Vol. 173, No. 11JOURNAL OF BACTERIOLOGY, June 1991, p. 3382-338800('21-9193/91/113382-07$02.00/0Copyright © 1991, American Society for Microbiology

Genetic Analysis of the Attenuator of the Rhizobium melilotitrpE(G) Gene

YOUNG MIN BAEt AND GEORGE V. STAUFFER*

Department of Microbiology, University of Iowa, Iowa City, Iowa 52242

Received 22 October 1990/Accepted 1 April 1991

It was previously reported that transcription of the Rhizobium meliloti trpE(G) gene starts at the adenineresidue of the AUG codon of the leader peptide coding sequence (trpL), suggesting that translation of the trpLsequence starts without the Shine-Dalgarno sequence. We constructed mutations replacing the AUG codon ofthe trpL sequence with AAG or ACG. These mutations reduced the expression of a trpL'-'lacZ fusion gene to0.1 and 0.2% of the wild-type level, respectively, indicating that the AUG codon is the translation initiationcodon for the trpL coding sequence. In addition, these mutations, as well as a mutation converting the eighthcodon (UCG) of the trpL sequence to UGA, abolished regulation by attenuation when introduced upstream ofthe tandem tryptophan codons in a trpE'-'lacZ fusion. Mutations affecting the stability of the probableantiterminator and terminator secondary structures in trpL mRNA were also constructed. Studies using thesemutations indicate that the attenuator of R. meliloti functions in a way analogous to that of the Escherichia colitrp attenuator.

In Escherichia coli and many other enteric bacteria, thegenes necessary for the biosynthesis of the amino acidtryptophan are clustered in an operon, and this operon isregulated transcriptionally by both repression and attenua-tion (30-32). The combination of repression and attenuationresults in regulation over a range of about 600-fold, withrepression regulating up to 70-fold and attenuation regulatingabout 8- to 10-fold (30). However, this kind of dual control isnot the case for Rhizobium meliloti (1). R. meliloti hastryptophan biosynthetic genes in three clusters on its chro-mosome (2, 4, 7). Only one of these genes [trpE(G)] isregulated solely by attenuation, while the others are notregulated.

In E. coli, repression of transcription initiation occurs bybinding of the tryptophan-activated repressor to the trpoperator, which overlaps the trp promoter (30). Attenuation,however, controls termination of the mRNA in the leaderregion through the formation of mutually exclusive stem-loop structures (30, 32). One of the most common features ofthe bacterial trp attenuator is the presence of a short leaderpeptide coding sequence (trpL) which contains two or threetandem tryptophan codons (31, 32), although exceptionshave been found (10, 23). A deficiency of the amino acidtryptophan increases the ratio of uncharged to chargedtRNATrP. If uncharged tRNATrP enters into the ribosometranslating the trpL coding sequence at the tandem tryp-tophan codons, the ribosome stalls (30). The stalled ribo-some masks region 1 of the 1:2 stem, allowing formation ofthe 2:3 stem (antiterminator) (Fig. 1) (30, 31). When there isexcess tryptophan, charged tRNATrP is abundant, and theribosome translating the trpL coding sequence reaches theleader peptide stop codon without stalling on the tandemtryptophan codons (30, 31). The ribosome on the leaderpeptide stop codon prevents formation of the 2:3 stem bymasking region 2, and this in turn facilitates formation of the3:4 stem structure (terminator), which along with the follow-

* Corresponding author.t Present address: Department of Cellular and Developmental

Biology, Harvard University, Cambridge, MA 02138.

ing stretch of U's serves as a rho-independent transcriptionterminator (30, 31). Therefore, formation of the alternativestem-loop structures and translation of the trpL sequence bya ribosome, which is following RNA polymerase closely, iscrucial for the function of the attenuator.

In R. meliloti, the trp genes are clustered at three differentchromosomal locations; trpE(G), trpDC, and trpFBA (2, 4,7). Of these three clusters, only the trpE(G) gene seems to beregulated in response to tryptophan (2, 6). Bae et al. (2)reported the presence of attenuatorlike structures in theleader region of the R. meliloti trpE(G) gene from DNAsequencing data. It was also shown that the R. melilotitrpE(G) gene is regulated solely by attenuation in response totryptophan (1). In this report, we examine the significance ofthe previously postulated secondary structures and theleader peptide coding region (trpL) in R. meliloti by intro-ducing mutations in the trpL coding sequence, the 2:3 stem,and the 3:4 stem.

