Báo cáo khoa học: Structure of FocB – a member of a family of transcription factors regulating fimbrial adhesin expression in uropathogenic Escherichia coli pdf

Amino acids known to be important for DNA-binding are exposed on the protein surface, although they are not located in the recognition helix of the HTH motif.. The results obtained from

factors regulating fimbrial adhesin expression in

uropathogenic Escherichia coli

Ulrika W Hultdin1, Stina Lindberg2, Christin Grundstro¨m1, Shenghua Huang1, Bernt Eric Uhlin2 and A Elisabeth Sauer-Eriksson1

1 Department of Chemistry, Umea˚ University, Sweden

2 Department of Molecular Biology, The Laboratory for Molecular Infection Medicine Sweden (MIMS), Umea˚ University, Sweden

Introduction

Fimbriae, which are long, adhesive, polymeric protein

structures, are expressed on the surface of many

patho-genic bacteria Uropathopatho-genic Escherichia coli (UPEC),

the primary cause of urinary tract infections, express

various types of fimbriae for adhering to (and

some-times invading) host cells [1,2] Because the interaction between the fimbrial adhesins and their receptors is specific, the expression of different types of fimbriae makes infection progress in different niches of the urinary tract

Keywords

fimbriae; FocB; repressor protein;

uropathogenic Escherichia coli; X-ray

crystallography

Correspondence

A E Sauer-Eriksson or B E Uhlin,

Department of Chemistry, Umea˚ University,

SE-90187 Umea˚, Sweden; Department of

Molecular Biology, Laboratory for Molecular

Infection Medicine Sweden (MIMS), Umea˚

University, SE-901 87 Umea˚, Sweden

Fax: +4690 7865944; +4690 772630

Tel: +4690 7865923; +4690 7856731

E-mail: elisabeth.sauer-eriksson@

chem.umu.se;

bernt.eric.uhlin@molbiol.umu.se

Database

The atomic coordinates and structure

factors for the Escherichia coli FocB

protein are available in the Protein Data

Bank database under the accession number

3M8J

(Received 28 January 2010, revised 5 May

2010, accepted 17 June 2010)

doi:10.1111/j.1742-4658.2010.07742.x

In uropathogenic Escherichia coli, UPEC, different types of fimbriae are expressed to mediate interactions with host tissue FocB belongs to the PapB family of transcription factors involved in the regulation of fimbriae gene clusters Recent findings suggest that members from this family of proteins may form homomeric or heteromeric complexes and exert both positive and negative effects on the transcription of fimbriae genes To elucidate the detailed function of FocB, we have determined its crystal structure at 1.4 A˚ resolution FocB is an all a-helical protein with a helix-turn-helix motif Interestingly, conserved residues important for DNA-binding are located not in the postulated recognition helix of the motif, but in the preceding helix Results from protein–DNA-binding stud-ies suggest that FocB interacts with the minor groove of its cognate DNA target, which is indicative of a DNA interaction that is unusual for this motif FocB crystallizes in the form of dimers Packing interactions in the crystals give two plausible dimerization interfaces Conserved residues, known to be important for protein oligomerization, are present at both interfaces, suggesting that both sites could play a role in a functional FocB protein

Structured digital abstract

l MINT-7901626 : focB (uniprotkb: Q93K76 ) and focB (uniprotkb: Q93K76 ) bind ( MI:0407 ) by x-ray crystallography ( MI:0114 )

Abbreviations

CRP, cAMP receptor protein; HNF, hepatocyte nuclear factor; HTH, helix-turn-helix; PDB, Protein Data Bank; UPEC, uropathogenic

Escherichia coli.

