Has DNA-BindingActivity Its Carboxy Terminus That ... · i2, andi3), andftrs (fl, f2, f3, f4, andf5) in thefliFandflaN-flbGpromoterregions; thethin line is genomicDNA,andthe thickline

Vol. 176, No. 19JOURNAL OF BACTERIOLOGY, Oct. 1994, p. 5971-59810021-9193/94/$04.00+0Copyright © 1994, American Society for Microbiology

FlbD Has a DNA-Binding Activity near Its Carboxy Terminus ThatRecognizes ftr Sequences Involved in Positive and Negative

Regulation of Flagellar Gene Transcriptionin Caulobacter crescentus

DAVID A. MULLIN,* SUSAN M. VAN WAY, CATHERINE A. BLANKENSHIP, ANDANN H. MULLINDepartment of Cell and Molecular Biology, Molecular and Cellular Biology Program,

Tulane University, New Orleans, Louisiana 70118-5698

Received 22 March 1994/Accepted 27 July 1994

FlbD is a transcriptional regulatory protein that negatively autoregulates fliF, and it is required forexpression of other Caulobacter crescentus flagellar genes, including flaN and flbG. In this report we haveinvestigated the interaction between carboxy-terminal fragments of FlbD protein and enhancer-like ftrsequences in the promoter regions offliFflaN, andflbG. FlbDc87 is a glutathione S-transferase (GST)-FlbDfusion protein that carries the carboxy-terminal 87 amino acids of FlbD, and FlbDc87 binds to restrictionfragments containing the promoter regions offliFflaN, andflbG, whereas a GST-FlbD fusion protein carryingthe last 48 amino acids of FlbD failed to bind to these promoter regions. DNA footprint analysis demonstratedthat FlbDc87 is a sequence-specific DNA-binding protein that makes close contact with 11 nucleotides inftr4,and 6 of these nucleotides were shown previously to function in negative regulation offliF transcription in vivo(S. M. Van Way, A. Newton, A. H. Mullin, and D. A. Mullin, J. Bacteriol. 175:367-376, 1993). Three DNAfragments, each carrying anftr4 mutation that resulted in elevatedfliF transcript levels in vivo, were defectivein binding to FlbDc87 in vitro. We also found that a missense mutation in the recognition helix of the putativehelix-turn-helix DNA-binding motif of FlbDc87 resulted in defective binding toftr4 in vitro. These data suggestthat the binding of FlbDc87 to ftr4 is relevant to negative transcriptional regulation offliF and that FlbDfunctions directly as a repressor. Footprint analysis showed that FlbDc87 also makes close contacts withspecific nucleotides inftr, tr2, andftr3 in thefjaN-flbG promoter region, and some of these nucleotides wereshown previously to be required for regulated transcription offlaN andflbG (D. A. Mullin and A. Newton, J.Bacteriol. 175:2067-2076, 1993). Footprint analysis also revealed a newftr-like sequence,ftr5, at -136 from thetranscription start site offlbG. Our results demonstrate that FlbD contains a sequence-specific DNA-bindingactivity within the 87 amino acids at its carboxy terminus, and the results suggest that FlbD exerts its effectas a positive and negative regulator of C. crescentus flagellar genes by binding to ftr sequences.

Caulobacter crescentus has a unique cell division cycle thatyields two morphologically distinct cell types, and it hasprovided a tractable and fruitful model for investigating inter-related aspects of temporal and spatial regulation that result incell differentiation (for reviews, see references 4, 39, and 53).Most of these studies with C. crescentus have concentrated onmorphogenesis of the flagellum, and they have been greatlyfacilitated by the availability of large numbers of nonmotilemutants (23).Approximately 50 genes are required for assembly and

function of the C. crescentus flagellum (16), and epistasisstudies have revealed that these flagellar genes are arranged ina regulatory hierarchy consisting of four levels or classes ofgenes (Fig. 1) (6, 40, 44, 58, 62). The flagellin genes flgK andflgL occupy level IV at the bottom of the hierarchy, and theirexpression depends on the level III genes (30, 44) whichdepend, in turn, on level II genes for their expression (9, 10, 35,40). Although trans-acting genetic requirements specific forexpression of the level II genes have been searched for byseveral laboratories, none has been reported. If found, such agene or genes would occupy level I.

In addition to being under hierarchical control, these flagel-lar genes are also under cell cycle control because their

* Corresponding author. Phone: (504) 865-5546. Fax: (504) 865-6785.

expression is restricted to a discrete interval in the cell cycle,and it has been suggested that a step in the DNA replicationpathway plays a role in timing the expression of flagellar genes(12, 46, 54). Studies to understand at the molecular level howexpression of the C. crescentus flagellar genes is tied to the cellcycle are still at a very early stage, and most of the work hasfocused on identifying and characterizing the flagellar genesand the cis- and trans-acting elements that regulate theirexpression.

fliF is one of the earliest flagellar operons to be expressed inthe cell cycle (43, 58), and its products include basal body andswitch proteins and a transcriptional regulatory protein calledFlbD (GenBank accession number M98855) (48). FlbD is anegative autoregulator of its own transcription (40) and apositive regulator of theflaN andflbG operons (10, 35, 44, 48).The inferred FlbD amino acid sequence suggests that itbelongs to a family of regulatory proteins that includes NtrCand NifA (48), which bind to enhancer-like sequences andstimulate promoter-bound C54 RNA polymerase to initiatetranscription (for reviews, see references 33 and 42). Consis-tent with this prediction, a flbD' plasmid restored glnAexpression in an Escherichia coli glnG (ntrC) mutant (48).The 5' transcriptional regulatory region of fliF contains a

novel promoter sequence that is also found in two other C.crescentus class II flagellar operons, fliQR and fliLM (12, 56,64). Immediately downstream from the fliF promoter, and

5971

on June 25, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

http://jb.asm.org/

5972 MULLIN ET AL.

Early

masterregulatory genes|

A. B.fliLM, fliQR, fliF flbF rpoNM-ring and protein sigmaswitch proteins export ? 54

cis-acting regulatory elements

a >promoter, ftr I

IflaN, fIbG

flgH,2 fIgF, fIgIbasal body andIhook protein

l|flgK, flgL

|flogellins

a 54promoter, ftr, ihf

a 54 promoter, ftr, ihf

FIG. 1. Flagellar gene regulatory hierarchy. The level II genes are classified as A or B because they utilize different types of promoters, andrpoN is listed separately because its promoter sequence has not been reported. fliF is the only level II gene known to have an ftr (58). flgH doesnot have an ihf between the a" promoter and ftr (11). The cis-acting elements and relative positions of the following genes in the hierarchy weresummarized from other sources: fibF (49, 50), rpoN (5), fliF (58), fliLM (64), fliQR (12), flaN and flbG (17, 36, 37), figH and flgF (11), flgI (24),and flgK and flgL (30).

overlapping the transcription start site, is a negative regulatorysequence called ftr4 that was proposed to constitute part of amolecular switch that is required to turn offfliF transcription atthe correct time in the cell cycle (58). The promoter regions offliQR and fliLM both apparently lack ftr sequences.

The divergently transcribed flaN and flbG operons at levelIII utilize (u" promoters whose activities depend on enhancer-like sequences called ftr (18, 30, 35-37, 41) and sequencescalled ihf that are located between the promoters and ftrs (17,18, 37). The ihf sequences conform to the consensus bindingsequence for E. coli integration host factor protein (29).Consistent with the prediction that FlbD functions directly asa transcriptional regulator (48) is a report that an epitope-tagged FlbD fusion protein binds to ftr sequences in thepromoter regions offlaN and flbG and the flagellin genes flgKand flgL as assessed by mobility shift analysis (60).

