The putative substrate recognition loop of Escherichia coli ribonuclease H is not essential for activity.

The RNase H family of enzymes catalyzes the hydrolysis of RNA from RNA·DNA hybrids in a divalent metal-dependent fashion. To date, structure/function studies have focused on two members of this family: Escherichia coli RNase HI, a small monomeric protein; and human immunodeficiency virus, type I (HIV) RNase H, a domain of HIV reverse transcriptase. The isolated RNase H domain from HIV reverse transcriptase can be expressed independently and shares significant structural homology with its E. coli homologue; however, unlike the bacterial protein, it is inactive. The most notable difference between the inactive domain from HIV and the active E. coli protein is a basic helix/loop sequence, present in E. coli but absent from the HIV homologue. Substitution of this basic region into the HIV domain partially restores its activity and increases its thermodynamic stability. By deleting the basic helix/loop region, we have modeled the structural difference between these two polypeptides onto the E. coli homologue. Surprisingly, the resulting mutant protein is active in Mn2+-dependent fashion. Therefore, the basic helix/loop is not required for RNase H activity.

RNase H, a family of structurally homologous enzymes from both prokaryotes and eukaryotes, selectively hydrolyzes the RNA strand of RNA⅐DNA hybrids in a divalent cation-dependent fashion. The best characterized members of this family are Escherichia coli RNase HI and the human immunodeficiency virus, type I (HIV) 1 RNase H domain (for review, see Ref. 1). E. coli RNase HI is a small (155 residues) single domain protein.
In HIV, however, this essential activity is carried out by a domain of the much larger heterodimeric protein, reverse transcriptase (1,2). In spite of the structural homology between the two (Fig. 1), the HIV domain is inactive when expressed independently of the rest of reverse transcriptase. The structural basis for this lack of activity is unclear.
A structural comparison of these two homologues reveals two primary differences that may contribute to the inactivity of the isolated retroviral domain. (i) Whereas the E. coli RNase H structure is well ordered throughout the protein, the COOHterminal region of the HIV RNase H is dynamic and disordered as shown by NMR (3, 4) and crystallographic (5) studies. This region includes a loop with a conserved histidine and the COOH-terminal-most helix (helix E). Mutational studies in both homologues demonstrate the importance of this region for activity (6 -10). (ii) Comparison of the sequences and structures of these two homologues also reveals a highly basic helix/loop region that is present in E. coli RNase H but missing from the HIV RNase H (5,11). Site-directed mutagenesis studies on the bacterial protein have demonstrated that this basic helix/loop region is important for the Mg 2ϩ -dependent activity of the enzyme (12): neutralizing mutations generally result in an increase in Michaelis constant (K m ). This observation has led to the belief that this region is important for substrate binding and that the inactivity of the isolated HIV RNase H domain is due to poor substrate recognition (5,12). Indeed, on transplantation of this region (residues 79 to 102) into the isolated domain from HIV, a Mn 2ϩ -dependent RNase H activity is restored (13,14). In addition to restoring activity, insertion of this E. coli basic helix/loop stabilizes the HIV RNase H domain (14). The molecular basis for this restoration of activity is unclear: has the basic helix/loop improved the substrate-binding affinity of the domain or its stability?
We have created a variant protein that removes the majority of this basic helix/loop from E. coli RNase H* (RNase H* is a cysteine-free version of E. coli RNase HI (15)). In essence, our deletion mutant models the structural features of HIV RNase H onto the E. coli homologue. This was accomplished by substituting 13 of the 16 residues comprising the basic helix/loop (including 5 basic residues) with a short linker of 6 glycines. The resulting variant protein folds and maintains a Mn 2ϩ -dependent RNase H activity. Hence, the putative substrate-binding loop of E. coli RNase HI is not essential for activity. Comparisons of the metal requirements for RNase H activity have revealed an unexpected and complex Mn 2ϩ -dependent activity. These activity studies help to determine whether this putative substrate recognition loop is essential for RNase H activity, and they further our understanding of the basis for inactivity in HIV RNase H.

