Quantitative determination of direct binding of b subunit to F1 in Escherichia coli F1F0-ATP synthase.

The stator in F(1)F(0)-ATP synthase resists strain generated by rotor torque. In Escherichia coli, the b(2)delta subunit complex comprises the stator, bound to subunit a in F(0) and to the alpha(3)beta(3) hexagon of F(1). To quantitatively characterize binding of b subunit to the F(1) alpha(3)beta(3) hexagon, we developed fluorimetric assays in which wild-type F(1), or F(1) enzymes containing introduced Trp residues, were titrated with a soluble portion of the b subunit (b(ST34-156)). With five different F(1) enzymes, K(d)(b(ST34-156)) ranged from 91 to 157 nm. Binding was strongly Mg(2+)-dependent; in EDTA buffer, K(d)(b(ST34-156)) was increased to 1.25 microm. The addition of the cytoplasmic portion of the b subunit increases the affinity of binding of delta subunit to delta-depleted F(1). The apparent K(d)(b(ST34-156)) for this effect was increased from 150 nm in Mg(2+) buffer to 1.36 microm in EDTA buffer. This work demonstrates quantitatively how binding of the cytoplasmic portion of the b subunit directly to F(1) contributes to stator resistance and emphasizes the importance of Mg(2+) in stator interactions.

F 1 F 0 -ATP synthase is the enzyme responsible for ATP synthesis by oxidative phosphorylation or photophosphorylation in mitochondria, chloroplasts, and bacteria and for ATP-driven generation of proton or sodium gradients in bacteria. Even before the discovery that it acts a molecular rotary motor (1), it was clear that this was a protein of considerable structural complexity. In Escherichia coli, the F 1 moiety lies peripheral to the membrane, with the subunit composition ␣ 3 ␤ 3 ␥␦⑀, and the F 0 moiety, consisting of subunits ab 2 c n (where n is not established at time of writing), is intrinsic to the membrane. Subunit complements in higher organisms can be considerably more complex, and there is notable variation in the number of c subunits between species (2)(3)(4)(5).
Until 1994, the consensus based on electron microscopy (for one example, see Ref. 6) and biochemical studies (for one example, see Ref. 7) was that F 1 was bound to F 0 by a single "central stalk" presumed to contain ␥, ⑀, ␦, and b subunits. The x-ray crystallography studies of F 1 by Walker and colleagues (8,9) changed that view by showing that the central stalk did contain ␥ and ⑀ subunits, but contrary to previous ideas, could not accommodate ␦ and b subunits for lack of space. Subsequently, electron microscopy studies (10) and biochemical studies (11)(12)(13)(14) have established that the ␦ and b subunits form a separate stalk structure, at the periphery of F 1 . With the emerging view of ATP synthase as a rotary motor, the central stalk is now referred to as the rotor (consisting of ␥,⑀ and the ring of c subunits) and the peripheral stalk (consisting of ␦ and b subunits) as the stator.
The role of the stator (b 2 ␦) is to hold one portion of the enzyme static (␣ 3 ␤ 3 a) while the other portion rotates (␥⑀c ring ), thus allowing the very efficient transfer of energy between the proton "channel" and the catalytic sites (15). Rotary strain can amount to Ͼ50 kJ/mol (15,16); therefore, strong interactions between stator subunits are required. ␦ subunit is known to interact with the N-terminal region of ␣ subunit at the very "top" of F 1 (i.e. most distant from the membrane surface) (17)(18)(19). The binding affinity (K d ) between ␦ and F 1 is ϳ1 nM, equivalent to 50 kJ/mol (20 -22) when measured directly, and is increased considerably, to possibly K d ϳ1 pM, when the cytoplasmic portion of the b subunit is included in the binding assay (22). The N-terminal domain of ␦ interacts with ␣ (19,23). The b subunit interacts with the C-terminal domain of ␦ (24 -26), with the C-terminal residues of the b subunit playing an important role. It is an interesting question whether the b subunit also interacts directly with F 1 ␣ and/or ␤ subunits. There is evidence that this is the case. For example, crosslinking studies showed that Cys residues introduced at positions 92, 109, or 110 in b could be cross-linked to ␣ and/or ␤ subunits (27), and other cross-linking studies indicated close proximity of b to ␣ (28). Native gel electrophoresis studies of F 1 and the cytoplasmic portion of b showed formation of a complex (24). Also, the addition of the cytoplasmic portion of the b subunit altered the conformation of the F 1 catalytic sites as shown by spin-labeling studies (29). However, these approaches could not quantitatively determine the strength of interaction between b and F 1 .
