THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1987 by The American Society of Biological Chemists, Inc.

Vol. 262, No. 4, Issue of February 5, pp. 1766-1771,1987 Printed in U.S.A.

Demonstration of Erythromycin-dependent Stalling of Ribosomes on the ermC Leader Transcript*

(Received for publication, July 7, 1986)

Chittampalli S. Narayanan and David Dubnau From the Department of Microbiology, The Public Health Research Institute of the City of New York, Inc., New York, New York 10016

ermC encodes a methylase that modifies 23 S rRNA, conferring resistance to macrolide-lincosamide-strep- togramin B antibiotics. The expression of this gene is induced by erythromycin using a translational mech- anism. We have employed the inherent RNase activity of a Bacillus subtilis S-30 extract as a probe for study- ing the interaction of ribosomes with ermC mRNA in the presence of antibiotics. 5’ end-labeled ermC runoff transcript is a substrate for this RNase activity, while the ribosome-bound region of the RNA appears to be protected. Erythromycin- and oleandomycin-depend- ent protection of fragments of length 79-81 was ob- served during the translation of end-labeled ermC transcript. This occurs only using unmethylated (erythromycin sensitive) ribosomes. Various other an- tibiotics including clindomycin, tylosin, and lincomy- cin do not show this specific protection. These effects parallel the in vivo specificity of ermC induction. The effect of erythromycin can be abolished by using oli- gonucleotides complementary to regions of the ermC transcript upstream from nucleotide 71 and not by using an oligonucleotide complementary to a region of ermC downstream from that position. These results are interpretable in terms of the translational atten- uation model and demonstrate that erythromycin- bound ribosomes initiate translation of the leader pep- tide, stall upstream from nucleotide 80 on the ermC mRNA, and thus make the ribosome-binding site for methylase message available for ribosome interaction.

In several cases of post-transcriptional gene regulation, control is exerted by regulating the initiation of protein syn- thesis. Certain ribosomal proteins bind to their own mRNA and prevent ribosomes from initiating translation (1). Other examples include the control of synthesis of T4 gene 32 protein (2) and MS, coat protein (3). An RNA complementary to mRNA prevents the ribosome from binding to the TnlO transposase mRNA (4). On the other hand in MS2, translation of the upstream coat protein gene renders the sequestered initiation region of the downstream replicase message active ( 5 ) . In all of these cases, mRNA-ribosome interaction is crucial to the regulatory mechanism.

The ermC gene of pE194 codes for an rRNA methylase

* This work was supported by United States Public Health Service Grants AI-17472 and GM-37137 (awarded to D. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adver- tisement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

which offers resistance to a group of MLS’ antibiotics (6). At low concentrations of erythromycin or oleandomycin, an in- crease in the production of methylase can be observed (7, 8). This effect is explained by the translational attenuation model. The ermC transcript folds into an inactive form in which the initiation region (SD2) of the methylase message is partially base paired to an upstream leader region which codes for a 19-amino acid polypeptide (9,lO) (Fig. 1). Accord- ing to the model, an erythromycin-bound ribosome translating the leader peptide stalls at an appropriate site, rendering the initiation region of the methylase message single-stranded. An erythromycin-free ribosome then initiates translation of methylase at SD2. Much evidence has been presented to support this model (7-ll), but two crucial assumptions remain untested an erythromycin-bound ribosome must translate the leader region of the ermC transcript, and this erythromycin- bound ribosome must stall while reading the leader peptide message.

The mechanism of action of erythromycin and of other macrolide antibiotics is not completely understood. It appears that erythromycin stimulates the dissociation of peptidyl- tRNA from the ribosome during translation (12) and can cause the freezing of polysomes (13). It also seems that erythromycin-bound ribosomes have more difficulty in trans- lating hydrophilic and bulky amino acids (14). However, the latter conclusion is based on studies with synthetic template RNA and must be interpreted cautiously. In this study, we have used the intrinsic RNase activity of crude S-30 extracts of Bacillus subtilis to probe the effects of erythromycin and of other antibiotics on translation through the ermC leader.

