Characterization of Mutants of βHistidine91, βAspartate213, and βAsparagine222, Possible Components of the Energy Transduction Pathway of the Proton-Translocating Pyridine Nucleotide Transhydrogenase of Escherichia coli

Archives of Biochemistry and Biophysics Vol. 388, No. 2, April 15, pp. 299 –307, 2001 doi:10.1006/abbi.2001.2298, available online at http://www.idealibrary.com on

Characterization of Mutants of ␤Histidine91, ␤Aspartate213, and ␤Asparagine222, Possible Components of the Energy Transduction Pathway of the Proton-Translocating Pyridine Nucleotide Transhydrogenase of Escherichia coli Philip D. Bragg 1 and Cynthia Hou Department of Biochemistry and Molecular Biology, University of British Columbia, 2146 Health Sciences Mall, Vancouver, BC, Canada V6T 1Z3

Received November 27, 2000, and in revised form January 12, 2001; published online March 20, 2001

The roles of three residues (␤His91, ␤Asp213, and ␤Asn222) implicated in energy transduction in the membrane-spanning domain II of the proton-translocating pyridine nucleotide transhydrogenase of Escherichia coli have been examined using site-directed mutagenesis. All mutations affected transhydrogenation and proton pumping activities, although to various extents. Replacing ␤His91 or ␤Asn222 of domain II by the basic residues lysine or arginine resulted in occlusion of NADP(H) at the NADP(H)-binding site of domain III. This was not seen with ␤D213K or ␤D213R mutants. It is suggested that ␤His91 and ␤Asn222 interact with ␤Asp392, a residue probably involved in initiating conformational changes at the NADP(H)binding site in the normal catalytic cycle of the enzyme (M. Jeeves et al. (2000) Biochim. Biophys. Acta 1459, 248 –257). The introduced positive charges in the ␤His91 and ␤Asn222 mutants might stabilize the carboxyl group of ␤Asp392 in its anionic form, thus locking the NADP(H)-binding site in the occluded conformation. In comparison with the nonmutant enzyme, and those of mutants of ␤Asp213, most mutant enzymes at ␤His91 and ␤Asn222 bound NADP(H) more slowly at the NADP(H)-binding site. This is consistent with the effect of these two residues on the binding site. We could not demonstrate by mutation or crosslinking or through the formation of eximers with pyrene maleimide that ␤His91 and ␤Asn222 were in proximity in domain II. © 2001 Academic Press

1 To whom correspondence should be addressed. Fax: (604) 8225277. E-mail: [email protected].

0003-9861/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.

Pyridine nucleotide transhydrogenase is a proton ⫹ pump which catalyzes the reversible reaction H out ⫹ ⫹ ⫹ ⫹ NADP ⫹ NADH N NADPH ⫹ NAD ⫹ H in, where reduction of NADP ⫹ by NADH is linked to an inward translocation of protons from the periplasm (in bacteria) or cytosol (in mammalian cells) into the cytosol or mitochondrial matrix, respectively. The properties of transhydrogenase have been reviewed recently (1– 4). The transhydrogenase of Escherichia coli is composed of two subunits, ␣ (510 residues) and ␤ (462 residues), organized as an ␣ 2␤ 2 tetramer (5–9). Three domains are recognized in all known transhydrogenase sequences (1, 10 –12). Domains I and III are extramembrane domains carrying the NAD(H)- and NADP(H)binding sites, respectively. Three-dimensional structures for domains I and III have been proposed (3, 4, 13–16). Domain II is inserted in the cell or mitochondrial membrane. In E. coli it is composed of the Cterminal 100 residues of the ␣-subunit, organized as four transmembrane ␣-helices, and the N-terminal 260 residues of the ␤-subunit, arranged as nine transmembrane ␣-helices (17–19) (Fig. 1). Binding of NADPH and NADP ⫹ causes a marked change in the conformation of the enzyme (20, 21). It is likely that the difference in binding energies of NADPH and NADP ⫹ is the principal driving force for proton pumping (1, 3, 22–25). The pathways by which protons are pumped across domain II are still uncertain. The crystal structures of cytochrome c oxidase and bacteriorhodopsin indicate that protons are translocated across biological membranes through a hydrogen-bonded network which involves the side-chains of charged and polar amino acids and water molecules 299

300

BRAGG AND HOU

FIG. 1. Model of the topology of the transmembrane domains (domain II) of the ␣- and ␤-subunits of the E. coli transhydrogenase, based on the work of Mueller and Rydstro¨m (17). The cytosolic side of the membrane is at the top of the diagram. The helices are numbered so as to be consistent with those of the mitochondrial transhydrogenase (1). The residues examined in this study are circled. T indicates tryptic cleavage sites.