To balance metabolic efficiency, the expression of

the different fimbrial types needs to be regulated The

control mechanisms are complex and act at multiple

levels Environmental factors may affect the on–off

switching of the fimbrial genes indirectly by influencing

the expression patterns of certain proteins In addition,

there is cross-regulation between the different fimbrial

gene clusters [3–5]

Many types of fimbriae expressed by UPEC strains

are structurally similar and their genetic organizations

show high resemblance The P fimbriae, commonly

associated with pyelonephritis [6], are the most studied

and have been described extensively The pap gene

cluster, sufficient for the production of P fimbriae,

consists of nine structural and two regulatory genes

The structural genes for P fimbriae encode both minor

and major subunits, building up the fimbrial rod, as

well as proteins important for the transport and

assem-bly of the fimbriae at the bacterial surface (Fig 1)

Recent studies have revealed additional genes in the

promoter distal region of the main fimbrial operons,

and their role also appears to be at the regulatory level [7,8]

The pap genes are transcribed from two divergent promoters, PB and PI, separated by an intercistronic region containing binding sites for the PapB protein, which is a key factor in the regulation of fimbriae [9,10] Other important regulators that operate in this region include cAMP receptor protein (CRP), Lrp, PapI, DNA adenine methylase and histone-like nucle-oid-structuring protein [11]

UPEC commonly express F1C fimbriae, a fimbrial type often found in strains also carrying genes for type 1

or P-fimbrial expression [12] The receptors for F1C fimbriae have been identified as glycosphingolipids [13,14] and their specific interaction allows the F1C fimbriae to adhere to the collecting ducts and distal tubules of the kidney [15]

The organization of the genetic determinant for F1C fimbriae, the foc operon, is very similar to that of the pap operon, with only a few transposed genes The foc gene cluster includes seven structural and two

pro-pap

J96

Usher Chaperone Minor subunits Adhesin

Regulation (?) Regulators Major

prs

subunit

prf

sfa

foc

H NS

CRP Lrp

H-NS

focB

+ –

Site 2 Site 1

B

Fig 1 Genetic organization of different fimbrial gene clusters: pap (P fimbriae), prs (Prs fimbriae), prf (P-related fimbriae), sfa (S fimbriae) and foc (F1C fimbriae) A representative E coli strain carrying the gene cluster is indicated The transcription factor FocB is acting in the int-ercistronic region between the focI and focB genes, together with a number of other regulatory proteins FocB has also been described to have a repressive effect on fim expression (encoding type 1 fimbriae) [5] Boxes with the same color represent genes with similar function Recent findings suggest that fimbrial gene clusters also include the previously unrecognized novel regulatory genes, X and Y [8] Modified from Sjo¨stro¨m et al [8].

moter-proximal regulatory genes [16,17] (Fig 1) One

of these regulatory genes encodes the FocB protein, a

member of the PapB family of fimbrial transcriptional

factors

The members of the PapB family identified so far

exist in both the E coli and Salmonella species and are

all involved in the regulation of fimbrial expression At

the amino acid level, FocB is 81% identical to PapB

and 100% identical to SfaB, the protein regulating the

expression of S-fimbriae [17], both of which are

associ-ated with E coli newborn meningitis strains [18] FocB

is also 47% identical to the PefB protein of

Salmo-nella typhimurium [19], and 34% and 28% identical to

the FaeB and FanB proteins, respectively The latter

two proteins are regulators of the K88 and K99

fimbri-al types that are expressed by enterotoxigenic E coli

and cause diarrhea in domestic animals In many

respects, the roles of the FaeB and FanB proteins in

transcriptional regulation are similar to that of PapB

in the regulation of the pap operon [20,21]

UPEC strains often carry several different fimbrial

operons within their genome [12] For example, the

UPEC strain J96 carries at least the operons fim, pap,

prs and foc [22–24] However, less than 10% of the

bacteria display more than one fimbrial type on their

surfaces simultaneously Thus, fimbrial expression

involves cross-regulatory interactions between the

dif-ferent operons Recent experiments suggest that an

intricate hierarchy exists with respect to the the

cross-regulation of the pap and foc operons because FocB

could stimulate the expression of pap, whereas PapB is

insufficient for stimulating the expression of foc by

itself [5] In concordance, the PapB family members

share a common core structure, as revealed by multiple

sequence alignments These homologous proteins share

almost completely conserved regions that are known to

be important for oligomerization and DNA-binding

[25]