EXPERIMENTAL PROCEDURES
Materials-All buffer components were from Sigma unless otherwise specified: ribonucleotides (Boehringer Mannheim); acetylated bovine serum albumin (BSA) (U. S. Biochemical Corp.); and heparin-Sepharose (Pharmacia Biotech Inc.). Restriction endonucleases and T4 ligase (New England Biolabs, Beverly, MA) were used as directed by the suppliers' recommendations. All synthetic oligonucleotides were made on an Applied Biosystems 392 RNA/DNA synthesizer. The HIV RNase H domain was purified as described previously (14). E. coli RNA polymerase was kindly provided by Michael Chamberlin (University of California, Berkeley). Purified E. coli RNase H* (E. coli RNase HI with alanine replacing the 3 free cysteines) was obtained as a gift from Jonathan Dabora (15).
Creation of pJMD101-pJMD101, a gift from Jonathan Dabora, was created by cassette mutagenesis of pSM101, a T7 expression plasmid that encodes for E. coli RNase H* (15). pJMD101 is the product of ligating the BamHI-BglII fragment of pSM101 with a synthetic cassette of two complementary oligonucleotides with BamHI and BglII ends, following standard cloning protocols (16). The resulting plasmid * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  (pJMD101) encodes for a variant of E. coli RNase H* (RNH110) in which residues 83-95 are substituted by 6 glycine residues ( Fig. 1; details of plasmid sequence available upon request). This deletion constitutes the majority of the basic helix/loop (residues 84 -99) as defined by Kanaya et al. (12). The sequence of JMD101 was confirmed by standard sequencing techniques.
Purification of RNH110 -E. coli BL21 (DE3) cells transformed with pLysS (Novagen, Madison, WI) and pJMD101 were grown at 37°C in Luria-Bertani medium (16) with 100 g/ml ampicillin and 25 g/ml chloramphenicol. Cells at midlogarithmic phase (A 600 0.5-0.6) were induced to overexpress RNH110 with 1 mM isopropyl ␤-D-thiogalactopyranoside and were harvested by centrifugation after 3 h of growth. Cell pellets were resuspended in 50 mM Tris, pH 8.0, 20 mM NaCl, 0.5 mM EDTA and lysed by sonication. RNH110 was found in the insoluble fraction of the sonicate (as determined by polyacrylamide gel electrophoresis), presumably in inclusion bodies. Sonication pellets were resuspended in a membrane solubilization buffer (50 mM Tris, pH 8.0, 1 mM EDTA, 1% (w/v) Nonidet P-40, 1% (w/v) deoxycholic acid) and sonicated further. Insoluble material (including RNH110) was pelleted by centrifugation and then resuspended in a protein solubilization buffer (200 mM Tris, pH 8.0, 6 M guanidine HCl (GdnHCl)). Solubilized RNH110 was refolded by dialysis against low salt buffer (50 mM Tris, pH 8.0, 20 mM NaCl, 0.5 mM EDTA). The dialysis retentate was loaded directly onto a heparin-Sepharose column and developed with a linear KCl gradient from 0 to 200 mM. Purified protein fractions (judged by a single band on polyacrylamide gel electrophoresis stained with Coomassie Blue) were pooled, concentrated to Ͻ5 ml, and dialyzed against 50 mM Tris, pH 8.0, 100 mM NaCl (for activity experiments) or 50 mM potassium phosphate, pH 8.0, 50 mM KCl (for CD studies). Protein concentrations were determined by UV absorption spectroscopy in 6.0 M GdnHCl (A 280 at 1 mg/ml ϭ 1.80).
