J. Mol. Biol. (1991) 217, 283-292

Decay of mRNA Encoding Ribosomal Protein S15 of Escherichia cob Is Initiated by an RNase E-dependent Endonucleolytic Cleavage that Removes the 3’ Stabilizing Stem and Loop Structure Philippe Rkgnier’**t and Eliane Hajnsdorf’ ‘Institut de Biologie Physic0 Ghimique 13 rue Pierre et Marie Curie, 75005 Paris, France 2Universite’ Paris (Received

7, Paris,

France

9 July 1990; accepted 11 October 1990)

The transcripts of the rpsO-pnp operon of Escherichia cd, coding for ribosomal protein 515 and polynucleotide phosphorylase, are processed at four sites in the 249 nucleotides of the intercistronic region. The initial processing step in the decay of the pnp mRNA is made by RNase III, which cuts at two sites upstream from the pnp gene. The other two cleavages are dependent on the wild-type allele of the me gene, which encodes the endonucleolytic enzyme RNase E. The cuts are made 37 nucleotides apart at the base of the stem-loop structure of the rho-independent attenuator located downstream from rps0. The cleavage downstream from the attenuator generates an rps0 mRNA nearly identical with the monocistronic attenuated transcript, while the cleavage upstream from the transcription attenuator gives rise to an rps0 message lacking the terminal 3’ hairpin structure. The rapid degradation of the processed mRNA in an me+ strain, compared to the slow degradation of the transcript that accumulates in an me- strain, suggests that RNase E initiates the decay of the rps0 message by removing the stabilizing stem-loop at the 3’ end of the RNA.

1. Introduction An important factor in the determination of the expression of a gene is the intracellular concentration of its mRNA. This level is set up by both the rate of synthesis and the rate of decay of the transcript. In bacteria, half-lives of individual mRNAs vary in a 40-fold range from 30 seconds to 20 minutes. Such variation can have a profound influence on the rate of protein synthesis (Nilsson et al., 1984; Pedersen et al., 1978). It is generally believed that the metabolic instability of mRNA (Gros et al., 1961) permits the rapid adaptation of the cell to environmental changes. While the mechanisms that control transcription and translation have been analysed extensively, the steps involved in mRNA decay and their regulation are still poorly understood. The secondary structures of RNAs are an important parameter in their sensitivity to ribonucleaaes. For example, stem-loops such as a rho-independent t Author addressed.

to whom all correspondence

should be

terminator at the 3’ end of RNA or repetitive extragenie palindromes (REP) protect the upstream RNA from the attack of 3’ to 5’ exonucleases (Chen et aZ., 1988; Gross & Hollatz, 1988; Hayashi & Hayashi, 1985; Klug et al., 1987; Mott et al., 1985; Newbury et al., 1987; Panayotatos &, Truong, 1985; Plamann & Stauffer, 1990; Schmeissner et al., 1984; Wong & Chang, 1986), probably polynucleotide phosphorylase and RNase II (Donavan & Kushner, 1983; Higgins et al., 1988; Mott et al., 1985; Plamann & Stauffer, 1990). On the other hand, secondary structures in the 5’ leader, or at the 3’ end of several RNAs are the targets recognized by several endonucleases. Two endonucleases, RNase III and RNase E, have been implicated in the endonucleolytic processing of mRNA. A cleavage by RNase III in regions that can fold into secondary structures is the limiting step in the decay of the transcripts of the rpsO-pnp (Portier et al., 1987), rnc-era (Bardwell et al., 1989) and met Y-nusA-infB (Regnier & Grunberg-Manago, 1989) operons of Escherichia coli and of the int gene of phage lambda (Schmeissner et al., 1984). Thus, RNase III negatively controls the expression of pnp, rnc, era and int. In addition to 283

0022-2836/91/020283-10

$03.00/O

0

1991 Academic

Preess Limited

284

P. Rkgnier

and E. Hajnsdorf

triggering mRNA decay, RNase III participates in the processing of both rRNA and mRNA (Bram et al., 1980; Gegenheimer & Apirion, 1980; Barry et al., 1980; Daniels et al., 1988; Downing & Dennis, 1987; Downing et al., 1990; Hyman & Honigman, 1986; Steege et al., 1987). Such processing regulates positively the expression of gene 1.2 of phage T7 (Saito & Richardson, 1981) and ~111 of phage lambda (Altuvia et al., 1987) and negatively that of gene ~11 of phage lambda (Krinke & Wulff, 1987). RNase III has been implicated in the processing of the transcripts of genes controlled by antisense RNAs (Krinke & Wulff, 1987; Simons & Kleckner, 1988). RNase E also cleaves RNA in the vicinity of secondary structures (Mudd et al., 1988; Tomcsanyi & Apirion, 1985). It has been shown to be involved in the trimming of the d&F RNA of E. coli (Faubladier et al., 1990) and in the processing of the transcripts encoding proteins 33, 59 and 32 of phage T4. While the processing of the mRNA of gene 32 by RNase E stabilizes the gene 32 message, it appears to accelerate the decay of the upstream mRNA (Belin et al., 1987; Carpousis et al., 1989; Mudd et al., 1988). Moreover, many transcripts of phage T4 are functionally and chemically stabilized in an RNase E deficient strain (Mudd et al., 1990). RNase E participates in the maturation of 5 S ribosomal RNA and cuts the RNA 1 of plasmid ColEl (Misra & Apirion, 1979; Tomcsanyi & Apirion, 1985). Other endonucleolytic cleavages occurring in the 5’ leader, and in the ribosome-binding site also have an influence on mRNA stability and gene expression (Melefors & von Gabain, 1988; Uzan et aE., 1988). However, the enzymes responsible for these events remain to be characterized (Nilsson et al., 1988; Ruckman et al., 1989). The rpsO-pnp operon of E. eoli, encoding ribosomal protein S15 and polynucleotide phosphorylase, has been a useful model system for the study of RNA processing and decay. Endonucleolytic processing by RNase III in the intercistronic region is the first step in the degradation of the pnp mRNA (Portier et al., 1987). Here, we demonstrate that extracistronic RNase E dependent maturation is critical in the degradation of the rps0 message.

