Retention of Heme in Axial Ligand Mutants of Succinate-Ubiquinone Oxidoreductase (Complex II) from Escherichia coli *

Succinate-ubiquinone oxidoreductase (SdhCDAB, complex II) from Escherichia coli is a four-subunit mem-brane-bound respiratory complex that catalyzes ubiquinone reduction by succinate. In the E. coli enzyme, heme b 556 is ligated between SdhC His 84 and SdhD His 71 . Con trary to a previous report (Vibat, C. R. T., Cecchini, G., Nakamura, K., Kita, K., and Gennis, R. B. (1998) Biochemistry 37, 4148–4159), we demonstrate the presence of heme in both SdhC H84L and SdhD H71Q mutants of SdhCDAB. EPR spectroscopy reveals the presence of low spin heme in the SdhC H84L (g z 5 2.92) mutant and high spin heme in the SdhD H71Q mutant (g 5 6.0). The presence of low spin heme in the SdhC H84L mutant suggests that the heme b 556 is able to pick up another ligand from the protein. CO binds to the reduced form of the mutants, indicating that it is able to displace one of the ligands to the low spin heme of the SdhC H84L mutant. The data

Succinate-ubiquinone oxidoreductases (SQR 1 ; succinate dehydrogenase, complex II) and menaquinol-fumarate reductase (QFR; fumarate reductase) are membrane-bound complexes that play critical roles in cellular metabolism in prokaryotic and eukaryotic organisms. The enzymes catalyze the reversible transfer of two electrons and two protons between succinatefumarate and the quinol-quinone couples (1); however, they normally are only expressed in aerobic (SQR) or anaerobic (QFR) environments (2,3). SQR directly connects the Krebs cycle with the aerobic respiratory chain by transferring reducing equivalents via quinone, whereas QFR is a terminal reductase in anaerobic respiration, where it oxidizes low potential quinols and reduces fumarate as the final electron acceptor (2, 4 -6).
Two distinct operons encode the subunits of SQR (sdhCDAB) (7,8) and QFR (frdABCD) (9,10) in Escherichia coli. X-ray structures for membrane-bound QFR from E. coli (11) and Wolinella succinogenes (12) have recently become available. Both QFR and SQR are composed of a conserved catalytic domain that consists of the two largest subunits (flavoprotein and iron-sulfur protein subunits). The flavoprotein subunit (SdhA, FrdA, or Fp) contains covalently bound FAD and the dicarboxylate binding site. The iron-sulfur protein subunit (SdhB, FrdB, or Ip) contains three distinct linearly arranged iron sulfur clusters ([2Fe-2S] 2ϩ,1ϩ , [4Fe-4S] 2ϩ,1ϩ , and [3Fe-4S] 1ϩ,0 ). The soluble dehydrogenase fragment (FpIp) binds to the hydrophobic anchor domain to form a membrane-bound complex (complex II) that is able to carry out electron transfer with quinone-quinol. The hydrophobic membrane anchor subunit pairs (SdhC, FrdC and SdhD, FrdD) are essential for forming the quinone binding sites and assembly of the whole complex (13)(14)(15)(16). The complex II membrane anchor subunits also coordinate one or two b-type hemes; however, E. coli QFR is a type of complex II that contains no heme (4). The single heme in E. coli SQR (17), as for the heme(s) in other complex IIs, has been shown to have bis-histidine axial ligation by EPR and near infrared magnetic circular dichroism (18,19), and from the x-ray structure (12). The complex II membrane anchor domains of different species share little similarity in amino acid sequence; however, the overall structure suggests a similar arrangement of the trans-membrane helices. A structural model for the membrane anchor domain of complex II has been proposed (20,21), and the recent high resolution structure of the diheme W. succinogenes QFR (12) is consistent with this model including the arrangement of the heme moieties. The redox properties of the heme b in complex II from different organisms also vary over a range of ϳ200 mV. The single heme in bovine heart complex II is not readily reducible by succinate due to its low redox potential at pH 7.0 (E m,7 ϭ Ϫ185 mV) (22). Bacillus subtilis complex II contains two b hemes, with the higher potential heme b H (E m,7 ϭ ϩ65 mV) reducible by succinate, whereas the lower potential heme b L (E m,7 ϭ Ϫ95 mV) is not (23). Succinate is able to reduce the single heme b 556 in E. coli SQR (E m,7 ϭ ϩ36 mV) (24). The varied presence and reducibility of the heme in complex IIs raises questions about whether catalysis is linked to the redox properties of cytochrome b. The heme, where present, has been shown to have an important role for proper assembly of complex II. In hemedeficient mutants of B. subtilis, the apocytochrome of SQR is made and inserted into the membrane, whereas the catalytic domain (FpIp) of the enzyme is accumulated in the cytoplasm (25). Similar results have also been reported for E. coli SQR when expressed in cells deficient in heme synthesis (26). The heme axial ligands for E. coli SQR have been shown to bridge the two membrane anchor subunits. SdhC His 84 and SdhD His 71 were identified as the heme ligands, and it was shown that succinate-quinone reductase activity was retained in the mutant enzymes despite the apparent absence of the heme b 556 (27). Herein, we present spectral and kinetic characterization of the SdhC His 84 and SdhD His 71 mutants of E. coli SQR that retain heme. The results show that an altered heme b does assemble in the isolated mutant enzyme although with significantly lowered redox potential. The data also show that the heme moiety is near the quinone binding domain.

