Báo cáo khoa học: Characterization of the tRNA and ribosome-dependent pppGpp-synthesis by recombinant stringent factor from Escherichia coli

Characterization of the tRNA and ribosome-dependent pppGpp-synthesis by recombinant stringent factor from Escherichia coli Rose-Marie Knutsson Jenvert1,2 and Lovisa Holmberg Schiavone1

1 Cell Biology Unit, Natural Science Section, So¨ derto¨ rns Ho¨ gskola, Huddinge, Sweden 2 Department of Cell Biology, Arrhenius Laboratories E5, Stockholm University, Sweden

Keywords pppGpp; RelA; ribosome; stringent response; tRNA

Correspondence L. Holmberg Schiavone, Cell Biology Unit, Natural Science Section, So¨ derto¨ rns University College, S-141 89 Huddinge, Sweden Fax: +46 8608 4510 Tel: +46 8608 4597 E-mail: lovisa.holmberg-schiavone@sh.se

(Received 25 August 2004, revised 4 November 2004, accepted 25 November 2004)

doi:10.1111/j.1742-4658.2004.04502.x

tetracycline and viomycin;

Stringent factor is a ribosome-dependent ATP:GTP pyrophosphoryl trans- ferase that synthesizes (p)ppGpp upon nutrient deprivation. It is activated by unacylated tRNA in the ribosomal amino-acyl site (A-site) but it is unclear how activation occurs. A His-tagged stringent factor was isolated by affinity-chromatography and precipitation. This procedure yielded a protein of high purity that displayed (a) a low endogenous pyrophosphoryl transferase activity that was inhibited by the antibiotic tetracycline; (b) a low ribosome-dependent activity that was inhibited by the A-site specific antibiotics thiostrepton, micrococcin, (c) a tRNA- and ribosome-dependent activity amounting to 4500 pmol pppGpp per pmol stringent factor per minute. Footprinting analysis showed that stringent factor interacted with ribosomes that contained tRNAs bound in classical states. Maximal activity was seen when the ribosomal A-site was presaturated with unacylated tRNA. Less tRNA was required to reach maximal activity when stringent factor and unacylated tRNA were added simultaneously to ribosomes, suggesting that stringent factor formed a complex with tRNA in solution that had higher affinity for the ribosomal A-site. However, tRNA-saturation curves, performed at two different ribo- some ⁄ stringent factor ratios and filter-binding assays, did not support this hypothesis.

alarmone results in a down-regulation of stable RNA synthesis (rRNA and tRNA) and up-regulation of the synthesis of mRNAs encoding enzymes involved in amino acid biosynthesis.

alarmone

Abbreviations DMS, dimethylsulfate; SF, stringent factor; A-site, amino-acyl site; P-site, peptidyl-site; TC-ribosomes, twice salt-washed tight-couple ribosomes; T4-mRNA, gene T4 mRNA-fragment.

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

685

Prokaryotic cells coordinate the rate of mRNA, rRNA and tRNA synthesis via the stringent response, which is activated upon nutrient deprivation or stress (reviewed in [1]). This physiological response is initi- ated when stringent factor (SF) binds to translating but stalled ribosomes that are starved for cognate amino-acyl tRNAs. The stringent factor is activated by the stalled ribosomal complex and starts to synthesize (p)ppGpp from GTP(GDP) using the ATP as a phosphate donor. Stringent factor is thus a ribosome-dependent ATP:GTP pyrophosphoryl trans- ferase that synthesizes (p)ppGpp. Production of this The stringent factor was first isolated from ribo- somal salt-wash fractions [2] and was identified as the producer of magic spots (ppGpp and I and II pppGpp [1]), in a (p)ppGpp synthesis assay. In this assay, purified stringent factor is incubated with ribo- somes, ATP and radiolabelled GTP. This is followed by separation of newly synthesized pppGpp from GTP by thin-layer chromatography [2]. The reported

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

([3]

activity of purified SF in the ribosome-dependent reaction varies extensively, between 100 and 10 000 pmol pppGppÆpmol SF)1Æmin)1 and references therein). The purified factor was also shown to dis- play low activity in a ribosome-independent reaction in the presence of 20% (v ⁄ v) alcohol [4].

protein and Ni-agarose beads according to Wendrich et al. [7]. However, because SDS ⁄ PAGE analysis showed that the resulting protein was contaminated by low molecular mass proteins (Fig. 1, compare lanes 5– 7) SF was further purified by precipitation [3]. Several low molecular mass contaminants were removed by this procedure (Fig. 1, lanes 5–7), and the protein could be further concentrated. The purified SF was stored in the freezer, at a concentration of 1.0 mgÆmL)1 in 20% (v ⁄ v) glycerol, without forming a pre- cipitate or loss of activity. This is in contrast to the results presented by Wendrich et al. [7] where purified recombinant SF was found to precipitate at protein concentrations exceeding 0.15 mgÆmL)1.

Early on it was shown that unacylated tRNA in the ribosomal amino-acyl site (A-site) stimulates pppGpp synthesis by SF [5,6] and the general belief is that un- acylated tRNA in the ribosomal A-site is required for the activation of SF by ribosomes [5–7]. It is unclear how unacylated tRNA enters the ribosomal A-site. At least two options are possible: (a) it could enter the A-site by simple diffusion; or (b) interaction of SF with unacylated tRNA could increase the affinity of the tRNA for the ribosomal A-site as originally sug- gested by Richter [8]. Activity of the recombinant protein in the complete system

independent synthesis pppGpp and

The activity of recombinant SF was first measured in a complete system containing twice salt-washed tight- couple ribosomes (TC-ribosomes), poly(U), tRNAPhe, radiolabelled GTP and unlabelled substrates. In this system (at 15 mm MgCl2) unacylated tRNAPhe may bind to all of the tRNA binding sites on the ribosome (i.e. A, P and E sites) [13,14]. As expected, the addi- tion of SF to the system resulted in pppGpp synthesis (Fig. 2). The speed of synthesis, calculated as des- cribed in Experimental procedures, amounted to 4800 pmol pppGppÆpmol SF)1Æmin)1 at the five-minute It has also been shown that ribosomal protein L11 is a stimulator of pppGpp synthesis in the ribosome- dependent reaction [7,9,10]. Moreover, the A-site-speci- fic antibiotic thiostrepton, which is dependent on L11 for ribosome-binding [11], inhibits ribosome-dependent pppGpp synthesis in vitro [2]. Another A-site-specific antibiotic, tetracycline, inhibits both ribosome-depend- ent [6,12]. Altogether this shows that the function of SF is closely connected to the ribosomal A-site, unacylated tRNA and ribosomal protein L11 (see for example [7]) but it is unclear how SF interacts with ribosomes and unacylated tRNA and how pppGpp synthesis is stimulated.

