Mutation of Tyr235 in the NAD(H)-binding subunit of the proton-translocating nicotinamide nucleotide transhydrogenase of Rhodospirillum rubrum affects the conformational dynamics of a mobile loop and lowers the catalytic activity of the enzyme.

The Tyr residue in the mobile loop region of the soluble, domain I polypeptide (called Ths) of the proton-translocating transhydrogenase from Rhodospirillum rubrum has been substituted by Asn and by Phe. The recombinant proteins were expressed at high levels in Escherichia coli and purified to homogeneity. The two well defined resonances at 6.82 and 7.12ppm, observed in the one-dimensional proton NMR spectrum of wild-type protein, and previously attributed to the Tyr residue, were absent in both mutants. In the Tyr235 --> Phe mutant Ths, they were replaced by two new resonances at 7.26 and 7.33 ppm, characteristic of a Phe residue. In both mutants, narrow resonances attributable to Met residues (and in the Tyr235 --> Phe mutant, resonances attributable to Ala residues) were shifted relative to the wild type, but other features in the NMR spectra were unaffected. The conformational dynamics of the mobile loop closure in response to nucleotide binding by the protein were altered in the two mutants. The fluorescence emission from Trp72 was unaffected by both Tyr substitutions, and the fluorescence was still quenched by NADH. The mutant Ths proteins bound to chromatophore membranes depleted of their native Ths with undiminished affinity. In these reconstituted systems, the Km values for thio-NADP+ and NADH, during light-driven transhydrogenation, were similar to those of wild-type, but the kcat values were decreased about 2-fold. In reverse transhydrogenation, the Kmvalues for NADPH were slightly decreased in the mutants relative to wild-type, but those for acetyl pyridine adenine dinucleotide were increased about 10- and 13-fold, respectively, and the kcat values were decreased about 2- and 5-fold, respectively, in the Tyr235 --> Phe and Tyr235 --> Asn mutants. It is concluded that Tyr235 may contribute to the process of nucleotide binding and that substitution of this residue prevents proper functioning of the mobile loop in catalysis.

In animal mitochondria and bacteria, transhydrogenase is driven in the direction of NADP ϩ reduction by the protonmotive force generated through the action of respiratory (or photosynthetic) electron transport chains. REACTION 1 Uniquely in Rhodospirillum rubrum, the NAD(H)-binding domain I of transhydrogenase exists as a separate polypeptide (1)(2)(3). This polypeptide can be expressed in large quantities in Escherichia coli and purified as a water-soluble protein (4). Like the native protein (called Th s ), 1 the recombinant form is dimeric and can restore transhydrogenation activity to everted membranes (chromatophores) of R. rubrum, which have been washed to remove native Th s . Th s binds NADH with a K d of about 20 M (4).
Domain I of transhydrogenase has a mobile loop straddling protease-sensitive sites (Lys 227 -Thr 228 and Lys 237 -Glu 238 in R. rubrum Th s ). It is detectable by NMR, and its conformation is altered when the protein binds nucleotides (5). A Gly-Tyr-Ala motif (residues 234 -236 in the R. rubrum protein) in this region is conserved in all known transhydrogenase sequences. It was proposed that 3,5 and 2,6 ring protons of Tyr in the motif give rise to resonances at 6.82 and 7.12 ppm, respectively, in the NMR spectrum of R. rubrum Th s , and equivalent resonances in the spectrum of E. coli domain I protein (5). Here we test this hypothesis by examining mutants of R. rubrum Th s in which Tyr 235 has been substituted by Asn or Phe. Because the residue is conserved, and approaches nucleotide to within 0.5 nm in domain I-AMP complex (13), Tyr 235 might have a role in catalysis. We examine the effect of Tyr 235 3 Asn and Tyr 235 3 Phe substitutions in Th s on the binding of NADH, as judged by quenching of fluorescence of the lone Trp residue of the protein, on conformational dynamics of the loop during the binding of NAD(H) and analogues as determined by NMR, and on catalytic activity and Michaelis constants of mutant protein in forward (energy-linked) and reverse transhydrogenation after reconstitution with depleted R. rubrum membranes bearing domain II/III proteins of the enzyme. The results are compared with those in which the effect on reverse transhydrogenation activity of mutating the equivalent residue in E. coli enzyme was measured in membrane fractions (6).