In this report we have investigated the DNA-binding activityof FlbD in order to better understand how this proteinfunctions as a regulator of transcription. We show that a

glutathione S-transferase (GST)-FlbD fusion protein calledFlbDc87, which contains the carboxy-terminal 87 amino acidsof FlbD, is able to bind toftri,fitr2,ftr3, andftr5 in theflaN-flbGpromoter region and to ftr4 in the fliF promoter region.Furthermore, our findings demonstrate that FlbDc87 makesclose contact with bases in*ri,ftr2, fr3, andft4 that have beenshown previously to be essential for regulating transcription invivo.

MATERIALS AND METHODS

Strains and culture conditions. The bacterial strains, plas-mids, and phage used in this work are listed in Table 1. C.crescentus CB15 was grown in a peptone-yeast extract (PYE)medium (47), and plasmid-containing strains were grown in

PYE broth supplemented with 2 pug of tetracycline per ml.Motility of C. crescentus merodiploid strains was tested bystabbing them into motility agar (23). The extent of swarmingwas compared with that of the motile wild-type strain. E. colistrains were grown in a yeast extract-tryptone (YT) medium(20), and strains containing plasmids were grown in YTmedium supplemented with the appropriate antibiotics: ampi-

TABLE 1. Bacterial strains, plasmids, and phage

Strain, plasmid, Relevant genotype or Referenceor phage phenotype or source

StrainsC. crescentus CB15 Wild type 47

E. coliJM107 Male 63DH5cx Male, Kmr BRLa

PlasmidspUC19 Apr 63pUC2.2RI Apr flbD+ This workpGEX-2T Apr, expresses GST 55pGEXc87 Apr, expresses FlbDc87 This workpGEXc48 Apr, expresses FlbDc48 This workpRK290 Cloning vector, replicates in 13

C. crescentuspRK290-c87K Apr Tcr, expresses FlbDc87 This workpRK290-c87(T4291) Apr Tcr, expresses This work

FlbDc87(T-429-->I)pRK290-c48 Apr Tcr, expresses FlbDc48 This work

Bacteriophage M13mpl9 Cloning vector 63

a BRL, Bethesda Research Laboratories Life Sciences, Inc.

III

LateIV

J. BACT1ERIOL.


.org/D

ownloaded from

http://jb.asm.org/

FlbD PROTEIN BINDS TO ftr REGULATORY SEQUENCES 5973

flaQ flaN fbG fkoJ fIoK 1.1 fliF 5G fbE fUN fbD-e :3-

B P B B By H B RI ~ ~11

- kb

~~ ~ ~ ~ ~ ~ R

H P f3 f2 f5 fl B

i3 a- Q2 i1 a54

floN fIbG

2

3

4 _

IL -1Ji LM1O-SV4

fliF

5' 5 5'6 * 3,

FIG. 2. Genetic and physical map of the hook gene cluster. (A) Restriction map and organization of genes and transcription units (GenBankaccession no. M98855) (35, 45, 49, 50). Restriction sites: B, BamHI; H, HindIII; P, PstI; and R, EcoRI. (B) Organization of promoters, ihfs (il,i2, and i3), andftrs (fl, f2, f3, f4, and f5) in the fliF andflaN-flbG promoter regions; the thin line is genomic DNA, and the thick line is polylinkerDNA from pUC18 (63). Arrows indicate the origin and direction of transcription. (C) Probes used in footprint reactions. *, 32P end label.

cillin, 50 jig/ml; kanamycin, 50 jig/ml; and tetracycline, 10pug/ml.Construction and expression of gst-flbD gene fusions. Plas-

mid pUC2.2RI consists of a 2.2-kb flbD' EcoRI fragmentcloned in the EcoRI site of pUC19. In-frame fusions betweenthe GST gene (gst) and flbD were constructed by a Bal 31exonuclease protocol or a PCR protocol. In the Bal 31exonuclease protocol, plasmid pUC2.2RI was digested at theunique polylinker HindIII site that is located 5' of flbD. Atvarious times after Bal 31 nuclease was added, aliquots of thereaction mixture were phenol extracted and recut with EcoRIto release an overlapping set of fragments that extend from theEcoRI site to various Bal 31-generated end points. The re-leased fragments were ligated to M13mp19 that had beendigested with SmaI and EcoRI, and constructs capable ofgenerating in-frame fusions to GST in plasmid pGEX-2T (55)were identified by nucleotide sequence analysis. By sequencingthe Bal 31-generated clones and using oligonucleotide primers,we determined the nucleotide sequence of both strands offlbD, and our sequence data match the published flbD se-quence (48). We deposited the flbD nucleotide sequence (withthe gene designation ftrC) in the GenBank under accessionnumber M30946 prior to the published report of Ramakrish-nan and Newton (48). In-frame gst-flbD gene fusions wereconstructed by ligating BamHI-EcoRI fragments from selectedM13mpl9 recombinant constructs to pGEX-2T that had beendigested with BamHI and EcoRI.A DNA fragment that encodes the 87 carboxy-terminal

amino acids of FlbD with BamHI sites near each end wassynthesized by PCR synthesis with pUC2.2RI as a templateand oligonucleotide primers 4100 (5'-CCGCGTGGATCCATGGCCCCGGCGCCGGAC) and 400 (5'-CCCTGGGATCCGTCTTAAGCGGCCGCGCCGACCCCGCCCTG). The PCRwas carried out with a commercially available kit from Perkin-Elmer Cetus, Inc., under conditions recommended by themanufacturer, except that exo+ Vent DNA polymerase fromNew England Biolabs was used in place of Taq DNA poly-merase. The reaction mixture contained 10 ng of template

DNA and 100 pmol of each oligonucleotide primer, and 15reaction cycles (1 min at 970C for denaturation, 30 s at 60'C forannealing, and 30 s at 720C for extension) were used. The289-bp PCR product was purified by electroelution from a 6%native polyacrylamide gel, and after cleavage with BamHI, itwas ligated to BamHI-digested pGEX-2T to generate pGEXc87.pGEXc48 expresses a GST-FlbD fusion protein that in-

cludes the carboxy-terminal 48 amino acids of FlbD, and it wasconstructed by the Bal 31 nuclease protocol described above.pRK290-c48 was constructed by ligating EcoRI-digestedpGEXc48 to EcoRI-digested pRK290. pRK290-c87 was con-structed by ligating EcoRI-digested pGEXc87 to EcoRI-di-gested pRK290. Plasmids were introduced into E. coli strainsby transformation (27), and electroporation was used to intro-duce plasmids into C. crescentus (36).

Purification of GST-FlbD fusion proteins. Recombinantplasmids that expressed soluble GST-FlbD fusion proteinswere introduced into E. coli DH5a by transformation, andovernight cultures were diluted 1:100 in 1 liter of YT brothcontaining ampicillin and grown at 370C with shaking until theoptical density at 650 nm was 0.6. High-level GST-FlbD fusionprotein expression was induced by adding isopropyl-o-D-thio-galactopyranoside (IPTG) to a final concentration of 0.5 mM,and the cells were cultured for an additional 4 h. Cells wereharvested by centrifugation, resuspended in 10 ml of ice-coldPBS (150 mM NaCl, 20 mM sodium phosphate [pH 7.3]), andbroken by sonication. Whole cells and cell debris were re-moved by centrifugation in a Beckman 50 Ti rotor at 36,000rpm, at 40C, and for 30 min, and the amber supernatant waspassed through a 0.2-jim-pore-size cellulose acetate filter. Thefiltered supernatant was adjusted to 1% (vol/vol) Triton X-100and applied to a glutathione-Sepharose 4B column. After thecolumn was washed with 10 bed volumes of PBS, the GST-FlbD fusion proteins were eluted with 5 mM reduced glutathi-one in 50 mM Tris-HCl (pH 8.0). Eluted proteins wereconcentrated by centrifugal filtration with a Centricon 30membrane (Millipore, Inc.), and the protein concentration wasestimated by using the Bradford assay (3). The apparent

A.

