Characterization of Escherichia coli RNase PH.

We have previously shown that the orfE gene of Escherichia coli encodes RNase PH. Here we show that the OrfE protein (purified as described in the accompanying paper) (Jensen, K. F., Andersen, J. T., and Poulsen, P. (1992) J. Biol. Chem. 267, 17147-17152) has both the degradative and synthetic activities of RNase PH. This highly purified protein was used to characterize the enzymatic and structural properties of RNase PH. The enzyme requires a divalent cation and phosphate for activity, the latter property indicating that RNase PH is exclusively a phosphorolytic enzyme. Among tRNA-type substrates, the enzyme is most active against synthetic tRNA precursors containing extra residues following the -CCA sequence, and it can act on these molecules to generate mature tRNA with amino acid acceptor activity; 3'-phosphoryl-terminated molecules are not active as substrates. The equilibrium constant for RNase PH is near unity, suggesting that at the phosphate concentration present in vivo, the enzyme would participate in RNA degradation. The synthetic reaction of RNase PH displays a nonlinear response to increasing enzyme concentrations, and this may be due to self-aggregation of the protein. Higher order multimers of RNase PH could be detected by gel filtration at higher protein concentrations and by protein cross-linking. The possible role of RNase PH in tRNA processing is discussed.

PAGE3 (6). N-terminal sequence analysis of the first 21 amino acids from this protein band revealed that RNase PH corresponded to the product of the orfE gene (6), a previously unidentified open reading frame upstream of, and co-transcribed with, pyrE (7). Jensen's (8) laboratory has recently developed an overproduction system in which milligram quantities of the OrfE protein can be obtained with only two purification steps; this work is reported in the accompanying paper. The availability of highly purified OrfE protein presented us the opportunity to assess the protein's catalytic properties.
In this paper we show, first of all, that the OrfE protein possesses both the degradative and synthetic activities of RNase PH. We then used the highly purified protein to perform a detailed characterization of RNase PH.

EXPERIMENTAL PROCEDURES
M~terials-[~H]Poly(A) and "C-amino acids were obtained from Amersham and [3H]CDP and [3H]CTP were obtained from Du Pont-New England Nuclear. Poly(A), used to dilute the radioactive material, was obtained from Sigma. Ovalbumin, bacterial alkaline phosphatase, and blue dextran 2000 were purchased from Sigma, Cooper Biomedical, and Pharmacia Inc., respectively. N-Ethylrnaleimide (NEM) and p-hydroxymercuribenzoate (PMB) were obtained from Sigma. Dimethyl suberimidate (DMS) was from Pierce Chemical Co. Ultrogel AcA 44 was purchased from Pharmacia LKB Biotechnology Inc. All other chemicals were reagent grade.
The tubes were placed in ice for 15 min and excess periodate was destroyed with 2 pl of ethylene glycol by incubation at room temperature for 20 min. These RNAs were used directly as substrates for RNase P H or were first treated with bacterial alkaline phosphatase (0.46 fig for 30 min at 45 "C) to remove any 3"terminal phosphate produced.
Partially purified RNase PH was prepared as described earlier (5). Highly purified OrfE/RNase PH was kindly provided by Dr. Kaj Frank Jensen, University of Copenhagen, and its preparation is described in the accompanying paper (8 tion of [3H]CDP into tRNA as acid-precipitable radioactivity (6). Reaction mixtures for the synthetic reaction contain in 100 p1: 50 mM glycine-NaOH, pH 9.75, 5 mM MgCl,, 75 mM KCl, 0.5 mM [%I-CDP, 17 pg of tRNA, and the indicated amount of RNase PH. Incubation conditions are presented in the legends. Acid-precipitable radioactivity was determined as described previously (6).