We have isolated a recombinant His-tagged SF by affinity-purification and by taking advantage of the natural ability of the protein to form a precipitate [3]. The purified protein is highly active in a complete sys- tem containing ribosomes, poly(U) and unacylated tRNAPhe and converts approximately 4500 pmol GTP to pppGppÆpmol SF)1Æmin)1. Here, the components that are needed for pppGpp synthesis by SF are sys- tematically mapped.

Results and Discussion

Stringent factor (SF) is a ribosome-dependent ATP:GTP pyrophosphoryl transferase that is encoded by the relA locus in Escherichia coli. We have cloned and purified SF from E. coli and examined the ribosome, tem- plate and tRNA-dependence of the pppGpp synthesis reaction.

Fig. 1. SDS ⁄ PAGE showing the purification of recombinant His- tagged SF, indicated by the arrow. The cell extract containing over- expressed SF (lane 1) was incubated with Ni-NTA agarose beads, unbound protein was removed (lane 2) and the beads washed (lanes 3, 4). SF was eluted with imidazole (lane 5) and precipitated by dialysis against low salt ⁄ high magnesium buffer. The precipita- ted protein was dissolved in high salt buffer and dialyzed into low salt buffer (lane 7). Lane 6 shows protein contaminants that did not precipitate and lane M contains protein markers. See Experimental procedures for more details.

Purification of recombinant SF

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

686

We started out by purifying SF by affinity-chromato- graphy using a His tag at the C-terminal end of the

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

Table 1. Characterization of SF-activity in the poly(U)-dependent system. Samples containing TC-ribosomes, poly(U) and tRNAPhe, as indicated in the table, were incubated for 10 min at 37 (cid:1)C before addition of SF and nucleotides. Incubation was for 5 min (complete system) or 20 min at 37 (cid:1)C. Nucleotides were separ- ated as described in the legend to Fig. 2 and the activity was calculated as described in Experimental procedures. The values are based on three independent experiments.

Activity (% of max)

Ribosome (1.7 lM)

Poly(U) (0.16 lgÆlL)1)

tRNAPhe (10 lM)

SF (0.2 lM)

Activity (pmol pppGppÆpmol SF)1Æmin)1)

+

+

+

+

+ + + +

+

+ + + + + + +

+ +

28 ± 17 163 ± 33 169 ± 57 170 ± 72 332 ± 124 327 ± 185 342 ± 162 4461 ± 475

0 3.5 3.6 3.6 7.1 7.0 7.4 100

Fig. 2. A time course of pppGpp synthesis by SF in the ribosome- dependent reaction. Ribosomal complexes were formed by incuba- ting TC-ribosomes (1.7 lM) with poly(U) (0.16 lgÆlL)1) and tRNAPhe (10 lM) for 10 min at 37 (cid:1)C. Radiolabelled GTP (0.6 lCi) was added to the samples together with unlabeled substrates (10 mM) and SF (0.2 lM). Samples were precipitated, with formic acid, at the indica- ted times, and spotted on TLC plates to separate the nucleotides. The pppGpp synthesis speeds (pmol pppGppÆpmol SF)1Æmin)1, m) and available substrate concentrations (j) were calculated as des- cribed in Experimental procedures and plotted as a function of time. The input of radioactive GTP in the reactions is indicated by the asterisk. See Experimental procedures for more details.

time-point. After that speeds dropped almost linearly with time (Fig. 2, triangles). The drop in synthesis speeds was caused by a shortage of substrate in the reaction mixtures (Fig. 2, squares). It is possible that the varying activity of SF reported in the literature ([3] and references therein) may in part be caused by limiting supplies of nucleotides in the reaction mix- tures, as the specific activity has often been measured after long incubation times when nucleotides should be limiting.

Fig. 3. A-site specific antibiotics inhibit the ribosome-dependent activity in the absence of added tRNA. Autoradiograms showing the inhibitory effects of viomycin, 0.1 mM (lane 3), 1 mM (lane 4), 10 mM (lane 5); tetracycline, (0.5 mM, lane 9); thiostrepton (10 lM, lane 10); and micrococcin (10 lM, lane 11) on TC-ribosome-dependent pppGpp synthesis. Antibiotics were omitted from samples 2 and 8. Endo- genous activity of SF (lanes 6, 14) in the presence of 10 mM viomycin (lane 7) and 0.5 mM tetracycline (lane 15). Ribosomes were incuba- ted with antibiotics before the addition of SF and nucleotides. Incuba- tion was for 30 min at 37 (cid:1)C. See Fig. 2 legend for more details.

Endogenous activity of SF

Ribosome-dependence of SF in the complete system

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

687

is difficult [7]. It Table 1 shows that the addition of template, unacy- lated tRNA and ribosomes to the activity assay stimu- lated SF 30-fold when using optimal conditions. When the ribosome-dependence was investigated more thor- oughly it was found that SF activity increased with increasing ribosome concentrations until a plateau was reached at a 10-fold excess of ribosomes over SF (results not shown). This is in accordance with results presented by Wendrich et al. to understand the biological significance of the ribosome The activity of SF in the presence of different transla- tional components is summarized in Table 1. It is shown that purified SF produced low amounts of pppGpp, amounting to 150 pmol pppGppÆpmol SF)1Æmin)1, in the presence of buffer and nucleotides. This synthesis is visible in the autoradiogram in Fig. 3 (lane 14). That SF has low synthetase ability in the absence of alcohol and ribosomes is in accordance with an earlier report [3]. The endogenous activity of SF was not affected by the addition of template or unacylated tRNA (Table 1) but was inhibited by the antibiotic tetracycline (Fig. 3, lane 15). Similarly, the alcohol-activated assay is inhibited by tetracycline [12]. The mechanism for this inhibition is unknown [1].