RNase H Activity Assays-Synthesis and purification of the RNA⅐DNA hybrid substrate were performed as described previously (17). RNase H assays were carried out at 37°C in a standard RNase H reaction buffer (14) (50 mM Tris, pH 8.0, 50 mM NaCl, 1 mM divalent cation (MgCl 2 or MnCl 2 ), 1.5 M BSA, 1 M (base pairs) RNA⅐DNA hybrid), unless otherwise specified. In addition, 1 l of a saturated solution of polyacrylamide absorbent gel was added per 20 l of reaction to help maintain substrate solubility in high MnCl 2 concentrations (final concentration, ϳ0.1 mg/ml, used only with 1 mM MnCl 2 reactions). Enzymes were diluted from concentrated stocks in 50 mM Tris, pH 8.0, 100 mM NaCl to their final stock concentrations in 50 mM Tris, pH 8.0, 100 mM NaCl, 1.5 M BSA, 50% glycerol just before assaying them for activity. At indicated time intervals, 20-l reaction aliquots were stopped by adding 50 mM EDTA, 200 g/ml Torula RNA (final concentration) on ice. The remaining substrate was precipitated with the addition of 5% (w/v) trichloroacetic acid, 20 mM sodium pyrophosphate (final concentration), followed by incubation on ice for 10 min. Precipitated substrate was pelleted for 10 min in a microcentrifuge, and the acid-soluble radioactivity of 90% of the supernatant was determined by liquid scintillation counting.
CD Measurements-All CD measurements were made on an Aviv 62DS spectropolarimeter with a Peltier temperature-controlled sample holder and with a 1-cm path length cuvette.
Protein denaturation studies were carried out by monitoring the ellipticity at 222 nm as a function of temperature or GdnHCl. Chemical denaturations were carried out at 25°C with RNH110 at 35 g/ml, E. coli RNase H* at 15-40 g/ml, or the HIV RNase H domain at 30 -40 g/ml in 50 mM potassium phosphate, pH 8.0, 50 mM KCl with the indicated GdnHCl concentrations. Thermal denaturations were performed in the same buffer with 0.5 M GdnHCl to allow reversibility (15). The free energy of unfolding was determined as described by Dabora and Marqusee (15) using a two-state assumption and linear extrapolation (18).

Construction of a Variant of E. coli RNase H* Lacking the
Basic/Helix Loop Region-A plasmid encoding a deletion mutant of E. coli RNase H* with its basic helix/loop region replaced by a glycine linker was created by cassette mutagenesis of pSM101 (see "Experimental Procedures"). The resulting plasmid (pJMD101) encodes the variant protein RNH110, which is composed of the E. coli RNase H* sequence from Met-1 to Ile-82, a Gly 6 linker, and then E. coli RNase H* from Lys-96 to Val-155. Essentially, we have removed the basic helix/loop of the enzyme (residues 83-95 (Fig. 1)) and replaced it with a flexible glycine linker. Six glycines were judged to be sufficient as a flexible linker after molecular modeling using INSIGHT (Biosym Technols., San Diego, CA). The deletion mutant was modeled to resemble the RNase H domain of HIV which lacks the basic helix/loop region (Fig. 1). The coding region of pJMD101 was confirmed by standard DNA sequencing techniques (16), and purification of the overexpressed protein RNH110 to apparent homogeneity is described under "Experimental Procedures.'' The Deletion Protein, RNH110, Is Folded and Stable in Solution-Since the recombinant protein, RNH110, was isolated from inclusion bodies and purified under denaturing conditions, it is especially important to evaluate its efficiency of renaturation. Far-UV CD spectropolarimetry was used to assay for refolding of RNH110; Fig. 2 shows the spectra of E. coli RNase H* and the deletion mutant, RNH110. Both proteins are well folded under these conditions, with the deletion mutant displaying slightly greater mean residue ellipticity than the parent protein. While somewhat unexpected, this difference may be the result of removing three tryptophan residues and/or partial removal of a loop region. Both of these factors can contribute a positive signal in this region of the spectrum. In spite of these small differences, the overall spectral features of

FIG. 1. Structural comparison of E. coli RNase H and the isolated HIV RNase H domain (A) and design of RNH110, an E. coli RNase H deletion mutant (B).
