Polyadenylation of stable RNA precursors in vivo

Abstract

Polyadenylation at the 3′ terminus has long been considered a specific feature of mRNA and a few other unstable RNA species. Here we show that stable RNAs in Escherichia coli can be polyadenylated as well. RNA molecules with poly(A) tails are the major products that accumulate for essentially all stable RNA precursors when RNA maturation is slowed because of the absence of processing exoribonucleases; poly(A) tails vary from one to seven residues in length. The polyadenylation process depends on the presence of poly(A) polymerase I. A stochastic competition between the exoribonucleases and poly(A) polymerase is proposed to explain the accumulation of polyadenylated RNAs. These data indicate that polyadenylation is not unique to mRNA, and its widespread occurrence suggests that it serves a more general function in RNA metabolism.

Cellular RNA molecules have long been divided into two groups. The first group, the stable RNAs, includes mainly rRNA, tRNA, and a variety of other small RNAs, and represents >95% of the total cellular RNA population. These RNAs have long lifetimes relative to the generation times of the cells in which they reside. The second group, unstable RNAs, consists primarily of mRNAs, which are a small fraction of the total RNA population and have half-lives that are usually much shorter than the generation time of the cell. The unstable RNAs generally contain a poly(A) tract at their 3′ ends that contributes to their turnover; for mRNAs, poly(A) contributes to their role in translation (1–5). However, polyadenylation of stable RNAs has rarely been seen, and has not been considered important for stable RNA metabolism.

In previous studies of the maturation of tRNA (6, 7), 5S RNA (8), and other small, stable RNAs (4.5S, 6S, M1, and tmRNA) (9) in Escherichia coli, it was shown that exoribonucleolytic trimming was a necessary step in the formation of the 3′ termini of these various molecules. Thus, in the absence of the requisite 3′ to 5′ exoribonucleases, precursor products with extra nucleotides at their 3′ ends accumulated. In some instances, however, these extra sequences were longer than expected. For example, 5S RNA is known to be released from long rRNA transcripts by RNase E endonucleolytic cleavages that leave three additional residues at each end of the processing intermediate (10–13). However, products were found to accumulate in some exoribonuclease-deficient strains that contained as many as 10 extra nucleotides at their 3′ ends (8). Likewise, in a recent study of the maturation of M1 RNA, the catalytic subunit of RNase P, products with up to six additional 3′ residues were observed (9), despite the fact that RNase E is thought to cleave this precursor at a position only one or two nucleotides downstream from the mature 3′ terminus (14, 15).

In this paper we provide an explanation for these unexpected observations. To do so, we determined the nucleotide sequences at the 3′ ends of the 5S and M1 RNA products that accumulate in the multiple exoribonuclease-deficient cells. This was accomplished by a procedure using reverse transcription–PCR (RT-PCR). Surprisingly, we found that both the 5S and the M1 RNA products contain oligo(A) tails of varying lengths at their 3′ ends that would account for the increased size of these molecules. Based on these findings, we extended the analyses to precursors of other stable RNAs–tRNA, 4.5S RNA, tmRNA, 6S RNA, and rRNA– and found that all of them accumulate products containing oligo(A) tails. These data indicate that polyadenylation is not unique to messenger or other unstable RNAs.

EXPERIMENTAL PROCEDURES

Bacterial Strains and Plasmid.

E. coli K12 strain CA244 I⁻ (lacZ, trp, relA, spoT, rna) was considered wild-type for these studies. Exoribonuclease-deficient derivatives of CA244 were described previously (6, 7). The poly(A) polymerase-deficient strain (PAP⁻) was constructed by using phage P1-mediated transduction. For this purpose, a pcnB–kan (kanamycin) mutation was introduced into CA244PH⁻D⁻BN⁻ (16). The resulting PAP⁻ transductant was made RNase I⁻, RNase T⁻ by consecutive transductions with the corresponding mutant alleles. All mutations were confirmed by using enzyme-activity assays. The mutations in polynucleotide phosphorylase (PNPase), RNase T, RNase PH, and RNase D are interruption and/or deletion mutations and are devoid of the relevant activity. The mutations in RNase I, RNase II, and RNase BN have not been defined, but they result in >95% loss of the relevant activity. Plasmid pJL89 carrying the wild-type poly(A) polymerase gene was kindly provided by Sidney Kushner (University of Georgia) (17).