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

titration curve as ribosomes should always be in molar excess of SF [3,15,16] far exceeding the 10 : 1 ratio that gives maximal synthesis speeds in the in vitro assay (results not shown; [7]).

However, a model has been proposed to explain how a few SF molecules can trigger the stringent response within a few minutes on a large population of ribosomes. In this model, SF molecules are sugges- ted to hop between different stalled ribosomal com- plexes and initiate pppGpp synthesis [7].

Ribosome-dependent but tRNA-independent pppGpp synthesis?

Fig. 4. Ribosomes stimulate pppGpp synthesis by SF in the absence of added tRNA. TC-ribosomes (1.7 lM, lanes 2, 4) or 50S (1.7 lM) and 30S (2.3 lM) particles reassociated to 70S ribosomes (lane 3, 5) were incubated with SF (0.3 lM). Poly(U) (0.16 lgÆlL)1) and tRNA (10 lM) were added to samples 4 and 5 before the addi- tion of SF. Endogenous activity of SF (lane 6). Incubation was for 10 min at 37 (cid:1)C. See Fig. 2 legend for more details.

Unacylated tRNA is incapable of stimulating SF in the absence of ribosomes (Table 1) but do ribosomes have an intrinsic ability of stimulating pppGpp synthe- sis? This might have been overlooked in some earlier studies where the activity of SF was 10-fold lower than in the experiments presented here.

The stimulatory activity of ribosomes that had been prepared by two different methods was analyzed: (a) TC-ribosomes that are competent in binding 95% tRNA in the ribosomal A-site [17]; and (b) ribosomes that have been reassociated from subunits. The sub- units have been exposed to low magnesium concentra- tions during preparation [18] to dissociate ribosomes with concomitant release of tRNA [5].

of pppGpp visible in Figs 3 and 4 and that this ribo- some-dependent activity is inhibited by antibiotics that target the ribosomal A-site and ⁄ or protein L11. However, reassociated ribosomes

low levels it cannot be excluded that ((cid:1) 5%) of contaminating tRNAs in the ribosome pre- paration caused the stimulatory effect. Here, it should also be mentioned that in the complete system, reasso- ciated ribosomes were 30% less efficient in stimulating pppGpp synthesis than TC-ribosomes (Fig. 4, lanes 4–5). Similarly, reassociated ribosomes are not as com- petent in binding tRNA [25] as TC-ribosomes [17]. Therefore, it appears that extensive purification of ribosomes impairs the tRNA-binding and pppGpp synthesis stimulating activity of ribosomes. The activity of SF was found to increase threefold lane 2; in the presence of TC ribosomes (Fig. 4, 450 pmol pppGppÆpmol SF)1Æmin)1) and twofold in the presence of (lane 3, 300 pmol pppGppÆpmol SF)1Æmin)1) compared to the endogenous activity of the enzyme (lane 6). Moreover, the TC-dependent activity was inhibited by antibiotics that target pppGpp synthesis and ⁄ or the ribosomal A-site (Fig. 3). Thus, viomycin (lanes 3–5), tetracycline (lane 9), thiostrepton (lane 10) and micrococcin (lane 11) inhibited pppGpp synthesis compared to control reactions carried out in the absence of antibiotics (lanes 2 and 8). Thiostrepton and micrococcin probably

Template-dependence of pppGpp-synthesis

inhibit pppGpp synthesis by blocking the function of L11 [7,9,19,20], whereas viomycin and tetracycline interfere with A-site related functions [20–24]. As mentioned earlier, tetracycline also inhibited SF in the absence of ribosomes (Fig. 3, lanes 14–15; [1]), whereas the other antibiotics did not inhibit this activity (Fig. 3, lanes 6–7; and results not shown). Table 1 shows that if SF is incubated with ribosomes, unacylated tRNA and nucleotides, but no template, the activity of the enzyme is similar to that in the absence of tRNA. This is not surprising because the template is needed to position tRNAs on the ribosome [26].

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

688

The activity of the poly(U)-dependent system was compared with a more natural system containing a gene T4 mRNA fragment (from now on referred to as It is known that stringent factor forms a stable com- plex with ribosomes in the absence of unacylated tRNA [7,8]. We speculate that formation of such com- plexes stabilises SF and leads to the small production

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

f

f

f

f

f

f

tRNAMet

twofold excess of tRNAMet , there was a clear DMS footprint at the E-site specific base C2394 in 23S rRNA (Fig. 6A, red line) compared to samples con- taining no tRNA (blue line). However, further analysis revealed that this footprint disappeared when ribo- somes were incubated with equimolar amounts of unacylated tRNAMet red line). Moreover, (Fig. 6B, -containing ribo- chemical modification of somal complexes with kethoxal showed that strong footprints were seen in the so called P-loop in 23S rRNA at nucleotides G2252 and G2253 (Fig. 6C, red line [28]). Altogether, this suggested that the ribosomal complexes that interacted with SF in the activity assay contained one tRNAMet that was bound in the P-site of 50S subunits and one tRNAMet that was bound in the E-site of 50S subunits. The P-site bound tRNAMet also gave clear DMS footprints on bases A794 and C795 in 16S rRNA that have been linked to the ribosomal P-site of the 30S subunit (results not shown; [28]).

f

Addition of a threefold excess of tRNAPhe to the programmed ribosomes resulted in a strong tRNAMet

A

f

f

B

Fig. 5. Autoradiograms showing the template and tRNA-depend- ence of the pppGpp-synthesis reaction. The activity of TC-ribo- somes (0.67 lM, lane 2) containing T4-mRNA (1.3 lM, lane 3) plus tRNAMet (1.3 lM, lane 4) plus tRNAPhe (6.7 lM, lane 5). Comparison of activity of the poly(U) (2.45 lg, lane 6) and T4-mRNA (4 lM, lane 7) dependent systems. Ribosomal complexes were formed by incu- bating TC-ribosomes with T4-mRNA and tRNAMet for 10 min at 37 (cid:1)C. tRNAPhe was added and incubation continued for 10 min. SF was added to the reactions and samples were taken at 10 (lanes 2–5) and 5 (lanes 6, 7) min. See Fig. 2 legend for more details.