In Vitro Manipulation and Analysis of DNA, and Generation of Th s
Mutants-Routine operations, including agarose-gel electrophoresis, preparation of plasmid DNA, growth and handling of E. coli strains, and preparation of competent cells and transformation, were carried out as described (7). Restriction enzymes were used according to manufacturers' instructions.
Mutagenesis was carried out by the gapped duplex method (8). The 700-base pair EcoRI-HindIII fragment of pCD1 (4) was cloned into the EcoRI-HindIII site of pMa (8). Following mutagenesis, the EcoRI-HindIII fragment was cloned back into pCD1 and transformed into E. coli C600. The fragment was then sequenced to check for errors; none were found.
DNA Sequencing-Plasmid DNA was isolated using the Qiagen-tip 20 kit. DNA within the EcoRI-HindIII sites was sequenced with an Applied Biosystems Inc. model 373A sequencer, using the manufacturer's kit employing Taq polymerase and dye terminators, according to their instructions.
Biochemical Procedures-E. coli cells carrying plasmids bearing either wild-type or mutant Th s genes were induced as described (4). Wild-type and mutant Th s were purified by column chromatography (1).
Everted membrane vesicles (chromatophores) from phototrophicallygrown cells of R. rubrum were prepared and depleted of their native Th s as described (1,10). Chromatophores of strains that overexpress transhydrogenase (11) were depleted of Th s by washing 3-4 times in 2 M NaCl, 10 mM Tris-HCl, pH 8.0.
Protein was estimated by the microtannin assay (12), without using the correction procedure determined (4), see (13). Bacteriochlorophyll was measured using the in vivo extinction coefficient (14). SDS-polyacrylamide gel electrophoresis was carried out as described (15).
Assays and Reconstitution Procedures-Reverse transhydrogenation activity of reconstituted membranes was measured at 375-450 nm as reduction of AcPdAD ϩ by NADPH. Forward transhydrogenation activity was measured at 395-450 nm as reduction of thio-NADP ϩ by NADH by membranes irradiated with photosynthetically-active light (16). Extinction coefficients are given (17). Reconstitution of depleted membranes and Th s was carried out by simple mixing followed by a few seconds of incubation at 30°C (18).
Spectroscopy-Fluorescence and NMR spectroscopy of purified Th s were carried out as described (4,5).

One-dimensional Proton NMR Spectra of Tyr 235 3 Asn and
Tyr 235 3 Phe Mutants of Ths-Mutant and wild-type R. rubrum Th s , the domain I polypeptide of transhydrogenase, were expressed in E. coli, and purified. During chromatography on QA-Trisacryl, Reactive-Green-19 agarose, and AcA-44 (1, 4), both mutant proteins displayed similar elution profiles to wildtype Th s (not shown). Notably, the proteins had an apparent molecular mass of approximately 74 kDa by non-denaturing gel filtration, indicating that, like the wild type, they are dimeric. On SDS-polyacrylamide gel electrophoresis, both ran as single polypeptides with an apparent molecular mass of 43 kDa, indistinguishable from wild-type Th s (data not shown; compare Refs. 1 and 4).
