Light-induced Changes in H' Binding to the Purple Membrane EFFECT OF pH, LIGHT, TEMPERATURE, AND IONIC STRENGTH*

Under continuous illumination, isolated planar sheets of purple membrane from Halobacterium halo- bium acidify the surroundings at alkaline pH. This light-induced change in H' binding to the purple membrane (Ah) was studied by differential titration under varying conditions of pH, temperature, ionic strength, salt composition, light intensity, and wavelength.

The purple membrane of HaZobacterium halobiurn contains a light-dependent transmembrane proton pump which can drive the synthesis of ATP (for a comprehensive review, see Ref. 1). Both the crystal structure (2) and amino acid sequence (3) of the light-absorbing protein bacteriorhodopsin have been determined, providing the basis for understanding a proton pump at the molecular level. When nonvesicular, planar fragments of the purple membrane are illuminated with a steady light-source, small pH changes may be measured in the surrounding medium, due to light-induced changes in proton binding to the purple membrane (Ah ). This effect was first measured by Oesterhelt and Hess (4), in ether-saturated membrane suspensions at high ionic strength. I subsequently showed (5) that, at alkaline pH, A& appears to depend on a group with a pK of 10. I also found A 6 to be a sensitive probe of alterations in proton pump activity resulting from chemical ' This work was supported by grants from the Robert A. Welch Foundation (AX-736), the National Institutes of Health (GM 25483), and the National Science Foundation (PCM 7822732). This paper was presented, in preliminary form, at the XI International Congress of Biochemistry, Toronto, 1979. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. modification of the purple membrane (6, 7). Garty et al. (8) have used the light-induced changes in proton binding as a measure of the relative rates of proton release and uptake.
Light-induced changes in proton binding may arise from a steady state build-up of protonated or deprotonated intermediates during proton pumping. Two types of proton binding sites could contribute to this effect (I) amino acid side chains that directly participate in the proton translocation process, and (11) side chains that are perturbed during pumping (for example, by conformational changes). There could be some sites that combine both properties I and 11, while others might be distinctly type I or type 11.
Measurement of Ah provides unique information about side chains that undergo changes in protonation during proton pumping. The experiments reported here were undertaken in order do: 1) test the previously proposed three-state model for A f i (5), as a function of light intensity and wavelength; 2) measure the temperature dependence ofAh; and 3) study the salt concentration dependence of Afi. The temperature-dependence studies were initiated in the hope of identifying the group of pK 10 that participates inA&. The enthalpy of proton dissociation could distinguish between lysine and tyrosine, since AHo = 6 kcal/mol for phenolic groups, while AHo =-10 kcal/mol for amino groups. The salt concentration dependence was studied in the hope of clarifying the unusual finding of a salt-dependent stoichiometry for proton pumping which was one proton/cycle at low ionic strength and two protons/ cycle at high ionic strength (9-11).

MATERIALS AND METHODS
Purple membranes were prepared by the method of Oesterhelt and Stoeckenius (12) from H. halobium S9. Experiments were done essentially as previously described (5). Unless otherwise indicated, the following conditions were used. Samples were 2.0 ml , with purple membranes suspended at a concentration of 1.0 X M bacteriorhodopsin in a specially-made 1.4-cm diameter vessel that was thermostatted to 15 "C by a Haake FK circulating constant-temperature bath. The incident light source was a quartz halogen movie projector lamp (Sylvania ELE) operated at 20 V, filtered through 7 cm of 1% CuS04 and an Oriel 560 nm narrow band interference filter to an incident intensity of 2.4 X IO4 erg cm-' s". Either a Beckman 39031 or Radiometer GK2321C electrode was used. The electrodes were wrapped in aluminum foil to within about 1 cm of the tip to prevent a light response. The pH was measured with either a Beckman 4500 or a Radiometer PHM 64 meter. The meter output was offset and amplified by a Gould 13-4615-10 differential amplifier, and recorded on either a Houston Instruments Omniscribe or Heath recorder at a scale of about pH/inch.