C

D

T4-mRNA [27]). The stimulatory activity of the T4-mRNA programmed ribosomes was only 60% of the poly(U)-programmed ribosomes in the presence of added tRNA (Fig. 5, lanes 6–7). These results may be explained by recent data where SF was found to inter- act more strongly with poly(U) and full-length mRNAs than with short mRNA fragments [7].

tRNA-dependence of pppGpp synthesis

E

Binding states of tRNAs

f

f

complexes in ribosomal that

f

Fig. 6. Footprinting analysis of the interaction of tRNAMet (A, B, C) and tRNAPhe (D, E) with the T4-mRNA programmed TC-ribosomes. Complexes were formed with twofold excess (A, C, D, E) or equi- molar amounts of tRNAMet (B) before addition of threefold excess of tRNAPhe (D, E). DMS (A, B, E) and kethoxal (C, D) modifications and primer extensions were performed according to Experimental (black line), samples without procedures. Mock-modified control added tRNA (blue lines) and samples containing tRNAs (red lines). See Fig. 5 legend for more details.

It was important to determine the binding states of tRNAs stimulated pppGpp synthesis by SF as tRNAs could either be bound in classical or hybrid states [28]. First, it was decided to monitor the state of the peptidyl site (P-site) bound tRNA as this state determines the state of the A-site bound tRNA. Ribosomal complexes, containing tRNAMet

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

689

, were footprinted with dimethylsulfate (DMS) and kethoxal at 15 mm MgCl2. Primer extension analysis of the T4-mRNA programmed ribosomes showed that, at a

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

f

f

f

f

A-site footprint at G530 in 16S rRNA (Fig. 6D, red line [28]) and A-site footprints in domain IV in 23S rRNA at nucleotides C1941 and C1942 (Fig. 6E, red line [29]). A footprint was also seen at position A1966 by binding of tRNAMet to the P-site of 50S subunits (visible in Fig. 6E).

From these results it can be concluded that tRNAs were bound in classical states with the tRNAMet bound in the P-sites of the 30S and 50S subunits and the tRNAPhe bound in the A-site of the 30S and 50S sub- unit [28]. Moreover, a tRNAMet was present in the 50S E-site. The E-site bound tRNA did not affect the stim- ulatory activity of ribosomes as ribosomal complexes formed with 1.2-fold or twofold excess of tRNAMet sti- mulated SF to the same extent upon addition of tRNAPhe to the activity assay (results not shown).

f

f

Binding of tRNAMet to ribosomes

Binding of tRNAMet to T4-mRNA programmed ribo- somes did not increase the ability of ribosomes to sti- lane 4). This is in mulate pppGpp-synthesis (Fig. 5, agreement with other data [5,7] and thus supports the notion that unacylated tRNA has to be bound in the ribosomal A-site for activation of SF to occur.

f

Titration of tRNAphe to the A-site of T4-mRNA programmed ribosomes

the activity of SF, increased

containing T4-mRNA-programmed TC-ribosomes, tRNAMet , were incubated with increasing amounts of tRNAPhe before the addition of SF. As can be seen in calculated as pmol Fig. 7A, ribosome)1Æmin)1, with pppGppÆpmol increasing amounts of tRNAPhe added until a plateau was seen at a threefold molar excess of tRNAPhe over ribosomes (Fig. 7A). This shows that the ribosomal A-site was saturated with unacylated tRNA [14] when maximal activation of SF occurred.

f

Fig. 7. tRNA-titration curves showing the tRNA-dependence of the pppGpp synthesis reaction. pppGpp synthesis speeds (pmol pppGppÆpmol ribosome)1Æmin)1) were plotted as a function of dif- ferent tRNA ⁄ ribosome ratios ⁄ TC-ribosomes (0.67 lM) programmed or (B, C) poly(U). (A, B) tRNAPhe was with (A) T4-mRNA + tRNAMet added, at the indicated concentrations, and incubation continued for 10 min at 37 (cid:1)C. SF was added at two different concentrations ; or 0.67 lM, m) together with tRNAPhe (C) and the incu- (0.13 lM, bation continued for 5 (A) or 10 (B, C) min at 37 (cid:1)C. Nucleotides were separated and speeds (pmol pppGppÆpmol ribosome)1Æmin)1) calculated as described in Experimental procedures. The curves are based on at least three independent experiments. Refer to the leg- ends of Figs 2 and 6 for more details.

Binding of tRNAPhe to the A, P and E-sites of poly(U) programmed ribosomes

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

690

In the second set of experiments poly(U) programmed ribosomes were incubated with increasing amounts of tRNAPhe before addition of SF. In Fig. 7B it can be seen that the tRNA-binding curves reached a plateau at a fivefold to 10-fold molar excess of tRNAPhe over in this system, more tRNAPhe was ribosomes. Thus, required to get maximal SF activity. This is not surpri- sing because tRNAPhe will bind to all three tRNA bind- ing sites on poly(U) programmed ribosomes [14] and two molar equivalents of tRNA are needed to saturate

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

the tRNA saturation curves might be affected by the amount of SF present in the reaction. Therefore, the P- and E-sites before A-site binding [13]. It can also be concluded from this experiment that maximal SF-activation required that the A-site was saturated with unacylated tRNA.

Is the tRNA-stimulatory effect affected by the order of addition of tRNA and SF to the reactions?

the tRNA-saturation curves were per- formed at two different ribosome ⁄ SF ratios: first, a fivefold molar excess of ribosomes (Fig. 7, triangles); or second, equimolar amounts of ribosomes and SF (Fig. 7, rectangles). (The concentration of ribosomes was constant in the experiments whereas the SF con- centration varied.)

the way that

The results show that maximal activity of SF was reached at similar tRNA levels independent of the ribo- some ⁄ SF level (Fig. 7). This was true for both the T4- mRNA-dependent system and the poly(U)-dependent systems. Therefore, this experiment does not support the notion that SF should form a complex with tRNA in solution before binding to ribosomes because similar amounts of tRNA were needed independent of the SF concentration. The higher activity of the systems con- taining more SF (1 : 1 ratio between SF and ribosome) may be attributed to the endogenous activity of SF (150 pmol pppGppÆpmol ribosome)1Æmin)1).

Does SF form a complex with unacylated tRNA in solution?

