Heme binding by a bacterial repressor protein, the gene product of the ferric uptake regulation (fur) gene ofEscherichia coli

Journal of Protein Chemistry, Vol. 15, No. 6, 1996

Heme Binding by a Bacterial Repressor Protein, the Gene Product of the Ferric Uptake Regulation (fur) Gene of Escherichia coli

Ann Smith, ''2 Nigel I. Hooper, 1 Natalya Shipulina, 1 and William T. Morgan I

Received June 6, 1996

The fur gene product, Fur, of Escherichia coli is a repressor when it binds Fe(II). Since heme and iron metabolism are closely linked and Fur is rich in histidine, a ligand for heine, the binding of heine to Fur was investigated. The oxidized Fur-he ine complex is stable and low spin with a Soret maximum at 404 nm and no 620-nm band. CO coordinates with the reduced he ine-Fur complex, causing a shift from 412 nm to 410 nm, and stabilizes it, increasing the half-life from 5 to 15 min. Circular dichroism (CD) spectra in the Soret region show heme bound in an asymmetric environment in Fur, both in the oxidized and reduced-CO forms. Quenching of tyrosine fluorescence by heine revealed rapid, tight binding (Kd < 1/xM) with an unusual stoichiometry of 1 heme:l Fur dimer. Fur binds Mn(II), a model ligand for the endogenous Fe(II), much more weakly (Kd>80/xM). Far-ultraviolet CD spectroscopy showed that the a-helix content of apo-Fur decreases slightly with heme binding, but increases with Mn(II) binding. Competition experiments indicated that heme interacts with Fur dimers at the same site as Mn(II) and can displace the metal. In contrast to Mn(II), Zn(II) did not quench the tyrosine fluoroescence of Fur, affected the CD spectrum less than Mn(II), but did bind in a manner which prevented heme from binding. In sum, Fur not only binds heme and Zn(II) with sufficient affinity to be biologically relevant, but the interactions that occur between these ligands and their effects on Mn(II) binding need to be taken into account when addressing the biological function of Fur.

KEY WORDS: Fur; heine; iron; transcription factor; binding; zinc.

1. INTRODUCTION

The growth of pathogens within body tissues and fluids is opposed in part by defense mechanisms of the host which restrict iron availability. Several strategies have been developed by pathogens to overcome these defenses and establish an infection. These include production of iron chelators, such as the catechol siderophores enterobactin and aerobactin, of the corresponding iron-siderophore

1Division of Molecular Biology and Biochemistry, School of Biological Sciences, University of Missouri-Kansas City, Kan- sas City, Missouri 64110.

2To whom correspondence should be addressed; e-mail: [email protected].

575

recognition proteins (receptors), and of other proteins which function in the transport of iron. In some strains of bacteria, outer membrane proteins which bind iron-transferrin (Brock et al., 1991), heme-hemopexin (Wong et al., 1994), or iron- lactoferrin (Mickelsen et al., 1982) are produced. 3

The protein components of these iron and heme acquisition systems are often regulated by extracellular iron and heme levels. One of the best-studied cases is the regulation of siderophore synthesis by Escherichia coli. The expression of

3Abbreviations: Heme, iron-protoporphyrin IX; mesoheme, iron-mesoporphyrin IX; cobalt-protoporphyrin IX; CD, circular dichroism.

0277-8033/96/0800-0575509.50/0 �9 1996 Plenum Publishing Corporation

576 Smith, Hooper, Shipulina, and Morgan

several genes, including flu, f epA , and f h u F , is regulated by the binding of a repressor protein, Fur, the Mr-17-kDa product of the fur gene (Schaffer et aI., 1985; Wee et al., 1988). Fur, like many DNA-binding proteins, is a homodimer at concentrations above I ~ M (Coy and Neilands, 1991). A model for Fur to function as a repressor of transcription has been developed based on mutational analysis and DNA footprinting. Upon ligation of Mn(II), as a model for Fe(II), Fur binds more tightly to the operator regions of several bacterial genes involved in iron uptake and in pathogenic virulence (Coy and Neilands, 1991; De Lorenzo et al., 1987) thereby downregulating the expression of these genes when extracellular iron is available. Examples of iron- regulated genes affected by Fur include those encoding proteins for siderophore synthesis, recognition, and transport (Coy and Neilands, 1991), for the production of toxins (Dubos and Geiger, 1946; Boyd et al., 1990), and for hemolysins (Stoebner and Payne, 1994; Pickett et al., 1992). Furthermore, a role for Fur in cellular metabolism is indicated by several observations, including the following: mutations in the fur gene also cause defects in dicarboxylic acid transport, a decrease in succinate dehydrogenase (Neilands, 1990), and abnormal expression of superoxide dismutase (Niederhoffer et al., 1989). At high concentrations (->100/xM) even apo-Fur binds to DNA with sufficient affinity to create a "footprint" in vitro (Saito and Williams, 1991b). Yersinia pestis also responds to iron in its environment and possesses a regulatory protein analogous to Fur that is capable of interacting with the E. coli fur system, even though there is significant nucleotide sequence divergence (Staggs and Perry, 1991). Other Fur analogs are being identified, e.g., in Neisseria gonorrhoeae (Berish et al., 1993).

