Rothery, Department of Biochemistry, 474 Medical Sciences Building, University of Alberta, Edmonton, Alberta T6G 2H7 Fax: +1 780 492 0886 Tel: +1 780 492 2229 E-mail: Richard.Rothery@UAl
Trang 1by site-directed mutagenesis, fluorescence quench
titrations and EPR spectroscopy
Richard A Rothery1, Andrea M Seime1, A.-M Caroline Spiers1, Elena Maklashina2,3,
Imke Schro¨der4, Robert P Gunsalus4, Gary Cecchini2,3and Joel H Weiner1
1 CIHR Membrane Protein Research Group, Department of Biochemistry, University of Alberta, Edmonton, Alberta, Canada
2 Molecular Biology Division, Veterans Affairs Medical Center, San Francisco, CA, USA
3 Department of Biochemistry and Biophysics, University of California, San Francisco, CA, USA
4 Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, CA, USA
Escherichia coli, when grown anaerobically with
fuma-rate as the respiratory oxidant, develops a respiratory
chain terminated by a membrane-bound
menaqui-nol:fumarate oxidoreductase (FrdABCD1) [1,2] The
enzyme comprises a catalytic dimer of the FrdA (65.8 kDa) and FrdB (27 kDa) subunits that is anchored to the inner surface of the cytoplasmic mem-brane by two small hydrophobic memmem-brane-anchor
Keywords
fumate reductase; Q-site; iron-sulfur;
menaquinol
Correspondence
R A Rothery, Department of Biochemistry,
474 Medical Sciences Building, University of
Alberta, Edmonton, Alberta T6G 2H7
Fax: +1 780 492 0886
Tel: +1 780 492 2229
E-mail: Richard.Rothery@UAlberta.ca
(Received 13 September 2004, revised 22
October 2004, accepted 1 November 2004)
doi:10.1111/j.1742-4658.2004.4469.x
We have used fluorescence quench titrations, EPR spectroscopy and steady-state kinetics to study the effects of site-directed mutants of FrdB, FrdC and FrdD on the proximal menaquinol (MQH2) binding site (QP) of Escherichia coli fumarate reductase (FrdABCD) in cytoplasmic membrane preparations Fluorescence quench (FQ) titrations with the fluorophore and MQH2 analog 2-n-heptyl-4-hydroxyquinoline-N-oxide (HOQNO) indi-cate that the QPsite is defined by residues from FrdB, FrdC and FrdD In
FQ titrations, wild-type FrdABCD binds HOQNO with an apparent Kdof 2.5 nm, and the following mutations significantly increase this value: FrdB-T205H (Kd¼ 39 nm); FrdB-V207C (Kd¼ 20 nm); FrdC-E29L (Kd¼
25 nm); FrdC-W86R (no detectable binding); and FrdD-H80K (Kd¼
20 nm) In all titrations performed, data were fitted to a monophasic bind-ing equation, indicatbind-ing that no additional high-affinity HOQNO bindbind-ing sites exist in FrdABCD In all cases where HOQNO binding is detectable
by FQ titration, it can also be observed by EPR spectroscopy Steady-state kinetic studies of fumarate-dependent quinol oxidation indicate that there
is a correlation between effects on HOQNO binding and effects on the observed Km and kcat values, except in the FrdC-E29L mutant, in which HOQNO binding is observed, but no enzyme turnover is detected In this case, EPR studies indicate that the lack of activity arises because the enzyme can only remove one electron from reduced MQH2, resulting in it being trapped in a form with a bound menasemiquinone radical anion Overall, the data support a model for FrdABCD in which there is a single redox-active and dissociable Q-site
Abbreviations
DmsABC, E coli dimethylsulfoxide reductase; FQ, fluorescence quench; FrdABCD, E coli fumarate reductase; FrdCAB, Wolinella
succinogenes fumarate reductase; HOQNO, 2-n-heptyl-4-hydroxyquinoline-N-oxide; LPC, oxidized lapachol [2-hydroxy-3-(3-methyl-2-butenyl)-1,4-naphthoquinone]; LPCH2, reduced lapachol; MQ, menaquinone; MQH2, menaquinol; NarGHI, nitrate reductase A; SdhCAB, Bacillus subtilis succinate dehydrogenase; SdhCDAB, E coli and ⁄ or eukaryotic succinate dehydrogenase.