In the present study, we describe the structure of

FocB, which is the first reported structure of a member

of the PapB protein family FocB crystallizes as

dimers, in which each subunit contains one

helix-turn-helix (HTH) motif Crystal packing interactions

suggest that there are two dimerization interfaces of

interest for the in vivo function of FocB Amino acids

known to be important for DNA-binding are exposed

on the protein surface, although they are not located

in the recognition helix of the HTH motif The results

obtained from DNA-shift assays suggest that FocB

binds to the DNA minor groove, thus indicating a

DNA-binding pattern different from that of classical

HTH motifs

Results

Structure determination The E coli FocB structure (109 amino acid residues) was determined at a resolution of 1.4 A˚ by multiwave-length anomalous diffraction [26] from a single crystal

of the selenomethionine labeled protein The crystal comprised two molecules per asymmetric unit Apart from several residues at the N- and C-terminal ends, all protein residues could be modeled into the electron density The final model contains residues 10–99 of chain A and residues 10–97 of chain B The final R-values, Rwork= 0.196 and Rfree= 0.222, are higher than expected for this resolution This is probably a result of the additional electron density ascribed to the N- and C-terminal residues of the protein, which were not modeled in the structure because of disorder Table 1 summarizes the statistics of X-ray data collec-tion and the results for the structural refinement of FocB The coordinates and structure factors are depos-ited in the Protein Data Bank (PDB) (accession code 3M8J)

The structure of FocB

As anticipated from CD measurements and secondary structure prediction software [27] (Fig 2), the amino acid chain of FocB forms an all a-helical structure that comprises five a-helices: a1 (Asp12-Ser21), a2 (30Glu-Ser40), a3 (Asp45-Gly58), a4 (Arg61-Tyr68), and a5 (Asn71-Tyr95) (Fig 3) Of these, helices a4 and a5 and the connecting turn (Gln69-Asn71) comprise an HTH motif, which appears to be quite different from the canonical HTH motifs found in many bacterial transcription factors [30–32] In particular, the second helix of the HTH motif, the putative recognition helix,

is much longer in the FocB structure than in a typical HTH motif Structural similarities with the helix-loop-helix motif in homeodomains also strengthen the view that FocB harbors an atypical HTH motif The helix a5, at the center of the domain, can be divided into two parts based on its amino acid compo-sition The N-terminal part, comprising residues Asn72

to Asn86, is amphiphatic, whereas the C-terminal part, starting at residue Val87 and extending to the last mod-eled residue Tyr95, is predominantly hydrophobic in nature, with the exception of one residue, Arg91 The hydrophobic surface of the amphiphatic N-terminal part of the helix is shielded from the solvent by three short anti-parallel helices a2–a4 situated perpendicular

to a5 The interactions between these four helices form

Table 1 Data collection and refinement statistics.

Range of resolution (A ˚ ) 47.40–1.40 ⁄ (1.47–1.40) 47.77–1.90 ⁄ (2.00–1.90) 47.80–2.00 ⁄ (2.11–2.00) 47.77–2.05 ⁄ (2.16–2.05)

R mergea 0.036 ⁄ (0.314) 0.058 ⁄ (0.381) 0.055 ⁄ 0.335) 0.056 ⁄ (0.428)

Refinement

Water molecules (A˚2 ) 28.1

Ramachandran plot favoured (%) 100

Ramachandran plot accepted (%) 0

Ramachandran plot outliers (%) 0

a For replicate reflections, R = RIhi) <I h >| ⁄ R<I h >; Ihi= intensity measured for reflection h in data set i, <Ih> = average intensity for reflection

h calculated from replicate data b R-factor = R||F o | ) |F c || ⁄ R|F o |; F o and F c are the observed and calculated structure factors, respectively.

c Rfreeis based upon 5% of the data randomly culled and not used in the refinement.

