INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

There is a growing consensus that general anesthetics interact with neuronal proteins (Eckenhoff and Johansson, 1997; Franks and Lieb, 1994). Electrophysiological studies have indicated that a superfamily of neurotransmitter-gated ion channels, including the nicotinic acetylcholine receptors (nAChRs), -aminobutyric acid_A (GABA_A) receptors, glycine receptors, and 5-hydroxytryptamine (5-HT₃) receptors, is particularly sensitive to general anesthetics (Franks and Lieb, 1996). Site-directed mutagenesis has implicated the existence of an inhibitory site within the aqueous pore of nAChR (Forman et al., 1995) and a potentiating site at the extracellular interfacial regions of transmembrane domains 2 and 3 (MII-MIII) on the glycine and GABA_A receptors (Mihic et al., 1997). Although these studies clearly indicate the involvement of these critical residues in anesthetic action, it remains unknown whether direct anesthetic binding to the receptor is involved in the anesthetic action.

Indeed, experimental results on binding between volatile anesthetics and membrane proteins are scarce. Photoaffinity labeling of [¹⁴C]halothane, a clinically important inhalational anesthetic, to native Torpedo membranes and isolated nAChR (Eckenhoff, 1996b) suggests that halothane binding to nAChR is saturable and that the conformational changes associated with receptor function and desensitization do not alter the binding domain for halothane. Moreover, while [¹⁴C]halothane incorporation demonstrates little subunit selectivity, the labeling pattern within -subunits (presumably also within other subunits) suggests that most binding occurs at the four putative transmembrane segments, MI-MIV. Whether halothane penetrates between transmembrane sequences to produce significant labeling of the putative pore-lining MII segments remains unclear. The kinetics of binding of volatile anesthetics to transmembrane proteins are largely unknown.

Using ¹⁹F nuclear magnetic resonance (NMR) spectroscopy and gas chromatography (GC), we analyzed the binding kinetics of isoflurane, currently the most important clinical anesthetic, to nACh receptors that were purified and reconstituted in 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) membranes. Based on a two-site exchange model between nonspecific and specific binding, chemical shift and transverse relaxation time (T₂) measurements were used to determine the equilibrium dissociation constant, K_d, and the dissociation rate constant, k₁, of isoflurane binding to nAChR.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Previously documented procedures (Ellena et al., 1983) for reconstituting nAChR into DOPC were adapted as follows. Briefly, the electric organ of Torpedo nobiliana (Biofish Associates, Georgetown, MA) was freshly dissected and processed (Chak and Karlin, 1992) to obtain nAChR-rich membranes, which were frozen at 70°C until use. For purification, nAChR-rich membranes from 1 kg of electric organ were thawed and suspended in 600 ml buffer A (pH 7.4), containing 100 mM NaCl, 10 mM 3-(N-morpholino)propanesulfonic acid (MOPS), 0.1 mM EDTA, and 0.02% sodium azide. All chemicals were obtained from Sigma Chemical Co. (St. Louis, MO) and used without further purification. The nAChR-rich membrane suspension was stirred for 2 h in the presence of 1% (w/v) cholate to solubilize the receptors. The mixture was centrifuged at 256,000 × g for 30 min, and the supernatant layer was collected. The solubilized receptors were divided into two equal portions, and each was gently stirred for 2 h with 50 ml AffiGel-10 (Bio-Rad Laboratories, Hercules, CA) that was freshly derivatized with bromoacetylcholine bromide (Research Biochemical International, Natick, MA). The receptor-containing AffiGel-10 was then packed into a 2.5-cm-diameter glass column and washed at a rate of 100 ml/h with 1% cholate in buffer A for 2 h, followed by 0.003% DOPC (Avanti Polar Lipids, Alabaster, AL) and 0.5% cholate in buffer A for an additional period of 2 h. The nAChRs were then eluted with buffer A containing 20 mM carbamylcholine chloride, 0.003% DOPC, and 0.5% cholate. The protein-rich fractions were pooled and dialyzed for 12 h against six changes of 6 liters of buffer A. The membrane suspension was then concentrated to a few milliliters using Centriprep-50 (Amicon Inc., Bevery, MA) and stored at 70°C until use. Immediately before NMR experiments, the reconstituted nAChR-rich membrane suspension was subjected to six cycles of a freeze-thaw process alternating between liquid nitrogen and room temperature to adjust the vesicle size. After the precipitation of large vesicle aggregates, the protein content in the homogeneous vesicle suspension was determined by the bicinchoninic acid method.