Trang 2subunits, FrdC (15 kDa) and FrdD (13.1 kDa) The
crystal structure of FrdABCD has been reported at
3.3 A˚ resolution [3,4], and has an overall architecture
similar to that of the E coli complex II homolog
SdhCDAB (succinate:ubiquinone oxidoreductase) [5,6]
Each enzyme contains a single FAD that is covalently
bound to the catalytic subunit (FrdA⁄ SdhA) and three
[Fe-S] clusters (a [2Fe-2S] cluster, a [4Fe-4S] cluster,
and a [3Fe-4S] cluster) coordinated by the
electron-transfer subunit (FrdB⁄ SdhB) [1] However, important
differences exist between the membrane-intrinsic
domains of the two enzymes [1,7] The
membrane-intrinsic domain of SdhCDAB coordinates a single
heme b (b556) that is sandwiched between the SdhC
and SdhD subunits [8,9] Quinone binding and
reduc-tion is believed to take place in the region between the
heme and the [3Fe-4S] cluster of SdhB [1,6] In the case
of FrdABCD, the membrane-intrinsic domain does not
contain heme, but instead contains two menaquinones
at discreet sites in the crystallized form of the enzyme
[3,4] In both enzymes, despite the available structures,
the number of functional quinone⁄ quinol binding sites
has yet to be unequivocally determined
The menaquinones identified in the crystal structure
of FrdABCD [3] are located at sites towards the inner
(cytoplasmic) and outer (periplasmic) sides of the
mem-brane-intrinsic domain of the enzyme (FrdCD) One
site, the QP site (the proximal Q-site), is located in the
interface region between the FrdCD subunits and
the [3Fe-4S] cluster coordinating region of FrdB on the
cytoplasmic side of the membrane The other site, the
QDsite (the distal Q-site) is located approximately 25 A˚
from the QP site on the opposite (periplasmic) side of
the membrane [3,10] The relatively large distance
between the two sites may preclude direct
electron-trans-fer through the protein medium, which is believed to be
limited to a distance of approximately 14 A˚ [11]
How-ever, a third region of electron density has been
identi-fied recently between the QPand QDsites (the ‘M’ site),
and is centered approximately 13 A˚ from each Q-site [4]
If this electron density corresponds to an additional
electron-transferring cofactor, it could provide a conduit
for electron-transfer from the QD site to the QP site
However, analyses of the bioenergetics of respiratory
growth of E coli on fumarate indicate that FrdABCD
turnover does not produce a transmembrane
electro-chemical potential [12], suggesting the presence of a
sin-gle dissociable and redox-active Q-site that is formally
located on the cytoplasmic side of the membrane
Menaquinol (MQH2) oxidation by FrdABCD has
been studied using a combination of site-directed
muta-genesis, enzymology, EPR spectroscopy and X-ray
crys-tallography Initial mutagenesis studies suggested that
there may be two Q-sites present – a polar QB site (equivalent to the QPsite), and an apolar QAsite (equiv-alent to the QDsite) [13–15] Investigation of the steady-state kinetics of quinol-dependent fumarate reduction
by FrdABCD suggests that MQH2 binding and oxida-tion occur at a single site [16] Kinetic studies carried out in the presence of HOQNO or alkylated dinitro-phenol derivatives also support the presence of a single MQH2oxidation site [17] By exploiting the fluorescent properties of HOQNO in fluorescence quench (FQ) titrations, we determined that this inhibitor binds at a single high-affinity site within FrdABCD [18,19] EPR studies indicate that this high-affinity site is conforma-tionally linked to the [3Fe-4S] cluster of FrdB [18] The emerging hypothesis that there is a single site for MQH2
or HOQNO binding has been complicated recently by the observation in crystallographic studies that the QD site is unoccupied when HOQNO or a dinitrophenol derivative is bound at the QPsite [4] Given the available structural information on FrdABCD, it would therefore
be of interest to examine the effects of a range of site-directed mutants on the HOQNO binding properties and enzymology of the enzyme
In this paper, we evaluate the effects of mutation of amino acid residues located in the vicinity of the QP site on HOQNO binding to FrdABCD We have deter-mined the effect of each mutation on HOQNO binding detected by FQ titration and EPR spectroscopy We have also investigated the effects the mutants have on the steady-state kinetics of fumarate-dependent quinol oxidation
Results
Selection of mutants of FrdB, FrdC and FrdD The following residues are located within approximately
5 A˚ of the menaquinone (MQ) observed at the QP-site
in the structure of FrdABCD: T205, F206, Q225 and K228 from FrdB; R28, E29, W86, L89 and A93 from FrdC; and W14, F17, G18, H80, R81 and H84 from FrdD [3,4,10] Site-directed mutants of some of these res-idues have been generated and partially characterized, including the following: E29L [14,20], FrdC-W86R, FrdD-H80K and FrdD-H84K [14] In the con-text of this study, mutants of the following residues located at a slightly greater distance from the QPsite are also potentially of interest: FrdB-V207 ( 8 A˚ from QP,
a FrdB-V207C mutant) [21], and FrdC-A32 ( 9 A˚ from QP, a FrdC-A32V mutant) [14] At an even greater distance away from the QPsite is FrdC-F38 ( 18 A˚), a mutation at this position (FrdC-F38M [14]), would be expected to have little effect on MQH2 binding and
Trang 3oxidation Finally, we generated a mutant of FrdB-T205
(FrdB-T205H) to assess the role of the [3Fe-4S] cluster
binding domain of FrdB in defining the QP-site This
residue is sandwiched between the [3Fe-4S] cluster and
the QP site All eight mutant enzymes were studied to