The two types of tRNA titration experiments showed that ribosomes must be saturated with tRNAPhe for maximal activation of SF to occur. A-site bound unacylated tRNAPhe should be stably bound in the experiments presented here as the half-life of dissoci- ation is more than 2 h [14]. The strong footprint at G530 in 16S rRNA supports this notion (Fig. 6E). This can be compared to the weak A-site binding needed for maximal SF activation in the system used by Wendrich et al. [7]. Curiously, there was one dif- ference it the experiments were per- formed, as in that system recombinant SF was added together with unacylated tRNAPhe to ribosomes, whereas in our system ribosomes were preincubated with unacylated tRNAPhe before the addition of SF. Is it possible that less tRNA is needed to reach maximal activation of SF by adding SF and tRNA together to ribosomes?

concentrations

We also tried to isolate a complex between tRNA and SF in solution by filter-binding assays. In this assay, SF was incubated with tritium-labelled unacylated tRNA in a buffer containing 20 mm MgCl2 and vary- (Table 2). Reactions were ing salt chilled on ice before filtration through a 0.45 lm Milli- pore filter. The results show that approximately 10-fold more tRNA was retained on the filter in the presence of SF (Table 2). The effect was specific to SF, as incu- bation of unacylated tRNA with recombinant ribo- somal protein L10 or ovalbumin did not increase the amount of tRNA retained on the filter (results not shown). Binding was slightly more efficient at lower (50 mm) than higher (100 mm) salt concentrations.

Table 2. Binding of unacylated tRNA to SF in solution. SF-tRNA complexes were formed as described in Experimental procedures and complexes were separated from unbound tRNA by filter- binding. The values are dependent on at least three independent experiments.

To investigate this, it was tested whether the tRNA- saturation curve would behave differently by adding SF and tRNAPhe simultaneously to poly(U)-pro- grammed ribosomes. Surprisingly, the experiments sug- gest that this prediction is true. The curves in Fig. 7C show that saturation was reached at a two- to threefold molar excess of tRNAPhe over ribosomes when SF was added together with tRNAPhe to the activity assay. In this last experiment the ribosomal A-site would not be saturated with tRNAPhe as P- and E-site binding pre- cedes A-site binding [13] and two molar equivalents of tRNAPhe are needed to saturate the ribosomal P- and E-sites. The experiments suggest therefore that only weak binding of unacylated tRNAPhe in the ribosomal A-site was needed when SF and tRNA were added sim- ultaneously to the reaction mixtures, in agreement with the results presented by Wendrich et al. [7].

Salt [KCl]

tRNA (pmol)

SF (pmol)

Binding (nCi)

tRNA bound (%)

100 100 100 50 50 50

60 30 60 60 30 60

– 20 20 – 20 20

777 ± 247 4019 ± 653 3849 ± 566 756 ± 127 5736 ± 770 5774 ± 628

0.3 3.4 1.6 0.3 4.8 2.4

The stimulatory effect of tRNA was not affected by two different ribosome ⁄ SF ratios tested

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

691

We found it intriguing that the order of addition of tRNA and SF to the activity assay affected the tRNA- saturation curve. This suggested that SF might form a complex with unacylated tRNA in solution that has higher affinity for ribosomes than unacylated tRNA by itself, as originally suggested by Richter [8]. If so,

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

complex when they discovered the stimulatory effect of adding unacylated tRNA to the ribosomal A-site. It would be interesting to compare the kinetics of SF in the physiological system with the system used here, con- taining only unacylated tRNAs. Here, it should also be mentioned that it has been suggested that the 50S sub- unit may contain a domain that senses the aminoacyla- tion state of the tRNA in analogy with the T-box in antitermination of transcription of amino acid biosyn- thetic enzymes [1,34]. We suggest that a putative T-box on the 50S subunit would be part of the 50S A-site, as unacylated tRNA is required for stimulation of SF by ribosomes and this tRNA sits in the 50S A-site.

The ribosome-dependence of SF has been known since the factor was first isolated more than 30 years ago. Despite this fact, there are still big gaps in our knowledge of how SF interacts with the ribosome and which ribosomal components are essential for the acti- vation of pppGpp synthesis. This might be partly due to the fact that SF is present in very low amounts in the cell [3,15] and has therefore been hard to purify. During the last few years, several different recombinant pyro- phosphoryl transferase have been cloned and isolated ([7,35]; this study). Work with the recombinant proteins have elucidated the endogenous activity of SF ([35]; this study), and how SF interacts with ribosomes ([7,35]; this study). Future experiments will reveal how SF binds to and is activated by ribosomes and unacylated tRNA.

Experimental procedures

f

f

Materials

reverse transcriptase was

tRNAMet was from Boehringer (Ingelheim, Germany) and tRNAPhe, poly(U), ATP, GTP, isopropyl thio-b-d-galacto- side and polyethyleneimine plates (Macherey & Nagel, Du¨ ren, Germany) were from Sigma-Aldrich (St. Louis, MO, USA). 3H-labelled acylated tRNA was prepared and stripped of amino acid according to [36]. The phage T4 gene 32 mRNA fragment was from Dharmacon (Lafayette, CO, USA). The sequence of the fragment is according to [27]. [32P]dGTP[aP] (10 mCiÆmL)1) and Hyperfilm MP was from Amersham Bioscience (Buckinghamshire, UK). DMS was from Sigma, kethoxal was from ICN (Irvine, CA, USA), Superscript from Life Technologies, Inc. (Rockville, MD, USA) and the DNA sequencing kit was from PerkinElmer (Boston, MA, USA).

