Bound Volatile General Anesthetics Alter Both Local Protein Dynamics and Global Protein Stability 

Halothane (2-bromo-2-chloro-1,1,1-trifluoroethane) was obtained from Halocarbon Laboratories (Hackensack, NJ). The thymol preservative present in the commercial halothane was removed with an aluminum oxide column. [6]Isoflurane (1-chloro-2,2,2-trifluoroethyl difluoromethyl ether) was purchased from Ohmeda PPD, Inc. (Liberty Corner, NJ). The nonimmobilizer 1,2-dichlorohexafluorocyclobutane was obtained from PCR Incorporated (Gainesville, FL). Fatty acid-free BSA (lot 119F9306) and myoglobin (lot 56F9306, from horse skeletal muscle) were purchased from Sigma Chemical Co. (St. Louis, MO). Guanidinium chloride (8.0 M) was obtained from Pierce (Rockford, IL). All other chemicals were reagent grade.

The buffer used for all experimentation was 130 mM sodium chloride, 20 mM sodium phosphate, pH 7.0. Protein solutions were equilibrated with halothane and isoflurane in gas-tight Hamilton (Reno, NV) syringes. Anesthetic-equilibrated protein was diluted with predetermined volumes of plain protein (not exposed to anesthetic agent but otherwise treated in the same manner) to achieve the final anesthetic concentrations shown in Figure 2, Figure 4, Figure 8and Figure 10. Equilibration of BSA with the nonimmobilizer 1,2-dichlorohexafluorocyclobutane was performed in the same manner.

Bovine serum albumin and myoglobin were used without further purification (purity >95% by electrophoresis). Myoglobin contains two tryptophan residues, at positions 7 and 14. [17]Protein concentrations were determined as described [6]after solutions were passed through 0.2 [micro sign]m nylon syringe filters (Gelman Sciences, Ann Arbor, MI).

The fluorescence anisotropy of BSA (5 [micro sign]M) was measured at 360 nm (bandwidth 8 nm) after selective excitation of tryptophan residues at 295 nm (bandwidth 2 nm) with a xenon arc lamp on a K2 multifrequency cross-correlation phase and modulation spectrofluorometer (ISS Inc., Champaign, IL), equipped with Glan-Thompson prism polarizers. A 320-nm emission cut-on filter was used to minimize the transmission of scattered excitation light. Anisotropy (r) was defined as Equation 1where I sub [double vertical bar] and I sub [perpendicular] are the emission intensities polarized parallel and perpendicular to the excitation, respectively. [14]During these conditions, grating (g) values, which correct for emission monochromator polarization effects, [18]ranged from 0.99-1.01. Polarizer alignment was verified [14]with a dilute solution of glycogen in deionized water, yielding an r value of 0.98-1.00 (n = 4). The temperature of the cell holder was controlled with a Haake F3 (Berlin, Germany) waterbath.

The concentration dependence of the anesthetic-induced fluorescence anisotropy change was fit using the following relationship:Equation 2where r is the measured fluorescence anisotropy, r¹is the fluorescence anisotropy in the absence of added anesthetic agent, r²is the additional limiting fluorescence anisotropy at an infinite anesthetic concentration, [A] is the anesthetic concentration, and K^dis the average dissociation constant for the anesthetic-protein interaction.

Steady-state near-ultraviolet spectra (240-330 nm) were recorded with a Model 62 DS spectropolarimeter (Aviv, Lakewood, NJ), using 10-mm path length quartz cells, sealed with Teflon stoppers. The cell holder was temperature controlled at 25 +/− 0.1 [degree sign]C. The bandwidth was 1.00 nm, with a scan step of 0.5 nm and an average scan time of 3.0 s.

Thermal denaturation of albumin (22 [micro sign]M) was followed by circular dichroism (CD) spectroscopy, measuring the ellipticity at 270 nm (Theta^270nm) as a function of temperature. The Theta^270nmwas measured at 1 [degree sign] intervals over the temperature range 50-75 [degree sign]C. The bandwidth was 1.00 nm, temperature equilibration time was 2 min, and the measurement time was 10 s.