MATERIALS AND METHODSBacterial strains, plasmids, and media. The bacterial

strains and plasmids used in this study are listed in Table 1.The complete medium was TY medium (3) for Rhizobiumstrains and Luria broth (21) or 2x YT medium (14) for E. colistrains. M9 medium (15) was used as the minimal medium forRhizobium strains. The antibiotics used were streptomycin(800 ,ug/ml), tetracycline (5 ,ug/ml), and cycloheximide (25,g/ml).

In vitro site-directed mutagenesis. Site-directed mutagene-sis using synthetic oligonucleotides was done as describedpreviously (5, 9). For mutagenesis of the trpL'-'lacZ fusion,we cloned the 380-bp BamHI-PstI fragment of pYB49 intothe BamHI-PstI sites of phage M13mpl9 replicative form.For the mutagenesis of the trpE'-'lacZ fusion, we cloned a569-bp BamHI-PstI fragment of pYB56 into the BamHI-PstIsites of phage M13mpl9. A series of mutations were thenintroduced into the leader sequences by site-directed muta-genesis (see Fig. 3). All the mutations were confirmed byDNA sequencing (22). The mutagenized fragments werethen cloned back into pYB49 BamHI-PstI sites and used to

3382

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

R. MELILOTI trpE(G) ATTENUATOR 3383

A

1

G AU U

G U

C GG AU GA AC=GU-AGAUC=G

GAC=GU AUCC=G

G=CG=CC=GG=C 112

AUGGCAAACACGCAGAACAUUUCGAUCUGGUGGUGGGCUCGCUGAG=CCGGAGAUUUCGAGGCGGCUUUUI

B

2:3Antiterminator

G AU GC=GG=CC-GUC=G

G=C

GG=CuU

G. GGa -AGG=CG=C

CGA U-AU UC=GU AA-U GAUUAC G=C A U

A-UA G UC-G G=CG=C C=GC=G C=GAA-U C=GC=G G=C

A A C=GA-U C=GA-U G-C

AUGGC=GAGAGAUGGA-UUUUUU1:2 3:4

Terminator

4

FIG. 1. Predicted secondary structures of the leader mRNA inTerminator.

transform E. coli S17-1. The mutagenized plasmids werethen transferred to R. meliloti YB41 by biparental conjuga-tion as described previously (1).DNA manipulations. Subcloning experiments were done

by using the low-melting-temperature agarose method ofStruhl (27) as modified by Bae et al. (2).

Construction of plasmids pYB49, pYB56, and pYB76. Plas-mid pYB49 was constructed from plasmid pYB40 (Fig. 2).Plasmid pYB49 carries an in-frame fusion of the 8th codon(UCG, encoding serine) of the trpL sequence to the 10thcodon of lacZ (GUC, encoding valine) with the connectingsequence of GGGGATCCC between them. Plasmid pYB56was constructed from plasmids pYB49 and pEG220 (Fig. 2).This plasmid carries an in-frame fusion of the 99th codon ofthe trpE(G) gene (GAT, encoding aspartic acid) to the 10thcodon of the lacZ gene with the connecting sequence ofCCCbetween them. Plasmid pYB76 was constructed from pYB56by deleting the 569-bp BamHI-PstI fragment, making bothends blunt, and ligating the two ends (Fig. 2). This plasmidcarries the lacZ, lacY, and lacA genes without a promoter

IR. meliloti. (A) Antiterminator. Regions 1 and 4 are underlined. (B)

TABLE 1. Bacterial strains and plasmids

Strain or plasmid Genotype or marker(s) Reference

E. coliBW313 dut ung thi-l relA spoTliF' lysA 9JM109 recAl endAl gyrA96 thi hsdR17 29

supE44 relAl X A(lac-proAB)[F' traD36 proAB lacIqZAM15]

S17-1 Tpr Smr pro 25R. melilotiYB41 Smr AtrpE(G) lacZ 1YB52 Same as YB41 but carrying pYB49 This studyYB94 Same as YB41 but carrying pYB76 This study

PlasmidspEG220 Tcr trpE(G) 2pYB40 Tcr mob trpL'-'lacZ 1pYB49 Tcr mob trpL'-'lacZ This studypYB56 Tcr mob trpE'-'lacZ This studypYB76 Tcr mob; a version of pYB56 with This study

the trpE(G) promoter deleted

VOL. 173, 1991

------

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3384 BAE AND STAUFFER

SspI

1. BamH I2. Fill-in3. Pst I linker + T4 DNA ligase4. Pstl + ClaI

Purify 15.0 kb fragment

T4 DNA ligase

1. SspI2. Pstl linker + T4 DNA ligase3. Psti + Clal

Purify 1.2 kb fragment

1)