2. Materials and Methods (a) Strains

and probes

The double mutant harbouring the me3071 and rnclO5 alleles, and the isogenic rnc105 single mutant, have been constructed by Pl transduction of the mc105 allele of BL321 with the n&B51 ::TnlO TetR marker into the N3431 and N3433 strains, respectively (Table 1). The rcn105 cotransductants were selected as small colonies on tetracycline selective medium. The rnc- phenotype was verified by looking at the overproduction of the polySDS/ u-subunit on phosphorylase nucleotide polyacrylamide gels, and at the accumulation of uncut primary transcripts of the rpsO-pnp and met Y-nusA-in.B operons (Regnier & Grunberg-Manago, 1989; Regnier & Portier, 1986). The DNA fragments were prepared as described (Regnier & Portier, 1986) and labelled, at the 5’ end, for

Table 1 Escherichia

coli strains used in this work

Strain

Genotype and relevant markers

Source/origin

BL322

thi-1, argH1, gal-6, galY1, mtl-2, ~$7, malAl, ara-13, str-9, tonA2, SupE44 BL322 ml05 HfrH, A(ptg-lac), qmT1, thi-1, nadB51:: TnlO BL321, m-105, nadB51:: TnlO. Transductant of BL321 (Pl vir grown on NK6042) F-, thi-1, argG6, argE3, his-4, mtl-1, ~~1-5, tsx-29, rpsL,

W. Studier

BL321

NK6402 IBPC496 IBPC5321

IBPC490

N3433 N3431 IBPC637

AlacX74 IBPC5321, ~105,

W. Studier J. P. Bouche J. Plumbridge

Plumbridge et al. (1985) J. Plumbridge

n&B51 :: TnlO. Transductant of IBPC5321 (Pl vir grown on IBPC496) HfrH, ZacZ43, Iz-, relA1, spoT1, thi- 1 N3433. me307 1(Ts) N3431, mc105, nadB51 ::TnlO

D. Apirion D. Apirion This work

Transductant of N3431 (Pl vir grown on IBPC490) IBPC633

N3433, ~105,

m&B51 :: TnlO

This work

Transductant of N3433 (Pl vir grown on IBPC490)

nuclease Northern

S, mapping blotting.

and

(b) Preparation

by

random

priming

for

of RNA

RNA was extracted from BL321 and BL322 cells grown on LB medium by the hot phenol technique (R&gnier & Portier, 1986). When time-course experiments were performed, cells were grown at 37°C (BL321 and BL322) or at 30°C (N3431, N3433, IBPC633 and IBPC637) in Mops-tricine medium supplemented with 1 pg thiamine/ ml, 64% (w/v) glucose, 2 mM-potassium phosphate, 05% (v/v) Casamino acids. Exponentially growing cultures were shifted at 43°C or supplemented with rifampicin (500 pg/ml); lo-ml portions were withdrawn at different times and were treated as described (Bardwell et aZ., 1989). (c) S, nuclease mapping

RNA was mixed with 5’-end labelled probe in hybridization buffer, incubated overnight and digested with 8, nuclease as described (Regnier & Portier, 1986). The DNA probe was added in a large excess to the hybridization mixture so that the amount of protected fragment was proportional to the amount of mRNA added. Relative amounts of RNA were estimated from densitometric analysis of autoradiographs with an LKB Ultroscan XL. (d) Northern

blotting

Total RNA prepared as described above was fractionated on 1.5 o/o (w/v) agarose/formaldehyde gels and transferred to Amersham Hybond N membrane for hybridization with 32P-labelled probe as described (Regnier t Grunberg-Manago, 1989). Relative amounts of RNA were estimated as for S, nuclease mapping.

RNase

E Initiates

mRNA

Decay

(e) Primer extension The method used is essentially that described by Gauss et al. (1987). The oligonucleotide, B’-GCCGCGCGAACCTCTGCAACGG-3’, complementary to the region of the RNA indicated in Fig. 1 was labelled at its 5’ end by polynucleotide kinase and [Y-~‘P]ATP (3000 Ci/mmol), purified on a denaturing gel and mixed in molar excess with 15 pg of total RNA prepared as described above in (pH &3), 10 m&r-Mgcl,, 5 Pl of 50 mM-Tris.HCl 80 mhr-ECl. After 4 min at 6O”C, the mixture was cooled on a solid CO,/ethanol bath, thawed on ice, adjusted to 05 mM each dNTP. 4 mw-dithiothreitol in 10 ~1 of the above buffer and incubated at 42°C for 30 min with 2 units of avian myeloblastosis virus (AMV) reverse transcriptase (Genofit). Then nucleic acids were precipitated with ethanol, dried and analysed on a 50% urea/8% polyacrylamide gel. The reaction was modified as follows to generate a sequence ladder: 32 pg of total RNA of strain BL322 was denatured and cooled as above before addition of the 4 dNY!Ps and dithiothreitol in the buffer described above. The mixture was then divided into 4 tubes, which were each adjusted to 05 mM of 1 of the 4 ddNTPs. The final mixtures contain 1 mM of the 4 dNTPs and 4 mllr-dithiothreitol in 5 ~1 of the buffer described above. After incubation with 4 units of AMV reverse transcriptase for 30 min at 42”C, the nucleic acids were precipitated and analysed as described above.

3. Results (a) Zdent@cation of endonucleolytic cleavage sites in the intercistronic region of the rpsO-pnp operon The transcripts of this operon can be either dicistronic or monocistronic. Transcription starting from the promoter Pl can either read through the rhoindependent attenuator tl downstream from rps0 or terminate at this site. Transcription initiation can occur also at the secondary promoter P2 lying between the two genes. In addition, the primary transcripts are processed by endonucleases at several sites in the intercistronic region (Fig. 1; and see Regnier & Portier, 1986). One of these cleavages made by RNase III gives rise to the predominant pnp RNA species identified by S, nuclease mapping in the wild-type strain (Fig. 2; and see Regnier & 1986). As indicated by previous data Portier, (Portier et al., 1987), the 5’ end of this mRNA decays much more rapidly than that of the unprocessed RNA (Fig. 2(a); and see Takata et al., 1989). Its half-life is estimated to be about 45 to 60 seconds. A similar experiment with RNA from an RNase III deficient strain (mc105) shows four RNA 5’ ends (Fig. 2(a)) among which two can be attributed to the initiation of transcription at the promoters Pl and P2, and one named M to an endonucleolytic processing independent of RNase III, occurring two nucleotides downstream from the main site of termination of transcription mediated by the rho-independent attenuator of rps0 (Fig. 1; and see Regnier & Portier, 1986). Precise mapping with reverse transcriptase shows that the fourth 5’ extremity corresponding to the band M2 of Figure 2 maps 37 nucleotides upstream from M at the base of the stem-loop of the rps0 attenuator (Figs 5(b) and