In this report, we titrated soluble F 1 containing introduced Trp residues in the ␣ or ␤ subunit with a soluble cytoplasmic portion of the b subunit. By following the changes in fluorescence, we ascertained the K d for direct interaction between b and F 1 . We then studied the influence of Mg 2ϩ cation and of the presence or absence of ␦ subunit on the b/F 1 interaction. In related studies, we studied the effects of Mg 2ϩ cation on the ability of the cytoplasmic portion of b subunit to promote binding of ␦ subunit to F 1 , and we introduced Trp residues into the b subunit at residues 92 and 109 to test the effects on interaction with F 1 .

EXPERIMENTAL PROCEDURES
Purification of F 1 , Purification of ␦ subunit, Preparation of ␦-depleted F 1 , Purification of b  , Competition Assay of Binding of b  to F 1 Using ATP-driven Proton Pumping in Reconstituted Membrane Vesicles, Routine Procedures-These were all as described previously (21,22).
Fluorescence Binding Assays-The buffer for fluorescence titrations was 50 mM HEPES/NaOH, pH 7.0, room temperature, with either 5 mM MgSO 4 or 0.5 mM EDTA. Excitation was at 295 nm. For measurement of binding of purified ␦ subunit to ␦-depleted F 1 , the ␦-depleted F 1 was the ␤W107 F 1 from strain pSWM86/DK8 as described previously (21). ␤W107 F 1 contains only the native Trp at residue ␤-107, it has normal functional properties, and it has a low Trp background signal in the ␦-binding assay. Binding of ␦ subunit was monitored by enhancement of fluorescence at 325 nm (21,22). For measurement of binding of b ST34 -156 1 to F 1 , a series of different F 1 enzymes was used in which novel Trp residues had been inserted into otherwise Trp-free F 1 (see below). The fluorescence signal at 335 nm was monitored. For most of the enzymes used, this was close to max. No significant shift (Ͼ3 nm) in max occurred upon the addition of b ST34 -156 to any F 1 . It was ascertained that the presence of EDTA or Mg 2ϩ had no effect on F 1 fluorescence spectra, nor did depletion of the ␦ subunit. F 1 concentration in the cuvette was varied from 100 to 500 nM. b ST34 -156 was added incrementally, usually up to 1.5 M concentration (equivalent to 0.75 M b 2 dimer) but higher when necessary to reach saturation. All of the enzymes are expected from previous work to have fully occupied noncatalytic sites. To be sure that the signal changes seen on the addition of b ST34 -156 were not due to changes in nucleotide occupancy of catalytic sites, we conducted experiments on enzyme that had been passed consecutively through two 1-ml Sephadex G-50 centrifuge columns in Tris/ SO 4 , pH 8.0, which removes all catalytic site nucleotide while leaving the noncatalytic sites fully occupied (30). The results were not altered in any way by this precaution. All fluorescence values were routinely corrected for the small Trp contamination in purified b ST34 -156 (see "Results"). K d values for binding of b ST34 -156 and ␦ to F 1 were determined by nonlinear regression analysis (21,22,31). Concentration of b ST34 -156 was expressed as the dimer concentration, since the dimer is the functionally active species (28,(32)(33)(34)(35)(36)(37).
E. coli Strains and b  Mutagenesis-Strains for purification of wild-type and mutant ␦ subunits were described in Refs. 21 and 22. Strains for purification of Trp-containing F 1 enzymes were described in Refs. 38, 39, and references therein. Wild-type F 1 was from strain SWM1 (40), and ␦W28L F 1 was from strain pSWM92/DK8 (21). Oligonucleotide-directed mutagenesis (41) of b ST34 -156 was used to generate the mutations bA92W and bI109W. The template was M13mp19 containing the EcoRI-HindIII fragment from plasmid pJB3 (34), which contains a synthetic gene encoding a Met-Ser-Thr leader sequence followed by residues 34 -156 of subunit b. The oligonucleotide for bA92W was 5Ј-GGACGAAGCGAAATGGGAGGCTGAGCAGGAACGT-AC-3Ј, in which the italicized bases encode the Trp mutation and the underlined bases introduce a BlpI site. The oligonucleotide for bI109W was 5Ј-GGCGCAGGCGGAATGGGAAGCCGAGCGTAAACGTGC-3Ј, in which the italicized bases encode the Trp mutation and the underlined base deletes a BlpI site. Mutant clones were identified by BlpI digestion, and the entire inserts were sequenced to verify the presence of introduced Trp residues and absence of unwanted base changes. Mutations were then moved into pJB3 on EcoRI-HindIII fragments, and the resultant plasmids pSWM128 (bA92W ST34 -156 ) and pSWM129 (bI109W ST34 -156 ) were transformed into strain DK8 (42) for expression and purification of the mutant b ST34 -156 .