(26 –30). Such a system needs a mechanism by which a conformational change mediated by a redox change or by ligand binding, in the case of the transhydrogenase, can induce changes in the pK a of one or more ionizable amino acids. In previous studies (31–34), we have altered by site-directed mutagenesis all conserved acidic, basic and other polar residues in domain II of the E. coli transhydrogenase. Only mutations of ␤His91, ␤Asp213, and ␤Asn222 resulted in significant effects on proton translocation (31–35). Residues ␤His91 and ␤Asn222 are located on transmembrane helices 9 and 13 of domain II (Fig. 1). Both residues appear to interface on an aqueous channel(s) accessible from the cytosolic surface of the membrane. This fact, together with the high degree of amino acid conservation in helices 9, 13, and 14 between the enzymes from different species, is consistent with a proton translocation pathway involving these helices (2, 19, 34). The residue ␤Asp213 is on the interhelical sequence connecting helices 12 and 13 (Fig. 1) and is close to helix 13. In the present paper we further characterize the properties of ␤His91, ␤Asp213, and ␤Asn222. Substitution of a lysine or arginine residue for ␤His91 or ␤Asn222 resulted in the transhydrogenase retaining bound NADP(H) at the substrate-binding site of domain III. Other amino acid substitutions for ␤His91 and ␤Asn222 had no, or much less obvious, effects on the retention of NADP(H) on domain III. However, almost all mutants of these two amino acids bind NADP ⫹ less rapidly than NADPH to the NADP(H)-

binding site. By contrast, lysine and arginine mutants of ␤Asp213 did not retain tightly bound NADP(H), and did not show the differences in the rate of binding between NADP ⫹ and NADPH. Although we could not show that ␤His91 and ␤Asn222 were in proximity in domain II, we suggest that both residues, either directly or indirectly, influence the state of protonation of the carboxyl group of ␤Asp392, a residue which may control the conformational changes at the NADP(H)binding site on domain III (16). MATERIALS AND METHODS Bacterial strains, plasmids, and mutagenesis. E. coli JM109 cells containing wild-type (pSA2) or mutant plasmids were grown overnight at 37°C in LB 2 broth containing ampicillin at a concentration of 100 ␮g/ml. The medium was shaken at 250 rpm in a New Brunswick Scientific Controlled Environment Incubator Shaker. Plasmid pSA2 contains the nonmutant pnt genes of the pyridine nucleotide transhydrogenase of E. coli introduced into the pGEM-7Zf(⫹) plasmid (8). Site-directed mutagenesis was carried out as described previously (34). Preparation of membrane vesicles containing transhydrogenase. Three types of membrane preparations were used in this study. Unwashed everted membrane vesicles were used for measurements of proton translocation rates since they were less leaky to protons than other preparations. Washed everted inner membrane vesicles 2 Abbreviations used: AcPyAD ⫹, 3-acetylpyridine adenine dinucleotide; Hepes, 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid; Mes, 2-(N-morpholino)ethane sulfonic acid; oPDM, o-phenylenedimaleimide; pPDM, p-phenylenedimaleimide; BMH, 1,6-bis(maleimido)hexane; DBB, dibromobimane; LB, Luria–Bertani.

MUTANTS OF ␤HISTIDWE91, ␤ASPARTATE213, AND ␤ASPARAGINE222 TABLE I

Reverse Transhydrogenation and Proton-Pumping Activities of Unwashed Membrane Vesicles of ␤H91N, ␤N222H, and ␤H91N/␤N222H Reverse transhydrogenation Mutation

Specific activity

%

Proton-pumping activity (%)

Nonmutant ␤H91N ␤N222H ␤H91N/␤N222H

2.6 0.46 0.18 0.04

100 18 7 2

100 4 7 8

Note. Specific activity is expressed as ␮mol AcPyAD ⫹ reduced/ min/mg protein. Proton-pumping activity is expressed in arbitrary units.

were used for most of the other measurements. The transhydrogenase composed over 80% of the protein in these vesicles. For analysis of bound NADP(H) the inner membrane vesicles were washed twice to remove any traces of unbound pyridine nucleotide. The third type of preparation used was Triton-washed inner membrane vesicles. Washing inner membrane vesicles with 1% Triton X-100 made the ␤-subunit of the transhydrogenase more accessible to reaction with trypsin (21), for example. The cell cultures were harvested by centrifugation at 4400g for 20 min. The cell pellets were washed by resuspension in 0.9% NaCl followed by centrifugation at 12,000g for 15 min. Cell pellets were resuspended in buffer A (50 mM Tris–HCl, pH 7.8, 1 mM EDTA) at 1 g wet weight/5 ml. All steps were performed at 0 – 4°C. The cells were lysed by passage through an AMINCO French Pressure Cell at 1400 kg/cm 2. Unbroken cells were removed by centrifugation at 1000g for 15 min. The supernatant was centrifuged at 252,000g for 2 h to give a pellet containing unwashed everted membrane vesicles. In some experiments these were used directly. For the preparation of washed inner membrane vesicles this pellet was suspended in buffer A at 1 g wet weight/5 ml. Membrane vesicles (1.5 ml) were layered on a 6-ml sucrose cushion (45% sucrose (w/w) in buffer A) and centrifuged in a Beckman Type 65 fixed angle rotor at 40,000 rpm (139,000g) for 1.25 h. The outer membrane fraction pelleted to the bottom of the tube and was discarded, while the everted inner membrane vesicles banded at the interface and were removed by a syringe. The vesicles were diluted with buffer A and then centrifuged at 252,000g for 1.5 h to give a pellet of washed inner membrane vesicles. Triton-washed inner membrane vesicles were obtained by sedimenting the interface layer of everted membrane vesicles from buffer A containing 1% (w/v) Triton X-100. All of the membrane vesicle preparations were suspended in 50 mM Hepes–KOH, pH 8, 5 mM magnesium acetate for use. Measurement of transhydrogenation activities. Transhydrogenation of AcPyAD ⫹ by NADPH (“reverse transhydrogenation”) was measured as described previously (32). An appropriate amount of membrane vesicle suspension was added to 1 ml of 50 mM sodium phosphate buffer (pH 7) containing 0.5 mM EDTA, 0.01% Brij 35, 0.5 mM AcPyAD ⫹, and 0.5 mM NADPH. Reduction of AcPyAD ⫹ was followed at 375 nm using a Perkin-Elmer Lambda 3A UV/vis spectrophotometer. The reduction of AcPyAD ⫹ by NADH in the presence of added NADP ⫹ or NADPH is termed “cyclic” transhydrogenation activity. Cyclic transhydrogenation involves reduction by NADH of NADP ⫹ bound at the NADP(H)-binding site, followed by reoxidation of the bound NADPH by AcPyAD ⫹. NADH and AcPyAD ⫹ alternately occupy the NAD(H)-binding site on the ␣-subunit. In contrast to reverse transhydrogenation, the cyclic reaction does not involve pro-