Materials.

[³H]ATP and [³H] poly(A) were purchased from Amersham. [γ-³²P]ATP was from DuPont/NEN. The RNase T substrate, tRNA-C-C-[³H]A, was prepared as described (18). Phage T4 RNA ligase was the product of New England Biolabs. Phage T4 polynucleotide kinase and Moloney murine leukemia virus reverse transcriptase were from GIBCO/BRL. RNasin was obtained from Promega. Sequagel for high-resolution Northern analysis was purchased from National Diagnostics. The GeneAmp PCR Reagent kit was purchased from Perkin–Elmer. The TA cloning kit was the product of Invitrogen. Oligonucleotides used for Northern blot analysis were described previously (7–9). The oligoribonucleotide (pUGGUGGUGGAUCCCGGGAUCp) was used as a linker ligated to the 3′ end of cellular RNA. An oligodeoxynucleotide complementary to the linker (GATCCCGGGATCCACCACCA) served as the primer for reverse transcription and as one of the primers for PCR. The second PCR primers specific for each RNA species were: 5S rRNA (CCGATGGTAGTGTGGGGTCTCC); M1 RNA (GCTGGCCTAGATGAATGACTG); tRNA^Tyr (CAGACTGTAAATCTGCCGTC); tmRNA (GCATGTAGTACCGAGGATGTAG); 4.5S RNA (CAAGGCAGATGACGCGTGTGCC); 6S RNA (GACGACACATTCACCTTGAACC); 23S rRNA (GATAGGCCGGGTGTGTAAGCG); and 16S rRNA (CCTGCGGTTGGATCACCTCC). RNA samples were prepared as described (8).

Northern Blot Analysis.

Northern blotting was performed as described (7–9). To obtain high resolution, RNA samples were separated on denaturing polyacrylamide gels of various concentrations and allowed to migrate various distances. Transferral to a membrane and hybridization were carried out by routine procedures.

RT-PCR Cloning.

An excess amount of the synthetic oligoribonucleotide linker (20 pmol) was ligated to 0.5 μg of total RNA in 10 μl by using 20 units of T4 RNA ligase (19). The ligation reaction was carried out in 50 mM Hepes (pH 7.5), 20 mM MgCl₂, 3 mM DTT, 0.1 mM ATP, 10% (vol/vol) dimethyl sulfoxide, and 10 μg/ml BSA (RNase-free), and incubated at 16°C overnight. Aliquots of the ligation products were annealed to the oligonucleotide complementary to the linker, and the cDNA strands were synthesized by using 400 units of Moloney murine leukemia virus reverse transcriptase under conditions described previously (8). The resulting RNA/DNA mixture was used directly in the following PCR reactions, with one primer the same as that used for making cDNA, and the other primer specific to the RNA of interest. The PCR products were amplified by using the TA cloning system under conditions provided by the manufacturer. All of the clones containing the inserts were sequenced; the original RNA sequences were deduced from the sequences preceding that of the linker. Similar procedures were reported previously (20–22).

Enzyme Activity Assays.

Poly(A) polymerase activity was determined by incorporation of [³H]ATP into poly(A) primers (23). The activities of RNase T and RNase I were measured against tRNA-C-C-[³H]A and [³H]poly(A), respectively (18, 24).

RESULTS

Polyadenylation of 5S RNA and M1 RNA.