NMR spectra of purified wild-type and mutant Th s are shown in Fig. 1. The general characteristics of the spectra, including broad humps of aromatic and methyl protons from 9 -6 ppm and 2.3-0.3 ppm, respectively, were similar, indicating that the overall fold of the mutant proteins was unchanged by the amino acid substitutions. Like wild-type Th s , those of the mutants' spectra displayed a number of narrower resonances superimposed upon the broad methyl and aromatic proton absorptions. Some or all of these resonances probably derive from amino acids in a mobile loop that straddles protease cleavage sites at the protein surface (5,13). On the basis of the effect on the NMR spectrum of cleavage of Th s by trypsin at Lys 227 -Thr 228 and Lys 237 -Glu 238 , the particularly clear resonances at 6.82 and 7.12 ppm in wild-type Th s were assigned to 3,5 and 2,6 ring protons of Tyr 235 (5). These resonances were absent in both the Tyr 235 3 Asn and Tyr 235 3 Phe mutants (Fig. 1). Evidently, resonances from other Tyr residues (at positions 120, 146, and 154) are very broad, as expected for residues in a molecule having the correlation time of an 80-kDa protein like the Th s dimer. The appearance of resonances in the Tyr 235 3 Asn mutant, between 2.6 and 2.8 ppm, corresponding to the ␤CH 2 of the Asn residue introduced into the loop, was obscured by those from dithiothreitol (and its breakdown products) in the sample. However, in the Tyr 235 3 Phe mutant, new narrow resonances were observed at 7.26 and 7.33 ppm, characteristic of ring protons of Phe. These resonances were considerably sharper than those in wild-type protein at 7.33 ppm, which were also attributed to Phe (possibly Phe 218 ).
Unexpectedly, the two Tyr 235 substitutions both had effects on resonances previously assigned to the CH 3 groups of Met residues (Fig. 1). The NMR spectra of wild-type Th s at 20°C reveal sharp resonances attributable to Met residues at 1.97 and 2.04 ppm with shoulders at 2.06 and 2.08 ppm (5,13); they are designated MetA, MetB, MetC, and MetD, respectively (Table I). In the Tyr 235 3 Asn mutant, MetB was unaffected, but MetA was shifted downfield by 0.04 ppm, and MetC was shifted 0.02 ppm downfield to overlap with MetD at 2.08 ppm. In the Tyr 235 3 Phe mutant, MetB again was unaffected, MetA was again shifted downfield (but only by 0.02 ppm), and a slight downfield shift of MetC resulted in a peak at 2.065 ppm with a shoulder at 2.08 ppm (MetD, Table I). In the Tyr 235 3 Phe mutant, but not the Tyr 235 3 Asn mutant, the region of the Ala resonances was resolved into two distinct components (each possibly comprising one or more Ala ␤CH 3 doublets). Note that trypsin treatment of wild-type protein also resulted in separation of two Ala components (5); the mobile loop region includes several Ala residues.
Resonances other than those assigned to Tyr and Met (and, in the Tyr 235 3 Phe mutant, Ala) were indistinguishable in mutant and wild-type proteins. Thus, the CH 3 of Thr at 1.25 ppm and ring protons of Phe at 7.37 ppm (although these were obscured in the Tyr 235 3 Phe mutant), the tentative Gly CH 2 at 3.96 ppm and Glu ␥CH 2 at 2.31 ppm, the resonances from unassigned amino acid residues at 7.6 -7.9 ppm, fine structure superimposed on the broad methyl absorption at 0.8 -1.0 ppm, and the ring-shifted methyl protons at approximately 0.17,  The Effect of NAD ϩ , NADH, and Analogues on NMR Spectra of Mutant Th s -The conformational dynamics, as revealed by NMR, of amino acid residues in the mobile loop of wild-type-Th s during binding of NAD(H) and analogues have been discussed (5,13). NMR spectra recorded in titrations of the Tyr 235 3 Asn and Tyr 235 3 Phe mutants with NADH and NAD ϩ are shown in Figs. 2 and 3, respectively. They reveal some differences from data for wild-type Th s .
