JOURNAL OF BACTERIOLOGY, Mar. 1991, p. 1634-1641 Vol. 173, No. 50021-9193/91/051634-08$02.00/0Copyright © 1991, American Society for Microbiology

Transcription Attenuation-Mediated Control of leu OperonExpression: Influence of the Number of Leu Control Codons

JOANNE M. BARTKUS,t BONNIE TYLER, AND JOSEPH M. CALVO*Section ofBiochemistry, Molecular and Cell Biology, Cornell University, Ithaca, New York 14853

Received 9 July 1990/Accepted 15 December 1990

Four adjacent Leu codons within the leu leader RNA are critically important in transcription attenuation-mediated control of leu operon expression in Salmonella typhimurium and Escherichia coli (P. W. Carter, D. L.Weiss, H. L. Weith, and J. M. Calvo, J. Bacteriol. 162:943-949, 1985). The leader region from S. typhimuriumwas altered by site-directed mutagenesis to produce constructs having between one and seven adjacent Leucodons, all CUA. leu operon expression was measured in strains containing six of these constructs, eachintegrated into the chromosome in a single copy. Operon expression was sufficiently high that all strains grewin minimal medium unsupplemented by leucine. Expression of the operon was measured in stains cultured insuch a way that their growth was limited by the intracellular concentration of either leucine or of leucyl-tRNA.In general, the leu operon for each construct responded similarly to the parent construct in terms of the degreeof expression as a function of the degree of limitation. However, a strain containing (CUA)1 and, to a certainextent, a strain having (CUA)2 responded somewhat more sluggishly and strains containing (CUA)6 and(CUA)7 responded more sensitively to limitations than did the parent construct. In addition, DNA fragmentscontaining the leu promoter and leader region were used as templates in in vitro transcription reactionsemploying purified RNA polymerase. With nucleoside triphosphate concentrations of 200 ,uM, RNA polymer-ase paused during transcription of the leu leader region at a site about 95 bp downstream from the site oftranscription initiation. The halftimes of the pause were 1 min at 37°C and 3 min at 22rC. The pause waslengthened substantially when the GTP concentration was lowered to 20 ,uM. Our results are interpreted mosteasily in terms of an all-or-none model. Given two Leu control codons, the operon responds with nearlymaximum output over a wide range of leucine limitation, and that outcome does not change much withincreasing numbers of control codons.

Expression of the leucine operon of Salmonella typhimu-rium is regulated by a transcription attenuation mechanism(9). The leu operon consists of a promoter, a leader region,and four structural genes. For cells grown in a minimalmedium, most transcription terminates at an attenuator sitelocated at the end of the leader region, giving rise to a160-nucleotide leader RNA. The leader RNA contains trans-lational start and stop signals, a cluster of four Leu codons(termed control codons), and overlapping regions of dyadsymmetry that are capable of forming stem-and-loop struc-tures. These features are common to a number of operonsknown to be regulated by transcription attenuation (for arecent review, see reference 14). Gemmill et al. (9) proposedthat whether or not the structural genes of the leu operon aretranscribed is determined by which of these stem-and-loopstructures form. Figure 1A defines these stem-and-loopstructures. Formation of a terminator stem and loop leads totranscription termination, whereas prior formation of a pre-emptor stem and loop is thought to preclude formation of theterminator, thereby allowing transcription to continue pastthe attenuator and through the structural genes of theoperon. Which of these stem-and-loop structures form isdependent upon the progress of a ribosome translating theleader transcript. The preemptor structure can form onlywhen a translating ribosome has stalled over one of the fourleucine control codons, as is the case when cells are grownunder conditions of leucine limitation.

* Corresponding author.t Present address: Department of Microbiology, University of

Minnesota Medical School, Minneapolis, MN 55450.

The results of a number of experiments support the modelfor attenuation summarized above (2, 3, 8, 22, 23). Ofparticular significance for the work described here is theevidence supporting an important role for the four tandemLeu codons. In a mutant strain in which the four Leu controlcodons were replaced by four Thr codons, the leu operon didnot respond to a leucine limitation but instead responded toa limitation for threonyl tRNAThr (3). The nature of the Leucontrol codons is also critical for proper function of theattenuation mechanism. In the S. typhimurium leu leader,three of the four tandem codons are CUA, whereas inEscherichia coli all four of the control codons are CUA (9,23). Changing the rarely used CUA codons of the S. typhi-murium leader to the more frequently used CUG codonsreduced the basal level of operon expression and the sensi-tivity with which the operon responded to leucine limitation(2).

In this paper we investigate another question related tocodon usage, namely, the relationship between the numberof control codons and the expression of the operon. Todetermine how many Leu codons are required for a sensitiveand specific response to a leucine limitation, the controlcodons in the leu leader were first changed from(CUA)3CUC to (CUA)4 and then the number ofCUA codonswas varied from one to seven. Each construct was intro-duced into the chromosome in a single copy, and theexpression of the leu operon was measured in cells grownunder different conditions of leucine limitation. To summa-rize our results, with the exception of (CUA)1, all of theconstructs responded similarly to the wild type in responseto leucine limitation. The implications of this finding to

1634

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

Leu CONTROL CODONS 1635

various models of regulation by transcription attenuation arediscussed.