MATERIALS AND METHODS
Bacterial Strains and Plasmids-E. coli strain DW35 (⌬frdABCD, sdhC::kan) and its recA derivative GV141 have been previously described (15,27). The deletion in the frd operon and the insertional mutation in the sdh operon eliminates strain background expression of any enzyme capable of succinate oxidase activity. Plasmid pSDH15 (sdhC ϩ D ϩ A ϩ B ϩ ) and the plasmid derivatives encoding mutations in the heme histidyl ligands (pSdhC H84L and pSdhD H71Q) have been described previously (27). Plasmids pFAS and pFGS (1) contain a frd promoter fusion to sdh (P FRD sdhC ϩ D ϩ A ϩ B ϩ ) such that SQR can be expressed anaerobically in E. coli with plasmid pFAS giving the highest expression level. In order to express the mutant enzyme anaerobically, the 0.57-kilobase pair EcoRI-KpnI fragment from pSdhC H84L or the fragment from pSdhD H71Q was inserted into the equivalent site in pFAS or pFGS to create plasmids pFAS or pFGS SdhC H84L (P FRD sdhC H84L D ϩ A ϩ B ϩ ) and pFAS or pFGS SdhD H71Q (P FRD sdhC ϩ D H71Q A ϩ B ϩ ).
Growth Conditions-Aerobically grown cultures were grown overnight in Luria-Bertani (LB) medium with appropriate antibiotics, and 150 ml was used as inoculum for a 10-liter fermentor (New Brunswick Scientific, Edison, NJ) containing the medium previously described (1) supplemented with 0.05% (w/v) casamino acids, 0.2% (w/v) tryptone, 0.1% (w/v) yeast extract, 0.2 mM MgCl 2 , 5 M CaCl 2 , 20 g/ml Fe 2 (SO 4 ) 3 , and 50 mM sodium succinate. Ampicillin (35 g/ml) and kanamycin (50 g/ml) were included in all media. Cultures were grown with high aeration and harvested at late exponential phase. Anaerobically grown cultures were grown overnight in the same medium described above (minus succinate) with 50 mM glycerol and 50 mM fumarate as electron donor and acceptor, respectively (1). The heme-deficient strain, E. coli SASX41B (HfrP02A hemA41 metB1 relA1), a ␦-aminolevulenic acid (ALA) auxotroph (28), was transformed with appropriate plasmids and grown in 50 ml of LB medium in the presence of ampicillin (50 g/ml) and ALA (50 g/ml). Cells were collected by centrifugation (10 min at 500 ϫ g) and gently resuspended in 20 ml of LB medium without ALA, and 1 ml was used to inoculate 1 liter of LB medium. Cultures were then grown aerobically in LB with ampicillin (35 g/ml) or anaerobically in the same medium with 50 mM glucose. When necessary, the medium was supplemented with 100 M ALA.
Preparation of Membrane Fraction and Enzyme Purification-Cells were collected by centrifugation, and the membrane fraction enriched in SQR was isolated as previously described (1) with the exception that the cells were disrupted by one passage with an EmulsiFlex-C5 homogenizer (Avestin, Inc., Ottawa, Ontario, Canada) at 18,000 p.s.i. at 4°C. Membranes containing wild type or mutant SQR were resuspended in 50 mM potassium phosphate, 0.2 mM EDTA (pH 7.2) to ϳ30 mg of protein/ml and frozen at Ϫ70°C. To purify wild type and mutant SQR, the membranes were extracted with 2% (w/v) of the nonionic detergent Thesit (polyoxyethylene 9-dodecyl ether) (Roche Molecular Biochemicals) as previously described (1). The solubilized extract was then applied to a HiLoad 26/10 Q-Sepharose Fast Flow column (Amersham Pharmacia Biotech) and eluted using a 600-ml linear gradient of 0.1-0.25 M NaCl in 10 mM potassium phosphate (pH 7.2) with 0.05% Thesit according to published procedures (24,27). The brownish fractions containing succinate dehydrogenase activity were pooled and concentrated with Centriprep-30 concentrators (Amicon Inc., Beverly, MA). The enzyme was washed with 100 mM potassium phosphate (pH 7.0), and reconcentrated to 15-20 mg of protein/ml and stored at Ϫ70°C.
Measurement of Enzyme Activity-Activity measurements for the succinate oxidase reaction were measured in the presence of 10 mM succinate in the assay cuvette as previously described (29). The succinate-phenazine ethosulfate (PES) reaction in the presence of dichlorophenolindophenol (DCIP) (⑀ 600 ϭ 21.8 mM Ϫ1 cm Ϫ1 , pH 7.8) was measured with 1.5 mM PES and 50 M DCIP. To measure kinetic parameters of succinate-quinone reductase activity of wild type and mutant SQR, 30 M Wurster's Blue (⑀ 612 ϭ 12 mM Ϫ1 cm Ϫ1 ) was used as final electron acceptor with varied amounts of quinone as previously described (29,30).
Spectrophotometric Measurements-Absorption spectra were recorded at 25°C with a Hewlett Packard 8453 diode array spectrophotometer (Palo Alto, CA) in a 2-ml anaerobic cuvette. The spectrum and concentration of cytochrome b 556 attributed to purified and membranebound SQR enzymes was determined as previously described (1,24,27). Spectra were routinely recorded of membranes suspended in 50 mM potassium phosphate (pH 7.0), 0.2 mM EDTA at a protein concentration of 0.25 mg/ml. Anaerobiosis was achieved by vacuum evaporation and saturation of the buffer with oxygen-free argon.
EPR Spectroscopy and Redox Potentiometry-EPR spectra were recorded using a Bruker ESP300 spectrometer equipped with an Oxford Instruments ESR-900 flowing helium cryostat. Samples were prepared as described in Fig. 4. Potentiometric titrations were carried out as previously described with 150-l samples being extracted from the tritrations into 3-mm internal diameter quartz EPR tubes (31,32). Titrations were carried out on membranes enriched in wild-type and mutant enzymes at a protein concentration of ϳ30 mg/ml in 100 mM MOPS and 5 mM EDTA (pH 7.0). EPR spectra were recorded as described in Fig. 6 legend.
