNase H activity, are not coupled in a 1:1 mode (11)(12)(13)(14). Gopalakrishnan et al. (7) suggested a temporal coordination of the two activities due to kinetic coupling. Cleavage can occur if the rate of DNA polymerization is lower than the rate of RNA hydrolysis. A similar concept might also explain the resistance of dsRNA toward cleavage. It has been shown that dsRNA formed by the tRNA Lys-3 and the viral RNA during initiation of minus-strand synthesis remains intact during DNA synthesis, but is cleaved, if the DNA synthesis is artificially stopped through use of chain terminating nucleotides (9). This can be explained by the fact that the rate of polymerization is higher than the rate of RNA template hydrolysis of dsRNA in line with the kinetic coupling model (9). Ribonuclease activity on dsRNA, which is believed to be mediated by the active site of the RNase H domain (9,15), has been termed RNase H * activity (16).
The high resolution crystallographic model of an RT-dsDNA primer/template complex has revealed that the space between the polymerization site and the RNase H site can be filled by a primer/template of 17 or 18 bases (17).
In the present study a biochemical approach has been applied, namely site-specific hydroxyl radical cleavage to localize the active site of HIV-1 RT-associated RNase H domain on a DNA template. This approach is based on findings by Metzger et al. (18) who showed that a hypersensitive cleavage site appears in the [Fe(EDTA)] 2Ϫ footprint close to the position where the RNase H active site is located at the template. This cleavage reaction requires an intact RNase H domain, since hyperreactivity was not observed with an RNase H-deficient enzyme (HIV-1 RT(E478Q)), whose ability to bind divalent metal ions is supposed to be suppressed (19). Hyperreactivity was previously interpreted as an enhanced accessibility of the template due to a putative conformational change of the DNA induced by the RNase H domain (18). This report shows that hyperreactivity is due to site-specific generated hydroxyl radicals released by Fe 2ϩ -ions after replacement of Mg 2ϩ at the RNase H active site.
Nucleic Acids-DNA was chemically synthesized using the phosphoamidite method. To remove impurities, oligonucleotides were electrophoretically purified using 12% polyacrylamide, 7 M urea gels containing 50 mM Tris borate, pH 8.0, 1 mM EDTA. DNA was visualized by UV shadowing at 254 nm and the fragment of correct length was eluted from excised gel slices with 0.5 M ammonium acetate, pH 6.5, 0.1% SDS. After ethanol precipitation, concentration of nucleic acids was determined spectrophotometrically.
RNA (designated as primer-binding site 1 by Götte et al. (9) was synthesized by in vitro transcription using T7 RNA polymerase. Purification of the transcript was achieved as described above. tRNA Lys-3 with the anticodon SUU (S ϭ 5-methoxycarbonyl-2-thiouridine) was isolated from rabbit liver and purified following the procedure from Raba et al. (33).
End labeling of dephosphorylated RNA transcript and the DNA template was conducted with [␥-32 P]ATP using T4 polynucleotide kinase (Boehringer Mannheim) according to the manufacturer's recommendation. 3Ј-End labeling with [ 32 P]pCp and T4 RNA ligase (Pharmacia) was carried out as described (36). To ensure homogeneity, labeled nucleic acids were again electrophoretically purified.

Methods
Formation of RT-Primer-Template Complexes and Polymerization-To achieve quantitative complex formation, primer/template sequences were hybridized before incubating with the enzyme. Annealing was conducted in a reaction volume of 10 ml. A mixture of 100 nM (final concentration) unlabeled DNA template, 50,000 cpm of 5Ј-end-labeled template, and 150 nM DNA primer in a buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl was heated to 90°C for 1 min followed by incubation at 60°C for 10 min and cooling for 20 min to room temperature. Complete hybridization was confirmed by analyzing aliquots of the annealed duplexes on native polyacrylamide gels containing 20 mM Tris-HCl, pH 7.8, 10 mM NaCl. The tRNA Lys-3 /RNA homoduplex and the DNA/RNA heteroduplex were prepared analogously.