A, ribbon diagrams of the crystal structures of E. coli RNase HI (28,29) and the isolated HIV-1 RNase H domain (5) (drawn with Molscript (20)). The dashed line on the HIV RNase H ribbon represents a region of the crystal structure for which no electron density was observed (5). B, amino acid sequence comparison of E. coli RNase H* and RNH110, a deletion mutant of E. coli RNase H*. The secondary structure of E. coli RNase HI (as determined in Ref. 29) is shown below the sequences, and active site residues are indicated by asterisks above the residues (7). The basic helix/loop region as defined by Kanaya et al. (12) is boxed. Residues shown in boldface result in at least a 2-fold increase in Mg 2ϩ -dependent K m on mutation to alanine (12). Italicized amino acids indicate the 3 cysteine to alanine mutations present in E. coli RNase H*.
The relative thermodynamic stability of the deletion mutant was determined by both thermal and chemical denaturation.  (Fig. 3). E. coli RNase H* is active in 1 mM MnCl 2 , but with a reduction of 100-fold in activity relative to 1 mM MgCl 2 (data not shown). Surprisingly, the deletion mutant, RNH110, is also active in the presence of MnCl 2 , in spite of its lack of Mg 2ϩ -dependent activity. At 1 mM MnCl 2 , the specific activity of RNH110 is similar to that of its E. coli RNase H* parent in the same conditions. This is surprising in light of the known importance of the deleted region for substrate binding (12) and the inability to detect Mg 2ϩ -dependent activity. As noted previously (14), the isolated HIV RNase H domain remained inactive under all of our experimental conditions.
Given the similar Mn 2ϩ -dependent activity observed for RNH110 and E. coli RNase H*, we investigated the divalent cation dependence of both enzymes (Fig. 4). The activity of the helix/loop deletion protein depends on both the concentration and the identity of the divalent cation used. Activity was observed only in MnCl 2 -containing reactions and only at concentrations above ϳ50 M divalent cation. No activity was observed for this mutant under any MgCl 2 conditions (Fig. 4A). The activity of RNase H* is also dependent on divalent cation concentration and identity. The Mg 2ϩ -dependent activity of RNase H* plateaus to a maximum at ϳ1 mM Mg 2ϩ . Activation of the enzyme with Mn 2ϩ is quite different, with maximal activity occurring at 2-4 M Mn 2ϩ and strong inhibition with higher MnCl 2 concentrations (Fig. 4B). The differential effects of Mn 2ϩ and Mg 2ϩ on E. coli RNase H* were not predicted from previous work and are examined under "Discussion."

DISCUSSION
In vitro studies of HIV RNase H activity are hampered by the fact that the RNase H domain is inactive when expressed independently of the rest of reverse transcriptase. Substitution of the basic helix/loop region from E. coli RNase HI into the RNase H domain from HIV has been shown to restore Mn 2ϩdependent RNase H activity (13,14). In addition to activating the domain, this substitution creates a chimeric domain with increased thermodynamic stability compared with the inactive HIV domain (14). Whether the role of the basic helix/loop in activating the HIV domain is stability, substrate affinity, or both is therefore unclear. In an attempt to dissect the potential roles of the basic helix/loop, we performed the converse experiment and modeled the structural characteristics of HIV RNase H onto the active E. coli RNase H* by creating a mutant form of the E. coli enzyme with its basic helix/loop region replaced by a short glycine linker. Our deletion mutant (RNH110) was expected to display the instability and inactivity that are observed in the isolated HIV RNase H domain. Instead we found that RNH110 retained a Mn 2ϩ -dependent RNase H activity and was more stable than the HIV domain.