For each run, the ellipticity (Theta) versus temperature (T) was fit using the relationship:Equation 3where R is the gas constant (1.987 cal/mol K) and T and T^m(defined later) are in kelvins. We thus obtained best fit values for the six parameters Eⁿ, mⁿ, E^u, m^u, Delta H, and T^m. This formula describes a standard model for protein unfolding. [19]In this model, the ellipticity at any temperature is given by Theta = Thetaⁿfⁿ+ Theta^uf^u, where Thetaⁿand Theta^urepresent the ellipticity of the native and unfolded states, respectively, and fⁿand f^urepresent the fraction of protein in the native and unfolded states, respectively (fⁿ+ f^u= 1). At low temperatures, in the pretransition region, f^uis close to zero, and Theta [almost equal to] Thetaⁿ, which is assumed to be a linear function of temperature (Thetaⁿ= Eⁿ+ mⁿ[middle dot] T). At high temperatures, in the posttransition region, f^uis close to 1, and Theta [almost equal to] Theta^u, which also is assumed to be a linear function of temperature (Theta^u= E^u+ m^u[middle dot] T). In the intermediate transition region, f^uvaries with temperature such that the equilibrium constant for protein unfolding, K = f^u/fn, satisfies the relationship:Equation 4

(Equation 4) describes an equilibrium with midpoint temperature T^m(at which point K = 1; i.e., f^u= fⁿ= 0.5), and - Delta H/R is the slope of the van't Hoff plot (lnK vs. 1/T). The enthalpy change for protein unfolding (Delta H) describes how sharply the unfolding equilibrium varies with temperature. Thus, for each experiment, a single fitting determined (1) the pre- and posttransition baseline values for Theta (ⁿ) and Theta^u, so the fraction unfolded (f^u) at each temperature could be calculated from the data using the formula f^u=(Theta - Theta (ⁿ))/(Theta^u- Thetaⁿ), [19]and (2) the parameters T^mand Delta H for determining the unfolding equilibrium constant at any temperature using Equation 4.

The dependence of T^mon anesthetic concentration, in conjunction with a value for Delta H, furnishes information about anesthetic binding parameters. We denote the unfolding equilibrium constant and midpoint temperatures in the absence of anesthetic agent by T^m0and K⁰, and those in the presence of anesthetic by T^mand K^app, in corcordance with the terminology used previously. [20]The normalized constant was calculated from the T^mand Delta H results using the equation Equation 5which follows from Equation 4evaluated at a temperature halfway between T^m0and T^m. Assuming that the native state of BSA has a single binding site for anesthetic agent A with dissociation constant K^d, and the unfolded state lacks this binding site, then the normalized unfolding equilibrium constant will be Equation 6

That is, the slope of a plot of K^0/Kappversus [A] gives the dissociation constant of the anesthetic-protein complex.

Buffer concentrations of halothane, isoflurane, and 1,2-dichlorohexafluorocyclobutane were determined using gas chromatography on an HP 6890 Series instrument (Hewlett Packard, Wilmington, DE), as described. [8,11,21] 

Best fit curves were generated with the KaleidaGraph (Abelbeck Software, Reading, PA) program, and MicroCal Origin 2.9 (MicroCal Software, Inc., Northampton, MA). Data are expressed as mean +/− SEM.

Earlier work using steady-state and time-resolved fluorescence measurements [6,8]and direct photoaffinity labeling [7]has implicated the IB and the IIA subdomains of BSA as volatile anesthetic binding sites. The effect of binding of halothane and isoflurane to the IB and IIA sites in BSA, each of which contains a single tryptophan residue, was examined using fluorescence anisotrophy measurements. Figure 2shows that both anesthetic agents cause a concentration-dependent increase in the tryptophan fluorescence anisotropy of the protein. For both agents, the data have been fit using Equation 2, yielding K^dvalues of 1.6 +/− 0.4 and 1.3 +/− 0.3 mM for isoflurane and halothane, respectively. Degassing of samples with nitrogen to remove added anesthetic agent resulted in a return of the tryptophan fluorescence anisotropy to control values. Serial dilution of the protein causes no change in the fluorescence anisotropy (although there is a progressive linear change in the steady-state fluorescence intensity), indicating that turbidity-induced light scattering is negligible over the protein concentration range 0.5-10.0 [micro sign]M and is not responsible for the observed increase in the fluorescence anisotropy in the presence of the volatile anesthetic agents (data not shown).