Fluorescence Titration of F 1 with the Soluble Cytoplasmic
Portion of the b Subunit-The cytoplasmic portion of the b subunit used here was b ST34 -156 , as described in Ref. 34. It consists of residues 34 through 156 (C terminus) of the b subunit with an additional Ser-Thr-sequence at the N terminus. We had previously confirmed (22) that it yields a single band on SDS gels after purification in soluble form and that it is fully functional as determined by the assay of Dunn (32) in which binding of F 1 to stripped E. coli membranes is competed for by added b ST34 -156 . Although theoretically b ST34 -156 contains no Trp residue, we found from its fluorescence spectrum in 6 M guanidine chloride that Trp contamination was present, in the amount of 0.09 mol/mol, and we routinely corrected for this. Titrations were carried out in either 5 mM MgSO 4 -or 0.5 mM EDTA-containing buffer.
The fluorophore in the titrations was either the native Trp residues in wild-type F 1 or Trp inserted at a single residue position in otherwise Trp-free F 1 . Wild-type F 1 contains nine native Trps, comprised of three in ␣, three in ␤, two in ␥, and one in ␦ subunit. We found that there was fluorescence enhancement on the addition of b ST34 -156 of maximally 8.6% (average of six experiments). This allowed calculation of K d values for binding of b ST34 -156 . Since the fluorescence enhancement seen in the ␤W107 enzyme (below) was ϩ5.3%, it appears that this residue contributes significantly to the fluorescence enhancement seen in wild-type F 1 . A series of F 1 variants in which all the native Trp had been substituted (43) and a novel Trp had been incorporated at a single residue position in the ␣ or ␤ subunit was also tested. In each case, we confirmed that an E. coli strain containing the mutant enzyme showed normal growth on succinate plates and normal growth yield in limiting glucose, to ensure that enzyme function was not impaired. None of these F 1 enzymes showed a fluorescence quench. Several of the enzymes gave no change in fluorescence when titrated with b ST34 -156 at concentrations up to 3.0 M and F 1 concentration of 100 nM. These were: ␣F291W, ␣I346W, ␤F17W, ␤Y297W, and ␤R398W. A further group gave a fluorescence enhancement of Յ6% at the highest b ST34 -156 concentration. These were: ␣F409W, ␤F148W, ␤Y331W, and ␤W107 (the last contains just the single natural Trp at residue ␤-107, with the other natural Trp residues substituted (21)). A third group of F 1 enzymes gave larger fluorescence changes on the addition of b ST34 -156 . These were ␣L55W (ϩ14.5%), ␣F406W (ϩ8.5%), ␤Y26W (ϩ16%), and ␤F410W (ϩ24%) (average values for between 4 and 20 experiments). We decided to use this last group for determination of K d values.
Typical titration curves for binding of b ST34 -156 to wild-type, ␣L55W, ␣F406W, ␤Y26W, and ␤F410W F 1 in the presence of Mg 2ϩ are shown in Fig. 1, A-E, filled symbols. Calculated K d values are shown in Table I. There was good agreement among K d calculated for all five F 1 enzymes, with the values being ϳ100 -150 nM. One would not expect that the introduced Trp residues in the mutant F 1 would themselves perturb K d significantly since all of the enzymes used show normal oxidative phosphorylation in vivo as judged by growth tests.