Tc R

SspI

ClaI '

BamH I

PstI BamHI+ PstI

Purify 15.6 kb fragment

T4 DNA ligase

y

I1 Bcl I

2. Fill-in3. Pst I linker + T4 DNA ligase4. PstI +XhoII

Purify 569-bp fragment

ori oriT

lacA IcR lacA I

lacYpYB 76

lacZ 16.2 kb 1. BamHI + PstI lacZ 15.6 kb

itrpL /2. T4 DNA polymerase

BamH I r 3. Purify 15.6kb fragment (BarmH V4. 14 DNA ligase (Bm>)

Pst I

FIG. 2. Construction of plasmids pYB49, pYB56, and pYB76. Both pYB49 and pYB56 have unique BamHI and Pstl sites, and pYB76 lostboth sites. pYB49 carries a trpL'-'lacZ fusion, and pYB56 carries a trpE'-'lacZ fusion. Plasmid pYB76 was constructed by deleting the 569-bpBamHI-PstI fragment of pYB56. The thin line represents vector DNA. Symbols: m, trpL DNA; , trpE(G) structural gene; El , E. colilacZ, lacY, and lacA structural genes.

and was used as a negative control for the ,3-galactosidaseassay.

Assay of I8-galactosidase activity. One loopful of a frozenculture of R. meliloti was used to inoculate 5 ml of TYmedium containing streptomycin, tetracycline, and cyclo-heximide. This was incubated at 30°C for 36 h with shaking.Three drops of this culture was used to inoculate 5 ml of M9medium supplemented with 0.2% glucose, 0.15% acid-hy-drolyzed casein, and 5 jig of tetracycline per ml (5 ,ug ofL-tryptophan per ml for tryptophan starvation and 40 ,ug/mlfor excess tryptophan). This was grown for another 36 h, and5 ml of fresh medium of the same composition was inocu-

lated with 3 drops of the culture and incubated for 36 h.These cultures were then used for the ,-galactosidase assayby the method of Miller (15).

Purification of the TrpL-LacZ fusion protein and determi-nation of its N-terminal amino acid sequence. R. melilotiYB52 cells (3.2 g [wet weight]) grown in 1 liter ofTY mediumwas harvested by centrifugation at 10,000 x g for 10 min.The cells were washed once with a solution of 100 mMTris-HCl (pH 7.3) and 10 mM EDTA and resuspended in 10ml of the same solution. The cells were then broken bysonication, and the cell debris was removed by centrifuga-tion at 15,000 x g for 15 min. Solid ammonium sulfate was

J. BACTERIOL.

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

R. MELILOTI trpE(G) ATTENUATOR 3385

AAG (L2A)

AUGGCAAACACGCAGAACAUI

ACG (L2C)

G AU U

G UC GG AAU GAA AC=GU-AGA_uC=G

GAC=GUGA (L23G-24A) U A-UUUUUU

UCC=G=CG=C=G

C (L49C) G=C=G -- C (L104C)U (L49T) ~ ~C=G=C

G=C=GJUPCAUCUGGUGGUGGGCUCGCUGAG=C=G

C=GAG=C

G UA UGAU

G AU GC=GG=C

40 - C=G - 49UC=GG=C

GG=CU UG GGU-AG=CG=CU-AC=G

wild type

C UG G

4-0 C AU GC=GG=CGG=C- 49U*GG=CG=C

UUG

U-AG=CG=CU-AC=G

L49C

BG A

U GC=GG=C

40 - C U- 49UC=GG=C

GG=CUU UG GU-AG=CG=CU-AC=G

L49T

FIG. 3. Locations of the mutations in the attenuator and the predicted 1:2 stem structures of the L49C and L49T mutations. (A) Theinitiator AUG codon, the eighth codon of the trpL sequence, and the UGA termination codon of trpL are underlined. The 2:3 and 3:4 stemstructures are also shown. (B) The predicted 1:2 stem structures of the wild type and of the L49C and L49T mutations. Only the top partsof the 1:2 stem are shown because the rest of the structures are predicted to be identical. See Table 2 for the free energies of these structures.