Fig 2 Sequence alignment ( BLAST ) [28] of FocB and PapB from E coli The sequence identity is 81% Residues not identical are highlighted

in yellow Secondary structural elements from the current structure of FocB are shown in bold with the HTH motif highlighted in red The first and last residue visible in the electron density of the FocB structure is highlighted in blue Residues Arg61 and Cys65, which were previ-ously found to be important for DNA-binding, are boxed in pink, whereas some residues previprevi-ously found to be important for oligomerization are boxed in green [25] The secondary structure elements predicted with JPRED3 [29] are shown in blue (H, helices; E, b-strands) Interest-ingly, the a1 helix in FocB was predicted to be a b-strand in PapB.

a stable hydrophobic core The first helix a1 is posi-tioned perpendicular to a5, but situated on the oppo-site side of helix a5 with respect to helices a2–a4 Only

a few contacts are formed between a1 and a5, and include one hydrogen bond between the main chain nitrogen atom of Leu23 and the side chain Oc1 atom

of Thr83, and one hydrophobic interaction between the side chain of Leu18 and the Cc2 atom of Thr83

FocB dimerization Cross-linking and size exclusion chromatography stud-ies showed that FocB forms predominantly dimers in solution [27] Furthermore, the packing of molecules in the crystal structure suggests that FocB is dimeric The asymmetric unit comprises two molecules of FocB Crystal packing contacts provide two alternative possi-bilities for homodimeric FocB interactions At the first dimer interface, the two monomers, chains A and B in the asymmetric unit, form extensive contacts between their helices a2 and a5 (Fig 4A, subunits colored in light and dark blue) These helices are positioned per-pendicular with respect to each other in a four-helix arrangement The two monomers are related by a non-crystallographic two-fold symmetry operation, and we refer to this interface as dimer Interface-I The

inter-Fig 3 Ribbon representation of the structure of FocB Helices

a1–a5 are displayed, with the HTH motif highlighted in blue The

conserved DNA-binding residues Arg61 and Cys65 are shown as

ball and sticks.

A

Fig 4 Crystal packing of the FocB struc-ture (A) Subunits forming Interface-I are colored in light and dark blue; subunits of Interface-II are colored in light and dark coral (B) Interactions in Interface-I (C) Interactions in Interface-II For clarity, only selected residues are shown as sticks Selected hydrogen bonds are shown in green (dotted line).

face involves predominantly residues positioned at the

hydrophobic C-terminal part of a5 An extensive

hydrophobic core is formed over this interface,

includ-ing four residues from a5 (AB-Leu85, AB-Leu88,

AB-Val89 and AB-Leu92) and two residues from a2

(AB-Leu35 and AB-Ile39) (Fig 4B) In addition, the

side chains of AB-Tyr95 and AB-Gln32 stack and

contribute to the hydrophobic core Side chains of

seven polar or charged residues positioned on helices

a2 and a5 are involved in hydrogen bond formation

over Interface-I (Table 2)

The crystal packing interactions revealed a second

putative dimer interface that we call Interface-II

(Fig 4A, subunits colored in light and dark coral) This

interface is also formed by a four-helix arrangement

comprising symmetry-related a1 and a5 helices

How-ever, at this interface, it is the N-terminal part of a5

that is involved The polar sides of the two

symmetry-related N-terminal parts of helix a5 are facing each

other, and hydrogen bonds and two salt bridges are

formed across the dimer interface (Table 2)

Hydropho-bic contacts also exist between symmetry-related

resi-dues AB¢-Phe14, AB¢-Leu15 and AB¢-Leu18, where B’

is a crystallographic symmetry-related copy of subunit

B (symmetry transformation: )x, y + 1 ⁄ 2, )z + 1 ⁄ 2)