345 390

nt P2) nt

‘8%

:f I

0.5 hb

.P,irnW WUOSeE ~2.8kb 0421aw

f

d2.3hb

Figure 1. Transcription and maturation of the rpsOpnp operon. The results of the S, mapping experiments of Figs 2 and 4 are summarized under the map of the E. coli chromosome showing rps0 and pnp. The promoters (Pl and P2), and RNase E maturation sites (M and M2), the transcription termination sites (tl and t2) and the RNase III recognition sites (RIII) are located on this map (Regnier et al., 1987; Regnier & Portier, 1986). The HpaI-HpaI probe used in the S, nuclease mapping experiment is shown by an open box. The protected DNA fragments are indicated by lines. Their lengths (in nucleotides) are shown together with the 5’ ends (transcription initiation or maturation sites) from which they originate. The small black square called primer locates the sequence complementary to the oligonucleotide used in the “primer” extension experiment of Fig. 5. The DraI-BgZI black box at the top of the Figure is the double-stranded DNA fragment used as probe in the Northern blot experiment for Figs 3 and 6. The wavy lines at the bottom of the Figure show the structure of the primary transcripts of the operon, whose lengths are indicated (in kb). The secondary structures of tl and t2 are shown by hairpins. The arrows indicate the sites at which processing by RNase E and RNase III occurs.

7). There is no obvious potential promoter consensus sequence that could account for initiation of transcription at this site (Regnier et al., 1987), suggesting that band M2 probably also arises from an endonucleolytic cleavage. (b) Decay of transcripts of the rpsO-pnp operon in a strain dejkient for RNase III If the 5’ ends mapped at M and M2 are generated by endonucleolytic processing of the Pl primary transcript, their intracellular concentration should result from an equilibrium between their rates of decay and their rates of formation from the precursor mRNA. To verify this prediction, the decay of the different RNA 5’ ends has been assayed by S1 mapping of RNAs at different times after the inhibition of RNA synthesis by rifampicin in an rncl05 strain. Utilization of a strain defective for RNase III allows the analysis of transcripts with 5’ ends at Pl, M2, M and P2, all of which are processed by RNase III in a wild-type strain (Fig. 2(a)). The 5’ ends corresponding to transcription initiation at Pl and P2 decrease exponentially (Fig. 2(b)) with half-lives estimated to be three and ten minutes, respectively. On the other hand, the kinetics of decay of the M and M2 species are biphasic (Fig. 2(b)). They first decrease slowly with half-lives of 15 to 20 minutes for M and M2 as estimated from

286

P. RtQnier

and E. Hajnsdorf

0

5 Time

IO otter

I

I

1

20

30

40

ritampicin

addition

(min)

(b)

Figure 2. Identification and decay of the 5’ ends of the pnp transcripts. (a) Total RNA was prepared from BL321 (~~105, RNase III deficient) and BL322 (VW+, wild-type) cultures grown at 37°C samples were harvested at different times (indicated in minutes above each lane) after rifampicin addition and submitted to S, nuclease analysis. Decay of the RNA was analysed over periods of 2 min (rnc’ and ~~105) and 40 min (~~105). The probe was the 843 base single-stranded HpaI-HpaI DNA shown in Fig. 1, labelled at its 5’ end. It was hybridized overnight at 52°C with 10 pg of total RNA and then incubated at 37°C for 45 min with Si nuclease. Protected fragments were analysed on a 50% urea/5% polyacrylamide gel. The locations of the 5’ extremities corresponding to the probe (pr), the promoters (Pl and P2) and the processing sites by RNase III (RIII) and by RNase E (M and M2) are shown beside the autoradiograph. (b) Relative amounts of the different RNA species have been estimated by scanning the autoradiograph showing the decay of pnp RNA of a rnc- strain over a period of 40 min. They are represented on a logarithmic scale as the percentage of the intracellular concentration of each RNA species at the time of rifampicin addition, (A) Pl; (0) P2; (m) M; (A) M2.

the two minute experiment of Figure 2(a). Then, when the Pl transcript drops to less than 36% of its original value they decay more rapidly with halflives of ten minutes (M) and three minutes (M2) (see the 40 min experiment of Fig. 2(a)). These kinetics (see also Takata et al., 1989) suggest that during the first ten minutes of the experiment the degradation of M and M2 RNAs is partly compensated for by their formation from the Pl mRNA precursor. Consistent with this hypothesis is the fact that the amount of Pl primary transcript lost during this period (Fig. 2) could be accounted for by its conversion to the M and M2 processed RNAs (data not shown). The similar half-lives of the Pl and M2 RNAs (about 3 min) suggest that the M2 RNA does not accumulate because it decays at about the same rate as it is produced from its precursor. This might explain why the M2 RNA is not always detected by S, mapping analysis (Regnier & Portier, 1986; Takata et al., 1989). The high level of the M RNA shown by the autoradiograph of Figure 2 is probably the consequence of its accumulation in the rncstrain because of its long half-life in this strain.

(c) The rpsO-pnp cotranscript by RNase

is processed

E

In order to verify that RNase E is implicated in the processing of the rpsO-pnp dicistronic transcript, total RNA from RNase III and RNase E deficient mutants were analysed by probing a Northern blot with rps0 DNA (Fig. 3). Two major transcripts (64 and 2.8 kbt) were identified on the autoradiograph of Figure 3. These sizes correspond well to the monocistronic and dicistronic rps0 primary transcripts initiated at Pl and terminated at the rho-independent terminators located downstream from rps0 (tl) and of pnp (t2) respectively (Fig. 1). The fact that the 2.8 kb RNA can be identified only in the double mutant deficient for RNase III and RNase E implies that the polycistronic transcript is processed by both endonucleases. The wild-type strain, as well as either of the single mutants deficient for RNase E or RNase III, have only one band of 64 kb that hybridizes to the rps0 DNA. In the rnc+ strains, it is likely that t Abbreviation

used: kb, 10’ bases or base-pairs.