added to the supematant to 35% saturation, and the mixturewas centrifuged at 15,000 x g for 15 min. The supernatantwas discarded, and the pellet was resuspended in the samesolution used above at about 20 mg of total protein per ml.The protein solution was diluted with 50 mM Tris-HCl (pH7.3) to about 4 mg/ml just before loading onto the ProtoSorblacZ immunoaffinity absorbent column (Promega Corp.,Madison, Wis.). The column was washed with 15 ml of asolution of 50 mM Tris-HCl (pH 7.3) and 0.2% Nonidet P-40,and the fusion protein was eluted with two successive 1-mlaliquots of the elution buffer (0.1 M NaHCO3-Na2CO3, pH10.8). The eluted protein was dialyzed against 4 liters ofdistilled water for 24 h and was concentrated by freeze-drying.Ten micrograms of the protein was loaded on a 6% sodium

dodecyl sulfate-polyacrylamide gel by the method of Laem-mli (12), and electrophoresis was continued until the bro-mophenol blue dye just ran off the gel. The fusion protein

was then transferred to a polyvinylidene difluoride mem-brane as described by Matsudaira (13). The N-terminalamino acid sequence was determined by the Protein Struc-ture Facility at the University of Iowa.

RESULTS

Modification of the fusion plasmids. Previously we con-structed plasmids pYB27 and pYB40 carrying trpE'-'lacZand trpL'-'lacZ fusions, respectively, to measure trpE andtrpL expression in R. meliloti strains (1). These plasmids,however, do not have convenient restriction enzyme cleav-age sites for frequent subcloning experiments. Therefore, weconstructed plasmids pYB49 and pYB56 (Fig. 2). PlasmidspYB49 and pYB56 carry trpL'-'lacZ and trpE'-'lacZ fusions,respectively, and both have unique BamHI and PstI cleav-age sites flanking the DNA segment where mutations were tobe introduced. Plasmid pYB76 was constructed as a negative

VOL. 173, 1991

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3386 BAE AND STAUFFER

TABLE 2. Calculated free energy values of the stems ofthe mutants

Calculated free energy (kcal/mol) of stem:Mutation

1:2 2:3 3:4

None (wild type) -33.7 -21.5 -23.8L23G-24A -34.7 NCa NCL49C -31.4 -13.6 NCL49T -27.8 -13.6 NCL104C NC NC -15.9L104T NC NC -15.9

a NC (no change) indicates that the free energy is identical to the freeenergy of the equivalent stem of the wild type.

control from pYB56 and does not carry trp sequences (Fig.2).

Construction of the mutants. It was previously reportedthat the trpL codi'ng sequence of the R. meliloti trpE(G)attenuator lacks the Shine-Dalgarno sequence, since an Sinuclease mapping analysis showed that transcription of thetrpE(G) gene starts at the adenine residue of the AUG codonof the trpL coding sequence (2). In E. coli, this kind oftranscript is very rare and its translation is very poor. Thetranscript of the cI gene from the bacteriophage lambda pRMpromoter lacks the Shine-Dalgarno sequence and is trans-lated very poorly (16). However, the P-galactosidase activ-ities of R. meliloti strains carrying the trpL'-'lacZ fusionwere consistently higher than 1,200 U, regardless of thetryptophan concentrations in the growth medium, indicatingthat the trpL sequence is translated in R. meliloti (1). Thesefacts and earlier results demonstrating that attenuation is thesole mechanism of regulation for the R. meliloti trpE(G) gene(1) raised the following questions. First, is the AUG codon atthe beginning of the mRNA the real translational startcodon, or is another codon used for translation initiation oftrpL? Second, is translation of trpL mRNA necessary forregulation at the trpE(G) attenuator? Finally, are the second-ary structures in the leader' sequence mRNA involved inregulation at the attenuator?To answer these questions, we used site-directed muta-

genesis to introduce mutations in trpL. To' test whether theAUG codon at the beginning of the mRNA is 'the realtranslational start codon, mutations were introduced in theAUG codon of the trpL'-'lacZ fusion, converting AUG 'toeither AAG or ACG by oligonucleotide-directed mutagene-sis (Materials and Methods) (Fig. 3). To determine whethertransiation of the trpL mRNA and secondary structures inthe leader mRNA are involved in regulation at the trpE(G)attenuator, mutations were introduced into the leader se-quence of the trpE'-'lacZ fusion. The AUG codon of thetrpL sequence was converted to AAG or ACG, a serinecodon (UCG) in the trpL sequence was changed to a UGAstop codon (L23G-24A), a guanine residue of region 2 waschanged to either C (L49C) or T (L49T), and a guanineresidue in region 4 was changed to C (L104C) (Fig. 3). Allthese mutagenized fragments were cloned back into plasmidpYB49 and transferred back into R. meliloti YB41 by bipa-rental conjugation (1).