MATERIALS AND METHODS

Oligonucleotide-directed mutagenesis. Mutations were con-structed by the uracil incorporation method of Kunkel (13)as follows. Bacteriophage PS15 contains a 5.1-kb fragmentof S. typhimurium DNA carrying leuPLABCD' (D' meansthat part of leuD is missing) in the unique EcoRI site ofsingle-stranded phage flR229 (3). DNA from phage PS15grown in E. coli BW313 (dut ung) was used as a template foroligonucleotide-primed DNA synthesis as described byGillam and Smith (10). After transfection of E. coli JM101(made competent by the method of Hanahan [11]), phagewere analyzed by hybridization to identify those having thedesired mutation. Procedures for transfection, growth ofphage from plaques, and binding of phage DNA to nitrocel-lulose filters were as described by Miyada et al. (18).Hybridizations to 32P-labeled oligonucleotides were carriedout at 37°C, and filters were washed with 3 M tetrameth-ylammonium chloride at a temperature appropriate to thelength of the oligonucleotide probe as described by Wood etal. (26). Single-stranded DNA from the mutant phage wassequenced by the dideoxy method of Sanger et al. (20) with5'-CCGCCCACCGGTCTA-3' as a primer. The transfer ofmutations from fl phage to lambda phage by recombinationwas as described by Carter et al. (3), except that hybridiza-tions to 32P-labeled oligonucleotides to verify the presence ofmutations were done as described above. An oligonucleotidewith the sequence 5'-ATGCGTTT'AGTAGTA-3' was usedto change the (CUA)3CUC control codons to (CUA)4. Witholigonucleotide 5'-TAGTAGTAGTAGTAGCCCAGT-3' asa primer and template DNA prepared from the fl phage with(CUA)4 control codons, mutant constructs were obtainedhaving either one, two, or three additional CUA codons. Thereason for this is that pairing between redundant CUAcodons on the template and TAGs on the primer can occur ina number of different ways. Mutants having 5, 6, and 7 CUAcodons were obtained from this single mutagenesis experi-ment in the ratio 1:11:2. The oligonucleotides used to makethe (CUA)1, (CUA)2, and (CUA)3 mutants were, respec-tively, 5'-ATGCGTTTAGCCCAG-3', 5'-GCGTTTAGTAGCCC-3', and 5'-TGCGTTTAGTAGTAGCCCAGT-3'.

Limitations for leucine and leucyl-tRNA. Cells were grownin SSA salts (1) containing 5 ,ug of thiamine per ml and 0.2%glucose (minimal medium). Starvation for leucine wasachieved in one of two ways. Growth in chemostats at 37°Cufider conditions of leucine limitation was done essentiallyas described by Carter et al. (3). Cells were removed foranalysis after at least three doublings. The second method ofachieving a leucine limitation involved growing cells in amedium containing excess isoleucine and valine and a limit-ing amount of leucine (5). An early-stationary-phase culturegrown in minimal medium containing 50 ,ug of L-leucine perml was centrifuged, washed with SSA salts, suspended in 0.2volume of SSA, and used to inoculate fresh minimal mediumto an A6. of 0.03. This latter medium contained L-isoleucine(50 ,ug/ml) and L-valine (100 ,ug/ml) and either 10, 11, 13, or15 ,ug of L-leucine per ml. Cells were grown at 37°C withshaking, and samples were removed for enzyme assays atA6ws between 0.18 and 0.26.

Limitation for leucyl-tRNA was achieved by growingstrains containing a temperature-sensitive leucyl-tRNA syn-thetase (leuS31) mutation (15) at temperatures between 20and 37°C. The introduction of the leuS31 mutation into

strains by cotransduction with TnJO was described previ-ously (2). Cells were grown at 20°C in minimal mediumcontaining 50 ,ug of L-leucine per ml to an A600 of 1.0 anddiluted 1:20 into the same prewarmed medium containing, inaddition, 50 ,ug of L-isoleucine and 100 ,ug of L-valine per mland incubated at an elevated temperature. The reasons forincluding isoleucine and valine in the medium were de-scribed earlier (2).Leucine operon expression was assessed by measuring the

specific activity of the leuB gene product, 1-isopropylmalatedehydrogenase (13-IPM dehydrogenase; EC 1.1.1.85). As-says, performed in duplicate, were done with a permeabi-lized cell assay (21) measuring the amount of a-ketoisocap-roate formed as the 2,4-dinitrophenylhydrazone derivative.Activity is the change in A540 in 15 min (1 U of activitycorresponds to 9 nmol of ketoisocaproate). Specific activityis activity per 5 x 108 cells. Cell number was determined bylight scattering at A6. with a Zeiss PM6 spectrophotometer.

In vitro transcription. Reaction mixtures contained 5 nMtemplate DNA, 20 mM Tris-acetate (pH 7.9), 4 mM magne-sium acetate, 0.1 mM EDTA, 0.1 mM dithiothreitol, 100 mMKCl, ribonucleotide triphosphates (NTPs; 200 ,uM unlessstated otherwise) and RNA polymerase holoenzyme (10 nM;Boehringer Mannheim). In some experiments, RNasin(Promega) was present at 0.5 U/4l. One 32P-NTP (finalspecific activity, 1 to 2 ,Ci/nmol) was included to radiolabelnewly synthesized RNA. All experiments consisted of asynchronized single round of transcription. An open com-plex was formed between RNA polymerase and templateDNA during a 10-min incubation in a reaction mixturecontaining only two or three NTPs, and then elongation wasinitiated by the addition of the remaining NTPs together withrifampin (10 jig/ml, final concentration). Transcription wasterminated by the addition of 3 volumes of an ice-cold stopbuffer (0.66 M Tris-Cl [pH 7.9], 13.3 mM EDTA, 266 p,g oftRNA per ml), and nucleic acid was precipitated withethanol. Samples were fractionated on a 6 or 8% polyacryl-amide gel (acrylamide/bisacrylamide ratio, 29:1) containing 8M urea. After autoradiography, slices were excised from thegel, and Cherenkov radioactivity was determined in a scin-tillation counter. When the time course of RNA synthesiswas examined, 25-,ul samples were removed from the tran-scription reaction at various times into 75 RI of stop buffer.The template DNA, a 460-bp HaeIII fragment from plas-

mid pCV21 (8), contained the leu promoter and the leaderregion from S. typhimurium. To simplify the preparation ofthe template, the indicated HaeIII fragment was cloned intothe HinclI site of plasmid pIB176, yielding plasmid pCV151.The latter plasmid was cut with EcoRI and HindIII, the endswere made blunt by treatment with reverse transcriptase anddeoxynucleoside triphosphates, and the desired fragmentwas separated from the vector DNA by differential precipi-tation with polyethylene glycol (16).