Polyhalogenated molecules that fail to act like conventional volatile general anesthetic agents have been introduced recently [22]and are termed nonimmobilizers. These compounds may serve as tools to test the utility of various model systems currently in use to understand mechanisms of anesthetic action. One example of a nonimmobilizer is 1,2-dichlorohexafluorocyclobutane. [22]This compound recently was shown to bind to human serum albumin [23]using steady-state tryptophan fluorescence measurements to determine binding. [6] Figure 3A shows the effect of 1,2-dichlorohexafluorocyclobutane on BSA steady-state tryptophan fluorescence. There is a concentration-dependent quenching associated with binding of the nonimmobilizer to BSA, with a maximum quenching of 31 +/− 3%(n = 3) of the tryptophan fluorescence, at a saturating aqueous concentration of 1,2-dichlorohexafluorocyclobutane of 180 +/− 30 [micro sign]M (n = 8). Figure 3B shows that binding of the nonimmobilizer 1,2-dichlorohexafluorocyclobutane to BSA is not associated with any apparent change in the fluorescence anisotropy of the tryptophan residues.

Experiments were performed with myoglobin to test the specificity of the anesthetic-induced change in the tryptophan fluorescence anisotropy of albumin. Myoglobin also contains two tryptophan residues but has been shown previously not to bind halothane. [6,24] Figure 4shows the lack of effect of halothane and isoflurane on myoglobin tryptophan fluorescence anisotropy. Therefore, in contrast to what is observed with albumin, the anesthetic agents have no effect on the myoglobin tryptophan fluorescence anisotropy, indicating an absence of binding to the native, folded form of myoglobin.

Induced hypothermia has been shown to decrease anesthetic requirements. [25]The effect of temperature on the tryptophan fluorescence anisotropy of BSA was examined and is shown in Figure 5. There is a progressive decrease in the fluorescence anisotropy (reflecting an increase in the mobility of the tryptophan residues) as the temperature is increased from 8 to 38 [degree sign]C. The effect of the bound anesthetic agents on indole ring mobility in BSA (Figure 2) is therefore comparable to what results from decreasing the temperature of the protein solution.

Circular dichroism is a form of spectrophotometry that makes use of polarized light and may give rise to positive and negative absorption bands that are conformation dependent. The differential absorption of left and right circularly polarized light after passing through a solution of optically active material (e.g., a protein) results in the transmission of elliptically polarized light. Near-ultraviolet (i.e., 250-330 nm) CD spectra reflect protein tertiary structure [26]and were obtained to examine further the effect of anesthetic agents on the local environment of the aromatic residues in BSA. The motionally constrained tryptophan, tyrosine, and phenylalanine side chains in the native folded form of albumin give rise to the CD spectra shown in Figure 6A with broad negative transitions in the 255- to 300-nm range. Adding either halothane or isoflurane to albumin results in only minor changes in these steady-state spectra, suggesting the absence of major structural changes (data not shown). In contrast, elimination of protein tertiary structural contacts by adding 4 M guanidinium chloride leads to the loss of most of the near-ultraviolet CD signal (Figure 6B) as the aromatic side chains are freed to assume additional conformations around the C sup [small alpha, Greek]-C sup [small beta, Greek] and C sup [small beta, Greek]-C sup [small gamma, Greek] bonds (Figure 1).

Anesthetic binding to BSA has minimal effects on the steady-state near-ultraviolet CD signal arising from the aromatic residues. Changes in the dynamics of the protein are apparent, however, because the bound anesthetic agent increases the tryptophan fluorescence anisotropy. To further investigate the effect of halothane and isoflurane binding on BSA dynamics, thermal denaturation (unfolding) of BSA was performed in the presence of the anesthetic agents.

In the absence of anesthetic agent, T^mfor albumin unfolding averaged 61.3 +/− 0.2 [degree sign]C (n = 4) for the isoflurane controls and 61.9 +/− 0.1 [degree sign]C (n = 4) for the halothane controls. Halothane and isoflurane increase T^mas shown in Figure 7. This finding implies that an increase in the overall stability of the native folded protein is observed in the presence of the anesthetic agent, with a shift of the thermal transition to higher temperatures in a concentration-dependent manner (Figure 8).

The increase in T^mwas accompanied by an increase in Delta H for unfolding as predicted, because d(Delta H)/dT = Delta C^p(the change in the heat capacity) is positive for protein unfolding, because of the exposure of previously buried hydrophobic residues to water. [27] Figure 9shows the best fit lines for Delta H versus T^m, yielding Delta H = 157 + 5.6(T^m- 61.3) for isoflurane and Delta H = 158 + 5.7(T^m- 61.9) for halothane, where Delta H is in kcal/mol and T^mis in [degree sign]C. These values for Delta H were used in Equation 5to calculate the normalized equilibrium constant K^0/Kapp, which was plotted against the anesthetic concentration, as shown in Figure 10, allowing calculation of dissociation constants for the anesthetics. Best fits to the data in Figure 10yield K^dvalues of 0.98 +/− 0.10 mM for isoflurane and 1.0 +/− 0.1 mM for halothane.

