Complementarity of Bacillus subtilis 16S rRNA with Sites of

Vol. 172, No. 11

Complementarity of Bacillus subtilis 16S rRNA with Sites ofAntibiotic-Dependent Ribosome Stalling in cat and erm Leaders

ELIZABETH J. ROGERS,1 NICHOLAS P. AMBULOS, JR.,1 AND PAUL S. LOVETT' 2*Department of Biological Sciences, University of Maryland Baltimore County, Catonsville, Maryland 21228,1*

and The Center of Marine Biotechnology, University of Maryland, Baltimore, Maryland 212022

Received 22 March 1990/Accepted 20 August 1990

Inducible cat and erm genes are regulated by translational attenuation. In this regulatory model, geneactivation results from chloramphenicol- or erythromycin-dependent stalling of a ribosome at a precise site inthe leader region of cat or erm transcripts. The stalled ribosome is believed to destabilize a downstream regionof RNA secondary structure that sequesters the ribosome-binding site for the cat or erm coding sequence. Herewe show that the ribosome stall sites in cat and erm leader mRNAs, designated crb and erb, respectively, are

largely complementary to an internal sequence in 16S rRNA of Bacillus subtilis. A tetracycline resistance genethat is likely regulated by translational attenuation also contains a sequence in its leader mRNA, trb, which iscomplementary to a sequence in 16S rRNA that overlaps with the crb and erb complements. An in vivo assayis described which is designed to test whether 16S rRNA of a translating ribosome can interact with the crbsequence in mRNA in an inducer-dependent reaction. The assay compares the growth rate of cells expressingcrb-86 with the growth rate of cells lacking crb-86 in the presence of subinhibitory levels of inducers of cat-86,chloramphenicol, fluorothiamphenicol, amicetin, or erythromycin. Under these conditions, crb-86 retardedgrowth. Deletion of the crb-86 sequence, insertion of ochre mutations into crb-86, or synonymous codonchanges in crb-86 that decreased its complementarity with 16S rRNA all eliminated from detection inducer-dependent growth retardation. Lincomycin, a ribosomally targeted antibiotic that is not an inducer of cat-86,failed to selectively retard the growth of cells expressing crb-86. We suggest that cat-86 inducers enable thecrb-86 sequence in mRNA to base pair with 16S rRNA of a translating ribosome. When the base pairing isextensive, as with crb-86, ribosomes become transiently trapped on crb and are temporarily withdrawn fromprotein synthesis to the extent that growth rate declines. Site-specific positioning of an antibiotic-stalledribosome is a hallmark of the translational attenuation model. The proposed rRNA-mRNA interaction mayprecisely position the ribosome on the stall site and perhaps contributes to stabilizing the ribosome leadermRNA complex.

The plasmid gene cat-86 is inducibly regulated by chlor-amphenicol through a control mechanism termed transla-tional attenuation (21). In this regulatory scheme, the anti-biotic inducer activates translation of cat-86 mRNA ratherthan transcription of the gene. Induction results from thechloramphenicol-dependent stalling of a ribosome in theleader region of cat-86 transcripts; the location of the stalledribosome that is a consequence of chloramphenicol additionplaces the aminoacyl site at leader codon 6 (1). This precisesite of ribosome stalling results in the destabilization of a

downstream RNA stem-loop structure that normally seques-ters the ribosome-binding site for the cat-86 coding se-

quence.The site-specific stalling of a ribosome in the leader region

of cat-86 transcripts is dependent not only on chloramphen-icol but also on the nature of leader codons 2 through 5 (27).Therefore, leader codons 2 through 5, which we define as thecrb-86 box, comprise the genetic determinant in the leaderthat enables the inducing antibiotic to precisely stall a

ribosome at the location needed to disrupt the downstreamRNA stem-loop.The role that the crb-86 box plays in ribosome stalling is

not well understood. However, it has been proposed that thesequence of the leader peptide amino acids specified by thecodons of the crb-86 box may contribute to the function ofthe box as a chloramphenicol-dependent ribosome stall

* Corresponding author.

sequence (21, 25, 27). We have also suggested that a secondfunction of the crb-86 box could involve complementarybase pair formation with a sequence that is internal inBacillus subtilis 16S rRNA (21). Such an interaction mightserve to precisely position a stalled ribosome on leadermRNA. In this report, we describe a chloramphenicol-dependent phenotype of B. subtilis cells expressing thecrb-86 box which requires extensive homology between thecrb-86 box and 16S rRNA.

MATERIALS AND METHODS

Bacteria and plasmids. B. subtilis BR151 (trpC2 lys-3metB10) was used throughout. Plasmids used were pC221(30) and pPL703 (23), which carry the genes cat-221 andcat-86, respectively. cat-86 lacks a promoter. Therefore,cat-86 was transcriptionally activated by inserting either theSpac (33) or P4 (23) promoter fragment in the linker ofpPL703 (Fig. 1). Throughout this report, plasmid designa-tions are used interchangeably with cat gene designations(13). Growth media and standard plasmid manipulationswere as described in reference 12.CAT and protein assays. Chloramphenicol acetyltransfer-

ase (CAT) was assayed by the colorimetric procedure ofShaw (29), and protein was measured as described byBradford (5). CAT specific activity is the number of micro-moles of chloramphenicol acetylated per minute per milli-gram of protein at 25°C.