RESULTS

Construction of mutants having different numbers of Leucontrol codons. The nucleotide sequence of the leu leaderRNA from S. typhimurium is shown in Fig. 1A, with the fourLeu control codons highlighted in boldface type. Our strat-egy for constructing mutants having different numbers ofLeu control codons (Fig. 1B) was to use oligonucleotide-directed mutagenesis to convert the wild-type sequence,(CUA)3CUC, to (CUA)4 and then to prepare constructshaving from one to seven CUAs from the construct having(CUA)4. An analysis of steady-state RNA secondary struc-

VOL. 173, 1991

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

1636 BARTKUS ET AL.

A

B

(CUA).

/) ~~~~~~~I/OR PROTECTOR TERMIN

I1 (CUA)2 (CUA)3

ATOR

(CUA)3CUC -4 (CUA)4

(CUA)S (CUA)e (CUA)7

FIG. 1. Control codons within the leu leader. (A) Nucleotidesequence within the leu leader (nucleotides 12 through 160). Thestructure at the right is folded in such a way as to emphasize theprotector and terminator stem and loop structures, whereas the oneon the left emphasizes the preemptor stem and loop. Dots representstandard base pairs. The four Leu codons are represented in larger,boldface letters. (B) Strategy by which the control codons were

changed by site-directed mutagenesis. Note that for each of theseven constructs, the sequence of the leader RNA is identical to thatshown in part A from nucleotides 1 through 53 and 66 through 160.

tures of leader RNAs transcribed from these constructs wasmade by using the algorithm of Williams and Tinoco (24).This analysis indicated that the changes introduced bymutation would have little effect on secondary structure ofthe leu leader. For each construction, once the mutation wasintroduced by oligonucleotide-directed mutagenesis andconfirmed by nucleotide sequencing, it was transferred byrecombination to the leu operon carried on a phage lambdaderivative, and single lysogens were prepared in a straindeleted for the leu operon (CSH73). The designations for theresulting strains are given in Table 1. Note that these strainsare leucine auxotrophs, because the leu operon on phagelambda is missing part of leuD.The effect of the mutations upon the expression of the leu

operon was assessed by measuring of one of the leu-specificenzymes, 1-IPM dehydrogenase, in cells grown with excessleucine or under conditions in which growth of the cells wasreduced because of a limitation for leucine or leucyl tRNA.Under such conditions of growth limitation, expression ofthe wild-type operon is elevated. Three methods were em-

ployed for achieving a growth limitation. The results ofexperiments employing each method are detailed below.

Limitation for leucine in chemostats. Leucine limitation

TABLE 1. Strains used in this study

Strain Control Phage leuS HotStrain codons designation allele Hosta

CSH73 None None + CSH73CV745 (CUA)3CUC PC-O + CSH73CV868 (CUA)3CUC PC-O IeuS31 CSH73CV943 (CUA)4 BC-4 + CSH73CV944 (CUA)6 BC-7 + CSH73CV946 (CUA)7 BC-6 + CSH73CV948 (CUA)4 BC-4 + C600CV958 (CUA)6 BC-6 IeuS31 CSH73CV959 (CUA)7 BC-7 leuS3l CSH73CV961 (CUA)3 BC-3 + CSH73CV962 (CUA)3 BC-3 leuS31 CSH73CV963 (CUA)1 BC-1 + CSH73CV964 (CUA)2 BC-2 + CSH73CV965 (CUA)1 BC-1 leuS31 CSH73CV966 (CUA)2 BC-2 leuS31 CSH73CV967 (CUA)1 BC-1 + C600CV968 (CUA)2 BC-2 + C600

0 CSH73, HfrH Alac (ara-leu) thi; C600, F- thi-l thr-l leuB6 lacYl tonA21supE44.

over a wide range of growth rates was achieved by growingauxotrophs in a chemostat under conditions in which thegrowth rate depended upon the rate at which fresh leucine-containing medium was introduced (3). After at least threedoublings at a particular growth rate, expression of the leuoperon was assessed by removing small samples from thechemostat and measuring the specific activity of ,B-IPMdehydrogenase (encoded by leuB). When the parent strain,CV745, was repeatedly grown in a chemostat at the samedegree of growth limitation, sizeable variation in leuBexpression was observed. For example, the average andstandard deviation for eight chemostats run at generationtimes between 135 and 152 min were 2.68 + 1.18. We werenot able to determine the source of this variation, but, afteranalyzing the results of strain CV745 run in many chemo-stats over a wide range of growth limitations, we concludedthat expression was independent of the degree of growthlimitation over the range of growth limitations that we used.We assume that this is also the case for the mutant con-structs that we studied. Examples of data obtained for twosuch constructs are shown in Fig. 2. Although the variationalluded to above is evident, it is clear that for the (CUA)construct, derepression in response to a leucine limitation isgreater than for the (CUA)1 construct. The data from che-mostat experiments for all of the constructs tested are givenin Table 2, presented as averages from a number of experi-ments. With the exception of the construct containing asingle CUA codon, all of the constructs behaved similarly interms of their response to a leucine limitation. The level ofoperon expression for the (CUA)1 construct was only a fifththat of the other constructs.Leudne lmitation achieved t comp for entry.