The light-induced change in proton binding to the membrane, Ah, was calculated from the measured pH change, ApH, and the measured buffering capacity (per mol of bacteriorhodopsin) of the membrane sample, 8: The buffering capacity was determined from the slope of the titration curve that was measured for each sample at the same time and under the same conditions as the light-induced pH changes. This procedure also directly converts H' activity to H' concentration without knowledge of the activity coefficient.
Wavelength was varied with Oriel narrow band interference filters. The light intensity was measured with a Yellow Springs Instruments Radiometer.
In fitting parameters to Equation 3 (see below), the quantum yield for proton pumping was assumed to be 0.25 (13), and the purple membrane extinction coefficient was assumed to be 63,000 (4) at 570 nm.

RESULTS
pH Dependence of Light-induced Proton Release-At alkaline pH, purple membrane sheets acidify the medium under low intensity steady illumination, while at acid pH there is a net alkalinization. This effect is more readily quantitated at alkaline pH than acid for two reasons: 1) the changes are larger per mol of bacteriorhodopsin, and 2) the membrane sheets are more constant in their physical properties above pH 7 than below. Near pH 3, the purple membrane turns blue and forms large aggregates. However, absorbance changes are not observed at alkaline pH up to at least 10.5, and aggregation above pH 7 only occurs at high ionic strength.
With decreasing H+ concentration, the magnitude of the steady state light-induced change in H+ binding to the purple membrane/mol of bacteriorhodopsin ( Ah ) increases. It reaches a maximum value and then, above pH 10, sharply decreases. The decrease is most obvious at lower temperatures ( Figs. 1 and 2 where K is the ratio of the forward and reverse rate constants for light-induced deprotonation (4). For example, at pH 10, where the measured light-induced change in proton binding in deionized water is approximately 0.025 eq/mol(5), Equation 2 predicts a value of 0.14 eq/mol. Moreover, Equation 2 also predicts a steep rise to a limit of Ah = 1 at very alkaline pH, even at low light intensity. The measured values of Ah are far below this predicted amount. The H' binding sites that dissociate in the light must, at some alkaline pH, also dissociate in the dark. Thus, a more complete model for the light-induced change in proton binding is (5): where RH and R are, respectively, protonated and deprotonated bacteriorhodopsin that are not photochemically activated, and R* is deprotonated bacteriorhodopsin that is photocycling. The rate constant for light-induced proton release has been measured by flash spectroscopy (11, 14). However, under steady illumination, this rate is simply QIn, where Q is the quantum yield for proton pumping and I, is the absorbed light intensity. At the relatively low light intensities of the experiments reported here, only about one photon is absorbed/bacteriorhodopsin/s. The rate constant, k,, refers to the reprotonation of bacteriorhodopsin during the protoll pump cycle. I assume that this proton uptake reaction occurs at the opposite side of the membrane from the light-induced proton release. Although this process can occur in the dark (i.e. for a few seconds after the light is turned off), the rate is dependent on the concentration of the photoproduct R*. Measurements of k, by flash spectroscopy indicate that the rate is diffusion-limited (11, 14). Thus, at the light intensities used in the present experiments, appreciable steady state concentrations of R* will be found only at very low hydrogen ion concentrations (i.e. where [H'lk, = 1). For convenience, K2 is defined as Qla/kr[RH], and is essentially constant, except as discussed below. K 1 is the dissociation constant for the photolabile H+-binding when measured by titration in the dark. I assume that both the dissociation and association steps of the K1 equilibrium are from the same side of the membrane (2.e. this equilibrium is not a transmembrane process, unlike the light-dependent proton release and uptake). Throughout this paper, I refer to the K , equilibrium as light-dependent proton dissociation, the QIa reaction as light-induced proton release, and the k, reaction as pump-dependent proton uptake.  FIG. 1 (left). Temperature dependence of Ah, 0.015 M NaCl. Steady state light-induced changes in H' binding to purple membrane sheets ( A l i ) , eq/mol of bacteriorhodopsin, in 0.015 M NaC1, at 0 "C (0), 15 "C (A) and 30 "C (0). Lines calculated from Equation 3. See Table I for constants. FIG . 2 (right). Temperature dependence of Ah, 3.0 M NaCl. Same as Fig. 1 Figs. 1 and 2 and also data at 10 and 20 "C (not shown), were fitted with lines using Equation 3. The constants Kl and k, were obtained by an iterative computer method.
