Isolating and sequencing the predominant 5'-ends of a specific mRNA in cells. II. End-labeling and sequencing

The reaction mechanism of ribonuclease II and its interaction with nucleic acid secondary structures

Ribonuclease II is a processive 3'- to 5'-exoribonuclease in Escherichia coli with two binding sites: a catalytic site associated with the first few 3'-nucleotides and an anchor site binding nucleotides approximately 15 to 25 from the 3'-end. When RNase II degrades single-stranded helical poly(C), the enzyme-substrate complex dissociates at discrete intervals of 12 nucleotides. RNase II stalled at the last rC of single-stranded 3'-(rC)(n)(dC)(m) oligonucleotides. The more residues released, the faster the stalled complex dissociated and the less it inhibited RNase II activity, i.e. the enzyme-substrate association weakened progressively. Using phosphodiesterase I (PDE I) as a probe, a method was developed to identify cytidine residues in (32)P-oligonucleotides interacting with a protein. PAGE bands corresponding to nucleotides 1-6 from the 3'-end were consistent with interaction at the catalytic site, and following a gap, bands approximately 15 to 25 from the 3'-end, with anchor site association. Both 3' and 5' binding were necessary to maintain the complex. Of most significance, the original anchor site nucleotides remained fixed at the anchor site while the 3'-end was pulled, or threaded, through the catalytic site, i.e. the substrate did not 'slide' through the enzyme. DNA oligonucleotides with double-stranded stem-loops were good competitive inhibitors of RNase II. A 3'-single-stranded arm was essential, while optimal binding required both 5'- and 3'-arms. PDE I probing indicated that the nucleotides at the anchor site were specified by the spatial distance from the catalytic site, and on only one of the duplex strands. When degradation of a structured RNA paused or stopped, the RNase II-product commenced cycles of dissociation-reassociation. Duplex strand binding by RNase II made complex DNA or RNA structures accessible to degradation by other nucleases and further verified the PDE I footprinting method.

The 5' ends of RNA oligonucleotides in Escherichia coli and mRNA degradation

The 5' ends of RNA oligonucleotides in Escherichia coli were identified to assess the contributions of specific endoribonucleases to the cleaving of bulk mRNA. About 60% of the total 5' ends have a 5' OH, and 40% a phosphate. Of those oligonucleotides with a 5'-OH end, 55% of the larger-sized molecules started with 5'-OH-A. With decreasing size there was a progressive decrease in its relative abundance, reaching 33% for the mononucleotide pool, close to its content in E. coli mRNA. In a mutant lacking RNase I* (a form of RNase I), the fraction starting with 5'-OH-A was even higher; 65-70% for oligonucleotides of any size, as well as the mononucleotides, whereas only 3-5% started with 5'-OH-U. Oligonucleotides with a 5'-P end were analyzed after pulse-labeling growing cells with 32Pi. Virtually all of them had a 5'-ppp-purine end which would result from transcription initiations, and there were four-times more G than A starts. The fraction of 5' ends with a monophosphate (5'-pN) was too low to measure. The known degradative enzymes of E. coli (RNases I, I*, M and R) release a 5'-OH oligonucleotide upon cleavage, whereas known processing endoribonucleases, e.g. RNases E, H, P and III, generate 5'-P oligonucleotides. Among these enzymes, RNase M is the only one known to enrich for 5'-OH-A ends, since its preference is for pyrimidine-A bonds [Cannistraro, V. J. & Kennell, D. (1989) Eur. J. Biochem. 181, 363-370]. It also gives a very low level of 5'-OH-U ends. These results are consistent with generalizations derived from our previous studies [Cannistraro, V. J., Subbaro, M. N. & Kennell, D. (1986) J. Mol. Biol. 192, 257-274] and suggest that RNase M is a primary endoribonuclease for mRNA degradation in E. coli. The results also indicate that RNase I* contributes a smaller fraction of cleavages to larger RNA oligonucleotides and accounts for most of the degradation of the very small oligonucleotides and almost all degradation of dinucleotide to mononucleotide.