Effects of the mutations on the calculated stability of thesecondary structures in tipL mRNA. We calculated the pre-dicted free energy of each stem-loop structure in the atten-uator of the wild-type and mutagenized plasmids using thecomputer program of Zuker and Stiegler (33) (Table 2). Onlyslight changes in the free energy occurred in the 1:2 stem asa result of the L23G-24A and L49C mutations. The mutation

TABLE 3. Effects of mutations in the trpL AUG codon ontrpL'-'lacZ expression

,B-Galactosidase activitya (Miller units)Mutation

Tryptophan starvation Excess tryptophan

None (wild type) 1,251.2 (100.0) 1,981.2 (100.0)L2A 4.2 (0.2) 6.0 (0.2)L2C 3.2 (0.1) 4.6 (0.2)Control 1.8 1.5

a The relative activities (percent) are shown in parentheses.

L23G-24A increased the stability of the 1:2 stem slightlywithout affecting the other stems. Both L49C and L49Tdestabilized the 1:2 stem slightly. The minor difference in thestability between the L49C and L49T mutants is likely due tothe formation of the different structures. The wild-typeattenuator has a G at position 49, which pairs w'ith the C atposition 40 in region 1. For mutations L49C and L49T, thecomputer program of Zuker and Stiegler predicted thatalternate 1:2 structures would form (Fig. 3B). In these newstructures, L49T does not pair with a base in region 1 to formhydrogen bonds and it remains in a bulge. However, L49Cpairs with the G at position 35 in region 1,' making the 1:2stem containing L49C more stable than the 1:2 stem contain-ing L49T. No changes were found in the stability 'of the 1:2stem for L2A, L2C, or L104C. L49C or L49T, however,reduced the stability of the 2:3 stem significantly withoutaffecting the 3:4 stem. The L104C mutation had no effect onthe 2:3 stem but reduced the stability of the 3:4 stemsignificantly.

Effects of the mutations on trpL'-'lacZ and tqpE'-'lacZexpression. R. meliloti YB41 carrying plasmid pYB49 orpYB56 or mutagenized plasmids was grown in M9 minimalmedium under both tryptophan starvation and excess con-ditions, and ,B-galactos'idase levels were measured. Themutations in the trpL AUG codon (L2A and L2C) nearlyabolished translation of the mRNA of the trpL'-'lacZ fusiongene'(Table 3) under both tryptophan starvation and excessconditions. After subtraction of the P-galacto'sidase activityof the negative control, the L2A mutation resulted in 520-and 440-fold-lower '3-galactosidase levels under tryptophanstarvation and excess con'ditions, respectively. Similarly,the L2C mutation resulted in 890- and 640-fold-lower a-ga-lactosidase levels under tryptophan starvation and excessconditions, respectively.The wild-type attenuator regulated expression of the

trpE'-'lacZ fusions about 6.2-fold under tryptophan starva-tion and excess conditions (Table 4).' Mutations L2A andL2C nearly abolished translation initiation of the trpL codingsequence (Table 3) and also nearly abolished regulation' bythe attenuator (Table 4). The L2C mutation, however, re-sulted in higher P-galactosidase levels than the L2A muta-tion, approaching wild-type levels under the excess tryp-tophan condition. A possible explanation for this will bediscussed below (see Discussion). Mutation L23G-24A gen-erates a stop codon (UGA) preceding the tandem tryptophancodons in the trpL sequence. This mutation resulted inP-galactosidase levels similar to the levels observed for theL2A mutation. Both the L49C and L49T mutations reducedthe overall expression of the trpE'-'lacZ fusion. However,these mutations did not totally abolish regulation but re-duced attenuation to a narrower range (3.2- and' 2.3-fold,respectively) than the wild type. The L104C mutation re-sulted in a complete loss of regulation, showing significantly

J. BACTERIOL.

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

R. MELILOTI trpE(G) ATTENUATOR 3387

TABLE 4. Effects of mutations in the attenuator ontrpE'-'lacZ expression

P-Galactosidase activity (Miller units)

Mutation Tryptophan starvation Excess

Uncorrected Correctedab tryptophanb

None (wild type) 504.7 807.5 (100.0) 132.9 (100.0)L2A 15.3 24.5 (2.7) 30.3 (21.9)L2C 62.6 100.2 (12.1) 117.6 (88.4)L23G-24A 18.7 29.9 (3.4) 33.1 (24.1)L49C 95.6 153.0 (18.7) 47.7 (35.2)L49T 89.9 143.8 (17.5) 62.8 (46.6)L104C 571.6 914.6 (113.3) 1,003.4 (762.5)Control 1.8 1.5

a The enzyme levels obtained under tryptophan starvation conditions werecorrected for the effect of insufficient translation. ,B-Galactosidase levels of thetrpL'-'lacZ fusion were 1.6-fold higher under excess tryptophan conditions(Table 3). Therefore, enzyme levels obtained under tryptophan starvationconditions were multiplied by 1.6.

b The relative activities (percent) are shown in parentheses.

higher levels of expression of the fusion gene under excesstryptophan conditions.