EXPERIMENTAL PROCEDURES

The source of some of the materials has been described in the preceding paper (19). [y3’P]ATP (7000 Ci/mmol) and [cY-~*P]ATP (3000 Ci/mmol) were from New England Nuclear. Fusidic acid was from Lovens Kemiske Fabrik, Copenhagen, Denmark; neomycin and kanamycin from Behring Diagnostics; kasugamycin from Sanraku Ocean Co., Ltd., Tokyo; lincomycin, spectinomycin, and clindomycin

chloramphenicol from Sigma; thiostrepton was a kind gift of I. Smith from Upjohn; streptomycin, tetracycline, tylosin, erythromycin, and

(of this department). The antibiotic solutions (1 mg/ml) were pre- pared in water. Thiostrepton was dissolved in dimethyl sulfoxide. Oligodeoxynucleotide fragments were purchased and purified as de- scribed in the preceding paper (19).

Methods-Uniformly labeled transcript was obtained by transcrib- ing pE194 DNA, cut with HaelII, with Escherichia coli RNA polym- erase in the presence of 10-20 GCi of [w3’P]ATP as described previously (9). Runoff transcript for end labeling was obtained by transcribing pE194 DNA cut with HaeII1, or UAA mutant plasmid DNA (11) cut with HaeIII, with E. coli RNA polymerase. End labeling

The abbreviations used are: MLS, macrolide-lincosamide-strep- togramin B; SD, Shine-Dalgarno site.

1766

Erythromycin-dependent Stalling of Ribosomes on ermC m R N A 1767

C A U A A

m u c A G

:3: C A110

U A A U A U

U A G C

A 9 0 A U

U A A U U A U A120 G C G C U A

A B

7 ‘0

F3

4 0 U A U A

U A

A U U A

U A

A U U A U A

3011 A U A I60

G

- .” .. A A A A A A U U G 6 C A U C A C A G U C A A A A

2 0 - I 170

FI FIG. 1. Secondary structure of the ermC leader (9, 10).

Regions complementary to the 3 oligonucleotides used in this study are shown. The sequences of the oligonucleotides are given, and the positions of SD1 and SD2 are indicated.

6

3

17

:I

the transcript using bacterial alkaline phosphatase, T4 polynucleotide kinase, and 50 pCi of [y3*P]ATP, and isolation of the labeled fragments have been described (9). Bacterial growth and preparation of S-30 extracts from BD404 and BD886 and conditions for in vitro translation are described in the preceding paper (19). The translation reaction mix (6 pl) contained the following in addition to the general components (with no [35S]methionine): 15 mM M$+, 5-10,000 Cer- enkov counts of labeled transcript, antibiotic (200 pg/ml), and 1 pl of S-30. Each mix was incubated at 37 or 25 “C for various times, an equal volume of formamide containing 0.001% of bromphenol blue and xylene cyano1 was added, and the mixtures electrophoresed on 15% acrylamide-urea sequencing gels (9). Following electrophoresis the gels were covered with Saran wrap and autoradiographed at -70 “C with intensifying screens. In the reactions where synthetic oligodeoxyribonucleotide fragments were used, these were added just prior to adding S-30.

RESULTS

Fig. 2 shows the effects of various antibiotics on mRNA cleavage during the translation of the 5’ end-labeled 271- nucleotide ermC runoff transcript with B. subtilis S-30. “Spon- taneous” degradation of the end-labeled transcript in the absence of S-30 occurs specifically at C or U residues, followed by A residues (lune R). We have observed this before, as have others with a different RNA sample (9, 15). Punel A shows

FIG. 2. Effect of antibiotics on 5‘ end-labeled ermC runoff transcript during translation. The reaction components are de- scribed under “Experimental Procedures.” Antibiotics (200 pg/ml) were added prior to S-30. The reaction was at 25 “C for 20 min. R, RNA alone. Lanes 1-15 show translations of the transcript containing no antibiotic ( I ) , erythromycin (2), clindomycin (31, oleandomycin ( 4 ) , tylosin (5) , chloramphenicol (6) , fusidic acid (7), kanamycin (8), kasugamycin (9), lincomycin (IO), neomycin (111, spectinomycin (12), streptomycin (13) thiostrepton (14), tetracycline (15). A, unmethyl- ated S-30 extract; B, methylated S-30 extract. Some of the “sponta- neous” degradation products of the transcript are marked.

