Ribonucleic acid processing activity of Escherichia coli ribonuclease III.

We have studied the nuclease activities present in preparations of Escherichia coli RNase III and the "sizing factor" responsible for specific processing of several RNA species. RNase III preparations contain three activities: one which solubilizes stable RNA:RNA duplexes; one which solubilizes the RNA of DNA:RNA hybrids; and one which processes the polycistronic mRNA of bacteriophage T7 in a manner identical with sizing factor. We show that the activity against the RNA of DNA:RNA hybrids can be removed, but that the activity which cleaves RNA:RNA duplexes and that responsible for specific processing of phage T7 polycistronic mRNA appear to be identical by several biochemical criteria. In addition, partially purified enzyme fractions from mutants lacking these two activities contain substantial amounts of activity against the RNA of DNA:RNA hybrids. We have also defined several properties of the two activities solubilize RNA:RNA duplexes and RNA of DNA:RNA hybrids. Average oligonucleotide chain length in an exhaustive digest of double-stranded RNA is about 15 bases, while that in a digest of the RNA in DNA:RNA hybrids is less than 10 bases. Direct analysis shows that both activities cleave RNA chains to yield 5'-phosphate and 3'-hydroxyl termini. All four bases can reside at the 5' end of the resulting oligonucleotides, although both activities show a mild preference for certain bases. These results and previous findings allow us to specify the probably size and structure of potential cleavage sites for these enzymes in biological RNA molecules.


Ribonucleic
Acid Processing Activity of Escherkhia coli Ribonuclease III * (Received for publication, May 16, 1974) HUGH D. ROBERTSON AND JOHN J. DUNN From the Rockefeller University, New York, New York lOOZ?l, ana! the Biology Department, Brookhaven National Laboratory, Upton, New York 11973 SUMMARY We have studied the nuclease activities present in preparations of Escherichia coli RNase III and the "sizing factor" responsible for specific processing of several RNA species. RNase III preparations contain three activities: one which solubilizes stable RNA:RNA duplexes; one which solubilizes the RNA of DNA:RNA hybrids; and one which processes the polycistronic mRNA of bacteriophage T7 in a manner identical with sizing factor. We show that the activity against the RNA of DNA:RNA hybrids can be removed, but that the activity which cleaves RNA:RNA duplexes and that responsible for specitic processing of phage T7 polycistronic mRNA appear to be identical by several biochemical criteria. In addition, partially purified enzyme fractions from mutants lacking these two activities contain substantial amounts of activity against the RNA of DNA:RNA hybrids.
We have also defined several properties of the two activities which solubilize RNA:RNA duplexes and RNA of DNA:RNA hybrids. Average oligonucleotide chain length in an exhaustive digest of double-stranded RNA is about 15 bases, while that in a digest of the RNA in DNA:RNA hybrids is less than 10 bases. Direct analysis shows that both activities cleave RNA chains to yield 5'-phosphate and 3'-hydroxyl termini. AU four bases can reside at the 5' end of the resulting oligonucleotides, although both activities show a mild preference for certain bases. These results and previous fiudings allow us to specify the probable size and structure of potential cleavage sites for these enzymes in biological RNA molecules.
The role of specific RNase activities in cells has recently been the object of intensive study. Of particular interest have been those RN&se activities which introduce cleavages at a limited number of specific sites during the maturation of RNA precursors.
In Escherichia coli, several such RNases have recently been iden-* Part of this research was supported by a grant to H. D. R. from the National Science Foundation. Research carried out at Brookhaven National Laboratory was under the auspices of the United States Atomic Energy Commission. tified and the proper cleavage of their normal substrates have been reproduced in vitro by isolated enzyme (l-3).
Dunn and Studier (1) originally observed that the E. coli "sizing factor" which processes early T7 mRNA copurified with RNase III (4, 5). With the subsequent availability of an E. coli mutant lacking that activity in crude extracts which cleaves dsRNAl (6), these authors were able to show that mutant cells failed to process early T7 mRNA and E. coli rRNA precursors which accumulated in c&o (2). These precursors could be processed in vitro with enzyme from wild type cells. Similar results have been obtained by Nikolaev et al. (7,8).