Effect of ␦ Subunit on Binding of b ST34 -156 to F 1 -Two approaches were used to investigate the effect of ␦ subunit on binding of b ST34 -156 to F 1 . First, we prepared ␦-depleted F 1 following the procedure in Ref. 21. This was done for several of the enzymes including ␣L55W, ␣F406W, ␤Y26W, ␤F410W, and wild type. In all cases, the absence of ␦ was checked on SDS gels. We then titrated the ␦-depleted F 1 with b ST34 -156 and found that the titration curves were essentially the same as for the corresponding intact F 1 , yielding similar K d values. Typical titration curves are shown in Fig. 2, A and B, for the ␤F410W and ␤Y26W enzymes, respectively. In a second approach, we added pure ␦ subunit containing the ␦W28L mutation at 500 nM concentration (i.e. 5-fold excess over F 1 ) to either intact F 1 or ␦-depleted F 1 and then carried out the b ST34 -156 titrations. Again, no change in the titration curve was seen (data not shown). Therefore, the presence or absence of ␦ did not affect the titration curves or the K d values.
Effect of EDTA on Binding of b ST34 -156 to F 1 -When 0.5 mM EDTA was included in the buffer instead of Mg 2ϩ , there was a clear effect upon the titration curves, as shown in Fig. 1, open symbols. For all of the enzymes, the degree of fluorescence enhancement was much reduced in the presence of EDTA, although there was indication of residual binding, and it appeared that the K d value was increased to Ն1 M. Of the five enzymes shown in Fig. 1, open symbols, only the ␤F410W enzyme had a signal enhancement in EDTA that was large enough (ϳ5%, Fig. 1E) to enable us to attempt to determine a K d value. By using higher concentrations of F 1 and b ST34 -156 , we determined a K d value of 1.25 M (average of four experiments using 250 -500 nM F 1 , K d value range was 870 nM to 1.71 M). A typical experiment is shown in Fig. 3.
Effect of b  to Increase Affinity of Binding of Isolated ␦ Subunit to ␦-depleted F 1 in Presence of EDTA-We had previously shown that b ST34 -156 strongly promotes the binding of isolated ␦ subunit to ␦-depleted F 1 in the presence of Mg 2ϩ (22). In that work, we had used a mutant ␦ subunit preparation (␦Y11W/W28L) that binds relatively weakly in the absence of added b ST34 -156 to see the effect. (With wild-type ␦, the K d is already so low without added b ST34 -156 that little effect is seen upon the addition of b ST34 -156 due to technical limitations of the assay.) The K d for binding of ␦Y11W/W28L ␦ was reduced from 0.5 M in the absence to Յ5 nM in the presence of b ST34 -156. The same effect was seen with other weakly binding mutant ␦ preparations (22). Here we repeated those experiments in EDTA-containing buffer. The enhancement of fluorescence of the ␦-Trp-11 residue upon binding of ␦Y11W/W28L subunit to ␦-depleted ␤W107 F 1 was measured (see "Experimental Procedures"). It was first confirmed that the addition of b ST34 -156 to the ␦Y11W/W28L subunit had no effect on ␦-Trp-11 fluorescence in the absence of F 1 . It is apparent (Fig. 4) that in the presence of EDTA, the Y11W/W28L mutant ␦ preparation showed essentially no binding to F 1 in the absence of b ST34 -156 , but in the presence of b ST34 -156 , there was measurable binding, and the calculated K d value for binding of ␦ was 745 nM (mean of five experiments). We confirmed that b ST34 -156 had the same effect to increase the affinity of other weakly binding mutant ␦ preparations (␦Y11A, ␦V79W/W28L, ␦V79A; data not shown).
To calculate the apparent K d for the effect of b ST34 -156 to increase the affinity of ␦-binding in EDTA, we carried out titrations in which b ST34 -156 was added incrementally to ␦-depleted F 1 (500 nM) in the presence of constant Y11W/W28L mutant ␦ (1 M). Control experiments showed that under these conditions, there was essentially no change in the fluorescence of F 1 due to the addition of b ST34 -156 when added ␦ was absent. Fig. 5 shows a typical experiment. From three such experiments, an average K d (apparent) for b ST34 -156 of 1.36 M was calculated. We also varied the concentration of added ␦ from 1 to 5 M; this did not significantly affect the result. Previously, in Mg 2ϩ -containing buffer, K d (apparent) of 150 nM for b ST34 -156 was reported (22). The effect of EDTA is therefore substantial, and we may conclude that binding of b ST34 -156 is much tighter in the presence of divalent cation.
