Bacteriophage T4 polynucleotide kinase triggers degradation of mRNAs

Abstract

The bacteriophage T4-encoded RegB endoribonuclease is produced during the early stage of phage development and targets mostly (but not exclusively) the Shine–Dalgarno sequences of early genes. In this work, we show that the degradation of RegB-cleaved mRNAs depends on a functional T4 polynucleotide kinase/phosphatase (PNK). The 5′-OH produced by RegB cleavage is phosphorylated by the kinase activity of PNK. This modification allows host RNases G and E, with activity that is strongly stimulated by 5′-monophosphate termini, to attack mRNAs from the 5′-end, causing their destabilization. The PNK-dependent pathway of degradation becomes effective 5 min postinfection, consistent with our finding that several minutes are required for PNK to accumulate after infection. Our work emphasizes the importance of the nature of the 5′ terminus for mRNA stability and depicts a pathway of mRNA degradation with 5′- to 3′-polarity in cells devoid of 5′–3′ exonucleases. It also ascribes a role for T4 PNK during normal phage development.

Although significant progress has been made in deciphering the different pathways of mRNA degradation in its host Escherichia coli, the way that bacteriophage T4 degrades its mRNAs and the contribution of host and phage factors are far less understood (1). Immediately after T4 infection, the early mRNAs are synthesized in a soup of host RNases. Among them, the 3′–5′ exoribonucleases, polynucleotide phosphorylase, RNase R, and RNase II together with an oligoribonuclease are responsible for the swift degradation of mRNAs into mononucleotides. Endonucleolytic cuts greatly accelerate mRNA decay by creating new entry sites for 3′–5′ exoribonucleases that bypass the structural barriers. The endoribonuclease RNase E plays a major role here. Indeed, examination of the decay of T4 gene 32 mRNA was instrumental in the discovery of the role of RNase E in mRNA decay (1). RNase G, a paralogue of RNase E, is also involved in mRNA decay, although to a much lesser extent than RNase E (2, 3).

RNases E and G share the remarkable property of sensing the nature of the 5′ terminus of their substrates. These endonucleases are significantly more active on RNA substrates bearing a 5′-monophosphate than a 5′-triphosphate or -hydroxyl (4, 5). This property accounts for a mechanism of mRNA degradation from the 5′ terminus. Indeed, the 5′-monophosphorylated RNA fragment produced by the first RNase E/G cleavage can be rapidly attacked again by RNase E or G, leading to a cascade of endonucleolytic cuts, often of 5′- to 3′-polarity (6). Many E. coli primary transcripts enter this pathway of degradation upon pyrophosphate removal from their 5′-terminal triphosphate group by the RNA pyrophosphohydrolase, RppH (7).

RegB is a sequence-specific endoribonuclease encoded by T4 that is expressed shortly after infection (8). It targets mostly but not uniquely Shine–Dalgarno sequences of early genes (9). The direct consequence of cleavages in Shine–Dalgarno sequences is the functional inactivation of mRNAs (10). In addition, RegB cleavages expedite the degradation of T4 early (but not middle or late) bulk mRNA, consistent with the quasiabsence of RegB consensus sequences in these two classes of mRNAs (11). By providing a mechanism that frees the translation apparatus from abundant early mRNAs, this endonuclease is thought to facilitate the transition between early and later stages of T4 development. The RegB nuclease produces 5′-OH and 2′,3′-cyclic PO₄ termini (8, 12) like two E. coli ribotoxins, YoeB and RelE, with which it shows structural and functional similarities (13). The mechanism by which RegB facilitates mRNA degradation is still unclear. One likely model is that RegB accelerates mRNA decay by increasing the number of entry sites for 3′–5′ exoribonucleases within long polycistronic mRNAs. Another mechanism was suggested by the finding that some early transcripts undergo cleavages by RNases G and E that depend on prior cleavage by RegB a few nucleotides upstream (10, 14, 15). These cuts, called secondary cuts, were detected in seven different early transcripts. In most cases, multiple adjacent cleavages were observed. In all seven cases, the primary RegB cut does not lie in a Shine–Dalgarno sequence but at some distance upstream. We interpreted these observations as meaning that RegB triggers a cascade of endonucleolytic cuts that leads to degradation of the transcript with 5′- to 3′-polarity (10). However, the dependence of RNase G and E cleavages on an RegB primary cut seemed paradoxical in view of the strong preference of both RNases G and E for 5′-monophosphorylated RNA substrates.