B.

Ll-603

C.

5'

3'

VOL. 176, 1994


.org/D

ownloaded from

http://jb.asm.org/

5974 MULLIN ET AL.

A.

B S/B>1.

B

R

c48

B

c87

B S R

B.

18 20 38 48 50 68 78 88 98 188 118 128MIRLLVVCKLNG8LSVRVKM1¶RNACRKVSHVETTEORTNRLRACOGROLLMVDYVLXIRGL I1RNERERMRlvPVVRCGVOROPM+RNRIRIKRCGKEF IPLPPORELI RRVLR8VTODEKPM

c371130 180 130 168 170 188 198 200 218 220 230 298

VVROPA1EOV IKLROQVRPSERSILI IESGSGKEVMARYVHGKS*RKRPF I S8NCRRIPENLLESELFCHEKCRFTGAMRRR ICKFEERDCGTLLLDE I SE¶OVRLORKLLRRIQERE

c308 c290 c289258 260 278 288 298 388 318 320 330 348 358 368

IORVGGSKPVKVNIRILRTSNRDLRORVKDGTFREOLLYRLNVVNLRLPPLRERPRDVISLCEFFVKKYSARNGIEEKPI SREAKRRLIRHRUPGNVRELENRMHRRVLLSRGPE IEEFA

378 380 39 9 88 418 '428 438 448 Y45IRLPDGCQPF RPDAPvRVRRcRonRRDAARRFVGSTvREVEQL ItD IEHCLGNRTHRRANILGISIRTLRNKLKEYSORGVOVPPPQGGVGRRR

= - -

c87A B

c48

FIG. 3. GST-FlbD fusion protein constructs. (A) Restriction map of plasmid pGEX-2T (55) and restriction fragments that encodecarboxy-terminal fragments of FlbD. S, SmaI. Other restriction sites are as indicated in the legend to Fig. 2. (B) The FlbD amino acid sequenceis inferred from the nucleotide sequence (48). Brackets mark the sites of fusion junctions between GST and FlbD. Fusion protein designationsindicate the length of the carboxy-terminal fragment of FlbD. Thus, c371 marks the site of a fusion protein that includes the carboxy-terminal 371amino acids of FlbD. The corresponding GST-FlbD fusion protein would be referred to as FlbDc371. The dashed line under the sequence indicatesa proposed nucleotide-binding motif; the thick line under the sequence indicates a potential helix-turn-helix (48), and the open bars under thesequence indicate a region that is similar to the A and B dimerization helices of Fis (25, 26, 42, 65).

molecular weights of fusion proteins were measured by sodiumdodecyl sulfate (SDS)-polyacrylamide gel electrophoresis(PAGE) for denatured proteins, and gel filtration chromatog-raphy was done on Sephadex G200 Superfine for nativeproteins.

Mobility shift and footprint assays. DNA probes used inmobility shift and footprint assays were 32P 3' or 5' end labeledas described previously (50), and they are shown in Fig. 2C. Asynthetic double-stranded oligonucleotide probe carrying theftr4 sequence was made by annealing equal molar amounts ofoligonucleotides ftr4l (5'-GATCCGTACTGGGTAAATCCT

GCCTACCA) and ftr42 (5'-AGCT-fGGTAGGCAGGAT-FTACCCAGTACG). The oligonucleotides in TES (50 mM Tris[pH 7.4], 50 mM NaCl, 5 mM EDTA) were heated to 95°C for5 min and then slowly cooled to 25°C. Mobility shift reactionmixtures were set up in a final volume of 20 ,ul, and theycontained FlbDc87 protein, 1 pg of sonicated calf thymusDNA, and 0.5 pmol of 32P-end-labeled probe DNA in a buffercontaining 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, and 1 mMdithiothreitol. After 15 min at 37°C, marker dye was added andthe reaction mixtures were fractionated by electrophoresis on6% (for restriction fragment probes) or 11% (for the oligonu-

J. BACT1ERIOL.


.org/D

ownloaded from

http://jb.asm.org/


A A1 2 3 4

B

66

45

31

22

4- FlbDC874- F bDc484- Gst

/

-O/-10.B

Kav

0.7

0.6 I

0.5 I

0.4

0.3

0.2

0.10.00-,10,000

FbDc48Gst

4- FIDCS7/,l F,A.I

~I.

100,000

Molecular WeightFIG. 4. Apparent molecular weights of GST-FlbD fusion proteins

and GST. (A) SDS-10% PAGE. Lane 1, molecular weight standards;lane 2, GST; lane 3, FlbDc48; lane 4, FlbDc87. (B) Apparent molec-ular weights of native proteins as determined by gel filtration chroma-tography. Open circles represent protein molecular weight standards.

ATaa

aTAAATCCTGCCT

ee

ee0

NM4

-10*

AHCCTATT

'a +

4

FIG. 5. FlbDc87 binding to ftr4. Footprint reactions are describedin Materials and Methods; DNA probes are shown in Fig. 2C. -,reduced reactivity toward DMS; +, enhanced reactivity toward DMS;0, protection from the hydroxyl radical. (A) Footprint on nontemplatestrand of ftr4, using probe 6. Lanes 1 to 3, hydroxyl radical reaction;lanes 4 and 5, DMS reaction; lanes 2 and 4, FlbDc87 protein; lanes 1,3, and 5, no FlbDc87 protein. (B) Footprint on the template strand offtr4, using probe 5. Lanes 1 and 2, hydroxyl radical reaction; lanes 3 and4, DMS reaction; lanes 2 and 3, FlbDc87 protein; lanes 1 and 4, noFlbDc87 protein.

cleotide probe) native polyacrylamide gels. The free andbound complexes were visualized by autoradiography.

Dimethyl sulfate (DMS) footprint analysis (61) was per-formed by combining 0.5 pmol of 32P-end-labeled probe DNAand 0.1 nmol of FlbDc87 protein under the conditions outlinedabove for mobility shift assays. After 15 min at 370C, 180 p1 ofice-cold DMS reaction buffer (50 mM sodium cacodylate [pH8.0], 1 mM EDTA) was added along with 1 Al of DMS, and thereaction mixture was incubated at 20'C for 2 min. The reactionwas stopped by adding 50 pl of ice-cold stop mix (3 Mammonium acetate, 1 M 2-mercaptoethanol, 20 mM EDTA,250 pug of tRNA per ml), and the DNA was ethanol precipi-tated twice, dried under a vacuum, and resuspended in 100 puIof 1 M piperidine. After the reaction mixture was heated in a950C water bath for 30 min, the DNA solution was frozen andlyophilized, and the DNA was then subjected to three addi-tional cycles of lyophilization from ice. After the final Iyophi-lization step, the DNA was dissolved in formamide marker dye.

Hydroxyl radical footprinting (14) was performed by com-bining 0.5 pmol of 32P-end-labeled probe DNA and 0.1 nmol ofFlbDc87 protein under the conditions outlined above formobility shift assays. The hydroxyl radical reaction was startedby adding 10 pul of 0.3% H202, 10 pl of 35 mM Fe-EDTA, and10 pA of 70 mM ascorbate, and after 2 min at room tempera-ture the reaction was stopped by adding 10 Ail of stop solution(0.1 M thiourea, 2 pug of yeast tRNA per ml). The nucleic acidswere phenol extracted, ethanol precipitated twice, dried, anddissolved in formamide marker dye.DMS and hydroxyl radical footprint reaction products in

marker dye were heated in a 950C water bath for 3 min andsubjected to electrophoresis in 4 or 8% denaturing polyacryl-amide gels. Footprints were visualized by autoradiography.Hydroxylamine mutagenesis. Plasmid DNA was dissolved at

40 g/ml in mutagenesis buffer (50 mM potassium phosphate,1 mM EDTA [pH 6.0], 0.4 M hydroxylamine) and incubated at

370C for 48 h (19). The mutagenesis reaction was stopped byethanol precipitating the plasmid DNA.