RNase P H Equilibrium Determination-The reaction conditions used to determine the equilibrium between the degradative and the synthetic reactions catalyzed by RNase PH were as follows: 50 mM Tris-C1, pH 8.0, 5 mM MgCIZ, 50 mM KC1, 0.1 mM NaP04, 10 pg of tRNA, and 0.4 mM [3H]CDP, or 10 pg of tRNA-CCA-[3H]C2.3, 0.4 mM CDP, and 0.1 pg of RNase PH. Samples were incubated for 60 min at 37 "C. The amount of CDP was determined as acid-soluble radioactivity, while the amount of tRNA was determined as acidprecipitable radioactivity, as described above.
Gel Filtration-A column of Ultrogel AcA 44 (1.0 X 51 cm) was equilibrated with the running buffer: 20 mM Tris-C1, pH 7.2, 10% glycerol, 0.1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 10 mM 2-mercaptoethanol. Varying amounts of purified RNase PH were each added to 50 p1 of ovalbumin (20 mg/ml) in the running buffer, and then loaded onto the column; the ovalbumin was found to stabilize RNase PH upon dilution on the column. Fractions (0.26 ml) were collected and assayed for both the synthetic and degradative activities. The column was standardized with ovalbumin, bacterial alkaline phosphatase, and blue dextran 2000.
Cross-linking of RNase PH-Varying amounts of RNase PH were added to 0.25 M triethanolamine, pH 8.4, 10% glycerol, 0.05 mM phenylmethylsulfonyl fluoride, 0.05 mM dithiothreitol, 0.05 mM EDTA, and 2.5 mg/ml DMS in a final volume of 50 pl. The crosslinking reactions were carried out at 24 "C for 3.5 h. The reaction was terminated by the addition of sodium dodecyl sulfate and 2mercaptoethanol was added to final concentrations of 1 and lo%, respectively. Each sample was then boiled for 3 min before dialyzing in 50 volumes of 0.125 M Tris-C1, pH 6.8, 1% SDS, and 1% 2mercaptoethanol for 1 h as described (11). Samples were run on 8.5% SDS-PAGE (12) and then silver stained.

RESULTS
Comparison of OrfE Protein and RNase PH-To ensure that the OrfE protein was appropriate for the characterization of RNase PH, we have compared the highly purified protein to partially purified RNase PH prepared in our laboratory on the basis of its RNase PH activity. Both proteins were found to migrate identically on 12.5% SDS-PAGE corresponding to a size of 33 kDa (data not shown). The degradative and synthetic activities of the two proteins were also directly compared using a variety of substrates. After first normalizing the two preparations by their activity against tRNA-CCA- [3H]Cz-3, it was found that the relative RNase PH degradative activities against tRNA-CC[14C]A, tRNA-C[14C]C, and 3Hlabeled poly(A) were identical for the two enzyme preparations (data not shown). The RNase PH synthetic activity of the two proteins was also the same. However, the OrfE protein was more highly purified since it had almost a 10-fold higher specific activity than the RNase PH prepared in our laboratory. These results confirm our previous conclusion that orfE encodes RNase PH (6), and they indicate that the highly purified OrfE protein is suitable for characterization of RNase PH activity. The purified protein will henceforth be referred to as RNase PH.
Requirements of the RNase PH Degradative Reaction-The highly purified RNase PH requires both Pi and a divalent cation for activity (Table I). In the absence of added M$+, RNase PH activity was eliminated almost entirely. Likewise, in the absence of Pi, RNase P H activity was undetectable.
This latter observation demonstrates that the purified RNase PH preparation is devoid of other ribonucleases. Also, as shown in Table I   Other divalent cations were tested to see if they could substitute for Mg2+ in the degradative reaction (Table   11). Mn2+, Co2+, and Zn2+ were tested at concentrations of 1 and 5 mM. Both Mn2+ and Co2+ were found to substitute for M$+ to some degree (about 40 and 30%, respectively). In both cases 1 mM cation was somewhat more effective than 5 mM, in contrast to the situation with M P . Essentially no activity was observed when Zn2+ was substituted for M$+.