Analytical Methods-Protein content in membranes was determined by the Biuret method and in isolated enzymes by the method of Lowry in the presence of 1% (w/v) SDS with bovine serum albumin as a standard. The protoheme IX content of cytochrome b was determined from the pyridine hemochromogen difference spectra (dithionite-reduced minus oxidized) (⑀ 558 -540 ϭ 23.98 mM Ϫ1 cm Ϫ1 ) as described (33). The histidyl-flavin concentration in purified SQR enzymes was determined as follows. Purified protein (0.5-0.7 mg) was precipitated with 1 ml of cold acetic acetone (8 l of 6 M HCl per 1 ml of acetone) to remove protoheme IX and the iron-sulfur clusters and then centrifuged for 30 s in a microcentrifuge. The yellowish pellet was washed three more times with the same volume of acidic acetone and suspended in 0.8 ml of 0.1 M sodium phosphate (pH 7.0) with 1% (w/v) SDS, and the precipitated protein was solubilized after 2 h at 38°C. The spectrum of the resulting solution shows two peaks at 354 and 445 nm attributed to histidylriboflavin. The covalent flavin concentration was determined using ⑀ 445 ϭ 12.0 mM Ϫ1 cm Ϫ1 for histidyl-riboflavin (34).

RESULTS
Anaerobic Expression of SQR Mutants-It has been previously shown that aerobic overexpression of SQR can be achieved in E. coli from plasmids that encode wild type SQR (24). Nevertheless, it has been shown that anaerobic expression of SQR driven from the frd promoter (P FRD ) enables even higher levels of SQR to be produced in the membranes of E. coli (1). Therefore, to facilitate expression of site-directed mutant forms of SQR and to aid in purification of the enzymes, constructs were cloned into plasmid pFGS so expression could be driven by the P FRD promoter. E. coli strain DW35, when transformed with pFGS, is capable of growth under anoxic conditions on glycerol-fumarate minimal medium, indicating that wild type SQR can replace fumarate reductase in the anaerobic respiratory chain (1). It has been shown that E. coli GV141 expressing SdhC H84L or SdhD H71Q mutant enzyme is able to grow aerobically on succinate minimal medium, indicating a functional complex II is formed (27). To test whether anaerobic growth is possible with these mutant SQR enzymes, DW35 containing either pFGS SdhC H84L or pFGS SdhD H71Q was grown anaerobically on glycerol-fumarate minimal medium as previously described (1). Both mutants supported anaerobic growth on glycerol-fumarate minimal medium in E. coli DW35 with a doubling time of 3.8 h for pFGS SdhC H84L and 3.2 h for pFGS SdhD H71Q as compared with 3.0 h for wild-type SQR (data not shown). These results indicate that a functional complex competent in catalysis for the menaquinol-fumarate reductase reaction in vivo is expressed from both the wild-type and mutant plasmids.
Properties of Isolated Mutant Membranes-Previous studies with aerobically grown E. coli cells encoding the SdhC H84L and SdhD H71Q SQR mutants suggested that these substitutions resulted in formation of catalytically active membraneassociated complexes that lacked heme (27). Therefore, it was surprising that membranes isolated from anaerobically grown DW35 cells transformed with both mutants plasmids had an intense color. Membranes from SdhC H84L were brownish red in color, similar to those from cells transformed with wild type SQR plasmids. The membrane fraction from the SdhD H71Q mutant was brownish green in color. The absorption spectra of membranes from strain DW35 enriched in wild-type or mutant SQR complex show a significant absorbance at the Soret region compared with membranes obtained from an E. coli control strain (MC4100) transformed with pBR322. MC4100, the parent strain of DW35 (15), contains chromosomal copies of the sdh and frd operons and under the anaerobic conditions used for growth does not express SQR. Under the anoxic conditions used for growth, the heme containing bd-oxidase is expressed (35) and contributes somewhat to the absorbance at the Soret region in the membrane fraction. Fig. 1 shows the dithionitereduced minus air-oxidized difference spectra of membranes enriched in wild-type and mutant SQR complexes as well as membranes from anaerobically grown cells of E. coli MC4100. The spectra show a significant ␣-absorption at 558 nm and a broader ␤-absorption between 526 and 528 nm as well as the Soret absorption (ϳ425 nm) characteristic of room temperature difference spectra for cytochrome b 556 from SQR (36). Analysis of protoheme IX content of the membranes by the pyridine hemochromogen method showed similar cytochrome b concentrations for wild-type SQR and both mutant SQRs (Table I). Mutant forms, however, show lower amplitudes at the Soret absorption and also changes in the line shape of the spectrum (Fig. 1). Both mutants demonstrate succinate-PES reductase and succinate-Q 1 reductase activity; however, SdhC H84L showed at least two times lower succinate-PES reductase activity than the wild-type or SdhD H71Q mutant membranes. Moreover, the succinate-quinone reductase activity was even lower in the SdhC mutant, and the ratio of Q 1 /PES activities indicated that this mutant is significantly impaired in its ability to interact with quinones.
