Vol. 175, No. 13JOURNAL OF BACTERIOLOGY, JUly 1993, p. 4137-41440021-9193/93/134137-08$02.00/0Copyright C 1993, American Society for Microbiology

Nucleotide Pool-Sensitive Selection of the TranscriptionalStart Site In Vivo at the Salmonella typhimunium pyrC

andpyrD PromotersK. I. S0RENSEN,"* K. E. BAKER,2 R. A. KELLN,2 AND J. NEUHARD'

Department of Biological Chemistry, Institute ofMolecular Biology, University of Copenhagen, Denmark,and Department of Chemistry, University of Regina, Regina, Saskatchewan, Canada2

Received 24 February 1993/Accepted 30 April 1993

Expression of the Salmonella typhimurium pyrC and pyrD genes is regulated in response to fluctuations in theintracellular CTP/GTP pool ratio. The repressive mechanism involves the formation of a stable secondarystructure (hairpin) at the 5' ends of the transcripts that precludes translational initiation by sequesteringsequences required for ribosomal binding. The potential for hairpin formation is controlled throughCTP/GTP-modulated selection of the transcriptional start site. Substitution of nucleotides in the region oftranscriptional initiation has revealed that selection of the transcriptional start point in vivo depends on thenucleotide context within the initiation region and the nucleoside triphosphate pool ratios. For maximal controlin response to CTP/GTP pool ratios, the wild-type CCGG start site motif appears to be optimal. Changing the-35 region in the pyrC promoter to the consensus sequence, or replacement of the pyrC promoter with the lacpromoter from Escherichia coli, has served to illustrate that the ability of the RNA polymerase to select theinitiation site in response to the intracellular nucleoside triphosphate pools is not promoter specific but isdetermined by the kinetic properties of the initiating RNA polymerase during the formation of the firstphosphodiester bond of the transcript.

The Salmonella typhimuriumpyrC andpyrD genes encodethe third and the fourth enzymes of the UMP biosyntheticpathway, namely, dihydroorotase and dihydroorotate dehy-drogenase. Expression of these unlinked genes is regulatedin response to fluctuations in the intracellular CTP/GTP poolratio through modulation of the leader region of the tran-script (4, 8, 25). High CTP/GTP pool ratios repress expres-sion as a consequence of the production of an mRNAinitiated with a CTP, 6 or 7 bp downstream of the -10 region(C at position -1 or + 1 [Fig. 1]); this transcript is ineffi-ciently translated because of its capacity to form a stablesecondary structure (hairpin) at the 5' terminus, therebysequestering sequences necessary for ribosomal binding.Under a low CTP/GTP pool ratio, the predominant start siteis not C at position -1 or + 1, but occurs with GTP, 2 or 3 bpfurther downstream (G at position +3). This shorter tran-script has reduced potential to form a stable secondarystructure at its 5' end, and translation occurs unhindered.Evidence that a similar regulatory mechanism controls ex-pression of the Escherichia colipyrC gene has recently beenpresented (26).The effect of the relative nucleoside triphosphate concen-

trations on the in vivo selection of the transcriptional startpoint is dependent on the nucleotide sequence in the initia-tion region. Relocating the site for the first G nucleotidewithin this region 1 bp closer to the -10 region of the S.typhimunium pyrC and pyrD promoters established it as thepredominant start point even when the CTP/GTP pool ratiowas high. However, when the G at position +3 was changedto a C, thereby effectively moving the first G 1 bp furtherdownstream, C at position +1 was the primary start pointeven when the CTP/GTP pool ratio was low (4, 25).

In the present work, the effect of nucleotide context

* Corresponding author. Electronic mail address: BIOK1M@MERMAID.MOLBIO.KU.DK.

alterations within the initiation region on the distribution oftranscriptional start sites has been further examined. Addi-tionally, the impact of structural changes in the promoter onsubstrate-sensitive selection of the transcriptional initiationsite by RNA polymerase has been investigated.

MATERUILS AND METHODS

Bacterial strains. The strains used are derivatives of S.typhimurium LT2: JL1280 (pyrG1611,ea4, cdd-7 udp-2 glp7)was from J. L. Ingraham, and strain SL4213 (metA22metE55 galE496 rpsLl20 ilv xyl-404 Fels2- Hl-b nml- H2-enx hsdL6 hsdSA29) was from B. A. D. Stocker; KP1725(pyrH16311eaky cdd-9 cod-8 deoD201 udp-11) has been de-scribed previously (21).