RNase III was originally characterized as an enzyme preparation with endonucleolytic activities against both classes of stable double helical RNA-RNA:RNA duplexes and the RNA of DNA : RNA hybrids (5). Subsequent studies confirmed the presence of both activities after a number of purification steps (9-l 1). Activities specifically able to digest the RNA of DNA:RNA hybrids have since been found in eukaryotic nuclei (12), tumor virus reverse transcriptase (13), and E. coli (1416), and have been given the generic name "RN&se H." Such RNase H activities could play a role in removing the RNA primer which has recently been implicated in the initiation of DNA synthesis in a number of systems (17-20).
Crouch (21) has employed sedimentation on sucrose density gradients to fractionate further E. coli RNase III preparations. He obtained fractions of higher S value which were largely free of RNase H activity, but were still capable of digesting dsRNA; activity of lower S value contained largely RNase H. In addition, Nikolaev et al. (7) have stated that the mutant lacking activity against dsRNA still retains wild type levels of RNase H. It should be noted, however, that several groups (14-16, 21) have identified three different enzyme fractions in E. coli capable of digesting hybrid RNA.
In this communication, we will present evidence complementary to that of Crouch (21) that it is possible to obtain RNase 1 The abbreviations used are: dsRNA, double-stranded RNA (used here to denote specifically RNA: RNA duplexes containing two sizable strands of RNA base paired over their entire length and having exactly complementary sequences (28); ssRNA, single-stranded RNA; hybrid RNA, the double helical RNA present in DNA:RNA hybrids containing one strand of DNA and one strand of RNA of exactly complementary sequence, basepaired over their entire length; PC, the mold Penicillium ckysogenum; pXp, a (2'-, 3'-), 5'-nucleoside diphosphate; pAp, pCp, pGp, and pup, the four common bases when involved in this configuration.
III preparations which lack RNase H activity. The key observa-and stored on ice (Fraction VII). Sizing factor was purified using tion of these studies is that these fractions (lacking RNase H) BioGel filtration and other procedures as before (1). are able both to solubilize stable dsRNA and to process phage T7 Pancreatic DNase (DPFF) was purchased from Worthington. polycistronic mRNA in vitro. In order to describe these activities RNase Tl was purchased from Calbiochem. Other Ma~etials-la-**Plribonucleoside triphosphates (ATP, in more detail, we have also carried out experiments on the speci-GTP, CTP, and UTP), all with specific radioactivities in the range ficity, end group polarity, and size limit of the digestion products. of 100 Ci/mmol, were purchased from New England Nuclear. Un-The results of these studies place limitations upon the kinds of labeled nucleoside triphosphates were from P-L Biochemicals.
cleavage sites which we might expect to find in biological RNA Standards of 3'-UMP, CMP, AMP, and GMP, as well as pup, molecules.
pCp, pAp, and pGp, were those used previously ( Fig. 1 and Tab& I and iI), the RNA was labeled using [a-aeP]UTP adjusted to a final specific radioactivity of 0.7 pCi/nmol. Incubation was for 30 min at 37" and the hybrid was purified using cellulose CFll chromatography according to Robertson (10). The extent of synthesis was 250/o, and the resulting hybrid regions had a DNA specific activity of 3.15 X lo6 cpm/pg and an RNA snecific activitv of 2.25 X 106 comluz. For the experiments in Figs. 3 and 4 and Table III, equal specific radioactivity for all four ribonucleoside triphosphates was sought so that the in vitro product would reflect accurately the base composition of the template. Therefore, in hybrid Preparation B, all four [&*P]ribonucleoside triphosphates were added in the above reaction conditions to give a final specific radioactivity of 3 X 107 cpm of 32P per pg.