Here, we show that formation of secondary cleavages and other RNase E/G cleavages farther downstream in transcripts depends on prior phosphorylation of the 5′-OH terminus left by RegB by T4 polynucleotide kinase/phosphatase (PNK). We, thus, describe a way of entering the host 5′-phosphate–dependent degradation pathway of RNases E and G using the RegB and PNK enzymes of T4. We also ascribe a role for T4 PNK during normal phage development other than in tRNA repair (16, 17).

Results

T4 PNK Allows RNases G and E to Act at Secondary Sites.

How can RNases G and E be responsible for secondary cleavages when RegB leaves a 5′-OH moiety after cleavage? Bacteriophage T4 codes for a 5′-PNK-3′ phosphatase that, a priori, could phosphorylate RNA termini during the phage cycle. PNK, encoded by the pseT gene, has three activities: 5′-PNK, 3′- phosphomonoesterase, and 2′,3′-cyclic phosphodiesterase. Both DNA and RNA are substrates of PNK (18, 19).

To test whether PNK was involved in the generation of secondary cuts, we analyzed the 5′-region of the motB and cef mRNAs by primer extension at various times after infection of strain CSH26 with WT T4 or a phage carrying an almost total deletion of the pseT gene (T4ΔpseT; Materials and Methods). A similar analysis performed after infection of a host deficient in RNase G (CSH26rng::cat) with WT T4 was included for comparison. The full-length motB mRNA (denoted by P in Fig. 1A) was converted by RegB into shorter transcripts (Fig. 1A, arrow 1) at the same rate in all three conditions of infection analyzed. Moreover, secondary cleavages (Fig. 1A, arrow 2), which appeared typically 4–5 min postinfection (10), were significantly reduced in the rng mutant and almost suppressed in the absence of a functional PNK (Fig. 1A and Fig. S1A). Concomitantly, the RegB processed transcripts accumulated to levels above those levels observed in infection of WT cells with WT T4 (Fig. 1A). A very similar pattern was observed for the cef mRNA (Fig. 1B). In this case, RNase G contributes to only one and RNase E contributes to the other of the two secondary cleavages (Fig. S1B) (15). Both cleavages disappeared in the absence of PNK (Fig. 1B). Similar analyses showed that the multiple secondary cleavages observed upstream of genes 39 and 43, contributed mostly by RNase G, are dramatically reduced in the T4ΔpseT infection (Fig. S2). Thus, RNase G and E cleavages at secondary sites depend on a functional PNK. In an infection by a phage carrying the pseT1 mutation, where the 3′-phosphatase activity is abolished but the kinase activity is unaffected (20), secondary cleavages upstream of genes 39 and 43 occur normally (Fig. S3). Taken together, these results show that cleavages at secondary sites depend on prior phosphorylation of the RegB-generated 5′-OH RNA termini by the kinase activity of PNK.

Fig. 1.

Effect of PNK or RNase G inactivation on the generation of secondary cleavages. Primer extension assays of the intergenic regions upstream of the motB (A) and cef (B) genes using primers 3 and 6 (Fig. 3A and Table S1), respectively, were performed on RNA extracted at the times indicated above the gel images after infection of CSH26 or CSH26 rng::cat by T4 WT or CSH26 by T4ΔpseT. Arrows on the left show RT stops corresponding to RegB cleavages (labeled 1), secondary cleavages (collectively labeled 2), and transcription starts (P). The dots in the sequences replace nucleotides not represented. Their number is shown below the sequence. RegB cleavage sites and start codons are boxed. Bent arrows indicate transcriptional starts. Vertical arrows show the positions of RegB, RNase G (G), and RNase E (E) cleavage sites. The intensity of the RT stops is indicated by the shading of the arrows; the darkest arrows represent the strongest stops.