E. coli has two transport systems for leucine that have Kmvalues in the micromolar range (6). Even at leucine concen-trations as low as 1 pg/ml (7.5 p,M), these transport systemsare nearly saturated, and the expression of the leu operon isexpected to be repressed. Thus, at concentrations of leucinethat are low enough to achieve derepression (<1 ,ug/ml),auxotrophs quickly deplete the medium of leucine, and it isnot practical to measure the extent of derepression undersuch conditions. However, Freundlich et al. (5) showed thata leucine limitation could be achieved at leucine concentra-

J. BACTERIOL.

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

Leu CONTROL CODONS 1637

0coc

0

0o

0

I-

'D

0

>00

0

co.

4.0 -

3.0 -

2.0 -

1.0 '

9,0 ;120

C:

00

0

0

0)0

c

I-

40

0.

~0

L.m-

0

4)C,)

2.5 -

2.0 -

1.5 -

1.0 -

0.5 -

0.0 -

0.06 0.07 0.08 0.09 0.10 0.11150 180 210

Generation time (min)FIG. 2. Ieu operon expression in strains CV963 (CUA)1 (E) and

CV946 (CUA)7 (*) grown in a chemostat under conditions ofincreasingly severe leucine limitation. Cells were grown in a che-mostat containing leucine at 8 ,ug/ml at the indicated generationtimes. The generation time is the time required to replace the culturevolume multiplied by the natural logarithym of 2. Specific activity isunits of enzyme per 5 x 108 cells.

tions in the range of 10 to 15 ,ug/ml by growing an auxotrophin a medium containing excess valine (100 ,ig/ml) and iso-leucine (50 ,ug/ml). The presumed basis for this method isthat valine and isoleucine compete with leucine for entrythrough the LIV system (6). This method is better suitedthan the chemostat for achieving a mild leucine limitation,but it is not well suited for studying cells that are severelylimited for leucine. For this reason, we applied this proce-dure to strains that we had reason to believe responded moresensitively to a mild degree of limitation, namely, strainshaving six or seven control codons.

Early-stationary-phase cells were diluted to an A6. of 0.03with fresh medium containing excess valine and isoleucineand a limited amount of leucine. After between two andthree doublings (cell density such that less than 20% of theleucine had been taken up), samples were removed andassayed for P-IPM dehydrogenase activity. In these experi-

TABLE 2. Expression of the leu operon in cells limited forleucine by growth in a chemostat

No. of CUA Sp act of P-IPM Generation timebStrain codons dehydrogenasea (min)

CV745 (CUA)3CUC 3.2 ± 1.4 (18) 80-212CV963 1 0.58 ± 0.19 (7) 105-162CV964 2 2.9 + 0.64 (7) 97-154CV943 4 2.9 ± 1.4 (8) 81-216CV944 6 3.8 ± 0.70 (4) 77-210CV946 7 3.0 ± 0.57 (6) 81-203

a Specific activity in units per 5 x 108 cells ± the standard deviation. Thenumber of determinations is given within parentheses.

b Time required to replace the culture volume of the chemostat multipliedby the natural logarithm of 2. For each strain, a number of chemostats wererun at different generation times. The range of generation times is shown.

[leuclne]V1FIG. 3. Ieu operon expression in strains grown under conditions

of leucine limitation achieved through competition for entry. Inoculawere grown in minimal medium containing 50 ,g of L-leucine per mland diluted to an A660 of 0.03 into minimal medium containingL-isoleucine (50 ,ug/ml) and L-valine (100 ,ug/ml). This culture wassplit into several parts, which received either 10, 11, 13, or 15 ,ug ofL-leucine per ml. Cells growth was followed by turbidity, andsamples were removed at an A6. between 0.18 and 0.22 fordetermination of enzyme activity. Specific activity is units per 5 x108 cells. Each point is an average from at least four experiments.Symbols: O, CV745 (CUA)3CUC; *, CV944 (CUA)6; *, CV946(CUA)7.

ments, comparisons were made for strains sampled at thesame cell density. As expected, expression of the wild-typeleu operon increased dramatically as cells became progres-sively starved for leucine (Fig. 3; note that the abscissa isexpressed as the inverse of the concentration of leucine and,therefore, that higher starvation is represented by morerightward points). The degree of starvation required to elicita given level of derepression depended upon the constructanalyzed: strains having six or seven CUAs required lessstarvation than did the parent construct having (CUA)3CUC.This result is consistent with the view that multiple controlcodons allow an operon to respond sensitively to a very lowdegree of amino acid limitation.

Limitation for leucyl tRNA in temperature-sensitive tRNAI'Usynthetase mutants. The leuS31 allele, which encodes atemperature-sensitive leucyl-tRNA synthetase, results intemperature-sensitive growth (17). This allele was trans-duced into stains described above carrying either the wild-type leu control region (CUA)3CUC or constructs havingfrom one to seven CUA control codons. Strains were grownat different temperatures, and the growth rates and differen-tial rates of syntheses of ,-IPM dehydrogenase were deter-mined for each temperature. Table 3 and Fig. 4A show thekind of data obtained for strain CV868 [leuS31, (CUA)3CUC] carrying the wild-type Leu control codons. At tem-peratures above 30°C, strain CV868 grew more slowly thanthe isogenic strain lacking the leuS31 allele (CV745; Table 3),presumably because the rate of synthesis of leucyl-tRNA instrain CV868 was not sufficient at the elevated temperaturesto sustain normal growth. However, even at 20°C, a temper-