In a reaction volume of 20 ml, pre-annealed primer/template substrates (100 nM) were incubated for 5 min at 37°C with HIV-RT (150 nM) or the RNase H-deficient enzyme, respectively, in a buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl. Polymerization was then initiated by addition of MgCl 2 (6 mM) and the appropriate dNTP (100 mM)/ddNTP (500 mM) mixture permitting primer extension by 1, 3, 9, and 18 nucleotides. Under these reaction conditions 95% of the primer is correctly elongated after 10 min. Fe 2ϩ -dependent incorporation of ddATP was initiated by addition of 0.4 mM Fe 2ϩ .
To characterize the influence of liganded versus unliganded Fe 2ϩ on the cleavage profile we changed the ratio of free Fe 2ϩ and [Fe(EDTA)] 2Ϫ by varying the EDTA concentration from 0 to 4 mM while keeping the concentration of all other reaction components constant (Fig. 1). The reaction was allowed to proceed for 5 min at 37°C and stopped with 40 ml of a solution containing 0.1 M thiourea, 200 ng of tRNA, 10 mM EDTA, and 0.6 M sodium acetate. Samples were subsequently precipitated with ethanol and loaded on a 12% polyacrylamide-urea gel. Products were visualized using a PhosphorImager or by autoradiography overnight.
ONOOK-dependent OH-radical Footprinting-A stable alkaline solution of ONOOK, pH 12.5, was synthesized as described (24). The ONOOK concentration was determined as 60 mM by absorbance measurement at 302 nm using an extinction coefficient of ⑀ 302 nm ϭ 1670 M Ϫ1 cm Ϫ1 . The peroxonitrite solution was stored at Ϫ80°C and thawed prior to use. Cleavage reactions with ONOOK were conducted by adding 1 ml of the peroxonitrite solution to the sample solution buffered at pH 7 (24,25). RT-primer-template complexes were prepared in a buffer containing 80 mM sodium kakodylate, pH 7, and 20 mM NaCl. Reactions were stopped and analyzed essentially as described above.
Fe 2ϩ /Mg 2ϩ Competition Experiment-Pre-formed RT-primer-template complexes in a buffer containing 50 mM Tris-HCl, pH 7.8, 50 mM NaCl, and 6 mM MgCl 2 were incubated in a total reaction volume of 25 l with a mixture of Fe(NH 4 )SO 4 ⅐6H 2 O (40 M) and 5 mM DTT. The reaction proceeds for 10 min at 37°C and was stopped as described in the previous paragraph. Replacement of Mg 2ϩ at the active site of RT-associated RNase H domain by Fe 2ϩ was studied by applying MgCl 2 in a concentration range from 0 to 10 mM and by keeping the Fe(NH 4 )SO 4 ⅐6H 2 O concentration constant. The size of the Fe 2ϩ -dependent cleavage products was assigned by a T-ladder generated after modifying the DNA with OsO 4 /bipyridine followed by cleaving the sugarphosphate backbone with piperidine.
RNase H/H * Activities in the Presence of Fe 2ϩ and Mg 2ϩ -RNase H * activity was monitored on the preformed RT-tRNA Lys-3 ⅐RNA complex in the presence of 6 mM MgCl 2 (9). RNase H activity was monitored analogously using the heteroduplex shown in Fig. 6B. To characterize cleavage specificity with Fe 2ϩ the complex was incubated in the absence of MgCl 2 using various concentrations of Fe(NH 4 )SO 4 ⅐6H 2 O (40 -400 M). The band intensities of the cleavage products did not depend on the presence of a reducing agent, such as DTT (data not shown). RNA fragment sizes were assigned using a homogeneous RNA ladder generated by partial alkaline hydrolysis with Na 2 CO 3 /NaHCO 3 , pH 9.5, and a G-ladder obtained by partial digestion of the RNA with ribonuclease T1.
Fe 2ϩ -dependent Hydroxyl Radical Cleavage of HIV-1 RT-To analyze the Fe 2ϩ -dependent hydroxyl radical cleavage products in the protein HIV-RT, dsRNA and dsDNA primer/template (concentration ϭ 100 nM) was incubated with equal amounts of RT (the wild type and the RNase H-deficient enzyme, respectively). In a reaction volume of 20 l, RT and primer/template were incubated with ddATP permitting elongation of the primer by one nucleotide. The complex was subjected to Fe 2ϩ treatment (0.4 mM Fe(NH 4 )SO 4 ⅐6H 2 O) at 37°C for 10 min in the presence of 5 mM DTT. The reactions were stopped by adding 1 volume of SDS-PAGE application buffer. The reaction mixture was heated for 5 min at 95°C and loaded on a 12% SDS-PAGE (34). The gel was stained with Coomassie Brilliant Blue R-250.