Spectral Properties of Isolated SdhC H84L and SdhD H71Q Mutants-Previous data had suggested that the SdhD H71Q SQR complex was less stable during purification. In the present studies, however (using slight modifications of the original protocol; see "Materials and Methods"), the chromatographic profiles for both SQR mutants and the wild-type enzyme were identical. A single brownish peak with succinate dehydrogenase activity appears at the end of the 0.1-0.25 M NaCl gradient (data not shown). On the basis of SDS-PAGE analysis, the purity of the wild-type and mutant SQRs also appear identical (data not shown). Table II indicates the protoheme IX content of the isolated SQR enzymes. Comparing the ratio of covalent FAD to protoheme IX content of the purified enzymes, it ap-pears that the SQR mutants show a 10 -15% deficiency in protoheme IX compared with the ratio of wild-type enzyme.
The absorption spectra of the wild-type and mutant SQR enzymes isolated from the anaerobically grown E. coli cells are shown in Fig. 2. The oxidized cytochrome in wild-type SQR shows a broad absorption at the ␣and ␤-regions with the Soret absorption at 412 nm. Incubation of wild-type enzyme with sodium dithionite reduces the cytochrome completely within half a minute, and an ␣ absorption at 558 nm and a broad ␤ absorption at 528 nm appear along with a sharp symmetrical Soret absorption at 424 nm. The air-oxidized spectrum for the SdhC H84L mutant SQR shows a broad absorption at 540 and 580 nm unlike wild-type SQR; however, the Soret absorption shows a maximum at 411 nm with similar intensity to the wild-type cytochrome b 556 . Complete reduction of the cytochrome by dithionite in this mutant takes 4 -5 min. The reduced enzyme displays an ␣ absorption at 559 nm and a broad ␤ absorption at 528 nm. The Soret absorption in the SdhC H84L mutant enzyme exhibits a maximum at 426 nm and a discernible shoulder at 445 nm, and its absorption intensity is some 2-fold lower compared with wild-type SQR (Fig. 2). The SdhD H71Q mutant SQR differs noticeably in color from wild type or the SdhC H84L mutant SQR; it is less reddish and more green-brown. As shown in Fig. 2, the oxidized spectrum of the SdhD H71Q mutant SQR shows no pronounced peak in either FIG. 1. Light absorption difference (dithionite-reduced minus air-oxidized) room temperature spectra of membranes from E. coli strains grown under anaerobic conditions. E. coli strain GV141 was transformed with wild type and mutant SQR-encoding plasmids, and control strain MC4100 was transformed with pBR322 and grown anaerobically under the same conditions. The spectra were determined in a 2-ml cuvette containing 1 mg of membrane protein/ml in 50 mM potassium phosphate (pH 7.2), and then 1-2 mg of solid sodium dithionite was added to reduce the membrane suspension. The horizontal line for each spectrum indicates zero absorbance for the individual spectra, with the absorbance units shown on the vertical axis on the left. the ␣ or ␤ regions, whereas the Soret absorption is similar to wild-type and the SdhC mutant SQR, although the maximum is shifted to 407 nm. The SdhD H71Q enzyme could be slowly reduced with dithionite, similar to the results with the SdhC mutant. The spectrum of the reduced SdhD mutant shown in Fig. 2 shows an ␣ absorption at 560 nm, a broad absorption at the ␤ region, and a Soret absorption at 423 nm similar in intensity to that of SdhC H84L.
Reduced cytochrome b 556 from E. coli SQR or the isolated SdhCD domain does not react with carbon monoxide (36), typical of low spin hexacoordinated hemes. As shown in Fig. 3, carbon monoxide causes noticeable alterations in the absorption spectra of both the SdhC and SdhD mutants. The Soret absorption is shifted to 422 nm in SdhC H84L and 423 nm in the SdhD H71Q mutant with a comparable increase in the absorption intensity and a more symmetrical shape to the Soret absorption. The ␣ absorption was unaffected in the SdhC mutant, whereas a decrease in absorption intensity is found in the SdhD mutant similar to that seen in the isolated cytochrome domain of beef heart succinate dehydrogenase (22). One interpretation of this data is that both mutations result in a change of ligation of the heme b 556 from hexa-to pentacoordinate (viz., from low to high spin).
In order to further investigate potential spin state changes elicited by the SdhC H84L and SdhD H71Q mutations, we subjected oxidized membranes enriched in these mutant enzymes to EPR analysis. Fig. 4A shows EPR spectra around g ϭ 2 that arise primarily from [3Fe-4S] clusters. Comparison of the spectrum shown in Fig. 4A (i) (background strain, DW35) and those of Fig. 4A (ii-iv) (overexpressing wild-type and mutant enzyme) indicates that high levels of the SdhB [3Fe-4S] cluster can be detected in the overexpressed wild-type and mutant enzymes. Fig. 4B shows equivalent spectra recorded around g ϭ 6. Noticeable in the spectrum of the background strain ( Fig. 4B (i)) is a typical high spin heme spectrum that probably arises from pentacoordinate hemes such as those found in cytochromes bo 3 (37) and bd (28). In membranes containing overexpressed wild-type SQR (Fig. 4B (ii)), there is a diminution of the g ϭ 6 signal compared with that observed in the background strain. This is likely to be due to the dilution of the proteins responsible for the background signal of Fig. 4B (i) by the overexpressed wild-type SdhCDAB, which contains no high spin heme. The spectrum of membranes enriched in SdhC H84L (Fig. 4B (iii)) is essentially identical to that of membranes containing wild-type SdhCDAB. The spectrum of membranes enriched in SdhD H71Q, however, has an intense signal at g ϭ 6.0, indicative of the presence of elevated amounts of pentacoordinate heme. Given that wild-type SdhCDAB contains hexacoordinate heme b 556 , it is likely that the intense g ϭ 6 signal arises from the loss of one of the histidine imidazole ligands of this heme, in agreement with the optical data presented herein. Given that no significant increase in g ϭ 6.0 signal intensity is observed in the SdhC H84L mutant, we also looked for low spin heme spectra in samples containing wildtype and mutant enzymes (Fig. 4C). Noticeable in the spectra of membranes lacking overexpressed enzyme (E. coli DW35, Fig.  4C (i)) is a peak at g z ϭ 3.3 similar to that assigned to heme b 558 observed in spectra of membranes containing the cytochrome bd ubiquinol oxidase (28,38). A broad peak is observed between g ϭ 3.65 and g ϭ 3.50 in spectra of membranes containing overexpressed wild-type enzyme (Fig. 4C (ii)). This feature appears to be essentially identical to the spectrum reported for the g z feature of low spin heme b 556 in purified E. coli SQR (36). The spectrum of the SdhC H84L mutant lacks the g ϭ 3.33 and g ϭ 3.76 -3.50 features and instead contains a distinct peak at g z ϭ 2.92, indicating that in this mutant the heme remains low spin in its oxidized state but has a significantly altered environment compared with the wild type.