A working model of Fur's three-dimensional structure has been derived from NMR data (Saito and Williams, 1991a; Saito et al., 1991a). The basic form is a relatively stable bundle of five a-helices connected to mobile loops and ends, which were too mobile to enable determination of the complete structure to high resolution. A series of antiparallel amphipathic helices occur at the protein surface, and the Fe(II)-binding site, identified using Mn(II) as a probe, involves principally His32, His131, as well as His142 and Hislaa, together with some carboxylate

side chains (Saito and Williams, 1991b; Saito et al., 1991b). Thus, the primary Fur-binding site appears to resemble that for iron in hemerythrin (Zhang et al., 1991) and that for Mn(II) in concanavalin A (Edleman et al., 1972). Binding of metal to Fur induces a conformational change which renders Fur more protease sensitive (Coy and Neilands, 1991), but binding of apo-Fur to DNA protected the protein from proteolysis, suggesting that the conformational changes caused by metal binding are not requisite for DNA binding, but rather influence the affinity of binding.

Heme and iron metabolism are clearly linked in bacteria, yeast, and mammalian cells (Smith, 1990). Heme has been shown to regulate its own uptake in several human bacterial pathogens, including Hemophi lus influenzae (Hanson et al., 1992; Wong et al., 1994) and Vibrio cholerae (Henderson and Payne, 1993) in part via the expression of outer membrane heme-binding proteins in H. influenzae (Wong et al., 1994, 1995; Hanson et al., 1992), Neiserria sp. (Lee, i992; Lee and Hill, 1992), and Porphyromonas gingivalis (Bramanti and Holt, i992, 1993). Heine also exerts pleiotropic regulatory effects on genes in yeast (Zhang et al., 1993) and mammals. For example, in mammalian cells heme, heme analogs, and heme-hemopexin (in a manner distinct from iron) induce the expression of heme oxygenase (Lutton et al., 1992; Mitani et al., 1993; Smith et al., 1993; Alam et a t , 1989; Alam and Smith, 1989; Morgan et al., 1988), 6-amino-levulinate synthase (Elferink et al., 1988; Srivastava et al., 1988), and metallothionein-1 (Smith et al., 1993; Alam and Smith, 1989).

Since Fur contains an unusually high number of histidine residues (Hantke, 1981), since histidine is a common heme ligand, and since histidines have been implicated in iron binding to Fur (see above), we used spectroscopic techniques to determine whether Fur interacts with heme and whether heme competes with metals for binding to Fur. A specific, high-affinity interaction of heme with Fur was found which prevents Mn(II) binding. We have also shown that, like many other transcription factors, Fur binds Zn(II) and, furthermore, that this binding is different from that of Mn(II) and heme, yet prevents heine from binding. This is the first instance of a low-spin heme-DNA-binding- protein complex in prokaryotes and points to the potential for heme-transcription factor interac-

Fur-Heme Interactions

tions in eukaryotes for heme-regulated genes like heme oxygenase-1.

2. MATERIALS AND METHODS

Fur (generously provided by Prof. J. Neilands, University of California, Berkeley, CA) was isolated as described (Wong and Saha, 1991) and lyophilized. The protein was dissolved in either 50mM sodium phosphate, pH7.4, or 50mM sodium borate, p H 8, immediately before use, and the concentration of Fur was determined spectrophotometricalty using an extinction of 0.4 A . cm-1 at 275 nm for a I mg/ml solution and a monomer molecular weight of 17kDa. The experiments with ZnC12, could not be carried out in phosphate buffer, so for studies at neutral pH, Fur was dissolved in HEPES-NaOH, p H 7.4. However, while absorbance spectra were valid, reliable CD spectra were limited and data at wavelengths below 220nm could not be obtained due to the absorbance of HEPES. Since Fur is predomin- antly dimeric at concentrations above 1/xM (Coy and Neilands, 1991), concentrations of monomer >-4/xM were routinely used, and additional details are given in the text and figure legends.