translation with an unmethylated S-30 extract from BD404. B presents results with a methylated S-30 extract from BD886. The latter strain carries a single chromosomal copy of ermC. The strain was grown in the presence of erythro- mycin as inducer, and the ribosomes are consequently fully methylated and unable to bind MLS antibiotics (6). When the runoff transcript is translated in the absence of antibiotic (lune I), there is some cleavage at positions 109-111, in addition to the cleavages observed without adding S-30. When the translation is carried out in the presence of various antibiotics, specific cleavages occur at various positions de- pending on the nature of the antibiotics. In the presence of erythromycin (lune 2) or oleandomycin (lune 4 ) (inducers of

1768 Erythromycin-dependent Stalling of Ribosomes on ermC mRNA

ermC methylase), heavy cleavage occurs clustered at 79-81 with some cleavage occurring several nucleotides before and after these positions. Although the experiments reported in this study were carried out with 200 pg/ml of the antibiotics, we have tested the effect of lower amounts of erythromycin. The cleavages at 79-81 are apparent with only 0.05 pg/ml erythromycin. This is a subinhibitory concentration that is sufficient to induce ermC in vitro (19). These erythromycin- dependent cleavages were time-dependent, appearing after only 2 min of incubation, were dependent on S-30 concentra- tion, and were not affected by the presence or absence of rifampicin and RNasin (not shown). With clindomycin (lune 3), slight cleavage occurs at positions 60-63. With, tylosin (lane 5 ) very slight cleavage occurs at 61-79. With chloram- phenicol (lane 6), slight cleavage occurs at 70-72. With fusidic acid (lane ?), kanamycin (lane 8), spectinomycin (lane 12), and streptomycin (lane 13), there is no significant change compared to the translation with S-30 containing no anti- biotic. With kasugamycin (lane 9) and neomycin (lane II), slight cleavages occur at positions 55-57 and with lincomycin (lane IO), at position 60-62. With thiostrepton (lane 14) and tetracycline (lane 15), cleavages somewhat heavier than those observed with other antibiotics (except erythromycin and oleandomycin) occur at positions 54-58.

Just as with unmethylated extracts, there is slight cleavage at positions 109-111 with methylated 8-30 preparations. The cleavages induced by antibiotics such as chloramphenicol, kasugamycin, neomycin, thiostrepton, and tetracycline re- main unchanged in the translations using methylated ex- tracts. With fusidic acid, kanamycin, spectinomycin, and streptomycin which induce no specific cleavages, the patterns with methylated and unmethylated extracts are again the same. When lincomycin was present in the translation mix- ture containing methylated ribosomes, cleavages occurred at positions 64-67 instead of at positions 60-62 as observed with unmethylated extracts. The specific cleavages observed in the translations using unmethylated ribosomes and the antibiot- ics erythromycin, clindomycin, oleandomycin, and tylosin are no longer observed with the methylated ribosomes, and in- stead, cleavages now occur at positions 109-111 just as with the translations containing no antibiotic. This is expected, since these antibiotics bind with lowered affinity to methyl- ated ribosomes. Using the B. subtilis in uitro translation system described in the preceding paper (19), we have exam- ined the effects of the antibiotics employed in Fig. 2, at 200 pg/ml. All were completely inhibitory in unmethylated ex- tracts except for kasugamycin. In methylated extracts, all but kasugamycin and the MLS antibiotics were inhibitory (not shown).

Fig. 3 shows the translation of the ermC transcript in the presence of synthetic oligodeoxyribonucleotides. These syn- thetic oligonucleotide fragments are complementary to spe- cific portions of the ermC leader region, as shown in Fig. 1. In the presence of as little as 0.025 pg of oligonucleotide fragment 1, a major portion of the radioactivity associated with end-labeled transcript is lost with or without erythro- mycin (lanes A 1-8). There are several positions at which cleavages occur due to the presence of the fragments, and these are different for different fragments. We will concen- trate only on the specific cleavages at positions 79-81 (eryth- romycin-dependent) and 109-111 (erythromycin-independ- ent). When oligonucleotide fragment 2 is present during translation (without erythromycin, lanes B 1-4) the specific cleavages occurring at positions 109-111 gradually decrease with increasing amounts of the fragment, almost completely disappearing when 1.25 pg of fragment is present. When these

