Characterization of the Lactococcus lactis transcription factor FlpA and demonstration of an in vitro switch

Molecular Microbiology (2000) 35(6), 1383±1393

Characterization of the Lactococcus lactis transcriptionfactor FlpA and demonstration of an in vitro switch

Colin Scott, John R. Guest and Jeffrey Green*

The Krebs Institute for Biomolecular Research,

Department of Molecular Biology and Biotechnology,

University of Sheffield, Western Bank, Sheffield S10 2TN,

UK.

Summary

The commercially important bacterium Lactococcus

lactis contains two FNR-like proteins (FlpA and FlpB)

which have a high degree of identity to each other

and to the FLP of Lactobacillus casei. FlpA was

isolated from a GST±FlpA fusion protein produced in

Escherichia coli. Like FLP, isolated FlpA is a homo-

dimeric protein containing both Zn and Cu. However,

the properties of FlpA were more like those of the E.

coli oxygen-responsive transcription factor FNR than

the FLP of L. casei. As prepared FlpA recognized an

FNR site (TTGAT-N4-ATCAA) but not an FLP site

(CCTGA-N4-TCAGG) in band-shift assays. In contrast

to FLP, DNA binding by FlpA did not require the

formation of an intramolecular disulphide bond.

However, despite containing only two cysteine

residues per monomer, FlpA was able to acquire an

FNR-like, oxygen-labile [4Fe 4S] cluster. But, whereas

the incorporation of a [4Fe 4S] cluster into FNR

enhances interaction with target DNA, it abolished

DNA binding by FlpA. An FlpA variant (FlpA 0) with an

N-terminal region designed to be more FLP-like failed

to incorporate an iron±sulphur cluster but could now

form an intramolecular disulphide. This simple exam-

ple of protein engineering, converting an oxygen-

labile [4Fe 4S] containing FNR-like protein into a

dithiol±disulphide FLP-like redox sensor demon-

strates the versatility of the basic CRP structure.

Attempts to demonstrate an FlpA-based aerobic±

anaerobic switch in the heterologous host E. coli

were unsuccessful. However, studies with a series of

FNR-dependent lac reporter fusions in strains of E.

coli expressing flpA or flpB revealed that both

homologues were able to activate expression of

FNR-dependent promoters in vivo but only when

positioned 61 base pairs upstream of the transcription

start.

Introduction

Lactic acid bacteria are of major economic importance

because of their use in the production and preservation of

many fermented foods. Advances in the understanding of

the genetics and molecular biology of these bacteria has

provided new opportunities for biotechnological exploita-

tion (Gasson, 1993). One aspect of their biology that may

prove amenable to manipulation are the global regulatory

circuits that respond to metabolic and environmental

stimuli. The CRP family of global transcription regulators

control gene expression in response to stimuli such as

glucose starvation (CRP) and anoxia (FNR) in a wide

range of bacteria. The first member of the CRP family to

be identified in a Gram-positive bacterium was the

FNR-like protein (FLP) of Lactobacillus casei (Irvine and

Guest, 1993). Although lacking two of the four essential

cysteine residues that act as ligands for an oxygen-labile

[4Fe 4S] cluster in FNR, FLP was designated FNR-like

because its DNA binding helix retained the E± ±SR motif

that was predicted to confer specificity for an FNR site

(TTGAT± ± ± ±ATCAA). However, isolated FLP did not

recognize an FNR site but a related sequence

(CCTGA± ± ± ±TCAGG) in which the TGA core motifs

are separated by only four bases, suggesting that the

DNA-binding helices of FLP penetrate deeper into the

major groove and thus must be more closely configured

than those of FNR (Gostick et al., 1998). Because FLP

lacks two of the four essential cysteine residues of FNR it

was not surprising that it failed to incorporate a [4Fe 4S]

cluster but operated a different redox-sensing switch

based on the reversible interconversion of an intramole-

cular disulphide bond (Gostick et al., 1998). Recently, two

transcriptional regulators of the CRP family (FlpA and

FlpB), each 41% identical to FLP from L. casei, have been

identified in Lactococcus lactis as the distal components of

two paralogous operons, each containing three genes

(Gostick et al., 1999). Both of the potential regulators in L.

lactis are predicted to share the same essential secondary

structural features of CRP and FNR. The similarity of the

Flp DNA binding helices with that of FNR and FLP

suggests that both lactococcal Flp regulators recognize

either an FNR or an FLP site and, intriguingly, both flp

operons have FNR sites appropriately positioned for

Q 2000 Blackwell Science Ltd

Received 25 September, 1999; revised 29 November, 1999;accepted 3 December, 1999. *For correspondence. E-mail [email protected]; Tel. (144) 114 222 4403; Fax (144) 114272 8697.

1384 The FlpA protein of L: lactis

Q 2000 Blackwell Science Ltd, Molecular Microbiology, 35, 1383±1393

regulatory activity (Gostick et al., 1999). The present work

indicates that the Flp proteins of L. lactis do indeed

recognize FNR sites, suggesting that the FNR boxes

within the promoter regions of the flp operons are

biologically significant sites of regulation. Furthermore,

the DNA binding activity of FlpA can be modulated in vitro

by the acquisition of an oxygen-labile [4Fe 4S] cluster,

despite possessing only two cysteine residues. The non-

cysteinyl ligands reside close to the N-terminal of FlpA and

deletion of this region converts FlpA from a [4Fe 4S] based

oxygen sensor into a dithiol±disulphide-based redox sensor.

Results

Overproduction of FlpA in Escherichia coli and properties

of the isolated protein

FlpA was overproduced in E. coli (JRG3507) as a

glutathione-S-transferase (GST) fusion protein. The over-

produced GST±FlpA fusion protein was mostly (75%) in

the cytoplasmic fraction and after induction by IPTG

constituted 17±20% of soluble cell protein, equivalent to

23 mg of GST±FlpA per litre of culture. However, despite

this efficient expression, cleavage of GST±FlpA with

thrombin only yielded 3.2 mg of pure FlpA per litre of

culture, representing only 21% of the theoretical

maximum (Fig. 1). A second expression vector,

pGS1035, based on pET16b, allowed the purification of

FlpA by a different procedure but, although expression

was once again good, FlpA was only recovered in low

yield. Consequently, because of the rapid purification

protocol developed for the GST±FlpA fusion it was this

expression system that was routinely used in these

studies. However, sufficient FlpA was produced by both

methods to allow some properties to be determined and

compared with those of FLP and FNR. As a consequence

of the strategy devised to purify FlpA an additional 15

amino acids were attached to the N-terminal of FlpA.