RNase E Initiates

287

mRNA Decay 0

30

60

0

30

60

&Z 2.6A6-

o-50.4

-

P2 -

mc me

+ +

+

+

Figure 3. Identification of rps0 mRNA in E. coli strains defective for RNase III and RNase E. Total RNA 3 pg (lanes 1, 3, 5 and 7) and 6 pg (lanes 2, 4, 6 and 8) prepared from strains N3433 (me’ rnc+; lanes 1 and 2), N3431 (me3071 me+; lanes 3 and 4), IBPC637 (me3071 rnclO5; lanes 5 and 6) and IBPC633 (me’ r&05; lanes 7 and 8) were separated on a 1.5% denaturing agarose gel, transferred to hybond-N, and probed with the 32P-labelled DraI-BgZI DNA fragment from the rps0 gene (Fig. 1). Sizes of the transcripts (indicated in kb next to the picture) were deduced from the migration of RNA size markers obtained from Gibco-BRL. Genotypes of strains used to prepare the RNA are indicated at the bottom of the Figure.

the 05 kb rps0 RNA generated by RNase III processing of the dicistronic mRNA (Fig. 1) is degraded by 3’ to 5’ exonuclease(s) up to the hairpin of rps0 attenuator to give an RNA that has the same length as the monocistronic terminated transcript. In agreement with this hypothesis, the data of Figure 5 show that RNA extending between the attenuator t 1 and the RNase III site is less abundant in the rnc+ strain than in the me105 mutant. (The experiment shown in Fig. 5 is described below.) In the RNase III deficient strain, the dicistronic rps0 transcript may yield a 04 kb RNA either because of an RNase E cleavage in the vicinity of the attenuator tl (at M and/or M2) or as a result of a 3’ exonucleolytic trimming up to the tl stem loop of a longer RNA. The 05 kb transcript found in the double mutant (Fig. 3, lanes 5 and 6) might result from 3’ to 5’ exonucleolytic degradation of the rpsO-pnp cotranscript up to the stem: loop in the RNase III site or from a cleavage by an unidentified endonuclease. (d) The processing at M and M2 depends on RNase E activity To confirm cistronic RNA

that RNase E cleaves the polyat sites M and M2 in vivo, the 5’

-me 3071 I-“=-

me+ mc -

mc +

Figure 4. Identification of the RNase E processing sites in the rpsO-pnp cotranscript. Total RNA from strains IBPC637 (me3071 mc105) and IBPC633 (me' rncl05) prepared at different times (indicated at the top of each lane) after a temperature shift of 43°C and of strain BL322 (me+ rnc’) grown at 37°C have been analysed by S, nuclease mapping as in Fig. 2. The promoters (Pl, P2) and the maturation sites (M, M2) corresponding to the different 5’ extremities are indicated beside the picture. A, B and C locate the 3 extremities mentioned in the text mapping in the S15 coding sequence that accumulates in the IBPC637 strain and (pr) shows the location of the probe. The phenotypes of the strain used in this experiment are indicated at the bottom of the Figure.

extremities of the pnp transcripts were compared in the temperature-sensitive me3071 mutant and the rne + isogenic strain after a temperature shift to 43°C. RNAs of the double mutant rne3071rnc105 and of the me+mcl05 control strain, have been analysed by S1 nuclease mapping (Fig. 4) as described above (Fig. 2). In the me3071 thermosensitive strain, the RNA 5’ end corresponding to M2 completely disappears and the end corresponding to M dramatically decreases after the temperature shift. Comparatively, the me+ control strain displays an increase in the relative amounts of both of these RNA 5’ ends at 43°C. The relative amount of the M RNA, which is about 15% of total

288

P. RAgnier and E. Hajnsdorf

-

M2

-

tl

--M

(b)

RNA in either strain at 3O”C, drops to 4 y. in the me3071 strain 30 minutes after the shift, while it rises to 23.5% in the qne+ strain. The long life-time of this RNA species probably explains its presence in the me3071 cells even one hour after the temperature shift. The increased amount of the PI rpsO-pnp RNA observed in the me3071 strain under conditions that inhibit the RNase E dependent cleavages suggests that the M and M2 RNA transcripts arise from it. Primer extension experiments performed with RNA extracted from the me3071 and me+ isogenic strains also indicate that cleavages at M and M2 are dependent on the activity of RNase E and that the M mRNA species is more abundant than the M2 RNA (Fig. 5(a)). Moreover, comparison of the relative intensities of bands given by RNA isolated from the isogenic rnc(Fig. 5(a), lanes 1 and 3) and rnc+ (Fig. 5(a), lanes 2 and 4) strains shows that the RNA hybridizing with the reverse transcriptase primer (Fig. 1) is much less abundant in the wild-type strain than in the isogenic deficient strain. In the mc+ strains, the region of the RNA complementary to the primer lying between the rps0 terminator and the RNase III site (Fig. 1) is probably the target of processive 3’ to 5’ exonucleases that initiate degradation at the 3’ end generated by RNase III. (e) The rps0 messenger is stabilized RNase E dejkient strain

I2

34 (a)

Figure 5. Precise mapping of the RNase E processing sites in the rpsO-pnp transcripts. (a) Equal amounts (15 pg) of the RNAs extracted from strains IBPC633 (me’ mcZ05; lane l), N3431 (rne3071 mc’; lane 2) IBPC637 (me3071 ~7~105; lane 3) and N3433 (mze’ rnc+; lane 4) harvested 30 min after a shift at 43°C were used as a matrix for reverse transcriptase primed with the oligonucleotide located on the map in Fig. 1. The genotype of each strain is indicated at the top of the lanes. (b) Similar experiments with RNA of the strains IBPC633 (lane me’) and IBPC637 (lane rne3071) grown for 30 min at 43°C have been run on a 50% urea/8% polyacrylamide gel in parallel with a sequencing ladder prepared by extension of the same oligonucleotide in the presence of the dideoxynucleotides indicated (A, C, G and T) at the top of the lanes. Because the sequence ladders obtained were hard to read, this experiment has been repeated several times in order to determine precisely the location of the M2 site. The identified signals corresponding to stops of reverse transcription are indicated next to the picture. Stop sites of reverse transcriptase elongation that are not identified by S, mapping of Figs 2 and 4 probably do not correspond to RNA 5’ ends. For example, the stronger of these signals maps in the distal sequence base-paired in the stem of the hairpin structure of the rps0 attenuator tl, which probably causes termination of reverge transcription. The intensities of radioactive signals are proportional to the amount of mRNA hybridizing with the primer. In these experiments, it measures the level of RNA lying between the rps0 attenuator tl and the RNase III site (see the text and Fig. 1).

