Topics:
chloroform, hexanes, linear regression, molecule, partition coefficient, solvents, sulfides, transfer technique, enflurane, isoflurane

THE sites of action of the volatile general anesthetics remain to be determined conclusively, despite extensive research over a number of decades. 1Nevertheless, current consensus favors membrane proteins that function as ion channels and neurotransmitter receptors as the likely targets, 2despite limited direct evidence for this. Indirect support comes from studies showing that volatile general anesthetics alter the activity  of a number of membrane proteins, such as voltage- and ligand-gated ion channels. 3,4 

The Meyer-Overton rule has played a central role in general anesthetic mechanisms research. 5This relation shows that the olive oil:gas partition coefficients of anesthetic molecules correlate positively with their potency in animals. By extension then, definition of a compound’s lipid solubility should allow a prediction to be made regarding anesthetic potency. Recently, a number of compounds have been described that are predicted to be potent anesthetics based on lipid-solubility criteria and their overall molecular configuration but fail to display full anesthetic effects (both immobility and amnesia) in animals. 6These highly lipophilic compounds have been termed nonanesthetics, or nonimmobilizers, because they may have amnestic properties. 7,8 

In a previous study, 9it was shown that four volatile general anesthetics (halothane, isoflurane, enflurane, and sevoflurane) interacted better with solvents displaying some polar characteristic (aromatic, alcohol, thiol, or sulfide), compared with the purely aliphatic solvent n -hexane. Herein, we report that the nonanesthetic molecules 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane interact more favorably with n -hexane compared with the polar solvents. Data obtained with the anesthetic molecules chloroform and 1-chloro-1,2,2-trifluorocyclobutane are also included. It is clear that using isotropic solvents to model anisotropic structures, such as proteins and lipid membranes, may only provide approximate interaction energies. Nevertheless, the use of octanol as a model for biologic membranes has been successfully and extensively used in medicinal chemistry, providing a framework for detailed quantitative structure–activity relations. 10The results of the current study suggest that nonanesthetic molecules are likely to occupy microenvironments in biologic membranes that differ from those favored by anesthetic molecules. The different microenvironment preferences of the anesthetics and the nonanesthetics are likely to account for their contrasting pharmacologic effects on intact animals.

Materials

The compounds 2,3-dichlorooctafluorobutane (F8), 1-chloro-1,2,2-trifluorocyclobutane (F3), and 1,2-dichlorohexafluorocyclobutane (F6) were purchased from PCR Incorporated (Gainesville, FL). Chloroform (99.8%), benzene (99.8%), and ethyl methyl sulfide (99%) were from Aldrich Chemical Co. (Milwaukee, WI), and n -hexane (99+%, capillary gas chromatography grade) was from Sigma Chemical Co. (St. Louis, MO). Methanol (99.9+%, high-pressure liquid chromatography grade) was obtained from Fisher Scientific (Fairlane, NJ).

Determination of Solvent: Gas Partition Coefficients and Transfer Free Energies

Partition coefficients and transfer free energies of the anesthetics and nonanesthetics between the gas and solvent phases were determined as described. 9 

Statistics and Linear Regression

Partition coefficients were determined between four and six times for each individual anesthetic–solvent and nonanesthetic–solvent combination. Data are expressed as mean ± SD.

The partition coefficients for chloroform, 1-chloro-1,2,2-trifluorocyclobutane, 2,3-dichlorooctafluorobutane, and 1,2-dichlorohexafluorocyclobutane, between the gas phase and the four organic solvents, are given in table 1. For the two anesthetics (chloroform and 1-chloro-1,2,2-trifluorocyclobutane), partitioning was found to be more favorable into the solvents that either have aromatic character or contain an alcohol group or a sulfide group, compared with the aliphatic solvent n -hexane. In contrast, the two nonanesthetic molecules (2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane) preferred to interact with n -hexane rather than with the somewhat polar solvents.

Table 1. Solvent:Gas Partition Coefficients for the Four Halogenated Alkanes for Four Different Solvents

For each case, the partition coefficient is given followed by n, the number of experiments. The errors are SD.

