Stabilization of discrete mRNA breakdown products in ams pnp rnb multiple mutants of Escherichia coli K-12

The degradation of mRNA in Escherichia coli is thought to occur through a series of endonucleolytic and exonucleolytic steps. By constructing a series of multiple mutants containing the pnp-7 (polynucleotide phosphorylase), rnb-500 (RNase II), and ams-1 (altered message stability) alleles, it was possible to study general mRNA turnover as well as the degradation of specific mRNAs. Of most interest was the ams-1 pnp-7 rnb-500 triple mutant in which the half-life of total pulse-labeled RNA increased three- to fourfold at the nonpermissive temperature. RNA-DNA hybridization analysis of several specific mRNAs such as trxA (thioredoxin), ssb (single-stranded-DNA-binding protein), uvrD (DNA helicase II), cat (chloramphenicol acetyltransferase), nusA (N utilization substance), and pnp (polynucleotide phosphorylase) demonstrated two- to fourfold increases in their chemical half-lives. A new method for high-resolution Northern (RNA) analysis showed that the trxA and cat mRNAs are degraded into discrete fragments which are significantly stabilized only in the triple mutant. A model for mRNA turnover is discussed.

The degradation of mRNA in Escherichia coli is thought to occur through a series of endonucleolytic and exonucleolytic steps. By constructing a series of multiple mutants containing the pnp-7 (polynucleotide phosphorylase), rnb-S00 (RNase II), and ams-l (altered message stability) alleles, it was possible to study general mRNA turnover as well as the degradation of specific mRNAs. Of most interest was the ams-1 pnp-7 rnb-500 triple mutant in which the half-life of total pulse-labeled RNA increased threeto fourfold at the nonpermissive temperature. RNA-DNA hybridization analysis of several specific mRNAs such as trxA (thioredoxin), ssb (single-stranded-DNA-binding protein), uvrD (DNA helicase II), cat (chloramphenicol acetyltransferase), nusA (N utilization substance), and pnp (polynucleotide phosphorylase) demonstrated twoto fourfold increases in their chemical half-lives. A new method for high-resolution Northern (RNA) analysis showed that the trxA and cat mRNAs are degraded into discrete fragments which are significantly stabilized only in the triple mutant. A model for mRNA turnover is discussed.
Although factors that affect mRNA decay can play an important role in controlling gene expression, analyzing the mechanism of mRNA degradation in procaryotes has been particularly difficult because most mRNAs undergo rapid chemical decay, defined as the degradation to oligonucleotides and mononucleotides. Thus, the average mRNA molecule in Escherichia coli has a chemical half-life of between 1.3 to 4.0 min depending on the growth temperature (12,18).
Despite this problem, previous studies have established some general aspects of mRNA decay. (i) The rate of mRNA turnover has no relation to the length of the gene (4). (ii) The segments of the message that decay most rapidly may be located anywhere on the mRNA (35). (iii) The growth rate affects the stability of some transcripts, but this phenomenon is not general (25). (iv) The nucleases responsible for stable RNA (ribosomal and transfer) processing in general do not seem to be involved in mRNA degradation (1,8).
The product of the altered message stability (ams) locus has been implicated in mRNA turnover, since the chemical half-life of total pulse-labeled RNA increased significantly at the nonpermissive temperature (26). However, the functional half-lives of several messages were not altered (26). The location of the ams gene at 23 min on the E. coli chromosome (27) makes it unlikely that it encodes any of the known RNases. Although a DNA fragment that reportedly complements the ams-J mutation has been cloned (6), the function of the Ams protein still has not been established.
As part of our continuing effort to characterize the biochemical pathway(s) by which mRNAs are degraded, we constructed a series of multiple mutants containing the ams-J, pnp-7, and rnb-500 alleles. All the multiple mutants * Corresponding author. analyzed showed increased stability of pulse-labeled mRNA. In addition, the chemical half-lives of specific mRNAs such as trxA (thioredoxin), ssb (single-stranded-DNA-binding protein), uvrD (DNA helicase II), cat (chloramphenicol transacetylase), nusA (N utilization substance), and pnp (PNPase) were also increased in these mutants. Northern (RNA) analysis of the trxA and cat mRNAs on polyacrylamide gels demonstrated the existence of discrete mRNA breakdown products which were dramatically stabilized in an ams-J pnp-7 rnb-500 triple mutant. This work provides conclusive evidence that the ams, pnp, and rnb gene products are required for normal mRNA turnover.