assess the effects of the mutations on MQH2 binding
using FQ titrations, EPR spectroscopy and steady-state
kinetic studies The locations of all the mutated residues
located within 10 A˚ of the QP site are illustrated in
Fig 1 HOQNO has a very similar structure to that of
MQ, and as a result appears to bind to the QPsite in an
almost identical way (compare Fig 1A and B with C
and D) This similarity in both structure and binding
renders HOQNO an excellent inhibitor with which to
characterize the QPsite of FrdABCD
FQ titrations of HOQNO binding to mutant
FrdABCD
HOQNO is a close structural analog of MQH2⁄ MQ
and is a very potent inhibitor of FrdABCD [16,18]
When excited at 341 nm, free HOQNO in aqueous
solution fluoresces with an emission wavelength of
479 nm Its fluorescence is completely quenched when bound to FrdABCD and certain other E coli respirat-ory chain enzymes (including dimethylsulfoxide reduc-tase and nitrate reducreduc-tase A [18,19,22–24]) This enables its binding to a Q-site to be analyzed by FQ titration Figure 2 shows representative titrations of membranes containing the wild-type and mutant enzymes studied herein Data for all of the mutants is presented in Table 1 DW35 membranes lacking FrdABCD (Fig 2A) do not exhibit high-affinity HOQNO binding The following FrdABCD mutants bind HOQNO with Kdvalues equivalent to that of the wild-type enzyme (Kd¼ 2.5 nm; Fig 2B): FrdC-A32V (2.5 nm; not shown), FrdC-F38M (2.5 nm; not shown) and FrdD-H84K (3.0 nm, not shown) At the opposite extreme, it is clear that the FrdC-W86R mutant does not exhibit high-affinity HOQNO binding (Fig 2E) This mutant appears to have a similar phenotype to that of the previously reported FrdC-H82R mutant [18,25] Intermediate effects are observed with the fol-lowing mutants: FrdB-T205H (Kd¼ 39 nm; Fig 2C), FrdB-V207C (20 nm; not shown), FrdC-E29L (25 nm; Fig 2D) and FrdD-H80K (20 nm; Fig 2F) Based on
Fig 1 Positions of the mutated residues
close to the Q P site studied herein A and B
show views of the MQ-bound form of
FrdA-BCD (1L0V), whereas C and D show views
of the HOQNO-bound form (1KF6) A and C
represent views from an identical
perspec-tive, as do panels B and D (Experimental
procedures) (A) Looking along the axis
defi-ned by the two keto-oxygens of the
prox-imal menaquinone (MQ) naphthoquinone
bicycle (B) Looking along the axis of the
MQ towards the isoprenoid chain (C) The
same perspective as A, but with HOQNO
bound (D) The same perspective as B, but
with HOQNO bound In all panels, FrdB and
FrdA are above the MQ ⁄ HOQNO plane, and
FrdC and FrdD are substantially below the
MQ ⁄ HOQNO plane Residues from FrdB,
FrdC and FrdD have labels starting with ‘B-’,
‘C-’, and ‘D-’, respectively.
Trang 4these observations and the FrdABCD structure [3,4], it
is clear that residues from FrdB, FrdC and FrdD play
important roles in defining the QP site In every case
where binding is detected, the data can be fitted to an
equation (Eqn 1) describing noncooperative binding at
a single site within FrdABCD
Table 1 shows the calculated specific concentration
of HOQNO binding sites for each mutant in which
binding is detected by FQ titration It also shows the
concentration of FrdABCD calculated by EPR spin
quantitation of both the [2Fe-2S] and [3Fe-4S] clusters
In each case, the estimated number of Q-sites per
enzyme is very close to unity, indicating that HOQNO
binding occurs at a single site within FrdABCD Based
on enzymes that bind HOQNO, 1.02 ± 0.12 sites were
observed per [3Fe-4S] cluster and 1.05 ± 0.09 sites
were observed per [2Fe-2S] cluster
Detection of HOQNO binding by EPR
spectroscopy
Figure 3 shows the effect of HOQNO on the EPR
spectrum around g¼ 2.0 of ferricyanide-oxidized
HB101 membrane samples containing wild-type and mutant FrdABCD EPR spectra of membranes lacking overexpressed FrdABCD exhibit low-intensity features around g¼ 2.0 upon which HOQNO has little effect (Fig 3A) Spectra of membranes containing over-expressed wild-type FrdABCD exhibit the EPR spec-trum of its oxidized [3Fe-4S] cluster (Fig 3B) This spectrum is nearly isotropic with a peak at g¼ 2.02 (gz) and a broad trough immediately up-field As has been reported previously [18,20], addition of HOQNO elicits the observation of an additional peak-trough at approximately g¼ 1.98 (gxy)
Both of the FrdB mutants studied herein (FrdB-T205H and FrdB-V207C) have significant effects on the EPR properties of FrdABCD In the case of the FrdB-T205H mutant, the [3Fe-4S] cluster line-shape is narrower than that of the wild-type (note the position
of the trough in the spectrum without HOQNO; Fig 3C) As is the case for the wild-type enzyme, addi-tion of HOQNO results in the resoluaddi-tion of a peak-trough on the high-field side of the g¼ 2.02 peak This peak-trough is centered at a g-value reflecting the narrower spectrum of the [3Fe-4S] cluster in the
Fig 2 Representative fluorescence quench titrations of HOQNO binding to wild-type and mutant FrdABCD in DW35 membranes Titrations were carried out using membranes from E coli DW35 transformed with plasmids encoding wild-type and mutant FrdABCD at total mem-brane protein concentrations of 0.2 (e), 0.3 (h), 0.4 (n), and 0.5 mgÆmL)1(s) Data were fitted to the following specific enzyme concentra-tions (nmolÆmg protein)1) and K d values (n M ): (A) background, 0.36, > 500; (B) wild-type, 3.54, 2.5; (C) FrdB-T205H, 3.13, 39; (D) FrdC-E29L, 3.26, 25; (E) FrdC-W86R, negligible binding; (F) FrdD-H80K, 3.61, 20 Note that in the cases of the background and FrdC-W86R mutant membranes, the data presented represent insignificant binding.