Response of BR151 to growth inhibition by chlorampheni-

6282

JOURNAL OF BACTERIOLOGY, Nov. 1990, p. 6282-62900021-9193/90/116282-09$02.00/0Copyright C) 1990, American Society for Microbiology

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

ROLE OF rRNA IN cat INDUCTION 6283

A AU

GACAUAA

Stop GUC

9Ser CUC

8 Ser CUC

AUCA

CAU

U

C

AGG RBS-3AGGAGAU

AAAAUUG - --

Hind III

Xbater

Bgl II/BamH I

FIG. 1. Plasmid pPL703 and the regulatory region for the chloramphenicol-inducible gene cat-86. Plasmid pPL703 contains thepromoterless cat-86 gene 144 bp downstream from a multicloning site linker. Sequences from the BamHI-BgIII junction located 3' to cat-86to the EcoRI site are from pUB110. The regulatory region for cat-86 induction extends from RBS-2 to the beginning of cat-86. Transcriptsof this region are predicted to form a stable stem-loop that sequesters the cat-86 ribosome-binding site. The regulatory leader consists ofRBS-2 and nine codons. A ribosome stalled in the leader by the action of chloramphenicol has its aminoacyl (A) site at leader codon 6 andits peptidyl (P) site at leader codon 5.

col: the growth retardation or ribosome-trapping assay. TheMIC of chloramphenicol was measured by placing 50 Rl oflog-phase cells into tubes containing 2 ml of broth andvarious concentrations of antibiotic. The tubes were shakenat 37°C for 20 h, and growth was recorded.For the growth retardation or ribosome-trapping assay,

growth rates in low levels of chloramphenicol were mea-sured by use of a Klett-Summerson colorimeter. In anindividual experiment, one batch of Penassay broth (Difco)was prepared and was supplemented with a particular con-centration of antibiotic. This batch of medium was used tosimultaneously determine the growth rates of cells carrying aversion of the cat leader that enhanced chloramphenicolsensitivity and of cells lacking a genetic determinant thatwould enhance sensitivity to chloramphenicol. Each batch

of medium was used only on the day of preparation. Abso-lute growth rates varied between experiments performed ondifferent days by as much as 15% despite our efforts tomaintain antibiotic levels at the same concentration. Regard-less of these slight variations in growth rates, the pattern ofenhanced susceptibility to chloramphenicol due to cat leaderexpression was consistently observed. In many experi-ments, fluorothiamphenicol was substituted for chloram-phenicol. Growth rates without chloramphenicol of all cellsstudied were indistinguishable.

cat gene induction. cat-86 was induced in BR151 by incu-bating cells in chloramphenicol (2 ,ug/ml), fluorothiampheni-col (0.5 ,ug/ml), or erythromycin (0.05 ,ug/ml). cat-221 wasinduced with chloramphenicol (1 ,ug/ml) or fluorothiam-phenicol (0.5 p,g/ml).

VOL. 172, 1990

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

6284 ROGERS ET AL.

B. subtilis 16S rRNA

crb boxes

crb-86, -66

crb-57

crb- 194

crb-221, -112

1305 1294

3----G A C T C T T G T C T A----5

AG

5T 3 -g TG-A LA-A CA-G AT - 4.2

g T G-A a A.g C A-G A T

a a G*A a A-g C A-G A c

a a acA a A-t C A-G A g

-10.8

- 6.8

- 6.6

FIG. 2. crb sequences from the leaders of six inducible cat genes. Data shown are taken from reference 21. Lowercase letters in crbsequences indicate a mismatch with 16S rRNA. The 16S rRNA sequence is from Green et al. (14). AG values were calculated as in reference32.

Calculation of RNA-RNA pairing. The free energy (AG) ofRNA-RNA interactions was calculated as described byTinoco et al. (32).

Site-directed mutagenesis. Oligonucleotide-directed muta-genesis and DNA sequencing were performed by using M13vectors as described previously (12, 31, 34).

Estimation of the percent contribution of crb-containingtranscripts to total cell mRNA. Nearly 40% of the cat-86transcripts initiated in vivo terminate immediately 3' to thestem-loop (4). The approximate location of the terminationsite is codon 2 of the cat-86 structural gene (4). Moreover,two other sites of partial in vivo transcription terminationwere found to exist between the XbaI and Hindlll siteswithin the cat-86 structural gene (4). Thus, a determinationof only full-length cat-86 transcripts will underestimate thenumber of crb-containing transcripts; the full-length andterminated transcript species each contain a single crb boxsequence. Therefore, we chose to quantitatively comparehybridization of a probe oligomer to total cell RNA withhybridization to an in vitro-prepared RNA standard thatcorresponded to the shortest naturally terminated in vivotranscript. By this approach, in vivo terminated transcriptsand full-length transcripts each weigh equally.RNA quantitation standards were prepared by making in

vitro runoff transcripts, using an SP6 RNA polymerasesystem. The template for transcription was the pSP65 plas-mid vector with a 637-bp insert that extends from the linkerin pPL703 to the Hindlll site (see reference 4 for details ofthe construction). The plasmid was cut with AhaIII (DraI),which cleaves between codons 2 and 3 of the crb-86 struc-tural gene. Consequently, the resulting in vitro transcriptswere 180 nucleotides (nt) in length and spanned the crb-86box (4). Various quantities of the in vitro RNA standard andRNA purified from BR151(pPL703-Spac) (3) were used fordot blot analysis. The 5'-end-labeled hybridization probe wasa 25-mer that was complementary to the leader mRNA se-quence 5'-GAAAGGAUGAUUGUGGUGGUGAAAA (Fig.1). Hybridization was at 42°C for 23 h by methods detailedelsewhere (3). Autoradiograms were prepared and scanned bydensitometry. By this method, 0.2% of the RNA fromBR151(pPL703-Spac) corresponded to the 180-nt crb-contain-ing transcript.