experiments were done in the presence of erythromycin (lanes B 5-8), the antibiotic-dependent cleavage at positions 79-81 gradually decreases with increasing amounts of fragment 2. The cleavages at positions 109-1 11 (erythromycin-independ- ent) and 79-81 (erythromycin-dependent) are unaffected with increasing amounts of fragment 3 (lanes C1-7). These effects are readily interpretable in terms of the translational atten- uation model. It appears that fragment 1 binds to SD1, preventing ribosome attachment and permitting wholesale loss of radioactivity from end-labeled mRNA. Fragment 3 binds to SD2, preventing translation of methylase (19), but permitting ribosomes to load at SD1. Protection of the mes- sage occurs, as well as the specific cleavage at 79-81, indicative of erythromycin-induced stalling. Fragment 2 binds so as to free SD2, permitting synthesis of methylase in the absence of erythromycin (19). It also apparently retards the movement of ribosomes through the leader peptide coding sequence, decreasing the frequency of erythromycin-dependent cleav- age.

Fig. 4 represents a study similar to the one described in Fig. 2, but with higher amounts of added oligonucleotide fragment (3 pg). With fragment 1 all the radioactivity is lost (lunes 8 and 9). This again indicates that prevention of ribosome loading at SD1 renders the mRNA susceptible to degradation at the 5’ end. With fragment 2 in the absence of erythromycin, no specific cleavage is observed at 109-111 (lane 10). In the presence of erythromycin (lane l l) , the cleavage at positions 79-81 is considerably less than in the control with no fragment and with erythromycin (lane 3). This is presumably due to decreased translation of the stall region due to binding of fragment 2. With fragment 3 (lanes 12 and 13) the erythro- mycin-dependent and -independent cleavages are still evident, as expected.

Fig. 5 shows the translation of a mutant transcript using unmethylated extracts and various antibiotics. Panel A shows the results with 5’ end-labeled ermC transcript. Erythromy- cin, oleandomycin, and clindomycin-dependent specific cleav- ages are obvious, Panel B shows the 5’ end-labeled ermC transcript with a UAA codon introduced at position 25 instead of GGC. This change interrupts translation of the leader peptide immediately after initiation at the preceding AUG codon, preventing induction of the methylase in vivo (11) and i n vitro (19). Since this transcript is similar in sequence to ermC, the spontaneous specific degradation occurring at C or U followed by A produces the same pattern as with the wild type ermC transcript. No antibiotic-specific cleavage can be seen at positions 79-81. There is, however, cleavage at posi- tions 54-57 with S-30 alone or with S-30 and antibiotics. This can be seen more clearly in a separate experiment shown in panel D. Panel C presents studies using uniformly labeled transcript. Erythromycin and oleandomycin-dependent cleav- ages at positions 79-81 are obvious (lanes 2 and 4 ) .

DISCUSSION

The rationale behind these studies was that during in vitro translation the translated portion of the message will be protected by ribosomes, while the untranslated portion will be sensitive to endogenous nucleases. With 5-30 alone there is some cleavage at positions 107-111. We do not understand this phenomenon, but it is specific for translating ribosomes. S-100 or salt-washed ribosomes alone have no such effect (not shown).

In the translation of the ermC runoff transcript in the presence of erythromycin there is an accumulation of frag- ments of average length from 79-81. We interpret this specific cleavage pattern as indicating that erythromycin-bound ri-