Insertion of Trp at Positions 92 and 109 of b Subunit-Previous work has shown that Cys residues, inserted at positions Ala-92 and Ile-109 of the b subunit, may be cross-linked FIG. 1. Fluorescence titration of wild-type, ␣L55W, ␣F406W, ␤Y26W, and ␤F410W F 1 with b ST34 -156 . Experiments using 100 nM F 1 are shown; higher concentrations were also used. Excitation was at 295 nm, and emission was measured at 335 nm. The solid lines are fits to the data by nonlinear regression analysis. Calculated K d values for binding are given in Table I  by a photoactivated cross-linking reagent to residues in the ␣ and/or ␤ subunits of F 1 (27), suggesting that these residues of b subunit, or the region around them, might normally be close to or directly interact with F 1 . Here we inserted Trp at these residues in b ST34 -156 to generate bA92W ST34 -156 and bI109W ST34 -156 . Both variant forms of b ST34 -156 could be purified in the same yield and purity as for parent b ST34 -156 . Fig. 6 shows the results of competition assays in which binding of wild-type F 1 to stripped membranes containing F 0 was measured in the presence of increasing concentrations of mutant b ST34 -156 . It is seen that although bA92W ST34 -156 competed as well as wild-type b ST34 -156 , bI109W ST34 -156 did not compete, demonstrating that the substitution of Trp at residue Ile-109 abrogates binding to F 1 , whereas a Trp at residue Ala-92 had no effect. It was also confirmed, using the same assay, that bA92W ST34 -156 and b ST34 -156 bound equally well to Trp-free F 1 (data not shown). As we discussed earlier (21), this assay can-not give absolute values of K d for b ST34 -156 binding; hence no fits to the curves were attempted in Fig. 6.
Both bA92W ST34 -156 and bI109W ST34 -156 had substantial fluorescence signals, with emission maxima at 347 and 342 nm, respectively. However, when bA92W ST34 -156 was added to Trpfree F 1 , only a small enhancement of the Trp-92 fluorescence signal was seen (maximally ϩ3%), which was too small for determination of K d values. DISCUSSION The first goal of this work was to determine whether and with what affinity the b subunit binds directly to F 1 in E. coli F 1 F 0 -ATP synthase. To do this, we used the soluble cytoplasmic portion of the b subunit named b ST34 -156 , which had previously been constructed by Dunn and Chandler (34). Using fluorimet- ric assays, we found that b ST34 -156 did bind in a saturable fashion to variant forms of F 1 containing inserted Trp residues and also to wild-type F 1 . K d values for binding were ϳ100 -150 nM in Mg 2ϩ -containing buffer ( Table I). The binding curves were essentially the same independent of the presence or absence of ␦ subunit in F 1 or in the presence of added excess ␦ subunit. This result is in apparent conflict with a previous report (24), which concluded that the presence of ␦ was necessary for binding of the cytoplasmic portion of b to F 1 , but we note that the previous work was done in buffer that did not contain Mg 2ϩ . The binding of b ST34 -156 to F 1 was strongly dependent on Mg 2ϩ , and in the presence of excess EDTA, binding was much reduced. Using the F 1 with the largest signal (␤F410W), a K d of 1.25 M for binding of b ST34 -156 in EDTA was calculated. In previous work, we showed that binding of ␦ subunit to ␦-depleted F 1 was also promoted by the presence of Mg 2ϩ . X-ray crystallography of F 1 has so far visualized Mg 2ϩ cation bound only in nucleotide binding sites. Our work predicts that bound "structural Mg 2ϩ " may be present in the stator and that it contributes in an important way to the ability of the stator to resist rotor strain. Methods in current use to release F 1 from membranes in soluble form invariably utilize conditions in which free divalent cation concentration is minimized, and it is well known that reconstitution of F 1 with F 0 is strongly favored by Mg 2ϩ ions.