301

ton translocation and is considered to measure hydride transfer activity only (36). For cyclic transhydrogenation of AcPyAD ⫹ by NADH, 0.5 mM NADH replaced NADPH and 0.1 mM NADP ⫹ was added. Assay of transhydrogenase activities at different pH values was carried out using a buffer of 17 mM Mes, 17 mM Hepes, 17 mM Tris, 0.5 mM EDTA, 2 mM dithiothreitol, and 0.1% Brij 35 adjusted to the appropriate pH with HCl or KOH. Protein concentration was determined by the method of Lowry et al. (37). NADP(H) was assayed as in Ref. (32). Assay of proton translocation. Proton translocation was measured by quinacrine fluorescence quenching as described by Glavas et al. (33). The rate of proton translocation was calculated from the initial rate of quenching and measured in arbitrary units.

RESULTS

Exchange of functional groups between ␤His91 and ␤Asn222. The amino acids ␤His91 and ␤Asn222 may be functional components of the proton pathway through domain II of the transhydrogenase (31–34, 38). Miller et al. (39) showed that the proton-translocating aspartic acid residue of subunit c of the ATP synthase of E. coli could be moved to an adjacent transmembrane helix with retention of biological activity. A similar result was obtained with the melibiose carrier by Wilson et al. (40). We therefore examined the activities of the double mutant ␤H91N/␤N222H in which the histidine and asparagine residues have been exchanged in position. The reverse transhydrogenation and proton-pumping activities of unwashed membrane vesicles were considerably reduced by the ␤H91N and ␤N222H mutations (Table I). Placing both mutations in the same protein did not significantly increase either activity. The cyclic transhydrogenation activities of the mutants in the presence of NADP ⫹ and NADPH were compared with those of the nonmutant over a range of pH values (Fig. 2). The cyclic transhydrogenation activities were decreased by the mutations ␤H91N and ␤N222H, and even more drastically by the double mu-

FIG. 2. pH/activity curves for the cyclic reduction of AcPyAD ⫹ by NADH in the presence or absence of NADPH (A) or NADP ⫹ (B). Washed inner membrane vesicles were from the nonmutant (a), ␤H91N (b), ␤N222H (c), and ␤H91N/␤N222H (d) strains. The enzymes were assayed as described under Materials and Methods.

302

BRAGG AND HOU

tation ␤H91N/␤N222H. NADPH gave higher activities than NADP ⫹. The mutations shifted the pH of maximum activity to lower values compared with the nonmutant enzyme. Spatial relationship of ␤His91 and ␤Asn222. The inability to restore transhydrogenation and protontranslocating activities by exchanging the residues at positions 91 and 222 of the ␤-subunit might be due to the lack of proximity of these residues in the membrane. This possibility was tested as follows. Tritonwashed inner membrane vesicles of the double mutant ␤H91C/␤N222C in the cysteine-free transhydrogenase background of pCH93 (38) were treated with one of the following crosslinking agents: 3 mM copper 1,10phenanthrolinate, 0.5 mM iodine, 2 mM diamide, 1 mM o-phenylenedimaleimide (oPDM), 1 mM p-phenylenedimaleimide (pPDM), 1 mM 1,6-bis(maleimido)hexane (BMH), or dibromobimane (DBB). Copper phenanthrolinate, iodine, and diamide are zero-length crosslinking agents catalyzing the formation of a disulfide linkage between two sulfhydryl groups in proximity. oPDM, pPDM, BMH, and DBB can crosslink between two sulfhydryl groups 6, 10, 16, and 5 Å apart, respectively. Internally crosslinked membrane proteins have been found to migrate differently from the noncrosslinked species on SDS–PAGE (41– 44). Examination of the rate of migration of the ␤-subunit in inner membrane vesicles treated with the above crosslinking agents gave no evidence that crosslinking had occurred. The Domain II region of the ␤ subunit migrates on SDS–PAGE as a 25-kDa polypeptide following treatment of the transhydrogenase with trypsin in the presence of NADPH (21). Examination of trypsin-treated vesicles following treatment with crosslinking agents showed that treatment did not affect the rate of migration of the 25-kDa polypeptide. As a further test for the proximity of ␤His91 and ␤Asn222, Triton-washed inner membrane vesicles of the ␤H91C/␤N222C double mutant were reacted with N-(1-pyrene)maleimide. This reagent reacts with sulfhydryl groups. If a pair of sulfhydryl groups on a protein are within 3.5 Å of one another, then the fluorescence emission spectrum of the disubstituted protein will show a broad emission band of the eximer at about 470 nm in addition to the bands of the monomer at about 378, 398, and 417 nm (45). We did not detect an eximer band in our experiments (results not shown). The above experiments gave no evidence that ␤His91 and ␤Asn222 are in proximity in domain II. Effect of substituting charged residues for ␤His91 and ␤Asn222. In preliminary studies we showed that the transhydrogenases of the mutants ␤H91K and ␤N222R retained tightly bound NADP ⫹ (32, 34). This was a significant observation since ␤His91 and ␤Asn222 in domain II are remote in sequence from the