RESULTS
To compare the hyperreactive cleavages on a DNA template with the RNase H-induced cuts recently obtained on an RNA template (9), we used a dsDNA fragment having the same sequence ( Fig. 1A). Complexes of primer/template DNA and HIV-RT were formed and subjected to [Fe(EDTA)] 2Ϫ -dependent OH-radical treatment, as described under "Methods." Lane 3 in Fig. 1B shows the OH-radical footprint of HIV-1 RT that was obtained using 5Ј-end-labeled DNA template. The schematic representation of the footprint in Fig. 1A indicates the contacts between HIV-RT and template. The region covered by HIV-RT reaches from base position ϩ4 to Ϫ18 containing a window of accessibility around base position Ϫ10 and a region of hyperreactive cleavage at base positions Ϫ15 to Ϫ18. This pattern essentially agrees with that previously published by Metzger et al. (18). However, that the observed hyperreactivity reflects an enhanced accessibility of the template DNA due to an RNase H-induced distortion of the template does not seem to be convincing. An accessibility which exceeds that of free DNA is difficult to envision. Therefore, we investigated whether the hypercleavage could be due to another reason. It was previously shown with other Mg 2ϩ -binding proteins such as tetracycline repressor (20) and E. coli RNA polymerase (21) that Fe 2ϩ can replace Mg 2ϩ and generate site-specific OH-radicals which cleave the DNA. Whether this mechanism is also applicable to the catalytic Mg 2ϩ of the RT-associated RNase H, is discussed below.
Fe 2ϩ Is Responsible for Hyperreactive Cleavage-In a typical OH-radical footprinting experiment, the Fenton reaction is used for generating OH-radicals which cleave the DNA ( Fig.  2A). The OH-radicals most likely attack the nucleotide sugars at the C-1 and C-4 positions, leading to excision of the base. EDTA is used in a 2-fold excess over Fe 2ϩ . Thus the negatively charged [Fe(EDTA)] 2Ϫ complex is obtained which prevents interaction of iron with the sugar phosphate backbone of the DNA (22). The Fe 2ϩ -dependent OH-radical footprint then reflects the accessibility of DNA toward diffusable OH-radicals generated by the metal complex. Hyperreactive cleavage is observed only when Fe 2ϩ is in excess. The role of excess Fe 2ϩ was analyzed by shifting the equilibrium between Fe 2ϩ and [Fe(EDTA)] 2Ϫ in the Fenton reaction. This was achieved by varying the EDTA concentration (see Fig. 2A). The role of Fe 2ϩ in the hyperreactive cleavage reaction was analyzed using a different, Fe 2ϩ -free OH-radical source for footprinting, namely potassium peroxonitrite (ONOOK) (23)(24)(25). The reaction scheme is depicted in Fig. 2B. The free acid has been disproportionated at neutral pH facilitating generation of OH-radicals. Fig. 3 shows a comparison of footprints obtained with ONOOK and [Fe(EDTA)] 2Ϫ , respectively. The latter was performed at high EDTA concentration to suppress hyperreactivity. The footprints obtained by both methods yield the same protected regions. The lack of enhanced cleavage at the RNase H position in the ONOOK footprint confirms that free Fe 2ϩ is responsible for the enhanced cleavage activity. The slight cleavage enhancement visible in the ONOOK pattern at position Ϫ6 might be due to an interaction of the protein with an active intermediate generated when disproportionating ONOOK (25).
Fe 2ϩ Displaces Mg 2ϩ at the RNase H Active Site-To prove that hypercleavage is due to a replacement of Mg 2ϩ , a competition experiment using wild type RT as well as the RNase H-deficient RT(E478Q) mutant enzyme was performed as described under "Methods." Fe 2ϩ was kept constant and the Mg 2ϩ concentration was varied. To obtain cleavage at a single position the standard conditions were slightly changed. EDTA was  2Ϫ and maintain the production of OH-radicals (22). B, the potassium peroxonitrite reaction. OH-radicals were generated by disproportioning of the free acid at neutral pH (23). omitted and the concentration of Fe 2ϩ was reduced from 400 to 40 M. Due to these changes only hypercleavage at position Ϫ17 and no OH-radical footprint was observed in contrast to the previously shown Fe 2ϩ assays ( Figs. 1 and 3). The result of the Fe 2ϩ /Mg 2ϩ competition experiment is shown in Fig. 4. The yield of hypercleavage decreases with increasing Mg 2ϩ concentration indicating that Mg 2ϩ displaces Fe 2ϩ .