Analysis of the TrpL-LacZ fusion protein. We purified theTrpL-LacZ fusion protein from R. meliloti YB52 carryingthe trpL'-'lacZ fusion gene on plasmid pYB49, and itsN-terminal amino acid sequence was determined (see Mate-rials and Methods). The sequence of the first 13 amino acidsof the fusion protein was Ala-Asn-Thr-Gln-Asn-Ile-Ser-Gly-Asp-Pro-Val-Val-Leu. Therefore, the first seven amino acidsof the fusion protein match perfectly with the N terminus ofthe predicted TrpL amino acid sequence (the first eightamino acids of the predicted sequence of the N terminus ofTrpL are Met-Ala-Asn-Thr-Gln-Asn-Ile-Ser [2]). The methi-onine at the N terminus was apparently removed. The nextthree amino acids determined (Gly-Asp-Pro) are encoded bythe connecting sequence of GGGGATCCC which was in-serted during construction of the plasmid. The last threeamino acids determined (Val-Val-Leu) match with the 10th,11th, and 12th codons of the lacZ coding sequence.

DISCUSSION

Most procaryotic genes that have been sequenced havetwo discernible elements for translation initiation, the Shine-Dalgarno sequence and the initiation codon (24). AUG is themost preferred initiation codon, although GUG and UUGare used infrequently (17). In R. meliloti, transcription of thetrpE(G) gene begins at the adenine residue of the AUGcodon of the leader peptide coding sequence (2). In thisstudy, we have shown that this AUG codon is the translationinitiation codon of the R. meliloti trpL coding sequence.Mutations converting this AUG codon to AAG or ACGessentially abolished translation of a trpL'-'lacZ fusion (Ta-ble 3), and amino acid sequencing data of the TrpL-LacZfusion protein support this conclusion.The cell's ability to recognize the levels of charged and

uncharged tRNATrP by the translating ribosome in the trpLcoding sequence is necessary for proper function of theattenuator (30-32). Zurawski et al. (34) isolated a mutationconverting the AUG initiation codon to AUA in the E. colitrpL sequence. When a strain carrying this mutation wasstarved of tryptophan, the rate of structural gene trp mRNAsynthesis increased only twofold above the level of un-starved cells. The wild-type strain showed a 10-fold increaseunder the same conditions. When transcribed in vitro, both

the DNA fragment carrying this mutation and the DNAfragment carrying the wild-type attenuator showed no differ-ence in the mRNA levels of the downstream genes. Theseresults, therefore, suggested that relief from transcriptiontermination requires translation of the leader transcript. Weconstructed mutations L2A, L2C, and L23G-24A in the R.meliloti trpL coding sequence to test whether translation ofthe trpL mRNA is essential for attenuator function. Muta-tions L2A and L2C essentially abolished translation initia-tion (Table 3). Mutation L23G-24A changed the eighth codon(serine) of the trpL sequence to a UGA codon that shouldresult in premature translation termination. The L2A andL23G-24A mutations reduced trpE'-'lacZ expression regard-less of the tryptophan concentrations and prevented relief oftranscription termination under tryptophan starvation con-ditions (Table 4). This suggests that translation of the trpLsequence beyond the seventh codon is required to relievetranscription termination in R. meliloti and that translationof the trpL coding sequence is required in order to respondto conditions that signal relief of transcription termination.One unexpected result is that the L2C mutant has about

fourfold-higher ,-galactosidase levels than the L2A mutant,despite the fact that both mutations are located at the sameposition (Table 4). Many potential stem-loop structurescould form in the region upstream of the 1:2 stem (33). TheL2C (ACG) transcript folds more stably than the L2A (AAG)transcript in many of these potential secondary structures(data not shown). These potential secondary structurescould compete with formation of the 1:2 stem, because thereis no ribosome translating this region for the L2A and L2Cmutations. Therefore, the formation of the 2:3 stem (antiter-minator) would be favored for the L2C transcript comparedwith the L2A transcript, which in turn would increaseread-through transcription into the trpE(G) gene.The role of the 2:3 stem is to cause tryptophan starvation-

induced transcription read-through (18-20). Roesser et al.(20) proposed that the basal level of expression woulddepend on the rate of ribosome dissociation at the trpL stopcodon and the timing of segment 4 synthesis. Although thereis no direct evidence demonstrating the formation of the 2:3stem during transcription of the R. meliloti trp leader region,several lines of evidence indicate the formation of the 2:3stem during transcription of E. coli trpL mRNA (11, 20). Theresults in Table 2 for L49C and L49T, which destabilize the2:3 stem, support a similar role in R. meliloti.