Metalloporphyrins were obtained from Porphyrin Products, Logan, UT. Concentrations of metalloporphyrin solutions were determined spectrophotometrically using the following extinction coefficients (A. M -1 �9 cm-1): mesoheme, 1.7 • 105 at 394nm in dimethysulfoxide and cobalt- protoporphyrin, 1.8• 105 at 424nm in 0.1M NaOH:pyridine:H20 (3:10:87) as previously described (Smith et al., 1993). Mesoheme was used rather than the naturally occurring protoheme because it is more stable, is biologically equivalent in nearly every aspect, and is less prone to aggregation and oxo-/x dimer formation. Mn(II) and Zn(II) solutions were prepared gravimetrically in l mM HC1 using reagent-grade MnC12 and ZnCI2, respectively. When feasible, borate buffer was used for experiments with Mn(II) and HEPES buffer for Zn(II). Experiments with Fe(II) were precluded by the tendency of ferrous ion to oxidize in the presence of even trace amounts of oxygen and the difficulty in deaerating and maintaining anaerobic Fur and buffer solutions. For the same reasons, Mn(II) has been routinely used in most work on Fur. Mesoheme-Fur complexes (at i heme:4 Fur monomer or lower molar ratio) in

577

50mM phosphate buffer, pH7.4, were bubbled gently with CO and then reduced with Na2S204 to form the CO-reduced heme-Fur complex. Complex formation was confirmed by absorbance spectroscopy. Treatment with dithionite before CO produced a less stable complex.

Absorbance and fluorescence spectra were recorded using a Hewlett-Packard 8452 or SLM Aminco 3000 diode-array spectrophotometer (Mil- ton Roy) and an SLM 4000 fluorimeter, respectively. Since Fur contains no tryptophan residues, ligand binding could be monitored by quenching of the intrinsic tyrosine fluorescence of Fur at 310 nm with excitation at 278 nm. Circular dichroism (CD) spectra were recorded in a Jasco 720 spectropolari- meter at room temperature using a 0.1-cm or 1.0-cm light path for the far-UV and Soret regions, respectively. CD data were averaged (usually ten scans per sample) and normalized using software supplied with the instrument. Additional ex- perimental details are given in the figure legends.

Titration data were analyzed by nonlinear regression using the "Enzfitter" program (Elsevier Publishing Co.). In some cases, due to difficulties in obtaining concentrations appropriate for the determination of stoichiometry of binding (i.e., concentrations >-10 times the Kd), a value could not be determined.

3. RESULTS

3.1. Demonstration of Heme Binding by Fur

Addition of Fur (Mr 17kDa) progressively shifted the Soret absorbance maximum of mesoheme in buffer from 389 to 404 nm until saturation was reached (Fig. 1). This red shift, together with the shape and properties of the visible bands, particularly the absence of a band near 620nm (Kaminsky et al., 1972), indicate that a low-spin heme-Fur complex is formed involving two strong-field ligands like histidine. Reduction of the stable Fur-heme complex with sodium dithionite produced a relatively unstable complex (half-life ca. 5 min) with a maximum near 412 nm and a shoulder near 425 nm. CO bound to the reduced heme-Fur (Soret maximum near 410nm) and stabilized it, increasing the half-life to ca. 15min. Both the heme-Fur and CO-reduced-heme-Fur complexes exhibited positive ellipticity bands in the Soret region (Fig. 2), indicating that the normally


0.3

r 0.2 e- r

O

J~ <[ 0.1

0.0

�9 J . , . i .

A 0.050 ~ . , . , �9 ,

B

0.025 ~ .

0 000 ~ ~ L

0 . 5 ,

0.4

0.3 t

0.2

0.1

0.0 [ 250

f . . /:

/ r

it ""

300 350 400 450 250 300 350 400 450 450 500 550 600 650

Wavelength (rim)

Fig. 1. Absorbance spectra of Fur and the Fur-heme complex. The Fur-mesoheme complex was formed by incubating Fur (16/zM) with mesoheme (4/xM) for 5 min at 22~ in 50 mM sodium phsophate, pH 7.4. The spectra shown are: Fur only (dashed line), mesoheme only (dotted line), and the Fur-ferric-mesoheme complex (solid line). At this concentration Fur would exist primarily as a dimer. The visible region is shown on an expanded scale in panel B. (C) Spectra of heme alone in sodium phosphate buffer, pH 7.4 (4 k~M, molar ratio 1:1; solid line), and lysozyme-heme mixtures (4/,tM, molar ratio 1:1; dashed-dotted line).

CD-silent heme was bound in an asymmetric environment in the protein. As noted above, at the concentrations of Fur used, the protein would exist primarily as a homodimer.

To help demonstrate that the association of Fur (Mr 17kDa) with heme involves specific ligation to heme-iron, lysozyme (which, like Fur, is a basic protein of low molecular weight, Mr 14.5 kDa) was used as a control for nonspecific adherence of heme to protein. Adding lysozyme to mesoheme did not produce any significant change in the wavelength maxima of the absorbance spectrum of heme (Fig. 1C). A small decrease in absorbance occurred with time, but is likely due to the progressive aggregation of heme that occurs in aqueous solution at neutral pH.