MATERIALS AND METHODS
Bacterial strains and plasmids. The relevant genotypes of the bacterial strains used are listed in Table 1. Mutants deficient in either PNPase (pnp-7) (31) or RNase II (rnb-500) (12) or with the ams-J allele (26) have been previously described. Both the ams-J and rnb-S00 alleles encode thermolabile proteins, while pnp-7 appears to be a nonsense mutation (S. R. Kushner and R. Ivarie, unpublished data). P1-mediated transduction (38) was used to construct isogenic strains of the desired genotypes in either E. coli C600 or MG1693. The ams gene is 30% cotransducible with pyrC (27), rnb is 80% linked with pyrF (2), and pnp is 77% cotransducible with argG (30). Transposable drug resistance elements inserted into or adjacent to these markers (specifically, TnJO insertions adjacent to argG and in pyrC, and TnS in pyrF) were used for the transductions. Drug-resistant transductants were selected and screened for either arginine or uracil requirements by replica plating to appropriate minimal agar medium. P1 lysates grown on strains containing the ams-J, pnp-7, or rnb-500 mutation were used to transduce the appropriate nearby marker to prototrophy. Transductants were screened qualitatively for PNPase and RNase II activities as described below. For the ams-J mutation, inability to grow at 44°C was determined.
Plasmid pDK39 (Cmr rnb-500) has been described previously (12). Probes for the RNA-DNA hybridizations were derived from the following plasmids: trxA, a 470-base-pair HincIl fragment from an M13mpll derivative (36); ssb, a  (20). Growth conditions. Luria broth (L broth) and K medium were as described by Dykstra et al. (13). Thymine (50 ,ug/ml) was added to the growth medium of strains carrying the thyA715 allele. Media were supplemented with appropriate antibiotics as needed. Antibiotics were obtained from Sigma Chemical Co. (St. Louis, Mo.), and final concentrations were 20 pLg/ml for chloramphenicol and tetracycline and 50 ,ug/ml for kanamycin.
Strains were grown at 30°C in a gyratory water bath. They were shifted to 44°C at 40 Klett units (1 x 108/ml) (green filter no. 54 for L broth and no. 42 for K medium). When cell cultures reached a Klett reading of 100 units, they were diluted 1:10 into fresh prewarmed medium. Generation times were determined in L broth at 30°C. Cell cultures were grown to 40 Klett units and diluted back to 15 Klett units, and then readings were taken every 25 min. To determine the viability of the strains, samples were removed periodically during the growth experiments, diluted in M56/2 buffer (21), plated on L agar plates, and incubated at 30°C for 24 to 36 h. Enzyme assays. PNPase was assayed in cell lysates by the Pir±ADP exchange reaction described by Grunberg-Manago (15) and modified by Reiner (30,31). RNase II activity was assayed qualitatively in vitro as described by Donovan and Kushner (10). Lysates were first incubated for 10 min at 44°C to inactivate the thermolabile RNase II protein (encoded by rnb-500) and then assayed for activity at 30°C. Determination of chemical half-lives. The chemical decay of pulse-labeled RNA was assayed essentially as described by Donovan and Kushner (12). Thus, in these experiments, mRNA decay was measured as the degradation of RNA to oligonucleotides too short to be precipitated with trichloroacetic acid. Exponentially growing cells were pulse-labeled for 1 min with 3 ,uCi (90 nmol) of [5, 6-3H]uridine (Dupont, NEN Research Products, Boston, Mass.) per ml. The labeling was stopped by the addition of 500 pLg of rifampin (Sigma) per ml, 20 ,ug of nalidixic acid (Sigma) per ml, and 200 pug of uridine per ml.