Plasmid constructions. The plasmids used are translationalfusions based on expression vectors pRAK81 and pRAK82(12, 25), derived from pNM481, pNM482 (20), and a lower-copy-number derivative of pJRD158 (3). The pRAK vectorscontain a multiple cloning site in front of the eighth codon oflacZ and are devoid of most of the lacY gene. The pyrCpromoter-leader and the first 10 codons were fused in frameto lacZ in pRAK81. The various pyrC promoter and leadermutations were transferred from recombinant M13 phages topRAK plasmids (12, 25). Translational pyrD-lacZ fusionsinclude the promoter-leader region and the first 17 codons ofpyrD fused in frame to lacZ of pRAK82. The cloningprocedure for the regulatory pyrD mutations has been de-scribed previously (4).

Plasmid pKIS42, which contains the lac promoter fused tothe pyrC transcriptional start region, was constructed fromplasmids pRZ8028 (27) and pKIS36 as illustrated in Fig. 2.Plasmid pRZ8028 carries the E. coli wild-type lac promoterwith an EcoRI linker inserted into the transcriptional startregion, placing the start site motif, CCGG, 7 bp downstreamfrom the -10 region of the promoter. To create an NgoMI

4137

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

4138 S0RENSEN ET AL.

promoter -10 +1 +3

pyrC 5'ACGTGCAAACGAAAACG1TTCCGC1TATCCTTTGTTCCCGGCAAAAAGACTATT CCTCCGGAG TATTA ATG ACTSD

ACUAUG U

EAA

A ACGG CG CC G

S'C-GAGCUlAUUA AUG ACUSD

closed for translation

mRNA at high CTP/GTP-

mRNA at low CTP/GTP

5'GGCAAAAAGACUAUUCCUCAGCG JUJAUUA AUG ACUSD

open for translation

promoter -10 -1+1 +3

5'ATTCCC1TITGCClFGTTATCGCCCATAATACGCCCCCGG1T[GCACACCGGGAATI CAGGAGA C ATG TAC,-,SD

mRNA at high CTP/GTP

G CU AU CU-AG-CG-CC-GC-G

5'C-GAAU jGAAUC AUG UACSD

closed for translation

mRNA at low CTP/GTP

5'GGUUUGCACACCGGGAAUiCAGGAGAG IUC AUG UACSD

open for translation

FIG. 1. Structure of the Salmonella typhimurium pyrC and pyrD promoter-leader regions and illustration of the proposed regulatorymechanism. Nucleotides are numbered relative to the promoter -10 region (overlined) where + 1 at the consensus spacing of 7 bp downstreamof the -10 element defines the in vivo transcriptional initiation point in repressing conditions. The Shine-Dalgarno regions (SD) are boxed,and the regions of hyphenated dyad symmetry are indicated by arrows above the sequence. Arrows below the sequence represent transcriptsarising in conditions of different CTP/GTP pool ratios. The putative secondary structures formed at the 5' ends of the transcripts are shown.

restriction site compatible with the NgoMI site of pKIS36,the 5' overhang of the unique EcoRI site of pRZ8028 wasremoved with mung bean nuclease, generating pKIS8028.The lac promoter was fused to thepyrC leader by ligating the2.8-kbp NgoMI fragment from pKIS8028 (carrying the lacpromoter and the bla gene) to the 3.2-kbp NgoMI fragmentfrom pKIS36 (carrying thepyrC leader and the first 6 codonsof the pyrC gene and lacZ).Media and growth conditions. Bacterial cultures used for

enzyme assays and for isolation of RNA were grown in ABminimal medium (2) containing 0.2% glucose and 0.2%Casamino Acids. Supplements, when included, were addedat the following concentrations: tetracycline, 10 ,ug/ml; am-picillin, 50 ,ug/ml; cytidine, 40 ,ug/ml; uracil, 20 ,ug/ml; andguanine, 15 ,ug/ml. Strain KP1725 contains a partially defec-tive UMP kinase due to the pyrH1631 mutation and istherefore starved for pyrimidine nucleotides even whengrown in the presence of uracil (9). When this strain is grownin the presence of cytidine, it contains high CTP and loweredUTP pools. The pyrG1611 allele of JL1280 encodes a par-tially defective CTP synthetase resulting in low intracellularcytosine nucleotide pools, and high UTP pools when thestrain is grown in the absence of cytidine. In the presence ofexogenous cytidine, the pools are normalized (16). Eitherstrain can be used to independently manipulate the CTP pool