PNK Controls Endonucleolytic Cuts at Some Distance from RegB Sites.

Given that PNK controls the accessibility of RNases G and E to secondary sites through phosphorylation of the 5′ terminus, we asked whether it could also control the accessibility of these RNases to sites located farther downstream. We, therefore, performed primer extension assays on WT T4 mRNAs using primers located either 200 (Fig. 2A) or 390 (Fig. S4) nt downstream of the initial RegB cleavage site in the cef transcript. These assays revealed many additional reverse transcriptase (RT) stops (locations given in Fig. 2B), all of which were weaker than those stops seen at the RegB or secondary site. The same analysis performed on RNA isolated after infection of host strains defective in RNase E (rne-1) or G (rng::cat) permitted attribution of most of these RT stops to cleavages by RNase E; only a few were RNase G-dependent (Fig. S5 A and B). After infection with T4ΔpseT, the cleavages located in the first 200 nt downstream of the RegB site disappeared or were dramatically weakened (Fig. 2A). The cuts located farther away (numbered 11–18 in Fig. S4) were also affected by the lack of PNK but to a lesser extent.

Fig. 2.

Effect of PNK or RegB inactivation on the occurrence of RNase E and G cleavages in the cef transcript. (A) Primer 4 (Fig. 3A and Table S1), located 200 nt from the RegB site, was used in primer extensions of RNAs extracted at the times indicated above the gel image after infection of CSH26 by T4 WT, T4ΔpseT, or T4regB R52L. Sequences of RNA extracted 4 min after infection of CSH26 by T4 regB R52L obtained with the same primers were run along the cDNAs (lanes labeled CTAG). RNase E and G cleavages are indicated by arrows; numbering starts with the RegB cleavage. *New degradation product appearing upstream of the RegB cleavage site in regB mutant strains. (B) The positions of RegB, RNase E (arrows below the sequence), and RNase G (arrows above the sequence) cleavages mapped in A and Fig. S5 in the cef sequence. The cef translation initiation and termination codons are underlined.

Because an RegB cut is the initial event that triggers all others, an regB mutation should have the same effect on the RNase G and E cleavages as the ΔpseT mutation. Primer extension carried out with primer 4 showed that the pattern of cleavage in T4regB mutant infection is strikingly similar to the pattern observed in the ΔpseT mutant (Fig. 2A, Right). The effect of the regB mutation on the cleavages located farther downstream in the cef mRNA is much less visible (Fig. S4, Right), similar to what was seen with T4ΔpseT.

In conclusion, our results show that, when the 5′-OH RNA end generated by RegB is not phosphorylated by PNK, the attack by RNases E and G is blocked or decreased over a distance of about 300 nt from the RegB site. However, after PNK has modified the 5′ terminus and RNase G (or E) has cut at secondary sites, the new 5′-monophosphorylated RNA ends can presumably activate RNases E and G in cascade.

PNK Controls the Decay of Early Transcripts Predominantly from Their 5′ Termini.

We asked about the consequences of PNK activity on the rate of degradation of RegB-processed transcripts. The primer extension experiments shown above already indicated that, in a T4ΔpseT infection, the RegB-processed RNA species accumulated to levels much higher than in the WT infection, suggesting mRNA stabilization (Fig. 1 and Fig. S2). Because the early promoters are abruptly turned off 2–3 min after infection (10, 21), any variation in the accumulation of early transcripts observed after this time in different genetic backgrounds should reflect modifications in their rate of degradation.

cef mRNA.