3.2. Characterization of Binding of Metailoporphyrin and of Mn(II) by Fur

To titrate the binding reactions, quenching of the intrinsic fluorescence of Fur by ligands was used. Fur contains no tryptophan, but four tyrosine residues; consequently its fluorescence spectra are characteristic of tyrosine, i.e., excitation maximum at 278nm and emission maximum at 310nm. Addition of mesoheme effectively quenches the tyrosine emission of Fur (Fig. 3A), and the affinity of Fur for heme is much higher (Kd < 1/xM) than for Mn(II) (see below). Interestingly, the binding stoichiometry is 1 heme:l Fur dimer, suggesting either an intersubunit heme-binding site or a steric hindrance exerted on the second subunit once one

heme binds to the first. Changes in absorbance (data not shown) showed that the heme analog cobalt-protoporphyrin (CoPP), which, like heme, is readily coordinated by histidine, also binds to Fur. There is a shift in the Soret maximum of CoPP from 419 to 427nm and a slight decrease in intensity. CoPP (data not shown) and Mn(II) (Fig. 3B) also quench the tyrosine fluoresence of Fur in a saturable manner, whereas ZnPP does not (data not shown)�9 The apparent Ka of the Fur-CoPP complex is 3.8/xM, comparable to that of Fur-heme, and the apparent Kj for Fur-Mn(II) is 80/xM, in good agreement with the affinity estimated previously for Fur-Fe(II) (Wong and Saha, 1991). The stoichiometry of Mn(II) binding to Fur could not be determined due to the low affinity and consequent inability to examine concentrations near ten times the Ka.

3.3. Changes in the Conformation of Fur upon Binding Heme or Mn(II)

The far-ultraviolet circular dichroism (CD) spectrum of Fur is typical of a protein with high (>50%) a-helix content and low/~-sheet secondary structure (Fig. 4), in agreement with the published NMR data (Saito et at., 1991b). Upon binding of either mesoheme or Mn(II), the CD spectrum of Fur is altered, indicating that conformational changes occur, but in distinct ways. Interestingly, as judged by the ellipticity at 222 nm, the CD band most clearly associated with a-helix content of proteins, the extent of a-helix in Fur is somewhat

0.8

1~0

0.6

~ 0.4

<

0.2

0.0

1.4

1.2

1.0

~ 0.8

~ 0.6

~1 0.4

0.2

0.0

35O

A / / ~ \\ CO-herne-Fur heme-Fur / / \ \

i i i l

i

400

i J i i

i CO-heme-Fur

450

Wavelength (nm)

Fig. 2. Absorbance and circular dichroism spectra of the heine- Fur and the CO-reduced-heme-Fur complexes. All spectra shown of heme-Fur and the CO-reduced-heme-Fur complexes were recorded at 22~ in 50mM sodium phosphate, pH7.4, buffer. The concentrations of Fur and heine used were 26/zM Fur and 7/xM mesoheme. The heme-Fur circular dichroism (millidegrees) spectrum represents the average of ten scans, and the CO-reduced-heme-Fur spectrum the average of two scans; instability of the complex precluded calculation of the molar ~llipticity due to uncertainty in complex concentration. The observed absorbance maxima of 404 and 410 nm (panel A) for the berne-Fur and CO-reduced-berne-Fur complexes, respectively, are very near the apparent maxima from the CD spectra of 403 and 411 nm (panel B).

decreased by heme, but increased by Mn(II) (Fig. 4).

3.4. Competition Between Heme and Metal for Binding to Fur

Competition experiments were carried out with heme and Mn(II) to confirm the tighter binding of heme and to examine whether these ligands compete at the same or different sites on Fur. As shown by absorbance spectroscopy, adding a ten-fold molar excess of Mn(II) over heme caused minimal changes in the preformed Fur-

0.9

0.8

0,7

=~ 0.6 0 U~

0

1.01 .>

"~ 0.9

, I r I , I

0 1 2 3

Mesoheme/Fur Dimer Ratio

i i i i

B

0.8

0.7

0.6

0.5

0.4

A

\ \

Fur-Heme Interactions 579

0.3 I [ L I i

0 I 0 20 30 40

Mn (II), I~M

Fig. 3. Quenching of the intrinsic tyrosine fluorescence of Fur by heme and Mn(II). The intrinsic fluorescence of Fur (3.8/xM) was measured in 50raM sodium phosphate buffer, pH7.4. Increments of (A) mesoheme or (B) Mn(II) were incubated with Fur for 3min before each measurement of fluorescence (excitation at 280 nm and emission at 310 nm) was recorded. The initial and final slopes of the biphasic quench curve of mesoheme were determined by linear regression analysis (d~ished lines).

heme complex (Fig. 5A). However, adding heme to preformed Fur-Mn(II) complexes [at a ratio of 1 heme:10 Mn(II)] rapidly generated the typical Fur-heme Soret absorbance spectrum (Fig. 5A). Similarly, adding heine to the preformed Fur- Mn(II) complex [at a ratio of 1 heme:25 Mn(II)] shifted the far-UV CD spectrum from that typical of the Fur-Mn(II) complex toward that typical of the Fur-heine complex (Fig. 5B), i.e., toward a lower helical content as shown by the less intense ellipticity at 222 rim. In contrast, Mn(II) added to


1.5

1.0

o 0.5 x

0.0 U

Q. -0.5

uJ

"6 -1.0

-1.5

-2.0 200

i L \ / . . ~ / . . '~.~ / / - - - - - - - - / . / '

. . . . .