All NMR experiments were conducted using a Chemagnetics CMXW-400SLI spectrometer (Fort Collins, CO), operating at 377.168 MHz for ¹⁹F resonance. Each NMR sample consisted of a freshly thawed 750-µl DOPC vesicle suspension in a 5-mm-diameter high-precision NMR tube (Wilmad Glass Co., Buena, NJ) with a total volume of 2500 µl (i.e., a vapor space of 1750 µl). NMR samples were prepared in pairs of the same DOPC concentrations and vesicle sizes, but with and without nAChR. The temperature was maintained at 15°C so that the receptor samples could be stable throughout the experiments. Isoflurane was added to the NMR samples with a microsyringe (Hamilton) in steps ranging from 0.03 to 0.2 µl. The equilibrium isoflurane concentrations in the membrane suspension were determined experimentally by NMR, with reference to one of two external standards containing 0.50 mM and 2.19 mM trifluroacetic acid (TFA) in 5- and 10-mm NMR tubes, respectively. The 10-mm standard was used coaxially with the 5-mm sample tube during concentration calibration. The external TFA resonance also served as a frequency reference for chemical shift measurements.

If  and _f are the ¹⁹F resonance frequencies of isoflurane in DOPC vesicle suspension with and without nAChR, respectively, and _b is the limiting resonance frequency for a hypothetical situation in which all isoflurane molecules were bound to nAChR, then, for a rapid exchange model between free and bound isoflurane, the mole fraction of isoflurane bound to nAChR, X_b, is given by (Xu et al., 1996)

(1)

From the definition of K_d, it can be shown that

(2)

or

(3)

where [A]₀ and [R]₀ are the total anesthetic and receptor concentrations, respectively. Thus, in a two-site free exchange model, expressing 1/(  _f) as a function of the total isoflurane concentration will yield a straight line, with an x-intercept of K_d. It should be noted that this chemical shift method for the determination of K_d does not require prior knowledge of X_b.

The binding K_d can also be determined by transverse relaxation time (T₂) measurements based on the rapid exchange model given by the following equation (Dubois and Evers, 1992; Xu et al., 1996):

(4)

where [A]₀ is the total isoflurane concentration at which T₂ is measured, and K_d is given by the negative intercept on the x axis. A modified Carr-Purcell-Meiboom-Gill (CPMG) spin-echo method (Meiboom and Gill, 1958) was used, in which T₂ was measured by incrementing the number of 180° pulses in a pulse train and acquiring a series of free induction decays at the end of each pulse train. After Fourier transformation, the spectral intensities (measured as area integration under the peaks) were fitted to an exponential decay function with respect to the time after the initial 90° pulse. Parallel T₂ measurements were made in paired samples of the same DOPC concentrations and vesicle size, but with or without nAChR. The samples without the receptors were used to determine T_2f, the T₂ in the absence of receptor binding.

To determine the binding kinetics, T₂ was also measured as a function of spin-echo time, _cp. The modified CPMG method overcame the hardware limitation at very short _cp values, so that measurements can be made at 180°-180° interpulse delays as short as 10 µs.