(Fig 4C) In this dimer, the hydrophobic residues

bur-ied in the core of Interface-I are surface exposed

pisais an interactive tool that can be used for

explo-ration of protein interfaces [33] pisa analysis of the

two interfaces identified in FocB crystals suggested

that both dimer alternatives are stable in solution The

solvation energy effects, DiG, of Interface-I and II were calculated to )16.1 kcalÆmol)1 and )8.1 kcalÆmol)1, respectively Furthermore, DGdiss, which indicates the free energy of assembly dissociation, was calculated to 8.3 kcalÆmol)1 for Interface-I, and 1.5 kcalÆmol)1 for Interface-II Contact areas of Interface-I and -II were estimated to 907 A˚2 and 851 A˚2, respectively Com-bined, the output data from pisa suggest that dimer Interface-I is significantly more stable than dimer Interface-II

FocB and PapB bind in the minor groove of double-stranded DNA

In a previous study, Xia et al [34] obtained evidence that PapB binds in the minor groove of DNA This minor groove-binding property makes the protein unusual in the perspective of transcriptional activa-tors To assess how the FocB protein interacts with DNA, we performed a electrophoretic DNA gel mobility shift test as a competition assay between FocB and distamycin, a minor groove-binding drug [35] and methyl green, a major groove-binding drug [36] That distamycin could bind to DNA under the conditions used was evident from the mobility shift observed when ‡ 4 lm of distamycin was added to the DNA in the absence of protein (Fig 5A, lane 4) At lower concentrations, partial occupancy of distamycin did not shift DNA (lanes 2 and 3) The DNA mobil-ity shift caused by 125 nm of FocB (lane 5) was com-pletely abolished when distamycin was present at concentration of 4 lm or higher (lanes 9–10) Addition

of 1, 2, or 3 lm distamycin resulted in a gradually reduced amount of FocB bound to the DNA (lanes 6–8) Combined, the results suggest that distamycin competes with FocB for DNA-binding, and that the FocB-binding site on the DNA was increasingly occu-pied by distamycin at the tested concentrations When challenging the FocB–DNA complex formation with methyl green at concentrations up to 100 lm, we observed no apparent inhibition of the DNA-binding, suggesting that FocB does not interact with the major groove (Fig 5B) When methyl green was added at concentrations up to 100 lm in the absence of the protein, we did not observe any shift in mobility of the DNA but, at a very high concentration (1 mm), the DNA did not enter the gel (data not shown) That methyl green could bind to DNA under the conditions used was evident from DNA-binding tests with CRP (i.e the cAMP receptor protein) CRP is a well char-acterized, major groove-interacting, DNA-binding pro-tein, used here as a control The addition of 100 lm methyl green abolished all of the CRP binding to

Table 2 Hydrogen bonds formed at the two plausible dimer

inter-faces of FocB.

Interface-I

Interface-II

DNA (data not shown) Taken together, our results

strongly suggest that FocB binds DNA by minor

groove interactions

The FocB structure shows a fold similar to

DNA-binding HTH proteins

Structural similarity searches using the dali server

[37,38] identified a number of structures similar to

FocB The top seven dali hits are presented in

Table 3 Generally, the protein structures identified by

dali comprise DNA-binding proteins with HTH

motifs and, for many of the hits, structures in complex

with DNA or RNA are available in the PDB Struc-tural comparisons of FocB with some of the top DALI hits are shown inFig 6

The dali server rates structural similarity by the z-score, where values above 2 are considered to be significant hits KorA, with a z-score above 7 and a sequence identity of 18%, can tentatively be considered relevant [38] The secondary structure organization of KorA with a recognition helix of a HTH motif flanked

by three short helices is also strikingly similar to the FocB structure (Fig 6A)

Furthermore, KorA is a homodimeric repressor pro-tein with two HTH motifs that bind in the major groove on opposite sides of the DNA [39] With the N-terminal end of its recognition helix, KorA recog-nizes and binds a 12 bp symmetric operator The side chains of Gln37 in the first helix of the HTH motif, and Arg48, Gln53 and Arg57 in its second helix, are particularly important for specific interaction with the DNA [39] In FocB, these residues correspond to Arg61, positioned on helix a4, and Asn72, Thr77 and Arg81, positioned on a5 Interestingly, substitution of Arg61 and Arg81 impair DNA-binding in PapB, indi-cating that these residues are critical for DNA-binding also in FocB [25]