Effect of Enzyme Concentration on the Synthetic and Degradatiue Reactions-Previously, RNase PH, in a partially purified fraction, was shown to respond linearly to enzyme concentration in the degradative reaction, but to give a distinctly nonlinear response in the synthetic reaction ( 5 ) . T o determine whether this unusual situation might have been dependent on other proteins in the cruder preparation, we have re-examined this phenomenon with the purified enzyme. The data in Fig. 1 demonstrate that the same result is found with the highly purified protein; i.e. the degradative reaction is linear with respect to the amount of enzyme added, whereas the synthetic reaction is greatly stimulated at higher amounts of RNase PH. These data show that the unusual response of the synthetic reaction to increasing enzyme is an intrinsic property of RNase PH.  CCA-[3H]Cz-3, respectively, when present at the same concentration. These data indicate that among tRNA-type molecules, the 3"extended tRNA precursor is the preferred substrate.
Crude preparations of RNase P H were previously shown to degrade poly(A) in a phosphate-dependent reaction (3). Polynucleotide phosphorylase also carries out such a reaction, but the ratio of the poly(A) to tRNA-CCA-[3H]C2-3 activity was very different for the two enzymes, about 2400 and 14 for polynucleotide phosphorylase and RNase PH, respectively. The overlap in activity with polynucleotide phosphorylase made it important to determine whether the activity of RNase P H against the nonspecific substrate, poly(A), was an intrinsic property of the enzyme or might have been due to a contaminating RNase in the crude preparation. As also shown in To determine the equilibrium between the synthetic and degradative activities, two separate reactions were carried out under the same conditions as described under "Experimental Procedures." In one, [3H]CDP and unlabeled tRNA were added to monitor the synthetic reaction, and in the other reaction, tRNA-CCA-[3H]C2-3 and Pi were used to follow the degradative reaction. Assuming the synthetic reaction is the forward reaction, the Keq value for the two reactions was found to be 0.25. These data indicate that the RNase PH reaction is freely reversible in vitro. Given the Pi concentrations present in vivo (-1 mM), these findings suggest that the degradative reaction probably is significant in vivo.
Temperature Sensitivity of RNase PH-Samples of RNase P H were diluted to a concentration of approximately 0.015 mg/ml and incubated at various temperatures for 10 min prior to determining the amount of degradative activity remaining. The data in Fig. 2 show that RNase PH was fairly stable up to 45 "C, that it lost slightly more than 50% of its activity when heated a t 55 "C, and that it was totally inactivated when incubated for 10 min at 65 "C. Thus, RNase P H is moderately stable to short term heating under these conditions.
Effect of Sulfhydryl Reagents on RNase PH-Based on the nucleotide sequence of rph (orfl), RNase P H contains 4 cysteine residues (assuming they are reduced). To determine whether these residues might play a role in RNase P H activ- ity, the enzyme was incubated with the sulfhydryl reagents, NEM and PMB, and the RNase PH activity assessed. The data in Table IV show that RNase PH is sensitive to both NEM and PMB, with the latter having the more profound effect. Thus, after a 10-min incubation with PMB, only 21% of RNase P H activity remained, and after 60 min the enzyme was completely inactivated. The sensitivity of RNase PH to sulfhydryl reagents suggests that one or more of its cysteine residues may be directly involved in RNase P H activity or that they significantly affect RNase P H structure.
Can RNase PH Generate Mature 3"Termini on tRNA?-An important question with regard to the function of RNase P H is whether the enzyme can generate the mature 3' terminus of tRNA when acting on a tRNA precursor. To test this point, tRNA-CCA-[3H]Cz_3 was treated with RNase PH, and the product of the reaction was examined for its amino acid acceptor activity in the presence of a mixture of 14Camino acids and aminoacyl-tRNA synthetases. Shown in Fig.  3 is the effect of increasing amounts of RNase P H on both nucleotide removal from the 3' terminus of tRNA-CCA-[3H] CZ-3 and aminoacylation of the treated tRNA precursor. The degree of nucleotide removal reached a level of about 90%. As nucleotide removal increased, the tRNA precursor acquired the ability to accept amino acids, increasing to a maximum level of close to 35% of that found with a comparable amount of mature tRNA.