The RNA fragment generated by RegB cuts upstream of the Shine–Dalgarno sequence of cef and downstream of orf39.1 (Fig. 3A) was chosen for half-life studies, because the RegB cleavage on its 5′-side gives rise to secondary cuts (discussed above). This transcript bears a hairpin structure immediately upstream of the distal RegB site with the stabilizing UUCG tetraloop, showing enough stability to block the progression of avian myeloblastosis virus reverse transcriptase in a primer extension assay (10). Northern blot analysis of RNA extracted after T4 infection using a probe for the cef mRNA (Fig. 3A, probe 6) revealed a unique band of the anticipated size (Fig. 3C). This 1.2-kb RNA species is hereafter designated cef mRNA.

Fig. 3.

Effect of the pseT deletion on cef and orf30.8 mRNA accumulation during T4 development. (A) Topology of the motB-cef-39 region. Gray rectangles represent ORFs. Gene 39 is truncated. The vertical lines show RegB cleavage sites. Hairpin structures are symbolized by a vertical line topped with a circle. (B) Topology of the orf30.8 region. Symbols are as in A. (C) Northern blot analysis of RNA extracted at the times (in minutes) indicated above the blot after infection of CSH26 with either T4 WT or T4ΔpseT. Equal amounts of RNA were loaded in each lane of the agarose gel. The 1.2-kb cef and orf30.8 RNA species were probed with ³²P-labeled oligonucleotides 6 and 7, respectively. The 23S rRNA was probed with the 23S oligonucleotide (Table S1). (D and E) Quantification of the cef and orf30.8 signals obtained in Northern blots from two independent experiments (one of which is shown in C) using ImageJ software after normalization to 23S rRNA. SD is indicated by error bars.

Quantification of the signals of Fig. 3C showed that the cef mRNA accumulated until the fifth minute of a WT T4 infection and declined thereafter (Fig. 3D). In a T4ΔpseT infection, the cef mRNA started to accumulate at a rate similar to the rate observed with WT phage; however, past the fourth minute, it continued to accumulate to reach levels two times those levels obtained with the WT and started to decay after the seventh minute, but it decayed more slowly than in the WT infection (Fig. 3D). Thus, the elimination of PNK leads to a significant stabilization of the cef mRNA, and therefore, 12 min postinfection with the pseT-deficient phage, the cef mRNA was fourfold more abundant than with the WT phage.

To determine the respective contribution of the kinase and phosphatase activities of PNK in cef mRNA degradation, we carried out a similar Northern analysis with RNA isolated after infection with the T4pseT1 mutant deficient for just the phosphatase activity (Fig. 4A). Quantification of the T4 WT vs. pseT1 Northern blot (Fig. 4B) showed that eliminating the phosphatase activity of PNK led to a pattern of cef mRNA accumulation similar to the pattern observed during WT infection, with a slight shift to earlier times. Thus, the lack of kinase activity is primarily responsible for the slower degradation rates.

Fig. 4.

Effect of inactivation of the phosphatase activity of PNK on cef and orf30.8 mRNA accumulation during T4 development. (A) Northern blot analysis using probe 6 (cef) or 7 (orf30.8) of RNA extracted after infection of CSH26 with either T4 WT or T4pseT1. Quantification of the cef (B) and orf30.8 (C) signals obtained in Northern blots. Two independent experiments were performed, one of which is shown in A, as indicated in Fig. 3. SD is indicated by error bars.

orf30.8 mRNA.