Previously, Bae and Crawford (1) showed that the mRNAlevel of the trp leader region in R. meliloti is constant,regardless of the tryptophan concentration. However, the,-galactosidase levels in a strain carrying the trpL'-'lacZfusion were lower under tryptophan-starved conditions thanunder excess tryptophan conditions. It was suggested thatthis phenomenon is due either to insufficient translation or torapid degradation of the fusion protein at low tryptophanconcentrations, rather than decreased transcription undertryptophan starvation conditions (1). The mutations L2Aand L23G-24A reduced the basal level of expression by 4.5-and 4.1-fold, respectively. The mutations L49C and L49Tdestabilized the 2:3 stem, changing the predicted AG valuefrom -21.5 to -13.6 kcal/mol (Table 2). These mutationsreduced the basal level of expression by 2.8- and 2.1-fold,respectively (Table 4), and increased transcription termina-tion during tryptophan starvation but did not completelyabolish read-through into the structural genes. These datasuggest that mutations L49C and L49T destabilize the 2:3stem without eliminating its formation, favoring formation of

VOL. 173, 1991

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

3388 BAE AND STAUFFER

the 3:4 stem (terminator), and thus increase transcriptiontermination.The mutation L104C destabilizes the 3:4 stem, changing

the predicted AG value of the stem from -23.4 to -15.9kcal/mol (Table 2). In the presence of excess tryptophan,relief of transcription termination by L104C is prominent(Table 4), and P-galactosidase levels increased 7.8-fold. Thisis in good agreement with the trpL153 and trpL154 mutationsin E. coli (26). These mutations destabilized the 3:4 stem ofE. coli trp attenuator (AG change is from -20.1 to -9.0kcal/mol) and had 3.8-fold-higher levels of transcription of adownstream gene. During tryptophan starvation, little effectby the L104C mutation is seen in R. meliloti (Table 4).

In E. coli, the methionine at the N terminus of manyproteins is removed. It has not been shown that this is alsothe case for the R. meliloti TrpL polypeptide. Our datashows that the methionine at the N terminus of the TrpL-LacZ fusion protein produced in R. meliloti is also removed,indicating that the methionine of the R. meliloti TrpL poly-peptide is likely to be removed.

R. meliloti is only very distantly related to E. coli (28). Inagreement with this are the low similarities of the trpE andtrpG sequences between these two bacterial species (2, 4),the different arrangement of the trp genes on the chromo-somes (2, 4, 7), and the different regulation of the trp genes(attenuation only in R. meliloti and a combination of repres-sion and attenuation in E. coli) (1, 30-32). The data pre-sented here indicate that, despite the distant relatedness ofthese two organisms, the R. meliloti trpE(G) attenuatorfunctions in a way very similar to that of the E. coli trpattenuator.

ACKNOWLEDGMENTS

We thank Mark L. Urbanowski for critically reading the manu-script.This work was supported by Public Health Service grant AI-20279

from the National Institute of Health.

REFERENCES1. Bae, Y. M., and I. P. Crawford. 1990. The Rhizobium meliloti

trpE(G) gene is regulated by attenuation, and its product,anthranilate synthase, is regulated by feedback inhibition. J.Bacteriol. 172:3318-3327.

2. Bae, Y. M., E. Holmgren, and I. P. Crawford. 1989. Rhizobiummeliloti anthranilate synthase gene: cloning, sequence, andexpression in Escherichia coli. J. Bacteriol. 171:3471-3478.

3. Beringer, J. E. 1974. R-factor transfer in Rhizobium legumi-nosarum. J. Gen. Microbiol. 84:188-198.

4. Crawford, I. P. 1989. Evolution of a biosynthetic pathway: thetryptophan paradigm. Annu. Rev. Microbiol. 43:567400.

5. Geisselsoder, J., F. Witney, and P. Yuckenberg. 1987. Efficientsite-directed in vitro mutagenesis. BioTechniques 5:786-791.

6. Holmgren, E., and I. P. Crawford. 1982. Regulation of thetryptophan genes in Rhizobium leguminosarum. J. Bacteriol.149:1135-1137.

7. Johnston, A. W. B., M. J. Bibb, and J. E. Beringer. 1978.Tryptophan genes in Rhizobium-their organization and theirtransfer to other bacterial genera. Mol. Gen. Genet. 165:323-330.