Broad-specificity endoribonucleases and mRNA degradation in Escherichia coli

Crude extracts from Escherichia coli were screened for any broad-specificity endoribonuclease after the cell proteins were fractionated by size. In a mutant lacking the gene for RNase I (molecular mass, 27,156 Da), the only such activities were also in the size range of 23 to 28 kDa. Fractionation by chromatography on a strong cation-exchange resin revealed only two activities. One of them eluted at a salt concentration expected for RNase M and had the specificity of RNase M. It preferred pyrimidine-adenosine bonds, could not degrade purine homopolymers, and had a molecular mass of approximately 27 kDa (V. J. Cannistraro and D. Kennell, Eur. J. Biochem. 181:363-370, 1989). A second fraction, eluting at a higher salt concentration, was active against any phosphodiester bond but was about 100 times less active than are RNase I and RNase I* (a form of RNase I) in the wild-type cell. On the basis of sizing-gel chromatography, this enzyme had a molecular mass of approximately 24 kDa. We call it RNase R (for residual). RNase R is not an abnormal product of the mutant rna gene; a cell carrying many copies of that gene on a plasmid did not synthesize more RNase R. Our search for broad-specificity endoribonucleases was prompted by the expectation that the primary activities for mRNA degradation are expressed by a relatively small number of broad-specificity RNases. If correct, the results suggest that the endoribonucleases for this major metabolic activity reside in the 24- to 28-kDa size range. Endoribonucleases with much greater specificity must have as primary functions the processing of specific RNA molecules at a very limited number of sites as steps in their biosynthesis. In exceptional cases, these endoribonucleases inactivate a specific message that has such a site, and they can also effect total mRNA metabolism indirectly by a global disturbance of the cell physiology. It is suggested that a distinction be made between these processing and degradative activities.

RNase I*, a form of RNase I, and mRNA degradation in Escherichia coli

A previously unreported endoRNase present in the spheroplast fraction of Escherichia coli degraded homoribopolymers and small RNA oligonucleotides but not polymer RNA. Like the periplasmic endoRNase, RNase I, the enzyme cleaved the phosphodiester bond between any nucleotides; however, RNase I degraded polymer RNA as fast as homopolymers or oligomers. Both enzymes migrated as 27-kDa polypeptides by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and could not be separated by various chromatographic procedures. In rna insertion mutants, both enzymes were completely missing; the spheroplast enzyme is called RNase I*, since it must be a form of RNase I. The two forms could be distinguished by physical treatments. RNase I could be activated by Zn2+, while RNase I* was inactive in the presence of Zn2+. RNase I was inactivated very slowly at 100 degrees C over a wide pH range, while RNase I* was inactivated slowly by heat at pH 4.0 but much more rapidly as the pH was increased to 8.0. In the presence of a thiol-binding agent, the inactivation at the higher pH values was much slower. These results suggest that RNase I*, but not RNase I, has free sulfhydryl groups. RNase I* activity in the cell against a common substrate was estimated to be several times that of RNase I. All four 2',3'-phosphomonoribonucleotides were identified in the soluble pools of growing cells. Such degradative products must arise from RNase I* activity. The activity would be suited for the terminal step in mRNA degradation, the elimination of the final oligonucleotide fragments, without jeopardizing the cell RNA. An enzyme with very similar specificity was found in Saccharomyces cerevisiae, suggesting that the activity may be widespread in nature.

Cloning and sequencing the gene encoding Escherichia coli ribonuclease I: exact physical mapping using the genome library

The amino acid (aa) sequence of the N terminus of Escherichia coli RNase I was determined. A mixed oligodeoxynucleotide coding for that sequence was used to probe the 476 lambda clones of Kohara et al. [Cell 50 (1987) 495-508]. DNA from these clones carry almost the entire E. coli chromosome in overlapping segments. Two overlapping clones hybridized to the probe sequence. From one of them DNA containing the rna gene was subcloned and sequenced. The inferred protein contains 245 aa residues and has an Mr of 27,156, which agrees with earlier estimates from sodium dodecyl sulfate-polyacrylamide-gel electrophoresis. RNase I is close to twice the size of pancreatic RNase A, but both enzymes contain eight Cys and four His; those aa are important for structure and function of RNase A. Proximal to the rna gene is a sequence that would code for a 23-aa peptide which conforms to consensus rules for signal peptides, and thus should transport this periplasmic enzyme. Sites for eight restriction enzymes had been mapped on each lambda clone. By relating to the map for that specific region, it was possible to position the rna gene exactly at 659 kb from the thr locus (time zero on a time scale of 100 min). This physical mapping gave a more precise (exact) map position based on distance than was possible using genetic mapping based on a time scale derived from conjugation, and should be applicable for mapping many other E. coli genes.

Purification and characterization of ribonuclease M and mRNA degradation in Escherichia coli

A previously unreported endoribonuclease has been identified in Escherichia coli, which has a preference for hydrolysis of pyrimidine-adenosine (Pyd-Ado) bonds in RNA. It was purified about 7000-fold to give a single band after SDS/polyacrylamide gel electrophoresis; the eluted protein gave the same RNase specificity. The sizes of the native and denatured enzymes agreed suggesting that the enzyme exists as a monomer of approximately 26 kDa. It is called RNase M. The only other reported broadly specific endoribonuclease in E. coli is RNase I, a periplasmic enzyme. Based on differences in charge, heat stability and substrate specificity, it was clear that RNase M is not RNase I. The specificity of RNase M was remarkably similar to that of pancreatic RNase A even though the two enzymes differ in charge characteristics and size. Earlier studies had shown that mRNA from the lactose operon of E. coli is hydrolyzed in vivo primarily between Pyd-Ado bonds [Cannistraro et al. (1986) J. Mol. Biol. 192, 257-274] We propose that this major RNase activity accounts for these cleavages observed in vivo and that it is the endonuclease for mRNA degradation in E. coli.