In the cases examined above (motB, cef, and genes 39 and 43), the PNK-dependent secondary cleavages were observed after a primary RegB cut in a non-Shine–Dalgarno sequence. Therefore, we asked whether the stability of a transcript processed in a Shine–Dalgarno sequence, such as sequence of orf30.8 (which does not undergo secondary cleavages), could also be influenced by PNK. RegB makes flanking cuts in the early T4 orf30.8 mRNA (one in the Shine–Dalgarno and the other 59 nt downstream of the end of the orf), producing a 333-nt RNA fragment visible in Northern blots using probe 7 (Fig. 3 B and C). This mRNA also differs from the cef mRNA in that no secondary structure can be formed on its 3′-side. In infection of E. coli with WT T4, accumulation of the orf30.8 mRNA reached a maximum 3 min postinfection and declined abruptly thereafter, indicative of a very short-lived mRNA (Fig. 3E). In infection with T4ΔpseT, the orf30.8 mRNA started to accumulate in a manner very similar to the manner observed in the WT infection. However, after the third minute, the transcript decayed in a biphasic mode: first, rather rapidly but less abruptly than in the WT infection until the sixth minute and then, extremely slowly. Between the minutes 6 and 12, the orf30.8 mRNA level was consistently about fourfold higher than the level in the WT infection (Fig. 3E). Preventing the removal of the 2′,3′-phosphate led to a very similar accumulation curve for this transcript in the T4pseT1 infection compared with WT (Fig. 4 A and C) but with a 1-min shift to earlier times, which was observed with the cef transcript. Thus, the kinase activity of PNK plays the main role. The biphasic mode of decay (Fig. 3E) in the pseT deletion mutant suggests that this small transcript is degraded through two pathways. The first pathway is independent of PNK and effective as soon as the promoter is shut down 3 min postinfection. The second pathway depends on the kinase activity of PNK and takes place after the fifth minute of infection, presumably after sufficient PNK has accumulated. This finding shows that, even in the absence of detectable secondary cuts, PNK plays a role in the degradation of RegB-processed transcripts from their 5′ termini. It is noteworthy that, with both cef and orf30.8 mRNAs, the destabilizing effect of PNK becomes effective after 5 min of phage development, around the time that secondary cleavages (by RNases G and E) start to be detectable (10, 14).

PNK Accumulates Slowly After Infection.

The above data shows that the effect of PNK on mRNA stability is revealed only after 5 min of infection. We, therefore, checked whether this finding reflected the timing of PNK biosynthesis during the phage cycle. This process was done by pulse-labeling of cells with [³⁵S]-methionine for 1 min at different time intervals after infection to identify newly synthesized phage proteins. The band corresponding to PNK was identified on protein gels as a band comigrating with purified T4 PNK and present after WT but not T4ΔpseT infection. PNK was barely visible before the fourth minute of infection and accumulated thereafter (Fig. 5). This delay in expression likely explains the timing of the effect of PNK on cef and 30.8 mRNA stability.

Fig. 5.

Rate of PNK synthesis during T4 development. Phage proteins synthesized after infection of CSH26 with T4 WT or ΔpseT (30 °C) were pulse-labeled with [³⁵S]-Methionine for 1 min at each of the times indicated after infection as described in Materials and Methods. The T4 proteins were run on SDS/PAGE along with purified T4 PNK (Biolabs). Aliquots of T4 WT or ΔpseT-infected cells were loaded alternatively for each pulse. The arrowheads indicate the position of the PNK band.