to monitor CTP-mediated transcriptional start site selection.The strains were transformed with the appropriate plasmidconstructs and grown in repressing (cytidine-added) or dere-pressing (cytidine-absent) conditions.DNA techniques. Methods for plasmid isolation, restriction

endonuclease digestion, ligation, and transformation havebeen described previously (22). Plasmid constructs wereinitially selected in an appropriate E. coli strain and subse-quently transformed into restriction-negative S. typhimu-num SL4213, before transformation to the final S. typhimu-rium host. The chain termination method of Sanger et al. (24)was used for DNA sequencing with M13 templates.

Site-directed mutagenesis. Specific mutations were intro-duced into thepyrC andpyrD promoter-leader regions by themethod of Kunkel (14). Uracil-containing DNA from phageM13mpl8/DZ1, harboring the pyrD promoter-leader regionand a limited amount of the coding sequence (4), or fromM13mpl8, carrying the comparablepyrC elements, was usedas the template. The oligonucleotides used for mutagenesiswere as follows (altered bases are underlined):

pKIS35, 5'CTTTTTGOCCGAAAACAAAGGATAAG3'pKIS36, 5'CTTTTTGCCGGCAACAAAGGATAAG3'pKIS39, 5'CTTTTTGCCGTAAACAAAGGATAAG3'

pyrD

J. BACT1ERIOL.

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

SELECTION OF TRANSCRIPTIONAL START SITE 4139

EcoRI-10 +1TATGTrGTGTGGCCGGAAITCCGGCTGAGCGlac promoter lacZ

pRZ8028

EcoRl digestionMung Bean nuclease treatmentLigation

NgoMI NgoMJ-10 +1 -10 +1 ITATGTTGTGTGGCCGGCCGGCTGAGCG TATCCMITG1TGCCGGCAAAAAGACTLac promoter lacZ pyrC promoter pyrC

pKIS8028 pKIS36

NgoMI digestion NgoMl digestion

Ligation

NgoMI-10 +1 I

TATG1TGTGTGGCCGGCAAAAAGACTlac promoter pyrC

pKIS42FIG. 2. Construction of pKIS42. pRZ8028 was linearized with EcoRI and treated with mung bean nuclease. The blunt ends were ligated

to generate pKIS8028 containing an NgoMI site compatible with the NgoMI site of pKIS36. The lac promoter on pKIS8028 was fused to the,pyrC leader of pKIS36 after digestion of both plasmids with NgoMI and ligation, yielding pKIS42.

pKIS40, 5'CGTTTTCGTTQTCAAGTAAAAAAAAGGG3'pKEB4, 5'CCCGGTGTGCAAACCOAGGGCG3'The mutations in the pyrD leader region of pMMF25 andpMMF29 pertain to pyrDr2O and pyrDY7, respectively, andwere cloned from the chromosome of the correspondingregulatory mutants as described previously (4).Enzyme assays. Crude cellular extracts for enzyme assays

were prepared by sonically disrupting the cells (21). Theprocedures for determining the activities of the followingenzymes have been published previously: dihydroorotase(11), dihydroorotate dehydrogenase (10), and 3-galactosi-dase (19). The level of plasmid-encoded P-lactamase wasused as a relative measure of plasmid copy number (7).Protein was determined by the method of Lowry et al. (15).Primer extension. Total RNA was extracted from expo-

nentially growing cells containing the appropriate plasmidsessentially as described by Lu et al. (16). Analysis of the 5'end of the pyrC and pyrD transcripts was carried out byprimer extension with Moloney murine leukemia virus re-verse transcriptase (4, 25). The primers used were comple-mentary to the first six codons of pyrC (5'CCTGGGATGGTGCAGTC3'), or to codons 4 to 9 ofpyrD (5'GGC1TJTlACGAACGAAGGG3'). Extension products were resolved onan 8% polyacrylamide sequencing gel alongside DNA se-quence ladders.