In wild-type Th s a two-step binding reaction is revealed in NMR spectra recorded during nucleotide titrations (5,13). It is characterized by specific broadening of MetA at low concentrations of nucleotide, followed, at higher concentrations, by broadening of other resonances assigned to the mobile loop. In titrations with NADH the two-step reaction is barely perceptible at 20°C, although easily resolved at 37°C (13). In the Tyr 235 3 Phe mutant Th s (Fig. 3), a similar sequence of events was observed to that in wild-type. Two differences were: (a) the new Phe resonances at 7.26 and 7.33 ppm broadened during the titration in the same way that the Tyr resonances broadened in wild-type protein, and (b) of the two Ala resonances, split in the mutant, the more upfield was more sensitive to broadening by NADH. Probably because in the Tyr 235 3 Asn mutant the MetA resonance is displaced downfield in the absence of nucleotides (see Table I), the two-step binding reaction was clearly observed in NADH titrations even at 20°C (Fig. 2). Thus, 30 M NADH led to more extensive broadening of MetA than, for example, the Ala or the Thr resonances. Higher concentrations (200 M) did lead to broadening of the latter. The dependence of resonance broadening on NADH concentration in both mutants was similar to that with wild-type protein. NMR spectra recorded in titrations of mutant Th s with the analogue, AcPdADH, were qualitatively similar to those from NADH titrations (not shown).
Also reflecting ligand-protein interaction, linewidths of the NADH (and AcPdADH) resonances remained broad during titration against both of the mutant proteins until added nucleotide reached concentrations approaching 10 Ϫ3 M (i.e. in considerable excess of protein concentration). Similar behavior was observed with wild-type protein and was suggested to result from decreased mobility of NADH in its protein-bound state and to an intermediate/fast exchange (5).
In titrations of the mutant proteins with NAD ϩ , the two-step reaction observed with wild-type protein (5) was again evident. Thus, moderately low concentrations of oxidized nucleotide had a specific effect on the MetA resonance, before other mobile loop resonances were broadened (Figs. 2 and 3). In the Tyr 235 3 Asn protein, in which the MetA resonance was displaced downfield (see above), addition of NAD ϩ led, not only to broadening, but also to a shift back upfield that was more extensive than that in wild-type Th s (Fig. 2; compare Ref. 5). Whereas the concentration dependence of resonance broadening was similar in NADH titrations for wild-type Th s and for mutant proteins, this was not the case with NAD ϩ , where higher concentrations were required for both mutants to give the response observed in the wild type.
In marked contrast to the considerable broadening of NADH resonances in the presence of either wild-type (5, 13) or mutant Th s (above), NAD ϩ resonances became evident in the wild-type titration spectra even at quite low concentrations, consistent with a higher K d value for oxidized nucleotide and faster exchange. In titrations with Tyr 235 3 Asn and Tyr 235 3 Phe mutants (Figs. 2D and 3D), NAD ϩ proton resonances were detectable at even lower concentrations of nucleotide, providing another indication of its weaker binding.
As with wild-type Th s , NMR spectra of the mutants titrated with AcPdAD ϩ and with 5Ј-AMP were similar to those with NAD ϩ (not shown). Notably, they revealed a two-step binding process; the MetA resonance was affected at lower concentrations of nucleotide than the other loop resonances. As with NAD ϩ , higher concentrations of both AcPdAD ϩ and 5Ј-AMP were required with both mutant proteins to produce an equivalent broadening of the narrow resonances, and again, nucleotide resonances were resolved at lower concentrations in the mutant than in wild-type titrations. 3 Asn and Tyr 235 3 Phe mutants was quenched upon addition of NADH. The dependences of fluorescence quenching on nucleotide concentration were broadly similar to that in wild-type (Fig. 4). Note, however, that because the K d is quite high, this analysis does not sensitively detect decreases in binding affinity. In stopped flow experiments (data not shown), it was observed that the time course of Trp 72 fluorescence quenching by NADH was similar in the Tyr 235 3 Asn mutant to that for wild-type Th s (4). Judging by the quenching of Trp fluorescence, wild-type Th s bound AcPdADH with a lower affinity than NADH (13). Fig. 4 shows that the Tyr 235 3 Asn and Tyr 235 3 Phe Th s mutants also bound AcPdADH with a low affinity. Because of limitations imposed by inner filtering by the nucleotide (see above), it was difficult to estimate precise K d values. In common with wild-type Th s , addition of either NAD ϩ or NADPH led to no fluorescence quenching up to about 60 M.