Nucleotide sequencing. Nucleotide sequences were deter-mined by the dideoxy nucleotide chain terminator method (51)using [ot-32P]dATP as the radiolabel, and 7-deazaguanosine5'-triphosphate was substituted for GTP in the sequencingreaction mixtures to reduce compression artifacts (1, 31). Thenucleotide sequences of the 642-bp (L1-603) and 285-bp(LM10-SV4) BamHI-HindIII fragments (Fig. 2B) were re-ported previously (36, 58), and their GenBank accessionnumbers are M26955 and M84717, respectively.

ftr4

e e -10fliF

eeEA CTGGGTAAATCCTGCCT ACT IGACCCATT TAGGACGGA. TG

+ -ee +

Mutation

f IF expression

T(-15)G

itA (-12G.-13G) C(-6)T T(-4)G C(-2)T T(1)G

it t it t itFIG. 6. Summary of ftr4 footprint data. ftr4 is boxed; the arrow

indicates the direction of transcription and the in vivo 5' end of thefliFtranscript (58). Reactivity toward DMS: -, reduced; +, enhanced; 0,bases protected from the hydroxyl radical. Some ftr4 mutations andtheir effects on fliF transcript levels are shown below the sequence.T(-15)G indicates a T-to-G mutation at -15 in fliF. T, elevated fliFtranscript levels (data summarized from reference 58).

VOL. 176, 1994


.org/D

ownloaded from

http://jb.asm.org/

5976 MULLIN ET AL.

A B C D

1 2 3 4 1 2 3 4 1 2 3 4 I 2 A A

bf-

E F

f-i

FIG. 7. ftr4 mutants are defective in binding to FlbDc87. (A)Mobility shift assays were used to compare the binding of FlbDc87protein to probe 5 (Fig. 2C) or to probe 5 carrying the indicated ftr4mutations. (B) Deletion of GG at -12 and -13; (C) mutation of G toT at -12; (D) mutation of T to G at -15. Reactions were carried outas described in Materials and Methods with the indicated amounts ofFlbDc87 protein. Lane 1, no protein; lane 2, 0.06 nmol; lane 3, 0.11nmol; lane 4, 0.28 nmol. f, free probe, b, bound probe (as shifted by theaddition of 0.11 nmol of FlbDc87). (E) Binding of wild-type FlbDc87to a synthetic double-stranded 29-merftr4-containing oligonucleotide(Materials and Methods). (F) Binding of FlbDc87(T-429--4) to the29-mer oligonucleotide. Lane 1, no protein; lane 2, 1.4 pmol; lane 3, 2.8pmol; lane 4, 14 pmol; lane 5, 28 pmol.

RESULTS

Construction of gst-flbD gene fusions and purification ofGST-FlbD fusion proteins. The inferred amino acid of FlbDsuggests that it belongs to a family of regulatory proteins thatbind to enhancer-like sequences and stimulate Cr54 RNApolymerase to initiate transcription (28, 48, 52, 59). Membersof this family of enhancer-binding proteins contain severalfunctional domains, including a helix-turn-helix DNA-bindingmotif near the carboxy terminus (for reviews, see references 32and 42), and experiments with chimeric constructs of NifA andNtrC have shown that carboxy-terminal fragments can interactspecifically with their respective enhancer-like sequences (8,15, 32). Previous work with mobility shift assays demonstratedthat an epitope-tagged FlbD fusion protein binds to fir se-

quences in flaN and flbG (60), and we have sought to deter-mine if carboxy-terminal fragments of FlbD can bind to ftrsequences.

Six in-frame gst-flbD gene fusions were constructed in plas-mid pGEX-2T (55) to facilitate the high-level production andpurification of polypeptides carrying carboxy-terminal frag-ments of FlbD (Fig. 3). The GST-FlbD fusion point of eachconstruct was determined by nucleotide sequencing, and theseare shown in Fig. 3B. Constructs were transferred to E. coliDH5a by transformation (27), and IPTG was used to inducesynthesis of these GST-FlbD fusion proteins. The four largestfusion proteins, c371, c308, c290, and c289 (Fig. 3B), were

insoluble, and they were not characterized further (data not

shown). The two smallest fusion proteins, FlbDc87 andFlbDc48, which carry the carboxy-terminal 87 and 48 aminoacids, respectively, were soluble, and they were purified byaffinity chromatography on glutathione-Sepharose 4B.SDS-PAGE revealed that purified FlbDc87 and GST each

contain a single major protein species with apparent molecularmasses of 35 and 26 kDa, respectively (Fig. 4A). Thesemolecular masses are consistent with the calculated molecularweights of 35,195.9 for FlbDc87 and 25,496.6 for GST based ontheir inferred amino acid sequences (48, 55). FlbDc48 prepa-rations contained the expected 31-kDa FlbDc48 protein and asmaller protein band of about 29 kDa (Fig. 4A). The 29-kDaprotein may have resulted from proteolysis of FlbDc48 or frompremature termination of transcription or translation ofFlbDc48. The apparent native molecular masses as estimatedby gel filtration chromatography on Sephadex G-200 are 130kDa for FlbDc87, 63 kDa for FlbDc48, and 60 kDa for GST(Fig. 4B). The native FlbDc87, thus, appears to be a homotet-ramer, whereas native FlbDc48 and GST appear to be ho-modimers.F1bD has a carboxy-terminal DNA-binding domain that

recognizesftr4. FlbDc87 protein formed complexes with probe6 (Fig. 2C) which carries the fliF promoter and ftr4, and DMSfootprint analysis demonstrated that FlbDc87 is in close con-tact with G residues at -3, -12, -13, and -14 on thenontemplate strand offtr4 (Fig. SA). Hydroxyl radical footprintanalysis revealed protection of nucleotides at -4, -5, -6,-16, and -17 on the nontemplate strand (Fig. SA). On thetemplate strand FlbDc87 protected G residues at -1 and -2from DMS and enhanced reactivity of G residues at +3 and -5to DMS (Fig. SB). Hydroxyl radical footprint analysis showedthat FlbDc87 protected nucleotides at +1 and -1 on thetemplate strand (Fig. SB). The protection of bases at +3 and- 17, which are outside of the ftr consensus sequence, suggeststhat the nucleotides recognized by FlbD extend beyond theconsensus sequence proposed previously for ftr (38). Com-plexes between FlbDc48 protein and probe 6 were not de-tected (data not shown). FlbDc87 and FlbDc48 both containthe carboxy-terminal helix-turn-helix motif of FlbD; however,FlbDc87 contains an additional 39 amino acids that appear tobe essential for binding to ftr4.The close contacts between FlbDc87 and nucleotides in ftr4

at + 1, -2, -4, -6, -12, and -13 are functionally significantbecause mutation of these positions has been shown previouslyto result in greatly elevated levels of fliF transcripts in vivo(Fig. 6) (58). We also investigated the in vitro binding ofFlbDc87 to DNA fragments carrying three different flr4 muta-tions that resulted in elevated fliF transcript levels. A total of0.06 nmol of FlbDc87 added to the DNA-binding reactionmixture was sufficient to shift all of the wild-type probefragment to a more slowly migrating form, although the shift isnot very pronounced (Fig. 7A, lane 2). A total of 0.11 nmol ofFlbDc87 yielded ftr4-specific footprints (Fig. 5) and a morepronounced mobility shift with probe 5, but probe 5 carryingpoint mutations of G to T at -12 and T to G at -15 and adeletion of the GG dinucleotide at - 12 and - 13 was defectivein binding to FlbDc87 (Fig. 7B, C, and D). At a level of 0.28nmol of FlbDc87 per reaction mixture, both wild-type andmutant probes were shifted equally, suggesting that this levelof protein leads to nonspecific binding. The correlation ofelevatedfliF transcript levels in vivo caused by mutations inftr4(58) to the reduced binding of FlbDc87 in vitro to fragmentscarryingftr4 mutations (Fig. 7) strongly suggests that the FlbDfunctions directly as a negative regulator of fliF transcriptionby binding toftr4. Consistent with this prediction, C. crescentus(pRK290-c87) which expressed FlbDc87 was nonmotile. In

J. BACT1ERIOL.


.org/D

ownloaded from

http://jb.asm.org/

VOL. 176, 1994

Af b

'4

--O/ 90.