In the absence of the exoribonuclease RNase T, 5S RNA products containing 1–10 extra 3′ residues are observed (ref. 8; Fig. 1, lane 2). Sequence analysis of clones derived by the RT-PCR procedure revealed the presence of multiple A residues at the 3′ ends of many of these 5S RNA products present in an RNase T⁻PH⁻D⁻BN⁻ mutant strain; no clones derived from mature 5S RNA were recovered (Table 1). It should be noted that the RT-PCR procedure may not quantitatively reflect the distribution of RNA molecules in the original preparation (e.g., Northern analysis indicates products primarily with two or three extra residues (ref. 8; Fig. 1, lane 2); however, the procedure is well suited for identifying qualitatively the diversity of 3′ termini present. The number of A residues observed varies from zero to seven among the 15 clones analyzed from the mutant strain (Table 1). Because three A residues normally are present in the 5S RNA precursor at positions 2–4 downstream of the mature 3′ terminus, it was not possible from the sequence data alone to ascertain whether all of the observed A residues actually represent posttranscriptional polyadenylation. However, as will be discussed below, all depend on the presence of PAP, strongly suggesting that they all arise by a posttranscriptional reaction. In contrast to the clones from the exoribonuclease-deficient cell, clones derived from wild-type RNA had primarily the mature 5S RNA sequence (Table 1). Interestingly, two incomplete 5S RNA chains (one with an extra A residue), one 5S RNA precursor, and one mature 5S RNA molecule with an extra A residue also were observed.

Figure 1.

Northern analysis of stable RNA species from wild-type and mutant E. coli strains. Total RNA from each strain was isolated, and high-resolution Northern blotting was performed as described in Experimental Procedures. PAP activities were measured for all of the cells. No detectable activity was seen in the PAP⁻ strain, and two- to-threefold elevation of activity was seen in the pJL89-containing strains. M shows the position of the mature RNA species. The length of the precursor residues was determined by running a DNA sequencing ladder side-by-side (7–9).

Table 1.

Sequences of RT-PCR clones derived from 5S RNA

Sequence	No. of clones
	Wild type	RNase T−PH−D−BN−
GGAACTGCCAGGCAT (mature)	8	0
GGAACTGCCAGGCA	1	0
GGAACTGCCAGGA	1	0
GGAACTGCCAGGCATA	1	1
GGAACTGCCAGGCATC	1	4
GGAACTGCCAGGCATCA	0	2
GGAACTGCCAGGCATCAA	0	2
GGAACTGCCAGGCATCAAA	0	1
GGAACTGCCAGGCATCAAAA	0	2
GGAACTGCCAGGCATCAAAAA	0	1
GGAACTGCCAGGCATCAAAAAAA	0	2

Sequences around the 3′ end of 5S RNA are shown, and are written 5′ to 3′. Plain text indicates sequences present in mature 5S RNA. Underlined nucleotides represent precursor residues present in the primary transcript. Residues in boldface type are not encoded by 5S RNA genes. Italicized A residues may either be encoded or added posttranscriptionally; data presented in Table 3 and Fig. 1 support posttranscriptional addition. 

A similar situation was found with M1 RNA. In the RNase T⁻PH⁻D⁻BN⁻ mutant strain, the major M1 RNA product that accumulates, based on Northern analysis, appears to be four nucleotides longer than the mature form present in the wild-type cell (ref. 9; Fig. 1, lanes 1 and 2). RT-PCR and sequence analysis of 12 clones derived from the mutant RNA and 7 clones derived from the wild-type RNA verified the conclusions from Northern blotting and also revealed the presence of extra A residues at the 3′ termini of the M1 RNAs accumulating in the exoribonuclease-deficient strain (Table 2). M1 RNA from the wild-type cell was primarily mature, although one shorter RNA clone and one clone containing 25 precursor residues were also observed (Table 2). The latter clone presumably was derived from a primary transcript that had not yet been cleaved by RNase E. No extra A residues were seen in any wild-type clones.

Table 2.

Sequences of RT-PCR clones derived from M1 RNA

Sequence	No. of clones
	Wild type	RNase T−PH−D−BN−
CGGTCAGTTTCACCT (mature)	5	0
CGGTCAGTTT	1	0
CGGTCAGTTTCACCTGATTTACGTAAAACCCGCTTCGGC	1	0
CGGTCAGTTTCACCTGAA	0	2
CGGTCAGTTTCACCTGAAA	0	10

Symbols are the same as in Table 1. 