Amino acid sequencing (GSPGISGGGGGILDSMEIK)

confirmed the presence of the additional amino acids

and that the isolated protein was FlpA (native FlpA begins

MKIK but a K2E substitution is created in the engineering

of the NcoI restriction site). In common with many other

proteins expressed as GST fusions a number of minor

contaminating polypeptides were copurified along with the

GST±FlpA, but were shown by N-terminal sequencing

and Western blotting to be truncated GST±FlpA products

(not shown).

Analysis of the oligomeric state of purified FlpA

revealed that FlpA is, like FLP, a homodimer, with a Mr

of 27 700 by SDS±PAGE and 54 700 by gel filtration

(Table 1).

FlpA does not operate an FLP-like switch

The FLP from L. casei has been shown to be associated

with substoichiometric quantities of zinc and copper

following isolation from E. coli (Gostick et al., 1998).

Total metal ion analysis by inductively coupled plasma

(ICP) mass spectrometry of FlpA revealed the presence

of 1.32 atoms of Zn and 0.25 atoms of Cu per FlpA

monomer (Table 1). No other metals were present in

significant quantities. Thus, the metal ion content of FlpA

was similar to that of FLP from L. casei and as both

proteins have two cysteine residues, it was predicted that

FlpA would operate the same dithiol±disulphide based

redox switch used by FLP (Gostick et al., 1998, 1999).

However, in contrast to FLP, FlpA, even after prolonged

Fig. 1. SDS±PAGE analysis of overproduced FlpA. Lane 1,molecular weight markers (sizes in kDa are indicated); lane 2, crudeextract (20 mg); lane 3, GST±FlpA (10 mg); lane 4, FlpA (10 mg).

Table 1. Properties of purified FlpA.Property

Native FlpA Mr (gel filtration) 54 700Subunit FlpA Mr (SDS±PAGE) 27 700Subunit GST±FlpA Mr (SDS±PAGE) 55 800Metal ion content (atoms per FlpA monomer) Zn, 1.32; Cu, 0.25; Fe, , 0.006Reactive sulphydryl groups(SH per FlpA monomer)

1.4±1.7

Iron content after reconstitution(atoms per monomer of GST±FlpA)

3.8±4.05

Acid labile sulphide after reconstitution(atoms per monomer of GST±FlpA)

3.2±3.9

C. Scott, J. R. Guest and J. Green 1385


exposure to air in the absence of reducing agents, failed

to acquire an intramolecular disulphide bond as indicated

by a reactive sulphydryl content of 1.4±1.7 per monomer.

This observation was confirmed using FlpA expressed

from pGS1035, a pET16b derivative, encoding FlpA

without the extra N-terminal amino acids associated with

FlpA isolated from the GST±FlpA fusion protein. Accord-

ingly, further analysis showed that neither the addition of

hydrogen peroxide, nor the removal of metal ions (by

treatment with EDTA, 20 mM), nor the addition of CuII

ions could induce disulphide bond formation as indicated

by an increase in FlpA mobility on non-reducing

SDS±PAGE (not shown). This suggests that, despite their

similarity, FlpA and FLP do not sense their respective

stimuli by the same mechanism and that they may respond

to different environmental cues.

FlpA can acquire an oxygen-sensitive [4Fe 4S] cluster

The FLP of L. casei senses redox stress via the formation

of an intramolecular disulphide bond (Gostick et al., 1998)

and the FNR protein of E. coli senses oxygen via the

assembly disassembly of a [4Fe 4S] cluster that requires

four cysteine ligands (Jordan et al., 1997; Popescu et al.,

1998). Thus, the cysteine residues of these related

transcription factors are the key to their sensory capabil-

ities. As FlpA failed to respond to oxygen in an FLP-like

manner (see above) attempts were made to assemble an

iron±sulphur cluster in FlpA under anaerobic conditions

using the protocol developed for FNR (Green et al., 1996).

In preliminary experiments an iron±sulphur cluster was

assembled in FlpA as judged by a broad absorbance

maximum around 420 nM (not shown). However,

because the FlpA from the purification protocol was dilute

(and was further diluted during the removal of excess

reconstitution components) it was not possible to obtain

meaningful data using isolated FlpA and therefore the

GST±FlpA fusion protein had to be used for reconstitution

studies. The presence of GST is unlikely to influence the

assembly of an iron±sulphur cluster as it is separated

from FlpA by an unstructured 15 amino acid linker and this

approach has been successfully applied to the FNR

homologue HlyX (Green and Baldwin, 1997). Cluster

assembly in GST±FlpA was monitored by optical spectro-

scopy which revealed the formation of a single broad

absorbance band at around 420 nM (Fig. 2A). The

spectrum closely resembled that of FNR (Green et al.,

1996) and other [4Fe 4S] containing proteins and lacked

Fig. 2. Absorption spectra of GST±FlpA. Allspectra were obtained with anaerobicallyreconstituted samples of GST±FlpA insealed cuvettes.A. Assembly of the GST±FlpA (17 mMdimer) iron±sulphur cluster. Spectra werecollected at intervals for 3 h.B. Dithionite (0.1 mM) mediated reductionof the GST±FlpA (9 mM dimer) iron±sulphur cluster: bold line, before dithioniteaddition; thin line, after dithionite addition.C. Disassembly of the FlpA [4Fe 4S] clustermonitored by the decrease in absorbanceat 420 nm following exposure ofreconstituted GST±FlpA (9.5 mM dimer) toair.



the additional features usually associated with [2Fe 2S]

proteins. A useful index to assess the iron±sulphur

content of a protein is the A420:A280 ratio. The 280 nm

absorbance of GST±FlpA consists of contributions from

four tryptophan, 13 phenylalanine and 20 tyrosine

residues yielding a predicted 1280 of 55 100 M21 cm21.

The model iron±sulphur compound [Fe4S4(S±Et)4]2± has

an A420:A280 ratio of 0.7 and an 1420 of 17 200 M21 cm21

(Green et al., 1996) therefore the ratio for FlpA containing

one [4Fe 4S] cluster per monomer (two per dimer) is

predicted to be 0.25. The observed value of 0.16

(A420:A280, 0.315:1.96 after anaerobic gel filtration to

remove unincorporated iron and other low molecular

weight components of the reconstitution mix) is lower than

expected, indicating either the presence of fewer than two

clusters per dimer or that the non-cysteine ligands reduce

1420 of FlpA. In support of the latter contention is the

observation that the molar extinction coefficient of [2Fe

2S] chromophores with two non-cysteine ligands (Rieske

iron±sulphur proteins) is only two-thirds that of [2Fe 2S]

clusters with four cysteine ligands such as ferredoxins

(Fee et al., 1984).