I i I i I f I i

210 220 230 240 250

Wavelength

Fig. 4. Circular dichroism spectra of Fur and the Fur-heme and Fur-Mn complexes. Circular dichroism (molar ellipticity [| deg.cm2.dmol -~) spectra of Fur (solid line), the Fur- mesoheme complex (dashed line), and the Fur-Mn(II) complex (dotted line) were recorded at 22~ in t0 mM sodium phosphate, pH7.4, or 50 mM borate buffer, prig.0 [for the Fur-Mn(II) complex only]. The concentrations of Fur and ligands used were 3.5/LM Fur, 3.5 txM mesoheme, and 87.5/xM Mn(II)C12. Identi- cal spectra were obtained for Fur in the phosphate and borate buffers (data not shown).

the F u r - h e m e complex had no effect on the ellipticity (data not shown). Thus, both absorbance and CD data indicate that heme competes with Mn(II) for binding to Fur and is bound more tightly than the metal. Since both heme and Mn(II) quench tyrosine fluorescence and alter the helical content of Fur, and in view of the measured binding affinities, heme most likely binds at or near the same site as Mn(II) rather than to an independent site.

3.5. Effect of Zn(II) on Fur and on Fur-Heme Complexes

Since many DNA-binding proteins bind Zn(II) via histidine and cysteine residues, and Zn(II) is likely to be present in buffers used in footprinting assays, we carried out a series of competition experiments with Zn(II). Absorbance spectroscopy revealed that Zn(II) (10/xM) prevented heine (1/xM) from binding to Fur (4.4/xM; Fig. 6A). Furthermore, addition of ZnC12 displaced heme when added to preformed h e m e - F u r complex, as shown by the decrease in absorbance at 404nm (Fig. 6A). However, Zn(II) does not change the tyrosine emission spectra of Fur (data not shown)

0.08

0.06

o r-

0 . 0 4 0 ,.Q <

0.02

0.00 300

, , , , 1 1 , , , I , , , , A , I , L h , I , , , ,

3 5 0 4 0 0 4 5 0

Wavelength (nm)

o 1

X

r 0 Q.

UJ

-1 :E

-2 200

I i i I

B

,t,, / f I;- ,'/"

I I I I

210 220 230 240 250

Wavelength (nm)

Fig. 5. Competition between heme and Mn(II) for binding to Fur. (A) Fur (10/xM) was incubated with 5/xM mesoheme for 5 rain at 22~ in 50mM sodium phosphate, pH7.4. Mn(II) (50/xM) was then added, and after another 5 min the sample was rescanned. The absorbance spectra shown are the Fur- mesoheme complex (solid line) and Fur with mesoheme added either before (Mn(II) (dashed line) or 5min after Mn(II) (dotted line). (B) CD spectra of Fur (3.5/xM, dotted line) and Fur-Mn(II) in the absence (solid line) and presence of heme (dashed line) were recorded at 22~ in 50mM borate buffer, pH8.0. Fur was incubated with 87.5/LM Mn(II) for 10min on ice before its CD spectrum was recorded, followed by addition of 3.5/xM mesoheme and after a further 10-min incubation on ice the spectrum was rerecorded.

and only slightly decreases the ellipticity at 222 nm in the far-UV CD spectra of Fur (Fig. 6B).

4. DISCUSSION The data presented here shown that one heme

molecule is bound tightly (Kd < 1/~M) by one Fur dimer. When heme binds to the Fur dimer, a

Fur-Heine Interactions 581

o r -

s o

0.08 [

0.04

0,02

1

A

0.0

, , , , """ - 7 . . . . . . 7" . . . . . . . . . . . . . ; . . . . . . . . . . . . .