It has been shown (Allerhand and Gutowsky, 1965; Dubois and Evers, 1992; Xu et al., 1996) that in the presence of receptor binding, T₂ is given by

(5)

where T_2b is the T₂ of isoflurane in the receptor-bound state,  = 2(_b-_f), _b is the lifetime of isoflurane in the bound state, and f(_cp) is a hyperbolic tangent:

(6)

We measured X_b by using GC (Xu et al., 1996), and determined from chemical shift measurements (Eq. 1). Nonlinear regression of measured 1/T₂ as a function of 1/_cp (Eqs. 5 and 6) will yield the best estimates of T_2b and _b. The off-rate constant of binding, k₁, equals 1/_b by definition, and the on-rate constant, k₊₁, can be calculated by k₁/K_d.

All fitting parameters from linear and nonlinear regression will be reported as "best estimate ± standard error of estimate" (Glantz, 1992).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Three separately prepared batches of reconstituted nAChR-rich membranes were used for repeated measurements. The receptor concentrations, as determined by the bicinchoninic acid method for the homogenous nAChR-rich vesicle suspensions, were 5.8, 10.7, and 14.0 µM, respectively. Receptors reconstituted as described above have been shown previously to retain all of their functional properties (Ellena et al., 1983; Chak and Karlin, 1992), including specific activity for toxin binding and ion flux properties. Figure 1 plots the reciprocal of the frequency change for the trifluoromethyl (-CF₃) and difluoromethyl (-CHF₂) resonance as a function of isoflurane concentration. Notice the difference in the scales of the two y-axes, reflecting a larger frequency change due to the higher sensitivity of the -CHF₂ resonance to isoflurane concentration. The solid lines are linear least-squares fits to the data at isoflurane concentrations lower than 1.5 mM. Linear extrapolation to the x axis (Eq. 3) yields a K_d of 0.38 ± 0.06 mM (mean ± SE) from the -CF₃ resonance and 0.34 ± 0.01 mM from the -CHF₂ resonance. The average of the two gives a K_d of 0.36 ± 0.03 mM (mean ± SD). At higher isoflurane concentrations, the data deviated from the linear relationship predicted by Eq. 3, indicating a certain degree of saturation at the receptor binding sites. At the nearly saturating concentration of 2.5 mM, the -CHF₂ resonance of isoflurane shifted by 30 and 65 Hz in samples with 5.8 and 14 µM nAChR, respectively. For the latter, the X_b was determined by six independent GC measurements to be 0.05 ± 0.03 (mean ± SEM, n = 6). Using Eq. 1 and   _f = 65 Hz, it can be estimated that _b  f  1300 Hz. Because of strong partitioning of isoflurane in lipids, which are used as the control, X_b measurements by GC in samples with lower nAChR concentrations were difficult. However, if one assumes that _b  f for the same resonance does not vary with receptor concentrations, then X_b for the sample with 5.8 µM nAChR can be estimated to be 0.023 by using Eq. 1. The results are summarized in Table 1.

View larger version (18K): [in this window] [in a new window]   FIGURE 1   Determination of the dissociation constant, Kd, by measurements of NMR resonant frequencies as a function of isoflurane concentration. Data at concentrations lower than 1.5 mM are fitted by linear least-squares regression based on a two-site exchange model (Eq. 3). Extrapolation of the linear fit to the x axis yields Kd. , -CF3 resonance, the left y axis; , -CHF2 resonance, the right y axis.

View this table: [in this window] [in a new window]   TABLE 1   Measured and calculated Xb and b  f (CHF2) for 2.5 mM isoflurane