To determine the decay of specific messages, we added the rifampin and nalidixic acid to stop initiation of new RNA synthesis just prior to shifting to 44°C. Samples of 7 ml were removed at the indicated times and harvested as described by Krzyzek and Rogers (19). Total cellular RNA was extracted by the procedure of Williams and Rogers (39), treated with DNase I (RNase-free; Boehringer Mannheim Biochemicals, Indianapolis, Ind.), and hybridized with specific probes (see RNA-DNA hybridization). Only curves having a least-squares measure of fit of -0.9 were used in half-life determinations. Errors were estimated from the standard deviation of the slope.
RNA-DNA hybridization. The purified RNA samples were analyzed by using dot blots, slot blots, and agarose or polyacrylamide gel electrophoresis. The dot-blot and slotblot techniques were a modification of the method of White and Bancroft (37). To an appropriate volume of RNA, 0.4 volume of formaldehyde (37%; J. T. Baker Chemical Co., Phillipsburg, N.J.) and 0.6 volume of 20x SSC (lx SSC is 0.15 M NaCl plus 15 mM sodium citrate) were added. The solution was placed in a 65°C water bath for 15 min and subsequently transferred to ice. The denatured RNA was diluted into 15 x SSC to the desired concentrations and spotted onto nitrocellulose paper (Schleicher & Schuell, Inc., Keene, N.H.). RNA blots were hybridized with radioactively labeled DNA fragments by the method of Thomas (34) and then autoradiographed at -70°C. All DNA fragments were radioactively labeled with [32P]dATP (Dupont, NEN Research Products) as described by Feinberg and Vogelstein (14). Quantitative determination of hybridization was done by two methods: (i) direct counting in a scintillation counter of the dots cut from the nitrocellulose paper; (ii) densitometric analysis of the autoradiograms. The plateau level of stable RNA was subtracted from all the time points. The percentage of specific mRNA hybridized was calculated from the radioactivity (counts per minute) at a given time divided by the maximum radioactivity in the unstable RNA fraction. The half-life of mRNA was estimated from a plot of the percentage of specific mRNA hybridized versus time after the addition of rifampin.
In the Northern analyses, total mRNA and molecular size standards were denatured in deionized formamide containing 0.3% xylene cyanol, 0.3% bromophenol blue, and 0.37% disodium EDTA. The RNA species were then separated in either 5 or 6% polyacrylamide gels containing 7 M urea unless noted otherwise. Molecular size standards were derived by AluI digestion of pBR322 plasmid DNA and end labeled with polynucleotide kinase (Boehringer Mannheim) or a 0.16to 1.77-kb RNA ladder obtained from Bethesda Research Laboratories, Inc. (Gaithersburg, Md.). After electrophoresis, the RNA was electroblotted onto nylon membranes (Biotrans; ICN Pharmaceuticals Inc., Irvine, Calif.) by a procedure developed by T. Fitzwater and B. Polisky (personal communication). Hybridizations were done as described by Thomas (34). Autoradiography with Kodak XAR-5 X-ray film was performed for various lengths of time at -70°C.

RESULTS
Construction of multiple mutants. A series of isogenic strains containing various combinations of the ams-J, pnp-7, and rnb-500 alleles were constructed in both the C600 and MG1693 genetic backgrounds (Table 1). Of particular interest was the fact that the pnp-7 rnb-SOO double mutant could only be constructed if the rnb-500 mutation was also present on a multicopy plasmid (pDK39). In contrast, the ams-J pnp-7 rnb-SOO multiple mutant was viable in the absence of pDK39.
Growth properties of multiple mutants. Growth experiments were performed in either L broth or K medium. In L broth, both SK5726 [pnp-7 rnb-SOO, pDK39 (rnb-SOO)] and SK5725 [ams-J pnp-7 rnb-SOO, pDK39 (rnb-SOO)] ceased growing within 60 to 85 min after the shift to the nonpermissive temperature (Fig. 1A). The profile of SK5704 (ams-J pnp-7 rnb-500) was interesting in that the growth rate dropped immediately after the shift to 44°C, in contrast to SK5725 [ams-J pnp-7 rnb-SOO, pDK39 (rnb-SOO)]. SK5665 (ams-J) and SK5715 (ams-J rnb-SOO) continued to grow in L broth for several generations, but their growth rates decreased progressively in comparison with that of the wildtype control, MG1693. In K medium (Fig. iB), SK5725 [ams-I pnp-7 rnb-500(pDK39) (rnb-500)] and SK5726 [pnp-7 rnb-500, pDK39 (rnb-500)] continued growing after the temperature shift, in contrast with the L broth results. Similarly, SK5715 (ams-l rnb-500) and SK5704 (ams-1 pnp-7 rnb-500) continued growing slowly in K medium at 44°C. The other strains did not show any significant difference in the K medium relative to their growth in L broth. Interestingly, SK5715 (ams-1 rnb-500) grew more slowly in L broth than SK5665 (ams-I) but matched its growth in K medium.