in the

In addition

to the effects on the processing of the the absence of RNase E has a striking effect on the intracellular concentration of the rps0 monocistronic mRNA. Densitometric analysis of the Northern blot of Figure 3 and of other similar experiments shows that there is approximately 23 times more monocistronic rps0 messenger in the me3071 strain than in the isogenic wild-type strain. To determine whether the increased concentration of the rps0 mRNA in the me- mutant strain is a consequence of a decrease in degradation rate, decay rates of this transcript were measured in Northern blots of RNAs from the me3072 and wildtype cells harvested at different times after inhibition of transcription initiation by rifampicin. The half-life of the rps0 mRNA, which is 50 seconds in the wild-type cells, increases to about 20 minutes in the mutant (Fig. 6). This increase of half-life in the mutant is sufficient to explain the increase of the rps0 mRNA. Previous work has demonstrated that this increase is specific and not the consequence of a generalized increase of mRNA stability in mebacteria (Apirion & Gitelman, 1980). rpsO-pnp

cotranscript,

4. Discussion (a) RNase

E processing

of rps0 transcripts

The data in this paper demonstrate that the cotranscript undergoes two endonucleo-

rpsO-pnp

RNase E Initiates

289

mRNA Decay A A A GC GU UA C A AU CG U TU UA AU AU CG

t1 A G G u u CG CG GC GC GC GC GU AU

E

RNose

VW

AURNaseE l(M)

3 1’ “AI

GU :;

GUUUCAG-CGG -UAAUUCUUGCGA -

CUAU

(a)

t1

4 I 2 Time after rifomplcin

e addition (min)

(b)

Figure 6. Decay of the rps0 mRNA in the me+ and me3071 strains. (a) Cultures of strains N3433 (me’) and N3431 (me3071) grown at 30°C were transferred to 43°C for 20 min before addition of rifampicin (500 pg/ml). Then total RNA was prepared from portions harvested at different times (indicated in minutes at the top of each lane). Time zero corresponds to the addition of the antibiotic: 14 pg of me+ RNAs (see the 6 left lanes) and 0% pg of me3071 RNAs (see the 6 right lanes) were fractionated on a 1.5% denaturing agarose gel, transferred and probed as described in Fig. 3. Genotypes of the strains are indicated beneath the picture of the autoradiograph. (b) The relative amounts of RNA obtained by scanning the autoradiograph were expressed as a percentage of the amount of RNA in each strain at the same time of rifampicin addition and plotted as a function of time. (0)

me+;

(m) rne3072.

lytic cleavages that are dependent on the activity of RNase F,. These cuts are made 37 nucleotides apart in the intercistronic region, on each side of the hairpin of the rho-independent attenuator of ripsO. All RNase E cleavages mapped thus far occur at sites that have some sequence similarity and are located upstream from potential secondary structures (Mudd et al., 1988; Tomcsanyi & Apirion, 1985). In the rps0 transcript, the location of the M2 site suggests that the stem-loop of the rho-independent terminator might be the structural motif recognized by RNase E (Fig. 7). The sequence downstream from M is predicted by computer analysis (Jacobson et al., 1984) to form a secondary

A G G u u CG CG GC GC GC GC GU AU AU AU AU AGu C % A

M2 +”

= AtM G

GC GCGA UGAC

GU CG GC GC UA CGA C z G C= UA WA

WA* CAGA UAAAA P

(b) of the RNase E cleavage sites; homology between the 5 S rRNA precursor and the rpsOpnp cotranscript. (a) The RNase E cleavage sites have been located on the sequence of the rpsO-pnp cotranscript folded as predicted by computer analysis of the RNA upstream from the RNase III processing site (Jacobson et al., 1964). Portions of the RNA appearing as singlestranded are predicted to anneal with upstream regions of the rps0 messenger that are not shown in the Figure. (b) The 2 complementary sequences of the rpsO-pnp cotranscript, located upstream and downstream from the attenuator tl, in which the RNase E cleavages are made are base-paired as previously suggested for bhe 5 S rRNA precursor (Roy et al., 1983), which is shown beside. Identical nucleotides at homologous locations in both structures are indicated by bold characters. Arrows indicate the target of RNase E (M and M2). The regions suggested to be the RNase E recognition sequences (Tomcsanyi & Apirion, 1985) are framed and the UAA terminator codon of rpa0 is underlined.

Figure 7. Locations

structure that could act as a second RNase E recognition site (Fig. 7). The sequences of the RNA at sites M (UUCAA\GCUGA) and M2 (GCGAG\ UUUCA) share only two and three nucleotides, respectively, with the ten nucleotides of the ACAGfi\AUUUG consensus recognition motif originally proposed (Tomcsanyi & Apirion, 1985), and do not fit with the minimal sequence (Pu\A\UU) reported by Mudd et al. (1990).