UU U

a U

U

VOL. 173, 1991

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

1638 BARTKUS ET AL.

TABLE 3. Growth rates of strains CV745 and CV868 atdifferent temperatures

Doubling time (min) for cells grown at theStraina following temp (°C):

20 25 30 33 36

CV745 (leuS+) 240 138 110 105 57CV868 (IeuS3J) 218 142 113 141 200b

a Strains CV745 and CV868 are isogenic except for the leuS31 mutation.b At 36°C, growth was exponential to an A6w of about 0.5 and then began

to slow. The differential rate of synthesis was not linear when these cells weregrown at 36°C, and therefore differential rate data are not reported in Fig. 4and Table 4 for this temperature.

ature at which growth was not restricted, the leucine operonof strain CV868 was already substantially derepressed incomparison with the isogenic strain lacking the leuS31 allele(Fig. 4A, dashed line near the bottom of the figure). Thus, at20°C (and 25°C), leucyl-tRNA synthetase in strain CV868was probably already partially inactivated, resulting in re-duced levels of leucyl tRNA. The extent of the reductionwas not sufficient to affect the growth rate at 20 and 25°C, butit was sufficient to cause nearly maximum derepression ofthe leucine operon.With these considerations in mind, it is possible to inter-

pret the results of similar experiments employing strainshaving different numbers of Leu control codons. Figure 4Bshows data for strain CV966 (CUA)2. The leu operon instrain CV966 was derepressed by leucyl-tRNA limitation,but the degree of derepression at 20 and 25°C was less thanthat for the wild-type operon. Thus, reducing the number ofcontrol codons from four (CUA)3CUC to two (CUA)2 low-ered the sensitivity with which the operon responded to aleucyl-tRNA limitation.The results of similar experiments with other strains,

represented as differential rates of synthesis (slopes ofcurves of type shown in Fig. 4B), are shown in Table 4. Forstrains having one or two CUA codons, the extent ofderepression was lower than that for the parent (CUA)3CUCat each of the temperatures employed. Strains having four,six, or seven CUA codons showed a different pattern ofexpression: at low degrees of leucyl-tRNA limitation (20 or25°C), expression was higher than that for the parent; but athigher degrees of limitation, expression was about the sameas that for the parent. These results suggest that an increasein the number of control codons increases the sensitivitywith which the operon responds to a leucyl-tRNA limitation.Note, however, that constructs with four, six, and sevenCUAs behaved similar to one another. Thus, it does notseem to be the case that each additional CUA codon abovefour imparts an additional measure of sensitivity to operonexpression.

Expression of the leu operon in cells grown with excessleucine. To investigate the influence of the number of controlcodons upon the basal level of operon expression, cells weregrown in a minimal medium containing excess leucine, acondition in which expression is maximally repressed. Con-structs having one or two CUA codons showed a reducedlevel of operon expression [as much as a twofold reductionfor (CUA)1] whereas the other constructs having three ormore CUA codons all showed about the same basal level ofexpression (Table 5).

In vitro transcription of the leu leader region. In consider-ing some models for transcription attenuation control of theleu operon (discussed below), it became important to know

E1-

Un0-

-

0

0m

C0

0.

10 -

8-

6-

0.25 0.50 0.75 1.00

0.25 0.50 0.75 1.00

OD 600FIG. 4. Differential rates of synthesis of 1-IPM dehydrogenase in

strain CV868 [leuS31; (CUA)3CUC] (A) and CV966 [leuS31;(CUA)2] (B) grown at different temperatures. For panel A, the lowerline is fit to the filled diamonds (-) (20°C). Data for strain CV745grown at 20, 25, 30, and 33°C (---) are included in both panels.Inocula were grown at 20°C in minimal medium containing 50 ,ug ofL-leucine per ml and diluted 1:20 into the same medium containing inaddition 50 Fg of L-isoleucine per ml and 100 ,ug of L-valine per mland incubated at the same or elevated temperature. Cell growth wasmonitored by A6.. Samples were removed at the indicated times fordetermination of enzyme activity. Symbols: *, 20°C; *, 25°C; A,30°C; O, 33°C.

whether RNA polymerase pauses at a specific site duringtranscription of the leu leader. At least in vitro, this is knownto be the case for the trp (25), thr (7), ilvB (12), ilvGMEDA(12), and his (4) operons. For these latter cases, the pausingsite is located immediately distal to a stem and loop withinthe leader RNA termed the 1:2 stem and loop. It waspostulated by Yanofsky and co-workers (14) that pausingduring transcription allows time for a ribosome to initiatetranslation and serves to coordinate the movements ofRNApolymerase and ribosome so that the attenuation mechanismworks efficiently.The time course of RNA synthesis from the leu leader

J. BACTERIOL.

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

Leu CONTROL CODONS 1639

TABLE 4. Expression of the leu operon in strains limitedfor leucyl-tRNA

,B-IPM dehydrogenase synthesisa inNo. of CUA cells grown at the following tempStrain codons ___

20 25 30 33

Exp 1CV868 (CUA)3CUC 1 1 1 1CV965 1 0.45 0.47 0.54 0.71CV966 2 0.36 0.64 0.67 0.63CV962 3 0.75 0.94 1.2 1.1

Exp 2CV868 (CUA)3CUC 1 1 1 1CV957 4 1.6 1.2 0.79 0.84CV958 6 1.7 1.4 0.98 1.1CV959 7 1.3 1.2 0.74 1.1a Differential rate of synthesis (slope of activity per milliliter versus cells

per milliliter). Data for strain CV868 and CV966 (experiment 1) are shown inFig. 4. Values for CV868 for each temperature were normalized to 1.Differential rates for CV868 for experiment 1 at 20, 25, 30, and 33°C were,respectively, 7.9, 15, 12, and 17. For experiment 2 they were 2.8, 4.8, 7.8, and7.1, respectively. The difference in the two sets of data is related to one of thecomponents of the assay, which was not optimal in experiment 2.