These experiments were carried out in the presence of tRNA nucleotidyltransferase to repair any tRNA chains in which RNase P H might have removed the terminal AMP residue from the -CCA sequence. That this was in fact occurring was indicated by a lower level of aminoacylation (20%) found in the absence of end repair by tRNA nucleotidyltransferase. Furthermore, a direct measurement of AMP removal from tRNA-CC[14C]A under the conditions used in this experiment revealed that as many as 70% of the tRNA chains were affected by RNase PH. Thus, even though RNase P H acts more slowly on mature tRNA than on tRNA-CCA-[3H] C2-3 (Table 111), when high levels of enzyme are used, as in this test for generation of the mature 3' terminus, entry into the -CCA sequence can be significant. In fact, the maximum aminoacylation level of -35% was probably due to tRNA chains that could not be repaired by tRNA nucleotidyltransferase plus ATP because nucleotide removal had proceeded past the 3'-AMP residue. Nevertheless, these data show that RNase P H is able to remove residues from tRNA precursors to generate a mature 3' terminus.
Comparison of Precursor and Mature tRNA Affinity for RNase PH-In view of the difference in RNase P H action on the synthetic precursor tRNA compared to that on mature tRNA, it was of interest to examine further why the rate of  tRNA-CC[14C]A degradation was almost an order of magnitude lower than that for tRNA-CCA-[3H]C2_3. To test the relative affinities of the two tRNAs for RNase PH, we measured the ability of mature tRNA to act as an inhibitor of tRNA-CCA-[3H]C2-3 degradation by the enzyme. Nonradioactive mature tRNA was added to the tRNA-CCA-[3H]C2-3 degradative assay in increasing amounts, and the level necessary for 50% inhibition of the reaction was found to be approximately 4 mg/ml when 0.1 mg/ml tRNA-CCA-[3H]C2-3 was present as substrate (data not shown). This suggested that the affinity of mature tRNA for RNase P H was considerably lower than that of the precursor tRNA. Measurement of the apparent K,,, for tRNA-CC[14C]A as a substrate of RNase P H confirmed this point. A value of approximately 10 p~, 10-fold higher than that for precursor tRNA, was found. These findings imply that upon maturation of the 3' terminus of a precursor tRNA by RNase PH, the processed tRNA would tend to dissociate from the enzyme rather than be acted upon further.
Native Molecular Weight of RNase PH-In an earlier study (3) the native size of RNase P H was shown by gel filtration to be 45-50 kDa. Inasmuch as Jensen et al. (8) have observed that the size of the purified protein is greater than 240 kDa, it was important to ascertain the explanation for these differences. The purified protein isolated by Jensen et al. (8) was run on an Ultrogel AcA 44 column using four different amounts of enzyme ranging from 2 to 65 pg. In each case the elution position of RNase P H was monitored by both its synthetic and degradative activities. As shown in Fig. 4, the more concentrated the RNase P H sample applied to the column, the larger the apparent size of the enzyme. Thus, at the lowest RNase P H concentration (Fig. 4A), which corresponded to that used previously in experiments with the partially purified enzyme (3), the size was as reported, 45-50 kDa. Likewise, at the highest concentration tested (Fig. 4 0 ) , the size approached that reported by Jensen et al. (8). Thus, these data indicate that RNase PH has a strong tendency to aggregate, and that the size determined is dependent on the protein concentration used for the measurement.
It is interesting to note that the ratio of synthetic to degradative activity increases as the protein concentration applied to the column increases, in full agreement with the data in Fig. 1. Moreover, when the most concentrated sample was run on the column (Fig. 4 0 ) , the synthetic activity was slightly shifted to a higher molecular weight than the degradative activity; to a lesser extent, this effect was also seen in Fig. 4, B and C. These data suggest that the larger aggregates of RNase PH may be responsible for the unusual response of the synthetic reaction to increasing enzyme concentration. At present, it is not clear how the different multimeric forms could catalyze different levels of synthetic and degradative activities.