Kinetic Properties of Isolated Enzymes-The ratio of quinone reductase activity to that with artificial electron acceptors such as PES and DCIP has been shown to indicate the ability of complex II to interact with quinones (2). Both mutants showed catalytic activity with PES and Q 1 , although the SdhC H84L mutant showed a 4-fold lower turnover number in its ability to reduce Q 1 (Table II). There was also an increase in the K m for Q 1 in the SdhC H84L mutant, whereas the K m was similar to wild-type enzyme for the SdhD H71Q mutant. Interestingly, the quinone site competitive inhibitor pentachlorophenol (PCP) (29) showed an increased K i for the SdhC mutant, and there was no change of the K i for the SdhD mutant.
The diheme cytochrome b of the B. subtilis SQR complex has its absorption spectrum perturbed by the addition of the quinone site inhibitor 2-n-heptyl 4-hydroxyquinoline-N-oxide (HQNO) (39). Although HQNO is a potent inhibitor of B. subtilis SQR and E. coli QFR, it does not inhibit E. coli SQR (29). The effects of quinone site inhibitors on the absorption spectra of E. coli SQR cytochrome b have not been reported, so it was of interest to determine if an inhibitor like PCP affected the   (36). When incubated with PCP, the isolated wild-type SdhCD subunits show differ-ence spectra identical to that seen in the intact complex (Fig.  5). The similar spectral changes in both the SQR complex and the isolated SdhCD subunits caused by PCP suggest that the inhibitor binds near to the cytochrome b in SQR. The effect of PCP on the mutant forms of SQR was also examined. The dithionite-reduced difference spectrum of the SdhD H71Q enzyme incubated with PCP is similar to that seen for the wildtype enzyme. The Soret absorption, however, shows no detectable maximum, and min ϭ 425 nm, maximum (⌬⑀ ϭ 8.7 mM Ϫ1 cm Ϫ1 ). There was only a very slight ␤ absorption, but the ␣ absorption was blue-shifted like wild type with a min ϭ 562 nm (⌬⑀ ϭ 1.9 mM Ϫ1 cm Ϫ1 ). It was not possible to reliably determine the K d PCP for the SdhD H71Q enzyme due to the instability of the mutant enzymes in the reduced state (see below). By contrast to wild-type and SdhD H71Q SQR, PCP had very little effect on the difference spectrum of the SdhC H84L enzyme (Fig. 5). There was a very broad absorbance with a minimum at 425 nm and a minor absorbance change at 560 -565 nm. These data along with the changes in K i and K m (Table II) suggest that the association of PCP with the enzyme has been affected in SdhC H84L.
Redox Properties of the Heme-The favorable redox potential of cytochrome b 556 of wild type E. coli SQR (E m ϭ ϩ36 mV) allows complete reduction by succinate under anaerobic conditions (36). By contrast, the cytochrome in the mutant SQRs investigated in this study are only partially reduced by succinate under anaerobic conditions. The SdhC H84L heme is only reduced some 20% (compared with dithionite reduction) after 12 min (pH 7.2, 25°C), and the SdhD H71Q mutant is reduced about 30% using the same conditions (data not shown). Overall, these results suggest that there is a thermodynamic or kinetic FIG. 2. Visible light absorption spectra of purified wild-type and mutant SQR complex. The oxidized spectrum (as isolated) trace is shown by the dashed line and the reduced enzyme is shown by the solid line. The spectra of the oxidized enzymes were scanned at 25°C in a 1-ml cuvette in 50 mM potassium phosphate, 0.05% (w/v) Thesit at a protein concentration of 2.65 M heme b. The dithionite-reduced (solid line) spectra were recorded following reduction of the enzyme with 1-2 mg of solid sodium dithionite after incubation for 5 min. The ␣ and ␤ absorbance (500 -700 nm) on the right is shown at 3 times the gain of the Soret absorbance (380 -500 nm).

FIG. 3. Effect of carbon monoxide on the visible absorption spectra of purified SdhC H84L and SdhD H71Q SQR enzymes.
The dithionite-reduced (solid line) and dithionite-reduced with carbon monoxide (dashed line) visible absorption spectra are shown. Conditions are the same as in Fig. 2 with the cuvette flushed with carbon monoxide gas for 2 min. CO has no effect on the wild type with the spectrum identical to that seen in Fig. 2 (top panel). block in electron transfer to the heme in both mutants. One possible explanation for this is that the E m,7 of the heme is significantly lowered in these mutants.