However, the amount of tRNA retained was low compared to amounts of SF and total tRNA present in the reaction (Table 2). Moreover, similar levels of tRNA were bound at the two concentrations of tRNA tested. It is difficult to estimate the significance of the data because it cannot be ruled out that tRNA was trapped on filters through a nonspecific interaction with SF. Of course, it is possible that SF formed a labile complex with tRNA in solution and that this complex was prone to dissociate upon dilution of the reactions before filtration, in analogy with the weak interaction of unacylated tRNA with the ribosomal E-site [30]. We are currently investigating this theory more thoroughly. Levels of unacylated tRNA are increased during the stringent response (reviewed in [1]). Experimental data suggest that unacylated tRNA interacts with the 30S A-site in vivo [31] but it is not understood how the tRNA is directed to the A-site in the cell. The data presented here does not support the hypothesis that SF directs unacylated tRNA to the ribosomal A-site for the following reasons: the tRNA-saturation curves were independent of the two ribosome ⁄ SF ratios tes- ted, and we were unable to isolate significant amounts of a putative tRNA–SF complex by filter-binding assays. In vitro, it is easy to manipulate binding of unacylated tRNA to the ribosome by increasing the magnesium concentration [14]. The magnesium concentration also affects the binding states of the tRNAs on the ribosome. Unacylated tRNAs can be bound in either classical or hybrid states [28]. We have shown here by footprinting analysis that in our buffer system at 15 mm MgCl2 SF interacted with ribosomes that contain tRNAs bound in classical states. This means that the 50S A- and P-sites interacted with the 3¢ end of unacylated tRNAPhe and tRNAMet , respectively. Thus, in this study, SF was acti- vated by unacylated tRNAPhe that sits in the 50S A-site. In contrast, if the tRNAs had been bound in hybrid states the CCA end of unacylated tRNAPhe would have been bound in the 50S P-site and the 50S A-site would have been empty [28].

The relA gene was amplified by PCR from E. coli MRE 600 genomic DNA with primers 5¢-CGGGAATTCCATATGGT TGCGGTAAGAAT-3¢ and 5¢-CCCGCTCGAGACTCCCG

Cloning of the E. coli relA gene

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

692

Most SF activity studies have been performed at 10– 20 mm Mg2+ [2,6,8,15,32,33] although Wendrich et al. [7] performed their studies at 6 mm Mg2+ with addi- tional spermine and spermidine. It is therefore imposs- ible to say whether tRNAs were bound in similar states in all of the above studies. In the cell, SF probably binds to a ribosome with a peptide in the P-site (P ⁄ P- state) and an unacylated tRNA in the A-site. This unac- ylated tRNA must therefore bind in the A ⁄ A-state (in analogy with this study) although, to our knowledge, the interaction of unacylated tRNA with ribosomes that are filled with peptide has not been structurally mapped. Haseltine and Block [5] used this type of ribosomal

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

purity of ribosomes was checked by denaturing gels con- taining 8 m urea and the activity of ribosomes was tested in poly(Phe) synthesis assays according to [38].

the primers

In

the

standard

TGCAACCGACG-3¢ containing NdeI and XhoI recognition sequences, respectively, and inserted into a TOPO-vector (Invitrogen, Carlsbad, CA, USA). Afterwards, the gene was subcloned into the pET24b vector to generate pET24b(relA). The correct sequence of relA was confirmed by sequencing using 5¢-AGCAATACGCTCCGCCAG-3¢, 5¢-TGGCGGATGCCAACGTAG-3¢, 5¢-CTCGACCGCGA ACACTAC-3¢, 5¢-CACCCAACTCTGCATCTTC-3¢, 5¢-TT TCGAACGCCCACGGC-3¢ and 5¢-TGTACTGAAATACC GCGCC-3¢.

pppGpp synthesis

f

10 mm 2-mercaptoethanol,

incubated with Ni-NTA agarose beads

pppGpp synthesis assays were essentially carried out according to Haseltine and Block [5] with the following assay TC-ribosomes modifications. (25 pmol) were programmed with poly(U) (2.45 lg) and tRNAPhe (150 pmol) in a buffer containing 20 mm MgCl2, 100 mm KCl, 30 mm Hepes pH 8.0, 10 mm 2-mercaptoeth- anol for 10 min at 37 (cid:1)C. A mixture of ATP and GTP (10 mm final concentrations) and [32P]GTP[aP] (0.6 lCi) was added to the reactions and the MgCl2 concentration was adjusted to 15 mm. SF (5 pmol, unless otherwise indi- cated) was added to the reaction mixtures (total volume 15 lL) and incubation was continued for the indicated and tRNAPhe were times at 37 (cid:1)C. T4-mRNA, tRNAMet used at the concentrations indicated in the figures. In some assays TC-ribosomes were preincubated with the antibiotics tetracycline (500 lm), micrococcin (10 lm), thiostrepton (10 lm) and viomycin (0.1–10 mm) before addition of nucleotides and SF. The reactions were stopped by the addition of 1 lL 88% (v ⁄ v) formic acid, incubated on ice for 15 min and centrifuged for 5 min at 16 000 g in an epp- endorf centrifuge at 4 (cid:1)C.

Expression and purification of stringent factor

SF was purified from BL21(DE3) cells, grown in 2· YT medium, containing the pET24b(relA) plasmid. Protein expression was induced with 0.5 mm isopropyl thio-b-d- galactoside, at D550 ¼ 0.7, for 4 h at 30 (cid:1)C. Cells were har- vested by centrifugation (11 000 g, 16 min, 4 (cid:1)C) and washed with 0.1 m NaCl, 10 mm Tris ⁄ HCl pH 8.0, 1 mm EDTA. The cell pellet was dissolved in lysis buffer: 50 mm NaH2PO4, 300 mm NaCl, 10 mm imidazole, 10% (v ⁄ v) gly- pH 8.0. Lysosyme cerol, (1 mgÆmL)1) was added and the cell suspension was left on ice for 30 min before sonication (10 · 15 s, 2 · 30 s, 1 min between cycles). Cell debris was removed by two centrifuga- tions for 15 min at 23 000 g, 4 (cid:1)C. The cleared cell lysate (Qiagen, was Valencia, CA, USA) for 1 h at 4 (cid:1)C and washed four times with lysis buffer containing 1 m NaCl and 20 mm imidazole. The suspension was transferred to a column and SF was elut- ed in 0.5 mL fractions with lysis buffer containing 250 mm imidazole. Fractions containing protein were dialyzed over- night against 10 mm Tris ⁄ HCl pH 8.0, 14 mm MgOAc, 60 mm KOAc, 0.5 mm EDTA, 10% (v ⁄ v) glycerol, 10 mm 2-mercaptoethanol. Using these conditions SF formed a pre- cipitate [3]. The precipitate was dissolved in 10 mm Tris ⁄ HCl pH 8.0, 1 m KCl, 1 mm EDTA, 10% (v ⁄ v) glycerol, 10 mm 2-mercaptoethanol and dialyzed overnight against SF buffer: 30 mm Hepes pH 8.0, 300 mm KCl, 20% (v ⁄ v) glycerol and 10 mm 2-mercaptoethanol. The protein concentration was determined according to Bradford and aliquots of the protein were quick-frozen and stored at )80 (cid:1)C. The His tag did not appear to interfere with the activity of the protein, as the recombinant SF was highly active in accordance with previ- ous results [7].