Trang 5FrdB-T205H mutant in the absence of inhibitor
(gxy¼ 2.0 in the presence of inhibitor rather than at
1.98) Figure 3D shows the spectrum of oxidized
mem-branes containing overexpressed FrdB-V207C mutant
enzyme In agreement with Manadori et al [21], little
or no [3Fe-4S] cluster is assembled into this mutant
enzyme (compare Fig 3A and D), and therefore
HOQNO binding cannot be detected by its
perturba-tion of the EPR spectrum of the oxidized enzyme (see
below)
In contrast to the results of Ha¨gerha¨ll et al [20], the
EPR experiments reported herein indicate that
HO-QNO elicits an effect on the EPR line-shape of the
[3Fe-4S] cluster of the FrdC-E29L mutant enzyme
(Fig 3E) This result is consistent with the observation
of HOQNO binding by FQ titration (Fig 2D and
Table 1) For the other mutations located within the
membrane anchor subunits (FrdC and FrdD), there is
a strong correlation between the observation of an
HOQNO-induced line-shape change and the
observa-tion of inhibitor binding in FQ titraobserva-tions (compare
Figs 2 and 3, Table 1) Thus, no EPR line-shape
change is elicited on the FrdC-W86R mutant [3Fe-4S]
cluster spectrum (Fig 3G)
HOQNO binding to reduced wild-type and
FrdC-V207C mutant enzyme
The EPR properties of reduced wild-type FrdABCD
are complicated by spin–spin interactions between the
paramagnetic [Fe-S] clusters present (viz between the
S¼ ½ [2Fe-2S] and [4Fe-4S] clusters and the S ¼ 2 reduced [3Fe-4S] cluster) [26] The clusters have mid-point potentials (Em values) of )79 mV ([2Fe-2S] c1uster [27]), )320 mV ([4Fe-4S] c1uster [26]), and )70 mV ([3Fe-4S] c1uster [18,21,26]) Because of the pairing of the [3Fe-4S] cluster with the [4Fe-4S] cluster
in a 7Fe ferredoxin-type motif, we examined the possi-bility that HOQNO binding to the QP site may affect the EPR properties of the fully reduced enzyme Figure 4A shows that HOQNO has no effect on the spectrum of dithionite-reduced HB101 membranes lacking overexpressed FrdABCD No differences are observed between the spectrum recorded in the absence
of HOQNO (Fig 4Ai) and that recorded in its pres-ence (Fig 4Aii) The spectrum of reduced membranes containing overexpressed wild-type FrdABCD recor-ded in the absence of HOQNO has an intense peak at
g¼ 2.02 (gz) and a peak-trough at g¼ 1.93 (gxy) (Fig 4Bi) These comprise the EPR spectrum of the [2Fe-2S] cluster of FrdB [27] The EPR spectrum of the [4Fe-4S] cluster manifests itself as a very broad, rapidly relaxing signal underlying that of the [2Fe-4S] cluster [21,26] with peaks at g¼ 2.18 and troughs at
g¼ 1.82 and g ¼ 1.66 No significant effect is elicited
on this spectrum by HOQNO (compare Fig 4Bi and Bii)
Figure 4C shows similar spectra recorded of mem-branes containing the overexpressed FrdB-V207C mutant that contains a [4Fe-4S] cluster in place of the [3Fe-4S] cluster of the wild-type enzyme [21] In this case, the broad underlying spectrum arises from the
Table 1 Effect of the FrdABCD mutations on HOQNO binding determined by FQ titrations and EPR spectroscopy in E coli strain DW35 The concentration of the dithionite-reduced [2Fe-2S] cluster was estimated by double integration of EPR spectra recorded at 40 K under nonsaturating conditions using a CuEDTA concentration standard [47] The concentration of the ferricyanide-oxidized [3Fe-4S] cluster was estimated by double integration of EPR spectra recorded at 9 K under nonsaturating conditions using a Cu-EDTA concentration standard [47] The effect of HOQNO on the [3Fe-4S] cluster EPR line-shape was determined using E coli HB101 membranes Samples and EPR con-ditions were as described for Figs 3 and 4 ND, not detected.