Erythromycin-dependent Stalling of Ribosomes on ermC mRNA 1769

FIG. 3. Effects of synthetic oligodeoxyribonucleotides on the 5‘ end-labeled ermC runoff transcript during translation. The reactions contained all the general components, end-labeled transcript (10,000 Cerenkov counts), oligonucleotides, erythromycin (where applicable), and 1 p1 of S-30. The reaction was at 25 “C for 20 min. Oligonucleotides 1, 2, and 3 were used in A, B, and C, respectively. In A and C the amounts of fragments used are (in pg): 0.025 (lanes I and 5); 0.125 (lanes 2 and 6); 0.25 (lanes 3 and 7); 1.25 (lanes 4 and 8). Lunes 1-4, no erythromycin, lanes 5-8 contained 200 pg/ml erythromycin. In B, the amounts of oligonucleotide 2 added were: 0.024 (lam 4 ) ; 0.125 (lanes 1 and 5); 0.25 (lanes 2 and 6); and 1.25 pg (lanes 3 and 7). B, 1-3, no antibiotic; and B, 4-7, contained erythromycin. C- and C’ refer to translations with (C’) or without (C-) erythromycin and without oligodeoxyribonucleotides. The arrow pointing left indicates the erythromycin-dependent cleavages at positions 79-81.

bosomes bind to SD1 and translate the leader peptide. Eryth- romycin-dependent stalling of ribosomes occurs within the leader peptide. Since the transcript downstream is naked, it is available to nucleases and is subject to degradation up to the region protected by the ribosomes, giving an accumulation of fragments of a specific size. This interpretation is supported by several findings. First, the specific cleavage is dependent on the availability of unmethylated (erythromycin-suscepti- ble) ribosomes. Thus, the effect is ribosome dependent and is not due to a direct interaction of erythromycin with transcript. Second, the specific cleavage is prevented by conditions that prevent translation of the leader peptide (fragment 2 binding and the UAA mutation). Finally the cleavages at 79-81 are specific for erythromycin and oleandomycin as expected from the induction specificity of ermC. An alternative interpreta- tion of the observed cleavages is that they represent an erythromycin-bound ribosome specific processing event that

may provide an explanation for ermC induction, since proc- essing at positions 79-81 would permanently free SD2. This interpretation has been ruled out since processing of the ermC transcript does not occur at this position in vivo in the presence of erythromycin.’

We cannot infer precisely where the actual erythromycin- induced inhibition of peptide synthesis occurs. We may as- sume that the cleavage site (79-81) defines the leading edge of the stalled ribosome or where the mRNA emerges from association with the ribosome. If the peptidyltransferase site is located near the “middle” of the translating ribosome, then the leader peptide made in the presence of erythromycin probably contains 12-15 rather than 19 amino acids. This is based on the assumption that 25-40 bases are protected by a stalled ribosome. If this site is near the “front” edge of the

* D. H. Bechhofer and D. Dubnau, unpublished data.

1770

10:

IOC

7 :

72

64

5€

4E

44

36

33

Eqthromycin-dependent Stalling of Ribosomes on ermC mRNA

c

FIG. 4. Effect of a high concentration of synthetic oligode- oxyribonucleotideson the 5' end-labeled ermC transcript dur- ing translation. The reactions were as described in Fig. 3. 3 pg of oligonucleotide were used. Lanes: 8 and 9, oligonucleotide I ; IO and 11, oligonucleotide 2; 12 and 13, oligonucleotide 3; 8, 10, 12, no erythromycin; 9, 11, 13, with erythromycin. 1-7 are controls with no oligonucleotide. Lane 1, RNA alone (no S-30); lane 2, translation of RNA with no antibiotic; lanes 3-8, translation of RNA with eryth- romycin (lune 3), clindomycin (lane 4) , oleandomycin (lane 5 ) , tylosin (lane 6), and chloramphenicol (lane 7). The arrows indicate the erythromycin-dependent and -independent cleavages.

ribosome, the leader peptide may be nearly completed. Since B. subtilis ribosomes bound at SD1 protect 25-40 nucleotides (9), protection of 79-81 bases from the 5' terminus of the transcript requires two ribosomes. Since the transcript is 5' end-labeled, the 5' end must be protected.