Fig. 2 shows stack plots of representative ¹⁹F spectra, acquired with the modified CPMG pulse sequence from samples without (A) and with 5.8 µM nAChR (B). The _cp for Fig. 2, A and B was 2 ms and 0.01 ms, respectively, as indicated by the labels next to the first (bottom) spectrum in each stack. The subsequent spectra in the stack plots are acquired at the multiples of _cp. Thus, spectra in Fig. 2 A were acquired after 1, 2, 3, 4, 5, 6, and 7 refocus pulses in 2-ms intervals, and those in Fig. 2 B were acquired after 1, 20, 30, 40, 50, 60, and 70 refocus pulses in 0.01-ms intervals. Spectral line broadening due to the presence of receptors is apparent by comparing spectra in Figs. 2, A and B. The T₂ at each _cp value was determined by fitting the spectral intensities, such as those in Fig. 2, to an exponential decay function with respect to the spin-echo time. In the receptor-free lipid vesicle suspensions, the T₂ values of the -CF₃ and -CHF₂ resonance were found to be ~330 ms and ~10 ms, respectively, and were independent of _cp and isoflurane concentration. Figure 3 shows the dependence of -CF₃ T₂ on isoflurane concentration in the presence of 10.7-µM nAChR. At low isoflurane concentrations, the T₂ dependence fulfills the linear relationship predicted by Eq. 4. This relationship, however, is violated at higher isoflurane concentrations. The solid line at concentration greater than 0 mM shows the best fit to the data using the saturation equation [a + bx/(c + x)], yielding a = 1.81 ± 0.13 ms, b = 1.15 ± 0.18 ms, and c = 0.22 ± 0.13 mM. The first derivative of the fitting curve with respect to concentration at 0 mM defines the slope for the linear extrapolation, resulting in a K_d of 0.35 ± 0.28 mM from the negative intercept on the x axis.

View larger version (15K): [in this window] [in a new window]   FIGURE 2   Representative stack plots of 19F-NMR spectra of isoflurane acquired in the absence (A) and presence (B) of nAChR in DOPC vesicles. The first (lowest) spectrum in each stack was acquired after a single refocusing pulse at the spin-echo time of cp as labeled. The subsequent spectra in the stack were acquired after multiple refocusing pulses in equal intervals of cp. Notice the significant line broadening due to the presence of nAChR.

View larger version (15K): [in this window] [in a new window]   FIGURE 3   Determination of the dissociation constant, Kd, by 19F T2 measurements of the -CF3 resonance as a function of isoflurane concentration. The cp was set at 25 µs. The error bars show the standard error associated with the exponential fitting of individual T2 curves. The solid line at a concentration greater than 0 mM shows the nonlinear regression, using a saturation function, [a + bx/(c + x)]. The value of the first derivative of the regression curve at 0 mM was used as the slope for linear extrapolation in Kd determination, according to Eq. 4.

Figure 4 depicts the 1/T₂ ~ 1/_cp dependence of the -CHF₂ resonance for 2.5 mM isoflurane in the presence of 5.8 and 14.0 µM nAChR. The solid lines show the three-parameter (i.e., _b, T_2b, and _b  f) nonlinear regression using Eq. 5. The X_b and _b  f values listed in Table 1 were used as the initial estimates. The regression was rapidly converged to give a _b  f value of 1329 ± 183 Hz and 1368 ± 238 Hz for the samples with 5.8 µM and 14.0 µM nAChR, respectively. Both values are in excellent agreement with the result of ~1300 Hz obtained from the independent GC and chemical shift measurements (Table 1). The best estimates for _b and T_2b are, respectively, 54.6 ± 5.6 (mean ± SE) and 29.3 ± 5.1 µs in the presence of 5.8 µM nAChR, and 62.0 ± 7.7 and 32.1 ± 6.7 µs in the presence of 14.0 µM nAChR. Note that although T₂ values for the two samples differ greatly because of the difference in receptor concentration, the _b and T_2b values are essentially the same within the experimental error, indicating the robustness of the approach. Table 2 compares the averaged kinetic and binding constants of isoflurane binding to nAChR and to a soluble protein, bovine serum albumin (BSA) (Xu et al., 1996).