The hypercleavage reaction in the [Fe(EDTA)] 2Ϫ assay was dramatically reduced, but not to zero (Fig. 4), when the RNase H Ϫ RT mutant (E478Q) (19) was used. The crystal structure model of the RNase H domain (26) indicates that the amino acid replacement in the mutant leads to a reduced affinity of the catalytically active Mg 2ϩ . This crystallographic finding along with our result that the same mutation in the RNase H affects the affinity for Fe 2ϩ , as well as for Mg 2ϩ , suggests that the binding sites for both ions are overlapping, if not identical.
The DNA Template Is Cleaved 17 Nucleotides Upstream of the Primer 3Ј-End-To determine the exact position of the hypersensitive cut, HIV-1 RT was arrested at different positions during synthesis by using an incomplete set of dNTPs along with a chain terminating ddNTP (Ref. 18, "Experimental Procedures"). After the primer was specifically extended by 1, 3, 9, and 18 nucleotides (Fig. 5A), complexes were subjected to treatment with Fe 2ϩ -generated OH-radicals. Determination of the cleavage position was facilitated by lowering the concentration of Fe 2ϩ so that only a single nucleotide was cleaved. The cleavage positions were assigned by means of a T-ladder. Thymidines of the single-stranded template strand were modified by OsO 4 -bipyridine, followed by strand scission with piperidine, as described under "Experimental Procedures." The pattern in Fig. 5B and the schematic representation in Fig. 5A shows that the template DNA strand is cleaved 17 base positions upstream of the primer 3Ј-end in all registers, indicating that the 17-base distance is sequence-independent.
The 15-base pair distance previously described by Metzger et al. (18) can be ascribed to an error in their base assignment. In that study higher [Fe(EDTA)] 2Ϫ concentrations were used, resulting in a larger spread of the cleaved nucleotides (see also Fig. 1B, lanes 1 to 3).
We conclude from the above finding that the Fe 2ϩ -dependent hyperreactive cut can be used to trace the position of the Mg 2ϩ -binding site of the RT-associated RNase H domain. Mg 2ϩ is replaced by Fe 2ϩ , which can generate a high local concentration of OH-radicals, resulting in enhanced cleavage of the nucleotide next to the bound Fe 2ϩ . The assumption that hyperreactivity is mediated by an oxidative mechanism is supported by the finding that a reducing agent (DTT) is necessary to maintain cleavage (data not shown). Reduction of Fe 2ϩ is a prerequisite for permanent generation of OH-radicals, as is the case for OH-radical production in solution by the [Fe(EDTA)] 2Ϫ -dependent Fenton reaction (22).
Analysis of the electrophoretic mobility of the 5Ј-end-labeled cleavage products further confirms an oxidative cleavage mechanism. As depicted in Fig. 5B, cleaved fragments migrate at the same position like the DNA fragments of the T-ladder. The latter was obtained by strand scission with piperidine, which gives rise to 3Ј-PO 4 groups as proposed for OH-radical cleavage (22,27).
Comparison of Cleavage Positions on RNA and DNA Templates-The observation that the Fe 2ϩ -dependent cut in different registers occurs at a constant distance from the primer 3Ј-end is reminiscent of the RNase H * -induced cuts obtained when analyzing complexes stalled on dsRNA during initiation of minus-strand synthesis (9). These studies showed that the RNA template of the tRNA Lys-3 /primer-binding site duplex was cleaved in the presence of Mg 2ϩ 18-base positions upstream of the primer 3Ј terminus, while DNA used in the present study was cleaved by the Fe 2ϩ -substituted HIV-RT 17-base positions upstream.