Discussion

We showed in this work that bacteriophage T4 PNK plays a key role in the degradation of phage mRNAs that are processed by the T4 RegB endonuclease. On cleavage, RegB leaves 5′-OH and 2′,3′-cyclic phosphate termini (8, 12). Phosphorylation of the 5′-OH terminus by the kinase activity of PNK permits attack by the host RNases G and E, two endoribonucleases that have strong preference for 5′-monophosphorylated RNA termini (4–6, 22). RNases E and G cut first in the vicinity of the initial RegB cleavage site, generating the so-called secondary cleavages (10, 14, 15). Because the cleavage products of RNases E and G bear 5′-PO₄ groups, they are themselves substrates for additional cleavage by these endonucleases, causing a cascade of endonucleolytic cuts with 5′- to 3′-polarity that leads to mRNA destabilization. Supporting this view is the fact that the RNase G and E cleavages that normally occur in the cef transcript over a distance of about 300 nt downstream of the RegB cut are abolished or significantly reduced in infections with a T4 PNK deletion mutant. An important observation of this paper is that the two RegB-generated transcripts, cef and orf30.8, are stabilized after 5 min of phage development in the absence of PNK. Although primary cuts by RegB can be observed as early as the first minute of infection (8), the secondary cleavages (by RNases G and E) start to be detectable only 3–4 min later (10, 14), concomitantly with the observed destabilizing effect of PNK. Here, we showed that the PNK protein accumulates slowly after infection, suggesting that the delay between primary and secondary cuts reflects the time necessary for PNK to accumulate in sufficient quantities after infection. The work by Zajanckauskaite et al. (15) proposed that a phage factor can modify RNases E and G, permitting them to attack the RegB-processed transcripts with a 5′-OH terminus. We showed here that a phage factor indeed plays a key role, but it modifies the 5′-end of mRNAs rather than the host RNases. Thus, RegB and PNK cooperate to create new entry sites for RNases G and E, leading to mRNA degradation in a stage-dependent manner. In an earlier work, we reported that RegB destabilizes (threefold) bulk early mRNA synthesized during the first 3 min of infection (11). Clearly, the RegB/PNK-dependent degradation pathway, which is effective after the fifth minute of infection, cannot account for this early wave of degradation. Other pathways may play a role in this phenomenon, such as the exonucleolytic attack of RegB-cleaved fragments from their 3′-end or in cases where translation is blocked after RegB cleavage, the greater access to RNase E in a 5′-independent mechanism.

The RegB/PNK degradation pathway described here unveils additional layers of regulation during T4 development. In the case of the cef mRNA and the six other transcripts showing secondary cleavages, RegB does not cut in translation initiation regions. Thus, until sufficient PNK has accumulated to trigger their degradation by RNases G and E, these relatively stable early transcripts should produce their encoded proteins long after early promoters have shut down (minutes 2–3 postinfection) (10, 21). In agreement with this hypothesis, we have previously observed that the biosynthesis of ComCα, one of the proteins encoded by the cef mRNA, persists until the onset of the late period, although with a decreasing rate of synthesis (23). Fig. 6 summarizes our view of the way that the cef mRNA is degraded by the RegB/PNK pathway. The orf30.8 mRNA illustrates the most common case, where RegB cleavage occurs in the Shine–Dalgarno sequence and leads to immediate translation shut down. The lack of a 3′-stabilizing structure in orf30.8 makes the transcript particularly susceptible to degradation from the 3′-end, which probably explains the inability to detect secondary cleavages in this transcript. We, nonetheless, saw that turnover of this transcript was PNK-dependent after the fourth minute of infection. Thus, in the first 3 min postinfection, expression of orf 30.8 is expected to be quite similar to the expression of other typical early genes (e.g., motA), with a burst of protein synthesis very soon after infection followed by abrupt silencing (10).

Fig. 6.

Schematic representation of the pathways of cef mRNA degradation during the first 6 min of infection. The vertical gray arrow on the left represents time (in minutes) after T4 infection. The horizontal lines represent the cef transcripts with triphosphate (PPP), monophosphate (P), and hydroxyl (OH) 5′-extremities. RegB cleavage sites (vertical white arrows) and the stem loop structure downstream of the cef transcription unit are shown. 3′–5′ exonucleases are represented by a white three-quarters circle symbol. The two endoribonucleases, RNases E and G, are collectively symbolized by a black hexagon. PNK is represented as a gray three-quarters circle symbol. Additional explanations are in the text.

A few minutes after infection, many T4 genes are transcribed from a proximal middle promoter in addition to the distal early and sometimes, middle promoters (24). This finding is the case for genes 39 and 43 coding for a subunit of topoisomerase and DNA polymerase, respectively. Both the polycistronic early transcripts and the monocistronic middle transcripts for genes 39 and 43 end at a strong terminator immediately downstream. What distinguishes the two overlapping transcripts in each case is that the polycistronic species can be cut by RegB in the upstream intergenic regions, whereas the monocistronic middle transcripts cannot (Fig. 3A). This finding leads to degradation of the polycistronic transcripts but leaves the overlapping middle mRNA intact. Thus, the RegB/PNK pathway of degradation also permits uncoupling of the fate of the overlapping mRNA species. It is noteworthy that this mechanism of degradation takes place 4–5 min after infection, precisely when middle transcription takes over from early transcription.