region was examined in vitro by using purified RNA poly-merase and a 460-bp restriction fragment containing the leupromoter and leader region. Figure 5 shows the RNA speciespresent at various times during a single round of transcrip-tion carried out at 37°C with 200 ,uM NTPs. The 160-bp leuleader RNA (8) was visable after incubation for 30 s. Attimes greater than 45 s, the leader RNA was seen to be threespecies, representing heterogeneity at the 3' end (8). Since invitro transcription of the leu leader region terminates at theattenuator site with an efficiency of about 98% (9), tran-scripts longer than the leader RNA are not present in largeamounts. Of particular interest is an RNA about 95 bases inlength that was seen as early as 15 s after the initiation oftranscription. The length of this species was estimated fromRNA size standards included in the same gel. This RNAincreased in amount over the next 45 s, and its disappear-ance correlated with an increase in the amount of leaderRNA. These kinetics are consistent with the idea that the95-base species is a nascent RNA that accumulates becauseof polymerase pausing at a site about 95 bp downstream fromthe site of transcription initiation. To confirm that the95-base species arose from transcription originating at theleu promoter, these experiments were repeated with tem-plates that were truncated at their promoter-distal end.

TABLE 5. Expression of the leu operon in cells grown in amedium containing excess leucine

Strain No. of CUA Sp act 3-IPMcodons dehydrogenasea

CV745 (CUA)3CUC 0.18 + 0.04 (20)CV963 1 0.08 ± 0.03 (19)CV964 2 0.12 t 0.04 (19)CV961 3 0.16 t 0.03 (18)CV943 4 0.17 t 0.09 (33)CV944 6 0.22 t 0.06 (15)CV946 7 0.22 ± 0.08 (14)

a Strains were grown in minimal medium containing 50 jLg of leucine per ml.Specific activity is units per 5 x 10i cells. Averages are given t the standarddeviation, and the number of determinations is given within parentheses.

Lane: 1 2 3 4 5 6 7 8 9

p 4111S0*- LT

Min - -.25.5 1 2 3 5 7.5FIG. 5. Time course of in vitro transcription from a leu-contain-

ing template. A HaeIIIf4 fragment from plasmid pCV151, contain-ing the leu promoter and leader region, was used as a template.Reactions were started by the addition of nucleotides and rifampin(zero time). Lanes 1 and 2 contain RNA standards, 90, 98, 111, 133,and 141 nucleotides long, synthesized in parallel transcription reac-tions. Arrows denote the position of the leu pause transcript (PT)and the leu leader transcript (LT).

Transcription from a template cut with TaqI (which createsa runoff transcript 120 bases long) gave rise to the 95-basespecies in addition to the expected runoff transcript, whereastranscription from a template cut with MspI (which creates arunoff transcript 90 bases long) gave rise to a single tran-script about 90 bases long (data not shown).The half time for pausing was measured for transcription

carried out under different conditions. Under standard tran-scription conditions at 37°C with 200 ,uM NTPs, the half timewas 1 min (data not shown). When the nucleotide concen-tration was maintained at 200 ,uM but the temperature waslowered to 22°C, the half time for pausing increased to 3 min(Fig. 6). At 22°C, the pause was further extended when theconcentration of CTP or GTP was reduced to 20 ,uM (halftime of 4.5 or >30 min, respectively). The exceptionally longpause associated with reduced GTP concentration suggeststhat GTP is the next nucleotide to be added to the nascenttranscript within a paused complex.

DISCUSSIONTo our knowledge, the work described here represents the

first study in which the number of control codons within anattenuation-controlled operon was systematically varied.The goal of these studies was to assess the influence of thenumber of such control codons on the efficiency and sensi-tivity of the control mechanism.We did observe some differences in operon response for

constructs having from one to seven CUA control codons.For example, in experiments in which growth in chemostatswas limited by the availability of leucine, a strain having onlya single CUA control codon was derepressed to only aboutone-fifth the level of the wild-type strain (Table 2). Inaddition, strains having one or two CUA control codons did

VOL. 173, 1991

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

1640 BARTKUS ET AL.

0)

'Eco

0.

0

4)

a)

o0

100

50

10

a

0 8 1 24

Time after rifampicin (min)FIG. 6. Rate of disappearance of pause transcript RNA during

transcription in vitro at 22°C. Experiments were performed as

described in the legend to Fig. 5. After autoradiography, regions ofthe gel containing the pause transcript were excised, and theradioactivity was quantitated. For each of these experiments, thesample having the highest radioactivity was normalized to 100%b.