TABLE II

Reverse Transhydrogenation and Proton-Pumping Activities of Unwashed Membrane Vesicles of Mutants of ␤His91, ␤Asp213, and ␤Asn222

Mutation

Reverse transhydrogenation activity (%)

Proton pumping activity (%)

Nonmutant ␤H91K ␤H91R ␤H91D ␤D213K ␤D213R ␤D213E ␤N222K ␤N222R ␤N222D

100 2 1 9 7 5 34 3 2 13

100 30 25 6 49 28 77 13 13 22

Note. Reverse transhydrogenase activities are expressed as a percentage of the activity in nonmutant membrane vesicles, where 100% activity ⫽ 2.6 ␮mol/min/protein. Proton pumping activities are expressed as a percentage of the initial rate of quenching of the fluorescence of quinacrine with nonmutant membrane vesicles. The values are the means of two separate determinations.

NADP(H)-binding site in domain III of the transhydrogenase molecule. It suggested that ␤His91 and ␤Asn222 were coupled to, and influenced the conformation of, the active site, most probably as components of a proton-translocating pathway. These observations have now been extended by systematically altering the charge at these two residues. Table II shows the reverse transhydrogenation and proton-pumping activities of unwashed membrane vesicles from mutants of ␤His91 and ␤Asn222. Arginine, lysine, and aspartic acid were used to replace the histidine and asparagine residues. All mutations resulted in significant loss of reverse transhydrogenation and proton-pumping activities. The pH/activity curves for reverse transhydrogenation in washed inner membrane vesicles of the charged mutants of ␤His91 and ␤Asn222 are shown in Fig. 3. Two patterns of behavior are evident. The lysine and arginine mutants of these two amino acids show maximal activity at pH 5.0. The aspartic acid mutants, in contrast, have higher activity with a maximum in the pH 7–9 range, which is more typical of the nonmutant enzyme. Mutants ␤H91S, H91C, ␤H91N, ␤N222H, and ␤N222C had pH/activity curves similar to that of ␤H91D (results not shown). Table III shows the content of tightly bound NADP(H) in inner membrane vesicles of mutants of ␤His91 and ␤Asn222. Five mutants, ␤H91K, ␤H91R, ␤N222K, ␤N222R, and ␤N222D, showed significant retention of NADP(H). Essentially all of the NADP(H) was present as NADP ⫹, although less than one of the potential two binding sites for NADP(H) in the ␣ 2␤ 2 tetramer was filled by pyridine nucleotide. Of interest is the fact that the mutants ␤H91D, ␤H91N, ␤H91S,

MUTANTS OF ␤HISTIDWE91, ␤ASPARTATE213, AND ␤ASPARAGINE222

FIG. 3. pH/activity curves of the reduction of AcPyAD ⫹ by NADPH (reverse transhydrogenation). The washed inner membrane vesicles were assayed as described under Materials and Methods. The aspartic acid, lysine, or arginine residues to which the indicated residue was changed are indicated by D, K, or R.

␤H91C, ␤N222H, ␤N222C, and ␤N222Y, as well as the nonmutant strain, retained little bound NADP(H). This suggests that the presence of positively charged residues, in particular, at these sites locked the TABLE III

Content of Bound NADP(H) in Double-Washed Inner Membrane Vesicles of Mutants of ␤His91, ␤Asp213, and ␤Asn222 Mutation

Bound NADP(H) (mol/␣ 2␤ 2 tetramer)

Nonmutant ␤H91K ␤H91R ␤H91D ␤H91N ␤H91S ␤H91C ␤D213K ␤D213R ␤D213E ␤N222K ␤N222R ␤N222D ␤N222H ␤N222C ␤N222Y ␤H91N␤N222H

0.01 0.48 ⫾ 0.06 0.80 0.02 ⫾ 0.02 0.035 ⫾ 0.02 0.21 0 0 0 0 0.58 0.54 ⫾ 0.04 0.44 ⫾ 0.08 0.09 ⫾ 0.01 0.18 0.11 0.07 ⫾ 0.03

Note. NADP(H) was assayed as described under Materials and Methods.

303

FIG. 4. pH/activity curves for the cyclic reduction of AcPyAD ⫹ by NADH in the presence of NADPH or NADP ⫹. Washed inner membrane vesicles of nonmutant and mutants of ␤His91 were assayed as described under Materials and Methods.