Estimations of the iron and acid labile sulphide content

of GST±FlpA after removing excess reconstitution reac-

tion components by anaerobic gel filtration indicated the

presence of 3.8±4.05 iron atoms and 3.2±3.9 atoms of

acid labile sulphur per FlpA monomer (Table 1). Iron was

not detectable in the GST±FlpA fusion protein before

reconstitution. Thus, on the basis of these measurements

and the appearance of the optical spectrum FlpA appears

to contain one [4Fe 4S] cluster per monomer with at least

two non-cysteine ligands.

Compared with FNR, the [4Fe 4S] cluster of FlpA can

be readily reduced as evidenced by the reduction in

absorbance at 420 nM upon anaerobic addition of 0.1 mM

dithionite (Fig. 2B). At least 1 mM dithionite is required

to reduce FNR (Green et al., 1996). Unfortunately it was not

possible to re-oxidize the cluster because it was destroyed

by the addition of potassium ferricyanide (0.1 mM).

The [4Fe 4S] cluster of FNR is a sensitive monitor of

environmental oxygen and is rapidly degraded upon

exposure to air (Jordan et al., 1997). Exposure of

anaerobically reconstituted FlpA to air caused the

destruction of the iron±sulphur clusters as estimated by

the decrease in absorbance at 420 nM (Fig. 2C). The

absence of any absorbance bands at longer wavelengths

and the absence of residual 420 nM absorbance after

45 min of air exposure indicated that the [4Fe 4S] cluster

was not converted to a [2Fe 2S] but was completely

disassembled. Thus, the cluster is both oxygen and redox

(potassium ferricyanide) sensitive.

To investigate the role of the N-terminal region of FlpA

in the assembly of the [4Fe 4S] cluster, an expression

vector encoding a GST±FlpA variant (GST-FlpA 0) lacking

the first 11 amino acids of FlpA and containing the

substitutions D12M, H13D, H14L was created by PCR.

This protein thus lacked the three N-terminal histidine

residues and failed to incorporate an iron±sulphur cluster

under the conditions developed for the unaltered

GST±FlpA fusion. This indicated that the N-terminal

region of FlpA contains the non-cysteine [4Fe 4S] cluster

ligands. Furthermore when these ligands are removed

from FlpA, the protein (FlpA 0) acquired the ability to form

an intramolecular disulphide as judged by increased

mobility on non-reducing SDS±PAGE. Titration of oxid-

ized FlpA 0 with dithiothreitol revealed that the midpoint of

the engineered FlpA 0 switch was 2420 mV, compared

with 2400 mV for FLP (Gostick et al., 1998) and 2185 mV

for OxyR (Zheng et al., 1998). Therefore, it would appear

that the presence of an extended N-terminal region is

sufficient to impede the interaction between C15 and C112

in FlpA preventing the operation of an FLP-like switch

(Gostick et al., 1998).

FlpA recognizes an FNR site

Following the characterization of FLP from L. casei the

prediction that members of the CRP family containing a

E± ±SR motif within the DNA binding helix will recognize

an FNR site (TTGAT± ± ± ±ATCAA) was modified to

include the related FLP site (CCTGA± ± ± ±TCAGG)

(Gostick et al., 1998). The DNA-binding specificity of FlpA

was tested in vitro by band-shift analysis with a promoter

region containing an FNR site (yfiD, Fig. 3A) and a

promoter containing both FNR and FLP sites FFmelR

(Fig. 3B). The mobility of both promoter regions was

retarded by the presence of FlpA indicating that it is the

FNR site and not the FLP site that is recognized. This was

confirmed by the failure of FlpA to retard an isolated FLP

site (Fig. 3C). Furthermore, in band-shift assays with the

FF, CC, NNmelR family of promoters in which each

member is identical except for symmetrically related base

pair replacements that create FNR (FF), CRP (CC) or

neutral (NN) consensus sites, only FFmelR was retarded

Fig. 3. FlpA retards the mobility of DNA containing FNR sites inband-shift assays.A. yfiD DNA incubated with FlpA as indicated: lane 1, no FlpA; lane2, 2 mM; lane 3, 10 mM.B. FFmelR DNA: lane 1, no FlpA; lane 2, 2 mM; lane 3, 10 mM.C. FlpA incubated with a synthetic FLP binding site (test site A,Gostick et al., 1998). Lane 1, no FlpA; lane 2, 2 mM; lane 3, 10mM. Free DNA and the retarded DNA:FlpA complexes areindicated, each experiment was carried out at least twice.



by FlpA, providing compelling evidence for FlpA±FNR-

site interaction. Like FNR, FlpA bound at the yfiD

promoter with higher affinity than at the FFmelR promoter.

However, whereas active FNR binds to target DNA at nM

concentrations, retardation with FlpA was only observed

in the 2±10 mM range, perhaps indicating that only a small

fraction of the FlpA present in the reaction is active,

although mM concentrations of FLP were required to

observe binding at its cognate binding site and low affinity

binding may be a feature of Flp:DNA interactions. Indeed

the affinity of FlpA for the promoter of the L. lactis flpA

operon was similar to that observed for FFmelR,

suggesting that the DNA context of the FNR site has little

effect on FlpA binding.

Attempts were made to increase the affinity of FlpA for

DNA by the addition of various types of coeffector known

to modulate the activity of other members of the CRP

family. Metal ions, cyclic nucleotide monophosphates,

reducing agents (DTT added to the sample anaerobically)

and cell-free extracts from L. lactis MG1363 were added

to band-shift reactions, but all failed to significantly

improve the efficiency of DNA binding by FlpA, although

a small, but reproducible enhancement was evident upon

addition of cGMP (data not shown).

The DNA recognition helices of FNR and Flp proteins

contain the E± ±SR motif but there are two significant

differences (Fig. 4). Valine 208 in FNR is replaced by a P

residue at the equivalent position in FlpA/B, and G216 in

FNR is replaced by a positively charged residue in FlpA/B

(K in FLP and FlpB, and R in FlpA). Because FlpA could

bind at an FNR site it was predicted that FNR variants with

these substitutions may have a reduced affinity for, but

should still recognize, the FNR target sequence. There-

fore, the FNR DNA binding helix was modified to carry a

V208P substitution, a G216K substitution, or both. The

ability to activate transcription from an FNR-driven

promoter was tested by b-galactosidase assay. The

FNR-V208P variant was capable of activating expression

from the FNR-dependent FF-41.5pmelR promoter

although at reduced levels (4120 Miller units) compared

with FNR (6990 Miller units). Therefore, the presence of P

at the beginning of the DNA binding helix of the Flp

proteins is not sufficient to prevent, but does impair,

recognition of an FNR site consistent with the lower affinity

of FlpA for an FNR site compared with FNR itself.