296

P. Rignier

and E. Hajnsdorf

The extended hairpin of the tl attenuator formed by annealing the two complementary sequences of the rpsO-pnp cotranscript cleaved by RNase E shares some identity with the potential stem and loop of the 5 S rRNA precursor processed in each strand by RNase E (Fig. 7(b); and see Roy et al., 1983). In both RNAs, the upstream cuts occur 5’ to three U nucleotides that are partly annealed with a downstream UCAA sequence, which is immediately 5’ to the second cleavage sites. Conservation among rRNA precursors of these sequences, which are partly absent from the mature 5 S rRNA, suggests that their complementarity might be important for maturation of 5 S rRNA by facilitating RNase E recognition (Roy et al., 1983). It is possible that in both the rpsO-pnp transcript and the 5 S rRNA precursor, the annealing of the two nearby sites processed by RNase E allows the enzyme bound to one site to recognize the other site without dissociating from RNA (Roy et al., 1983). It is possible that RNase E is not directly involved in the me dependent processing of the rpsO-pnp RNA at M and/or M2 (Subbarao & Apirion, 1989). For example, a cleavage by another endonuclease at M might be dependent upon a prior RNase E cleavage at M2. Alternatively, the me3071 mutation might impair the activity of other endonucleases associated with RNase E in a multienzyme complex (Jain et al., 1982; Pragai & Apirion, 1982). In spite of the small distance (80 nucleotides) that separates the sites processed by RNase III and RNase E, cleavages by each of these enzymes occur independently. (b) Destabilization

of the rps0 mRNA

The dramatic accumulation and stabilization of the rps0 mRNA observed in the RNase E deficient strain compared to the wild-type strain suggests that processing by RNase E is the initial step in the decay of the rps0 message. The stability of the rps0 mRNA in the absence of RNase E may be due, in part, to the secondary structure of the rho-independent attenuator located at its 3’ end (Mott et al., 1985; Plamann & Stauffer, 1996; Schmeissner et al., 1984). In a wild-type strain, the rps0 mRNA might be degraded very rapidly because the 3’ secondary structure is removed as a result of ‘the RNase E dependent cleavage at M2. It is unlikely that the RNase E cleavage at M that generates an RNA nearly identical with the terminated monocistronic transcript is responsible for the destabilization of the rps0 message. The failure to detect the in viva 3’ end of rps0 mRNA resulting from the processing at M2 (Regnier & Portier, 1986) is consistent with a very rapid degradation by 3’ to 5’ exonucleases of RNA processed at this site. Moreover, the absence of any other detectable RNase E sites in the rps0 message upstream from the terminator hairpin (Fig. 4) is also consistent with a destabilizing effect due to the M2 cleavage. The rapid and monophasic decay kinetics of the rpa0 mRNA observed in the

presence of RNase E (Fig. 6), strongly suggests that the cleavage at M2 occurs at the same rate in all the rps0 mRNAs. These mRNAs include the readthrough cotranscript processed by RNase III and the monocistronic transcript terminated at the rhoindependent transcription attenuator located downstream from rps0. The observation implies that RNA downstream from attenuator hairpin tl, containing the M site, is dispensable for the upstream RNase E recognition, and that cleavage of the RNA at M2 occurs independently from the cleavage downstream. This hypothesis is consistent with what has been observed for the maturation of the 9 S rRNA precursor into the 5 S rRNA by RNase E (Roy et al., 1983). (c) The role of RNase E processing in the expression of rps0 An important question is whether the decay initiated by RNase E has any effect on the expression of rps0. As several other ribosomal proteins, 515 negatively autoregulates the translation of its mRNA by binding at its own translation initiation site (Portier et al., 1996). In this kind of control, the level of gene expression is determined by the intracellular concentration of free protein, and is slightly dependent on gene dosage and mRNA concentration. Therefore, it would not seem likely that the rate of mRNA degradation plays a major part in the expression of rps0. However, the increased rate of decay of several ribosomal protein mRNAs observed under conditions of repression suggests that the rapid decay of non-translated messengers presents some advantage for the cells (Cole & Nomura, 1986; Mattheakis & Nomura, 1988). In fact if the newly synthesized mRNA was not turned over under these conditions it would accumulate and titrate out free ribosomal protein necessary to repress its translation. Therefore, the rapid degradation of repressed mRNA may be a way to minimize the pool of free 515 necessary to repress translation (Cole & Nomura, 1986). The bands A, B and C (Fig. 4) that accumulate in the rnclO5 me3071 double mutant correspond to cotranscripts cleaved within the rps0 coding sequence. They suggest that rps0 message might be inactivated by endonucleolytic cleavages occurring in the 5’ part of the gene independently of the 3’ to 5’ chemical decay initiated by RNase E. (d) The role of the RNase E cleavage at M We now have a precise idea of the effect of the RNase III processing and of the RNase E cleavage at M2 on the fate of the different segments of the rpsO-pnp transcripts. However, the consequences of the other cut made by RNase E at M, downstream from the terminator, is still unclear. Available, data suggest that this cleavage is independent from M2 processing (see the discussion above). Trimming of the messenger a few nucleotides downstream from the 3’ protecting secondary structure might prevent

RNa.se E Initiates the 3’ to 5’ processive degradation initiated by cleavage at the RNase III site from reaching the rps0 mRNA. The rapidity with which RNase III cleaves downstream from M (Regnier & Portier, 1986) argues against the involvement of RNase E in the physical separation of the rps0 and pnp messages. (e) Conclusion

Secondary structures at the end of mRNA appear to be involved in the termination of transcription and in the resistance of mRNA to exonucleolytic degradation. The data presented here suggest that they are probably also the targets for the endonucleolytic enzymes that trigger their degradation. Whereas RNase III has been implicated in the destabilization of phagic mRNA (Guameros, 1988), these data are the first to demonstrate that RNase E is involved in the turnover of a cellular mRNA. An intriguing question is whether RNase E, which has the ability to cut RNA upstream from secondary structures, has a more general role in the regulation of gene expression by removing stabilizing structures at the 3’ end of other mRNAs (Mudd et al., 1996). We are extremely grateful to M. Grunberg-Manago for advice on the manuscript and for providing support and facilities. We are particularly indebted to J. Plumbridge for advice on genetic methods and to H. Krisch, C. Olsson and A. Carpousis for critical reading of the manuscript. B. Bachmann, D. Apirion and J. Plumbridge are acknowledged for the gift of strains, H. Krisch and J.-P. Bouche for communicating preprints before publication, and Y. (Gif-sur-Yvette, d’Aubenton-Carafa France) for performing computer analysis. This work was supported by grants from Universite Paris 7 to P.R., from CNRS (URA1139), from the Association pour la recherche sur le cancer and the C.E.E. number SC1*/0194(AM) to M. Grunberg-Manago and from INSERM number 891017 to M Springer.