The obligate Mn 2ϩ dependence of the deletion protein is   interesting in light of the very different metal dependence of its parent protein, E. coli RNase H*. In vitro, the activity of RNase H* varies as a function of both the identity and the concentration of the divalent metal (Fig. 4). Mg 2ϩ -dependent activity plateaus at ϳ1 mM Mg 2ϩ . Its Mn 2ϩ -dependent activity, however, is maximal at 2-4 M Mn 2ϩ with subsequent inhibition at higher concentrations of Mn 2ϩ . This inhibition at high Mn 2ϩ concentrations correlates with a previous study showing greater activity in 0.2 mM than in 1 mM MnCl 2 (21). Inhibition of the Mn 2ϩ -dependent activity of RNase H* by high divalent concentrations presents an interesting possible model for the regulation of the general RNase H mechanism. In contrast to RNase H*, RNH110 shows no Mg 2ϩ -dependent activity but retains a weakened Mn 2ϩ activity. This Mn 2ϩ -dependent activity is shifted toward a higher metal concentration dependence (Fig. 4), suggestive of weaker metal binding in RNH110. Although the basic helix/loop is distant from the active site, its removal may result in a decrease in metal-binding affinity through subtle structural changes in the active site residues involved in metal coordination.
This observed shift in metal dependence may also help to explain the inactivity of the isolated HIV RNase H domain. The HIV domain is less stable than either the active E. coli RNase H* or RNH110 (Table I). Decreased stability may result in a lower metal-binding affinity which would then account for the observed inactivity of the domain. Substitution of the E. coli basic helix/loop into the HIV RNase H domain both activates and stabilizes the domain (14); this stabilization may, therefore, stabilize the metal-binding pocket(s) of the domain. These data suggest that the inactivity seen in the RNase H domain from HIV is due, at least in part, to its poor thermodynamic stability.
Obligate Mn 2ϩ -dependent activity is not uncommon among retroviral RNase H mutants. An active-site mutant in HIV reverse transcriptase RNase H abolishes Mg 2ϩ -dependent activity but still maintains a weakened Mn 2ϩ -dependent activity (22). A histidine-tagged version of the HIV RNase H domain displays Mn 2ϩ -dependent activity (23,24). A recent study has shown that a series of mutations in the Moloney murine leukemia virus reverse transcriptase RNase H domain abolish its Mg 2ϩ activity but allow a Mn 2ϩ -dependent activity (25). As already described, substitution of a basic helix/loop in the HIV domain restores a Mn 2ϩ -dependent activity (13,14). Furthermore, inhibitors affect HIV reverse transcriptase Mg 2ϩ -and Mn 2ϩ -dependent RNase H activities differently, implying a difference between the two activities (26). In this report, we describe the first mutation in E. coli RNase H (removal of the basic helix/loop) that results in a strictly Mn 2ϩ -dependent activity.
The importance of the basic helix/loop region for substrate binding in E. coli RNase HI has been investigated by Kanaya et al. (12). Their findings, that neutralizing mutations in the basic helix/loop lead to increases in K m but small changes in catalyzed reaction velocity (k cat ), have indicated the general importance of this region for substrate binding. In contrast, a number of observations have implied that this region is not critical for proper folding and activity of the E. coli RNase HI molecule. Kanaya et al. (27) have demonstrated the ability to refold two proteolytic fragments of E. coli RNase HI (the protease site is within the basic helix/loop) reforming an active RNase H. Furthermore, we have found two fragments of E. coli RNase H* that can refold by dialysis from GdnHCl to form an active RNase H molecule lacking the entire basic helix/loop region. 2 Our observation of a Mn 2ϩ -dependent RNase H activity in RNH110 directly addresses the role of this region in the activity of E. coli RNase H. Taken together with the observations of Kanaya et al. (12,27), our work suggest that the basic helix/ loop serves two roles: substrate affinity and RNase H stability. Future studies of both the E. coli and HIV RNase H homologues should help to further elucidate this important relationship between stability and activity in the RNase H family of enzymes.