DNA sequencing of the PCR product of the rnpB gene encoding M1 RNA indicated that the sequence of the gene downstream of the mature 3′ terminus in the mutant is GATTT, the same as that published previously (14, 15), and the same as the wild-type precursor sequence (Table 2). Thus, we conclude that the A residues in the third and fourth positions must be added posttranscriptionally. The origin of the A residue in the second position cannot be ascertained from these data. Nevertheless, these data show that for both 5S and M1 RNAs the longer-than-expected products observed are likely the result of the posttranscriptional addition of adenylate residues.

Polyadenylation of Other Stable RNAs.

Our findings with 5S RNA and M1 RNA prompted us to examine all of the other stable RNAs of E. coli for the presence of nonencoded A residues at their 3′ termini. For most of these other stable RNAs, the expected lengths of the processing intermediates are not known. The stable RNA species examined included tRNA₁^Tyr and tRNA₂^Tyr, 4.5S RNA, tmRNA, 6S RNA, and 23S and 16S rRNAs. These data are summarized in Table 3; they provide a description of all of the RNA products observed for the various stable RNA species (tRNA₁^Tyr and tRNA₂^Tyr are considered to be representative of all the tRNAs).

Table 3.

Summary of sequences of RT-PCR clones derived from stable RNA species from wild-type and exoribonuclease-deficient cells

RNA species	Strain	Shorter than mature	Mature	Mature + adenylates	Precursor	Precursor + adenylates
5S	Wild-type	2 {1}*	8	1 {1}	1	0
	T−PH−D−BN−	0	0	1 {1}	4	10 {1–7}
	T−PH−D−BN−PAP−	3	1	0	6	0
M1	Wild-type	1	5	0	1	0
	T−PH−D−BN−	0	0	0	0	12 {1–3}
tRNATyr	Wild-type	0	5	0	0	0
	T−PH−D−BN−/tyr1	1	1	1 {5}	8	19 {1–3}
	T−PH−D−BN−/tyr2	0	8	10 {3–5}	0	0
tmRNA	Wild-type	0	5	0	0	0
	T−PH−D−BN−	0	10	6 {3–4}	0	0
4.5S	Wild-type	0	4	0	0	0
	T−PH−D−BN−	1 {5}	1	0	2	13 {3–4}
	T−PH−D−BN−PAP−	0	4	0	5	0
6S	Wild-type	1	5	0	0	0
	T−PH−D−BN−	0	4	1 {4}	9	5 {1–4}
23S	Wild-type	0	7	0	0	0
	T−PH−D−BN−	2	1	0	10  (5)	9 {1–5}
16S	Wild-type	2	5	0	1  (1)	0
	T−PH−D−BN−	1	9	0	3  (3)	1 {4}

The values presented are the number of RT-PCR clones recovered derived from RNA with the indicated 3′ terminal sequence. Values in parentheses ( ) indicate the number of clones having a single 3′ terminal adenylate residue that either can be encoded or added by polyadenylation. Values in brackets { } are the lengths of adenylate tails shown to be added posttranscriptionally. 

^*One of the two clones for 5S RNA has a nonencoded adenylate residue at its 3′ end. 

In general, RNA from wild-type cells consists primarily of the mature species, although molecules shortened by one or several residues or precursors are seen at a low frequency in a few cases. This observation is in complete agreement with analyses carried out by using Northern blotting (refs. 7–9; Fig. 1) and indicates that in wild-type cells stable RNA species are accurately and efficiently processed at their 3′ termini.