Symbols: C, all NTPs present at 200 FM; *, CTP present at 20 FMand all other NTPs present at 200 FM; *, GTP present at 20 ,uM andall other NTPs present at 200 ,uM.

not respond as sensitively as the wild-type strain to a slightlimitation for leucine (Table 4, Fig. 3). On the other hand,strains having six or seven CUA codons responded some-

what more sensitively to low degrees of leucine than didstrains having fewer numbers of control codons (Fig. 3,Table 4). Overall, however, we were struck more by howsimilarly these constructs behaved than by differencesamong them. With the (CUA)1 construct excepted, strainscontaining these constructs responded very similarly to a

leucine limitatipn created by growth in a chemostat (Table2). For experiuhents employing the leuS31 allele, the differ-ences between the construct-s were generally less than two-fold (Table 4).We considered, the results of these experiments in the

context of a model for attenuation control of transcription(9). In this model, the regulatory mechanism depends uponthe position of a ribosome on the leader RNA at the momentthat RNA polymerase reaches the attenuator site. If theribosome were upstream of the control codons, then termi-nation of transcription was predicted, due to formation of asecondary structure within the leader RNA termed theprotector stem and loop (Fig. 1A). If the ribosome were

downstream of the control codons, then termination oftranscription was again the predicted result, in this case

because the preemptor stem and loop was unable to form.The only situation in which readthrough was the predictedoutcome was the case in which translation was slowed orarrested at the Leu control codons. In this last-mentionedcase, with part of the protector region masked by theribosome, the preemptor region was free to form (Fig. 1A).Given this general model, it is relevant to ask how the

number of control codons might affect the regulatory mech-

anism. One possibility is that the number of control codonsdetermines the sensitivity of the response of an operon toleucine limitation. Under conditions of leucine limitation,the RNA polymerase may continue transcription as theribosome pauses at successive Leu codons. Pauses at eachcodon are summed until the polymerase reaches the attenu-ator (about 2 s is required for a polymerase moving at 45nucleotides per s), and the decision is made regardingtermination. A short pause time caused by a very smalldegree of leucine limitation would be summed seven timesfor a leader having seven codons but only twice for a leaderhaving two control codons. Thus, the operon having thelarger number of control codons might be expected torespond to a lower degree of starvation than the operonhaving fewer control codons.The reasoning developed above leads to an interesting

prediction: operons having a large number of control codonsmight respond to a very strong limitation for leucine morepoorly than operons having fewer control codons. Considera degree of starvation sufficiently strong that a ribosomewere stalled over the first of seven Leu codons within the leuleader. In that position, the ribosome would not be in aposition to mask the protector, and the outcome is predictedto be transcription termination. This prediction is not borneout by the results. As demonstrated in Tables 2 and 4,constructs having six or seven CUA control codons did notresult in reduced expression under conditions of severeleucine limitation in comparison with the parent construct.Another possibility is that this attenuation mechanism is

relatively insensitive to the number of control codons andthat, once a limitation for leucine is sensed, the operonresponds fully. Such an all-or-none mechanism can be imag-ined in concert with ideas developed by Yanofsky andco-workers about the importance of polymerase pausing tothe attenuation mechanism (14). They pointed out thatpausing during transcription allows time for a ribosome toinitiate translation and that subsequent movement of theribosome may release the polymerase from the paused state.Thus, pausing may serve to coordinate the movements ofRNA polymerase and ribosome so that the attenuationmechanism works efficiently. The evidence for pausing andfor the significance of pausing to the trp attenuation mecha-nism is compelling (15).

Pausing occurs in vitro during transcription of the trp (25),thr (7), ilvB (12), ilvGMEDA (12), and his (4) operons; ineach of these cases, the pausing site is located immediatelydistal to a stem and loop within the leader RNA termed the1:2 stem and loop (denoted as protector stem and loop in Fig.1A). We show here that during transcription of the leu leaderregion in vitro, RNA polymerase pauses at a site about 95 bpdownstream of the site of transcription initiation at a posi-tion that is comparable to that for the trp, thr, ilvB, andilvGMEDA operons. The tendency of RNA polymerase topause at this site is relatively strong; the halftime for pausingis 1 min at 37°C with 200 ,uM NTPs. For some other systemsstudied, equivalent pause times were only obtained at lowertemperatures or lower concentrations of NTPs (4, 12, 25).

If the ribosome indeed accelerates release of the polymer-ase from a paused complex, at what point during translationdoes this effect take place? Conceivably, a ribosome pausedover the first of seven Leu control codons does not providethe signal for polymerase release. Even under conditions ofa severe limitation for leucine, the ribosome is expected tomove to the next codon at some point, either because ofmisincorporation (19) or more likely because of leucinereleased during protein turnover. By this view, the polymer-

J. BACTERIOL.

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

Leu CONTROL CODONS 1641

ase would remain paused until the ribosome approachedclose enough to signal its presence. If this is the case, thenextra Leu codons beyond some minimum number may havelittle effect upon the overall mechanism.An analysis of the basal level of operon expression leads

to a similar conclusion. There was a significant reduction inthe basal level of expression with a construct having one, orpossibly two, CUA codons. This suggests that an importantdeterminant of basal level expression is the rate of ribosomemovement through the leader peptide-coding region at nor-mal levels of amino acyl tRNAs. The fact that basal-levelexpression was nearly the same for constructs having fromthree to seven CUA codons may be rationalized by postu-lating that, for constructs having more than three or fourCUA codons, RNA polymerase did not escape the pause siteuntil the ribosome had moved past the first several CUAcodons.Our results are interpreted most easily in terms of the

all-or-none model. Given two Leu control codons, the op-eron responds with nearly maximum output over a widerange of leucine limitation and that outcome does not changemuch with increasing numbers of control codons.

ACKNOWLEDGMENTS

This work was performed with the technical assistance of PatriciaElliott.The research was supported by Public Health Service grant

GM38898 from the National Institutes of Health to J.M.C.

REFERENCES1. Calvo, J. M., M. Freundlich, and H. E. Umbarger. 1969.

Regulation of branched-chain amino acid biosynthesis in Sal-monella typhimurium: isolation of regulatory mutants. J. Bac-teriol. 97:1272-1282.

2. Carter, P. W., J. M. Bartkus, and J. M. Calvo. 1986. Transcrip-tion attenuation in Salmonella typhimurim: the significance ofrare leucine codons in the leu leader. Proc. Natl. Acad. Sci.USA 83:8127-8131.