Table 1. Solvent:Gas Partition Coefficients for the Four Halogenated Alkanes for Four Different Solvents
View large
Table 1. Solvent:Gas Partition Coefficients for the Four Halogenated Alkanes for Four Different Solvents
View Large

Compared with an aliphatic environment (n -hexane), the presence of an aromatic group, an alcohol group, or a sulfide group improved solvent:gas partitioning, by factors of 1.9–2.4 for chloroform and 2.7–4.6 for 1-chloro-1,2,2-trifluorocyclobutane. As shown in figure 1A, the most favorable environments for chloroform were ethyl methyl sulfide, a model for methionine, and benzene, a model for the aromatic amino acid side-chains. Figure 1Bshows that 1-chloro-1,2,2-trifluorocyclobutane partitioned to the greatest extent into benzene, a model for the aromatic amino acid side-chains. Figures 2A and Bshow that the two nonanesthetic molecules 1,2-dichlorohexafluorocyclobutane and 2,3-dichlorooctafluorobutane partitioned most favorably into n -hexane, a model for the aliphatic amino acid side-chains of alanine, valine, leucine, and isoleucine.

Fig. 1. Transfer free energies for (A ) chloroform and (B ) 1-chloro-1,2,2-trifluorocyclobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the anesthetics between the gas and solvent phases were calculated as ΔGg→s=−RTln(Cs/Cg), where R is the gas constant (1.987 cal/mol K), T is the temperature in kelvins, and Csand Cgare the solvent and gas phase concentrations of the anesthetic molecules at equilibrium.
View largeDownload slide

Fig. 1. Transfer free energies for (A ) chloroform and (B ) 1-chloro-1,2,2-trifluorocyclobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the anesthetics between the gas and solvent phases were calculated as ΔGg→s=−RTln(Cs/Cg), where R is the gas constant (1.987 cal/mol K), T is the temperature in kelvins, and Csand Cgare the solvent and gas phase concentrations of the anesthetic molecules at equilibrium.

Fig. 1. Transfer free energies for (A ) chloroform and (B ) 1-chloro-1,2,2-trifluorocyclobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the anesthetics between the gas and solvent phases were calculated as ΔGg→s=−RTln(Cs/Cg), where R is the gas constant (1.987 cal/mol K), T is the temperature in kelvins, and Csand Cgare the solvent and gas phase concentrations of the anesthetic molecules at equilibrium.
View largeDownload slide

Fig. 1. Transfer free energies for (A ) chloroform and (B ) 1-chloro-1,2,2-trifluorocyclobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the anesthetics between the gas and solvent phases were calculated as ΔGg→s=−RTln(Cs/Cg), where R is the gas constant (1.987 cal/mol K), T is the temperature in kelvins, and Csand Cgare the solvent and gas phase concentrations of the anesthetic molecules at equilibrium.

Close modal
Fig. 2. Transfer free energies for (A ) 1,2-dichlorohexafluorocyclobutane and (B ) 2,3-dichlorooctafluorobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the nonanesthetics were calculated as defined for the anesthetics in the legend to figure 1.
View largeDownload slide

Fig. 2. Transfer free energies for (A ) 1,2-dichlorohexafluorocyclobutane and (B ) 2,3-dichlorooctafluorobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the nonanesthetics were calculated as defined for the anesthetics in the legend to figure 1.

Fig. 2. Transfer free energies for (A ) 1,2-dichlorohexafluorocyclobutane and (B ) 2,3-dichlorooctafluorobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the nonanesthetics were calculated as defined for the anesthetics in the legend to figure 1.
View largeDownload slide

Fig. 2. Transfer free energies for (A ) 1,2-dichlorohexafluorocyclobutane and (B ) 2,3-dichlorooctafluorobutane between the gas and four solvent phases that model various microenvironments present in biologic membranes. Error bars are SD. The transfer free energies of the nonanesthetics were calculated as defined for the anesthetics in the legend to figure 1.

Close modal

The overall free energy of solvation of these halogenated alkanes can be divided into two components. 11The first is related to the energy required to form a cavity in the solvent and is proportional to the volume of the solute. The second term is a favorable energy term associated with solute–solvent interactions. Molecular volumes for the four halogenated alkanes examined in this study are listed in table 2. Figures 3 and 4show plots of the free energy of solvation versus  the volume of the solvated molecule for the four halogenated alkanes examined in the current study, along with data for four additional anesthetic molecules (halothane, isoflurane, enflurane, and sevoflurane) taken from an earlier report. 9 

Table 2. Volumes of Six Anesthetics and Two Nonanesthetics

Volumes calculated using the approach outlined by Abraham and McGowan. 20 

Table 2. Volumes of Six Anesthetics and Two Nonanesthetics
View large
Table 2. Volumes of Six Anesthetics and Two Nonanesthetics
View Large
Fig. 3. Plots of the overall free energy of solvation into (A ) benzene and (B ) ethyl methyl sulfide for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8933 and (B ) 0.8920. The nonanesthetic data are the filled circles.
View largeDownload slide

Fig. 3. Plots of the overall free energy of solvation into (A ) benzene and (B ) ethyl methyl sulfide for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8933 and (B ) 0.8920. The nonanesthetic data are the filled circles.