Cross-linking of RNase PH-To definitively demonstrate that RNase PH self-aggregates, the enzyme was exposed to the cross-linking agent, DMS, at different protein concentrations, and the treated samples were then run on SDS-PAGE. As shown in Fig. 5, an extended series of RNase PH multimers forms upon treatment with DMS, and these are seen at all protein levels tested. Thus, RNase PH can readily self-associate to form multimeric complexes of varying sizes.

DISCUSSION
The data presented here show that the OrfE protein has both the degradative and synthetic activities of RNase PH, and they confirm our previous suggestion that orfE (now called rph) encodes this ribonuclease (6 in which it is encoded remains to be established. The availability of large amounts of highly purified RNase PH has allowed us to carry out a detailed characterization of the enzyme. The properties of the purified enzyme that were retested were essentially the same as determined earlier with less pure material ( 5 ) . In addition, much new information that could not be determined with impure preparations was obtained. Most importantly, the highly purified enzyme was shown to be devoid of activity in the absence of Pi, indicating that RNase PH is a strictly phosphorolytic nuclease, and that the low levels of hydrolytic activity observed earlier must have been due to contaminating enzyme(s).
The studies presented here also showed that, as expected for an enzyme with an equilibrium constant close to unity, purified RNase P H can catalyze the synthesis of RNA. However, the unusual activity versus enzyme profile observed for the synthetic reaction remains to be explained. One possibility is that at higher enzyme levels the longer chains synthesized are better substrates for further addition of nucleotides. This effect would not be apparent in the degradative reaction in which the size of the substrate, if anything, decreases. However, it would be consistent with the higher degradative activity on poly(A) compared to tRNA substrates. A second possibility is that the aggregation of RNase PH that should be preferred at higher enzyme concentrations somehow leads to the synthesis of a different product that can be made a t a more rapid rate.
The self-association of RNase P H was an unexpected finding. Both by gel filtration of the enzyme at varying protein concentrations and by cross-linking, RNase P H was shown to aggregate readily to higher order multimers. Whether this aggregation serves any physiological role remains to be determined. Nevertheless, it does explain the discrepancies in molecular weight determinations made by us and Jensen et al. (8). This finding also suggests that the smallest active form observed by gel filtration (45-50 kDa) may be due to a dimer that partially dissociates to monomer during the chromatography. It is also possible that the larger size on gel filtration compared to SDS-PAGE is due to an elongated shape for the monomer.
The preference of RNase P H for tRNA precursor molecules compared to mature tRNA suggests that this enzyme could participate in tRNA processing. Moreover, its ability to gen-erate mature tRNAs that can accept amino acids supports this idea. However, compared to RNase D (13), which carries out a similar reaction, RNase P H is less fastidious with regard to entering the -CCA sequence. Thus, RNase P H generated less mature tRNA compared to RNase D (13), and the amount of mature tRNA produced was increased by the presence of tRNA nucleotidyltransferase. On the other hand, in vivo in the absence of RNase D, RNase P H may play a significant role in the processing of the 3' terminus of tRNA precursors, and this is supported by the inviability of strains lacking RNase P H when this mutation is in combination with other RNase mutations.' The ability of RNase PH to act on poly(A) suggests that it may act preferentially on unstructured RNA molecules, and its higher activity on tRNA precursors compared to mature tRNA may reflect the distance of the residues removed from the double-stranded aminoacyl stem of tRNA. The greater difficulty in removing residues as the aminoacyl end is approached, suggests that in vivo final 3' processing of tRNA may not be the primary function of RNase PH, although it may serve as a backup enzyme for this process. The availability of mutants lacking RNase PH, alone or in different combinations with mutants lacking other RNases, should now make it possible to sort out the functions of these multiple E. coli exoribonucleases.