The DNA-recognition domain of RNA polymerase

rE-factor, another of the FocB structurally similar proteins, shows a DNA interaction similar to that of KorA The rE-factor is also a homodimer, with one HTH motif per subunit, interacting exclusively with the major grooves of two different DNA strands [40] (Fig 6B) Hepatocyte nuclear factor-1 (HNF-1b) also shares structural similarity to FocB (Fig 6C) Differ-ent from KorA and the RNAP rE-factor, this protein binds to the major groove using both helices of its HTH motif [41] Also, DNA bound to HNF-1b is ori-ented  90 with respect to the position of the DNA bound to KorA or RNAP rE-factor

Among the top DALI-hits, we also found the struc-ture of the G1⁄ S specific cyclin-D1 protein In this protein, the two helices that share structural similarity

to a HTH motif do not have DNA-binding function

A

B

Fig 5 Gel mobility shift assay of the FocB protein binding to DNA

in the absence or presence of the DNA-binding compounds

distamy-cin (minor groove-binding) and methyl green (major groove-binding),

respectively A 401 bp DNA fragment containing four repeats of the

9 bp long sequence constituting the primary FocB-binding sequence

[5] was used as target DNA The shifted bands in the gel

represent-ing different DNA complexes with protein or the tested compounds

are indicated along the left side (A) Effect of distamycin on

FocB-binding (B) Lack of effect of methyl green on FocB-FocB-binding.

Table 3 DALI search results.

[42] Other structures, such as domain 4 of the helicase

Hel308 (Fig 6D) and the RNA-binding M-domain of

Ffh (Fig 6E), gave examples of nucleic acid

interac-tions different from classical HTH–major groove inter-actions Domain 4 of Hel308 consists of a seven-helix bundle and together with other domains it forms a ring around the 3¢ tail of the unwound DNA oligonu-cleotide [43] In this interaction, the central helix of domain 4 (corresponding to a5 in FocB) provides a ratchet for directional transport of the product DNA tail across the protein Arg592 and Trp599 in the ratchet helix, corresponding to Thr77 and Arg84 in FocB, stack on base moieties of the single-stranded DNA Amino acids that are positioned in the helix corresponding to helix a4 in FocB are not involved in DNA interaction Ffh is a protein constituent of the signal recognition particle [44] and its M-domain con-tains a HTH motif that binds to the minor groove of signal recognition particle RNA with the small first helix of the motif (analogous to a4 in the FocB struc-ture) (Fig 6E)

Residues important for oligomerization and DNA-binding

Alanine substitutions in PapB, made at positions con-served throughout the PapB family, have revealed a number of specific amino acids that appear to be par-ticularly important with respect to the ability of pro-teins to bind to their target DNA, and for their ability

to form oligomeric complexes [25,34] Two residues found to be important for DNA-binding in PapB (i.e Arg61 and Cys65) are conserved and located on helix a4 in FocB (i.e the helix preceding the presumed rec-ognition helix, a5) (Fig 3) PapB binds DNA in an oligomeric fashion, probably in the form of dimers or tetramers [34] Conserved residues shown to be impor-tant for oligomerization in PapB are spread out over the FocB structure Some of these residues (i.e Asp53,

A

B

C

D

E

Fig 6 Protein–DNA and –RNA complex structures of the top five structural homologs of FocB identified by DALI [38] Ribbon repre-sentations are shown to the left and overlays of Ca-traces of the structurally similar proteins (light blue) and FocB (dark red) are shown to the right DNA strands are shown in yellow and orange, and the RNA strand is shown in coral Parts of the structures superimposing on the HTH motif of FocB are highlighted in blue in the ribbon representation To visualize the variety of DNA ⁄ RNA-binding sites, all proteins are shown with the same orientation of their HTH motif (A) KorA (PDB code 2W7N), chain B, residues 3–

97 (38–67 in blue) (B) RNA polymerase r E -factor (PDB code 2H27), chain D, residues 122–190 (185–196 in blue) (C) HNF-1b (PDB code 2H8R), chain B, residues 90–185 (157–184 in blue) (D) Hel308 (PDB code 2P6R), chain A, residues 508–612 (576–612 in blue) (E) Ffh M-domain (PDB code 1HQ1), chain A, residues 13–

83, (48–82 in blue).