RESULTS

Effects of mutations in the start site motif on transcriptionalinitiation. The transcriptional start point is typically located6 to 9 bp downstream of the 3' end of the -10 region of E.coli promoters (5, 6). More than 90% of all transcripts areinitiated with a purine nucleoside triphosphate, suggestingthat ATP and GTP are kinetically favored as the initiatingnucleotide by the RNA polymerase. However, the mecha-nism by which the RNA polymerase selects nucleotideswithin the transcriptional initiation region is unknown. Pre-vious results have demonstrated that both the relative intra-cellular concentrations of CTP and GTP and the nucleotidecontext in the initiation region influence the selection of thetranscriptional initiation site for the S. typhimurium pyrCandpyrD promoters (4, 25). To gain further insight into theseaspects, several nucleotides within the transcriptional startregion ofpyrC andpyrD were mutated and the effect in vivoof these mutations on the distributions of the mRNA 5' endswas determined under conditions facilitating variations inthe intracellular nucleoside triphosphate pools.The wild-type and mutant promoter-leader regions were

fused in frame with lacZ on a multicopy plasmid. Theplasmids were introduced into the appropriate host strains,and total RNA was extracted and used to determine the 5'ends of the transcripts through primer extension analysis.

VOL. 175, 1993

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

4140 S0RENSEN ET AL.

The results obtained are summarized in Fig. 3. With anelevated CTP pool, the start sites with CTP at position +1predominated in the wild-typepyrC situation (experiment 1),whereas the preferred start sites in the wild-type pyrDsituation included C at position -1 as well as at position + 1(experiment 2). When the CTP pool was low, initiation withGTP at position +3 predominated for both promoters.

Mutation of the C at position + 1 to a T in either pyrC(pKIS35) or pyrD (pMMF25) resulted in U at position + 1being the predominant start site in either growth condition(experiments 3 to 5). The efficient use of UTP as theinitiating nucleotide was further demonstrated by the resultsobtained with pMMF29, which has a C to T change atposition +2 (experiment 6). A strong preference for the useof the mutated site as the transcriptional start point (i.e.,initiation with UTP) occurred in either growth condition.With pKEB4, in which C at position + 1 and C at position +2of pyrD were both changed to T, multiple transcriptionalstarts were evident (experiment 7), with the primary startsoccurring with UTP at position + 1 and with GTP at position+3. This initiation pattern was unaffected by changes in thelevel of the CTP pool.Primer extension analysis of RNA from pKIS39 (C at

position + 1 of pyrC was changed to an A) showed that thepredominant transcriptional start was with ATP at position+ 1 in both growth conditions. A limited amount of initiationwith GTP at position +3 was also observed after prolongedexposure of the autoradiogram, but only in the low-CTPcondition (experiment 8).Experiment 9 (Fig. 3) shows the transcriptional initiation

patterns from pKIS36, which has T at position -1 in pyrCreplaced by G, thereby creating a GCCGG initiation region.In repressing conditions (high CTP), transcription was initi-ated with GTP at position -1. Significantly, however, thepredominant start in the low-CTP condition occurred withGTP at position +3.

Transcriptional initiation specified by other promoter struc-tures. Previous studies with the lacUVS promoter haveshown that mutations in the nontranscribed region of the lacpromoter may alter the transcriptional initiation pattern (1).Since the -35 region of thepyrC andpyrD promoters differssignificantly from that for the E. coli consensus promoter,consideration was given to whether this could be an impor-tant factor in promoting flexibility in the selection of thetranscriptional initiation site. To address this aspect, the -35region of thepyrC promoter was changed from GTGCAA tothe consensus sequence TTGACA. The mutated -35 regionwas introduced into the pyrC-lacZ fusion of pJRC49, yield-ing pKIS40. Expression of pyrC-lacZ from pKIS40 wasfound to be twofold higher than from the wild-type promoterof pJRC49, and expression from both plasmids was re-pressed by exogenous pyrimidines (Table 1). In accordancewith these results, primer extension analysis of RNA ob-tained from JL1280 harboring the individual plasmidsshowed that the overall initiation pattern was similar for thetwo constructs (Fig. 4).To investigate whether the substrate-sensitive selection of

the transcriptional initiation site is promoter specific, thepyrC promoter of pKIS36 was replaced with the E. coli lacpromoter from pRZ8028, yielding plasmid pKIS42. Thepromoters of pKIS36 and pKIS42 have identical transcrip-tional start site motifs (Fig. 2) and express pyrC-lacZ withalmost the same level of pyrimidine regulation (Table 1).Primer extension analysis of RNA extracted from JL1280containing pKIS36 and pKIS42 revealed that the distributionof transcriptional start sites was virtually identical from the

pyrC and lac promoters (Fig. 5). In high-CTP conditions, theG at position -1 was preferred as the start point, whereasstarts with G at position +3 were selected when the CTPpool was low. However, for the pyrC promoter a slightlyhigher level of transcripts initiated with G at position -1 wasobserved in low-CTP conditions.