The Effect of Substituting Tyr 235 of Th s with Asn and with Phe on Catalytic Properties of Reconstituted Transhydrogenase
Complex-Isolated Th s does not catalyze transhydrogenation between NADPH and AcPdAD ϩ (1,4) or between NADH and AcPdAD ϩ (data not shown). The suggestion (19) that reduction of AcPdAD ϩ by NADH in intact transhydrogenase takes place between the domain I polypeptides is therefore unlikely. Other explanations for this reaction are possible (20,21). We have only observed enzyme activity of Th s when it is associated with domain II/III components of transhydrogenase.
The ability of wild-type and mutant Th s to reconstitute reverse transhydrogenation activity to R. rubrum membranes depleted of native Th s is compared in Fig. 5. Depleted membranes were prepared by salt washing chromatophores isolated from a strain of R. rubrum that overexpresses wild-type transhydrogenase (see "Materials and Methods"). Experiments were performed with close-to-saturating concentrations of nucleotide substrates. Rates of reverse transhydrogenation with the Tyr 235 3 Asn and Tyr 235 3 Phe mutants of Th s were about 18% and 44%, respectively, of wild-type protein, but docking affinities revealed by double-reciprocal plots (data not shown) were undiminished.
Dependences of the rate of reverse transhydrogenation on the concentration of AcPdAD ϩ (saturating NADPH) in the reconstituted systems of depleted chromatophores, and either wild-type or mutant Th s , are shown in Fig. 6A. Double-reciprocal plots (not shown) yielded K m values for AcPdAD ϩ of approximately 800, 600, and 60 M in the Tyr 235 3 Asn and Tyr 235 3 Phe mutants and wild type, respectively.
Because the K m values for AcPdAD ϩ in the mutants were high, it was not practicable to carry out experiments with saturating concentrations of this nucleotide. Thus, Fig. 6B shows the dependence of reverse transhydrogenation rate on NADPH concentration at 1.1 mM AcPdAD ϩ . Double-reciprocal plots (not shown) gave an approximate apparent K m for NADPH of 15 M for the Tyr 235 3 Asn mutant, 15 M for the Tyr 235 3 Phe mutant, and 30 M for wild-type Th s .
Rates of light-driven reduction of thio-NADP ϩ by NADH (forward transhydrogenation) in depleted chromatophores reconstituted, either with mutant or wild-type Th s , were also investigated (Fig. 7). The light drives photosynthetic electron transport, generating a proton electrochemical gradient, which leads to enhanced proton flux through transhydrogenase in its physiological (forward) direction. Depleted membranes were prepared from wild-type chromatophores washed under mild conditions to remove Th s whilst preserving coupling activity. Because the overexpressing strain could not be used, the level of accuracy was lower than in Fig. 6. Fig. 7 shows that maxi- mum rates of light-driven forward transhydrogenation in reconstituted systems were about 2-fold lower for both mutants than for wild-type Th s . Dependences of rates of forward transhydrogenation on nucleotide concentrations (Fig. 7, A and B) show that in both mutants the K m values for thio-NADP ϩ and for NADH were not significantly different from wild-type K m values of approximately 5 and 4 M, respectively. Fig. 8 shows that mutant Th s proteins displaced the wildtype protein from its binding site on chromatophores. Addition of wild-type Th s to chromatophores led to a small increase in the rate of reverse transhydrogenation, presumably because some domain I protein was lost from the membranes during preparation (1,4). However, addition of either the Tyr 235 3 Asn or the Tyr 235 3 Phe mutant Th s resulted in substantial loss of activity, indicating that association-dissociation of domain I with domains II/III of transhydrogenase can occur on the time scale of the experiment. DISCUSSION By substituting Tyr 235 of wild-type Th s with Phe and Asn, we have tested our prediction that 1 H NMR signals at 6.82 and 7.12 ppm in the wild-type protein are attributable to that residue. Complete loss of those resonances from the spectra of both mutants unambiguously confirms the assignment. New, well defined resonances at 7.26 and 7.33 ppm in the Tyr 226 3 Phe mutant are characteristic of Phe ring protons, and further indicate that the Phe has adopted the mobile nature of the original Tyr. The emergence of new resonances in the Tyr 226 3 Asn mutant was masked