FNR variants containing the G216K substitution (FNR-

G216K and FNR-V208P,G216K) proved to be insoluble

(as judged by Western blotting) and as such were

incapable of activating transcription (data not shown).

Therefore, based on the data presented here for both FlpA

and the FNR variants, it seems clear that FlpA recognizes

an FNR site.

Recognition of target DNA by FNR is dramatically

enhanced by acquisition of a [4Fe 4S] cluster

(Khoroshilova et al., 1995; Jordan et al., 1997). However,

while unreconstituted FlpA bound target DNA, albeit with

low affinity (see above) rather than increasing affinity for a

FNR site, incorporation of an iron±sulphur cluster into

FlpA abolished DNA binding (Fig. 5). However, DNA

binding was restored after destruction of the [4Fe 4S]

Fig. 4. The interactions of CRP and theproposed interactions of FNR, FLP, FlpAand FlpB proteins with their respective DNAtargets. The DNA recognition helices ofCRP, FNR and Flp proteins are shown inhelical wheel format numbered from thefirst residue of the helix.A. The core interactions between Glu atposition 2 and Arg at position 6 with theconserved C±G and T±A base pairs of theFNR and CRP sites are shown. Theproposed discriminatory interactionsbetween Ser at position 5 and the A±Tbase pair of the FNR site and the Arg atposition 1 and the C±G base pair of theCRP site are also indicated.B. Comparison of the amino acidsequences of the DNA binding helices ofthe Flp proteins. The Glu, Arg and Serresidues that are proposed to be involved inFNR site recognition are indicated (italics).The Pro residue at position 1 that ispresent in all the Flp proteins and a CRP-like positively charged residue (Lys or Arg)at position nine are boxed.



cluster by exposure to air. Thus, the incorporation/loss of

the iron±sulphur cluster constitutes an in vitro switch for

FlpA that operates in the opposite direction to that of FNR.

Flp proteins can activate gene expression in E. coli

The in vitro data obtained for FlpA suggested that it may

sense oxygen via an iron±sulphur cluster, active in the

absence of the cluster (aerobic) and inactive upon cluster

acquisition (anaerobic) via binding to an FNR site.

Therefore, a strain of E. coli lacking FNR, expressing

either flpA or flpB from a plasmid, and containing an

FNR-dependent lac reporter should allow Flp function to

be tested in vivo. Two ptac85 derivatives carrying flpA

(pGS1115) or flpB (pGS1149) were created to facilitate

IPTG-inducible expression of the Flp proteins in E. coli

strain JRG1728 (Dfnr) carrying lac reporter plasmids in

which the activating FNR site is positioned at 241.5 (Class

II promoter, pRW5/FF-41.5pmelR); 261.5 (Class I

promoter, pRW5/FF-61.5pmelR); or 271.5 (Class I,

pRW5/FF-71.5pmelR). The activity of b-galactosidase

estimated for both aerobic and anaerobic cultures expres-

sing either flpA, flpB or neither (empty ptac85 vector

control) were not significantly different when the FNR site

was positioned at 241.5 or at 271.5, indicating that the Flp

proteins were unable to activate transcription from these

promoters in E. coli (Table 2). However, a five- to

sevenfold increase in b-galactosidase activity was

observed when the Flp proteins were tested against the

FF-61.5pmelR promoter, indicating that FlpA can be active

in E. coli provided the FNR site is centred at 261.5.

Attempts to reproduce in vivo the oxygen-mediated

FlpA switch observed in vitro were unsuccessful. FlpA or

FlpB driven expression from the FF-61.5pmelR promoter

was similar for both aerobic and anaerobic cultures, nor

was it affected by the addition of hydrogen peroxide, or

cAMP, or the membrane permeable cGMP homologue

8-bromo-cGMP (data not shown).

Discussion

At the outset of this work it was expected that the Flp

proteins of L. lactis would respond to oxidative stress via

the formation of an intramolecular disulphide bond and

recognize an FLP site. However, the inability of FlpA to

form an intramolecular disulphide bond indicates that FlpA

cannot independently operate as an FLP-type redox

sensor (Gostick et al., 1998). It is possible that other

factors (perhaps the products of the other genes orfX and

orfY of the flpA operon) may be required for the oxidation

of FlpA, but it is clear that E. coli contains no analogue of

such a factor. The Zn binding chaperone Hsp33 responds

to oxidative stress via the formation of intramolecular

disulphide bonds (Jakob et al., 1999). The coordination of

Zn by the reactive cysteine sulphydryls is thought to

prevent disulphide bond formation under normal redox

conditions (Aslund and Beckwith, 1999). The Zn asso-

ciated with FlpA may play a similar role, although

treatment with EDTA did not facilitate the formation of a

FlpA disulphide.

In vitro assembly of an iron±sulphur cluster in FlpA

raises a number of questions because each FlpA

monomer has only two cysteine residues to act as ligands

for the cluster. Although the exact identity of the non-

cysteinyl ligands is unknown, it is likely that they are

located within the first 14 amino acids of FlpA. This region

Fig. 5. Cluster mediated switching of DNA binding by GST±FlpA. The effect of iron±sulphur cluster acquisition on the ability of GST±FlpA tobind at the yfiD promoter was assessed by band-shift analyses. Anaerobic incubations containing decreasing concentrations of GST±FlpA(lanes: 2, 6 and 10, 10 mM; 3, 7 and 11, 5 mM; 4, 8 and 12, 2 mM; 5, 9 and 13, 1 mM). The proteins used were: unreconstituted GST±FlpA(lanes 2±5); reconstituted GST±FlpA (lanes 6±9); and reconstituted GST±FlpA that had been exposed to air for 30 min (lanes 10±13). Lane 1contains yfiD DNA alone. The positions of the free DNA and GST±FlpA:DNA complexes are indicated.