The protection of 79-81 bases was specific for erythromycin and oleandomycin. Other macrolide and lincosamide anti-

A B C

R I 2 345Rl2 345R I 23 45 "-

107

1

D E -n R 12 12

79 7 2

I

53 I

44

36

33

FIG. 5. ermC and the UAA mutant transcripts during trans- lation. The reactions contained the general translation components, 15,000 Cerenkov counts of labeled RNA, antibiotics, and 1 pl of S- 30. The reaction was at 37 "C for 10 min. A, 5' end-labeled ermC transcript. B, 5' end-labeled UAA mutant transcript. C, uniformly labeled ermC transcript. R, RNA alone (no S-30). Lanes 1-5 show translations of RNA with no antibiotics (lane I), erythromycin (lune 2), clindomycin (lane 3), oleandomycin (lune 4 ) , tylosin (lane 5) . On the right ( D and E ) is shown an electrophoretic analysis of the UAA mutant transcript after translation, run for a short time. These reactions were at 25 "C for 20 min. D, UAA mutant RNA. E, ermC RNA. R, RNA alone (no S-30). Lane I , RNA with S-30; lane 2, RNA with S-30 and erythromycin. The numbering on the left edge applies only to A, B, and C. The arrows indicate the erythromycin-dependent and -independent cleavages in panels D and E.

biotics, such as clindomycin, tylosin, and lincomycin, do not exhibit this effect. Thus, the i n vitro cleavage pattern mimics the in vivo ermC induction specificity. In the presence of clindomycin, cleavage occurs a t 58-61, and the frequency of cleavage is low compared to that with erythromycin or olean- domycin. It is possible that clindomycin-bound ribosomes slow down a t some point before 58-61 and then either fall off or continue. These events with clindomycin might free the methylase initiation codon but not the entire ribosomal bind- ing site, explaining the failure of clindomycin to induce ermC. In addition, as noted, the clindomycin-specific cleavages occur at low frequency. Similar arguments apply to lincomycin, which causes cleavage at positions 60-62. With chloramphen- icol, cleavage occurs at nucleotides 70-72. Since this region is

Erythromycin-dependent Stalling of Ribosomes on ermC mRNA 1771

opposite SD2, this should free SD2 and chloramphenicol might act as an inducer. However, chloramphenicol-specific cleavage occurs to a much lesser extent than with erythro- mycin. As with clindomycin, it is possible that chloramphen- icol-bound ribosomes merely slow down without stalling. We have been unable to induce methylase synthesis in vitro with chloramphenicol (not shown).

Using methylated extracts, the effects of erythromycin, oleandomycin, tylosin, and clindomycin were abolished. Since the ermC specific methylation of 23 S rRNA confers resistance to only MLS antibiotics, the effects of other antibiotics on the observed cleavages should not be affected by the use of methylated S-30, as observed. However, with lincomycin, a lincosamide antibiotic, cleavages using methylated extracts occurred at positions 64-67 instead of at 60-62. Although the pattern is affected by the ermC specific methylation as ex- pected, the interaction with lincomycin may be more complex than with the other antibiotics in this group. In order to demonstrate that these effects require translation of the ermC leader peptide, a mutant containing UAA instead of GGC following the AUG of the leader peptide was used. With S-30, new cleavages occur at positions 40-42 and 54-57. These may reflect protection by ribosomes that form initiation complex a t SD1. With the antibiotics erythromycin, oleandomycin, clindomycin, or tylosin no change in this pattern is observed. Thus, the specific cleavages observed with the wild type transcript with these antibiotics require translation in the leader peptide reading frame, as noted above.

Further insight into these cleavage and stalling events was gained by using synthetic oligodeoxyribonucleotide probes. Since SD2 is presumably not involved in erythromycin-de- pendent ribosome stalling, oligonucleotide complementary to SD2 (fragment 3) should have no effect on this process. Conversely, since SDl is needed for translation of the leader peptide, oligonucleotide complementary to SD1 (fragment 1) should abolish the stalling effect. Oligonucleotide complemen- tary to nucleotides 53-68 (fragment 2) should interfere with ribosome movement and stalling, although this oligonucleo- tide by itself induces methylase synthesis (19). These excep- tions were fulfilled. With as little as 0.25 pg of fragment 1, a major portion of the radioactivity associated with ermC runoff transcript was lost. A likely explanation is that the fragment prevents ribosome interaction at SD1 and does not comple- ment the first 9 oligonucleotides. The 5’ end label would thus become susceptible to nucleases as observed. If this notion is true, the use of uniformly labeled runoff transcript with fragment 1, with and without erythromycin, should show no wholesale degradation of RNA. This has been confirmed (not shown).