Double-stranded RNA containing sequences from polyoma DNA was synthesized in vitro using the DNA (a gift of Dr. A. R. Hunter, Salk Institute, La Jolla, Calif.) from a defective polyoma virus strain which yields upon transcription by E. coli RNA polymerase large proportions of complementary RNA when assayed as described by Haas et al. (24). Synthesis was carried out as for DNA:RNA hybrids, substituting the polyoma DNA for the fl DNA template. For the polyoma dsRNA used in Fig. 3  final specific activity 3 X 10' cpm/pg). To maximize dsRNA content in the product of such reactions, they were heated at 60" for 2 hours followed by mild RNase Tl digestion for !O min at 37" (0.01 rg/ml in 0.1 M NaCl, 0.05 M Tris-HCl (pH 7.0), 0.001 M EDTA). Extraction with phenol and chromatography on cellulose CFll were as described previously (25).
Polycistronic T7 [**P]mRNA was synthesized in vitro using E. coli RNA polymerase with [a-aeP]GTP as label. Reaction conditions and subsequent purification by gel electrophoresis were as described previously (1, 2). Unlabeled RNA from the bacteriophage f2 was the gift of Dr. N. D. Zinder. Ia2P]RNA from bacteriophage fl mRNA ribosome binding sites was synthesized and isolated by H. D. R. and Drs. G. Pieczenik and P. Model (26) using an adaptation of published procedures (27).
Enzllmes-RNase III was prepared from E. coli MREGOO as previously described (5, 28) and from E. coli A19 with minor modifications: DEAE-cellulose and CM-cellulose were emoloved instead of DEAE-Sephadex and CM-Sephadex. Enzyme it this stage of purification is referred to as Fraction VI. An additional fractionation step was carried out on part of Fraction VI. Portions (2 ml) were applied to a column (0.9 X 130 cm) of Sephadex G-100 equilibrated and developed with 0.02 M Tris-HCl (pH 7.9), 0.1 M KCl, 0.1 mu EDTA, 0.1 mM dithiothreitol, and 10% glycerol. Fractions (1.75 ml) were collected and assayed for their ability to solubilize [aH]poly(A-U) (1, 3). The peak fractions were pooled Methods Solubilization of double helical RNAs was assayed by precipitation with 5% tricbloroacetic acid and filtration on Whatman GF/A filters as before (5). Sizing factor activity (processing of T7 polycistronic mRNA) was assayed as previously described (1).
Fingerprinting analysis of digested RNA was carried out according to the procedure of Brownlee and Sanger (29). End group analysis of dsRNA and hybrid RNA was performed using a modification first suggested by Dr. P. G. N. Jeppesen of the standard fingerprinting procedure for RNase Tl digests which has been reviewed by Barrel1 (30). RNA samples were exposed to alkaline digestion by resuspending them in 10 pl of 0.2 Y NaOH and incubating them 18 hours at 37" in sealed capillaries. Digested RNA was then spotted near one end of a strip (2.5 X 85 cm) of Whatman No. 3 MM paper along with standard tracking dyes (30). The paper was moistened with pH 3.5 buffer containing 5% glacial acetic acid, 0.5yo pyridine, and 0.005 M EDTA, and exposed to electrophoresis at 5000 volts until the blue dye (xylene cyan01 FF) had migrated 13 cm towards the anode. The nucleotides were then blotted onto a sheet (46 X 85 cm) of Whatman DE81 DEAE-paper, washed with 95% ethanol, and exposed to electrophoresis in a second dimension in 7% formic acid as described previously (30). These and other two-dimensional analyses were exposed to autoradiography with DuPont Cronex-2 x-ray film. Radioactive spots were located, cut out and their Cerenkov radiation was determined in a liquid scintillation counter set for tritium counting (efficiency of Cerenkov counting of **P, 10%).

RESULTS
Section A: Further Puri$catien of RNase III-As shown in Fig. la, RNase III preparations at the Fraction VI stage of purification contain activities which digest stable dsRNA and hybrid RNA. In the present discussion we refer to these activities as "RNase III" and "RNase H," respectively.* As demonstrated earlier (5), at the Fraction VI stage of purification there is no DNase activity, and the DNA strands of the DNA:RNA hybrid are not digested. When the DNA:RNA hybrid is preincubated with pancreatic DNase sufficient to solubilize all of the hybrid DNA, the RNA which was originally contained in the hybrid is rendered resistant to digestion by the RNase H activity (Fig.  lb), presumably through its release in single-stranded form as suggested earlier (10). Fig. lb also shows that pancreatic DNase has no effect on the ability of RNase III to digest stable dsRNA.