3. Carter, P. W., D. L. Weiss, H. L. Weith, and J. M. Calvo. 1985.Mutations that convert four leucine codons of the Salmonellatyphimurium leu leader to four threonine codons. J. Bacteriol.162:943-949.

4. Chan, C. L., and R. Landick. 1989. The Salmonella typhimu-rium his operon leader region contains an RNA hairpin-depen-dent transcription pause site. J. Biol. Chem. 264:20796-20804.

5. Freundlich, M., J. Trela, anid W. Peng. 1971. Evidence that themajority of leucine transfer ribonucleic acid is not involved inrepression in Salmonella typhimurium. J. Bacteriol. 108:951-953.

6. Furlong, C. E. 1987. Osmotic shock-sensitive transport sys-tems, p. 768-796. In F. C. Neidhardt, J. L. Ingraham, K. B.Low, B. Magasanik, M. Schaechter, and K. E. Umbarger (ed.),Escherichia coli and Salmonella typhimurium: cellular andmolecular biology, vol. 1. American Society for Microbiology,Washington, D.C.

7. Gardner, J. F. 1982. Initiation, pausing, and termination oftranscription in the threonine operon regulatory region of Esch-erichia coli. J. Biol. Chem. 257:3896-3904.

8. Gemmill, R. M., J. W. Jones, G. W. Haughn, and J. M. Calvo.1983. Transcription initiation sites of the leucine operon ofSalmonella typhimurium and Escherichia coli. J. Mol. Biol.170:39-59.

9. Gemmill, R. M., S. R. Wessler, E. B. Keller, and J. M. Calvo.1979. The leucine operon of Salmonella typhimurium is con-trolled by an attenuation mechanism. Proc. Natl. Acad. Sci.USA 76:4941-4945.

10. Gillam, S., and M. Smith. 1979. Site specific mutagenesis usingsynthetic oligodeoxyribonucleotide primers. I. Optimum condi-tions and minimum oligodeoxyribonucleotide length. Gene8:81-97.

11. Hanahan, D. 1983. Studies on transformation of Escherichia coliwith plasmids. J. Mol. Biol. 166:557-580.

12. Hauser, C. R., J. A. Sharp, L. K. Hatfield, and G. W. Hatfield.1985. Pausing of RNA polymerase during in vitro transcriptionthrough the ilvB and ilvGEDA attenuator regions of Escherichiacoli K12. J. Biol. Chem. 260:1765-1770.

13. Kunkel, T. A. 1985. Rapid and efficient site-specific mutagenesiswithout phenotypic selection. Proc. Natl. Acad. Sci. USA82:488-492.

14. Landick, R., and C. Yanofsky. 1987. Transcription attenuation,p. 1276-1301. In F. C. Neidhardt, J. L. Ingraham, K. B. Low,B. Magasanik, M. Schoechter, and K. E. Umbarger (ed.),Escherichia coli and Salmonella typhimurium: cellular andmolecular biology, vol. 1. American Society for Microbiology,Washington, D.C.

15. La Rossa, R., G. Vogeli, K. B. Low, and D. Soll. 1977. Regula-tion of biosynthesis of aminoacyl-tRNA synthetases and oftRNA in Escherichia coli. II. Isolation of regulatory mutantsaffecting leucyl-tRNA synthetase levels. J. Mol. Biol. 117:1033-1048.

16. Lis, J. T. 1980. Fractionation of DNA fragments by polyethyl-ene glycol induced precipitation. Methods Enzymol. 65:347-353.

17. Low, B., F. Gates, T. Goldstein, and D. Soll. 1971. Isolation andpartial characterization of temperature-sensitive Escherichiacoli mutants with altered leucyl- and seryl-transfer ribonucleicacid synthetases. J. Bacteriol. 108:742-750.

18. Miyada, C. G., X. Soberon, K. Itakura, and G. Wilcox. 1982.The use of oligodeoxyribonucleotides to produce specific dele-tions in the araBAD promoter of Escherichia coli B/r. Gene17:167-177.

19. Parker, J., J. W. Pollard, J. D. Friesen, and C. P. Stanners.1978. Stuttering: high level misstranslation in animal and bacte-rial cells. Proc. Natl. Acad. Sci. USA 75:1091-1095.

20. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.USA 74:5463-5467.

21. Searles, L. L., and J. M. Calvo. 1988. Permeabilized cell andradiochemical assays for ,-isopropylmalate dehydrogenase.Methods Enzymol. 166:225-229.

22. Searles, L. L., S. R. Wessler, and J. M. Calvo. 1983. Transcrip-tion attenuation is the major mechanism by which the leuoperon of Salmonella typhimurium is controlled. J. Mol. Biol.163:377-394.

23. Wessler, S. R., and J. M. Calvo. 1981. Control of le.u operonexpression in Escherichia coli by a transcription attenuationmechanism. J. Mol. Biol. 149:579-597.

24. Williams, A. L., and I. J. Tinoco. 1986. A dynamic programmingalgorithm for finding alternative RNA secondary structures.Nucleic Acids Res. 14:299-315.

25. Winkler, M. E., and C. Yanofsky. 1981. Pausing of RNApolymerase during in vitro transcription of the tryptophanoperon leader region. Biochemistry 20:3738-3744.

26. Wood, W. I., J. Gitschier, L. A. Lasky, and R. M. Lawn. 1985.Base composition-independent hybridization in tetramethylam-monium chloride: a method for oligonucleotide screening forhighly complex gene libraries. Proc. Natl. Acad. Sci. USA82:1585-1588.

VOL. 173, 1991

on February 11, 2021 by guest

http://jb.asm.org/

Dow

nloaded from