References Altuvia, S., Locker-Giladi, H., Koby, S., Been-Nun, 0. & Oppenheim, A. B. (1987). RNase III Stimulates the Translation of the ~111 Gene of Bacteriophage Lambda. Proc. Nat. Acud. Sci., U.S.A. 84, 6511-6515. Apirion, D. & Gitelman, D. G. (1980). Decay of RNA in RNA Processing Mutants of E. c&i. Mol. Gen. Genet. 177, 339-343. Bardwell, J. C. A., Regnier, P., Chen, S., Nakamura, Y., Grunberg-Manago, M. BE Court, D. (1989). Autoregulation of RNase III Operon by mRNA Processing. EMBO J. 8, 340-3407. Barry, G., Squires, C. & Squires, C. L. (1980). Attenuation and Processing of RNA from the rplJL-rpoBC Transcription Unit of E. coli. Proc. Nat. Acad. Sci., U.S.A. 77, 3331-3335. Belin, D., Mudd, E. A., Prentki, P., Yi-Yi, Y. & Krisch, H. M. (1987). Sense and Antisense Transcription of Bacteriophage T4 Gene 32. Processing and Stability of the mRNAs. J. Mol. Biol. 194, 231-243.

mRNA Decay

291

Bram, R. J., Young, R. A. & St&z, J. A. (1980). The Ribonuclease III Site Flanking the 23 S Sequences in the 30 S Ribosomal Precursor RNA of E. e&i. Cell, 19, 393401. Carpousis, A. J., Mudd, E. A. t Krisch, H. M. (1989). Transcription and Messenger RNA Processing Upstream of Bacteriophage T4 Gene 32. Mol. Gen. Genet. 219, 3948. Chen, C. A., Beatty, J. T., Cohen, S. N. & Belasco, J. G. (1988). An Intercistronic Stem-Loop Structure

Functions as an mRNA Decay Terminator Necessary but Insufficient for puf mRNA Stability. Cell, 52, 609-619. Cole, J. R. & Nomura, M. (1986). Changes in the Half-life of Ribosomal Protein Messenger RNA Caused by Translational Repression. J. Mol. Biol. 188, 383-392. Daniels, D. L., Subbarao, M. N., Blattner, F. R. & Lozeron, H. A. (1988). & Mediated Late Gene Transcription of Bacteriophage Lambda: RNA Start Point and RNase III Processing Sites in Viwo. Virology, 167, 568-577. Donavan, W. P. & Kushner, S. R. (1986). Polynucleotide Phosphorylase and Ribonuclease II Are Required for Cell Viability and mRNA Turn Over in E. coli K-12. Proc. Nat. Ad. Sci., U.S.A. 83, 120-124. Downing, W. L. & Dennis, P. P. (1987). Transcription Products from the rplKAJL-rpoBC Gene Cluster. J. Mol. Biol. 194, 609-620. Downing, W. L., Sullivan, S. L., Gottesman, M. E. & Dennis, P. P. (1990). Sequence and Transcriptional Pattern of the Essential E. coli secE-nusG Operon, J. Bacterial. 172, 1621-1627. Faubladier, M., Cam, K. & Bouche, J. (1990). E. coli Cell Division Inhibitor DicF RNA of the dicB Operon. Evidence for Its Generation in Vivo by Transcription Termination and by RNase III and RNase E Dependent Processing. J. Mol. Biol. 212, 461-471. Gauss, P., Gayle, M., Winter, R. B. & Gold, L. (1987). The Bacteriophage T4 dexA Gene: Sequence and Analysis of a Gene Conditionally Required for DNA Replication. Mol. Gen. Genet. 206, 24-34. Gegenheimer, P. & Apirion, D. (1980). Structural Characterization and in Vitro Processing of E. coli RNA Transcripts Ribosomal Containing 5’ Triphosphate, Leader Sequences, 16 S rRNA, and Spacer tRNAs. J. Mol. Biol. 143, 227-257. Gros, F., Hiatt, H., Gilbert, W., Kurland, C. G., Risebrough, R. W. & Watson, J. D. (1961). Unstable Ribonucleic Acid Revealed by Pulse Labelling of Escherichia c&i. Nature (London), 190, 581-585. Gross, G. & Hollatz, I. (1988). Coliphage Lambda t0 Terminator Lowers the Stability of Messenger RNA in E. coEi Hosts. Gene, 72, 119-128. Guameros, G. (1988). Retroregulation of Bacteriophage Lambda int Gene Expression. Curr. Topics Mierobiol. Zmmunol. 136, 1-19. Hayashi, M. N. & Hayashi, M. (1985). Cloned DNA Sequences that Determine RNA Stability in Bacteriophage ox174 in Vivo Are Functional. Nucl. Acids Res. 13, 5937-5948. Higgins, C. F., McClaren, R. S. & Newbury. S. F. (1988). Repetitive Extragenic Palindromic Sequences, mRNA Stability and Gene Expression: Evolution by Gene Conversion?-a Review. Gene, 72, 3-14. Hyman, C. H. & Honigman, A. (1986). Transcription Termination and Processing Sites in the Bacteriophage Lambda pL Operon. J. MOE. Biol. 189, 131-141. Jacobson, A. B., Good, L., Simonetti, ,J. & Zuker, M.