View larger version (19K): [in this window] [in a new window]   FIGURE 4   Dependence of 1/T2 on 1/cp is used to determine b and T2b. The error bars show the standard errors of mean of multiple measurements (n = 2-5) at each given cp. Where there are no error bars, the experiments were performed only once at those points. The solid lines are best fit to the data, using Eq. 5. The rate constants from the fitting are summarized in Table 2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A practical challenge faced in studies of volatile anesthetic interaction with membrane proteins is to distinguish specific interaction with the protein from nonspecific interaction with the surrounding lipids. The high sensitivity to nAChR exhibited by both the resonant frequencies and T₂ values of the isoflurane ¹⁹F-NMR allows for quantitative analysis of binding kinetics that can be directly attributed to the presence of the receptors in the lipids. The observed association and dissociation constants, therefore, provide direct measures of specific binding that occurs either directly on the receptors themselves, or at particular domains created at the interface between the receptor and the lipids.

The K_d of specific binding determined in Fig. 1 is within the clinical concentration range for isoflurane and comparable to the value of 0.18 ± 0.04 mM for halothane binding to nAChR obtained by photoaffinity labeling (Eckenhoff, 1996b). Although the T₂ method for K_d determination (Eq. 4) yielded a similar K_d value by linear extrapolation at very low isoflurane concentrations (Fig. 3), the method showed a lower apparent saturation concentration than the chemical shift method. At first glance, this is rather unexpected because the T₂ method is supposed to be more sensitive to isoflurane concentration, given the significant difference between T_2b and T_2f. The linear relationships predicted by both the chemical shift (Eq. 3) and T₂ (Eq. 4) methods are derived from a two-site rapid exchange model. Whether a given exchange rate can be considered to be rapid depends on the method of measurements. For isoflurane binding to nAChR, we estimated that the exchange rate is on the order of 10⁴ to 10⁵ s¹ based on the _b value. This rate is much greater than the limiting chemical shift difference (~10³ s¹, see Table 1) between the bound and free states, but comparable to the difference between 1/T_2b and 1/T_2f (~10⁴ s¹, see Table 2). Therefore, the rapid exchange condition is fulfilled for the chemical shift method but only marginally for the T₂ method. It should be noted that in the extreme of rapid exchange, the T₂ relaxation process is governed by the fast-relaxation component because the exchange allows all spins to experience the environment causing the fast relaxation. The measured T₂ is then an average of two T₂ components weighted by X_b. At the other extreme when the exchange is slow, one would observe two distinct relaxation components, or biexponential decay. In such a case, T₂ measurement would be dominated by the slow-relaxation (i.e., sharp) component. For isoflurane binding to nAChR measured in this study, we have an intermediate situation in which exchange is sufficient to average two relaxation components into a single exponential decay, but insufficient to completely remove the dominance of the sharp spectral component in the T₂ measurements. Thus, unless the isoflurane concentration is significantly smaller than K_d, the measured T₂ is biased by the narrower spectral component in the intermediate exchange regime. The apparent early saturation shown in Fig. 3 as compared to Fig. 1 reflects the deviation from the rapid exchange model, in addition to the possible saturation at a receptor site. In contrast, because the exchange rate is much greater than the limiting frequency difference, the deviation from the linear relationship in Fig. 1 reflects the true saturation at the receptor binding sites (i.e., X_b becomes invariant at high isoflurane concentrations).

There are other practical concerns that favor the chemical shift method over the T₂ method for K_d determination of anesthetic binding to membrane-associated proteins. For soluble proteins in large quantities, the T₂ method is reliable (Dubois and Evers, 1992; Xu et al., 1996) as long as the difference between 1/T_2b and 1/T_2f is much smaller than the exchange rate (i.e., in the rapid exchange regime). However, with membrane-associated proteins where high protein concentration is difficult to attain, the T₂ method is critically dependent on the T₂ values at low isoflurane concentrations, at which the measurements are less accurate and are time consuming. Sample stability is also of some concern when T₂ data acquisition becomes too long. The apparent early saturation (Fig. 4) due to a biased weighting of the narrower spectral component (1/T_2f) limits the use of T₂ values at higher isoflurane concentrations for K_d determination. In contrast, even at very low anesthetic concentrations, chemical shift can be measured rapidly and accurately. Another advantage of using the method of Eq. 3 is that K_d can be determined solely by chemical shifts, independent of X_b. Thus, it is preferable to use the chemical shift method of Eq. 3, rather than the T₂ method of Eq. 4, for K_d determination of anesthetic binding to membrane proteins.