Table 2. FlpA-dependent transcription activation in vivo.

b-Galactosidase activity (Miller units)

FF-71.5pmelR FF-61.5pmelR FF-41.5pmelR

Aerobic Anoxic Aerobic Anoxic Aerobic Anoxic

RegulatorFlpA 68.5 65.6 226.5 186.9 139.9 99.2

(9.1) (6.2) (13.9) (10.9) (3.8) (16.8)

FlpB 97.4 58.1 287.6 195.4 110.4 101.3(11.7) (3.5) (14.3) (9.2) (19.4) (10.2)

No regulator 72.0 61.2 41.2 23.3 127.8 97.1(3.0) (5.8) (2.4) (2.4) (14.7) (16.5)

b-Galactosidase activities (Miller units) expressed from plasmidscontaining model Class I (FF-61.5pmelR and FF-71.5pmelR) or ClassII (FF-41.5pmelR) promoters upstream of a promoterless lac operonin E. coli strain JRG1728(Dfnr) expressing either flpA, flpB, fnr,fnrV208P or no regulator. Aerobic and anoxic cultures were assayedduring the exponential phase of the growth cycle. The values quotedare from duplicate measurements of three independent cultures.Figures in parentheses are the standard deviations.



contains three histidine, three aspartate and a glutamate

residue, any of which may be involved in iron coordination

(Fig. 6). It is also possible that at least one ligand could be

contributed by an external thiol (Jung et al., 1996), such

as the dithiothreitol present in the reconstitution reaction

or from GST. However, if this were the case, it might be

expected that the GST±FlpA 0 protein would acquire a

cluster and that isolated FlpA would not and not vice

versa. The influence of the non-cysteinyl ligands on the

properties of the FlpA cluster is manifest in a relatively low

1420 compared with a conventional [4Fe 4S] cluster and

increased oxygen stability compared with FNR. The

enhanced stability of the [4Fe 4S] cluster of FNRL28H

(Khoroshilova et al., 1995) may be specifically due to the

presence of H28. These observations may indicate that

[4Fe 4S] clusters with non-cysteinyl ligands are more

stable than all C liganded clusters.

An iron±sulphur cluster can not be assembled in the

FLP of L. casei, which may be because FLP lacks all but

one of the N-terminal histidine residues of FlpA and so

lacks sufficient ligands to form the cluster. Accordingly

altering the N-terminal region of FlpA to make the protein

(FlpA 0) more FLP-like prevented anaerobic incorporation

of an iron±sulphur cluster but allowed the formation of an

intramolecular disulphide bond. Thus, this simple example

of protein engineering (deletion of 11 N-terminal amino

acids and substitution of D12, H13 and H14) converts

FlpA from a protein with an oxygen-labile FNR-like [4Fe

4S] cluster to one resembling FLP capable of operating a

disulphide±dithiol switch poised close to that of the FLP of

L. casei. Therefore, by gaining or losing N-terminal iron±

sulphur cluster ligands, different mechanisms can be

adopted by FlpA and FLP to sense changes in redox state

in vitro.

Although FlpA can acquire an iron±sulphur cluster in

vitro this does not necessarily reflect the situation in vivo,

especially when it is recalled that lactic acid bacteria do

not require iron to grow, although at least one example of

an iron±sulphur containing protein is known in L. lactis

(Rowland et al., 1997). Attempts to reproduce the in vitro

oxygen responsive FlpA switch in E. coli were unsuccess-

ful because although FlpA (and FlpB) could drive lac

expression from a Class I FNR-dependent promoter there

was no reduction in b-galactosidase activity in anaerobic

cultures. This indicates that the iron±sulphur cluster is not

assembled in E. coli, particularly as FlpA isolated from E.

coli is associated with Zn and Cu which may act to block

cluster assembly in vivo much as the Zn in Hsp33

prevents disulphide bond formation (Aslund and

Beckwith, 1999). Alternatively, the in vivo switch may not

operate by cluster assembly disassembly but by oxidation

reduction and that a suitable electron donor/acceptor may

not be present in E. coli. It is clear from previous work that

FlpA is involved in the response of L. lactis to oxidative

stress by hydrogen peroxide (Gostick et al., 1999) but

whether this stress is sensed by an FlpA [4Fe 4S] cluster is

not yet established. Once more these observations

suggest that the proteins encoded by the additional

genes of the L. lactis flp operons (orfX and orfY) may be

required for activating/deactivating FlpA in vivo and in

order to model the FlpA switch expression of the whole

operon may be necessary (Gostick et al., 1999).

A further constraint on the ability to observe an

FlpA-mediated switch in E. coli is the interaction between

the host RNA polymerase and the heterologous regulator.

The FlpA/B proteins were able to activate transcription

from a Class I but not a Class II promoter in E. coli.

However, in L. lactis expression of the flpB operon is

driven from a Class II promoter (Gostick et al., 1999). This

suggests that Flp proteins are adapted to the much smaller

s70 of L. lactis (GenBank accession no. X71493). This

suggestion is supported by the lack of conservation of the

FNR Activating Region 3 (AR3) determinant D86 (Ralph

et al., 1998) in Flp proteins. The CRP adopts a different

strategy to activate transcription from Class II promoters

by using AR2 and aNTD. Only one of the amino acids that

form the AR2 of CRP (Rhodius et al., 1997) is conserved in

FlpA and FlpB (none are conserved in FLP). Thus, the lack

of an effective AR2 or AR3 contact could account for the

lack of Flp activity at E. coli Class II promoters. A small

Fig. 6. Comparison of the N-terminalregions of the Flp proteins with othermembers of the CRP family and the FNR*protein FNRL28H. The sequences are:DNR (Pseudomonas aeruginosa); DnrD,DnrE and DnrS (Pseudomonas stutzeri);FlpA and FlpB (L. lactis); FLP(Lactobacillus casei); FnrL (Rhodobactersphaeroides); and FnrP (Paracoccusdenitrificans). Also included are: FNRL28H(an E. coli FNR variant, Kiley and Reznikof,1991); and the FlpA(protein generated inthis work). Potential metal ion ligands areboldfaced.



region (AR1) of CRP centred around H159 is required for

regulation of Class I promoters (Savery et al., 1996).

However, the FNR AR1is much larger extending along the

whole of one face of the molecule. The pattern of

transcription activation observed with FlpA and FlpB

suggests that they share the extensive AR1 of FNR

permitting expression from the Class I promoter FF-

61.5pmelR in E. coli.

That FlpA recognizes an FNR site was unexpected, as

FLP interacts with a palindrome different from, but related

to, that of the consensus FNR binding site (Gostick et al.,

1998). Altering the FNR DNA-binding helix (aF) so that it

more closely resembled that of the Flp regulators was not

sufficient to alter target specificity, suggesting that there

are amino acids outside the DNA binding helices that are

involved in determining site specificity and binding affinity.