Until now, the only role ascribed to T4 PNK was the role of a healing enzyme in a process of tRNA repair (ref. 17 and references therein). The work presented here shows clear evidence that PNK can also participate in a mechanism of mRNA degradation through 5′-phosphorylation. The status of the 5′ terminus has recently been highlighted as an important factor for mRNA degradation. Indeed, the number of endo- and exoribonucleases that require 5′-monophosphorylated RNA as substrate has increased significantly these last years. The RNase E/G family of endonucleases in the γ-proteobacteria (4, 5, 22), the 5′–3′ exonucleases RNases J1 and J2, the endonuclease RNase Y in Bacillus subtilis (25, 26), and the 5′–3′ exonucleases Xrn1, Rat1, and to a lesser extent, Rrp17, in yeast (27–30) are among them. In E. coli and B. subtilis, 5′-monophosphorylated RNAs are generated by cleavages by most endoribonucleases or derived from 5′-triphosphorylated primary transcripts after removal of a pyrophosphate group by the RppH pyrophosphohydrolase (7, 31). In yeast, an enzyme with pyrophosphohydrolase activity to 5′-triphosphorylated RNA (called Rai1) has been proposed to stimulate Rat1 activity in the nucleus by a modification of the 5′-end of decapped RNA (32). RNA kinases have been reported to influence the processing or degradation of some RNAs in Eukaryotes, but their mechanism of action is not yet clear (33–35).

Materials and Methods

Bacteria and Bacteriophages.

E. coli BE (sup°) was our standard strain for growing bacteriophage T4. CTr5x is restrictive for the growth of T4 pseT mutants (36). CSH26 [ara, Δ(lac-pro), and thi] and GC11, which is CSH26rng::cat (37), were described. CSH26rne-1 and the CSH26(rng::cat, rne-1) double mutant in RNases G and E were constructed by transduction with P1 grown on strain GW20 (rne-1) (38). Bacteriophage T4 WT was T4D. T4regB R52L and T4pseT1 were described (11, 19, 36). T4ΔpseT mutant was obtained using the directed insertion/substitution mutagenesis method described in ref. 39. The resulting phage carries an 853-nt deletion of the pseT gene (903 nt) starting at the AUG translation initiation codon.

Experiments with RNA.

For RNA extraction, E. coli cells were grown in Mops–Tricine medium supplemented with 1% casamino acids and infected at a cell density of 2.5 × 10⁸/mL at a multiplicity of 10 at 30 °C. When the effect of RNase E inactivation had to be analyzed, the temperature of the cell cultures (CSH26 rne-1 or CSH26 rne-1, rng::cat) was raised to 43 °C for 30 min before phage infection. Methods for RNA extractions, primer extension, and RNA sequencing are described in ref. 8–10. Routinely, 700 μL infected cells were added to 77 μL SDS (10% wt/vol) and Na-EDTA (20 mM) and 500 μL water-saturated phenol maintained at 70 °C. RNA was then phenol-extracted three times. Northern blots were performed essentially as described in ref. 40. The membrane was scanned with a PhosphorImager. Quantification of the band intensities was done with ImageJ software (41) after removing the background according to the rolling ball method. The values were then normalized to 23S rRNA, which was considered constant throughout infection. The oligonucleotides used in this study are listed in Table S1.

T4 Protein Labeling and Electrophoresis.

In vivo labeling and electrophoresis of T4 proteins were performed as described in ref. 10. Proteins were pulse-labeled for 1 min by the addition of l-[³⁵S]-methionine (>1,000 Ci/mmol, 20 μCi/mL; Amersham) to the medium. T4 proteins were separated by electrophoresis on an SDS-EDTA 12% polyacrylamide gel.