The goal of this study was to gain an understanding of fundamental mechanisms whereby a bound anesthetic molecule on a protein might modify protein function. The effect of bound anesthetic agents on protein dynamics and stability was examined because of the intimate link between protein flexibility and function. [28]Bovine serum albumin was selected as the model anesthetic binding protein because there is a growing literature supporting direct interactions of various volatile anesthetic agents with this protein using several different techniques, such as¹⁹F-NMR spectroscopy, [4,29,30]direct photoaffinity labeling, [5,7]fluorescence spectroscopy, [6,8]differential scanning calorimetry, [15]and isothermal titration calorimetry. [31]Although BSA is clearly not of direct relevance to the anesthetic state, it is predicted that the types of interactions with model systems apply equally well to in vivo sites of action, such as the large multiple subunit membrane proteins, which are not currently amenable to detailed biophysical analyses.

Two volatile general anesthetic agents that bind to BSA were found to cause an increase in the fluorescence anisotropy of the tryptophan residues. Fluorescence depolarization of protein tryptophan side chains may result from three distinct processes:(1) Brownian rotational diffusion of the protein as a whole;(2) internal fluctuations of the indole rings relative to the protein backbone; and (3) energy transfer between excited state moieties. [32]The contribution of Brownian rotational diffusion is small in the case of a relatively large protein such as albumin (66 kDa) because the average excited state lifetime of the tryptophan residues of 6.1 ns [33]is short compared with the rotational correlation time of albumin of 42 ns, [34]implying that no significant rotational diffusion occurs before fluorescence emission. This follows from the Perrin Equation r(^o/r)= 1 +([small tau, Greek]/[small phi, Greek]), where r^ois the limiting fluorophore anisotropy in the absence of rotational diffusion, [small tau, Greek] is the fluorescence lifetime, and [small phi, Greek] is the rotational correlation time. [14]Because [small phi, Greek] much > [small tau, Greek], the measured anisotropy r approximates r^o. Energy transfer between tyrosine and tryptophan residues was minimized by selecting an excitation wavelength of 295 nm. Further, the effect of energy transfer is to decrease fluorescence anisotropy. Although it is not possible to completely exclude energy transfer at this point, it is concluded that the main effect of halothane and isoflurane binding to BSA is to decrease the mobility of the indole rings. In the absence of anesthetic agent, the rotations of the indole rings are limited by van der Waals repulsions because of collisions with neighboring residues. The presence of the bound anesthetic agent therefore further increases the likelihood that the indole rings will experience steric inhibition to rotation about the dihedral angles chi¹and [chi squared](Figure 1).

The anesthetic-induced attenuation of indole ring mobility in BSA is not reproduced by the nonimmobilizer 1,2-dichlorohexafluorocyclobutane, despite apparent binding to the same site(s) in the protein occupied by anesthetic molecules, as determined by steady-state fluorescence measurements. Although 1,2-dichlorohexafluorocyclobutane is a direct quencher of indole fluorescence, [35]it is possible that the observed quenching of BSA steady-state tryptophan fluorescence by the nonimmobilizer is secondary to binding at different sites in subdomains IB and IIA than those favored by anesthetic molecules. Alternatively, the changes in steady-state tryptophan fluorescence and lack of effect on fluorescence anisotropy may reflect preferential binding of the nonimmobilizer to a partially unfolded conformation of BSA rather than to the native protein conformation.

The measurement of the tryptophan fluorescence anisotropy of BSA as a function of added anesthetic agent allowed a quantification of the binding energetics of isoflurane and halothane. Figure 2A shows that isoflurane binds with a K^dof 1.6 +/− 0.4 mM. This value is comparable to the K^dvalues of 1.4 +/− 0.2 and 1.3 +/− 0.2 mM determined using¹⁹F-NMR spectroscopy [4,30]and the value of 1.5 +/− 0.2 mM reported by Eckenhoff and Shuman [5]using photoaffinity labeling. Halothane bound to BSA with a K^dvalue of 1.3 +/− 0.3 mM using fluorescence anisotropy measurements (Figure 2B), again comparable to the value of 1.3 +/− 0.2 mM obtained using¹⁹F-NMR spectroscopy, [29]the value of 1.8 +/− 0.2 mM determined using intrinsic fluorescence quenching measurements, [6]and the value of 0.3-0.5 mM reported using photoaffinity labeling. [5]The thermal denaturation experiments yielded K^dvalues of 0.98 +/− 0.10 mM for isoflurane and 1.0 +/− 0.1 mM for halothane, in reasonable agreement with the studies cited here.