Whether the cleavage position depends on the type of metal ion used was analyzed by cleavage studies using the same RNA primer/template, tRNA Lys-3 /primer-binding site, in the presence of Mg 2ϩ as well as of Fe 2ϩ . Cleavage was performed using wild type RT. The conditions were the same as described in the previous experiment. The cleavage products were analyzed by means of a 5Ј label in the template RNA which provides information about the chemical nature of the 3Ј terminus by gel electrophoretic mobility studies, as described above. The obtained cleavage patterns are depicted in Fig. 6A. Comparison of lane 3 with lanes 4 and 5 in Fig. 6A shows that the electrophoretic mobility of the major cleavage products obtained in the presence of Fe 2ϩ and Mg 2ϩ are the same, indicating that cleavage occurred at the same base position and that the 3Јends are identical.
Size and nature of the 3Ј-ends of the cleavage products were characterized by electrophoresis comparing their mobility with that of reference fragments obtained by alkaline hydrolysis of the template. Lane 2 in Fig. 6A shows that the RT-dependent cleavage products migrate between position Ϫ18 and Ϫ19, indicating a difference of the 3Ј-ends. Alkaline hydrolysis products contain 3Ј-PO 4 ends, like those generated with RNase T1 (lane 1 in Fig. 6A), thus migrating a little faster than fragments containing 3Ј-OH ends.
We conclude from the above findings: first, that the RNase H activity of HIV-1 RT cleaves the RNA template at position Ϫ18 with Fe 2ϩ as well as with Mg 2ϩ leaving a 3Ј-OH group, and second, that Fe 2ϩ can replace the catalytically active Mg 2ϩ at the RNase H active site without affecting the enzymatic cleavage mechanism. In other words, the absence of cleavage products containing 3Ј-PO 4 ends suggests that there is no Fe 2ϩ -dependent cleavage due to site-specific generated hydroxyl radicals when dsRNA is used as primer/template. The minor cleavage product at position Ϫ17 appearing in the Fe 2ϩ -dependent assays (lanes 4 and 5 of Fig. 6A) might be attributable to a 5Ј-directed cleavage activity of HIV-RT due to a polymerase-independent cleavage mechanism previously described for the Mg 2ϩ -dependent assay (6, 7, 9 -12).
To visualize Fe 2ϩ -dependent cleavages on a DNA/RNA substrate, we have used a 3Ј-end-labeled RNA template. This allowed us to determine specifically the polymerase-dependent RNase H cut (9), that reflects the binding mode in which RT's polymerase active site is located in the vicinity of the primer 3Ј-end. Cleavage positions were compared on dsRNA and DNA/ RNA hybrids using Mg 2ϩ and Fe 2ϩ , respectively. The electrophoretic pattern shown in Fig. 6B indicates that the main cleavage site at the RNA template is located 18 nucleotides upstream of the primer 3Ј-end, regardless of whether the primer is RNA (lanes 1 and 2) or DNA (lanes 3 and 4). Additionally, the cleavage position is also identical in regard to the type of the metal ion used, although the efficiency of the cleavages is modulated. Fe 2ϩ -dependent cuts are weaker than their Mg 2ϩ -dependent counterparts. If the primer is tRNA Lys-3 (lanes 1 and 2), a single cleavage site is observed. This is in line with previous studies which showed that dsRNA is cleaved in the polymerization-dependent mode 18 bases upstream of the 3Ј-primer position (9). If the primer is a DNA fragment (lanes 3 and 4), more than one cleavage is observed; this is probably due to cleavage of the template in a polymerization-independent mode indicated by small arrows in Fig. 6B.
To elucidate whether Fe 2ϩ may also serve as a catalyzing ion in regard to RT's polymerase activity, we have followed the extension of a 5Ј-end-labeled DNA primer with ddATP using the DNA/DNA substrate depicted in Fig. 1A. Fig. 6C shows that Fe 2ϩ can in fact substitute Mg 2ϩ without largely affecting DNA synthesis of the wild-type enzyme (lanes 3 and 4 compared with 10 and 11) and the RNase H-deficient enzyme (lanes 6 and 7  compared with 13 and 14), respectively. Thus, Fe 2ϩ catalyzes cleavage and polymerase activities, although less efficiently compared with the Mg 2ϩ -dependent reactions.
It is worth noting that the Fe 2ϩ concentrations chosen in this experiment is a compromise to fulfill two requirements. The concentration must be sufficiently high to facilitate the Fe 2ϩdependent enzymatic functions, but below 0.4 mM to avoid attack of DNA or RNA by OH-radicals generated by Fe 2ϩ free in solution.