by dithiothreitol present in the sample. Substitution of Tyr 235 for Asn also led to changes in resonances assigned to Met residues, notably a marked downfield shift of MetA and a smaller shift in MetC. The assignment of these residues is not yet possible, but is pertinent because the behavior of the MetA resonance reflects events at an intermediate stage of nucleotide binding (Refs. 5 and 13; see below). It was suggested (5) that MetA might derive from Met 239 , and experiments are now in progress to test this. Substitution of Tyr 235 for Phe also led to shifts of the MetA and MetC resonances and to separation of Ala resonances. The fact that the amino acid residue at position 235 influences the NMR-detectable Met and Ala residues indicates that there is structural organization in the loop even in the absence of nucleotides, but the nature of the interactions is not understood. Evidently protons of the Ala, MetA, and MetC residues can sample more than one environment on the NMR time scale; the effect of the Tyr 235 substitution might be to alter the exchange rate between different conformations, or it might result in changes in the chemical shift of the Met and Ala resonances in one of the conformational states, e.g. by altering positions of charged or aromatic groups relative to methyl groups of the amino acid residues.
The change from Tyr 235 to Asn or to Phe in Th s is not accompanied by gross changes in molecular structure; the protein retains its ability to form dimers and to dock with the domain II/III components of transhydrogenase, the short-wavelength emission of Trp 72 is preserved, and, on the basis of NMR spectra, the protein fold and environments of amino acids in the mobile loop (with the exception of MetA, MetC, and Ala residues) are unaffected. Thus, effects of the mutations on nucleotide binding and catalytic properties of the enzyme are likely to be a direct consequence of altered properties of the loop.
For both mutants, higher concentrations of NAD ϩ than for wild-type Th s were required to broaden resonances ascribed to the mobile loop. This might mean either (a) that the K d for NAD ϩ is increased by the amino acid substitution, or (b) that differences in exchange rate(s) between Th s , Th s -NAD ϩ , and Th s * -NAD ϩ (see (5)) alter linewidths without affecting the affinity for nucleotide. Similar observations and interpretations apply also to AcPdAD ϩ and 5Ј-AMP in wild-type and mutants. The very large K m for AcPdAD ϩ of both the mutant proteins, relative to the wild-type, during reverse transhydrogenation (after reconstitution with Th s -depleted membranes) might be another indication of increased K d for oxidized nucleotide. Thus it is possible that Tyr 235 contributes to the binding affinity of Th s for NAD ϩ , AcPdAD ϩ , and 5Ј-AMP. This is consistent with the observation that, in the two-dimensional 1 H NMR spectrum of wild-type 5Ј-AMP-Th s complex, NOE interactions were detected between Tyr 235 and bound nucleotide (13).
It cannot be determined with confidence whether or not the Tyr 235 3 Asn or Tyr 235 3 Phe mutations had a significant effect on binding of reduced nucleotides by Th s . There were no clear differences between the mutant and the wild-type proteins in either the K m values for NADH in forward transhydrogenation (Fig. 7) or the dependences on NADH of the protein-NMR spectra (Figs. 2 and 3), but neither of these give an unambiguous indication of the K d . The quenching of Trp 72 fluorescence by reduced nucleotides (Fig. 4) must also be interpreted with care; because of inner filtering, K d becomes more subject to error as its value increases beyond 10 -20 M. However, it is reasonable to conclude that substituting Tyr 235 with either Asn or Phe does not have a large effect on the affinity of Th s for NADH or AcPdADH. The fact that, in NADH titrations, the new Phe resonances in the Tyr 235 3 Phe mutant broadened in a similar way to Tyr resonances in wild-type Th s , indicates that the residue can participate in mobile loop closure either with, or without, the 4-OH group. There were minor differences in the NMR spectra recorded in NADH titrations that arose from the fact that alteration of Tyr 235 caused shifts in Met and Ala resonances (see above), but evidently the perturbations were not enough greatly to affect binding affinity or loop closure.