Membrane
preparation
HOQNO
K d (n M )
[Q-sites] (nmolÆmg)1)
by FQ
[2Fe-2S] (nmolÆmg)1)
by EPR
[3Fe-4S] (nmolÆmg)1)
by EPR
Q-sites per [2Fe-2S]
Q-sites per [3Fe-4S]
EPR effect
a Features clearly attributable to either a [2Fe-2S] cluster or a [3Fe-4S] are not detected in spectra of reduced and oxidized membrane sam-ples from E coli strain DW35 b The FrdB-V207C mutant contains a [4Fe-4S] cluster in place of the [3Fe-4S] cluster of the wild-type enzyme.
c
In this case, the effect of HOQNO was determined by analyses of spectra of dithionite-reduced samples recorded as described in the legend to Fig 4.
Trang 6spin-coupled pair of [4Fe-4S] clusters and comprises a peak at g ¼ 2.29, and troughs at g ¼ 1.87 and 1.67 Addition of HOQNO causes the appearance of a peak
at g¼ 1.98 (compare Figure 4Ci and ii) Overall, these data are consistent with there being a perturbation of the engineered [4Fe-4S] cluster in the FrdB-V207C mutant by HOQNO, and with there being no pertur-bation of the [4Fe-4S] cluster of the wild-type enzyme
Fig 4 Effect of HOQNO on the engineered [4Fe-4S] cluster EPR spectrum of FrdB-V207C FrdABCD in HB101 membranes Mem-branes were incubated in the absence of (i) or presence of (ii) 0.5 m M HOQNO for 5 min, then reduced with 5 m M dithionite under argon for 5 min prior to being frozen in liquid nitrogen Spec-tra are presented of membranes containing no overexpressed enzyme (A), and membranes containing overexpressed wild-type (B), and FrdB-V207C (C) EPR spectra were recorded as described for Fig 3.
Fig 3 Effect of HOQNO on the [3Fe-4S] cluster EPR spectrum of
wild-type and mutant FrdABCD in HB101 membranes Membranes
were incubated with 0.5 m M HOQNO (thick lines) or an equivalent
volume of ethanol for 5 min (thin lines), then oxidized with 0.2 m M
ferricyanide for two minutes prior to being frozen in liquid nitrogen.
Spectra are shown of membranes containing no overexpressed
enzyme (A), and membranes containing overexpressed wild-type
(B), FrdB-T205H (C), FrdB-V207C (D), FrdC-E29L (E), FrdC-A32V (F),
FrdC-W86R (G), and FrdD-H80K (H) EPR spectra were recorded
under the following conditions: temperature, 12 K; microwave
power, 20 mW at 9.47 GHz; modulation amplitude, 10 Gppat 100
KHz Spectra were normalized to a nominal protein concentration
of 30 mgÆmL)1 In addition, the absolute intensity of the g ¼ 2.02
peaks were normalized for each pair of spectra.
Trang 7Effect of the mutations on the quinol:fumarate
oxidoreductase activity of FrdABCD
In order to gain a broader understanding of the effects
of the mutants on the physiological quinol oxidation
reaction catalyzed by FrdABCD, we studied their
effects on the steady-state kinetics of the quinol:
fumarate oxidoreductase reaction using the MQH2 analog lapachol [2-hydroxy-3-(3-methyl-2-butenyl)-1,4-naphthoquinone; LPC] When reduced, this substrate (LPCH2) has significant structural similarity to MQH2, and in its oxidized form has a convenient absorbance peak in the visible region at 481 nm in aqueous solu-tion [16] Figure 5 shows representative Eadie–Hofstee plots describing the steady-state kinetic behavior of wild-type and a subset of the mutants of FrdABCD in DW35 membranes The wild-type enzyme has a Km for LPCH2 of approximately 225 lm and a kcat of approximately 71 s)1 The FrdB-T205H and FrdD-H80K mutants have increased Km values (of 355 lm and 670 lm, respectively), but have similar kcatvalues
to that of the wild-type (68 s)1 and 67 s)1, respect-ively) The FrdC-A32V mutant exhibits quite different behavior, with a decrease observed in both the Kmand the kcat values (to 115 lm and 31 s)1, respectively) Likewise, the FrdB-V207C mutant also displayed a decrease in both Km and kcat (Table 2) Despite the HOQNO binding observed both by EPR and FQ titra-tion, the FrdC-E29L mutant exhibited no quinol:fuma-rate oxidoreductase activity Kinetic data for all of the mutants are summarized in Table 2
Detection of a menasemiquinone radical anion
in the FrdC-E29L mutant The FrdC-E29L mutant is unusual because it retains high-affinity HOQNO binding (Table 1 and Fig 2), but demonstrates no fumarate-dependent LPCH2 oxi-dation It has been demonstrated previously by redox potentiometry to stabilize a menasemiquinone radical
Fig 5 Determination of steady-state kinetic parameters for
wild-type and mutant FrdABCD e, wild-wild-type, K m ¼ 225 l M , k cat ¼
71 s)1 h , FrdAB T205H CD; Km¼ 355 l M , kcat¼ 68 s)1 s ,
FrdABCD H80K , Km¼ 670 l M , kcat¼ 67 s)1 n, FrdABC A32V C, Km¼
115 l M , k cat ¼ 31 s)1 Assays at a range of LPCH 2
concentra-tions were carried out as described in the Experimental
proce-dures.