In order to determine if the heme present in the two mutants does have a lower midpoint potential (E m,7 ) than the low spin heme of the wild-type enzyme, we subjected the two mutant and wild-type enzymes to potentiometric analysis in combination with EPR spectroscopy. Fig. 6A shows that the high spin heme signal of membranes containing wild-type SdhCDAB titrates with an E m,7 of Ն210 mV, consistent with this signal arising from high potential pentacoordinate hemes present in either cytochrome bo 3 (37) or cytochrome bd (28). The g ϭ 6 signal from membranes enriched in SdhD H71Q titrates as a single species with an E m,7 of approximately Ϫ97 mV. Potentiometric titration of the g ϭ 2.92 signal of the SdhC H84L mutant reveals that its E m,7 is approximately ϩ15 mV. Similarly, analysis of the g ϭ 6.0 signal of membranes enriched in this mutant reveals that it titrates with two components, one major one at E m,7 ϭ Ն210 mV and a minor one at E m,7 ϭ ϩ15 mV. However, given the low concentration of high spin heme in membranes enriched in this mutant, it is unlikely that the high spin E m,7 ϭ ϩ15 mV component contributes significantly to the analyses reported herein (cf. Figs. 4B and 6B). Overall, the data suggest that the heme b 556 is overwhelmingly in a low spin hexacoordinate state in membranes enriched in the SdhC H84L mutant.
In order to investigate the possibility that changes in the coordination of heme b 556 have any effect on the properties of the [3Fe-4S] cluster located in SdhB, we also determined the E m,7 of the [3Fe-4S] cluster in membranes enriched in wildtype, SdhC H84L, and SdhD H71Q SQR. The E m,7 values for the [3Fe-4S] cluster were determined to be ϩ75, ϩ65, and ϩ83 mV for the wild type and SdhC H84L and SdhD H71Q mutants, respectively (data not shown). These values are in reasonable agreement with those previously reported for the SdhB [3Fe-4S] cluster (E m ϭ ϩ65 mV (41)).
Stability of the Mutant Enzymes-Both the SdhC H84L and SdhD H71Q mutant enzymes in the oxidized state and neutral pH remain catalytically active for several days at 4°C; however, incubation at 30°C and pH 7.8 results in inactivation of both mutants (Fig. 7A). As seen in Fig. 7, the quinone reductase activity is lost at about twice the rate of the succinate oxidase activity. The succinate oxidase activity measured with PES/ DCIP decreased with higher pH and temperature. Incubation of the SdhD H71Q mutant enzyme with 10 mM succinate increased the rate of inactivation of the quinone reductase activity some 10-fold; however, succinate oxidase activity was affected to a lesser extent (Fig. 7B). The inset in Fig. 7B shows the amplitude of the dithionite-reduced signal attributable to the heme during incubation of the SdhD H71Q enzyme with succinate. The decrease of the signal of the reduced heme b 556 , but not its spectral nature, indicates the release of protoheme IX from apocytochrome. The rate of release of heme b is more rapid in aerobic than anaerobically incubated enzyme, and the loss of heme directly correlates with the decrease of succinate-Q 1 reductase activity. The correlation of the rapid loss of quinone reductase activity with the decrease in the spectral signal for cytochrome b 556 suggests that the dissociation of SdhAB caused the loss of protoheme IX from SdhCD.
Effect of Growth Conditions on Heme Assembly-The data reported above indicates that pentacoordinated heme is assembled in the SdhC H84L and SdhD H71Q mutants when grown anaerobically in strain DW35 or GV141. Previous studies using FIG. 4. [3Fe-4S] cluster EPR spectra (A), high spin heme spectra (B), and low spin heme spectra (C), of membranes enriched in SdhCDAB, SdhC H84L DAB, and SdhCD H71Q AB. Illustrated are EPR spectra of membranes containing no SdhCDAB (i) (E. coli DW35), SdhCDAB (ii), SdhC H84L DAB (iii), and SdhCD H71Q AB (iv). Samples were prepared in 100 mM MOPS and 5 mM EDTA (pH 7.0). Oxidation was achieved by incubation in the presence of 0.8 mM dichlorophenolindophenol for 5 min. Spectra were normalized to a protein concentration of 30 mg of protein per ml and were recorded under the following conditions: temperature, 12 K; microwave power, 20 milliwatts at 9.47 GHz; modulation amplitude, 10 Gpp. For A and B, only a single scan was necessary; five scans were accumulated for C. these same mutants expressed in GV141 had suggested that heme was not assembled in the mutants, although SQR assembled in the membrane and was functionally active (27). In these studies, SQR was expressed aerobically, conditions that are different from those in the current paper. Therefore, wild-type SQR and the SdhC H84L and SdhD H71Q mutant enzymes were expressed aerobically in minimal medium with succinate in strain GV141 or DW35. The content of cytochrome b in membranes isolated from the aerobically grown cells was 1.1 nmol/mg of protein for the SdhC H84L mutant and 1.5 nmol/mg of protein for the SdhD H71Q mutant for strain GV141 (data not shown). This is less than half the amount found in anaerobically grown GV141 used in the current studies (see Table I). Even higher levels of heme were found in the membranes from aerobically grown DW35 containing the mutant plasmids, consistent with previous results showing higher expression levels for SQR in this strain (1). The enzyme isolated from membranes of aerobically grown cells was found to have identical catalytic properties, protoheme IX content, and spectral and redox characteristics as to the enzyme from anaerobically grown cells (data not shown).