300 400 500 600

O -0.5

X

o - -

ul

_~ A.0 o

B

/

0,00 -1 ,5 . . . . ' . . . . ' . . . . 2 2 0 2 3 0 2 4 0 250

Wavelength (nm)

Fig. 6. Interaction of Zn(II) with Fur using absorbance and CD spectroscopy. (A) Fur (4.4/xM) was incubated with 10 ixM ZnC12 for 10 rain at 22~ followed by addition of 1 ~M heine for 30 min. The spectra shown are Fur-Zn (dotted line) and Fur-Zn after addition of heine (dashed line). Fur-heine complexes were formed by incubating Fur (4.4/xM) with heine (1 txM) for 60 rain at 22~ in 50 mM sodium phosphate buffer, pH 7.4 (solid line), followed by incubation for 30 min with 10p~M ZnC12 (dashed and dotted lines). (B) Circular dichroism spectra of 4/~M Fur monomer (solid line) in HEPES buffer, pH 7.4, and the Fur-Zn(II) complex (dashed line) generated by the addition of 5/xM ZnC12 to Fur. Spectra were recorded ten times to accumulate data and recording was started 5 min after the last addition to the sample in the cuvette.

low-spin heme complex is formed, possibly with a bis-histidyl coordination. This conclusion is supported by the features of the oxidized and reduced visible absorbance spectra of heme-Fur. Further- more, upon binding of heme, CoPP (data not shown), or Mn(II), the intrinsic tyrosine fluorescence of Fur is quenched in a saturable manner, and the CD spectra indicate that distinct conformational changes are induced in Fur by heme and Mn(II). As judged by the ellipticity at 222nm, heine binding causes a significant decrease, and metal binding a small increase, in the helical content of Fur. An increase in helix upon binding Mn(II) is consistent with a more folded form with potentially tighter monomer-monomer interactions which may influence the protease sensitivity for various functional domains of Fur, e.g., DNA binding (Coy and Neitands, 1991), whereas the decrease in helix content upon binding heine suggests a more extended structure possibly from the Fur dimerization. Previous NMR studies were interpreted to show that Fur is a relatively rigid protein, but dimerization could not be addressed at that time (Saito and Williams, 1991b; Saito et al., 1991b). Electron spin resonance data suggested the presence of Mn(II) in one site per Fur monomer in a low-symmetry environment, whereas Cu(II) was

bound in two different environments: a major site with nitrogen and a minor site with sulfur (Hamed and Neilands, 1994). Detailed studies of the Fur interaction with DNA sequences within the regulatory regions of the aerobactin and hemolysin operons indicated that metal binding is required for the specific interaction with DNA and that the protein polymerized on its DNA-binding sites (Frechon and Le Cam, 1994).

Heine competes with Mn(II) for binding to Fur in vitro and presumably with Fe(II) in vivo. Proteins that bind both metal and heine at the same site are uncommon. However, this may explain the unusual stoichiometry of one heme per Fur dimer. Heme, Mr 640, is larger than Mn(II) or Fe(II) and is a hydrophobic, planar molecule with a distinct asymmetry: a hydrophilic face due to propionate side chains on the C and D pyrrole rings and a hydrophobic face formed by the remainder of the molecule. Evidence for heme binding in bis-histidyl coordination by synthetic helical bundle dimers (Robertson et al., 1994) and in bis-methionine coordination in side-by-side helices from two different subunits of bacterioferritin was recently reported (Cheesman et aL, 1990). The fact that the bound heme displays positive ellipticity in the Soret region indicates that the heine is in an


asymmetric environment in the Fur dimer. The pH dependence of Mn(II) binding provides evidence that histidines are involved with pKa values in the range p H 6 - 9 (Saito et al., 1991a). Heine may ligate with His31, His32, or His131, all of which lie in the primary Mn(II)-binding site, and the tyrosine residues whose fluorescence is affected by binding could be Tyr39, Tyr127, or Tyr129, which are near the His residues noted.

The interaction of Fur with Zn(II) is more difficult to interpret, since Fur contains four free sulfhydryl groups, which, like the 12 histidine residues, are avid ligands for Zn(II), but much less so for Mn(II). In fact, recombinant Fur has been purified using Zn(II)-chelate affinity chromatog- raphy (Wee et al., 1988). The CD and absorbance data suggest that Zn(II) is bound at least as tightly as heme, but since Zn(II) did not quench the intrinsic tyrosine fluorescence of Fur as do heme and Mn(II), Zn(II) is probably bound at alternate site(s). Nevertheless, Zn(II) prevents heme binding and bound heine is released by Zn(II), yet without greatly altering the secondary structure of Fur. Thus, whether the Zn(II) directly competes with heme for histidine side chains or interacts with sulfhydryl groups to block binding will require further study.

Another precedent for heine binding at the interface of multimeric proteins is the heme- induced dimerization and activation of the transcription factor HAP-1 in yeast (Zhang et al., 1993), and recently heme has been suggested to release HAP-1 from a higher molecular weight complex (Fytlovich et al., 1993). Fur does not contain the proposed HAP-1 heine-binding motif KCPVDH (Creusot et al., 1988; Pfeifer et al., 1989), which is too short to provide more than one heme-iron ligand (potentially K, C, or H). Moreover, the reported absorbance spectra (Cre- usot et al., 1988) are unusual and definitely not characteristic of low-spin heme, e.g., as displayed by cytochrome b5 and hemopexin. Thus, this work is the first instance of a low-spin heme-DNA- binding-protein complex consistent with the coordination of the heme iron by two strong-field ligands like histidine. However, together with the observations in yeast noted above, the possibility that heme binds directly to one or more transcription factors to displace a regulatory protein in both eukaryotic and prokaryotic cells in which there is heme-regulated expression of genes like heine oxygenase-1 merits consideration.