NADP ⫹ in the binding site of domain III in an occluded form. The effect of mutation of ␤His91 and ␤Asn222 to charged residues on the pH/activity curves for the cyclic reaction with NADPH or NADP ⫹ is shown in Figs. 4 and 5. Two patterns of behavior are evident. The cyclic transhydrogenase activities of inner membrane vesicles of ␤H91K, ␤H91R, and ␤N222R resemble those of the nonmutant vesicles in that the cyclic activities with NADP ⫹ were similar or greater than that obtained with NADPH. By contrast, the cyclic activity with NADP ⫹ was much lower than that with NADPH in vesicles of ␤H91D, ␤N222D, and ␤N222K. Moreover, the pH optimum with NADP ⫹ in the latter cases was shifted to pH 5.5 compared to a pH optimum of pH 5.5 to 6.0, with significant activity in the pH 6.5 to 7.0 range, with NADPH. The enzyme in membrane vesicles of mutants ␤H91N, ␤H91S, ␤N222H, and ␤N222C behaved similarly to those of mutants ␤H91D, ␤N222D, and ␤N222K (results not shown). Effect of substituting other charged residues for ␤Asp213. It has been proposed that ␤Asp213 has a role in the energy transduction pathway of the transhydrogenase (35). The effect of mutating this amino acid to a lysine, arginine, or glutamic acid residue on reverse transhydrogenation and proton translocation is shown in Table II. Mutation to basic residues resulted in considerable loss of transhydrogenation activity and in a shift in the pH optimum to an acidic pH

304

BRAGG AND HOU

FIG. 5. pH/activity curves for the cyclic reduction of AcPyAD ⫹ by NADH in the presence of NADPH or NADP ⫹. Washed inner membrane vesicles of nonmutant and mutants of ␤Asn222 were assayed as described under Materials and Methods.

(Fig. 3). Mutation of aspartic acid to glutamic acid had much less effect on reverse transhydrogenation and proton translocation than the other mutations (Table II). The pH/activity curve was more like that of the nonmutant enzyme (Fig. 3). In the pH/activity curves for the cyclic reaction of ␤D213E and ␤D213K, the activities with NADP ⫹ were not significantly lower than those with NADPH in the range of pH 5.5 to 6.5 (Fig. 6). Inner membrane vesicles of ␤D213E, ␤D213K, and ␤D213R did not retain bound NADP(H) (Table III). Kinetics of the cyclic reduction of AcPyAD ⫹ by NADH. The reduction of AcPyAD ⫹ by NADH in the presence of NADPH or NADP ⫹ by nonmutant transhydrogenase

FIG. 6. pH/activity curves for the cyclic reduction of AcPyAD ⫹ by NADH in the presence of NADPH or NADP ⫹. Washed inner membrane vesicles of mutants of ␤Asp213 were assayed as described under Materials and Methods.

FIG. 7. Cyclic reduction of AcPyAD ⫹ by NADH in the presence of NADP ⫹ (A, C) or NADPH (B, D) by washed inner membrane vesicles of mutants ␤N222K (A, B) and ␤H91R (C, D). The time of preincubation (0, 5, or 15 min) of the vesicles with NADP ⫹ or NADPH before addition of AcPyAD ⫹ and NADH to complete the assay mixture is indicated.

was linear with time at all pH values tested between pH 5.0 and pH 9.0. This was also true when inner membrane vesicles of mutants ␤H91K, ␤D213E, ␤D213K, and ␤D213R were examined. However, the transhydrogenases of mutants ␤H91C, ␤H91S, ␤H91N, ␤H91D, ␤H91R, ␤N222D, ␤N222H, ␤N222R, ␤N222C, and ␤N222K gave nonlinear plots of reduction versus time (see Fig. 7 for typical examples). These plots were characterized by a lag period, but ultimately the curved plots became linear. Typically, the extent of the lag increased with an increase in the pH at which the cyclic activity was assayed. The plots at pH 5.0 and 5.5 were generally almost linear. The lag was more pronounced with NADP ⫹ than with NADPH with inner membrane vesicles of ␤H91N, ␤H91D, ␤H91S, ␤H91C, ␤N222D, ␤N222H, ␤N222C, and ␤N222K. The lag was similar with NADPH and NADP ⫹ with ␤N222R and ␤H91R. Figure 7 illustrates this behavior with ␤N222K and ␤H91R. There was a correlation between the difference in the extent of the lag with NADPH versus NADP ⫹ and the pH/cyclic transhydrogenase activity curves shown in Figs. 4 and 5 (and in others not

MUTANTS OF ␤HISTIDWE91, ␤ASPARTATE213, AND ␤ASPARAGINE222 TABLE IV

K m Values for the Cyclic Reduction of AcPyAD ⫹ by NADH in the Presence of NADPH or NADP ⫹ K m (␮M) Mutant

NADPH

NADP ⫹

Nonmutant ␤H91K ␤H91D ␤H91N ␤N222R ␤N222D ␤N222H

0.7 ⫾ 0.2 1.9 ⫾ 0.4 1.7 ⫾ 0.8 1.6 ⫾ 0.5 1.5 0.7 ⫾ 0.2 0.8 ⫾ 0.6

1.5 ⫾ 0.9 4.0 2.8 ⫾ 0.8 8.6 ⫾ 0.8 2.6 1.4 ⫾ 1.9 2.0 ⫾ 0.9

Note. The assays were carried out at pH 6.0.