The situation is quite different in the exoribonuclease-deficient strain. Most of the RNA species have low levels of the mature form, whereas species with 3′ extra residues have increased dramatically. Among these products are some that contain just precursor sequences; others contain precursor sequences plus runs of adenylate residues. Although the proportion of clones with adenylate residues varies among the different RNA species, for many RNAs they represent the majority of the forms identified. Based on these data, it appears that every stable RNA species in E. coli can be polyadenylated when 3′ processing is slowed or cannot be completed because of RNase deficiency. The length of the adenylate sequences varies from as few as one to as many as seven residues, and as shown above for 5S and M1 RNA, these lengths can vary even for a single RNA species. The data in Table 3 also show that adenylate residues are less frequently found added to the 3′ end of the mature form of the RNA, indicating that this process is confined largely, although not exclusively, to immature forms.

An additional, interesting piece of information that emerged from these studies concerns the 3′ maturation of 23S and 16S rRNAs. Sequences of 23S RNA clones derived from the RNase T⁻PH⁻D⁻BN⁻ mutant strain consisted almost entirely of precursors containing three to eight additional residues; about half of the clones recovered also had poly(A) tails following the precursor-specific residues (Table 3). All of the clones recovered from wild-type cells had the mature 23S RNA sequence. These data strongly suggest that exoribonucleolytic trimming is required for 23S RNA maturation. In contrast, maturation of 16S RNA was not affected by the absence of exoribonucleases, and the few precursor sequences that were obtained from both wild-type and mutant cells contained the full 33 extra 3′ residues present after RNase III cleavage. Thus, the mature 3′ end of 16S RNA most likely arises from an endonucleolytic cleavage. However, further work will be necessary to identify the RNases responsible for 3′ maturation of 23S and 16S RNAs.

To ensure that the addition of adenylate residues to stable RNA precursors was not a peculiarity of the RNase T⁻PH⁻D⁻BN⁻ strain, other RNase-deficient strains were also examined. For example, 13 clones derived from 4.5S RNA from an RNase T⁻PH⁻ strain were sequenced. Of these, 12 contained adenylate runs varying from two to seven residues in length that must have originated from posttranscriptional addition (data not shown). Northern analysis of 4.5S, 5S, 6S, M1, and tmRNAs from an RNase T⁻PH⁻D⁻BN⁻PNP⁻ mutant strain showed that products of the same length accumulated as in the quadruple mutant containing PNP (data not shown). Northern analysis of RNA from an RNase T⁻D⁻BN⁻II^tsPNP⁻ strain at the nonpermissive temperature also revealed that the absence of both RNase II and polynucleotide phosphorylase do not alter the length of the A tracts on the stable RNA precursors, in contrast to what has been observed for poly(A) tails on mRNA (25, 26).

Role of PAP I.

Based on sequence analysis of the RT-PCR clones, the additional residues observed on stable RNAs appear not to be encoded. To confirm this conclusion, and also to determine the origin of those A residues which may result from either transcription or posttranscriptional addition, we carried out Northern blotting and RT-PCR analyses on RNAs from a mutant strain lacking the exoribonucleases and PAP I, the enzyme most likely required for the polyadenylation process (27). As shown in Fig. 1, lane 3, the absence of PAP I (PAP⁻) reduces the length of the M1, 5S, 4.5S, tRNA₁^Tyr, tRNA₂^Tyr, and tmRNA products that normally accumulate in the multiple exoribonuclease-deficient strain. However, products longer than wild-type RNA are still present in the M1, 5S, 4.5S, and tRNA₁^Tyr products because of the poor processing activity of the mutant cell. Only mature sized tRNA₂^Tyr and tmRNA were found, suggesting that all of the adenylate residues on these two RNAs are added directly to the mature 3′ end by PAP I. Overexpression of PAP I, encoded by plasmid pJL89 (17), also leads to increased lengths or increased amounts of some of the longer RNA species that accumulate in the RNase mutant (Fig. 1, lane 5). Overexpression of PAP I in a wild-type strain, in contrast, has no effect (Fig. 1, lane 6). These data confirm that polyadenylation occurs on all stable RNAs when they cannot be processed to the mature form, and they show that PAP I is the enzyme responsible for this modification.