NN'4'4.

/1An/---I

,130'1

B

f b

ICGGC

.AAGTTTC

CCGG

6

TT

CGCGG

TTITGCCaAG

i/go/J,.90

ft,2

I+ftr3

G

a

C

G

C

TCAAAGCGGCC

' CJ

C

CCCA

A

C

aaG

CT

'4 CG


Cf b

+

_'W

-G

-cT

G

A-110 AIAIA

A

IG*G

Df b

+

ftrl

MIT Ftrl

FIG. 8. FlbDc87 binding to ftr sequences in the flaN-flbG promoter region. DMS footprint reactions are described in Materials and Methods;DNA probes are shown in Fig. 2C. -, reduced reactivity toward DMS; +, enhanced reactivity toward DMS. Lanes b, FlbDc87 protein added; lanesf, no FlbDc87 protein. (A) Nontemplate strand of ftr2 and ftr3, using probe 4. (B) Template strand of ftr2 and ftr3, using probe 3. (C) ftrl(transcribed strand of flbG), using probe 2. (D) ftrl and ftrS (nontranscribed strand of flbG), using probe 1.

contrast, pRK290-c48 which expresses the FlbDc48 proteinthat is defective in ftr4 binding had no effect on motility of C.crescentus, as determined by swarm plate analysis and micros-copy.To investigate the loss of motility that resulted from expres-

sion of FlbDc87, we mutagenized pRK290-c87 with hydroxyl-amine in vitro and screened for tetracycline-resistant electro-porants of C. crescentus that remained motile (Mot').Sequence analysis of theflbD portion of pRK290-c87 from onesuch Mot' strain revealed that it had acquired a C-to-T pointmutation that converted threonine 429 to isoleucine. Threo-nine 429 forms part of the recognition helix of the predictedhelix-turn-helix DNA-binding motif of FlbD (48), and consis-tent with this prediction is the finding that purified FlbDc87(T-429-I) was defective in binding to a double-stranded ftr4-containing oligonucleotide probe in vitro (Fig. 7E and F).These results suggest that FlbDc87 expression inhibits motilityin vivo by binding to ftr4 and preventing transcription of fliF.FlbDc87 binds to ftr sequences in the flaN and flbG pro-

moter regions. FlbD is required for transcription of the flaNandflbG operons (35,44), and both of these transcription unitscontain fir sequences that are required for positive and nega-tive regulation of transcription (18, 36, 37). We previouslyinvestigated the role of individualftrs in regulation offlaN andflbG transcription by site-directed mutagenesis (36, 37), and inthis study we used DMS footprint analysis to investigate theinteraction of FlbDc87 with specific residues in fri, ftr2, ftr3,and firS.

ftr3 is located at + 120 from the transcription start site offlaN, and it is a site of negative transcriptional regulation ofboth flaN and flbG (37). On the nontemplate strand of ftr3FlbDc87 binding resulted in enhanced reactivity of DMStoward the G residue at + 136 (Fig. 8A). Figure 8B shows that

FlbDc87 bound to ftr3 and protected G residues at positions+ 132 and + 133 on the flaN template strand from DMS.Footprint analysis also revealed enhanced reactivity towardDMS of the G residues at position + 137, which is one residuedownstream fromftr3 (Fig. 8B). Protection of the G residue at+ 133 from DMS reaction is functionally significant because amutation of G to T at this position resulted in elevated levelsof the flaN and flbG transcripts (Fig. 9) (37).

Site-directed mutagenesis demonstrated that ftr2 at +86 inflaN is a cis-acting requirement forflaN transcription (18, 37),and Fig. 8B shows that FlbDc87 binding to the template strandof ftr2 resulted in protection of G residues at positions +89,+98, and +99 and enhanced DMS reactivity at residue +86.DMS footprint analysis of FlbDc87 binding to the nontemplatestrand offtr2 identified protected G residues at positions +87,+97, + 101, and + 102 (Fig. 8A). The functional significance ofthe interaction of FlbDc87 at +89, +98, and +99 is supportedby the previous finding (37) that mutations at +89 and +98and a deletion of the GG dinucleotide at +98 and +99abolished transcription offlaN in vivo (Fig. 9).frl is located at -101 from the transcription start site of

flbG, and site-directed mutagenesis has shown that it is a site ofpositive and negative regulation offlbG and that it is a negativeregulator offlaN (36, 37). DMS footprint analysis showed thatFlbDc87 protected G residues inftrl at -104, -113, and -114from the flbG transcription start site on the nontemplatestrand and enhanced the reactivity of DMS with the G at -118(Fig. 8C). On the other strand offtrl, FlbDc87 binding resultedin protection of the G residues at -100 and -112 andenhanced reactivity of DMS towards the G at -115 (Fig. 8D).The functional significance of the interaction of FlbDc87 at- 113 in vitro is supported by the finding that a point mutationat this residue abolishedflbG transcription in vivo (Fig. 9) (36).


.org/D

ownloaded from

http://jb.asm.org/

5978 MULLIN ET AL.

ftr3++M

ftr2+90

CICACCCCTTTCCCCC GGCCGCTTGAAAHC+

ftr5 ftrl-140 -no fbG

,3 + -. ;2 11elITCfAAlAGflI

ICCCCCCTTCSIAAciuA! C GAOCCCTTTT TCCCCCCC U U

(54 + a 54

flaN

Mutation G(133)T

1fIaN expression

fibG expression

A (G98.G99) G(98)T G(89)T

I l IG(-145)T

w

G(-113)T

w

w w w

G(-11)T

wFIG. 9. Summary footprint data for the flaN-fibG promoter region. ftr, ihf, and u54 promoters are shown; the flaN and flbG transcripts are

indicated by arrows. Shown below the sequence are some mutations, and their effects on flaN and flbG transcript levels are summarized fromreferences 36 and 37. W, wild-type level; T , elevated level; I, reduced level.

We also detected protection of G residues at -145 and -150from the transcription start site of flbG (Fig. 8D), and thesenucleotides are part of a sequence previously designated ftr5(38) that matches the fir consensus sequence at 14 of 17positions (Fig. 9).

DISCUSSION

The flagellar genes of C. crescentus are organized into aregulatory hierarchy that consists of four levels of genes (Fig.1), and expression of these genes is under cell cycle control. Tobetter understand the regulation of these flagellar genes at themolecular level, we have investigated some of the cis- andtrans-acting elements that are required for their expression. Inthis study we have focused our attention on the DNA-bindingactivity of the FlbD regulatory protein and its interaction withfir sequences that are required for regulated expression offlaNand flbG (36, 37) at level III and fliF (58) at level II.The flaN and flbG operon genes code for proteins that are

required for morphogenesis of the flagellar hook (45). Se-quence analysis, mutagenesis, and transcriptional mappinghave shown that the cis-acting elements needed for regulatedexpression offlaN andflbG include S54 promoters and enhanc-er-like ftr and ihf sequences (Fig. 2B) (10, 17, 18, 35-37).trans-acting requirements for flaN and flbG transcription in-clude RpoN, which is the a5" sigma factor (5); FlbF (10, 35, 44,50); and FlbD (48).flbD was first identified as a trans-acting requirement for

transcription of the flaN (35) and flbG operons (10, 44), andfurther studies showed thatflbD is a negative autoregulator ofits own transcription (40, 58). Nucleotide sequencing andtranscriptional analysis revealed that flbD is the last of fivegenes in thefliF operon (GenBank accession no. M98855) (48,50, 58), and flbD codes for a 52-kDa polypeptide that belongsto a family of enhancer-binding transcriptional regulatoryproteins (48). These enhancer-binding proteins contain con-served amino acid sequence motifs, including an amino-termi-nal domain that is the target of regulatory signals, a largecentral domain that plays a role in activation of u54 RNApolymerase, and a carboxy-terminal enhancer-binding domainthat contains a helix-turn-helix motif (7, 15, 33, 57). In the caseof two such enhancer-binding proteins, DctD and NifA, thetranscriptional activation domain and the DNA-binding do-mains retain their respective activities when expressed sepa-rately (21, 22, 32).