Radiolabelled nucleotides were separated by thin layer chromatography. Supernatants (10 lL) were spotted on polyethyleneimine cellulose plates and the nucleotides were allowed to migrate using 1.5 m KH2PO4 (pH 3.4) as a buf- fer. The radioactive spots corresponding to GTP and pppGpp were identified by autoradiography using a phos- phorimager or Hyperfilm MP. The amounts of pppGpp synthesized were quantified by phosphorimager analysis or by counting radioactivity in a liquid scintillation counter after the spots were cut out. Turnover rates were calculated as percent of radioactive (p)ppGpp of the total amount of radioactivity. This was then normalized in relation to the time and amount of SF ⁄ ribosome used, yielding the rate pmol pppGppÆpmol SF)1Æmin)1 or pmol pppGppÆpmol ribosome)1Æmin)1.

Separation of pppGpp from GTP and calculation of synthesis speeds

Purification of ribosomes

f

tRNAMet TC-ribosomal complexes containing T4-mRNA, and tRNAPhe were modified with DMS according to [39]. Alternatively, samples were modified with kethoxal (18 mm for 15 min at 37 (cid:1)C. The kethoxal final concentration)

Tight-couple ribosomes from E. coli strain MRE600 were purified according to [17], except that cells were lysed by sonication (6 · 15 s; 2 · 20 s, 1 min between cycles). Ribo- somes were suspended in 20 mm Tris ⁄ HCl (pH 7.6), 10 mm MgCl2, 100 mm NH4Cl, 0.5 mm EDTA, 6 mm 2-mercapto- ethanol and stored in small aliquots at )80 (cid:1)C. Ribosomal 30S and 50S subunits were purified according to [37]. The

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

693

Chemical modification and primer extension analysis

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

ribosomal wash of stringent Escherichia coli. Proc Natl Acad Sci USA 70, 2145–2148.

5 Haseltine WA & Block R (1973) Synthesis of guanosine tetra- and penta phosphate requires the presence of a codon-specific, uncharged transfer ribonucleic acid in the acceptor site of ribosomes. Proc Natl Acad Sci USA 70, 1564–1568.

6 Pedersen FS, Lund E & Kjeldgaard NO (1973) Codon specific tRNA dependent in vitro synthesis of ppGpp and pppGpp. Nat New Biol 243, 13–15.

7 Wendrich TM, Blaha G, Wilson DN, Marahiel MA & Nierhaus KN (2002) Dissection of the mechanism for the stringent factor RelA. Mol Cell 10, 779–788.

adduct was stabilized with 25 mm K-borate (pH 7.0) at all times. RNA was precipitated with ethanol and extracted from protein according to [39]. The positions of the modi- fied sites were identified by primer extension according to [40]. The primers used were according to [28] except that a fluorescent label was included at the 5¢ end of the probe. The following primers were used: 5¢-CCGAACTGTCT CACGAC-3¢ (906, 16S rRNA), 5¢-TGTTATCCCCGGAG TAC-3¢ (2437, 23S rRNA), 5¢-GCATTTCACCGCTACAC-3¢ (683, 16S rRNA) and 5¢-TCCGTCTTGCCGCGGGT-3¢ (2042, 23S rRNA). The primer extension products were analyzed on 5% (w ⁄ v) polyacrylamide sequencing gels in an Applied Biosystems 377 DNA sequencer as described previously [40].

8 Richter D (1976) Stringent factor from Escherichia coli directs ribosomal binding and release of uncharged tRNA. Proc Natl Acad Sci USA 73, 707–711.

9 Friesen JD, Fiil NP, Parker JM & Haseltine AW (1974)

A new relaxed mutant of Escherichia coli with an altered 50S ribosomal subunit. Proc Natl Acad Sci USA 71, 3465–3469.

10 Yang X & Ishiguro EE (2001) Involvement of the N ter- minus of ribosomal protein L11 in regulation of the RelA protein of Escherichia coli. J Bacteriol 183, 6532– 6537.

11 Thompson J, Cundliffe E & Stark M (1979) Binding of thiostrepton to a complex of 23S rRNA with ribosomal protein L11. Eur J Biochem 98, 261–265.

SF (20 pmol) was incubated with 3H-labelled unacylated tRNA (30 or 60 pmol; specific activity 1.7 nCiÆpmol)1) in a buffer containing 50 mm KCl, 20 mm MgOAc, 30 mm He- pes pH 7.8, 0.5 mm EDTA and 6 mm 2-mercaptoethanol for 10 min at 30 (cid:1)C. The final volume of the reactions was 10 lL. Reactions were cooled on ice for 10 min, diluted to 1.5 mL with the same buffer and filtered through a 0.45 lm Millipore filter. Filters were washed with 3 · 1.5 mL buffer and the samples were counted in a liquid scintillation counter.

Filter-binding assays

Acknowledgements

12 Silverman RH & Atherly AG (1978) Unusual effects of 5a,6-anhydrotetracycline and other tetracyclines. Inhibi- tion of guanosine 5¢-diphosphate 3¢-diphosphate meta- bolism, RNA accumulation and other growth-related processes in Escherichia coli. Biochim Biophys Acta 518, 267–276.

13 Rheinberger HJ, Sternbach H & Nierhaus KH (1981)

Three tRNA binding sites on Escherichia coli ribosomes. Proc Natl Acad Sci USA 78, 5310–5314.

14 Lill R, Robertson JM & Wintermeyer W (1986) Affi-

nities of tRNA binding sites of ribosomes from Escheri- chia coli. Biochemistry 25, 3245–3255.

Odd Nyga˚ rd is thanked for critical reading of the manuscript and general support. Ma˚ ns Ehrenberg is thanked for helpful discussions and Harry Noller is thanked for insightful comments. This research is sup- ported by a grant from the Swedish Research Council (Dnr 5 ⁄ 42 ⁄ 2001).