In a few cases, RT-PCR was used to examine the effect of PAP I in more detail. As shown in Table 3, all of the 5S RNA clones examined in the PAP⁻ strain lacked adenylate residues, including the three A residues that might have been encoded; this absence indicates that they all arose by posttranscriptional polyadenylation. For 4.5S RNA (Table 3), the absence of PAP also eliminated the extra adenylate residues. However, in this case, each of the five precursor clones recovered retained a single A at the 3′ terminus, indicating that this residue arises from transcription. Taken together, these data demonstrate that almost all of the extra A residues found on stable RNA precursors are the products of polyadenylation by PAP I.

DISCUSSION

The extensive polyadenylation of stable RNA precursors in E. coli indicates that this process is not limited to mRNA and other unstable RNAs, as has long been believed. Rather, it may be a common feature of all RNAs under certain conditions, such as when they cannot be processed to their mature forms, as described here. The presence of nonencoded adenylate residues at the 3′ terminus of some stable RNA species has been seen on occasion (22, 23, 30–33). Generally, however, little has been made of these observations, and the origin of the adenylate residues was not ascertained. In a recent paper it was shown that 5–10% of telomerase RNA in yeast is polyadenylated, and that the same machinery that polyadenylates mRNA is used to accomplish polyadenylation in telomerase RNA (34). However, the information presented here showing that there is widespread polyadenylation of precursors to essentially all stable RNAs is quite unexpected and may have important implications for a complete understanding of RNA metabolism in general and of the role of polyadenylation in particular. It is clear that the poly(A) tails on stable RNAs seen here are considerably shorter than those previously observed for E. coli mRNAs, which range from 10 to 60 residues in length (25, 26). The explanation for these length differences between the two classes of RNA is not yet understood.

Polyadenylation of stable RNA precursors is most easily explained by a stochastic competition model in which processing exoribonucleases and PAP I compete for the accessible 3′ ends of RNA precursors. Under normal circumstances the exoribonucleases are much more efficient and RNA precursors are matured. Thus, as we showed, in a wild-type cell with the full complement of RNases present, even overexpression of PAP I has no effect on RNA maturation. However, when most of the exoribonucleases needed to process stable RNA are absent, maturation slows down dramatically, leaving RNA 3′ ends that are accessible to polyadenylation by PAP. Residual exoribonuclease activities may then shorten the poly(A) tails so that what we observe represents a steady state of a dynamic lengthening and shortening process.

Polyadenylation of mature RNAs was also observed, but only at a low level. We presume that mature RNAs generally are not subject to polyadenylation because their 3′ termini are somehow protected, perhaps by aminoacylation or burial within a ribonucleoprotein particle. In fact, the process of RNA maturation may have evolved as a mechanism to ensure that only a properly synthesized and folded molecule would fit correctly within a ribonucleoprotein, and that if not properly synthesized and folded, the RNA’s 3′ terminus would remain accessible.

This model provides another example that the enzymes acting on RNA are exquisitely balanced in vivo such that the correct outcome for overall RNA metabolism is achieved (7–9). Any enzyme that can act on a particular RNA substrate may do so. What determines the correct outcome are the relative levels of the different enzymes, the structure of the RNA itself, and mechanisms for protecting the RNA. When an occasional error still is made, repair or degradation processes come into play to correct or eliminate the aberrant RNA. This explanation raises the interesting question of the function that may be served by the polyadenylation of stable RNAs. We propose that this process may provide the cell with a mechanism for eliminating “bad” RNAs. If an RNA precursor molecule were incorrectly synthesized or folded such that it could not be converted to the mature form (22, 31, 35–41), a situation analogous to that described here would ensue, i.e., polyadenylation of the accumulated RNA precursor. Inasmuch as polyadenylation is thought to promote degradation of RNAs in prokaryotic cells (3, 4, 25, 41, 42), such a signal would serve to remove stable RNAs that were nonfunctional and would thereby prevent their accumulation, which might be deleterious to cells.

ABBREVIATIONS

PNP

polynucleotide phosphorylase

PAP

poly(A) polymerase

RT-PCR

reverse transcription–PCR