Mutation of Tyr 235 to either Asn or Phe led to decreases in k cat for both forward and reverse transhydrogenation, when the Th s was reconstituted with depleted membranes (Figs. 6 and  7). The loop clearly has a role in catalysis in addition to its fine-tuning effect on the binding affinity of the protein for nucleotide. Whether this is in the hydride transfer reaction, or in conformational coupling with domain II/III components of transhydrogenase, is not known. Although the Tyr 235 3 Phe mutant Th s had a substantially decreased activity, the Tyr 235 3 Asn protein was considerably more inhibited, particularly in reverse transhydrogenation. This might indicate that both the aromatic ring and the 4-OH group of Tyr 235 are important in the conformational dynamics of the loop in catalysis by wildtype protein.
The equivalent of Tyr 235 in bovine transhydrogenase (Tyr 245 ) is sensitive to modification by 5Ј-[p-(fluorosulfonyl)benzoyl]adenosine (22). Modified enzyme had reduced catalytic activity; NADH protected against modification. On the basis of this, the equivalent residue in E. coli transhydrogenase (Tyr 226 ) was substituted with His, Leu, Phe, and Asn (6). In crude membrane fractions isolated from bacteria carrying the mutation, specific activities (mg Ϫ1 membrane protein) of AcPdAD ϩ reduction by NADPH were 33% (or 51% in another strain), 38%, 45% and 42%, respectively, lower than rates in membrane fractions prepared from bacteria carrying wild-type transhydrogenase gene; K m values for AcPdAD ϩ for the His, Leu, and Phe mutant membranes were, respectively, 3-, 1.9-, and 3-fold larger than those in wild-type membranes. Comparison between mutant and wild-type transhydrogenase in that study is complicated by uncertainty about the level of expression of the enzyme; the transhydrogenase content of the bacterial membranes was assessed from their appearance on SDS-polyacrylamide gels. Nevertheless, results of those experiments are broadly consistent with results reported here. In our experiments the statistical significance is assured because we used Th s purified to homogeneity, and the same preparation of membranes for reconstitutions with both wild-type and mutant proteins. Because the domain I protein of E. coli does not exist as a discrete polypeptide, this strategy is unavailable in that system. In our experiments, reverse transhydrogenation activities of Tyr 235 mutants of R. rubrum transhydrogenase were more inhibited, relative to wild-type, than Tyr 226 mutants of E. coli transhydrogenase, and K m values of the mutants were increased by a much larger factor. There might also be real species differences between the two enzymes. The GYA motif of the loop (5) is conserved in known transhydrogenase sequences, but there is only low homology among other residues in the region; there are no other invariant amino acid residues, although small and charged residues preponderate. There is often greater variation in amino acid sequence of surface loops because individual residues make only a small contribution to the global structure. It is likely in transhydrogenase that, during closure, multiple contacts are made between the loop and the rest of the protein or bound nucleotide, and therefore, in individual species, single amino acid substitutions in the loop might have greater or lesser effects on catalysis. Uniquely in the R. rubrum enzyme effects of mutations on nucleotide binding, interaction between domains, conformational dynamics of the mobile loop, and the Michaelis parameters can be assessed separately. The present experiments establish that Tyr 235 is important in the dynamics of mobile loop closure, and that substitution of the residue profoundly affects the Michaelis parameters, thus, for the first time, establishing a pivotal role for loop closure in catalysis.