References

15 Justesen J, Lund T, Pedersen FS & Kjeldgaard NO

1 Cashel M, Gentry DR, Hernandez VJ & Vinella D

(1986) The physiology of stringent factor (ATP:GTP 3¢ diphosphotransferase) in Escherichia coli. Biochimie 68, 715–722.

(1996) The stringent response. In Esherichia coli and Salmonella: Cellular and Molecular Biology (Neider- hardt FC, Curtis R, Ingraham JL, Lin ECC, Low KB, Magasanik B, Reznikoff WS, Riley M, Schaech- ter M & Umbarger HE, eds), pp. 1458–1496. ASM Press, Washington DC.

16 Howe JG & Hershey WB (1983) Initiation factor and ribosome levels are coordinately controlled in Escheri- chia coli growing at different rates. J Biol Chem 258, 1954–1959.

2 Haseltine WA, Block R, Gilbert W & Weber K (1972)

MSI and MSII made on ribosome in idling step of pro- tein synthesis. Nature 238, 381–384.

3 Pedersen FS & Kjeldgaard NO (1977) Analysis of the

relA gene product of Escherichia coli. Eur J Biochem 76, 91–97.

17 Lancaster L, Kiel MC, Kaji A & Noller HF (2002) Orien- tation of ribosome recycling factor in the ribosome from directed hydroxyl radical probing. Cell 111, 129–140. 18 Moazed D, Stern S & Noller HF (1986) Rapid chemical probing of conformation in 16S ribosomal RNA and 30S ribosomal subunits using primer extension. J Mol Biol 187, 399–416.

4 Sy J, Ogawa Y & Lipmann F (1973) Nonribosomal synthesis of guanosine 5¢, 3¢-polyphosphates by the

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

694

R.-M. Knutsson Jenvert and L. Holmberg Schiavone

tRNA and ribosome-dependent synthesis by stringent factor

30 Lill R, Robertson JM & Wintermeyer W (1984) tRNA binding sites of ribosomes from Escherichia coli. Bio- chemistry 23, 6710–6717.

19 Wimberly BT, Guymon R, McCutcheon JP, White SW & Ramakrishnan V (1999) A detailed view of a riboso- mal active site: the structure of the L11-RNA complex. Cell 97, 491–502.

31 Gao W, Jakubowski H & Goldman E (1995) Evidence that uncharged tRNA can inhibit a programmed trans- lational frameshift in Escherichia coli. J Mol Biol 251, 210–216.

32 Howard GA & Gordon J (1976) Stringent factor binds to Escherichia coli ribosomes only in the presence of protein L10. FEBS Lett 68, 211–214.

20 Cundliffe E (1981) Antibiotic inhibitors of ribosome function. In The Molecular Basis of Antibiotic Action (Gale EF, Cundliffe E, Reynolds PE, Richmond MH & Waring MH, eds), pp. 402–547. Wiley, New York. 21 Modolell J & Vazquez D (1977) The inhibition of ribo- somal translocation by viomycin. Eur J Biochem 81, 491–497.

33 Ramagopal S & Davis BD (1974) Localization of the

22 Rodnina MV, Savelsbergh A, Katunin VI & Winter-

meyer W (1997) Hydrolysis of GTP by elongation factor G drives tRNA movement on the ribosome. Nature 385, 37–41.

stringent protein of Escherichia coli on the 50S riboso- mal subunit. Proc Natl Acad Sci USA 71, 820–824. 34 Goldman E & Jakubowski H (1990) Uncharged tRNA, protein synthesis, and the bacterial stringent response. Mol Microbiol 4, 2035–2040.

23 Hausner T-P, Geigenmu¨ ller U & Nierhaus KH (1988)

35 Avarbock D, Avarbock A & Rubin H (2000) Differen-

The allosteric three-site model for the ribosomal elonga- tion cycle. J Biol Chem 263, 13103–13111.

tial regulation of opposing RelMtb activities by the ami- noacylation state of a tRNA.ribosome.mRNA.RelMtb complex. Biochemistry 39, 11640–11648.

f

24 Brodersen DE, Clemons WM, Carter AP, Morgan-War- rem R-J, Wimberly BT & Ramakrishnan V (2000) The structural basis for the action of the antibiotics tetracy- cline, pactamycin, and hygromycin B on the 30S riboso- mal subunit. Cell 1143–1154.

36 Nyga˚ rd O & Hultin T (1977) Accumulation of tRNAMet on 80S ribosomes in vitro under the influence of a Met- tRNA deacylase from rat liver microsomes. Eur J Bio- chem 72, 537–542.

37 Moazed D, Van Stolk BJ, Douthwaite S & Noller HF

25 Culver GM & Noller HF (1999) Efficient reconstitution of functional Escherichia coli 30S ribosomal subunits from a complete set of recombinant small subunit ribo- somal proteins. RNA 5, 832–843.

(1986) Interconversion of active and inactive 30S riboso- mal subunits is accompanied by a conformational change in the decoding region of 16S rRNA. J Mol Biol 191, 483–493.

38 Traub P, Mizushima S, Lowry CV & Nomura M (1981)

26 Nirenberg M & Leder P (1964) RNA code words and protein synthesis. The effect of trinucleotides upon the binding of SRNA to ribosomes. Science 145, 1399–1407.

Reconstitution of Ribosomes from Subribosomal Components. Academic Press, New York.

27 Cate J, Yusupov MM, Yusupova GZ, Earnest TN & Noller HF (1999) X-ray crystal structures of 70S ribo- some functional complexes. Science 285, 2095–2104. 28 Moazed D & Noller HF (1989) Intermediate states in

39 Holmberg L & Noller HF (1999) Mapping the riboso- mal RNA neighborhood of protein L11 by directed hydroxyl radical probing. J Mol Biol 289, 223–233. 40 Larsson SL & Nyga˚ rd O (2001) Proposed secondary

the movement of transfer RNA in the ribosome. Nature 342, 142–148.

29 Moazed D & Noller HF (1989) Interaction of tRNA

structure of eukaryote specific expansion segment 15 in 28S rRNA from mice, rats, and rabbits. Biochemistry 40, 3222–3231.

with 23S rRNA in the ribosomal A, P, and E sites. Cell 57, 585–597.

FEBS Journal 272 (2005) 685–695 ª 2005 FEBS

695