Fe 2ϩ -dependent OH-radical Cleavage of HIV-RT-The cleavage pattern does not differ if HIV-RT is bound to dsRNA or dsDNA but changes if HIV-RT is analyzed without nucleic acid primer/template. It is surprising that no OH-radical cleavage products are observed with dsRNA as a primer/template in contrast to dsDNA. To exclude the possibility that the lack of oxidative cleavage products in dsRNA is due to a reduced OH-radical production, we analyzed whether and how Fe 2ϩ -dependent OH-radicals cleave also the protein, the HIV-RT.
HIV-RT, wild type, and RNase H-mutant HIV-RT(E478Q), was incubated with dsDNA, dsRNA and without primer/template in the presence of Fe 2ϩ . The incubation and cleavage conditions were the same as in the experiment described in Fig.  1 for nucleic acid cleavage. SDS-PAGE (Fig. 7) was used to analyze the OH-radical cleavage on HIV-RT. Cleavage products are observed in isolated HIV-RT (Fig. 7A, lane 2) and in the complex with dsRNA ( lane 4) and dsDNA (lane 3).
The molecular weight of the cleavage products is in the range of 51 to 66 kDa. The size of the cleavage products indicates that only the p66 subunit is cleaved. This is in line with the observation on dsDNA that OH-radicals are generated solely by Fe 2ϩ bound at the catalytic site of the RNase H site in p66. The cleavage efficiency is dramatically reduced (Fig. 7A, right  panel) if the RNase H Ϫ mutant HIV-RT is used, indicating that intactness of the RNase H site is required for generation of OH-radicals by Fe 2ϩ , which is in accord with the conclusion drawn from dsDNA cleavage experiments by Fe 2ϩ -substituted HIV-RT.
The protein cleavage is much less efficient than nucleic acid cleavage, as a comparison of the cleavage patterns in Fig. 7 and Fig. 1A shows. Protein is obviously less susceptible to OHradical attack than nucleic acid.
HIV-RT is cleaved by OH-radicals to the same extent, regardless of whether dsRNA (Fig. 7A, lane 3) or dsDNA (lane 4) is used. This excludes the possibility mentioned above, that the lack of OH-radical cleavage in dsRNA can be explained by a reduced OH-radical production with dsRNA as primer/template.
The cleavage patterns obtained with dsRNA (Fig. 7A, lane 3) and dsDNA (lane 4) are identical. Two cleavage products are observed having a molecular mass of 55 and 60 kDa. The fact FIG. 6. Cleavage and polymerization activity of HIV-RT in the presence of Fe 2؉ . Fe 2ϩ -substituted HIV-RT was incubated with primer/templates as indicated and the cleavage products were analyzed by gel electrophoresis as described under "Methods." A, cleavage of an RNA template/primer in register 0 using 5Ј-end labeled template. A RNA primer/template having the same sequence as that used for DNA cleavage shown in Fig. 1a was used except that the primer was tRNA Lys-3 . The intensity scans were obtained from the electrophoretic cleavage pattern shown as an inset. Lanes 4 (40 M Fe 2ϩ ) and 5 (400 M Fe 2ϩ ) show the patterns obtained with different Fe 2ϩ concentrations, namely 0.04 mM (lane 4) and 0.4 mM (lane 5). Lane 3 shows the cleavage pattern obtained with Mg 2ϩ -substituted HIV-RT. The size of the cleavage products was determined using a G-ladder (lane 1) and an OH ladder (lane 2). B, comparison of the cleavage activity of 3Ј-labeled RNA template hybridized with RNA or DNA using Fe 2ϩ and Mg 2ϩ . Fe 2ϩsubstituted HIV-RT was incubated as in the previous experiment using as primer tRNA Lys-3 or a DNA fragment. The DNA primer has previously been described (9). Its sequence corresponds to the first 24 nucleotides of tRNA Lys-3 .  5-7). The same set of experiments performed in the presence of Mg 2ϩ was performed in the presence of Fe 2ϩ (lanes 9 -14).
that the cleavage patterns with dsDNA and dsRNA are the same suggests that the conformation of HIV-RT at the Fe 2ϩbinding site does not change, regardless of whether the primer/ template is dsRNA or dsDNA.
It is interesting to note that the cleavage pattern differs depending on whether HIV-RT is isolated or bound to a nucleic acid primer/template. The OH-radical pattern of isolated HIV-RT shows in Fig. 7A, lane 2, an additional band having a molecular mass of about 57 kDa, indicating a conformational difference of the isolated and complexed HIV-RT around the Fe 2ϩ /Mg 2ϩ -binding site of the RNase H domain.