shown). There was a much greater lag with NADP ⫹ compared with NADPH with those mutants in which the NADPH-dependent cyclic reaction was greater than the NADP ⫹-dependent cyclic reaction (see Figs. 4 and 5). K m values for NADP ⫹ and NADPH in the cyclic transhydrogenase were determined from the rate of reduction when it had become linear. With the exception of ␤H91N, the K m values for NADP ⫹ were approximately twofold higher than the K m values for NADPH (Table IV). Preincubation of the inner membrane vesicles with NADP(H) decreased the extent of the lag (Fig. 7). NADPH was more effective than NADP ⫹ in those mutants in which NADPH gave a greater activity than NADP ⫹. DISCUSSION

Three amino acids of domain II have been proposed to take part in the energy transduction process which couples the difference in the binding energy of reactants and products to the translocation of protons across domain II (31–35). The amino acids ␤His91 and ␤Asn222 are situated in transmembrane helices 9 and 13, respectively, whereas ␤Asp213 is located in the loop connecting transmembrane helices 12 and 13 (Fig. 1). Previous results had suggested that replacement of these amino acids by others affected the binding of NADP(H) to domain III. Thus, Yamaguchi and Hatefi (35) found that the transhydrogenase of the ␤D213I mutant had a threefold increased affinity for NADPH. The enzymes of mutants ␤H91K and ␤N222R contained tightly bound NADP ⫹ (32, 34). The effect of an introduced positive charge at ␤His91, ␤Asn222 and ␤Asp213 was investigated further. Tightly bound NADP(H) was found in the transhydrogenases of ␤H91R, ␤H91K, ␤N222K, and ␤N222R. It was not found in the enzymes of ␤D213K or ␤D213R. Significantly, bound NADP(H) was not found in the nonmu-

305

tant transhydrogenases or in the enzymes of mutants ␤H91D, ␤H91N, ␤H91C, and ␤D213E. Only low amounts were found in the transhydrogenases of ␤H91S, ␤N222H, ␤N222C, ␤N222Y, and the double mutant ␤H91N/␤N222H. A higher level was present in the enzyme of ␤N222D. Since ␤H91D and ␤D213E did not contain bound NADP(H), the presence of bound NADP(H) in the enzyme of ␤N222D is likely not due to the negative charge of the aspartate residue. It is perhaps due to a particular perturbation of the enzyme structure caused by this residue. The absence of bound NADP(H) in the mutants of ␤Asp213 distinguishes this residue from ␤His91 and ␤Asn222, although mutations at all three residues impaired hydride transfer and coupled proton translocation. The mutation ␤D213E had a lesser effect than other mutations, indicating that glutamic acid could in some measure take over the function of aspartic acid at this position. Cyclic transhydrogenation of AcPyAD ⫹ by NADH in the presence of NADP(H) is considered to be a measure of hydride transfer activity without the complication of coupled proton translocation, inherent in reverse transhydrogenation (36). Examination of the pH/activity curves for the mutants revealed significant differences between the behavior of the mutants and the nonmutant enzyme. These differences were attributed to differences in the binding of NADP ⫹ and NADPH at the NADP(H)-binding site of domain II. In the cyclic reaction either NADP ⫹ or NADPH needs to be bound at the NADP(H)-binding site since they act as intermediaries in the transfer of hydride between NADH and AcPyAD ⫹ (see Results for a description of the mechanism of the cyclic reaction). In the nonmutant enzyme NADP ⫹ was as effective at promoting the cyclic reaction as NADPH. It was also the case with the enzymes of mutants ␤H91K, ␤H91R, and ␤N222R. In the other mutants of ␤His91 and ␤Asn222, NADPH was much more effective than NADP ⫹ in stimulating the cyclic reaction as was reported for the mutant ␤H91E by Hu et al. (46). Moreover, in this latter group of mutants the pH optimum for the cyclic reaction was shifted from about pH 6 to pH 5.5. With the exception of the enzyme of ␤H91K, the reduction of AcPyAD ⫹ by NADH by the mutant transhydrogenases was characterized by a lag before the full rate of reaction was achieved. The lag was abolished by preincubation of the enzyme with NADP(H). However, a longer period of preincubation was required with NADP ⫹ than with NADPH. Thus, NADPH binds more rapidly than NADP ⫹. In the cyclic reaction at pH 6 there was a much greater lag with NADP ⫹ than with NADPH in those mutants in which the mutation had caused a shift in the pH optimum of the NADP ⫹-dependent reaction to lower pH values. Since the K m values for NADP ⫹ were only two-fold higher than those with NADPH, it seems likely that the pH/activity curves reflect the relative binding prop-