The size of the fluorescence responses of inserted Trp residues that occurred upon binding of b ST34 -156 to F 1 should not be taken as evidence of direct interaction (or not) of that specific residue with b ST34 -156 ; nevertheless, by comparing the location of these Trp residues with the fluorescence responses seen, a general correlation is obvious. Those residues that gave zero or small fluorescence response lie buried inside the F 1 molecule (e.g. ␣F291W, ␣F346W, ␤Y297W, ␤F148W, ␤Y331W, ␤R398W) or are cryptically sited close to ␥ (␣F409W). On the other hand, those that gave strong signals are located on the external surface of the ␣ 3 ␤ 3 hexagon (␣L55W, ␣F406W, ␤Y26W, ␤F410W). One apparent exception could be ␤F17W, which lies on the outside and gave no signal, but it is actually located in the N-terminal ␤-barrel domain right at the top of the molecule, probably well away from where the b subunit binds (see e.g. Ref. 44). As noted under "Results," the signal seen in wild-type F 1 is to a significant extent referable to the naturally occurring ␤-Trp-107, which is also located on the exterior surface. Therefore, the residues that reported binding of b ST34 -156 by fluorescence enhancement are all located on the exterior surface of the ␣ 3 ␤ 3 hexagon. It may be noted that no attempt was made here to specifically design Trp-containing enzymes for the purpose of measuring b subunit binding; rather, we tested a group of enzymes that had been previously constructed for a variety of purposes in our laboratory. Our work shows that in principle, it should be possible, with the advent of high resolution structure information, to design probes with better signals that could report conformational changes at the F 1 /b subunit interface during rotation.
The functionally active form of the b subunit, and of cytoplasmic portions of the b subunit such as b ST34 -156 , has previously been shown to be a dimer (32)(33)(34)(35)(36)(37). In recognition of this fact, we plotted all our titration data as a function of b 2 concentration, and the K d values for b ST34 -156 listed in Table I are those calculated for the b ST34 -156 dimer. Still, there is one puzzling point, which is that the K d for b dimer formation is reported to be in the 1 M range (36), which seems rather high (equivalent to 17 g/ml b subunit in cells). We found K d values of 100 -150 nM for binding of b ST34 -156 to F 1 . Our titration curves were well fitted by a hyperbolic binding curve with n ϭ 1 (for b 2 dimer) over the range of b concentration from zero to 3.0 M, or higher in some instances; thus, we saw no concentration dependence of binding that would indicate a dependence on monomer-dimer equilibrium. Most previous experiments in which dimerization of b or cytoplasmic portions of b was studied were done in buffers devoid of Mg 2ϩ , and the actual concentration dependence (K d ) for dimer formation in Mg 2ϩ has not been studied carefully. This might be one explanatory factor. In Ref. 35, the possibility was also raised that the presence of F 1 may provide a surface on which dimerization of  6. Investigation of binding of bA92W ST34 -156 and bI109W ST34 -156 to wild-type F 1 . The competition assay of Dunn (32) was used. Briefly, stripped E. coli membranes containing F 0 (500 g) were preincubated with wild-type F 1 in an amount sufficient to almost saturate the F 0 sites (40 g, determined beforehand by titration). The degree of F 1 binding was determined by the percentage of quench of acridine orange in the ATP-driven proton pumping assay. Increasing amounts of b ST34 -156 , bA92W ST34 -156 , or bI109W ST34 -156 were included in the preincubation as noted. q, wild-type b ST34 -156 ; ‚, bA92W ST34 -156 ; f, bI109W ST34 -156 . b might be facilitated in cells, and this appears to provide the best current explanation of our results.
In previous work, we had shown that b ST34 -156 greatly promoted binding of ␦ subunit to ␦-depleted F 1 in the presence of Mg 2ϩ ions, with a K d apparent of 150 nM for the b dimer. It is interesting that this is close to the K d values measured here for direct binding of b ST34 -156 to F 1 in the presence of Mg 2ϩ . We extended the previous work by measuring the promotion of ␦ subunit binding to ␦-depleted F 1 in the presence of EDTA. The new studies showed that the b ST34 -156 did still promote ␦-binding but with a higher K d apparent for the b dimer of 1.36 M. Interestingly, this is quite similar to the K d value determined for the direct binding of b ST34 -156 to F 1 in the presence of EDTA (1.25 M). Together the data reinforce the two conclusions evident above, that divalent cation (Mg 2ϩ ) is important in stator structure and that direct binding of the cytoplasmic portion of b subunit to F 1 provides an important contribution to stator stability.
Finally, we found that insertion of a novel Trp at position b-Ile-109 in b ST34 -156 prevented binding to F 1 . Although this could be due to prevention of dimerization of b ST34 -156 or to disruption of normal folding, it is consistent with the idea that this is a specific site of interaction of b with F 1 as adumbrated by cross-linking studies (27). However, substitution of residue b-Ala-92 by Trp did not affect F 1 binding. In neither case could the fluorescence signal of the introduced Trp be used to assay binding to F 1 or to isolated ␦ subunit.