The unusual DNA recognition properties of FLP from L.

casei may not be unique. Paracoccus denitrificans has

two FNR homologues, FnrP and NNR, with distinct and

non-overlapping roles in respiratory adaptation that are

predicted to recognize FNR sites on the basis of

sequence similarity (van Spanning et al., 1997). The

lack of cross-talk between NNR and FnrP could result

from a similar differentiation in binding specificity/affinity to

that evident between FNR, FLP and FlpA, and the

presence of a P residue at the start of the NNR DNA

binding helix may suggest that this residue is required for

altering DNA target recognition, although from the data

presented here it is not sufficient. Comparing the DNA-

binding specificities of FNR, FLP and FlpA clearly

Table 3. E. coli strains, phages and plasmids.

Strain, phage orplasmid Relevant characteristics Source or reference

JRG1728 Dlac x74 D(araA±leu) D(tyrR±fnr±rac±trg)17zdd-230::Tn9

Spiro and Guest (1987)

DH5a_ D(argF-lac)U169 (F80±lacZM15) recA Sambrook et al. (1989)BL21 F2 ompT rB

2 mB2 Novagen

JRG3506 DH5a(pGS1021) This workJRG3507 BL21 (pGS1021) This workJRG3900 BL21 (pGS1289) This workJRG3800 JRG1728 (pRW5/FF-41.5) (ptac85) This workJRG2844 JRG1728 (pRW5/FF-71.5) (ptac85) This workJRG2855 JRG1728 (pRW5/FF-61.5) (ptac85) This workJRG3402 JRG1728 (pRW5/FF-41.5) (pGS330) This workJRG4128 JRG1728 (pRW5/FF-41.5) (pGS1161) This workJRG4129 JRG1728 (pRW5/FF-41.5) (pGS1162) This workJRG4130 JRG1728 (pRW5/FF-41.5) (pGS1266) This workJRG3801 JRG1728 (pRW5/FF-41.5) (pGS1115) This workJRG3919 JRG1728 (pRW5/FF-71.5) (pGS1115) This workJRG3921 JRG1728 (pRW5/FF-61.5) (pGS1115) This workJRG3902 JRG1728 (pRW5/FF-41.5) (pGS1149) This workJRG3920 JRG1728 (pRW5/FF-71.5) (pGS1149) This workJRG3922 JRG1728 (pRW5/FF-61.5) (pGS1149) This workf115c l-zap carrying the flpA operon and flanking region Gostick et al. (1999)pGEX±KG GST-fusion expression vector, ApR Amersham-Pharmaciaptac85 Expression vector, ApR Marsh (1986)pGS330 ptac85 containing the fnr coding region Green et al. (1991)pGS652 pUC118 containing the FFmelR semisynthetic promoter, ApR Gostick et al. (1998)pGS1021 pGEX±KG derivative with the flpA coding region,

ApRThis work

pGS1035 pET16b derivative with the flpA coding region, ApR This workpGS1063 pUC118 containing the modified yfiD promoter with

a single FNR binding site (Y2), ApRGreen et al. (1998)

pGS1115pGS1149pGS1161

ptac85 derivative with the flpA coding region, ApR

ptac85 derivative with the flpB coding region, ApR

ptac85 containing the fnr coding region encoding theV208P FNR variant, ApR

This workThis workThis work

pGS1162 As pGS1161 but encoding the G216K variant FNR This workpGS1266 As pGS1161 but encoding FNR V208P G216K This workpGS1289 pGEX±KG derivative with an altered flpA encoding

FlpA 0that lacks 11-terminal amino acids and containsD12 M, H13D and H14 l substitutions, ApR

This work

pRW5/FF-41.5 FF-41.5pmelR±lac operon fusion in low copy. ColE1compatible, broad host range vector, with consensusFNR site centred at 241.5, TcR

Lodge et al. (1990)

pRW5/FF-61.5 As pRW5/FF-41.5, but with FNR site centred at 261.5, TcR Wing et al. (1995)pRW5/FF-71.5 As pRW5/FF-41.5, but with FNR site centred at 271.5, TcR Wing et al. (1995)



illustrates the pitfalls in attempting to predict DNA targets

on the basis of amino acid sequence.

The characterization of the lactococcal Flp regulators

described demonstrates the inherent flexibility of the basic

CRP structure to accommodate a variety of sensory

domains and how DNA recognition must be influenced by

amino acid residues outside the DNA binding helix. The

ability of FlpA to acquire an oxygen-labile iron±sulphur

cluster despite retaining only two of the four essential

cysteine residues of typical FNR proteins indicates that

the FNR coordination pattern is not the only competent

one. FNR-like proteins such as DNR (Pseudomonas

aeruginosa) FnrP (Paracoccus denitrificans) and FnrL

(Rhodobacter sphaeroides) that have histidine rich

N-terminal regions but either lack or have altered

arrangements of cysteine residues may use histidine

ligands to coordinate iron±sulphur clusters with altered

redox properties and switch points (Fig. 6).

Thus, while many questions remain concerning the

physiological role of Flp regulators in lactic acid bacteria

the in vitro characterization of FlpA has revealed a

number intriguing features that merit further investigation

to provide new insights into both transcription regulation in

lactic acid bacteria and the CRP family of transcription

factors.

Experimental procedures

Bacterial strains, plasmids and microbiological methods

The bacterial strains, phages and plasmids used in this workare summarized in Table 3. E. coli strains were grown in Lmedium (Lennox, 1955) at 378C. Media were supplementedwith ampicillin (20 mg ml21), tetracycline (35 mg ml21) andisopropyl-b-D-thiogalactoside (IPTG, 30 mg ml21) as appro-priate. Aerobic cultures were grown either in 100 ml, 250 mlor 2000 ml flasks and shaken at 250 r.p.m. Anaerobiccultures were grown in anaerobic gas jars (Oxoid) andsupplemented with glucose (0.2%). Transcriptional activationwas analysed in vivo by estimating b-galactosidase produc-tion by the method of Miller (1972).

Standard methods for the manipulation of DNA werefollowed (Sambrook et al., 1989). The flpA gene wasamplified and isolated as a 750 bp product from the flpAoperon of L. lactis by PCR using the l-phage F115C as thetemplate and primers S540 (GATTGGATCCCCATGGAGAT-TAAAGATTTTGATGAGCATTTAAGTG) and S541 (TTTCTGCAGTCGACTAATTTCTCCCAATCCCCCAGTTTAC) containingunique NcoI and SalI restriction sites (underlined) to facilitatecloning into the expression vector pGEX-KG. The resultantplasmid (pGS1021) was transformed into DH5a(JRG3506) andthen transferred into BL21 (JRG3507). The latter strain wasused for expression because of its protease deficientphenotype (Table 3).