The viability of single and double mutant strains deficient in PNPase or RNase II or both at 44°C has previously been reported (12). When the viability of SK5704 (ams-1 pnp-7 rnb-500) was determined after a shift to the nonpermissive temperature, a 10-fold drop in viability was observed (Fig.  2). It took only 28 min at 44°C to reduce the number of CFU at 30°C by 50%.
Chemical half-life of total pulse-labeled RNA. The chemical half-life of total pulse-labeled RNA was measured at 44°C in strains carrying various combinations of the ams-1, pnp-7, and rnb-500 alleles ( Table 2). Significant differences were noted between the wild type (3.4 to 3.6 min) and the ams-1 pnp-7 rnb-500 triple mutant (11.4 to 15.8 min). The presence of the ams-1 allele had the most striking effect on the measured mRNA half-life (Table 2). There was no obvious correlation between generation time at 30°C and the halflives obtained after the shift to 44°C (Table 2). Of interest was that the ams-J single mutant had a slower rate of decay (9 min) than either the ams-J pnp-7 (7.3 min) and or the ams-l rnb-500 (7.5 min) double mutant. In the same genetic background, the pnp-7 rnb-500 double mutant had a half-life of 4.8 min. The results were comparable in both the C600 and MG1693 genetic backgrounds. In addition, the presence of pDK39 (rnb-S00) in ams-J pnp-7 rnb-500 genetic backgrounds led to an increase in chemical half-life (SK5704 [ams-I pnp-7 rnb-500], 8.9 min; SK5725 [ams-1 pnp-7 rnb-500, pDK39 (rnb-500)], 11.4 min) (  nalidixic acid were added as described in Materials and Methods and the cultures were immediately shifted to 44TC. The chemical decay of specific mRNAs was analyzed by hybridization of the isolated RNAs with DNA fragments containing the coding sequences of the genes indicated (Materials and Methods).
The decay of trxA mRNA at 44°C was significantly slower in the triple mutant SK5704 (Fig. 3). Analysis of these data yielded a chemical half-life for trxA mRNA of 6.8 min in SK5704 (ams-J pnp-7 rnb-SOO) versus 2.6 min in the wildtype control ( Table 3). The half-lives observed in ams-J and pnp-7 rnb-S5O control strains were shorter than that of the triple mutant (Table 3). At 30°C, the chemical half-lives were similar: 5 min for SK5704 (ams-J pnp-7 rnb-500) and 4.5 min for MG1693 (wild type). When the chemical half-lives were determined for the ssb, uvrD, pnp, nusA, and cat genes, the results were consistent: the transcripts were always degraded two-to fourfold more slowly in the triple mutant (Table 3). Northern analysis of trxA mRNA. Northern analysis employing RNA denaturation with formaldehyde and agarose gel electrophoresis produced a single broad band at zero time in the range of 450 to 500 nucleotides (nt) (Fig. 4A). A second, smaller species appeared in the triple mutant approximately 4 min after the shift (Fig. 4A). Although this result suggested some stabilization of the mRNA breakdown products upon a shift to 44°C, it was clear that agarose gel electrophoresis would not provide adequate resolution. Accordingly, a different method based on DNA sequencing technology was used. RNA samples were denatured in 99% formamide and electrophoresed in 6% polyacrylamide gels containing .7 M urea.