In the first functional analysis of FlbD, Wingrove et al. used

mobility shift assays to show that an epitope-tagged FlbDfusion protein binds toftr sequences inflaNflbG,flgL, andflgK(60), and in this study we found that a GST-FlbD fusionprotein called FlbDc87 carrying the 87 amino acids whichcomprise the carboxy terminus of FlbD bound to fir sequencesin the flaN-flbG promoter region. DMS footprint analysisidentified G nucleotides in ftrl, flr2, and flr3 that are in closecontact with FlbDc87, and close-contact bases in each of thesethree fir sequences were shown previously to play a role inregulation offlaN and flbG expression in vivo (summarized inFig. 9) (18, 35, 38).DMS footprint analysis also revealed that FlbDc87 pro-

tected the G residues at -145 and -150 from the flbGtranscription start site, and these nucleotides are located in anftr-like sequence that was provisionally designated ftr5 (Fig. 8and 9) (37). We introduced a G-to-T mutation at -145;however, this mutation had no apparent effect on the in vivolevels offlaN orflbG transcripts (34). This result suggests thatthe G residue at position -145 is not essential for expressionofflaN orflbG, although it is possible that other nucleotides inftrS play a role in expression, and a more thorough mutagenicanalysis will be required to determine if firS is required forregulated expression offlaN and flbG.

Other level III flagellar genes that have a54 promotersinclude flgI, which codes for the P-ring protein (24); flgH,which codes for the L-ring protein; and flgF, which codes forthe proximal rod protein (11). The promoter regions of flgFand flgI both contain ihf and fir sequences (Fig. 1) (11, 24, 38),and DNA fragments carrying the ftr-like sequences offlgF andflgI bound to FlbDc87 in mobility shift assays (34).

The level II operons fliF, fliQR, and fliLM are among theearliest flagellar transcription units to be expressed in the cellcycle, and they code for proteins required for synthesis,assembly, and function of the basal body (GenBank accessionno. M98855) (11, 43, 58, 64). These transcription units have anovel promoter that is different in nucleotide sequence fromthe a5 promoters used by the level III and level IV genes, andit is different from other previously described promoter motifs(12, 58, 64). These level II promoters contain a stronglyconserved 18-bp sequence, and in the case of fliF and fliLM,point mutations that altered these conserved sequences re-sulted in reduced levels of their respective transcripts (56, 58).The fliF transcription start site lies within an ftr-like se-

quence calledftr4 that functions as a site of negative regulation(58). fliF is the only level II gene known to have an firsequence, and ftr appears to be the only cis-acting regulatory

J. BACTERIOL.


.org/D

ownloaded from

http://jb.asm.org/


sequence shared byfliF and the level III and level IV flagellargenes (Fig. 1). DMS and hydroxyl radical footprint analysesdemonstrated that FlbDc87 made close contacts with 11nucleotides inftr4 and 2 nucleotides outside of ftr4 (Fig. 5). Sixof the closely contacted nucleotides in ftr4 are directly relevantto negative control because they are required for negativeregulation of fliF transcription in vivo (58), and mutations atthree of these nucleotides reduced the ability of FlbDc87 tobind ftr4 in vitro (Fig. 7). These results suggest that FlbD actsdirectly as a transcriptional repressor by binding to ftr4.Consistent with this prediction is the finding that expression ofFlbDc87 in C. crescentus caused the cells to become nonmotile,presumably because flagellar gene expression was repressed.Our results demonstrate that the DNA-binding activity ofFlbD is carried within the carboxy terminus, and the findingthat a T-429-- mutation in the recognition helix of theproposed helix-turn-helix abolishes binding to ftr4 in vitroprovides evidence that the ftr-binding activity of FlbD ismediated by this helix-turn-helix sequence. While this workwas in review, Benson et al. published a paper demonstratingthat the intact FlbD makes close contact with ftrl,ftr4, and ftrS(which they referred to as ftrl*) (2).Although FlbDc87 bound to ftr sequences in flaN, flbG, and

fliF, several different preparations of a shorter polypeptide,FlbDc48 (Fig. 3), which contains the putative helix-turn-helixsequence of FlbD, failed to bind to DNA fragments carryingftr sequences in mobility shift assays (34). Consistent withthe failure to detect binding of FlbDc48 to firs and unlikeFlbDc87, expression of FlbDc48 in C. crescentus did not inhibitmotility. The apparent molecular masses of FlbDc87, FlbDc48,GST as determined by SDS-PAGE are 35, 31, and 26 kDa,respectively (Fig. 4A), which are consistent with their calcu-lated molecular weights based on the inferred amino acidsequences (48, 55). The native molecular weights as estimatedfrom gel filtration chromatography suggest that FlbDc87 is ahomotetramer, whereas FlbDc48 and GST appear to be ho-modimers, and it seems likely that dimerization of FlbDc48 ismediated by the GST domain. Included within the 39 aminoacids of FlbDc87 that are missing from FlbDc48 are sequencesthat are similar to the A and B helix sequences of the E. coli Fisprotein (Fig. 3B) which are proposed to form the dimerizationdomain of Fis (25, 26, 65) and NtrC (42). The failure to detectbinding of FlbDc48 to ftrs may, therefore, reflect a lack ofamino acid sequences that are needed to dimerize the FlbDDNA-binding domains.How does FlbD function as a temporal regulator of flagellar

gene transcription? In the case of the fliF operon, FlbDappears to act as a repressor of transcription by binding toftr4,and this negative regulation may help to restrict fliF operonexpression to the predivisional period during which its proteinproducts are needed for flagellum synthesis.A previously proposed model to explain the coordinate

activation of flaN and flbG transcription predicted that FlbDmight bind at fir sequences and interact with promoter-boundu5 RNA polymerase, and this interaction between FlbD andcr RNA polymerase would then lead to active transcription(17, 18, 37). Consistent with a prediction of this model is thefinding here that FlbDc87 made close contact with nucleotidesinftrl and ftr2 that are required for positive regulation offlaNandflbG (Fig. 9). An additional level of positive control offlaNand flbG expression by the c5A sigma factor was suggested bythe finding that the temporal transcription pattern of rpoNparallels those offlaN and flbG (5).According to the proposed model (37), coordinate negative

transcriptional regulation offlaN andflbG would result from aprotein-mediated interaction of theflaN andflbG transcription

complexes leading to repression of transcription. Nucleotidesin irl and Jfr3 have been shown previously to play a role innegative regulation (summarized in Fig. 9) (36, 37), suggestingthe possibility that in addition to its role in positive regulation,FlbD also acts as a negative regulator of flaN and flbGtranscription.

In summary, we have demonstrated that the C. crescentusFlbD regulatory protein has a sequence-specific DNA-bindingdomain near its carboxy terminus that recognizes nucleotidesin ftr sequences that are involved in positive and negativeregulation of flaN and flbG and in negative regulation of fliF.