As shown in Fig. 2, RNase III preparations at this stage of purification are able to process T7 polycistronic early mRNA to yield the five monocistronic RNAs, indicated in the figure by the numbers 1, 0.7, 0.3, 1.3, and 1 .I. A comparison of this activity with that of an Escherichia coli protein fraction "sizing factor," originally isolated solely on the basis of its ability to carry out this processing reaction (1)) resulted in nearly identical RNA patterns following electrophoresis of the reaction mixtures on polyacrylamide gels. In particular, when reaction mixtures are adjusted to contain equal units of activity against dsRNA (5) the extent and I The convention of designating as "RNase H" the E. coli activity against the RNA ofbNA:RNA hybrids while reassigning the name "RNase III" to the activitv aeainst RNA:RNA duplexes has been proposed by Crouch (21) (A-A) was digested with 1 unit of Fraction VI. I'CA. trichloroacetic acid. b, effect of preincubation with pancreatic DNase. Identical aliquots of the substrates described in a were pretreated with pancreatic DNase at a concentration of 20 pg/ml for 30 min at 37", after which one unit of Fraction VI activity was added to the reactions and incubation continued for an additional 50 min.
specificity of cleavage is the same for both (Fig. 2, Lanes b, b' and c, c'). Competition esperiments suggest that processing activity and the ability to digest &RNA are reactions catalyzed by the same enzyme. As shown in Fig. 2 the addition of a 60.fold weight excess of unlabeled dsRNA abolished cleavage (Lanes f, f') while neither a 60-fold nor a 600-fold weight excess of f2 ssRNA had any effect on the reactions (Lanes d, d'; e, e'). Similar weight excesses of tRNA or E. coli mature rRNAs also did not prevent cleavage of the T7 polycistronic mRNA (data not shown). Table I shows digestion of dsRNA or hybrid RNA by three different enzyme preparations which were capable of specific cleavage of T7 polycistronic mRNA. All three (Fraction VI, Fraction VII, and sizing factor) were able to solubilize 60 ng of PC dsRNA at about the expected rate. RXase H activity is present in Fraction VI. However, Fraction VII, the peak of dsRNA cleaving and T7 mRNA processing activity chromatographing on Sephadcx G-100, is unable to digest hybrid RNA. Apparently gel filtration removes some component required for RNase H activity. Furthermore, the sizing factor preparation which is shown in Table I to be lacking RNase H, had been chromatographed on RioGel A.5M prior to ion exchange chromatography (1). Here again apparently a protein sizing step has removed RNase H activity. Attempts to date to recover RNase H activity following Sephadex G-100 chromatography have failed. Table II shows an independent experiment which suggests that the digestion of dsRNA or hybrid RNA must be due to different enzymes. In the presence of sufficient excess unlabeled dsRNA to abolish completely the ability of Fraction VI to solubilize dsRNA, its ability to digest hybrid RN.4 is retained with only an 117; drop in the amount solubilized.
If the activities against dsRNA and hybrid RNA are related, there might be an effect of the Rh'ase III mutation (6)  0.1 mM EDTA. 16 ne (13.600 cnm) of nolv-I " cistronic T7 early RNA, synthesized 2)) W%O as described under "Experimental Procedure," and purified by preparative acrylamide gel electrophoresis and precipitation with ethanol. Incubations were carried out for 10 min at 3G", and reactions were terminated by addition of 0.1 ml of 0.05 M EDTA, and precipitated by adding 2.5 volumes of ethanol in the presence of 20 pg of tltNA and 2 pg Escherichia coli rRNA as carrier. After 1 hour at -15', RNA was collected by centrifugation and subjected to electrophoresis on a 2cj;, acrylamide-0.5c/c agarose gel as described previously (1, 2).