Tyr54, Leu55 and Val56) are located in helix a3.

Because this helix appears necessary for stabilizing the

HTH motif, which is in itself apparently insufficient

for independent folding, these mutations most likely

affect the conformation of the whole protein structure

Other residues important for oligomerization include

residues Leu35 and Leu36, which are buried in the

dimer Interface-I In this dimer constellation, the side

chains of residues Tyr74, Phe75 and Ser76, also

previ-ously shown to be important for oligomerization, are

exposed on the surface of the protein These residues,

however, are buried at Interface-II

Discussion

The FocB protein is a transcription factor involved in

the regulation of genes for production of F1C fimbriae

that are commonly expressed by UPEC FocB belongs

to the PapB family of adhesin regulators Within this

family, PapB is currently the most thoroughly

charac-terized member and shares 81% sequence identity with

FocB The most significant difference in their

sequences is located at the N-terminal ends of the

pro-teins, a region that is predicted to form a b-strand in

the PapB protein (Fig 2) In the present study, we

have structurally characterized residues 10–99 of the

109 amino acid residue protein FocB at 1.4 A˚

resolu-tion The structure comprises five a-helices, including a

HTH motif that is quite common for many

DNA-binding proteins From the amino acid sequence alone,

no obvious recognition motifs similar to other

DNA-binding proteins could be identified in FocB

Compar-ing the PapB sequence with the FocB structure shows

that differences in their sequences are predominantly

located in the a1-loop-a2 region of the N-terminus as

well as in the C-terminal part of the two proteins

Thus, the core structure of PapB is very likely identical

to that of FocB

Results from DNA-shift assays suggest that FocB,

similar to PapB, is a minor groove-binding protein In

general, minor groove-binding proteins show various

degrees of sequence specificity when binding to DNA

The lack of unique chemical features present within

the minor groove requires various strategies for

recog-nition For example, the TATA box-binding protein

finds and interacts with the minor groove of the TATA

element, onto which it binds analgous to a saddle on a

horse [45,46] The integration host factor, on the other

hand, uses a so-called winged helix motif to bind to

regions of unusually narrow minor grooves [47] A

common feature of minor groove-binding proteins is

their ability to bend DNA For TATA box-binding

protein and integration host factor, intercalation of

hydrophobic residues between base pair steps in the DNA is the main cause of the extreme DNA-bending ability of these proteins [48] Also, LacI [49,50] and PurR [51], members of the LacI family, bend their operator DNA Both LacI and PurR form dimers attached tail-to-tail, where each dimer consists of an N-terminal DNA-binding domain and a C-terminal oligomerization domain The DNA-binding headpieces

of the N-terminal domains contain conventional major groove-binding HTH motifs, but symmetric hinge a-helices immediately adjacent to the headpieces bind deep in the minor groove, and intercalate leucine resi-dues into the central base pair step, which cause the DNA to bend

The structure of the FocB subunit contains one HTH motif Generally, this motif is in its smallest functional form part of a core of three helices that form a right-handed helical bundle with a partly open configuration The HTH motif, as it is known from transcription factors in both prokaryotes and eukary-otes, recognizes a specific DNA-sequence and is mostly associated with major groove-binding Through inter-action between the second helix of the HTH motif (i.e the recognition helix) and the DNA major groove, the sequence information is accessible for the protein [32,52,53] Recently, proteins with minor groove-bind-ing HTH motifs have been identified One example is the human DNA repair protein O6-alkylguanine-DNA-alkyltransferase [54,55] Because DNA damage

is not sequence dependent, it is favorable for DNA repair, as well as other types of nucleotide-flipping proteins, not to bind specifically to DNA In O6-alkyl-guanine-DNA-alkyltransferase, one arginine residue, which is centrally positioned on the recognition helix, stacks between DNA bases in the minor groove, caus-ing the DNA to bend In addition, small hydrophobic residues of the recognition helix are presented to the minor groove of the DNA, providing a nonsequence-specific interaction [54]