8. Kolter, R., and C. Yanofsky. 1984. Genetic analysis of thetryptophan operon regulatory region using site-directed muta-genesis. J. Mol. Biol. 175:299-312.

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

10. Kuroda, M., H. Shimotsu, D. J. Henner, and C. Yanofsky. 1986.Regulatory elements common to the Bacillus pumilis and Bacil-lus subtilis trp operon. J. Bacteriol. 167:792-798.

11. Kuroda, M. I., and C. Yanofsky. 1984. Evidence for the tran-script secondary structures predicted to regulate transcriptionattenuation in the trp operon. J. Biol. Chem. 259:12838-12843.

12. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

13. Matsudaira, P. 1987. Sequence from picomole quantities ofproteins electroblotted onto polyvinylidene difluoride mem-branes. J. Biol. Chem. 262:10035-10038.

14. Messing, J. 1983. New M13 vectors for cloning. MethodsEnzymol. 101:20-78.

15. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

16. Ptashne, M., K. Backman, M. J. Humayun, A. Jeffrey, R.Maurer, and R. T. Sauer. 1976. Autoregulation and function ofa repressor in bacteriophage lambda. Science 194:156-161.

17. Reddy, P., A. Peterkofsky, and K. McKenney. 1985. Transla-tional efficiency of the Escherichia coli adenylate cyclase gene:mutating the UUG initiation codon to GUG or AUG results inincreased gene expression. Proc. Natl. Acad. Sci. USA 82:5656-5660.

18. Roesser, J. R., Y. Nakamura, and C. Yanofsky. 1989. Regulationof basal level expression of the tryptophan operon of Esche-richia coli. J. Biol. Chem. 264:12284-12288.

19. Roesser, J. R., and C. Yanofsky. 1988. Ribosome release mod-ulates basal level expression of the trp operon of Escherichiacoli. J. Biol. Chem. 263:14251-14255.

20. Roesser, J. R., and C. Yanofsky. 1990. The RNA antiterminatorcauses transcription pausing in the leader region of the tryp-tophan operon. J. Biol. Chem. 265:6055-6060.

21. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.

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

23. Shimotsu, H., M. Kuroda, C. Yanofsky, and D. J. Henner. 1986.Novel form of transcription attenuation regulates expression ofthe Bacillus subtilis tryptophan operon. J. Bacteriol. 166:461-471.

24. Shine, J., and L. Dalgarno. 1974. The 3'-terminal sequence ofEscherichia coli 16S ribosomal RNA: complementarity to non-sense triplets and ribosomal binding sites. Proc. Natl. Acad.Sci. USA 71:1342-1346.

25. Simon, R., U. Priefer, and A. Puhler. 1983. A broad host rangemobilization system for in vivo genetic engineering: transposonmutagenesis in gram negative bacteria. Bio/Technology 1:784-791.

26. Stauffer, G. V., G. Zurawski, and C. Yanofsky. 1978. Singlebase-pair alterations in the Escherichia coli trp operon leaderregion that relieve transcription termination at the trp attenua-tor. Proc. Natl. Acad. Sci. USA 75:4833-4837.

27. Struhl, K. 1985. A rapid method for creating recombinant DNAmolecules. BioTechniques 3:452-453.

28. Woese, C. R. 1987. Bacterial evolution. Microbiol. Rev. 51:221-271.

29. Yanisch-Perron, C., J. Vieira, and J. Messing. 1985. ImprovedM13 phage cloning vectors and host strains: nucleotide se-quences of the M13mpl8 and pUC vectors. Gene 33:103-119.

30. Yanofsky, C. 1981. Attenuation in the control of expression ofbacterial operons. Nature (London) 289:751-758.

31. Yanofsky, C. 1984. Comparison of regulatory and structuralregions of genes of tryptophan metabolism. Mol. Biol. Evol.1:143-161.

32. Yanofsky, C. 1988. Transcription attenuation. J. Biol. Chem.263:609-612.

33. Zuker, M., and P. Stiegler. 1981. Optimal computer folding oflarge RNA sequences using thermodynamics and auxiliaryinformation. Nucleic Acids Res. 9:133-148.

34. Zurawski, G., D. Elseviers, G. V. Stauffer, and C. Yanofsky.1978. Translational control of transcription termination at theattenuator of the Escherichia coli tryptophan operon. Proc.Natl. Acad. Sci. USA 75:5988-5992.

J. BACTERIOL.

on February 15, 2020 by guest

http://jb.asm.org/

Dow

nloaded from