Sample Lanes a and a' represent control incubations to which no enzyme or competing RNA was added. Mattes b to f received RNase III Fraction VI prepared as described under "Experimental Procedure." I,a,re b received 3.5 units and Laj1e.s c to f received 7 units. Laaes b' to f' received sizing factor prepared as before (1). Lane b' received an amount which also contained 3.5 units (assayed with poly(A-U), Ref. 3) while I,a~es c' tof' received the equivalent of G units. Addition of potential competitor RNAs was as follows: La?res d and d' received 1 pg of phage f2 ssRNA; Lanes e and e' received 10 pg of f2 RNA; I,atles f and f' received 1 pg of PC dsRNA. Electrophoresis is represented as going from the bollon of the photograph to the fop, while 1, 0.7, 1.5, 0.3, and 1.1 refer to locations of T7 monocistronic mRNAs (1). Fig. 1 and Table I with extracts of the RXase III-deficient strain which had been fractionated through the ammonium sulfate stage as before (3, 5) reveal that while less than 55;; of the parent,al st,rain's level of activity agaitwt dsRNA remains, there appears to be a full amount of RNase H activity still present (&a not shown), The presence or absence of RNase H activity seems t,o have no effect on the ability of enzyme fractions to cleave '17 RSA. Roth the RNase 111 preparation which had abundant RNase II activity and the sizing factor preparat.ion which contained little or no RNase H act.ivity cleaved the 'I'7 polycistronic mRSA with a high degree of fidelity.

Section B: Size Distribution of Complete Digests oj &RNA and
Hybrid RlVA- Fig.  3 shows autoradiographs of two-dimensional analyses of several RNAs after nuclease treatment,. The fingerprinting technique employed is one in which oligotmcleotides of a size 20 bases long or less will bc denatured and incapable of by guest on March 24, 2020 http://www.jbc.org/  (29,30). This is the case because both dimensions are run in 7 M urea, while the second dimension is also carried out at 60". It is the second dimension, which separates oligonucleotide chains according t'o size (29), which allows us to estimate the average size of limit digestion products.
Since different homochromatography mixtures do not give exactly the same patterns, it was necessary to run the t'hree analyses shown in Fig. 3 at the same time and in the same chromatography tank. Fig. 3a shows the pattern following RNase Tl digestion of the three principal ribosome binding sites of bacteriophage fl. Each of the major oligonucleotides shown here has been sequenced (26), and in particular, the three prominent spots lowest in the picture (about one-third from the origin to the top) are 11, 12, and 13 bases long, while the prominent topmost dark spot at the right side corresponds to GMP.   Fig. 3b can be compared directly with regard to mobility with the three 11-to 13-base long oligonucleotides shown in Fig. 3~. We conclude that the median length of individual chains in the digested dsRNA of Fig. 3b is about 15 bases. Previous experiments (data not shown) have demonstrated that oligonucleotides larger than 20 bases did not leave the origin under the chromatography conditionh used here. It is remarkable that the vast majority of radioactivity is distributed in a narrow size range between about 10 and 18 bases in length. Fig. 3c hhows a similar two-dimensional analysis of hybrid RNA digested by Fraction VI. This fingerprint suggests that the median size for the digested hybrid RNA is quite a bit smaller than that observed in Fig. 36 and that oligonucleotides of a wider size range (5 to 15 bases) are present. RNA-One way to distinguish among the various nuclease activities in Fraction VI might be cndgroup analysis. If some activities release 5'-, and others 3'-phosphate termini, it would be unlikely that they could be caused by the same enzyme. Fig. 4 shows huch analyses for the two activities which release acid-soluble fragments, RNase III and RNase H. Fig. 4c shows a drawing of the separation obtained using the two-dimensional base composition analysis method described under "Esperimental Procedure." The four common (a'-, 3'-) nucleoside