The results of a typical Northern analysis by this technique are shown in Fig. 4B. At the time of shift to 44°C, four major mRNA species ranging in size from 493 to 369 nt were observed. In the wild-type strain (MG1693), only the 387-nt RNA molecule was still present after 7 min at the nonpermissive temperature (Fig. 4B). In the ams-J pnp-7 rnb-500 triple mutant (SK5704), the same four species were observed at the time of shift. However, by 4 min after the shift, a series of discrete breakdown products began to appear (Fig.   4B). Further processing occurred at longer times at 44°C with RNA products as small as 185 nt appearing (Fig. 4B). By 60 min after the shift, all the RNA species had been degraded in both strains.
Since the chemical half-life of total pulse-labeled RNA also increased in the ams-) single mutant and the pnp-7 rnb-SOO double mutant (Table 2), the degradation of trxA mRNA in strain SK5665 (ams-1) and strain SK5726 (pnp-7 rnb-500) was also examined. While the kinetics of trxA mRNA decay changed in SK5665 (ams-J) compared with that of the wild-type control, no discrete breakdown products appeared (Fig. SA). The pattern in the pnp-7 rnb-500 double mutant (SK5726) initially also appeared identical to that of the wild-type control (Fig. SB). However, upon significantly longer exposure, it became apparent that breakdown products similar to those observed in the triple mutant were present (Fig. SC). Longer exposure of the autoradiograms of the RNA isolated from the ams-J strain did not show this pattern (data not shown). No hybridization was observed when total RNA was isolated from a trxA deletion mutant (data not shown).
Northern analysis of cat mRNA. To determine the generality of the above-described observations, we tested addi-  tional mRNAs in the ams-l pnp-7 rnb-500 genetic background. For the cat mRNA, four discrete large transcripts of 1,020 to 1,200 nt were observed after electrophoresis in a 5% acrylamide gel (Fig. 6). By 10 min after the shift to 44°C, all these transcripts had been degraded in the wild-type strain without any evidence of smaller discrete breakdown products. In contrast, in the triple mutant, the full-length transcripts appeared to be sequentially processed into a series of smaller mRNA species, some of which were still visible 30 min after the shift to 44°C. The degradation pattern of the cat mRNA was clearly distinct from that observed with trxA (Fig. 4). Similar results have been obtained with additional mRNAs (data not shown).

DISCUSSION
The results described above demonstrate that RNase II, PNPase, and the Ams protein are required for the terminal steps of mRNA breakdown. In their absence, the appearance of discrete breakdown products of the trxA and cat mRNAs can be visualized on Northern blots (Fig. 4 to 6). It is particularly important to note that these breakdown products were only significantly stabilized in the triple mutant. While the absence of the Ams protein alone increased the chemical half-life of the trxA mRNA (Table 3), the degradation pattern observed on Northern blots closely resembled that of the wild-type control (Fig. SA). In contrast, in the PNPase-RNase 11 double mutant, products similar to those seen in the triple mutant could be seen after very long exposure of the autoradiograms (Fig. SC), even though the chemical half-life of the trxA mRNA was closer to that observed in the wild-type control strain.
Several important questions regarding the nature of the breakdown products need to be answered. Perhaps most important are the oligonucleotides observed in the triple mutant normal mRNA breakdown products that are being stabilized by the absence of the Ams protein, PNPase and RNase II? Analysis of a wild-type strain overproducing thioredoxin mRNA showed the transient appearance of fragments identical in size to those observed at longer times in the triple mutant (data not shown). This result implies that the patterns shown in Fig. 4 are an accurate representation of trxA mRNA degradation.
A second important issue relates to how the initial breakdown products are generated. Although it remains to be proven, the simplest explanation would be endonucleolytic cleavage. Recently, Portier et al. (28) have suggested that RNase III plays a role in the functional inactivation of the pnp mRNA by cleaving the transcript 5' to the coding sequence. It has been previously shown that RNase IIIprocessing sites exist in some intercistronic regions (3,29). To determine whether RNase III is an important enzyme in mRNA turnover, we are currently constructing ams pnp rnb rnc multiple mutants. .v. .w : . .,!F , .iZ w : : . . : : .

FIG. 5.