FocB and PapB bind to the minor groove of their target DNA This is in agreement with their ability to bind A⁄ T-triplets occurring at 9 bp intervals [34], where a classical, direct major groove interaction is less expected We found that the conserved residues Arg61 and Cys65, known to be important for DNA-binding, are surface exposed and located in the helix preceding the putative recognition helix (Fig 3) Also in the major groove-binding protein KorB, side chains that are not located in the recognition helix have been shown to be essential for the binding specificity [56] In that case, the specific Thr211 and Arg240 are situated outside of the HTH motif, whereas the HTH itself is typically placed with the recognition helix running

along the major groove Taken together, our findings

suggest that the HTH motif in FocB is involved in

DNA interactions that very likely are different from

the classical HTH–major groove contacts

From dali searches, several proteins with similar

structures to FocB were identified We tried to identify

the DNA-binding site in FocB based on the proteins

identified by dali; however, the latter proteins

dis-played very different types of DNA or RNA

interac-tions (Fig 6) Therefore, at this point, we refrain from

speculating on where the exact DNA-binding site of

FocB

FocB and PapB bind to DNA in an oligomeric

fash-ion [34] From the crystal structure of FocB, two

pos-sible dimeric arrangements were identified Of these,

Interface-I is significantly more hydrophobic

Further-more, one of the dali hits, the hypothetical

DNA-binding UPF0122 protein SAV1236 from

Staphylococcus aureus (PDB code 1XSV; unpublished)

(Table 3), is a dimer with a packing interaction

resembling that of Interface-I in FocB (Fig 7)

Mutants that impair PapB oligomerization [25] are

localized both at Interface-I and -II We therefore

con-sider that Interface-I represents the most stable dimeric

form of FocB, but that Interface-II might play a role

in the formation of larger oligomeric structures

neces-sary for DNA-binding [34] We hypothesize that

several FocB dimers can bind to DNA side by side

The crystal structure of the FocB transcription

factor provides an important starting point for further

analyses aiming to understand the mechanisms of fimbrial gene regulation at the molecular level not just for FocB, but also for the entire family of related proteins

Experimental procedures

Protein expression and purification of native and Se-Met FocB

The overexpression and purification of native FocB (109 amino acid residues) has been described previously [27] The full-length FocB protein was cloned into pETM11 and overexpressed in E coli strain BL21 (DE3) The protein was purified on a Ni-NTA (Qiagen, Valencia, CA, USA) column followed by a Superdex 75 column (GE Healthcare, Milwaukee, WI, USA) During purification, the 6-His tag was removed with tobacco etch virus protease, leaving two extra residues, Gly and Ala, followed by Met1 correspond-ing to the native N-terminus Pure fractions of the protein

in 10 mm Hepes (pH 7.9), 500 mm NaCl, 5 mm EDTA and 0.1% b-mercaptoethanol were pooled and concentrated to

16 mgÆmL)1, filtered through a 0.2 lm filter and stored at

4C

Selenomethionyl labeling of FocB (Se-Met FocB) was performed using the protocol as described previously [57] FocB Se-Met was overexpressed in E coli BL21 (DE3) cells

KH2PO4, 9 mm NaCl, 19 mm NH4Cl, 2 mm MgSO4, 0.1 mm CaCl2, 4 gÆL)1 glucose) with the addition of kanamycin (0.1 mgÆmL)1) at 37C until D600= 0.6 The temperature

Fig 7 Structural similarity between (A) the

dimer Interface-I of FocB and (B) the dimer

interface of protein UPF0122 (PDB code

1XSV, unpublished) Residues Leu13-Leu76

of the protein matched residues

Ser27-Ala93 of FocB with a z-score of 5.1

(Table 2) The structures are shown in two

orientations The visual alignment is based

on the position of the HTH motif.