If the postulated replacement mechanism of Fe 2ϩ and Mg 2ϩ as observed with template cleavage is correct, it should be also possible to suppress the Fe 2ϩ -dependent OH-radical cleavage of the protein by excess of Mg 2ϩ . Fig. 7B shows that the cleavage of the p66 subunit is fully suppressed in the presence of 20 mM Mg 2ϩ (lane 8), indicating that the OH-radical generating Fe 2ϩ can be replaced by Mg 2ϩ .

DISCUSSION
The [Fe(EDTA)] 2Ϫ -dependent OH-radical footprint of HIV-RT on dsDNA shows hyperreactive cuts (18) at the upstream edge of the footprint. To determine whether this cut can be used to trace the position of the active center of RT-associ-ated RNase H, we have analyzed the mechanism of the hypercleavage reaction. We show that the template DNA is cleaved by OH-radicals which are generated at the metal-binding site of the RNase H active site after replacement of Mg 2ϩ by Fe 2ϩ .
The conclusion that unliganded Fe 2ϩ is responsible for the hypercleavage is based on the finding that the hyperreactive cleavage is observed only in the [Fe(EDTA)] 2Ϫ -dependent OHradical footprint and not in the Fe 2ϩ free (ONOOK)-dependent OH-radical footprint; the intensity of the hyperreactive cleavage depends on the availability of free Fe 2ϩ in the Fenton reaction.
The conclusion that the hyperreactive cleavage is facilitated by OH-radicals is supported by the finding that cuts are spread over a region of four to five nucleotides centered around the main cleavage site at position Ϫ17, in line with the assumption that OH-radicals diffuse from the Fe 2ϩ source and cleave the most proximate base most efficiently; analysis of the cleavage product, which shows that the 3Ј-end contains PO 4 group as expected for hydroxyl radical mediated cuts (28,29). That a reducing agent such as DTT is required further supports the view that hyperreactivity is mediated by an oxidative process (data not shown).
The conclusion that Fe 2ϩ binds at the metal-binding site of RT-associated RNase H active site is based on the findings that hyperreactive cleavage is not observed with the RNase H Ϫ mutant (E478Q) in which the Mg 2ϩ -binding site is destroyed by replacement of glutamic acid 478 by glutamine (6); that the intensity of the hyperreactive cleavage can be modulated by the Fe 2ϩ /Mg 2ϩ ratio in the assay. Excess Mg 2ϩ reduces the hypercleavage, indicating that Fe 2ϩ and Mg 2ϩ probably compete for the same site. But it is not clear whether one or both of the two Mg 2ϩ ions evidenced by the crystal structure of RNase H (26) can be replaced by Fe 2ϩ .
Mg 2ϩ ions are essential not only for the function of the RT-associated RNase H active site but also for the DNA polymerization site. In contrast to other Mg 2ϩ -dependent polymerases, such as E. coli RNA polymerase (21), HIV-RT shows no hypercleavage of the template at the polymerization site after replacement of Mg 2ϩ by Fe 2ϩ . That Mg 2ϩ is replaced in HIV-RT also at the DNA polymerization active site is clear from DNA polymerization studies in the presence of Fe 2ϩ . DNA polymerization takes place, but with lower efficiency, if 0.4 mM Fe 2ϩ is used instead of the standard 6 mM Mg 2ϩ (Fig. 6C).
There is no obvious reason for the lack of hyperreactive cleavage of the template at the polymerization site by OHradicals. It might be that Fe 2ϩ at the polymerization site has a lower redox potential or that the accessibility of Fe 2ϩ by oxygen or diffusion of H 2 O 2 is reduced due to sterical reasons.
Our studies show that hypercleavage on DNA can be spatially and mechanistically linked with the Fe 2ϩ substitution at the metal-binding site of the RT-associated RNase H active site. This permits conclusions to be drawn about the orientation of HIV-RT on a dsDNA. The orientation has been controversially discussed, since the crystal structure of rat DNA polymerase b with a DNA primer/template and ddCTP showed that RT binds its dsDNA substrate in the opposite orientation (30). In this binding mode, the RNase H domain would contact the single-stranded template downstream of the primer terminus. Our hypercleavage study excludes such a binding mode for HIV-RT in line with previous results from differential DNase I footprinting studies using murine leukemia virus RT with and without RNase H domain by Wöhrl et al. (31).