The temperature dependence of K 1 and k, are shown in Figs Table I.
The results are unusual in two respects. The value of K1, as previously noted, is approximately 10"' M, near the ionization constant expected for a lysine or tyrosine side chain. However, AHo clearly is of opposite sign to that expected for dissociation of a proton from lysine or tyrosine. The value of k, is about 10" M" s-', a rate constant for a diffusion-limited reaction. The activation energy for the diffusion-limited proton transfer reaction H' + -OH + HZO is 2 or 3 kcal/mol (15). But the apparent E , for k , is 5 to 10 times too large for a diffusion-   (Fig. 2). Least squares analysis gives apparent AHo of -6.0 and -6.5 kcal/mol, respectively.     (5) where F = Faraday's constant, R = the gas constant, and T = absolute temperature. This increased surface cation concentration can affect both the measurement of ion release from, and uptake by, a transmembrane cation pump. The effect on release will be to diminish the Concentration of released ions measured in bulk solution, since some ions that are pumped across the membrane could stay near the surface in the diffuse double layer. Ion uptake is affected by surface charge because it is a second-order process. The uptake rate will be enhanced by an increased surface concentration of the ion to be translocated.
Under the conditions of the experiments reported here, there is only a negligible effect on the rate of light-induced proton release, for the following reasons. The effect of surface charge on the light-induced proton release reaction will appear to be simply a buffering effect. The magnitude of Ah in these experiments is quantitated by measuring the buffering capacity of the membrane. Thus, protons released into the diffuse double layer are largely taken into account. However, any light-induced changes in buffering capacity would not be known, since the buffering capacity is measured in the dark. But this is also a negligible effect. Carmeli et al. (18) found about one negative charge appears on the purple membrane surface/bacteriorhodopsin during the formation of photointermediate M. A t the bacteriorhodopsin concentrations and pH used in the experiments in Fig. 7 , this light-induced change in surface charge would not attract a measurable amount of H+ (5 X lo-@ mol of H+/mol of bacteriorhodopsin, or approximately lo-' of a pH unit, assuming all bacteriorhodopsins are pumping.) The action of surface charge on proton uptake can readily account for the observed effects on light-induced changes in H' binding. The exponential factor in Equation 5 introduces a correction to the H' concentration in Equation 3 as shown below (19).
where Y = exp(-+F/RT) and [H+]b is the bulk solution concentration of H+. The surface potential, +, will vary with the salt concentration, and consequently so will Y and Ah.
Although a variety of models can be used (16,17), a simple  Table I to calculate K', and Kr2. -, assuming K I I = 3 X 1 0 " ' M and K'z = 2 x 10"O.
discrete charge model was used to fit the data in Fig. 7.

( r ) =n q exp(-rtr) (7)
where n = number of charges, r = distance from proton uptake site, q = unit charge, = dielectric constant of water, EO = permittivity of free space, and F2 4me0r 1/2 = (= C d ) (8) where ci = concentration of ith ionic species and zi = charge of ith ionic species. The H' uptake site of the proton pump (which is assumed to be the same as the photosensitive H' , the results presented here show that above pH 9 , A h reaches a peak and then decreases (Figs. 1 and 2). The temperature dependence of A h may be explained by linear van't Hoff and Arrhenius plots of the equilibrium and kinetic constants in Equation 3 (Figs. 1-4). The light intensity dependence is consistent with Equations 3 and 4 (Fig. 5). The wavelength dependence indicates that bacteriorhodopsin is the only significant absorbing species in the steady state at the actinic wavelength (Fig. 6). A membrane surface charge model (Equations 5-9) quantitatively accounts for the salt concentration dependence of A A.