monophosphates produced by alkaline hydrolysis are readily distinguishable from each other and from the four (2'-, 3'-), 5'-nucleoside diphosphates (pxp's) as indicated. analysis of dsRNA and hybrid RNA following alkaline hydrolysis. Reactions carried out in 5 ~1 of 0.01 M Tris-HCl (pH 7.6), 0.01 M magnesium acetate, 0.13 M NH&l, 5% sucrose and were terminated by mixing with 5 ~1 of 0.4 M NaOH and incubated as described under "Experimental Procedure." After 18 hours, two-dimensional analysis was carried out also as described under "Experimental Procedure." a, polyoma IazP]dsRNA (3 X lo6 cpm, specific activity 3 X 107 cpm/rg) was incubated for 50 min in the above buffer at 37". b, polyoma [8zP]dsRNA was incubated as in a for 50 min with 0.3 unit of Fraction VI, rendering 85% of the radioactivity soluble in 5% reaction conditions in the absence of enzyme. Only the four (2'-, 3'-) nucleoside monophosphates are evident. After RNase III digestion, additional spots appear at positions characteristic of all four pXp's (Fig. 4b). This observation indicates that the activity in Fraction VI which cleaves dsRNA leaves a 5' phosphate terminus. Fig. 4, cl and e shows analyses of similar incubations of fl DNA: RNA hybrids. Again, digestion releases all four pXp's in detectable quantities. Table III shows the radioactivity recovered in the various nucleoside mono-and diphosphates for the two digestions trichloroacetic acid. c, drawing of Fig. 4b, indicating locations and identities of the spots visible in the autoradiographs. These locations were established both with unlabeled marker nucleotide compounds as before (25) and also with specific digests of ["*PI-RNAs containing only one of the four pXp's (3, 32). d, fl I>NA: RNA hybrid (1.5 X lo6 cpm [3*P]RNA, specific activity 3 X 10' cpm/pg of hybrid) was incubated for 50 min in the buffer described above; e, fl DNA:RNA hybrid was incubated as in d for 50 min with 0.3 unit of Fraction VI as in b. In these pictures, the first dimension is from right to left, while the second is from the bottom of the picture to the lop. analyzed in Fig. 4, b and e. Several further conclusions of interest may be dratin from this data. First, the 5' end groups released do not reflect the overall base composition of the dsRNA or hybrid RNA. In particular, digestion products of dsRNA are enriched for A and U at their 5' termini, while in hybrid RNA, pyrimidines are mildly favored. Finally, the data of Fig. 4 and Table III allow us to calculate  independently the size distribution of oligonucleotides in these digests. Although the digestion conditions were much less eshaustive than those in Fig. 3, we can see t,hat the average size of  Fig. 4, b and e were cut out and their Cerenkov radiation was determined as described under "Experimental Procedure." b Chain length was calculated by dividing the total RNA (loOo/,) by the per cent in pXp's and multiplying the quotient by 2.
oligonucleotide produced by RNase H is 10.7 bases, in agreement with Fig. 3. However, the estimate for dsRNA in Table III (31 bases) suggests that the milder digestion conditions employed fell short of producing a true limit digest.

DISCUSSION
Evidence presented here shows that RNase III and the enzyme responsible for processing polycistronic T7 mRNA cannot be separated. Earlier preparations of RNase III purified through the Fraction VI stage contained RNase H activity thereby explaining the activity of those preparations on DNA:RNA hybrids.
Zdenti&ation of Activities-The data in Section A of "Results" agree with observations of Crouch (21) that sizing of the proteins of an RNase III preparation is sufficient to bring about separation of RNase H from RNase III. Our preliminary observations on partially purified fractions from the RNase lIIdeficient strain suggest that RNase H activity is present through several steps of the RNase III purification (5). If this suggestion is confirmed with highly purified fractions, it will indicate in agreement with Nikolaev et al. (7), that RNase H activity is not affected by mutation in a cistron which is required for cleavage of dsRNA.