Stable RNA (rRNA plus tRNA) constitutes about 97% ofthe total RNA from rapidly growing Escherichia coli cells(6). By applying this value to B. subtilis cells, 3% of the totalRNA examined is mRNA. Consequently, crb-86-containingtranscripts (180 nt in length) constitute 6.6% of the mRNAfraction of cells.

RESULTS

Experimental rationale. Figure 1 depicts the regulatoryregion of cat-86 transcripts and the locations of the ami-noacyl and peptidyl sites of a stalled ribosome that isactivating the induction mechanism. It is believed that thestalled ribosome destabilizes the RNA stem-loop structure,which in turn frees ribosome-binding site 3 (RBS-3) for theinitiation of translation of the cat-86 coding region (21).Leader codons 2 through 5, which we define as the crb-86box, of the cat-86 leader mRNA are essential to the chlor-amphenicol-dependent stalling of a ribosome with its ami-noacyl site at leader codon 6 (27). Conceivably, initiation ofribosome stalling on the crb-86 box might be due to an actionof the leader peptide upon a translating ribosome that hadpreviously bound a molecule of chloramphenicol to the 50Ssubunit. Alternatively, the interaction of the leader peptidewith a drug-free, translating ribosome might enhance theaffinity of the ribosome for binding chloramphenicol. Ineither case, the ribosome stalls during leader translation inthe presence of low levels of chloramphenicol.

Refinement of ribosome positioning on leader mRNA orstabilization of the mRNA-ribosome complex could beachieved if leader mRNA could guide or stabilize the ribo-some through complementary base pairing with rRNA. Asearch for such base-pairing potential revealed that thenucleotide sequence of the crb-86 box is largely (83%)complementary to a 12-nt sequence that is found at aninternal location in B. subtilis 16S rRNA (21; Fig. 2). Thecalculated free energy (AG) of the interaction is -14.2 (31).Complementarity between putative crb boxes in other catleaders and the same 12-nt sequence in rRNA is also evident,but generally to a lesser degree (Fig. 2). 16S rRNA present inE. coli lacks most of the region of complementarity with thecrb box sequence, and it is well established that genes suchas cat-86 which are chloramphenicol inducible in B. subtilisfail to significantly induce in E. coli (17, 22).

If complementary pairing of the crb-86 box with 16S rRNAtook place in vivo during drug-dependent ribosome stalling,we suspected that the number of free ribosomes available forgeneral protein synthesis could decrease. The working hy-pothesis was that low levels of chloramphenicol would stalla ribosome on the crb box, but dissociation of chloramphen-icol from the stalled ribosome would not immediately releasethe ribosome if a crb-rRNA pairing had occurred. If asignificant number of the ribosomes of cells were tempo-rarily trapped on the crb-86 box as a result of complementarybase pairing, the growth rate of the cells should decline.

J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

ROLE OF rRNA IN cat INDUCTION 6285

200 F

In

NC

0)

too

50

20

10

Cm Ftm

,x

-/

x /X

X1

'°°t

u' 50a

.0_w

Cm

x

TAA-5*

x/

Xx /

lo I 'I0 l

I

0 1 2 3 4 1 2 3 4

HoursFIG. 3. Growth rates of leader-containing and leader-deficient

cells in low levels of chloramphenicol and fluorothiamphenicol.BR151(pUB110) and BR151 containing pPL704X-Spac or pPL703-Spac were grown at 370C in broth supplementeba with chloramphen-icol (0.5 ,ug/ml) or fluorothiamphenicol (0.25 ,ug/ml), respectively.Turbidity was measured in a Klett-Summerson colorimeter. BR151containing no plasmid grew at a rate identical to that of BR151(pUB110) in both drugs. -L, Growth curve of BR151(pUB110); +L,growth curve of BR151(pPL703AX-Spac) in chloramphenicol (Cm)or of BR151(pPL703-Spac) in fluorothiamphenicol (Ftm).

Phenotype conferred by expression of the crb-86 box in B.subtUis. To determine whether the crb-86 box served as a

trap for chloramphenicol-sensitized ribosomes, it was nec-

essary to prevent leader-containing cells from inactivatingthe agent of stalling, chloramphenicol. This was accom-plished in two ways. When we chose to use chloramphenicolas the agent of ribosome stalling, cells carried a plasmid inwhich the cat-86 structural gene had been inactivated bydeleting an internal XbaI fragment. This plasmid is desig-nated pPL703AX-Spac. Alternatively, when we chose to usea plasmid containing an intact cat-86 gene, such as inpPL703-Spac, fluorothiamphenicol was the agent of stallingbecause this derivative of chloramphenicol cannot be acety-lated by. the CAT enzyme.The MICs of chloramphenicol and fluorothiamphenicol for

BR151 were 2 and 1 ,ug/ml, respectively. These valuesdecreased twofold when the test cells were expressing thecat-86 leader. Thus, leader expression increased the appar-ent sensitivity of the cells to growth inhibition by theantibiotics. The effect of leader expression on cell growth inchloramphenicol and fluorothiamphenicol was more obviouswhen we compared the growth rate of cells expressing thecat-86 leader with that of cells lacking the leader in thepresence of low levels of the stalling antibiotics. Under theseconditions, expression of the leader reduced the growth ratesubstantially (Fig. 3).To determine whether the enhanced sensitivity of BR151

to chloramphenicol and fluorothiamphenicol was a functionof the crb-86 box, appropriate plasmids containing ochremutations at leader codons 3, 5, and 6 were constructed (1).Replacement of cat-86 leader codon 6 with the ochre muta-tion permits chloramphenicol to stall a ribosome at theproper leader site needed to induce cat-86, whereas ochremutations at leader codons 3 and 5 prevent induction byblocking access of the ribosome to the crb-86 box (1). Aplasmid with an ochre codon replacement of leader codon 6slowed the growth rate of host cells in chloramphenicol and