From X_b and initial receptor concentrations (Table 1), it can be estimated that there are on average 9-10 specifically bound isoflurane molecules per nACh receptor. Assuming that the binding of isoflurane to nAChR resembles that of halothane in that subunit selectivity is asbent (Eckenhoff, 1996b), then, for a pentameric channel structure of nAChR, each subunit on average has at most one specific binding site for two isoflurane molecules or two specific binding sites for one isoflurane molecule each. It is interesting to note that mutagenesis studies (Mihic et al., 1997) have identified as few as two amino acid residues in glycine or the GABA_A receptor subunit that are essential for the receptor's sensitivity to general anesthetics. Given the photolabeling finding (Eckenhoff, 1996b) that most halothane binding to nAChR is in transmembrane domains, it is conceivable that the saturable isoflurane binding identified in this study also occurs within transmembrane domains.

Although at any given time the majority of isoflurane molecules are unbound, the presence of nAChR in the membrane greatly broadens the line width of the isoflurane resonance (compare stack plots in Fig. 2, A and B). This indicates that significant immobilization occurs to the bound isoflurane molecules. The exchange between the bound and free states is fast enough to result in a partial averaging effect. The T₂ values observed depend, again, on how rapidly T₂ is measured relative to the characteristic time of the exchange process. As shown in Fig. 4, the dependence of 1/T₂ on 1/_cp can be well characterized by the exchange model of Eqs. 5 and 6. The nonlinear regression in Fig. 3 shows that T_2b, the T₂ of -CHF₂ resonance of bound isoflurane, is ~30 µs, compared to 10,000 µs for the T₂ of the same resonance in receptor-free lipid vesicle suspensions. The reduction in T₂ by nearly three orders of magnitude suggests that the sites for binding must be rather structurally suited (or structurally specific) for isoflurane, because isoflurane molecules are significantly immobilized once they are bound. It should be noted that high specificity is not necessarily equivalent to tight binding. In fact, compared with isoflurane binding to BSA (a globular protein), the binding of isoflurane to the transmembrane nAChR is nearly diffusion limited, as revealed by our analysis of binding kinetics. As shown in Table 2, both the on- and off-rate constants are an order of magnitude faster for isoflurane binding to nAChR than to BSA. One possible interpretation is that the anesthetic binding sites are more easily accessible from the lipid phase to nAChR than from the aqueous phase to BSA. We showed previously that anesthetic binding sites are amphipathic in character (Xu and Tang, 1997; Tang et al., 1997; 1999a,b; Xu et al., 1998). In globular proteins, such amphipathic sites may be special folds (Eckenhoff, 1996a; Johansson et al., 1999; Franks et al., 1998) with hindered access directly from the aqueous phase. In contrast, in transmembrane proteins, the two-dimensional lateral diffusion in the lipid bilayer may provide an effective passage and orienting device for anesthetic binding to receptor sites. Thus, our finding of rapid binding kinetics for isoflurane to nAChR may suggest that the presence of lipids is important for realistically and accurately evaluating the binding kinetics between volatile anesthetics and neuronal receptors.

In summary, isoflurane binding to nAChR is specific and saturable. There are at most two specific binding sites for each subunit in a pentameric receptor, assuming that the binding has no subunit preference. The motion of the bound isoflurane molecules is greatly restricted, as judged by a decrease in T₂ upon binding by nearly three orders of magnitude. The binding to nAChR is nearly diffusion limited, in contrast to previous findings with soluble proteins. Thus, the presence of lipids probably contributes to the rapid kinetics of binding.