The tight binding of heme by Fur, evident from the data presented here, points to a previously unrecognized physiological role for Fur -he ine complexes in regulating bacterial metabolism. Several possibilities can be envisaged. If high intraeellular heme concentrations are an alternate signal that the cell is iron-replete, then, like binding of Fe(II), heme binding to apo-Fur might increase its affinity for DNA and thus its repressive effect. On the other hand, since binding of heine to Fur prevents Mn(II) binding in oitro, then heine might inhibit binding of Fe(II) to Fur in vivo, thus lowering the Fur-operator interaction and dere- pressing the iron uptake systems. The latter would allow iron uptake for the synthesis of heme needed for aerobic metabolism. In addition, changes in metabolism and the external milieu of microorgan- isms which affect Zn(II) uptake and intracellular pools are potentially capable of influencing the DNA binding activity of Fur in vivo.

Notably, repression of genes by Fur depends both upon the metal bound and the gene regulated. For example, Fe(II), but not Mn(II), acts as a corepressor of Fur-dependent Mn-superoxide dismutase biosynthesis, while both metals serve as corepressors for aerobactin synthesis (Privalle and Fridovich, 1993). Heme binding to Fur may similarly be more important in the regulation of genes other than those for the siderophore systems. In this regard it is interesting that the expression of Fur itself is controlled in two ways: by iron levels and by the cAMP-catabolite activator protein (CAP) system (De Lorenzo et al., 1988). Thus, modulation of iron absorption and the metabolic state of the cell are both implicated in the regulation of Fur. Perhaps heme binding by Fur, and possibly to proteins related to Fur, has a more general role in the regulation of genes involved in intermediary metabolism, in defense against heme- mediated oxidative stress, or in certain pathogens even in the acquisition of heme-iron from heme-hemopexin or hemoglobin than in the regulation of iron-siderophore uptake systems per s e .

ACKNOWLEDGMENTS

The authors wish to thank Prof. J. P. Neilands (Department of Molecular and Cell Biology, University of California, Berkeley, CA) for his generous gifts of Fur protein and for sharing

Fur-Heme Interactions 583

unpublished data with us. This work was supported in part by USPHS grant DK-37463 to A.S.

A preliminary account of portions of this work was presented at the International Bio-Iron Conference, Oxford, England, 1992, and the ASMBM/Biophysical Society Annual Meeting, Houston, Texas, 1992.

REFERENCES Alam, J., and Smith, A. (1989). J. Biol. Chem. 264, 17637-

17640. Alam, J., Shibahara, S., and Smith, A. (1989). J. Biol. Chem.

264, 6371-6375. Berish, S. A., Subbaroa, S., Chen, C.-Y., Trees, D. L., and

Morse, S. A. (1993). Infect. Immun. 61, 4599-4606. Boyd, J., Oso, M. N., and Murphy, J. R. (1990). Proc. Natl.

Acad. Sci. USA 87, 5968-5972. Bramanti, T. E., and Holt, S. C. (1992). J. BacterioL 174,

5827-5839. Bramanti, T. E., and Holt, S. C. (1993). J. Bacteriol. 175,

7413-7420. Brock, J. H., Williams, P. H., Lic6aga, J., and Wooldridge, K. G.

(1991). Infect. lmmun. 59, 3185-3190. Cheesman, M. R., Thomson, A. J., Greenwood, C., Moore, G.

R., and Kadir, F. H. (1990). Nature 346, 771-773. Coy, M., and Neilands, J. B. (1991). Biochemistry 30, 8201-8210. Creusot, F., Verdiere, J., Gaisne, M., and Slonimski, P. (1988).

J. MoL Biol. 204, 263-276. De Lorenzo, V., Wee, S., Herrenro, M., and Neilands, J. B.

(1987). J. Bacteriol. 169, 2624-2630. De Lorenzo, V., Herrero, M., Giovannini, F., and Neilands, J.

B. (1988). Eur. J. Biochem. 173, 537-546. Dubos, R. J., and Geiger, J. W. (1946). J. Exp. Med. 84,

143-156. Edelman, G. M., Cunningham, B. A., Reeke, G. N., Becker, J.

W., Waxdal, M. J., and Wang, J. L. (1972). Proc. Natl. Acad. Sci. USA 69, 2580-2584.

Elferink, C. J., Sassa, S., and May, B. K. (1988). J. Biol. Chem. 263, 13012-13016.

Frechon, D., and LeCam, E. (1994). Biochem. Biophys. Res. Commun. 201, 346-355.