306

BRAGG AND HOU

erties of the mutants for NADP ⫹ and NADPH. The lag was not shown by the nonmutant enzyme, or by the enzymes of ␤D213K, ␤D213R, and ␤D213E. It is concluded that all mutations at ␤His91 and ␤Asn222 affect the rate of binding of substrate at the NADP(H)binding site of domain III. As discussed above, the presence of an introduced positive charge at ␤His91 and ␤Asn222 locks the NADP(H) site to occlude this substrate. Recent structural studies of domain III have defined the structure of the NADP(H)-binding site (3, 4, 13–16). Jeeves et al. (16) have suggested that during the normal catalytic cycle of the enzyme the NADP(H)-binding site undergoes conformational changes between an “open” state, in which the pyridine nucleotide can readily bind or from which it can readily dissociate, and an occluded state. A key residue in the conformational change is likely the highly conserved ␤Asp392. This residue is positioned close to the pyrophosphate of bound NADP(H). These workers proposed that protonation and deprotonation of ␤Asp392 during the catalytic cycle might be associated with the transition between occluded and open states of the binding site. Mutation of ␤Asp392 resulted in complete loss of proton-pumping activity and a great reduction in hydride transfer activity (47). Although ␤His91 and ␤Asn222 are located on transmembrane helices, they are accessible to the cytosolic surface near domain III by an aqueous cavity (38). We suggest that the positive charges substituted for ␤His91 or ␤Asn222 affect the protonation of the free carboxyl group of ␤Asp392, stabilizing it in the anionic form and thus locking the NADP(H)-binding site in the occluded state. This suggestion implies that ␤His91 and ␤Asn222 communicate with ␤Asp392 in the nonmutant enzyme, possibly to deliver protons to, or receive protons from, the free carboxyl group of ␤Asp392. The different behavior of mutants of ␤Asp213, particularly their inability to cause occlusion of NADP(H), indicates that these mutations inhibit the transhydrogenase by another mechanism. Nothing is presently known about the disposition in space of the transmembrane helices of the ␤ subunit. Even though ␤His91 and ␤Asn222 are on different transmembrane helices of domain II, it is possible that they are close together (2, 19, 34). The fact that both amino acids face an aqueous cavity, and that their mutants show similar patterns of behavior, favors the idea that they are in proximity. The experiments described here have not produced evidence supporting this hypothesis. Thus, generating the double mutant ␤H91N/␤N222H, in which the two amino acids were exchanged in position, did not reconstitute the functions lost in the single mutants, as found with other systems (39, 40). Furthermore, we were not able to crosslink between two cysteine residues introduced at ␤His91 and ␤Asn222 using zero-length crosslinkers

and those spanning up to 16 Å. Further evidence is required on the disposition of these two residues. In conclusion, mutants of ␤His91, ␤Asn222, and ␤Asp213 were examined for effects of the mutations on transhydrogenase activities, proton translocation, occlusion of NADP(H) at the NADP(H)-binding site, and indirectly for effects on the binding of NADP(H). All mutations affected transhydrogenase activities and proton translocation, although to various extents. Replacing either ␤His91 or ␤Asn222 by arginine or lysine residues resulted in occlusion of NADP(H) at the NADP(H)-binding site of domain III. Other mutations had no, or much smaller, effects on occlusion. NADP(H) was not occluded in ␤D213K or ␤D213R mutants. It is suggested that the introduced positive charge at ␤His91 or ␤Asn222 affects the dissociation of the carboxyl group of ␤Asp392, a residue probably involved in initiating the conformational changes between “open” and occluded forms of the NADP(H)-binding site in the normal catalytic cycle of the enzyme. Most mutants of ␤His91 and ␤Asn222 bound NADP(H) more slowly at the NADP(H)-binding site than nonmutant transhydrogenase and that of the ␤D213K, ␤D213R, and ␤D213E mutants. We were not able to demonstrate by mutation, crosslinking, or through the formation of eximers with pyrene maleimide, that ␤His91 and ␤Asn222 were in proximity in domain II. ACKNOWLEDGMENTS This work was supported by a grant from the Medical Research Council of Canada. We thank Dr. Xiang Hu for preliminary work on this research project.

REFERENCES 1. Hatefi, Y., and Yamaguchi, M. (1996) FASEB J. 10, 444 – 452. 2. Bragg, P. D. (1998) Biochim. Biophys. Acta 1365, 98 –104. 3. Jackson, J. B., Peake, S. J., and White, S. A. (1999) FEBS Lett. 464, 1– 8. 4. Bizouarn, T., Fjellstro¨m, O., Meuller, J., Axelsson, M., Bergkvist, A., Johansson, C., Karlsson, G., and Rydstro¨m, J. (2000) Biochim. Biophys. Acta 1457, 211–228. 5. Clarke, D. M., and Bragg, P. D. (1985) J. Bacteriol. 162, 367–373. 6. Clarke, D. M., and Bragg, P. D. (1985) Eur. J. Biochem. 149, 517–523. 7. Clarke, D. M., Loo, T. W., Gillam, S., and Bragg, P. D. (1986) Eur. J. Biochem. 158, 647– 653. 8. Ahmad, S., Glavas, N. A., and Bragg, P. D. (1992) Eur. J. Biochem. 207, 733–739. 9. Hou, C., Potier, M., and Bragg, P. D. (1990) Biochim. Biophys. Acta 1018, 61– 66. 10. Olausson, T., Fjellstro¨m, O., Meuller, J., and Rydstro¨m, J. (1995) Biochim. Biophys. Acta 1231, 1–19. 11. Jackson, J. B., Cotton, N. P. J., Williams, R., Bizouarn, T., Hutton, M. N., Sazanov, L. A., and Thomas, C. M. (1993) Biochem. Soc. Trans. 21, 1010 –1013. 12. Williams, R., Cotton, N. P. J., Thomas, C. M., and Jackson, J. B. (1994) Microbiology 140, 1595–1604.