Previous studies with E. coli SQR had shown that assembly of the enzyme was perturbed when the sdhCDAB genes were introduced into a heme synthesis mutant (26), suggesting the importance of heme in the assembly of complex II. These results were consistent with those found for B. subtilis complex II (25). The data suggesting that a heme did not assemble in E. coli SQR in the SdhC H84L and SdhD H71Q mutants but that a functional enzyme complex was formed (27) were thus not consistent with the results using heme-deficient mutants. To reinvestigate this question in light of the above results, a heme-deficient strain of E. coli (SASX41B) (42) was transformed with the SdhC H84L-and SdhD H71Q-encoding plasmids. The strain is unable to grow aerobically on LB medium unless supplemented with ALA or with a fermentable substrate such as glucose. In agreement with the results reported by Nakamura et al. (26), it was found that SQR could not assemble in the membrane in aerobically grown cells (data not shown). This was found for either wild-type or mutant forms of SQR unless the medium was supplemented with ALA. Anaerobic growth of SQR using the P FRD promoter results in a higher yield of membrane bound SQR (1), so the effect of anaerobic growth in the heme-deficient strain was investigated. The same results were found; i.e. unless E. coli SASX41B grown anaerobically with glucose was supplemented with ALA, no assembled SQR was found in the membranes of the cell based on catalytic activity using either wild-type or mutant SQR plasmids for expression of the enzyme. (Anaerobically, fumarate reductase is expressed from the chromosomal copy of the frd operon in E. coli SASX41B. The fumarate reductase activity in the membrane can be discriminated from the succinate quinone reductase activity of SQR based on the sensitivity of fumarate reductase to HQNO, an inhibitor to which SQR is insensitive (29).) These results are consistent with the requirement of heme for assembly of SQR and based on the data presented in Tables I and II and Figs. 2, 3, and 5 indicating that heme is assembled in the SdhC H84L and SdhD H71Q mutants.

DISCUSSION
The role of the heme in SQR has yet to be established. It has been suggested that the heme has a structural and/or catalytic role in the function of Complex II (4,21). The structural role for the heme in B. subtilis SQR has been well documented (4,25), and the majority of available evidence has also suggested such a structural role for E. coli SQR (26). The recent crystal structure of the diheme W. succinogenes QFR shows that heme b H (equivalent to the single heme of E. coli SQR) has amino acid side chains from four of the five ␣-helices in the membrane domain that aid in binding the heme (12). This suggests the importance of the heme in the assembly and structure of the complex. The results obtained with mutants of the histidyl ligands of the heme in E. coli SQR that suggested that the enzyme assembled and functioned in the membrane, in the absence of heme, were therefore not entirely consistent with the other available data (27). The results reported in this study show that heme is indeed assembled in the SdhC H84L and SdhD H71Q enzymes. In the case of the SdhC H84L mutant, the heme appears to be retained in a hexacoordinate low spin form, with a g z ϭ 2.92, whereas in the SdhD H71Q mutant, the heme appears to become pentacoordinate and high spin with a g xy at 6.0. For the latter mutant, the E m,7 is also significantly lowered to approximately Ϫ97 mV. In the case of the SdhD H71Q mutant, it is not surprising that the high spin heme is able to bind CO. However, CO binding also occurs in the SdhC H84L mutant, indicating that it is able to displace the ligand that presumably replaces the imidazole nitrogen of His 84 . The results reported here are in agreement with those of others that show that the heme (26), when present, is important for proper assembly of complex II.
The retention of low spin heme in the SdhC H84L mutant suggests that His 84 might not be the heme ligand or that another residue within SQR can serve as an axial ligand for heme b 556 . Given the results of sequence alignments of complex FIG. 6. Potentiometric titrations of high and low spin signals attributable to wild-type and mutant enzyme. A, comparison of potentiometric titrations of the g ϭ 6.0 high spin signals from membranes enriched in wild-type and SdhD H71Q enzyme. ‚, titration of the g ϭ 6.0 (high spin heme) from membranes enriched in the SdhD H71Q mutant (E m,7 ϭ Ϫ97 mV); Ⅺ, titration of the g ϭ 6.0 signal from membranes enriched in wild-type enzyme (E m,7 Ն ϩ210 mV). B, comparison of potentiometric titrations of the g ϭ 2.92 low spin (‚; E m,7 ϭ ϩ15 mV) and the g ϭ 6.0 high spin (Ⅺ; E h ϭ ϩ15 mV (33%) and Ն210 mV (67%)) signals of membranes enriched with the SdhC H84L mutant. Data were obtained from EPR spectra recorded as described in the legend to Fig. 4, except that the microwave power used was 2 milliwatts, and three scans were accumulated for titration of the g ϭ 2.92 signal.
II from many sources (4,20) and the high resolution structure of the heme containing W. succinogenes QFR (12), it is unlikely, however, that an SdhC residue other than SdhC His 84 provides one of the ligands to heme b 556 in the wild-type enzyme. This conclusion is also supported by the site-directed mutagenesis studies of E. coli SQR, where substitution of histidyl residues in the membrane binding domain suggests that SdhC His 84 and SdhD His 71 are the axial ligands to the heme (16,27). An example of the swapping of the heme axial ligands has been observed in both beef (43) and E. coli SQR (26), when the small membrane anchor subunit QPs3 or SdhD respectively, is expressed alone. In these examples, it appears that the histidyl ligands come from two different molecules of the respective small subunit (26,43). In this study, the observation of a g z at 2.92 (Fig. 4) is also typical for a heme with bis-histidine ligation with a small interplanal angle between the planes of the ligating imidazoles (44,45). This suggests that the replacement ligand is another residue, possibly His 91 or His 30 from SdhC or alternatively His 14 of SdhD. Future site-directed mutagenesis experiments will address this question. Such ligand displacement has also been observed in the CO-sensing CooA protein from Rhodospirilium rubrum (46).