DISCUSSION

Initiation of transcription from the S. typhimurium pyrCand pyrD promoters exhibits heterogeneity in the selectionof the transcriptional start site. Selection of the start sitewithin the (C)CCGG motif depends on the intracellularconcentrations of the initiating nucleoside triphosphates,CTP and GTP. Heterogeneity in transcriptional initiation hasbeen described previously for in vitro transcription from thelac (1) and the aroH (28) promoters and for in vivo transcrip-tion from some rRNA cistrons (17). However, regulatoryconsequences of 5' end heterogeneity specified from a singlepromoter have been reported only for the pyrC and pyrDgenes of S. typhimunum (4, 25) and thepyrC gene of E. coli(26). At a high intracellular CTP concentration (high CTP/GTP pool ratio), expression of pyrC and pyrD is repressedbecause the predominant transcript produced is initiatedfrom the C at position + 1 and has the potential to form ahairpin structure capable of inhibiting translational initiation.Under derepressing conditions, when the level of CTP islow, RNA polymerase is unable to initiate efficiently withCTP because of the lowered substrate concentration. Sincethe first two residues in the start site motif are C's, initiationdefaults to using GTP at position +3. The resulting shortertranscript is unable to form a stable secondary structure atthe 5' end, and the transcript remains open for translationalinitiation.

Previous studies have indicated that transcriptional initia-tion with GTP is strongly favored over initiation with CTPwhen the start site is similarly positioned with respect to the-10 region (4, 25). Therefore, the T to G mutation of pKIS36could be expected to direct transcriptional starts with GTP atposition -1 in either repressing or derepressing growthconditions, resulting in low-level constitutive expression.The results obtained (Fig. 3, experiment 9) showed thatinitiation with GTP at -1, was, in fact, predominant whenthe CTP pool was high. However, with a low CTP pool,transcriptional initiation defaulted to G at position +3. As aconsequence, expression from pKIS36 in JL1280 was pyri-midine regulated almost to the same extent as that of the wildtype (Table 1). This CTP pool-dependent switch in start sitefrom G at position -1 to G at position +3 indicated that theintracellular concentration of the second nucleoside triphos-phate to be incorporated into the mRNA (i.e., formation ofthe first internucleotide bond) also has a profound effect onthe selection of the start site. This was further supported bythe observation that changes in the CTP pool imparted analteration in the initiation pattern from pMMF25 (Fig. 3,experiment 5) and pKIS35 (Fig. 3, experiments 3 and 4),both of which have TCGG as the start site motif. With eitherplasmid, a low CTP pool still mediated initiation with GTP at+3, and for pKIS35, the increased UTP pool of JL1280relative to KP1725 did not preclude a shift to G at position+3. When the C of this sequence was changed to a T, as inpKEB4, the initiation pattern was CTP pool independent(Fig. 3, experiment 7), illustrating that the pattern seen forpMMF25 and pKIS35 was dependent on the C residuepresent in the start site motif.The initiation pattern observed with pKIS36 compared