Ftm

TAA-6

2 3 4 0

Hours1 2 3 4

FIG. 4. Growth rates of leader-containing cells with an ochremutation at leader codon 5 or 6. Experimental details and abbrevi-ations are as described in the legend to Fig. 2 except that the cellscontained a version of the leader with an ochre mutation at eitherleader codon 5 (TAA-5) or leader codon 6 (TAA-6). BR151(pUB110)grew identically to TAA-5 in both drugs, as did leader-containingBR151 with an ochre mutation at leader codon 3 (TAA-3).

fluorothiamphenicol relative to plasmids that contained anochre codon replacement for leader codons 3 or 5 or aplasmid, pUB110, which lacks the crb-86 box (Fig. 4). Thus,the enhanced susceptibility to the chloramphenicol com-pounds required that leader translation proceed to the samesite as that required for induction of cat expression.To further establish the identity of the plasmid sequence

responsible for the reduced growth rate of host cells influorothiamphenicol, we compared the rates of growth ofBR151 harboring two deleted derivatives of pPL703-Spac.The A60 plasmid mutant lacks the region of pPL703 extend-ing from the linker to a site immediately 5' to RBS-2,whereas pPL703C2 contains a deletion that extends from thelinker to a site between the inverted repeats (2, 20). Thegrowth rates in fluorothiamphenicol (0.25 ,ug/ml) of BR151cells harboring either pPL703C2-Spac or pUB110 were iden-tical, whereas BR151 harboring the A60 mutant grew at asubstantially reduced rate (Fig. 5). Since the deletion end-points of the two mutants flank the cat-86 leader, weconclude that a sequence within the regulatory leader isresponsible for the slowing of growth in the presence of astalling antibiotic.

cat-86 is induced not only by chloramphenicol and itsderivatives but also by the nucleoside antibiotic amicetin andby erythromycin (13, 28). In contrast, lincomycin does notinduce cat-86 (13). The cat-86 gene does not confer resis-tance to amicetin, erythromycin, or lincomycin. Therefore,BR151 cells harboring pUB110 or pPL703-Spac were grownin broth containing amicetin (1 ,ug/ml), erythromycin (0.05jig/ml), or lincomycin (2 pug/ml). Cells carrying pPL703-Spacgrew more slowly in amicetin and erythromycin than didcells carrying pUB110, but no difference in growth rates wasdiscernible in lincomycin (data not shown). Thus, the en-hanced growth-inhibitory response due to cat-86 leaderexpression was limited to those antibiotics that induced thegene.The crb-86 box functions when displaced from its normal

location in the cat-86 leader. Since expression of the cat-86leader in B. subtilis heightened the response of cells tochloramphenicol, it was possible to determine whether thisphenotype was a function of only the crb-86 box or whether

VOL. 172, 1990

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

6286 ROGERS ET AL.

-

c

41-

y

100

50 cat- 86

201

10

HoursFIG. 5. Use of deletions into the cat-86 leader to define the

sequence responsible for growth retardation of B. subtilis in fluo-rothiamphenicol. BR151 harboring pUB110, pPL703A&60-Spac (des-ignated cat-86A60), pPL703C2-Spac (designated cat-86C2), orpPL703-Spac (designated cat-86) was grown in broth containingfluorothiamphenicol (0.25 ,ug/ml).

surrounding sequences in the leader were also involved. Thefour leader codons that constitute the crb-86 box, codons 2through 5, permit chloramphenicol to induce cat-86 if aprecise spatial relationship is maintained between the crbbox and the RNA stem-loop. Insertion of one extra codon,codon 5A, between codons 5 and 6 blocks induction, and wesuggested that this result was obtained because the crb boxwas displaced from the RNA stem-loop (19). Thus, theinserted codon was thought to displace the site of stalling butnot to prevent stalling. To test this idea, a version of cat-86containing the 5A codon insertion was transformed intoBR151, and the growth rate of the cells in fluorothiam-phenicol (0.25 jig/ml) was compared with the growth rate, inthe presence of the antibiotic, of BR151 carrying a version ofcat-86 in which leader translation was prevented by replace-ment of leader codon 5 with an ochre codon. The resultsdemonstrated that displacement of the crb box from theRNA stem-loop did not abolish the function of the crb box asa site of ribosome stalling and trapping, as evidenced by thereduced growth rate of host cells in fluorothiamphenicol(Fig. 6).To further test the role of surrounding leader sequences on

the function of the crb-86 box, a copy of the crb-86 box wastransplanted to a site within a gene. pPL703C2 contains aconstitutively expressed version of cat-86 that was gener-ated by deleting the regulatory leader; the deletion endpointis between the inverted repeats (20). This version of cat-86,designated cat-86C2, lacks the crb box, and no sequencecomparable to the crb box exists within the cat-86 structuralgene (16). By site-directed mutagenesis, we replaced codons3 through 6 of the cat-86C2 coding region with the crb-86box. This mutation reduced the activity of the correspondingCAT enzyme by a factor of 20. This mutant version ofcat-86, designated cat-86C2 crb-86, was inserted into BR151,and the growth rate of the cells in fluorothiamphenicol (0.25jig/ml) was compared with that of BR151 carrying cat-86C2in the same level of fluorothiamphenicol. The results dem-onstrated that the crb-86 box inserted into the cat-86 codingregion increased the susceptibility of host cells to fluorothi-