The reason for the discrepancy between the results reported here and the previous results (27) with the SdhC His 84 and SdhD His 71 mutants is not entirely clear. In the case of the SdhD H71Q mutant, the succinate-reduced Ϫ TMPD/ascorbate-reduced difference spectrum would not be expected to show the presence of the heme because, as the present studies document, the heme is of much lower potential. In the case of the SdhC H84L mutant, there may be a blockage in electron transfer from the [3Fe-4S] cluster to the heme, possibly as a result of disruption of the quinone binding site. More difficult to explain is the reported inability to detect heme extracted from purified SdhC H84L and SdhD H71Q mutant SQR (27). As shown in this paper (Fig. 7) and as noted in the previous study (27) the mutant SQR complexes are less stable than wild type and are particularly sensitive to temperature and aeration. Therefore, it is conceivable that in the previous study the heme could have been lost during the purification procedure, resulting in the inability to detect it in the final samples obtained from the chromatographic column. Nevertheless, as shown in the current study, heme is present in the purified mutant enzymes whether it is produced from anaerobically or aerobically grown E. coli.
The results in Table II and Fig. 5 show that PCP affects the ability of SQR to interact with quinones and perturbs the heme environment. In Table II, it can also be seen that the SdhC H84L enzyme is much more severely affected in its ability to interact with Q 1 than is the SdhD H71Q mutant. Also, the inhibitor PCP shows a 5-fold increase in its K i in the SdhC H84L mutant, whereas the SdhD H71Q enzyme shows no change in inhibition. These results are consistent with SdhC His84 being part of the quinone binding site, in addition to being a ligand of the b heme of SQR. Alternatively, the apparent ligand displacement observed might result in the disruption of the quinone binding site being a secondary effect; i.e. whichever residue replaces SdhC H84L may in fact be essential for ubiquinone binding and oxidation. Azidoquinones have been used to label the SdhC subunit of SQR and have implicated Ser 27 and Arg 31 of SdhC as being part of a quinone binding site in the enzyme (16). Although the primary sequence of the membrane anchor subunits of complex IIs are not highly conserved, the available structures and models (11,12,20,21) all suggest a very similar transmembrane topology. In two subunit membrane anchors, like E. coli SQR, this would place SdhC His 84 in helix II on the cytoplasmic membrane face and on the opposite side of a pocket from SdhC Ser 27 and SdhC Arg 31 as previously suggested (16). The structure of QFR from E. coli shows two quinone binding pockets on the opposite side of the membrane (11), and SdhC His 84 would be localized near the Q p (quinone-proximal) binding site. PCP perturbs the heme environment in E. coli SQR similar to that seen with HQNO in B. subtilis SQR, and it has been suggested for B. subtilis SQR that the cytochrome participates in binding and stabilization of the semiquinone generated during electron transfer in the enzyme (39). The semiquinone radical attributed to the Q p site demonstrates rapid relaxation behavior during EPR analysis, and this has been attributed to a relaxation pathway involving the heme and the [3Fe-4S] center of the enzyme, suggesting FIG. 7. Kinetics of inactivation of SdhC H84L and SdhD H71Q mutants and wild type SQR during incubation at 30°C at pH 7.8. Purified wild-type and mutant SQRs (at a concentration of 50 M enzyme based on cytochrome b content) were activated with 10 mM malonate at pH 7.0 as described (29) and then diluted with 50 mM potassium phosphate (pH 7.8) to 2 M concentration of b 556 and incubated at 30°C in the absence (Fig. 7A) or presence (Fig. 7B) of 10 mM succinate. A, succinate-Q 1 reductase activity (E and q) and succinate-PES reductase activity (Ⅺ and f) of purified mutant SdhC H84L (open symbols) and SdhD H71Q (closed symbols) enzymes. Succinate-Q 1 reductase activity for control wild-type SQR purified enzyme (ࡗ) is also shown in A. B, the affect of anaerobic versus aerobic incubation on the SdhD H71Q mutant purified enzyme. SdhD H71Q was incubated with 10 mM succinate aerobically (filled symbols) and anaerobically after saturation of the enzyme incubation mixture with argon (open symbols). Succinate-Q 1 reductase activity for SdhD H71Q (E and q) and succinate-PES reductase activity (f) are shown. Inset, the effect of 10 mM succinate on the dissociation of the heme b from SdhD H71Q during incubation at 30°C under aerobic (E) and anaerobic (q) conditions. The amplitude of the difference of the dithionite-reduced spectra at 559 -575 nm shows the loss of the heme from the enzyme. that they are near to the Q p site (47). The need for proper assembly of the heme to maintain the integrity of the quinone binding sites is also supported by the data in Fig. 7 showing that alteration of the histidyl ligands affects the stability of the enzyme. The more rapid loss of quinone reductase activity, as compared with succinate-PES reductase activity, suggests that if the heme is lost upon incubation of the mutant enzymes then the catalytic SdhAB subunits dissociate from the enzyme. Taken together, these data are consistent with Q p being located near the [3Fe-4S] cluster in the SdhB subunit and near the SdhC His 84 histidyl ligand of cytochrome b of SQR.
These studies do not directly address the role of the heme in electron transfer in complex II. QFR from E. coli catalyzes with similar efficiency the same reactions as SQR and does so in the absence of heme. This suggests that heme is not essential for electron transfer, but the results presented demonstrate that heme is present in SdhC H84L and SdhD H71Q mutants and that the mutations also have an effect on quinone reductase activity. This leaves open the possibility that heme is directly involved in reaction with quinone/quinol. A mutation in the human SdhD gene equivalent to the E. coli SdhD His 71 residue results in hereditary paraganglioma (48). As shown in Fig. 6, such a mutation may drop the redox potential of the heme b by more than 100 mV. If SQR heme acts as a sensor for the redox state of the quinone pool or as an oxygen sensor, the hypoxic phenotype observed (48) would be consistent with a lowered potential of human heme b. Structural information for SQR and further characterization of the cytochrome b should help resolve these issues.