J. BACTERIOL.

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

EXPERIMENT

1. pJRC49 (JL1280)wild-type pyrC

2. pMMFJ1 (KP1725)wild-type pyrD

vTATCCTTTGTTTCCGGCAAAAA

PB I A

VvCATAATACGCCCCCGGTTTGCA

PB A

high CrP *-

low CP

+1 +3

high CTP

low CTP Ii.I -1,

3. pKIS35 (JL1280)pC

VTATCCTTTGTTTTCGGCAAAAA

PB AI

4. pKIS35 (KP1725)pyrC

5. pMMF25 (KP1725)pyrD

VTATCCTTTGTTTTCGGCAAAAA

PB I I

VvCATAATACGCCCTCGGTTTGCA

PB IAA

high CTP f I

low CTP

4-1 -3

high CTP

low CTP tC+1 +3

6. pMMF29 (KP1725)pyrD

VVCATAATACGCCCCTGGTTTGCA

PB A

7. pKEB4 (KP1725)pyrD

8. pKIS39 (JL1280)pyrC

VvVCATAATACGCCCTTGGTTTGCA

PB AIA

VTATCCTTTGTTTACGGCAAAAA

PB A

high CTP

low CTP I,1 -+3

high CTP

low CTP

1

9. pKIS36 (JL1280)prC

VTATCCTTTGTTGCCGGCAAAAA

PB A

-10 +1

FIG. 3. Primer extension analysis with RNA obtained from cells grown in repressing or derepressing conditions. Strains used were eitherJL1280 or KP1725 containing plasmids with wild-type or mutantpyrC orpyrD promoter-leader regions fused in frame to the lacZ gene. Foreach experiment, the plasmid (with host strain in parentheses), the pertinent plasmid-borne transcriptional initiation region, and theautoradiogram of the primer extension analyses are shown. Numbering is in accordance with Fig. 1. The promoter -10 region (PB) isunderlined, and the mutations are indicated in boldface type. From the results of the primer extension analyses, the preferred in vivotranscriptional start sites were deduced. Open arrowheads, repressing growth conditions (high CTP); filled arrowheads, derepressingconditions (low CTP).

4141

high CTP

low CTP

high CTP

low CTP

high CTP

low CTP

-10 +1

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

4142 S0RENSEN ET AL.

TABLE 1. Expression of a pyrC-lacZ translational fusion fromvarious promotersa

Plasmidb Growth condition Sp act of,B-galactosidasec

pJRC49 Derepression 2,300Repression 400

pKIS40 Derepression 5,400Repression 1,800

pKIS36 Derepression 1,800Repression 250

pKIS42 Derepression 1,800Repression 350

a JL1280 containing the indicated plasmids was grown in glucose ABmedium without pyrimidine supplements (derepression) or with 40 jig ofcytidine per ml and 20 pLg of uracil per ml (repression).

b All plasmids contain an in-frame pyrC-lacZ fusion based on pRAK81:pJRC49, wild-typepyrC promoter-leader; pKIS40, pyrC promoter-leader witha consensus -35 region; pKIS36, pyrC promoter-leader; and pKIS42, lacpromoter-pyrC leader.

c Specific activities are in micromoles per minute per milligram of protein.Assays were done in duplicate, and the values represent the average of twoassays having less than 20% variance. Values have been corrected for plasmidcopy number.

with that for the wild-type situation of pJRC49 (Fig. 3,experiment 9 versus experiment 1) demonstrated that initia-tion with GTP at position -1 is preferred relative to initiationwith UTP at the same position. Primer extension mapping ofthe mutants with a C to T change in the start site motifindicated that UTP is preferred over CTP as the initiatingnucleotide. With pKIS35 and pMMF25, where both have thestart site sequence TCGG, starts with U at position + 1occurred at least as often as starts with G at position +3,regardless of the growth conditions (Fig. 3, experiments 3and 5). This is distinct from the wild-type situation in whichtranscription initiated from G at position +3 is stronglyfavored over that from C at position + 1 when the CTP poolis low (Fig. 3, experiments 1 and 2). The preferred usage ofUTP over that of CTP can be further illustrated by examin-ing the results for pMMF29, for which initiation was almostexclusively confined to UTP at position +2 (Fig. 3, experi-

-pJRC49-- *- pKIS40-0

-CR +CR -CR +CRC T A G 1 2 3 4

1g.

_ _ ! G~~+3

FIG. 4. Primer extension mapping of pyrC transcripts isolatedfrom JL1280 harboring various plasmids: pJRC49, wild-type pyrCpromoter-leader; and pKIS40, pyrC promoter-leader with a consen-sus -35 region. Cultures were grown without (-CR) or with (+CR)cytidine and uracil. Lanes 1 and 2, pJRC49; lanes 3 and 4, pKIS40.The lanes for the sequence ladder have been labeled as the comple-ment to facilitate direct correspondence with the transcript.