The most important observation.to emerge from the experiments in Figs. 1 and 2 and Table I is that the RNase III activity is probably responsible for processing polycistronic T7 mRNA. If this is the case, it puts limitations on both the sorts of cleavage sites we might expect within ssRNA and the endgroups we would 3055 obtain. The most straightforward model for cleavage sites would suggest that among many potential hairpin loops in an RNA precursor, only a few are large and stable enough to be recognized as dsRNA. Robertson and Hunter (28) have shown that RNase III preparations purified through the Fraction VI stage do not cleave ssRNA containing numerous potential hairpin loops, nor do such RNAs compete for digestion of stable dsRNA substrates. This point is also illustrated in Fig. 1, Lanes d, d' and e, e'. In addition, isolated bacteriophage f2 band 21, a 57-nucleotide RNA which can form a potential hairpin loop with 18 base pairs, is not cleaved by RNase III preparations (28). Since most hairpin loops found in Escherichia coli RNA are not cleaved by RNase III, and since little, if any, stable dsRNA is found in normal E. coli (31), there may be other features required for specific processing by RNase III. This is especially likely since preliminary studies on end groups on RNase III cleavage sites indicate that the same particular sequences are cleaved in each case.* Thus a combination of structure and specific sequence may be required for the processing event to occur.
RNase III is the second E. coli enzyme to which specific processing functions have been ascribed. Factors which determine the specificity of the other such enzyme, RNase P which cleaves the precursor to tyrosine tRNA (3, 34), have also not been defined. However, it is likely that, as with RNase III, some combination of sequence and structure will be involved (3, 35). Size Distribution of Oligonucleotide Digestion Products-The data in Fig. 3, a and b establish that an exhaustive digest of dsRNA by RNase III contains oligonucleotides averaging about 15 bases long, in agreement with previous observations (11). Crouch (21) has estimated 13 bases as the chain length of an exhaustive digest of poly (A-U) by RNase III. Using milder conditions of digestion, Crouch has estimated a chain length of 25 bases, in good agreement with our findings for such digests in Table III. The most striking aspect of the size distribution of RNase III digestion products of dsRNA is its narrow range, resulting in oligonucleotides between about 10 and 18 bases long (Fig. 3b).
End Group Analysis- Fig.  4 shows that both RNase III and RNase H cleave to form 5' phosphate end groups. Robertson et al. (5) found that an exhaustive digest of poly (A-U) contained about 10% of its radioactivity in the form of 3'-UMP, while no 5' UMP was released. Fig. 3b shows that no significant radioactivity was released into any mononucleotides.
Since the original purification of RNase III utilized E. coli K38, which contains wild type levels of RNase I, it is likely that the release of 3'-UMP was caused by low levels of contamination by this enzyme. With subsequent use of RNase I-deficient strains (28) it was possible to remove this contamination.
These observations and conclusions are in agreement with those of Crouch (21).
Our conclusion that RNase H activity present in E. coli RNase III preparations also forms 5' end groups (Fig. 4, d and e) is also in agreement with the suggestions of others using indirect assays (12,13) or RNase H's from eukaryotic sources (15). Finally, as observed by Rosenberg et al. (33) in vitro cleavage products of T7 polycistronic mRNA contain the same end groups (bearing 5'-phosphates) observed in monocistronic mRNAs isolated from infected cells. Furthermore, the two cleavages * In particular, Paddock and Abelson (32) have shown that bacteriophage T4 RNA Species I is cleaved specifically by E. coli extracts, and the activity responsible has been identified as RNase III (Robertson and Abelson, unpublished experiments). Furthermore, Rosenberg et al. (33) have found that similar specific endgroups are reproducibly generated in several of the RNase III mediated cleavages of T7 polycistronic mRNA.
made by highly purified RNase III preparations in T4 RNA Species I both yield 5'-phosphate and 3'-hydroxyl termini.3 Our :t, finding that RNase III digests have the same end groups further 13' ' increases the probability that cleavage of dsRNA and processing of T7 and rRNA precursors reflect different aspects of the same 14. enzyme.
In conclusion, we can expect the specific sites in cellular RNAs 15' which are processed by RNase III to have substantial double 16. helical structure; to be greater than 20 base pairs in length; to contain 5'-phosphate and 3'-hydroxyl endgroups after cleavage; and to contain, in all probability, at least one further charac-17.
teristic feature, either a common sequence or an additional struc-18. tural element, to differentiate them from the many regions of potential secondary structure now thought to reside at frequent 19. intervals in biological RNA sequences (36).