Table 2 Effect of the FrdABCD mutants on the kinetic parameters for lapachol oxidation in E coli strain DW35 Growth, ability of the DW35 based strains used herein to support anaerobic growth using glycerol as carbon source and fumarate as respiratory oxidant HOQNO binding is as judged by the data presented in Table 1 Group, classification of mutant phenotypes: 0, no quinol oxidation, no high-affinity HOQNO binding, does not support growth; 1, normal or modulated Kmand normal kcatfor quinol oxidation, high-affinity HOQNO binding, supports growth; 2 ) normal or modulated K m with decreased k cat , high-affinity HOQNO binding, supports growth; 3, no quinol oxidation, high-affinity HOQNO binding, does not support growth NA, not applicable Membranes from the background strain, E coli DW35, do not contain FrdABCD ND, not detected.
a Kinetic parameters were determined from Eadie–Hofstee plots such as those presented in Fig 5.
Trang 8anion [20] Thus, a plausible explanation for the lack
of quinol:fumarate oxidoreductase activity is that this
mutant becomes trapped in a state in which a
mena-semiquinone radical anion is bound to the QP site We
tested this hypothesis by attempting to observe
turn-over-induced radical species in the wild-type and
FrdC-E29L mutant enzymes Figure 6 shows EPR
spectra recorded at 150K of variously treated mem-brane preparations No g ¼ 2.00 radical signal is detected in oxidized and dithionite-reduced mem-branes containing overexpressed wild-type enzyme (Fig 6A,B) Addition of fumarate to dithionite-reduced membranes containing wild-type enzyme elicits the observation of a small g¼ 2.006 signal consistent with the appearance of a menasemiquinone radical intermediate under turnover conditions As is the case for the wild-type enzyme, dithionite-reduced mem-branes containing the FrdC-E29L mutant enzyme exhi-bit no radical signal A significant signal is observed
in oxidized membranes containing mutant enzyme An intense g¼ 2.006 signal is observed when the FrdC-E29L mutant enzyme is reduced with dithionite and then oxidized with fumarate, consistent with this mutant becoming trapped in a menasemiquinone bound form when enzyme turnover is attempted (Fig 6G,H)
Discussion
We have investigated the effects of a number of point mutations on the affinity of FrdABCD for HOQNO
In each case where HOQNO binding is detected, there
is a striking correlation between the concentration of binding sites and the concentration of enzyme deter-mined by EPR spin quantitation of the [2Fe-2S] and [3Fe-4S] clusters (Table 1) Where modulation of the
Kdfor HOQNO is detected, the FQ data can be fitted
to a binding equation describing noncooperative bind-ing at a sbind-ingle site within FrdABCD These observa-tions are consistent with the presence of a single redox-active dissociable Q-site in FrdABCD, and indi-cate that this site coincides with the QP site observed
in the crystal structures of Iverson et al [3,4] The HOQNO binding data agree with the structure of FrdABCD incubated in the presence of HOQNO, in which the inhibitor is bound exclusively at the QPsite
We previously reported the effect of HOQNO on the EPR line-shape of the [3Fe-4S] cluster of FrdB, and showed that a point mutation in FrdC, FrdC-H82R, eliminated both this effect and HOQNO bind-ing detected by FQ titration [18] However, the posi-tion of FrdC-H82 within the hydrophobic core of FrdC (> 5 A˚ away from QP), along with the relatively severe Arg substitution, warranted re-examination of HOQNO binding to FrdABCD using a range of avail-able mutations It is quite possible that the FrdC-H82R mutation causes relatively gross conformational changes that could affect both the QP and QD sites While some of the mutations studied herein may fall into the same category as the FrdC-H82R mutant (i.e
Fig 6 Demonstration that turnover of the FrdC-E29L mutant is
stalled with a menasemiquinone radical-bound form in E coli
DW35 membranes EPR spectra were recorded of DW35
mem-branes containing wild-type enzyme (A–D) and FrdC-E29L mutant
enzyme (E–H) (A, E), membranes reduced with 5 m M dithionite for
2 min; (B) and (F), oxidized membranes (C) and (G), membranes
reduced with dithionite for 2 min, then treated with 25 m M
fuma-rate for 30 s (D) and (H), as for (C) and (G), but with the incubation
with fumarate for 1 min EPR spectra were recorded at 150 K using
a microwave power of 20 mW at 9.44 GHz and a modulation
ampli-tude of 1.2 G pp Spectra were normalized to a protein concentration
of 30 mgÆmL)1.