ACKNOWLEDGMENTS

Brian Meyer is acknowledged for providing excellent technicalassistance.

This investigation was supported by grants to D.A.M. from theBoard of Regents of the State of Louisiana [LEQSF(87-90)-RD-A-19],the Newcomb College Foundation, and the Tulane University Molec-ular and Cellular Biology Graduate Program.

REFERENCES1. Barr, P. J., R. M. Thayer, P. Laybourn, R. C. Najarian, F. Seela,

and D. R. Tolan. 1986. 7-Deaza-2'-deoxyguanosine-5'-triphos-phate: enhanced resolution in M13 dideoxy sequencing. BioTech-niques 4:428-432.

2. Benson, A. K., G. Ramakrishnan, N. Ohta, J. Feng, A. J. Ninfa,and A. Newton. 1994. The Caulobacter crescentus FlbD protein actsat ftr sequence elements both to activate and to repress transcrip-tion of cell cycle-regulated flagellar genes. Proc. Natl. Acad. Sci.USA 91:4989-4993.

3. Bradford, M. M. 1976. A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72:248-254.

4. Brun, Y. V., G. Marczynski, and L. Shapiro. 1994. The expressionof asymmetry during Caulobacter cell differentiation. Annu. Rev.Biochem. 63:419-450.

5. Brun, Y. V., and L. Shapiro. 1992. A temporally controlled a factoris required for polar morphogenesis and normal cell division inCaulobacter. Genes Dev. 6:2395-2408.

6. Bryant, R., D. Glaser, and L. Shapiro. 1990. Genetic regulatoryhierarchy in Caulobacter development. Adv. Genet. 27:1-31.

7. Buikema, W. J., W. W. Szeto, P. V. Lemley, W. H. Orme-Johnson,and F. M. Ausubel. 1985. Nitrogen fixation specific regulatorygenes of Kiebsiella pneumoniae and Rhizobium meliloti sharehomology with the general nitrogen regulatory gene ntrC of Kpneumoniae. Nucleic Acids Res. 13:4539-4555.

8. Cannon, W., W. Chariton, and M. Buck. 1991. Organization andfunction of binding sites for the transcriptional activator NifA inthe Kiebsiella pneumoniae nifE and nifU promoters. J. Mol. Biol.220:915-931.

9. Champer, R., A. Dingwall, and L. Shapiro. 1987. Cascade regula-tion of Caulobacter flagellar and chemotaxis genes. J. Mol. Biol.194:71-80.

10. Chen, L.-S., D. Mullin, and A. Newton. 1986. Identification,nucleotide sequence, and control of developmentally-regulatedpromoters in the hook operon region of Caulobacter crescentus.Proc. Natl. Acad. Sci. USA 83:2860-2864.

11. Dingwall, A., D. J. Garman, and L. Shapiro. 1992. Organizationand ordered expression of Caulobacter genes encoding flagellarbasal body rod and ring proteins. J. Mol. Biol. 228:1147-1162.

12. Dingwall, A., A. Zhuang, K. Quon, and L. Shapiro. 1992. Expres-sion of an early gene in the flagellar regulatory hierarchy issensitive to an interruption in DNA replication. J. Bacteriol.174:1760-1768.

13. Ditta, G., S. Stanfield, D. Corbin, and D. R. Helinski. 1980. Broadhost range DNA cloning system for gram-negative bacteria: con-struction of a gene bank of Rhizobium meliloti. Proc. Natd. Acad.Sci. USA 77:7347-7351.

14. Dixon, W. J., J. J. Hayes, J. R. Levin, M. F. Weidner, B. A.Dombrowski, and T. D. Tullius. 1991. Hydroxyl radical footprint-ing. Methods Enzymol. 208:380-415.

VOL. 176, 1994


.org/D

ownloaded from

http://jb.asm.org/

5980 MULLIN ET AL.

15. Drummond, M. H., A. Contreras, and L A. Mitchenal. 1990. Thefunction of isolated domains and chimaeric proteins constructedfrom the transcriptional activators NifA and NtrC of Klebsiellapneumoniae. Mol. Microbiol. 4:29-37.

16. Ely, B., and T. W. Ely. 1989. Use of pulsed field gel electrophoresisand transposon mutagenesis to estimate the minimal number ofgenes required for motility in Caulobacter crescentus. Genetics123:649-654.

17. Gober, J. W., and L. Shapiro. 1990. Integration host factor isrequired for the activation of developmentally regulated genes inCaulobacter. Genes Dev. 4:1494-1504.

18. Gober, J. W., and L Shapiro. 1992. A developmentally regulatedCaulobacter flagellar promoter is activated by 3' enhancer and IHFbinding elements. Mol. Biol. Cell 3:913-926.

19. Hashimoto, T., and M. Seklguchi. 1976. Isolation of temperature-sensitive mutants of R plasmid by in vitro mutagenesis withhydroxylamine. J. Bacteriol. 127:1561-1563.

20. Howard-Flanders, P., E. Simson, and T. Theriot. 1964. A locusthat controls filament formation and sensitivity to radiation inEscherichia coli. Genetics 49:237-246.

21. Huala, E., and F. M. Ausubel. 1989. The central domain ofRhizobium meliloti NifA is sufficient to activate transcription fromthe R. melioti nifH promoter. J. Bacteriol. 171:3354-3365.

22. Huala, E., J. Stigter, and F. M. Ausubel. 1992. The central domainof Rhizobium leguminosarum DctD functions independently toactivate transcription. J. Bacteriol. 174:1428-1431.

23. Johnson, R., and B. Ely. 1979. Analysis of nonmotile mutants ofthe dimorphic bacterium Caulobacter crescentus. J. Bacteriol.137:627-634.

24. Khambaty, F. M., and B. Ely. 1992. Molecular genetics of the figIregion and its role in flagellum biosynthesis in Caulobacter cres-centus. J. Bacteriol. 174:4101-4109.

25. Kostrewa, D., J. Granzin, C. Koch, H.-W. Choe, S. Raghunathan,W. Wolf; J. Labahn, R. Kahmann, and W. Saenger. 1991. Three-dimensional structure of the E. coli DNA binding protein FIS.Nature (London) 349:178-180.

26. Kostrewa, D., J. Granzin, D. Stock, IL-W. Choe, J. Laban, and W.Saenger. 1992. Crystal structure of the factor for inversion stimu-lation FIS at 2.0 A resolution. J. Mol. Biol. 226:209-226.

27. Lederberg, E. M., and S. N. Cohen. 1974. Transformation ofSalmonella typhimurium by plasmid DNA. J. Bacteriol. 119:1072-1074.

28. Lee, H.-S., D. K. Beefger, and S. Kustu. 1992. Activity of purifiedNifA, a transcriptional activator of nitrogen fixation genes. Proc.Natl. Acad. Sci. USA 90:2266-2270.

29. Leong, J. M., S. Nunes-Duby, C. F. Lesser, P. Youderian, M. M.Susskind, and A. Landy. 1985. The 4+80 and P22 attachment sites.Primary structure and interaction with Escherichia coli integrationhost factor. J. Biol. Chem. 260:4468-4477.

30. Minnich, S. A., and A. Newton. 1987. Promoter mapping and cellcycle regulation of flagellin gene transcription in Caulobactercrescentus. Proc. Natl. Acad. Sci. USA 84:1142-1146.

31. Mizusawa, S., S. Nishimura, and F. Seela. 1986. Improvement ofthe dideoxy chain termination method of DNA sequencing by useof deoxy-7-deazaguanosine triphosphate in place of dGTP. Nu-cleic Acids Res. 14:1319-1323.

32. Morett, E., W. Cannon, and M. Buck. 1988. The DNA-bindingdomain of the transcriptional activator protein NifA resides in itscarboxy terminus, recognizes the upstream activator sequences ofnif promoters and can be separated from the positive controlfunction of NifA. Nucleic Acids Res. 16:11469-11488.