MUTANTS OF ␤HISTIDWE91, ␤ASPARTATE213, AND ␤ASPARAGINE222 13. Johansson, C., Berguist, A., Fjellstro¨m, O., Rydstro¨m, J., and Karlsson, B. G. (1999) FEBS Lett. 458, 180 –184. 14. Prasad, G. S., Sridhar, V., Yamaguchi, M., Hatefi, Y., and Stout, C. D. (1999) Nat. Struct. Biol. 6, 1126 –1131. 15. White, S. A., Peake, S. J., McSweeney, S., Leonard, G., Cotton, N. P. J., and Jackson, J. B. (2000) Structure 8, 1–12. 16. Jeeves, M., Smith, K. J., Quirk, P. G., Cotton, N. P. J., and Jackson, J. B. (2000) Biochim. Biophys. Acta 1459, 248 –257. 17. Meuller, J., and Rydstro¨m, J. (1999) J. Biol. Chem. 274, 19072– 19080. 18. Bragg, P. D., and Hou, C. (2000) Biochem. Biophys. Res. Commun. 273, 955–959. 19. Bizouarn, T., Meuller, J., Axelsson, M., and Rydstro¨m, J. (2000) Biochim. Biophys. Acta 1459, 284 –290. 20. Yamaguchi, M., and Hatefi, Y. (1989) Biochemistry 38, 6050 – 6056. 21. Tong, R. C. W., Glavas, N. A., and Bragg, P. D. (1991) Biochim. Biophys. Acta 1080, 19 –28. 22. Hutton, M., Day, J. M., Bizouarn, T., and Jackson, J. B. (1994) Eur. J. Biochem. 219, 1041–1052. 23. Fjellstro¨m, O., Johansson, C., and Rydstro¨m, J. (1997) Biochemistry 36, 11331–11341. 24. Venning, J. D., and Jackson, J. B. (1999) Biochem. J. 341, 329 – 337. 25. Peake, S. J., Venning, J. D., Cotton, N. J. P., and Jackson, J. B. (1999) Biochim. Biophys. Acta 1413, 81–91. 26. Tsukihara, T., Aoyama, H., Yamashita, E., Tomizaki, T., Yamaguchi, H., Shinzawa-Itoh, K., Nakashima, R., Yaono, R., and Yoshikawa, S. (1996) Science 272, 1136 –1144. 27. Yoshikawa, S., Shinzawa-Itoh, K., Nakashima, R., Yaono, R., Yamashita, E., Inoue, N., Yao, M., Fei, M. J., Libeau, C. P., Mizushima, T., Yamaguchi, H., Tomizaki, T., and Tsukihara, T. (1998) Science 280, 1723–1729. 28. Ostermeier, C., Harrenga, A., Ermier, U., and Michel, H. (1997) Proc. Natl. Acad. Sci. USA 94, 10547–10553. 29. Pebay-Peyroula, E., Rummel, G., Rosenbusch, J. P., and Landau, E. H. (1997) Science 277, 1676 –1681.

307

30. Luecke, H., Richter, H. T., and Lanyi, J. K. (1998) Science 280, 1934 –1937. 31. Holmberg, E., Olausson, T., Hultmann, T., Rydstro¨m, J., Ahmad, S., Glavas, N. A., and Bragg, P. D. (1994) Biochemistry 33, 7691–7700. 32. Glavas, N., and Bragg, P. D. (1995) Biochim. Biophys. Acta 1231, 297–303. 33. Glavas, N. A., Hou, C., and Bragg, P. D. (1995) Biochemistry 34, 7694 –7702. 34. Bragg, P. D., and Hou, C. (1999) Arch. Biochem. Biophys. 363, 182–190. 35. Yamaguchi, M., and Hatefi, Y. (1995) J. Biol. Chem. 270, 16653– 16659. 36. Bizouarn, T., Stilwell, S., Venning, J., Cotton, N. P. J., and Jackson, J. B. (1997) Biochim. Biophys. Acta 1322, 19 –32. 37. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, A. J. (1951) J. Biol. Chem. 193, 265–275. 38. Bragg, P. D., and Hou, C. (2000) Arch. Biochem. Biophys. 380, 141–150. 39. Miller, M. J., Oldenburg, M., and Fillingame, R. H. (1990) Proc. Natl. Acad. Sci. USA 87, 4900 – 4904. 40. Wilson, D. M., Hama, H., and Wilson, T. H. (1995) Biochem. Biophys. Res. Commun. 209, 242–249. 41. Loo, T. W., and Clarke, D. M. (1996) J. Biol. Chem. 271, 27482– 27487. 42. Loo, T. W., and Clarke, D. M. (2000) J. Biol. Chem. 274, 5253– 5256. 43. Kubo, Y., Konishi, S., Kawabe, T., Nada, S., and Yamaguchi, A. (2000) J. Biol. Chem. 275, 5270 –5274. 44. Kimura-Someya, T., Iwaki, S., Konishi, S., Tamura, N., Kubo, Y., and Yamaguchi, A. (2000) J. Biol. Chem. 275, 18692–18697. 45. Jung, K., Jung, H., Wu, J., Prive´, G. G., and Kaback, H. R. (1993) Biochemistry 32, 12273–12278. 46. Hu, X., Zhang, J., Fjellstro¨m, O., Bizouarn, T., and Rydstro¨m, J. (1999) Biochemistry 38, 1652–1658. 47. Meuller, J., Hu, X., Bunthof, C., Olausson, T., and Rydstro¨m, J. (1996) Biochim. Biophys. Acta 1273, 191–194.