D 50C t- OB + 5A

50/

-+1_) /,#'cat- 86

20 -

10 I I I

0 1 2 3 4Hours

FIG. 6. Enhancement of host sensitivity to fluorothiamphenicolby insertion of a codon immediately downstream of the crb-86 box.BR151 harboring pUB110 (control), pPL703-Spac (cat-86), orpPL703 + 5A-Spac (cat-86 + 5A) was grown in broth containingfluorothiamphenicol (0.25 jig/ml). The +5A mutation is a singlecodon (Asp) inserted between leader codons 5 and 6.

amphenicol (Fig. 7). Thus, the crb box appeared to functionwhen placed in a site different from the leader.

Base substitution mutations in the crb box which increase ordecrease complementarity with 16S rRNA. The crb-86 and -66boxes show the highest extent of complementarity with 16SrRNA (i.e., 10 of 12 matches) in comparison with crb boxesin other cat leaders (Fig. 2). By comparison, the crb box inthe cat-221 leader, designated crb-221, contains complemen-tary matches with only 6 of the 12 nt in the 16S rRNAsequence. To test the influence of crb-221 on the growth ofB. subtilis with a stalling antibiotic, the TaqI fragment thatspans cat-221 (30) was cloned from pC221 into pUB110,yielding p110-221. This plasmid failed to measurably reducethe growth rate of host bacteria in the presence of fluorothi-

1001

-0

.)

0 1 2 3 4 5 6Hours

FIG. 7. Enhancement of the sensitivity of host cells to fluorothi-amphenicol by insertion of the crb box into a structural gene.Codons 3 through 6 of cat-86C2 were converted to the crb box bysite-directed mutagenesis. This version of the gene, designatedcat-86C2crb-86, and the parent gene cat-86C2 were transformed intoBR151 as plasmids pPL703C2 crb-86-Spac and pPL703C2-Spac,respectively. Growth rates of the cells were determined in brothcontaining fluorothiamphenicol (0.25 ,g/ml).

J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

ROLE OF rRNA IN cat INDUCTION 6287

16S r

1305

3-G A CTRNA

1294

C T T G T C T A--5'

crb-86 5g T G A a A A C A G A TAA V K T D

crb-86-1 g T t A a A A C A G A T

crb-86-2 g T a A a A A C A G A T

crb-86-3 g T a A a g A C A G A T

crb-86-4 g T a A a g A C g G A T

crb-86-5 g T a A a g A C g G A c

crb-221 a a a A a A t C A G A gAA K K S E

crb-221-1 g T G A a A t C A G A gAA V K S E

Growth AG

R -14.2

R - 11.8

R - 11.8

N -10.6

N - 4.2

N - 2.4

N - 6.6

R - 9.0

FIG. 8. Summation of the properties of wild-type and mutated crb sequences. R, The crb box retarded cell growth in fluorothiamphenicol(0.25 ,ug/ml); N, the sequence had no detectable effect on cell growth in the antibiotic; AA, sequence of amino acids in the correspondingleader peptide. All mutants of cat-86 are predicted to specify the same leader amino acids as does the wild type.

amphenicol (0.25 ,ug/ml) compared with growth of the samecells containing pUB110. We then mutagenized the crb-221box to increase its complementarity with 16S rRNA (Fig. 8)and thereby generated a version of the crb-221 box, crb-221-1, that caused host cells to grow more slowly in fluo-rothiamphenicol. The mutations introduced into the crb-221-1 box did not detectably increase or decrease theinducibility of cat-221 by either chloramphenicol or fluo-rothiamphenicol.The foregoing results suggested that the extent of comple-

mentarity between a crb box and the corresponding region of16S rRNA determined whether or not a crb box would bemeasurably active in the trapping assay. To test this idea, thecrb-86 box was mutagenized to decrease its complementaritywith 16S rRNA (Fig. 8). The mutations chosen were changesin the crb-86 box that would produce a leader peptide withthe same amino acid sequence as the wild-type version.Several of the mutant versions of the' crb-86 box showed 'aloss of function in the trapping assay (Fig. 8). Mutations thatreduced the strength of the interaction between crb and 16SrRNA to a AG of' -4.2 decreased cat-86 inducibility byfluorothiamphenicol by about one-third (Fig. 9). The samemutations virtually eliminated inducibility by erythromycin(Fig. 9).