33. Morett, E., and L Segovia. 1993. The "54 bacterial enhancer-binding protein family: mechanism of action and phylogenicrelationship of their functional domains. J. Bacteriol. 175:6067-6074.

34. Mullin, D. A. Unpublished data.35. Mullin, D. A., S. A. Minnich, L.-S. Chen, and A. Newton. 1987. A

set of positively regulated flagellar gene promoters in Caulobactercrescentus with sequence homology to the nif gene promoters ofKlebsiella pneumoniae. J. Mol. Biol. 195:939-943.

36. Mullin, D. A., and A. Newton. 1989. Ntr-like promoters andupstream regulatory sequence ftr are required for transcription ofa developmentally regulated Caulobacter crescentus flagellar gene.

J. Bacteriol. 171:3218-3227.37. Mullin, D. A., and A. Newton. 1993. A (r54 promoter and down-

stream sequence elements ftr2 and ftr3 are required for positiveand negative regulation of divergent transcription units flaN andflbG in Caulobacter crescentus. J. Bacteriol. 175:2067-2076.

38. Mullin, D. A., and A. H. Mullin. 1993. cis- and trans-actingelements required for regulation of flagellar gene transcription inthe bacterium Caulobactercrescentus. Cell. Mol. Biol. Res. 39:361-369.

39. Newton, A., and N. Ohta. 1990. Regulation of the cell division cycleand differentiation in bacteria. Annu. Rev. Microbiol. 44:689-719.

40. Newton, A., N. Ohta, G. Ramakrishnan, D. A. Mullin, and G.Raymond. 1989. Genetic switching in the flagellar gene hierarchyrequires both positive and negative regulation of transcription.Proc. Natl. Acad. Sci. USA 86:6651-6655.

41. Ninfa, A. J., D. A. Mullin, G. Ramakrishnan, and A. Newton. 1989.Escherichia coli sigma-54 RNA polymerase recognizes Caulobactercrescentus flbG and flaN flagellar promoters in vitro. J. Bacteriol.171:383-391.

42. North, A. K., K. E. Klose, K. M. Stedman, and S. Kustu. 1993.Prokaryotic enhancer-binding proteins reflect eukaryote-like mod-ularity: the puzzle of nitrogen regulatory protein C. J. Bacteriol.175:4267-4273.

43. Ohta, N., L S. Chen, D. A. Mullin, and A. Newton. 1991. Timingof flagellar gene expression in the Caulobacter cell cycle isdetermined by a transcriptional cascade of positive regulatorygenes. J. Bacteriol. 173:1514-1522.

44. Ohta, N., L.-S. Chen, E. Swanson, and A. Newton. 1985. Transcrip-tional regulation of a periodically controlled flagellar gene operonin Caulobacter crescentus. J. Mol. Biol. 186:107-115.

45. Ohta, N., E. Swanson, B. Ely, and A. Newton. 1984. Physicalmapping and complementation analysis of TnS mutations inCaulobacter crescentus: organization of transcriptional units in thehook cluster. J. Bacteriol. 158:897-904.

46. Osley, M. A., M. Shefery, and A. Newton. 1977. Regulation offlagellin synthesis in the cell cycle of Caulobacter: dependence onDNA replication. Cell 12:393400.

47. Poindexter, J. S. 1964. Biological properties and classification ofthe Caulobacter group. Bacteriol. Rev. 28:231-295.

48. Ramakrishnan, G., and A. Newton. 1990. FlbD of Caulobactercrescentus is a homologue of NtrC (NRI) and activates oC5dependent flagellar gene promoters. Proc. Natl. Acad. Sci. USA87:2369-2373.

49. Ramakrishnan, G., J.-L. Zhao, and A. Newton. 1991. The cellcycle-regulated flagellar gene flbF of Caulobacter crescentus ishomologous to a virulence locus (1crD) of Yersinia pestis. J.Bacteriol. 173:7283-7292.

50. Sanders, L A., S. M. Van Way, and D. A. Mullin. 1992. Charac-terization of the Caulobacter crescentus flbF promoter and identi-fication of the inferred FlbF product as a homolog of the LcrDprotein from a Yersinia enterocolitica virulence plasmid. J. Bacte-riol. 174:857-866.

51. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencingwith chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA74:5463-5467.

52. Sasse-Dwight, S., and J. D. Gralla. 1988. Probing the Escherichiacoli glnALG upstream activation mechanism in vivo. Proc. Natl.Acad. Sci. USA 85:8934-8938.

53. Shapiro, L 1985. Generation of polarity during Caulobacter celldifferentiation. Annu. Rev. Cell Biol. 1:173-207.

54. Shefery, M., and A. Newton. 1981. Regulation of periodic proteinsynthesis in the Caulobacter crescentus cell cycle: control ofinitiation and termination of flagellar gene expression. Cell 24:49-57.

55. Smith, D. B., and K. S. Johnson. 1988. Single-step purification ofpolypeptides expressed in Escherichia coli as fusions with glutathi-one S-transferase. Gene 67:31-40.

56. Stephens, C. M., and L. Shapiro. 1993. An unusual promotercontrols cell-cycle regulation and dependence on DNA replicationof the Caulobacter ftiLM early flagellar operon. Mol. Microbiol.9:1169-1179.

J. BAcrERIOL.


.org/D

ownloaded from

http://jb.asm.org/


57. Stock, J. B., A. J. Ninfa, and A. M. Stock. 1989. Protein phosphor-ylation and regulation of adaptive responses in bacteria. Micro-biol. Rev. 53:450490.

58. Van Way, S. M., A. Newton, A. H. Mullin, and D. A. Mullin. 1993.Identification of the promoter and a negative regulatory element,ftr4, that is needed for cell-cycle timing offliF operon expression inCaulobacter crescentus. J. Bacteriol. 175:367-376.

59. Weiss, D. S., J. Batut, K. E. Klose, J. Keener, and S. Kustu. 1991.The phosphorylated form of the enhancer-binding protein NTRChas an ATPase activity that is essential for activation of transcrip-tion. Cell 67:155-167.

60. Wingrove, J. A., E. K. Mangan, and J. W. Gober. 1993. Spatial andtemporal phosphorylation of a transcriptional activator regulatespole-specific gene expression in Caulobacter. Genes Dev. 7:1979-1992.

61. Wissmann, A., and W. Hillen. 1991. DNA contacts probed by

modification protection and interference studies. Methods Enzy-mol. 208:365-379.

62. Xu, H., A. Dingwall, and L Shapiro. 1989. Negative transcriptionalregulation in the Caulobacter flagellar hierarchy. Proc. Natl. Acad.Sci. USA 86:6656-6660.

63. Yanisch-Perron, C., J. Vieira, and J. Messing. 1985. ImprovedM13 cloning vectors and host strains: nucleotide sequences of theM13mpl8 and pUC19 vectors. Gene 33:103-119.

64. Yu, J., and L Shapiro. 1992. Early Caulobacter crescentus genesfliL andfliM are required for flagellar gene expression and normalcell division. J. Bacteriol. 174:3327-3338.

65. Yuan, H. S., S. E. Finkel, J.-A. Feng, M. Kaczor-Grzeskowiak,R. C. Johnson, and R. E. Dickerson. 1991. The molecular structureof wild-type and a mutant Fis protein: relationship betweenmutational changes and recombinational enhancer function orDNA binding. Proc. Natl. Acad. Sci. USA 88:9558-9562.

VOL. 176, 1994


.org/D

ownloaded from

http://jb.asm.org/

Has DNA-BindingActivity Its Carboxy Terminus That ... · i2, andi3), andftrs (fl, f2, f3, f4, andf5) in thefliFandflaN-flbGpromoterregions; thethin line is genomicDNA,andthe thickline

Documents