Temperature Dependence-The apparent enthalpy of ionization for K , (Equation 3 and Fig. 3) is of opposite sign to the expected value for amino acid side chains of pK -10 (phenolic or amino groups). However, anomalous heats of proton dis-sociation are known to occur in enzyme-inhibitor complexes (21) or in proteins undergoing conformational changes (22, 23). The Arrhenius plot for the apparent rate of proton uptake (Fig. 4) gives an activation energy of 14-18 kcal/mol. This energy barrier is an order of magnitude too high for a diffusionlimited proton transfer reaction from water (15). Ort and Parson (24) reported a large activation energy for pumpdependent proton uptake, observed by measuring the temperature dependence of rates of light-induced volume changes (A V) of purple membrane fragments. They found a nonlinear Arrhenius plot, with an apparent inflection point at 17 "c.
The data presented in Fig. 4 appears to be linear, but it is not entirely comparable to the A V rates. The apparent rate constant k, in Fig. 4 is a second order rate constant, while that of Ref. 24 is f i s t order. The rate of A V was shown by Ort and Parson (25) to be pH-dependent during H+ uptake suggesting a second-order component. The A V rate also includes contributions from the buffer, while the rate constant k, was measured in the absence of buffer. Despite these limitations of comparison, the apparent activation energy for proton uptake obtained from A V is in good agreement with the results reported here obtained from steady state pH changes. The origin of a large activation energy for proton uptake may be a large conformational change in bacteriorhodopsin during pumping. Ort and Parson (24) find a large decrease in entropy during proton release and suggest this increase in order is due to a conformational change. Similarly, the large temperature dependence for pump-dependent proton uptake reflects the motions of bacteriorhodopsin toward its initial conformation. The slowing of this conformational motion at lower temperatures would thus slow proton uptake. Bacteriorhodopsin conformational changes have also been previously proposed to explain a variety of spectroscopic (26), kinetic (27), and reactivity (28) observations.
Can the conformational change and proton uptake processes be resolved into separate rates? Modifications of Equation 3 were derived with separate rate constants for conformational change ( k c ) and proton uptake ( kr ).

R '
QIa/" However, it was not possible to fit the temperature dependence of A h through the variation of kc alone. Instead, equations for A h with the correct temperature dependence must have the temperature-dependent terms as products of [H']. This strongly implies that the conformational change is directly linked to proton uptake. It is not yet clear whether the conformational change occurring upon light-independent proton dissociation is related to that occurring with pump-dependent proton uptake. In theory, bacteriorhodopsin need not participate in proton pumping other than to provide two wells leading to a Schiff base impeller at the center of the membrane. However, the evidence presented here strongly supports the notion of active participation of bacteriorhodopsin through conformational changes linked to proton transfer. Salt Concentration Dependence-The results show that the Gouy-Chapman theory quantitatively accounts for the effect of salt concentration on the light-induced changes in H' binding. At low ionic strength, a significantly higher hydrogen ion concentration will be present in the diffuse double layer at the membrane surface than in bulk solution, while at high ionic strength, the surface and bulk hydrogen ion concentrations will be similar. The second order proton uptake reaction of the proton pump is very sensitive to changes in the surface pH. In contrast, the effect of surface charge on the protonrelease reaction is not apparent, since the measurement technique corrects for salt-induced changes in buffering capacity.
Several workers have observed, under a variety of conditions and by a variety of methods, that the proton pump stoichiometry is apparently affected by salt concentration (9)(10)(11). Although I have assumed a value of one proton/cycle in order to calculate the kinetic constants given in Table I, the results presented here are not dependent on a particular pump stoichiometry. It is interesting that the surface charge model accounts for the salt concentration dependence ofAK without assuming any salt-induced changes in the quantum yield (contained in KIP of Equation 6; cf. Equation 4). It may be worthwhile to re-examine the methods used to measure the stoichiometry of proton pumping. The sensitivity of indicator dyes (11, 14) or buffers (9) could be influenced by surface charge at low ionic strength.