An FlpA variant (FlpA 0) that lacks 11 N-terminal aminoacids and contains the substitutions D12M, H13D and H14L(numbering as for FlpA) and thus retains both cysteine

residues but lacks the three N-terminal histidine residues ofFlpA was encoded by pGS1289. This plasmid was created bya two-step PCR using the following mutagenic primers(mismatches in lower case): S573, CATTAAAccatCCAC-CATTGTATCCAG; and S574, CATTAAACCATggACCtTTG-TATCCAG.

The ptac85 derivatives used in the investigation of FlpAand FlpB function in vivo were constructed by subcloning theNcoI/SalI fragment from pGS1021 to create the flpAexpression plasmid, pGS1115, and by cloning a PCR productcontaining the flpB coding sequence amplified from L. lactisssp. cremoris MG1363 genomic DNA using the primers S562(GAGAATCCCCACCATGGGTAG) and S563 (GGTTTTAGCTGATGTCGACAT) containing unique NcoI and SalI sites(underlined) to create pGS1149 (Table 3).

The singly mutated fnr variants were made by site-directedmutagenesis using oligonucleotides directing the desiredcodon changes according to the AlteredSites protocol(Promega) followed by subcloning into ptac85. The doubleFNR variant, V208P, G216K (pGS1266) was created by PCRand subcloning into ptac85 (Table 3). Automated DNAsequencing was used to confirm the desired codon altera-tions had been made.

GST±FlpA overexpression and purification

FlpA was amplified as a GST±FlpA fusion in E. coliJRG3507. Aerobic cultures (500 ml) were grown at 378C toan A600 0.2±0.3. At this point expression was induced by theaddition of IPTG (30 mg ml21) and incubation was continueduntil the cultures reached A600 1 ^ 0.1. The bacteria werethen collected by centrifugation and used immediately orstored at 2208C.

Clarified cell-free extracts were produced by resuspendingthe bacteria in 10 mM Tris±HCl (pH 8.0) containing 10 mMNaCl (10 ml l21 of original culture), lysis by two passagesthrough French pressure cell, followed by centrifugation. TheGST±FlpA fusion protein was adsorbed onto a column(1 ml l21 of culture) of GSH-Sepharose (Amersham-Pharma-cia) equilibrated with the resuspension buffer. The FlpA wasthen eluted either as a GST fusion, by washing withresuspension buffer containing 10 mM glutathione (pH 8´0),or as FlpA by incubation at 258C with thrombin (25 units) andsubsequent elution in resuspension buffer. The FlpA variantFlpA 0 was isolated by essentially the same procedure exceptthat induction by IPTG (200 mg ml21) was carried out at 258C.

Protein analysis

Protein concentration was estimated using the Bio-Rad proteinassay, using bovine serum albumin as standard. Proteinpurity was assessed by SDS±PAGE (Laemmli, 1970) andstaining with Coomassie brilliant blue. The oligomeric stateof FlpA was determined by gel filtration, using a calibrated(protein standards: BSA, 67 000; ovalbumin, 43 000; chymo-trypsinogen, 25 000; and ribonuclease A, 13 700) SephacrylS-200 column equilibrated with 10 mM Tris±HCl, 10 mMNaCl pH 8.0. Metal ion content was estimated by ICP-massspectroscopy of FlpA that had been samples dialysed



against10 mM EDTA to remove adventitious metals. Thedialysis buffer was used as a blank.

Purified FlpA was tested for the presence of a disulphidebond by SDS±PAGE fractionation under non-reducing,reducing and oxidizing conditions (Green and Guest, 1993).Free thiol content was determined by sulphydryl grouptitration as described Thelander (1973).

Western blotting with anti-FNR serum (Spiro and Guest,1987) was used to assess the cellular location and content ofthe FNR variants with substitutions in their DNA-bindinghelices.

Iron±sulphur centre reconstitution and analysis

An iron±sulphur cluster was assembled in GST±FlpA(. 1 mg ml21) under anaerobic reducing conditions asdescribed for FNR (Green et al., 1996) in stoppered 1 mlmatched quartz curettes, and cluster formation was mon-itored by optical spectroscopy using a Unicam UV4 UV/VISspectrometer. After reconstitution, unincorporated iron,cysteine and DTT were separated from FlpA by anaerobicchromatography on Sephadex G25.

Iron content and acid labile sulphur were analysed asdescribed (Beinert, 1983; Woodland and Dalton, 1984).Stability to oxygen was analysed by exposing desalted FlpAto air and following the changes in absorbance over a periodof 45 min. All anaerobic manipulations were carried out in ananaerobic workstation (Don Whitley Scientific Mk3).

DNA binding and site recognition

DNA binding in vitro was tested by band-shift analysis on 6%non-denaturing TBE-buffered PAGE, using radiolabelledyfiD, FF, CC, NNmelR, flpA operon promoter DNA or FLPtest site A DNA (Gostick et al., 1998). Co-incubation of FlpAprotein and probe was permitted for < 1 min before loading.All band-shift experiments were carried out at least twice.

Acknowledgements

We would like to thank: Dr A. J. G. Moir (Sheffield) for DNA and

amino acid sequencing; Mr A. Cox (Sheffield) for ICP mass

spectroscopy. This work has been supported by postgraduate

studentship (CS) and Advanced Fellowship (JG) awards from the

BBSRC.

References

Aslund, F., and Beckwith, J. (1999) Bridge over troubled waters:sensing stress by disulfide bond formation. Cell 96: 751±753.

Beinert, H. (1983) Semi-micro methods for analysis of labilesulfide and for labile sulfide plus sulfane sulfur in unusually

stable iron-sulfur proteins. Anal Biochem 131: 373±378.

Fee, J.A., Findling, K.L., Yoshida, T., Hille, R., Tarr, G.E.,

Hearshen, D.O., et al. (1984) Purification and characterization

of the Rieske iron-sulfur protein from Thermus thermophilus. JBiol Chem 259: 124±133.

Gasson, M.J. (1993) Progress and potential in the biotechnologyof lactic acid bacteria. FEMS Microbiol Rev 12: 3±20.

Gostick, D.O., Green, J., Irvine, A.S., Gasson, M.J., and Guest,

J.R. (1998) A novel regulatory switch mediated by the FNR-likeprotein of Lactobacillus casei. Microbiology 144: 705±717.