Fytlovich, S., Gervais, M., Agrimonti, C., and Guiard, B. (1993). EMBO J. 12, 1209-1218.

Hamed, M. Y., and Neilands, J. B. (1994). J. Inorg. Biochem. 53, 235 -248.

Hanson, M. S., Slaughter, C., and Hansen, E. J. (1992). Infect. Immun, 60, 2257-2266.

Hantke, K. (1981). Mol. Gen. Genet. 182, 288-292. Henderson, D. P., and Payne, S. M. (1993). Mol. Microbiol. 7,

461-469. Kaminsky, L. S., Byrne, M. J., and Davison, A. J. (1972). Arch.

Biochem. Biophys. 150, 355-361. Lee, B. C. (1992). J. Med. Microbiol. 36, 121-127. Lee, B. C., and Hill, P . (1992). J. Gen. Microbiol. 138,

2647-2656.

Lutton, J. D., da Silva, J.-L., Moqattash, S., Brown, A. C., Levere, R. D., and Abraham, N. G. (1992). J. Cell. Biochem. 49, 259-265.

Mickelsen, P. A., Blackman, E., and Sparling, P. F. (1982). Infect. lmmun. 35, 915-920.

Mitani, K., Fujita, H., Fukuda, Y., Kappas, A., and Sassa, S. (1993). Biochem. J. 290, 819-825.

Morgan, W. T., Alam, J., Deaciuc, V., Muster, P., Tatum, F. M., and Smith, A. (1988). J. Biok Chem. 263, 8226-8231.

Neilands, J. B. (1990). In Pseudomonas: Biotransformations, Pathogenesis, and Evolving Biotechnology, American So- ciety of Microbiology, Washington, D.C., pp. 382-395.

Niederhoffer, E. C., Naranjo, C. M., and Fee, J. F. (1989). In Metal Ion Homeostasis: Molecular Biology and Chemistry (Hamer, D. H., and Winge, D. R., eds.), Alan R. Liss, New York, pp. 149-158.

Pfeifer, K., Kim, K., Kogan, S., and Guarente, L. (1989). Cell 56, 291-301.

Pickett, C. L., Auffenberg, T., Pesci, E. C., Sheen, V. L., and Jusuf, S. S. D. (1992). Infect. Immun. 60, 3872-3877.

Privalle, C. T., and Fridovich, I. (1993). J. Biol. Chem. 268, 5178-5181.

Robertson, D. E., Farid, R. S., Moser, C. C., Urbauer, J. L., Mulholland, S. E., Pidikiti, R., James, D. L., Wand, A. J., DeGrado, W. F., and Dutton, P. L. (1994). Nature 368, 425-432.

Saito, T., and Williams, R. J. P. (1991a). Eur. J. Biochem. 197, 43-47.

Saito, T., Duly, D., and Williams, R. J. P. (1991a). Eur. J. Biochem. 197, 39-42.

Saito, T., Wormald, M. R., and Williams, R. J. P. (1991b). Eur. J. Biochem. 197, 29-38.

Schaffer, S., Hantke, K., and Braun, V. (1985). Mol. Gen. Genet. 201, 204-212.

Smith, A. (1990). In Biosynthesis o f Heine and Chlorophylls (Dailey, H. A., Jr., ed.), McGraw-Hill, New York, pp. 435-489.

Smith, A., Alam, J., Escriba, P. V.I and Morgan, W. T. (1993). J. Biol. Chem. 268, 7365-7371.

Srivastava, G., Borthwick, I. A., Maguire, D. J., Elfernik, C. J., Bawden, M. J., Mercer, J. F., and May, B. K. (1988). J. Biol. Chem. 263, 5202-5209.

Staggs, T. M., and Perry, R. D. (1991). J. Bacteriol. 173, 417-425.

Stoebner, J. A., and Payne, S. M. (1994). Infect. lmmun. 56, 2891-2895.

Wee, S., Neilands, J. B., Bittner, M. L., Hemming, B. C., Haymore, B. C., and Seetharam, R. (1988). Biol. Metals 1, 62-68.

Wong, J. C. Y., Holland, J., Parsons, T., Smith, A., and Williams, P. (1994). Infect. lmmun. 62, 48-59.

Wong, J. Y. C., Patel, R., Kendall, D., Whitby, P. W., Smith, A. Holland, J., and Williams, P. (1995). Infect. Immun. 63, 2327-2333.

Zhang, J.-H., Kurtz, D. M., Jr., Xia, Y.-M., and Debrunner, P. G. (1991). Biochemistry 30, 583-589.

Zhang, L., Bermingham-McDonogh, O., Turcotte, B., and Guarente, L. (1993). Proc. Natl. Acad. Sci. USA 90, 2851-2855.

Documents

Heme binding by a bacterial repressor protein, the gene product of the ferric uptake regulation (fur) gene ofEscherichia coli