Strength of the proposed interaction between crb and rRNAthat is required for detectable ribosome trapping. Figure 8summarizes the effects of mutations in the crb-86 box on thecalculated strength of the association of crb with 16S rRNA.Each mutation resulted in the replacement of a codon with asynonymous codon, and therefore the amino acid sequenceof the wild-type leader peptide is maintained in each mutantderivative. The data suggest that the crb-86 box sequenceretarded growth in fluorothiamphenicol (0.25 ,ug/ml) whenthe free energy, AG, of the base pairing with 16S rRNA was-11.8 or below.Figure 8 also shows the calculated AG of the interaction of

crb-221 and crb-221-1 sequences with 16S rRNA. Althoughleader amino acids specified by crb-221-1 differ from wild

type, both versions of crb allowed fluorothiamphenicol in-duction of cat-221. crb-221, which shows a AG of binding of-6.6, did not yield a positive result in the trapping assay,whereas crb-221-1 (AG = -9) gave a positive result in thetrapping assay. It is unclear why crb-221-1 (AG = -9)elevated host sensitivity to fluorothiamphenicol whereascrb-86-3 (AG = -10.6) did not. Possibly other features of thecrb-221 region contribute to stabilizing a stalled ribosome.For example, leader amino acids 4 and 5 of the cat-221 leader

FTM ERY

12

4-

-14.2 -I11 . -10.6 -4.2- -2.4 "14.2 -I I.8 -10S -4.2 -2.4

AGFIG. 9. Inducibility of crb mutants of cat-86 by fluorothiam-

phenicol (FTM) and erythromycin (ERY). Wild-type cat-86 andcrb-2, -3, 4, -5 were induced for 1 h with fluorothiamphenicol (0.25,ug/ml) or erythromycin (0.02 ,ug/ml). Values are fold induction overthe uninduced controls. AG values represent the calculated strengthof the interaction between each crb sequence and 16S rRNA.

VOL. 172, 1990

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

6288 ROGERS ET AL.

I B. subtilis specific -1310 1300 1290

16SrRNA 3 TA G G C TT GA CT C T TG T CTA AA C AC C CT AAC 5

crb-86 5'

erbC 5'

erbG 5'

erb A 5'

2 , 3 , 4 , 5g T G A a A A C A G AT

6 , 7 8 1 9AT TT t T G t a AT c

ATTT t T G t a AT a

A T a T t T G t t A T T

9 10 , 1 12 13erbD 5 t T C C c A A CT t t G A AC

trb-181 5'6 7 8 9 l0

A a A a A T T T G g G G a A T

AG

-14.2

- 4.2

- 4.2

- 3.0

-14.4

- 9.0

FIG. 10. Region of 16S rRNA that is complementary to crb, erb, and trb domains in the leaders for cat, erm, and tet genes. The sequencefor B. subtilis 16S rRNA is from Green et al. (14). The sequence noted as B. subtilis specific is absent from E. coli 16S rRNA (26). crb, erb,and trb are regions within the cat, erm, and tet leaders that are complementary to the 16S rRNA sequence. erb and trb sequences wereidentified in published sequences for erm and tet gene leaders (16, 18, 24). Codons are numbered with respect to their positions in the specificleader.

differ from the amino acids found at the same positions in thecat-86 leader. Conceivably, these amino acids may allow thecat-221 leader peptide to participate in ribosome stabiliza-tion.

DISCUSSIONcrb boxes in the leader sequences of six inducible cat

genes show 50 to 83% complementarity with a 12-nt se-quence in B. subtilis 16S rRNA. This could in theory be a

fortuitous occurrence, or it could indicate that cat leadersequences originated from rRNA gene sequences. However,the role of the crb box is to stall a chloramphenicol-sensitized ribosome at the precise leader site needed toinduce cat gene expression. This fact led us to suspect thatthe observed complementarity reflected a function enablingthe ribosome either to precisely align on the cat leadermRNA or to stabilize the ribosome on the mRNA or both.The trapping assay was developed on the premise that if

the crb-86 box paired with rRNA, a ribosome might becometransiently trapped on the cat-86 leader and thereby bewithdrawn from protein synthesis. Furthermore, the cat-86leader sequence is present on a high-copy-number plasmid,and transcription is driven by a strong promoter. Hence thecrb-86 box sequence in mRNA is highly amplified within acell. The results demonstrated that expression of the crb-86box enhanced the growth-inhibitory response of host bacte-ria to low levels of those antibiotics that stall ribosomes onthe box. This effect required translation of the box and wasobserved when the crb box was placed at a site within agene.A positive result in the trapping assay required extensive

complementarity between the crb box and 16S rRNA. Thus,several mutations in crb-86 which decreased its complemen-tarity with 16S rRNA eliminated the observed trapping.Conversely, the crb box that is present in the leader for thecat-221 gene was not detectably active in the trapping assayuntil the complementarity with 16S rRNA was increased bymutagenesis of the box sequence. The conservative muta-tions made in crb-86 that eliminated the observed ribosome

trapping diminished inducibility of gene expression by fluo-rothiamphenicol and erythromycin by up to 30 and 90%,respectively. These results suggest that the crb-rRNA inter-action that is minimally essential to cat induction by fluo-rothiamphenicol must be below the level that is detectableby the relatively insensitiVe trapping assay. By contrast,erythromycin induction of cat-86 was highly dependent onthe extent of complementarity between crb and 16S rRNA.

It is acknowledged that mechanisms other than ribosometrapping on crb could conceivably account for the reducedrate of growth of cat leader-containing cells in the presenceof the stalling antibiotics. However, inspection of the avail-able information seems to eliminate likely alternatives to theribosome-trapping model. For instance, perhaps the leaderpeptide facilitates the entry of the stalling antibiotics intocells and thereby elevates cell sensitivity to the drugs.Alternatively, perhaps the leader peptide binds to ribosomesand increases their sensitivity to chloramphenicol. Thesepossibilities seem unlikely because mutations within crb-86that decreased its complementarity with 16S rRNA but didnot alter the predicted amino acid sequence of the leaderpeptide prevented crb-86 from reducing the growth rate ofcells in fluorothiamphenicol. Alternatively, leader codons 1through 5 might encode a peptide that is, in itself, slightlytoxic to B. subtilis. This idea also seems unlikely becausein-frame fusions between the cat-86 leader (and the cat-112leader) and a reporter gene resulted in constitutive transla-tion of the reporter gene (8, 11). Thus, leader translationpresumably is constitutive, yet leader-containing cells grewat the same rate as did cells lacking the leader in the absenceof a stalling antibiotic.The classical interaction between mRNA and rRNA in-

volves the pairing of the ribosome-binding site sequence inmRNA to a complementary sequence at the 3' end of 16SrRNA. However, recent studies indicate that more internalsequences in 16S rRNA play a role in probing mRNAsequences during translation (9). An interesting aspect of thecrb-rRNA interaction is the appearance of the complementto crb in B. subtilis 16S rRNA, whereas E. coli 16S rRNA