We conclude from these studies that the various macrolide and lincosamide antibiotics affect the interaction of ribosomes with ermC leader transcript a t characteristic positions during translation of the leader peptide. We interpret this effect as stalling or retarding the movement of ribosomes. The speci- ficity and localization of these effects are consistent with the properties of ermC induction and are as predicted by the translational attenuation model. We do not understand the factors responsible for determining these specificities. We have suggested that the location of the stalling event may be determined by a preference of erythromycin for interrupting

translation a t bulky hydrophilic amino acid side chains (16). The sequence of the leader peptide (17, 18) and the apparent location of the stall site conforms to this idea. Other factors that may be involved include RNA sequence and structure, and codon usage.

These antibiotic-dependent cleavages are clearly secondary, are apparently artifacts of the in uitro system, and do not necessarily reflect the main effect of the antibiotic on trans- lation, namely inhibition. For instance, although the macro- lide and clindomycin-dependent cleavages, and the inhibitory effects on protein synthesis are abolished by methylation of 23 S rRNA, lincomycin-specific cleavages are observed in the methylated extracts. In these extracts lincomycin does not inhibit translation (not shown). Also, kasugamycin-specific cleavage is observed with both methylated and unmethylated extracts with no apparent inhibition of translation occurring in either extract (not shown). In general, however, the meth- odology described in this communication may provide a pow- erful new approach to the investigation of antibiotic action in certain cases, such as with the MLS agents, where the use of methylated extracts provides a means of demonstrating the specificity of the cleavages observed.

Acknowledgments-We thank L. Mindich for valuable discussions and A. Howard for expert secretarial assistance.

REFERENCES 1. Jinks-Robertson, S., Baugham, G., and Nomura, M. (1984) in

Gene Expression, Alfred Benzon Symposium I9 (Clark, B. F. C., and Petersen, E. H. U., eds) pp. 395-412, Munksgaard, Copen- hagen

2. Lemaire, G., Gold, L., and Yarus, M. (1978) J. Mol. Biol. 126,

3. Bernard, A., and Spahr, P.-F. (1972) Proc. Natl. Acad. Sci. U. S.

4. Simons, R. W., and Kleckner, N. (1983) Cell 34, 683-691 5. Min Jou, W., Haegeman, G., Ysebaert, M., and Fiers, W. (1972)

6. Shivakumar, A. G., and Dubnau, D. (1981) Nucleic Acds Res. 9,

7. Weisblum, B. (1983) in Gene Function in Prokaryotes (Beckwith, J., Davies, J., and Gallant, J. A., eds) pp. 91-121, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

73-90

A. 69, 3033-3037

Nature 2 3 7 , 8 2 4 8

2549-2562

8. Dubnau, D. (1984) CRC Crit. Reu. Biochem. 16, 103-132 9. Narayanan, C. S., and Dubnau, D. (1985) Nucleic Acids Res. 13,

10. Mayford, M., and Weisblum, B. (1985) J. Mol. Biol. 185, 769-

11. Dubnau, D. (1985) EMBO J. 4,533-537 12. Menninger, J. R., and Otto, D. P. (1982) Antimicrob. Agents

13. Cundliffe, E., and McQuillen, K. (1967) J. Mol. Biol. 30, 137-

14. Mao, J. C-H., and Robishaw, E. E. (1972) Biochemistry 11,4864-

15. Carbon, P., Ehresmann, C., Ehresmann, B., and Ebel, J. P. (1978)

16. Dubnau, D., Grandi, G., Grandi, R., Gryczan, T. J., Hahn, J., Kozloff, Y., and Shivakumar, A. G. (1981) in Molecular Biology, Pathogenicity, and Ecology of Bacterial Plasmids (Levy, S. B., Clowes, R. C., and Koenig, E. L., eds) pp. 157-167, Plenum Publishing Corp. New York

17. Gryczan, T. J., Grandi, G., Hahn, J., Grandi, R., and Dubnau, D. (1980) Nucleic Acids Res. 8, 6082-6097

18. Horinouchi, S., and Weisblum, B. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, 7079-7083

19. Narayanan, C. S., and Dubnau, D. (1987) J. Biol. Chem. 262,

7307-7326

780

Chemother. 2 1, 811-818

146

4872

FEBS Lett. 94, 152-156

1756-1765