.---pKIS36 -_

SELECTION OF TRANSCRIPTIONAL START SITE 4143

use as an initiating nucleotide is less favorable than that ofUTP, GTP, or ATP. Additionally, the low binding efficiencyof CTP as the second nucleotide was implicated as being alimiting factor for the in vivo selection of the start site. Thus,the production of a pppCpC dinucleotide must be highlydisadvantageous at low CTP concentrations, and (C)CCGGmight, therefore, be the most sensitive start site motif inpromoting CTP/GTP-dependent switching in selection of thesite for transcriptional initiation. As a corollary, the lack ofsensitivity imparted by the ACGG start site, as in pKIS39, inwhich essentially all starts were with ATP (Fig. 3, experi-ment 8), even when the CTP pool was low, is notable. Thiscould be due to more efficient binding of ATP, offering anincreased opportunity for formation of the pppApC dinucle-otide.

Carpousis and coworkers (1) found that looseness ofbinding and flexibility in selection of the start site can beaffected by mutations in the promoter. The deviation of thepyrC and pyrD -35 regions from consensus called intoquestion the role this region might have in start site selec-tion. However, changing the pyrC -35 region to consensus,as in pKIS40, did not lead to a significant change in tran-scriptional initiation pattern. The observed increase in pro-moter strength as reflected in the elevated levels of 13-galac-tosidase (Table 1) and transcripts (Fig. 4) is consistent withprevious findings (13). Replacing the entire pyrC promoteron pKIS36 with the E. coli lac promoter, without alterationof the start site motif, did not result in any significantdifference in the distribution of transcripts (Fig. 5). Consis-tent with this, the levels of synthesis of hybrid ,B-galactosi-dase from pKIS42 and pKIS36 were similar and werepyrimidine regulated to the same extent. Since the E. coliwild-type lac promoter has -35 and -10 regions with goodhomology to consensus, and with an 18-bp spacing betweenthem, the lac promoter is different both in sequence and instructure from the pyrC and pyrD promoters. Thus, theability of the initiating RNA polymerase to select the tran-scriptional start site in response to fluctuations in the nu-cleoside triphosphate pools is apparently independent ofpromoter structure and sequence. Rather, selection is de-pendent on the nucleoside triphosphate concentration andthe sequence of the start region, with the precise transcrip-tional start point defined as the site where the concentrationof the nucleotide(s) corresponding to the first and secondresidues of the transcript facilitates productive initiation.

In thepyrC andpyrD promoters, the initiation kinetics ofRNA polymerase have been utilized to create a regulatorymechanism optimized to sense changes in the level of theCTP pool. Inclusion of the (C)CCGG start motif within aregion of dyad symmetry allows for fluctuations in theintracellular CTP pool to be reflected through the productionof variable transcripts having different potentials for trans-lation. Consequently, a change of any nucleotide within thestart site motif can affect both the pattern of transcriptionalinitiation and the potential for translation, resulting in alteredgene expression and regulation.

ACKNOWLEDGMENTS

We would like to express our appreciation to Mogens Kilstrup formany fruitful discussions and to Lisbeth Stauning for excellenttechnical assistance. We also want to thank W. S. Reznikoff forkindly providing plasmid pRZ8028.

This investigation was supported by grants from the DanishNatural Science Council to K.I.S., the Danish Center of Microbiol-ogy to J.N., and the Natural Sciences and Engineering ResearchCouncil (NSERC) of Canada to R.A.K., and by a jointly held NATO

International Collaborative Grant. K.E.B. wishes to thank NSERCfor the award of a student scholarship.

REFERENCES1. Carpousis, A. J., J. E. Stefano, and J. D. Gralla. 1982. 5'

nucleotide heterogeneity and altered initiation of transcriptionat mutant lac promoters. J. Mol. Biol. 157:619-633.

2. Clark, D. J., and 0. Maal0e. 1967. DNA replication and the celldivision cycle in Escherichia coli. J. Mol. Biol. 23:99-112.

3. Davidson, J., M. Heusterspreute, M. Merchez, and F. Brunel.1984. Vectors with restriction-site banks. I. pJRD158, 3903-bpplasmid containing 28 unique cloning sites. Gene 28:311-318.

4. Frick, M. M., J. Neuhard, and R. A. Kelln. 1990. Cloning,nucleotide sequence and regulation of the Salmonella typhimu-rium pyrD gene encoding dihydroorotate dehydrogenase. Eur.J. Biochem. 194:573-578.

5. Harley, C. B., and R. P. Reynolds. 1987. Analysis of E. colipromoter sequences. Nucleic Acids Res. 15:2343-2361.

6. Hawley, D. K., and W. R. McClure. 1983. Compilation andanalysis of Escherichia coli promoter DNA sequences. NucleicAcids Res. 11:2237-2255.