J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

ROLE OF rRNA IN cat INDUCTION 6289

contains the potential complement to only 5 of the 12 basesof crb. Inducible cat genes are typically detected in gram-positive bacteria such as Bacillus spp. and not in gram-negative organisms. It is therefore possible that cat leadersequences have evolved to utilize a region of host rRNA forprecise positioning of a stalled ribosome.

Translational attenuation, the regulator of cat genes, wasinitially proposed as the control mechanism through whicherythromycin induces erm gene expression (10). Further-more, the leader sequence 5' to the tetracycline resistancegene on plasmid pT181 suggests that tet-181 may be similarlyregulated (18). We were therefore not surprised to find thatthe leaders for these genes also contain sequences, desig-nated erb and trb, respectively, that are complementary to16S rRNA (Fig. 10). It is noteworthy that the erb sequence inthe ermC leader is at the precise site previously proposed asthe site of erythromycin-dependent ribosome stalling (22).Thus, both crb-86 and erbC sequences are spanned by aribosome stalled by the action of the appropriate inducingantibiotic.The region of 16S rRNA that contains the complementary

sequences to crb, erb, and trb is 30 nt long (Fig. 9). 16SrRNA of B. subtilis consists of 1,550 nt, as determined fromthe sequence of the gene (14). Thus, the proposed interac-tions involve less than 2% of the 16S rRNA molecule.A rigorous test of the minimum crb-rRNA interaction

needed for chloramphenicol induction of cat-86 cannot beunambiguously obtained by further mutagenesis of crb-86.Potentially meaningful base changes beyond those alreadymade will result in amino acid replacements in the leaderpeptide and complicate interpretation. Mutagenesis of thecrb complement in 16S rRNA, as has been done in E. coli(17), is a theoretical possibility. Unfortunately, we have notas yet been successful in cloning and expressing an intact16S rRNA gene in B. subtilis.

ACKNOWLEDGMENTS

This investigation was supported by Public Health Service grantGM42925 from the National Institutes of Health and grant DMB-8802124 from the National Science Foundation.

LITERATURE CITED1. Alexieva, Z., E. J. Duvall, N. P. Ambulos, Jr., U. J. Kim, and

P. S. Lovett. 1988. Chloramphenicol induction of cat-86 requiresribosome stalling at a specific site in the leader. Proc. Natl.Acad. Sci. USA 85:3057-3061.

2. Ambulos, N. P., Jr., E. J. Duvall, and P. S. Lovett. 1986.Analysis of the regulatory sequences needed for induction of thechloramphenicol acetyltransferase gene cat-86 by chloramphen-icol and amicetin. J. Bacteriol. 167:842-849.

3. Ambulos, N. P., Jr., E. J. Duvall, and P. S. Lovett. 1987. Methodfor blot hybridization analysis of mRNA molecules from Bacil-lus subtilis. Gene 51:281-286.

4. Ambulos, N. P., Jr., S. Mongkolsuk, and P. S. Lovett. 1985. Atranscription termination signal immediately precedes the cod-ing sequence for the chloramphenicol-inducible plasmid genecat-86. Mol. Gen. Genet. 199:70-75.

5. Bradford, M. M. 1976. A rapid and sensitive method forquantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72:248-252.

6. Bremer, H., and P. P. Dennis. 1987. Modulation of chemicalcomposition and other parameters of the cell by growth rate, p.1527-1542. In F. C. Neidhardt, J. L. Ingraham, B. Magasanik,K. B. Low, M. Schaechter, and H. E. Umbarger (ed.), Esche-richia coli and Salmonella typhimurium: cellular and molecularbiology. American Society for Microbiology, Washington, D.C.

7. Bruckner, R., T. Dick, and H. Matzura. 1987. Dependence ofexpression of an inducible Staphylococcus aureus cat gene on

the translation of its leader sequence. Mol. Gen. Genet. 207:486-491.

8. Bruckner, R., T. Dick, H. Matzura, and E. Zyprian. 1988.Regulation of inducible Staphylococcus aureus cat gene bytranslational attenuation, p. 263-266. In A. T. Ganesan andJ. A. Hoch (ed.), Genetics and biotechnology of bacilli. Aca-demic Press, Inc., New York.

9. Dahlberg, A. E. 1989. The functional role of ribosomal RNA inprotein synthesis. Cell 57:525-529.

10. Dubnau, D. 1984. Translational attenuation: the regulation ofbacterial resistance to the macrolide-lincosamide-streptograminB antibiotics. Crit. Rev. Biochem. 16:103-132.

11. Duvali, E. J., N. P. Ambulos, Jr., Z. Alexieva, U. J. Kim, andP. S. Lovett. 1988. Ribosome stalling in the cat-86 regulatoryleader induces gene expression, p. 257-262. In A. T. Ganesanand J. A. Hoch (ed.), Genetics and biotechnology of bacilli.Academic Press, Inc., New York.