HIV-RT appears rather flexible with respect to the requirements of the type of metal ions catalyzing the cleavage reaction on RNA. Mg 2ϩ as well as Fe 2ϩ is accepted. The cleavage mechanism is an enzymatic mechanism for both metal ions, as  , lanes 2-4), and RNase H Ϫ mutant (E478Q) (right panel, lanes 6 -8), were incubated with dsRNA primer/ template (lanes 4 and 8), with dsDNA primer/template (lanes 3 and 7), and without primer/template (lanes 2 and 6), treated by Fe 2ϩ -dependent hydroxyl radicals, as described under "Experimental Procedures," and subjected to SDS-PAGE. Lanes 1 and 5 show the wild type and RNase H Ϫ -mutant, respectively, without OH radical treatment, as a reference. The subunits and the cleavage products were stained by Coomassie. B, competition experiment with Mg 2ϩ . The Fe 2ϩ -substituted HIV-RT has been incubated with dsDNA primer/template as shown in Fig. 1A  indicated by the analysis of the 3Ј termini of the cleavage products which is a 3Ј-OH (Fig. 6A in this study and Götte et al. (9)). It is not quite clear why no oxidative cleavage products by OH-radicals are observed with the Fe 2ϩ -substituted HIV-RT with RNA. We can exclude the possibility that no OH-radicals are generated if dsRNA is the template, since protein cleavage is observed regardless of whether the template/primer is dsRNA or dsDNA. The reason for the absence of oxidative cleavage products could be due to greater efficiency of the enzymatic cleavage process and/or due to the fact that OHradicals cleave at the same position.
The cleavage on dsDNA and dsRNA differs with respect to both position and mechanism, although the same catalyzing ion, namely Fe 2ϩ , was used for both primer/templates. It is clear that dsDNA can only be cleaved by an oxidative process, since enzymatic cleavage is not possible on DNA. DsRNA is cleaved essentially by an enzymatic mechanism, although OHradicals are generated. This can be explained by the fact that the oxidative cleavage process is less efficient than the enzymatic process.
The most interesting question is why DNA and RNA are cleaved at different positions. While the template strand of dsDNA is cleaved 17 base positions upstream of the 3Ј-primer terminus, dsRNA and DNA/RNA is cleaved 18 base positions upstream.
It is unlikely that the difference in the position can be attributed to difference in the cleavage mechanisms for reasons pointed out above or due to a conformational difference of HIV-RT when bound to dsDNA, dsRNA, or DNA/RNA. The latter is supported by the finding that the OH-radical cleavage pattern of HIV-RT is the same with dsDNA and dsRNA. The most likely explanation for the difference in the cleavage position is a conformational difference of the RNA and the DNA primer/template in the complex with HIV-RT. The high resolution structure model of HIV-RT in complex with a dsDNA fragment (17) shows that the template/primer consists of two segments, a 7-base pair region in A-type conformation located at the 3Ј-primer terminus and a 11-base pair region in B-type conformation further upstream having a kink in between. It was suggested that the reason for the B-like conformation of the upstream located in the 11-base pair segment is its enhanced solution accessibility and the reason for the A-like conformation of the downstream located in the 7-base pair region is its strong interaction with HIV-RT (17). Based on the data of our study we suggest that dsRNA and DNA/RNA have an A-like conformation not only in the 7-base pair downstream region but also in the 11-base pair upstream region. This suggestion is supported by the postulated solution accessibility of the upstream region which would mean that dsRNA and DNA/ RNA adopts in this region the canonical solution structure which is the A-conformation (35,37,38). Since A-RNA has a lower pitch than B-DNA (about 30 Å and 34 Å, respectively (35), it permits the accommodation of one additional base pair within a helical turn. This interpretation is in line with our observation that dsDNA primer/template is cleaved 1-base position closer to the 3Ј-end than dsRNA and DNA/RNA. Therefore, we suggest that differences in the cleavage pattern obtained by Fe 2ϩ -dependent enzymatic and oxidative cleavage of RNA and DNA can be used to trace differences of the helical pitch of the primer/template region between the polymerization site and the RNase H site.