Evidence for endonucleolytic cleavages in decay of lacZ and lacI mRNAs

S1 nuclease mapping revealed lacZ mRNA molecules whose 5' and 3' ends were internal to the transcription start and consistent with cleavages at pyrimidine-adenosine bonds 20 to 50 nucleotides apart. With the net 5'-to-3' direction known, lacZ mRNA is probably degraded by sequential cleavages of naked mRNA at vulnerable sites exposed by transit of the last translating ribosome.

Isolating and sequencing the infrequent 3'-ends of a specific mRNA

Procedures are described for identification of very infrequent in vivo 3'-ends of RNA. After purification by filter hybridization, the 3'-ends were labeled with [5'-32P] cytosine-3'-P in the RNA ligase reaction. Significantly fewer counts were incorporated in the ligase reaction than in the polynucleotide kinase reaction to label 5'-ends. The incorporation was increased by increasing the RNA concentration 5-10 fold by using only one round of filter hybridization. Non-specific RNA binding could be eliminated by RNase A treatment of the filter if a great excess of denatured heterologous DNA was immobilized along with the DNA probe. Significant amounts of DNA were released when eluting the hybrid RNA from such filters. DNA inhibited the ligase reaction, while its DNase products were even more inhibitory. Treatment of the DNase products with alkaline phosphatase completely eliminated the inhibition. We detected no spurious 5'- or 3'-ends generated in the hybrid RNA by RNase A activity used to reduce the non-specific RNA. Also, RNase T1 could be used in place of RNase A to eliminate non-specific RNA binding, but about 25 times more RNase T1 (microgram/microgram) was needed. We used partial alkali digestion to sequence 3'-ends. A major (one hit) and minor (two hit) set of products were produced which could be distinguished from each other by alkaline phosphatase treatment and homochromatography of the products.

Specific endonucleolytic cleavage sites for decay of Escherichia coli mRNA

The polycistronic lac mRNA of Escherichia coli contains three messages. The rate of degradation of the second (lacY) message was observed to be equal to that of the third (lacA), and each decayed twice as fast as did the first (lacZ). Specific 5'- and 3'-ended lacY mRNA molecules could be recovered from cells; most likely, they are generated from endonucleolytic cleavages that are a part of the degradative process. They were observed by S1 nuclease mapping, and the exact 5'- and 3'-end oligonucleotides of many of them were identified by direct sequencing. Almost all of the molecules started with a 5' adenosine that would be preceded by a pyrimidine. The specificity was further restricted by neighboring nucleotides, and analysis of the data suggested that 5'-U-U decreases-A-U- is especially vulnerable. Also, computer analyses predicted the most stable secondary structures of selected segments of the mRNA and suggested that cleavages may only occur in regions of single strandedness. A model of mRNA degradation is proposed based on these observations and earlier ones. There is no unique target on a message for the initial inactivating attack: any region free of ribosomes is vulnerable, but for statistical reasons the initial attack of most molecules is near the ribosome-loading site. With no further ribosome loading, the newly unprotected 5' ends are "chopped off" at one of the next preferred target sites almost as fast as the last ribosomes moves down the mRNA.

Isolating and sequencing the predominant 5'-ends of a specific mRNA in cells. I. Purification by filter hybridization

Procedures are considered for purification of a specific procaryotic RNA by successive hybridizations to DNA immobilized to nitrocellulose with special consideration of problems associated with subsequent end-labeling in the T4 polynucleotide kinase reaction. (1) Inhibitors of the kinase can be associated with the plasmid but were removed by electrophoresis of the DNA fragment through polyacrylamide. (2) Residual soluble acrylamide, contaminating the DNA and preventing its efficient retention to nitrocellulose, could be removed by DE52 chromatography. (3) Short denatured DNA required high salt (0.9 M) to bind to nitrocellulose but reannealed quickly at those salt concentrations unless applied at less than or equal to 0.3 micrograms/ml at 4 degrees C with a flow rate of 1 ml/min. (4) The kinetics of the hybrid reaction were a function of DNA length, concentration, and temperature. (5) Formamide was a more effective denaturing agent to remove hybrid RNA from the filter than either 12 M urea or 8 M guanidine-HCl, but caused significant release of DNA from the nitrocellulose as well as another potent inhibitor of the kinase reaction. The release of DNA and other kinase inhibitors was greatly reduced by eluting in boiling water.