Trang 9the FrdC-W86R mutant), we were able to study a
range of mutations that are more likely to have local
effects within the protein Overall, there is a good
cor-relation between the location of the mutated residues
and the severity of the observed effects on HOQNO
binding (compare Figure 1 and Table 1)
An effect on the EPR spectrum of the [3Fe-4S]
clus-ter is clearly observed in each case where HOQNO
binding is detected by FQ titration In addition, we
were able to observe that this effect is not propagated
beyond the location of the [3Fe-4S] cluster (Fig 4)
The FrdB-V207C mutant contains a [4Fe-4S] cluster in
place of the [3Fe-4S] cluster of the wild-type enzyme,
so that the mutant enzyme contains two [4Fe-4S]
clus-ters coordinated by a motif similar to those found in
the bacterial 8Fe ferredoxins [21] In this mutant, the
converted cluster is paramagnetic in its reduced state,
but its spectroscopic analysis is complicated by spin–
spin interactions with the other two reduced clusters of
the enzyme (Fig 4) Despite this, we were able to
dem-onstrate that HOQNO elicits a line-shape change on
the EPR spectrum of the fully reduced FrdB-V207C
mutant Overall, the combination of FQ and EPR data
confirm that the QP site is defined by residues from
FrdB, FrdC and FrdD
Our observation that the QP site is closely coupled
to the [3Fe-4S] cluster of FrdB bears interesting
com-parison with data reported for the membrane-bound
E coli dimethylsulfoxide reductase (DmsABC) This
enzyme is a complex iron–sulfur molybdoenzyme that,
like FrdABCD, contains no heme within its
mem-brane anchor domain (DmsC) [28] The electron
transfer subunit of DmsABC (DmsB) contains four
[4Fe-4S] clusters, and one of these can be changed to
a [3Fe-4S] cluster by site-directed mutagenesis (in a
DmsB-C102S mutant) [29] Treatment of this
mutant with HOQNO results in a perturbation of the
[3Fe-4S] cluster EPR spectrum that is similar to that
reported for the [3Fe-4S] cluster of FrdABCD [18,30]
It is therefore likely that the dissociable Q-site of
DmsABC is located in the interface region between
the membrane-anchor (DmsC) and the
electron-trans-fer subunit (DmsB)
Comparison of the FQ titration, EPR and
steady-state kinetic data on the FrdABCD mutants reported
herein supports their assignments to the following
groups:
0 – no enzyme activity, no high-affinity HOQNO
binding, unable to support growth Members: the
FrdC-W86R mutant and the FrdC-H82R mutant
pre-viously reported by us [18,25]
1 – normal or modulated Km, normal kcat, high-affinity
HQONO binding, able to support growth Members:
the wild-type enzyme, the FrdB-T205H, FrdC-F38M, FrdD-H80K and FrdD-H84K mutants
2 – normal or modulated Km, decreased kcat, high-affinity HOQNO binding, able to support growth Members: the FrdB-V207C and FrdC-A32V mutants
3 – no quinol oxidation, high-affinity HOQNO bind-ing, unable to support growth Member: the FrdC-E29L mutant
Overall, the kinetic data presented herein are consis-tent with the occurrence of simple Michaelis–Menten kinetics, with LCPH2 binding and oxidation occurring
at a single Q-site (Fig 5) However, it is notable that mutants that appear to have little effect on HOQNO binding can modulate the observed steady-state kinet-ics of the enzyme For example, the FrdC-A32V mutant significantly decreases the observed kcat A possible explanation for this is that the increased bulk
of the hydrophobic sidechain is able to stabilize qui-nol⁄ quinone species at the QP site, decreasing the rate
of substrate entry and product egress The other mutant with a significantly decreased kcat, the FrdB-V207C mutant contains a low potential [4Fe-4S] clus-ter (Em of )370 mV [21] in place of the native [3Fe-4S] cluster with an Emof)70 mV) In this case,
it is likely that the relative inefficiency of the low-potential [4Fe-4S] cluster in accepting electrons from reduced quinol explains the decreased kcat
The two FrdD mutants studied herein produced somewhat unexpected results: both are HisfiLys resi-due changes (FrdD-H80 and FrdD-H84), yet only the FrdD-H80K mutant has a significant effect on both the Kd for HOQNO and the Km for LPCH2 Careful examination of the structure of FrdABCD (PDB file L0V [4], Fig 1) reveals a possible explanation for this Whilst the sidechain of FrdD-H84 is marginally closer
to the MQ at the QP site than that of FrdD-H80, the axis of the His-84 imidazole points slightly away from the MQ naphthoquinone bicycle, whereas that of the His-80 imidazole appears to be pointing at least parti-ally towards it Thus, it is more likely that the side-chain of the Lys substitution of FrdD-H80 elicits an effect on HOQNO binding and LPCH2oxidation than the Lys substitution of FrdD-H84 Although this explanation appears plausible, it should be noted that
it is based on structural data of fairly low resolution (3.3 A˚) [3,4]
The FrdB-T205H mutant is of interest in establish-ing the role of FrdB in definestablish-ing the QP site As men-tioned previously (Results), this mutation was chosen because of the location of FrdB-T205H with respect to the QP site, the [3Fe-4S] cluster and the interface between FrdB and the membrane anchor subunits With the exception of the FrdC-W86R mutant, the
Trang 10FrdB-T205H mutant has the largest effect on the Kd