12. Duvall, E. J., N. P. Ambulos, Jr., and P. S. Lovett. 1987.Drug-free induction of a chloramphenicol acetyltransferase genein Bacillus subtilis by stalling ribosomes in a regulatory leader.J. Bacteriol. 169:4235-4241.

13. Duvall, E. J., S. Mongkolsuk, U. J. Kim, P. S. Lovett, T. M.Henkin, and G. H. Chambliss. 1985. Induction of the chloram-phenicol acetyltransferase gene cat-86 through the action of theribosomal antibiotic amicetin: involvement of a Bacillus subtilisribosomal component in cat induction. J. Bacteriol. 161:665-672.

14. Green, C. J., G. C. Stewart, M. A. Hollis, B. S. Vold, and K. F.Bott. 1985. Nucleotide sequence of the Bacillus subtilis ribo-somal RNA operon, rrnB. Gene 37:261-266.

15. Gryczan, T., M. Israel-Reches, M. DelBue, and D. Dubnau.1984. DNA sequence of ermD, an MLS resistance element fromBacillus licheniformis. Mol. Gen. Genet. 194:349-356.

16. Harwood, C. R., D. M. Williams, and P. S. Lovett. 1983.Nucleotide sequence of a Bacillus pumilus gene specifyingchloramphenicol acetyltransferase. Gene 24:163-169.

17. Hui, A., and H. DeBoer. 1987. Specialized ribosome system:preferential translation of a single mRNA species by a subpop-ulation of mutated ribosomes in Escherichia coli. Proc. Natl.Acad. Sci. USA 84:4762-4766.

18. Khan, S. A., and R. P. Novick. 1983. Complete nucleotidesequence of pT181, a tetracycline-resistance plasmid fromStaphylococcus aureus. Plasmid 10:251-259.

19. Kim, U. J., N. P. Ambulos, Jr., E. J. Duvali, M. A. Lorton, andP. S. Lovett. 1988. Site in the cat-86 regulatory leader thatpermits amicetin to induce expression of the gene. J. Bacteriol.170:2933-2938.

20. Laredo, J., V. Wolff, and P. S. Lovett. 1988. Chloramphenicolacetyltransferase specified by cat-86: gene and protein relation-ships. Gene 73:209-214.

21. Lovett, P. S. 1990. Translational attenuation as the regulator ofinducible cat genes. J. Bacteriol. 172:1-6.

22. Mayford, M., and B. Weisblum. 1989. ermC leader peptide.Amino acid sequence critical for induction by translationalattenuation. J. Mol. Biol. 206:69-79.

23. Mongkolsuk, S., Y.-W. Chiang, R. B. Reynolds, and P. S.Lovett. 1983. Restriction fragments that exert promoter activityduring postexponential growth of Bacillus subtilis. J. Bacteriol.155:1399-1406.

24. Monod, M., S. Mohan, and D. Dubnau. 1987. Cloning andanalysis of ermG, a new MLS resistance element from Bacillussphaericus. J. Bacteriol. 169:340-352.

25. Mulbry, W. W., N. P. Ambulos, Jr., and P. S. Lovett. 1989.Bacillus subtilis mutant allele sup-3 causes lysine insertion atochre codons: use of sup-3 in studies of translational attenua-tion. J. Bacteriol. 171:5322-5324.

26. Noller, H. F. 1984. Structure of ribosomal RNA. Annu. Rev.Biochem. 53:119-162.

27. Rogers, E. J., U. J. Kim, N. P. Ambulos, Jr., and P. S. Lovett.1990. Four codons in the cat-86 leader define a chlorampheni-col-sensitive ribosome stall sequence. J. Bacteriol. 172:110-115.

28. Rogers, E. J., and P. S. Lovett. 1990. Erythromycin inducesexpression of the chloramphenicol acetyltransferase gene cat-

VOL. 172, 1990

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.

6290 ROGERS ET AL.

86. J. Bacteriol. 172:4694-4695.29. Shaw, W. V. 1975. Chloramphenicol acetyltransferase from

chloramphenicol-resistant bacteria. Methods Enzymol. 43:737-755.

30. Shaw, W. V., D. G. Brenner, S. F. J. LeGrice, S. E. Skinner,and A. R. Hawkins. 1985. Chloramphenicol acetyltransferasegene of staphylococcal plasmid pC221. FEBS Lett. 179:101-106.

31. Taylor, J. W., J. Ott, and F. Eckstein. 1985. The generation ofoligonucleotide-directed mutations at high frequency usingphosphorothioate-mediated DNA. Nucleic Acids Res. 13:8765-

8785.32. Tinoco, I., Jr., P. N. Borer, B. Dengler, M. D. Levine, 0. C.

Uhlenbeck, D. M. Crothers, and J. Gralla. 1973. Improvedestimation of secondary structure in ribonucleic acids. Nature(London) 246:40-41.

33. Yansura, D. G., and D. J. Henner. 1984. Use of the Escherichiacoli lac repressor and operator to control gene expression inBacillus subtilis. Proc. Natl. Acad. Sci. USA 81:439-443.

34. Zoller, M. J., and M. Smith. 1983. Oligonucleotide-directedmutagenesis of DNA fragments cloned into M13 vectors. Meth-ods Enzymol. 100:468-500.

J. BACTERIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/j

b on

17

Dec

embe

r 20

21 b

y 18

6.64

.71.

54.