Anesthetic binding therefore has a stabilizing effect on BSA, in terms of local dynamics and global stability. This findings is in agreement with the effect of anesthetic agents on global BSA stability as measured by differential scanning calorimetry. [15]The effect of anesthetic agents on protein dynamics suggests a plausible mechanism for how anesthetic binding to proteins may alter protein function. The folded, biologically active, conformations of proteins are only marginally more stable (5-10 kcal/mol) than their unfolded counterparts, [36]a situation believed to have evolved to allow the structural changes required for normal protein function. [37]The ability of both protein receptors to bind ligands, and for enzymes to transform substrates, is dependent on protein motion. [28]Any alteration in the stability of a given conformational substate is predicted to alter the equilibrium between the different conformations, by raising the free energy barrier for subsequent conformational changes required for function. By limiting protein dynamics, volatile general anesthetic agents might alter the binding of native ligands to their receptors and thus disturb the function of central nervous system membrane proteins. This concept is shown schematically in Figure 11, in which the binding of successive anesthetic molecules to the protein results in additional stabilizing forces, trapping the protein in progressively lower free energy minima. Random thermal motion at physiologic temperatures in the presence of a bound anesthetic molecule therefore becomes insufficient to allow the protein to change into the active conformation. Experimental evidence for this concept is provided by work on the Torpedo nobiliana nicotinic acetylcholine receptor showing that general anesthetic agents increase the proportion of receptors in the desensitized state. [38]Bound volatile general anesthetic agents therefore may perturb the equilibrium between conformational substates of the protein.

It should be noted that the binding location of the anesthetic agent and the natural substrate need not be the same for apparent competition to occur. Binding of the anesthetic agent at one site may alter backbone dynamics at distant sites in the protein, preventing substrate binding, as seen in the case of the trp aporepressor protein after tryptophan binding. [39]In the latter case, this change in dynamics at distant sites is not accompanied by changes in secondary structure and may instead involve changes in the dihedral [small phi, Greek] and [small psi, Greek] angles (describing the rotations about the N-C sup [small alpha, Greek] and C sup [small alpha, Greek]-C' bonds, respectively;Figure 1) along the backbone without the breaking of hydrogen bonds. Similarly, binding of 1,2-dichloroethane to insulin crystals displaces sulfate ions from a distant site in the protein. [40] 

The number of volatile anesthetic binding sites on BSA is controversial but ranges from two [6-8]to four [4]and clearly depends on the aqueous concentration of agent. Therefore, at the lower anesthetic concentrations examined in the current study, only a single binding site was determined based on the thermal denaturation data. Occupancy of the site(s), although not associated with changes in protein secondary structure, [6,8]presumably causes some local, more subtle, structural or dynamic changes because the binding of other ligands such as thiopental, [41]coumadin, [42]and bilirubin [43]is altered by volatile anesthetic agents. Of these, both coumadin and bilirubin bind to the same IIA subdomain as halothane. [44]This finding suggests that volatile anesthetic binding to proteins may compete with binding of endogenous ligands as previously suggested. [45]As indicated earlier, however, this is not necessarily attributable to competition for occupancy of exactly the same physical site on the protein.

Anesthetic agents have been shown previously to have a fluidizing effect on lipid bilayers. [46]The finding that halothane and isoflurane have the opposite effect on local protein dynamics and global stability suggests that the interaction with protein targets may be fundamentally different. Protein stability is generally increased on ligand binding, as reflected by measures of dynamics such as fluorescence anisotropy, [47]hydrogen-deuterium exchange, [39]and Debye-Waller factors. [48]In this regard, halothane and isoflurane appear to behave like conventional ligands. Further work is needed, however, to confirm this finding because in the case of firefly luciferase there is evidence that anesthetic agents actually destabilize the protein and bind most favorably to a partially unfolded conformation. [16]Nevertheless, anesthetic-induced attenuation of protein dynamics as described in the current study and by others [24]is compatible with the clinical observations that (1) hypothermia decreases anesthetic requirements, [25]and (2) increased pressures, which denature proteins, [49]reverse the effects of anesthetic agents. [50] 

The authors thank Dr. Roderic G. Eckenhoff for helpful discussions.