for HOQNO, raising it from 2.5 nm to 39 nm (Fig 2
and Table 1) In addition to its effect on HOQNO
binding, this mutant is also of interest for the
follow-ing reasons: (a) it has a subtle effect on the [3Fe-4S]
cluster EPR line-shape of both the untreated and
HO-QNO treated enzyme (the linewidth is significantly
nar-rowed, compare Fig 3B and C) and (b) it changes the
sequence of the [3Fe-4S] cluster-coordinating Cys
group so that it contains the critical His residue that is
present after the first Cys in the carboxin-sensitive
complex II enzymes [31] We are currently
investi-gating the effect of this mutation on the
carboxin-sen-sitivity of FrdABCD (E Maklashina, RA Rothery, JH
Weiner and G Cecchini, unpublished data)
Of the mutants classified above, the single member of
the Class 3 subgroup is particularly interesting The
FrdC-E29L mutant has no quinol:fumarate
oxidoreduc-tase activity, yet it retains HOQNO binding measured
by both the FQ and EPR methods (Figs 2 and 3)
Ha¨gerha¨ll and coworkers [20] demonstrated by
potenti-ometric titration and EPR spectroscopy that a
mena-semiquinone radical anion is stabilized in this mutant
Examination of FrdABCD structure reveals that the
position of FrdC-E29 is suitable for it to act as a proton
acceptor during enzyme turnover [3,4] Furthermore, it
is widely believed that HOQNO represents a good
ana-log of the menasemiquinone radical intermediate
[32,33] Our observation of a radical when enzyme
turn-over is attempted indicates that the mutant is only able
to accept a single electron from MQH2, resulting in a
bound and stabilized menasemiquinone intermediate,
thus explaining the observed binding of HOQNO and
the lack of quinol:fumarate oxidoreductase activity
In addition to the E coli complex II homologs
(FrdABCD and SdhCDAB), a high-resolution
struc-ture is available for one additional bacterial complex
II homolog This is the Wolinella succinogenes
fuma-rate reductase (FrdCAB) [34,35] which belongs to a
distinct class of complex II homologs that includes the
Bacillus subtilis succinate dehydrogenase (SdhCAB)
[33] These enzymes have a single membrane anchor
subunit (FrdC and SdhC, respectively) that contains
two hemes The structure of the W succinogenes
Frd-CAB [35] reveals that one heme is proximal to the
membrane-extrinsic dimer (heme bP), whilst the other
is distal to it (heme bD) It has been demonstrated that
a point mutation (FrdC-E66Q) that eliminates MQH2
oxidation by FrdCAB is located at a site (a QDsite) in
close proximity to heme bD towards the periplasmic
side of FrdC [34] In B subtilis SdhCAB, the heme bD
is essential for electron-transfer to MQ [36], and this
heme is the only one that appears to be affected by
HOQNO [32] Thus, in contrast to the case in E coli FrdABCD, in W succinogenes FrdCAB and B subtilis SdhCAB, available evidence points towards a model for quinone⁄ quinol binding in which the redox-active dissociable Q-site is located towards the periplasmic side of the membrane anchor domain (at a QD site), and that electron-transfer across the membrane
to⁄ from the catalytic dimer is mediated by the two hemes in a manner similar to that observed in E coli nitrate reductase A (NarGHI) [24,37–40] and suggested for formate dehydrogenase N [41]
The role of the QDsite in FrdABCD remains unre-solved The data presented herein suggest a model for the enzyme in which quinol binding and oxidation occur exclusively at the QP site This is supported by theoretical models of through-protein electron transfer which indicate that the 25 A˚ distance between the QP and QD menaquinones identified in the protein struc-ture is too far to allow for physiologically relevant electron transfer between these sites [11] Our prelimin-ary investigations of mutants (such as FrdD-F57V and FrdC-V35A) surrounding the MQD observed in the protein structure indicate that these have no effect on the HOQNO binding detected by FQ titration and by EPR; and have little effect on quinol:fumarate oxidore-ductase activities A full description of these mutants will appear in a later communication (E Maklashina,
RA Rothery, JH Weiner and G Cecchini, unpublished data) Thus, it is likely that the QDsite plays no direct role in menaquinol oxidation
Overall, by using a range of FrdB, FrdC, and FrdD mutants, we have demonstrated that in every case where HOQNO binding is detected, it occurs at a sin-gle site within FrdABCD In agreement with the struc-tural data of Iverson and coworkers [3,4], we provide biochemical and biophysical evidence for the location
of the dissociable and redox-active Q site of FrdABCD being in the interface region between the FrdCD mem-brane-intrinsic domain and the FrdB electron-transfer subunit These studies provide important information
on the mechanism of MQH2oxidation by FrdABCD
Experimental procedures
Bacterial strains and plasmids
E coli DW35 (zjd::Tn10D(frdABCD)18 sdhC::Kan araD139 D(argF-lac)U169 rpsL150 relA1 flbB5301 deoC1 pfsF25 rbsR [14] does not express FrdABCD or SdhCDAB E coli HB101 (supE44 hsdS20 (rB–mB–) recA13 ara-14 proA2 lacY1 galK2 rpsL20 xyl-5 mtl-1) is a wild-type strain that expres-ses plasmid-encoded FrdABCD to very high levels and generates more consistent EPR data than that obtained