7. Hove-Jensen, B. 1985. Cloning and characterization of the prsgene encoding phosphoribosylpyrophosphate synthetase ofEscherichia coli. Mol. Gen. Genet. 201:269-276.

8. Jensen, K. F. 1989. Regulation of Salmonella typhimurium geneexpression: effect of changing both purine and pyrimidinenucleotide pools. J. Gen. Microbiol. 135:805-815.

9. Justesen, J., and J. Neuhard. 1975. pyrR identical to pyrH inSalmonella typhimurium: control of expression of thepyr genes.J. Bacteriol. 123:851-854.

10. Karibian, D. 1978. Dihydroorotate dehydrogenase (Escherichiacoli). Methods Enzymol. 51:58-63.

11. Kelln, R. A., J. J. Kinahan, K. F. Foltermann, and G. A.O'Donovan. 1975. Pyrimidine biosynthetic enzymes of Salmo-nella typhimurium repressed specifically by growth in the pres-ence of cytidine. J. Bacteriol. 124:764-774.

12. Kelln, R. A., and J. Neuhard. 1988. Regulation ofpyrC expres-sion in Salmonella typhimurium: identification of a regulatoryregion. Mol. Gen. Genet. 212:287-294.

13. Kobayashi, M., K. Nagata, and A. Ishihama. 1990. Promoterselectivity of Escherichia coli RNA polymerase: effect of basesubstitutions in the promoter -35 region on promoter strength.Nucleic Acids Res. 18:7367-7372.

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

15. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall.1951. Protein measurement with the Folin phenol reagent. J.Biol. Chem. 193:265-275.

16. Lu, C.-D., M. Kilstrup, J. Neuhard, and A. Abdelal. 1989.Pyrimidine regulation of tandem promoters for carAB in Salmo-nella typhimurium. J. Bacteriol. 171:5436-5442.

17. Lund, E., and J. E. Dahlberg. 1979. Initiation of Escherichia coliribosomal RNA synthesis in vivo. Proc. Natl. Acad. Sci. USA79:5480-5484.

18. McClure, W. R., C. L. Cech, and D. E. Johnson. 1978. A steadystate assay for the RNA polymerase initiation reaction. J. Biol.Chem. 253:8949-8956.

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

20. Minton, N. P. 1984. Improved plasmid vectors for the isolationof translational lac gene fusions. Gene 31:269-273.

21. Neuhard, J., R. A. Kelln, and E. Stauning. 1986. Cloning andstructural characterization of the Salmonella typhimuriumpyrCgene encoding dihydroorotase. Eur. J. Biochem. 157:335-342.

22. Neuhard, J., E. Stauning, and R. A. Kelln. 1985. Cloning andstructural characterization of the pyrE gene and ofpyrE::Mudl(ApRlac) fusions from Salmonella typhimurium.Eur. J. Biochem. 157:335-342.

23. Nierman, W. C., and M. J. Chamberlin. 1979. Studies of RNAchain initiation by Escherichia coli RNA polymerase bound toT7 DNA. J. Biol. Chem. 254:7921-7926.

24. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-

VOL. 175, 1993

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

4144 S0RENSEN ET AL.

ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.USA 74:5463-5467.

25. Sorensen, K. I., and J. Neuhard. 1991. Dual transcriptionalinitiation sites from thepyrC promoter control expression of thegene in Salmonella typhimunium. Mol. Gen. Genet. 225:249-256.

26. Wilson, H. R., C. D. Archer, J. Liu, and C. L. Turnbough. 1992.Translational control of pyrC expression mediated by nucleo-tide-sensitive selection of transcriptional start sites in Esche-

richia coli. J. Bacteriol. 174:514-524.27. Xiong, X., N. de la Cruz, and W. S. Reznikoff. 1991. Down-

stream deletion analysis of the lac promoter. J. Bacteriol.173:4570-4577.

28. Zurawski, G., R. P. Gunsalus, K. D. Brown, and C. Yanofsky.1981. Structure and regulation of aroH, the structural gene forthe tryptophan-repressible 3-deoxy-D-arabino-heptulosonic ac-id-7-phosphate synthetase of Escherichia coli. J. Mol. Biol.145:47-73.

J. BACTERIOL.

on May 31, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/