P. Rbqnier

and E. HczjnsdoTf

(1984). Some Simple Computational Methods to Improve the Folding of Large RNAs. Nucl. Acids Res. 12, 45-52. Jain, S. K., Pragai, B. & Apirion, D. (1982). A Possible Complex Containing RNA Processing Enzymes. Biochem. Biophys. Res. Commun. 196, 768-778. Klug, G., Adams, C. W., Belasco, J., Doerge, B. $ Cohen, S. N. (1987). Biological Consequences of Segmental Alterations in mRNA Stability: Effects of Deletion on the Intercistronic Hairpin Loop Region of the Rhodobacter Capsulatus puf Operon. EMBO J. 6, 35 15-3520. Krinke, L. & Wulff, D. L. (1987). Oop RNA, Produced from Multicopy Plasmids, Inhibits Lambda cI1 Gene Expression through an RNase III Dependent Mechanism. Genes Develop. 1, 1005-1013. Mattheakis, L. C. & Nomura, M. (1988). Feedback Regulation of the spc Operon in E. coli: Translational Coupling and mRNA Processing. J. Bacterial. 170, 4484-4492. Melefors, 0 & von Gabain, A. (1988). Site Specific Endonucleolytic Cleavages and the Regulation of Stability of E. coli ompA mRNA. Cell, 52, 893-901. Misra, T. K. & Apirion, D. (1979). RNase E, an RNA Processing Enzyme from E. coli. J. Biol. Chem. 254, 11154-11159. Mott, J. E., Galloway, J. L. & Platt, T. (1985). Maturation of E. coli Tryptophan Operon: Evidence for 3’ Exonucleolytic Processing after Rho-independent Termination. EMBO J. 4. 1887-1891. Mudd, E. A., Prentki, P., Belin, D. & Krisch, H. M. (1988). Processing of Unstable Bacteriophage T4 Gene 32 mRNAs into a Stable Species Requires E. wli Ribonuclease E. EMBO J. 7, 360-3607. Mudd, E. A., Carpousis, A. J. & Krisch, H. M. (1990). E. coli RNase E Has a Role in the Decay of Bacteriophage T4 mRNA. Genes Develop. 4,873-881. Newbury, S. F., Smith, N. H., Robinson, E. C., Hiles, I. D. & Higgins, C. F. (1987). Stabilization of Translationally Active mRNA by Prokaryotic REP Sequences. Cell, 51, 297-310. Nilsson, G., Belasco, J. G., Cohen, S. N. & von Gabain, A. (1984). Growth Rate Dependent Regulation of mRNA Stability in E. coli. Nature (London), 312, 75-77. Nilsson, G., Lundberg, U. & von Gabain, A. (1988). In Vivo and in Vitro Identity of Site Specific Cleavages in the 5’ Non Coding Region of ompA and bla mRNA in E. coli. EMBO J. 7, 2269-2275. Panayotatos, N. & Truong, K. (1985). Cleavage within an RNase III site can Control mRNA Stability and Protein Synthesis in Vivo. Nucl. Acids Res. 13> 2227-2241. Pedersen, S., Reeh, S. & Friesen, J. D. (1978). Functional mRNA Half Lives in E. coli Mol. Gen. Genet. 166, 329-336. Plamann, M. D. & Stauffer, G. V. (1990). E. coli gZyA mRNA Decay: the Role of 3’ Secondary Structure and the Effects of the pnp and mb Mutations. Mol. Gen. Cenet. 220, 301-306. Plumbridge, J. A., Dondon, J., Nakarmura, Y. $ Grunberg-Manago, M. (1985). Effects of NusA Protein on Expression of the nwA-injB Operon in E. coli. Nucl. Acids Res. 13, 3371-3388. Portier, C., Dondon, L., Grunberg-Manago, M. & Regnier, P. (1987). The First Step in the Functional Polynucleotide E. coli Inactivation of the Edited

Phosphorylase Messenger is a Ribonuclease III Processing at the 5’ End. EMBO J. 6, 21652170. Portier, C., Dondon, L. & Grunberg-Manago, M. (1990). Translational Autocontrol of the E. wli Ribosomal Protein S15. J. Mol. Biol. 211, 407-414. Pragai, B. & Apirion, D. (1982). Processing of Bacteriophage T4 Transfer RNAs: Structural Analysis and in Vitro Processing of Precursors that Accumulate in RNase E- Strains. J. Mol. Biol. 154, 465-484. Regnier, P. & Grunberg-Manago, M. (1989). Cleavage by RNase III in the Transcripts of the met Y-nusA-infB Operon of E. coli Release the tRNA and Initiates the Decay of the Downstream mRNA. J. Mol. Biol. 210, 293-302. Regnier, P. & Portier, C. (1986). Initiation, Attenuation and RNase III Processing of Transcripts from the E. co& Operon Encoding Ribosomal Protein S15 and Polynucleotide Phosphorylase, J. Mol. Biol. 187, 23-32. Regnier, P., Grunberg-Manago, M. & Portier, C. (1987). Nucleotide Sequence of the pnp Gene of E. coli Encoding Polynucleotide Phosphorylase; Homology of the Primary Structure of the Protein with the RNA Binding Domain of Ribosomal Protein Sl. J. Biol. Chem. 262, 63-68. Roy, M. K., Singh, B., Ray, B. K. & Apirion, D. (1983). Maturation of 5 S rRNA: Ribonuclease E Cleavages and Their Dependence on Precursor Sequences. Eur. J. Biochem. 131, 119-127. Ruckman, J., Parma, D., Tuerk, C., Hall, 1). H. & Gold, L. (1989). Identification of a T4 Gene Required for Bacteriophage mRNA Processing. New Biol. 1, 54-65. Saito, H. & Richardson, C. C. (1981). Processing of mRNA by Ribonuclease III Regulates Expression of Gene 1.2 of Bacteriophage T7. Cell, 27, 533-542. Schmeissner, U., McKenny, K., Rosenberg, M. & Court,, D. (1984). Removal of a Terminator Structure by RNA Processing Regulates int Gene Expression. J. Mol. Biol. 176, 39-53. Simons, R. W. & Kleckner, N. (1988). Biological Regulation by Antisense RNA in Prokaryotes. Annu. Rev. Genet. 22, 567-600. Steege, D. A., Cone, K. C.; Queen, C. & Rosenberg, M. (1987). Bacteriophage Lambda Gene Leader RNA: RNA Processing and Translational Initiation Signals. J. Biol. Chem. 262, 17651-17658. Subbarao, M. N. & Apirion, D. (1989). A Precursor for a Small Stable (10Sa RNA) of E. coli. Mol. Gen. Genet. 217. 499-504. Takata, R., Izuhara, M. & Hori, K. (1989). Differential Polynucleotide the E. coli Degradation of mRNA. Nucl. Acids Res. 17, Phosphorylase 7441-7451. Tomcsanyi, T. & Apirion, D. (1985). Processing Enzyme Ribonuclease E Specifically Cleaves RNA1 an Inhibitor of Primer Formation in Plasmid DNA Synthesis. J. Mol. Biol. 185, 713-720. Uzan, M., Favre, R. & Brody, E. (1988). A Nuclease That Cuts Specifically in the Ribosome Binding Site of Some T4 mRNAs. Proc. Nat. Aead. Sci., U.S.A. 85, 8895-8899. Wong, H. C. & Chang, S. (1986). Identification of a Positive Regulator That Stabilizes mRNAs in Bacteria. Proc. Nat. Acad. Sci., U.S.A. 83, 3233-3237.

by P. Chambon