Gostick, D.O., Griffin, H.G., Shearman, C.A., Scott, C., Green, J.,

Gasson, M.J., et al. (1999) Two operons that encode FNR-like

proteins in Lactococcus lactis. Mol Microbiol 31: 1523±1535.

Green, J., and Baldwin, M.L. (1997) HlyX, the FNR homologue ofActinobacillus pleuropneumoniae, is a [4Fe 4S]-containing

oxygen-responsive transcription regulator that anaerobically

activates FNR-dependent Class I promoters via an enhanced

AR1-contact. Mol Microbiol 24: 593±605.

Green, J., Baldwin, M.L., and Richardson, J. (1998) Down-

regulation of Escherichia coli yfiD expression by FNR occupy-

ing a site at -93.5 involves the AR1-containing face of FNR. Mol

Microbiol 29: 1113±1123.

Green, J., Bennett, B., Jordan, P., Ralph, E.T., Thompson, A.J.,and Guest, J.R. (1996) Reconstitution of the [4Fe-4S] cluster in

FNR and demonstration of the aerobic-anaerobic switch in

vitro. Biochem J 316: 887±892.

Green, J., and Guest, J.R. (1993) Properties of FNR proteinssubstituted at each of the five cysteine residues. Mol Microbiol

8: 61±68.

Green, J., Trageser, M., Six, S., Unden, G., and Guest, J.R.

(1991) Characterisation of the FNR protein of Escherichia coli,an iron binding transcription regulator. Proc R Soc Lond Ser B

244: 137±144.

Irvine, A.S., and Guest, J.R. (1993) Lactobacillus casei contains

a member of the CRP-FNR family. Nucleic Acid Res 21: 753.

Jakob, U., Muse, W., Eser, M., and Bardwell, J.C.A. (1999)Chaperone activity with a redox switch. Cell 96: 341±352.

Jordan, P.A., Thompson, A.J., Ralph, E.T., Guest, J.R., and

Green, J. (1997) FNR is a direct oxygen sensor having a

biphasic response curve. FEBS Lett 416: 349±352.

Jung, Y.S., Vassiliev, I.R., Qiao, F., Yang, F., Bryant, D.A., andGoldbeck, J.H. (1996) Modified ligands to FA and FB in

photosystem I. J Biol Chem 271: 31135±31144.

Khoroshilova, N., Beinert, H., and Kiley, P.J. (1995) Association

of a polynuclear iron-sulfur center with a mutant FNR proteinenhances DNA-binding. Proc Natl Acad Sci USA 92: 2499±

2505.

Kiley, P.J., and Reznikof, W.S. (1991) Fnr mutants that activate

gene expression in the presence of oxygen. J Bacteriol 173:

16±22.

Laemmli, U.K. (1970) Cleavage of structural proteins during the

assembly of the head of bacteriophage T4. Nature 277: 680±

685.

Lennox, E.S. (1955) Transduction of linked genetic characters of

host by bacteriophage P1. Virology 1: 190±206.

Lodge, J., Williams, R., Bell, A., Chan, B., and Busby, S. (1990)

Comparison of promoter activities in Escherichia coli and

Pseudomonas aeruginosa-use of a new broad range promoter

probe plasmid. FEMS Microbiol Lett 67: 221±225.

Marsh, P. (1986) ptac85, an Escherichia coli vector for expres-sion of non-fusion proteins. Nucleic Acids Res 14: 3603.

Miller, J.H. (1972) Experiments in Molecular Genetics. Cold

Spring Harbor, NY: Cold Spring, Harbor Laboratory Press.

Popescu, C.V., Bates, D.M., Beinert, H., Munck, E., and Kiley,P.J. (1998) Mossbauer spectroscopy as a tool for the study of

activation/inactivation of the transcription regulator FNR in

whole cells of Escherichia coli. Proc Natl Acad Sci USA 95:

13431±13435.

Ralph, E.T., Guest, J.R., and Green, J. (1998) Altering theanaerobic transcription factor FNR confers a hemolytic

phenotype on Escherichia coli K12. Proc Natl Acad Sci USA

95: 10449±10452.



Rhodius, V.A., West, D.M., Webster, C.L., Busby, S.J.W., andSavery, N.J. (1997) Transcription activation at Class II CRP-

dependent promoters: The role of different activating regions.

Nucleic Acids Res 25: 326±332.

Rowland, P., Nielsen, F.S., Jensen, K.F., and Larsen, S. (1997)

Crystallization and preliminary X-ray analysis of the hetero-

dimeric dihydrooratase dehydrogenase B of Lactococcus

lactis, a flavoprotein enzyme system consisting of two PyrDBsubunits and two iron-sulphur cluster containing PyrK subunits.

Acta Cryst 53: 802±804.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989) Molecular

Cloning: a Laboratory Manual, 2nd Edn. Cold Spring Harbor,NY: Cold Spring, Harbor Laboratory Press.

Savery, N., Rhodius, V., and Busby, S. (1996) Protein±protein

interactions during transcription activation: the case of theEscherichia coli cyclic AMP receptor protein. Phil Trans R Soc

Lond B 351: 543±550.

van Spanning, R.J.M., De Boer, A.P.N., Reijnders, W.N.M.,Westerhoff, H.V., Stouthamer, A.H., and Van Der Oost, J.

(1997) FnrP and NNR of Paracoccus denitrificans are bothmembers of the FNR family of transcriptional activators but

have distinct roles in respiratory adaptation in response to

oxygen limitation. Mol Microbiol 23: 893±907.

Spiro, S., and Guest, J.R. (1987) Regulation and over-expressionof the fnr gene of Escherichia coli. J Gen Microbiol 133: 3279±

3288.

Thelander, L. (1973) Physicochemical characterisation of ribo-nucleoside diphosphate reductase from Escherichia coli. J Biol

Chem 248: 4591±4601.

Wing, H.J., Williams, S.M., and Busby, S.J.W. (1995) Spacing

requirements for transcription activation by Escherichia coliFNR protein. J Bacteriol 177: 6704±6710.

Woodland, M.P., and Dalton, H. (1984) Purification and proper-

ties of component A of the methane monooxygenase of

Methylococcus capsulatus (Bath). J Biol Chem 259: 53±59.Zheng, M., Aslund, F., and Storz, G. (1998) Activation of the

OxyR transcription factor by reversible disulfide bond forma-

tion. Science 279: 1718±1721.

Documents

Characterization of the Lactococcus lactis transcription factor FlpA and demonstration of an in vitro switch