Northern analysis of trxA mRNA in ams-J or pnp-7 rnb-500 genetic backgrounds. Strains were grown at 30°C in L broth. At 40 Klett units, cell cultures were shifted to 44°C with the concomitant addition of rifampin. Cell culture samples were removed at the indicated times, and total RNA was extracted from SK5665 (ams-J), SK5626 (pnp-7 rnb-500), or SK5704 (ams-J pnp-7 rnb-500) as described in Materials and Methods and electrophoresed through a 6% polyacrylamide-7 M urea gel. In panel A, the first four lanes were from SK5665 (ams-J) and the last four lanes were from SK5704 (ams-J pnp-7 rnb-500). In panel B, the first four lanes were from SK5626 [pnp-7 rnb-500, pDK39 (rnb-500)] and the last four lanes were from SK5704 (ams-1 pnp-7 rnb-500). Panel C is a longer exposure of panel B. Numbers above each lane indicate the time (minutes) after the temperature shift at which the RNA was extracted. The sizes (nucleotides) of the two largest transcripts are indicated.
Other RNases are known to exist in the cytoplasm, but most of them have not been characterized in relation to mRNA turnover (9). It is possible that the discrete RNA fragments observed in Fig. 4 to 6 could have arisen from an exonucleolytic activity which is inhibited by specific secondary structures. Such secondary structures have been demonstrated to influence the rate at which nucleases degrade o given species of mRNA (7,23,24,32).
Determination of the nucleotide sequences where cleavages occur will help answer the question of whether unique sequences or special secondary structures are recognized. The work of Cannistraro et al. (5) suggested that the lacz mRNA was preferentially cleaved at UUAU sequences.
There are three of these sequences in the trxA mRNA. Two of them occur in the untranslated 5' region and the third occurs between amino acids 5 and 6. It is thus not possible for all the fragments to have arisen from cleavage only at these sites.
Taken together, these data indicate a complex mechanism for the breakdown of mRNA molecules. We propose that full-length transcripts are initially processed by endonucleolytic cleavages, with each mRNA having a unique series of sites. Complete degradation of the initially cleaved transcripts occurs through possible additional endonucleolytic steps, followed by exonucleolytic degradation by RNase II, PNPase, and possibly the Ams gene product.
The results presented above indicate a complex role for the Ams protein in mRNA decay. While the absence of the protein alone leads to increased chemical half-lives of mRNAs (Table 2), discrete breakdown products were not visualized (Fig. 5). Furthermore, while the viability of ams-J rnb-SOO (Fig. 1) and ams-J pnp-7 (data not shown) mutants was reduced at 44°C compared with that of the ams-J single mutant, the chemical half-lives measured in these strains rnb-500 triple mutant. Strains SK6605 (wild type containing pDDK1) and SK6606 (ams-J pnp-7 rnb-500 containing pDDK1) were grown and treated as described in the legend to Fig. 4. At the times indicated, total RNA was extracted from the cells as described in Materials and Methods and electrophoresed through a 5% polyacrylamide-7 M urea gel. The same amount of RNA (4 p.g) was loaded in each lane. After electrophoresis, the RNA was transferred by electroblotting to Biotrans and the filters were hybridized with a 1.2-kb DNA fragment containing the cat gene. The first six lanes show RNA extracted from SK6606 (ams-J pnp-7 rnb-500), and the last six lanes show RNA extracted from SK6605 (wild type). The sizes (nucleotides) of representative bands are indicated. (Table 2) were actually shorter. In addition, even though the mRNA breakdown products are stabilized in the triple mutant, by 60 min after the shift all the RNA was degraded (Fig. 4). Taken together, it is not clear whether the Ams protein acts directly as an RNase or indirectly as a regulatory protein. Although Chanda et al. (6) apparently cloned the ams structural gene, its published nucleotide sequence is in fact identical to a segment of the groEL gene (17). Preliminary results in our laboratory suggest that the Ams protein regulates its own synthesis (M. Torres-Diaz, F. Claverie-Martin, and S. R. Kushner, manuscript in preparation).
In conclusion, it is worth pointing out that many unresolved questions remain regarding the mechanism of mRNA turnover. Some of these relate to the unusual interaction between the ams-l allele and mutations in RNase II and PNPase regarding viability, mRNA half-lives, and growth rates. In addition, even in the absence of these three gene products, mRNA is still being degraded, although more slowly, to mononucleotides. Thus other RNases still to be identified are clearly involved.