Fig. 3. Plots of the overall free energy of solvation into (A ) benzene and (B ) ethyl methyl sulfide for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8933 and (B ) 0.8920. The nonanesthetic data are the filled circles.
View largeDownload slide

Fig. 3. Plots of the overall free energy of solvation into (A ) benzene and (B ) ethyl methyl sulfide for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8933 and (B ) 0.8920. The nonanesthetic data are the filled circles.

Close modal
Fig. 4. Plots of the overall free energy of solvation into (A ) methanol and (B ) n -hexane for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8192, and (B ) 0.9298 (if anesthetics and nonanesthetics are separated) versus  0.4406 for the entire data set (for B  only). The nonanesthetic data are the filled circles.
View largeDownload slide

Fig. 4. Plots of the overall free energy of solvation into (A ) methanol and (B ) n -hexane for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8192, and (B ) 0.9298 (if anesthetics and nonanesthetics are separated) versus  0.4406 for the entire data set (for B  only). The nonanesthetic data are the filled circles.

Fig. 4. Plots of the overall free energy of solvation into (A ) methanol and (B ) n -hexane for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8192, and (B ) 0.9298 (if anesthetics and nonanesthetics are separated) versus  0.4406 for the entire data set (for B  only). The nonanesthetic data are the filled circles.
View largeDownload slide

Fig. 4. Plots of the overall free energy of solvation into (A ) methanol and (B ) n -hexane for six anesthetic and two nonanesthetic molecules (from table 2) as a function of molecular volume. Linear regression least squares fits were generated using KaleidaGraph version 3.0.5 (Abelbeck Software, Reading, PA, 1994). The correlation coefficients (r) are (A ) 0.8192, and (B ) 0.9298 (if anesthetics and nonanesthetics are separated) versus  0.4406 for the entire data set (for B  only). The nonanesthetic data are the filled circles.

Close modal

For benzene (fig. 3A) and ethyl methyl sulfide (fig. 3B), the overall free energy of solvation correlates with the molecular volume of the eight solutes. For methanol (fig. 4A), the correlation between the overall free energy of solvation and the volume of the solvated molecule is not quite as good (correlation coefficient, r = 0.8192), and this is probably related to the occurrence of favorable hydrogen bonding between the ether anesthetics (enflurane and isoflurane with molecular volumes of 133 Å3and sevoflurane with a molecular volume of 142 Å3) and methanol. For n -hexane (fig. 4B), the correlation between the overall free energy of solvation and the volume of the solvated molecule is poor (r = 0.4406) unless the anesthetic and the nonanesthetic molecules are separated (r = 0.9298 for the six anesthetics), suggesting that the nonanesthetic molecules 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane form particularly favorable interactions with n -hexane.

The nonanesthetic molecules 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane do not influence γ-aminobutyric acid type A α1β2and α1β2γ2sreceptor currents in Xenopus  oocytes, whereas the structurally related 1-chloro-1,2,2-trifluorocyclobutane potentiates chloride currents in the same manner as other anesthetic molecules. 12Similarly, current conducted through human neuronal nicotinic acetylcholine receptors expressed in Xenopus  oocytes is inhibited by isoflurane and 1-chloro-1,2,2-trifluorocyclobutane but is insensitive to the nonanesthetics 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane. 13These studies suggest that ligand-gated ion channels may represent central nervous system sites contributing to the immobility component of anesthesia. In contrast, mouse skeletal muscle nicotinic acetylcholine receptor currents are inhibited by both enflurane and the nonanesthetics 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane, although the kinetics of inhibition differ, indicating that the two classes of halogenated compounds may favor different conformations of the protein. 14Binding studies using human serum albumin 14indicate that 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane bind to the same overall site as halothane and chloroform, 15–17suggesting that this site in subdomain IIA may serve as a better model of an in vivo  target responsible for the amnestic component of the anesthetic state.

A 19F nuclear magnetic resonance study on the distribution of 1-chloro-1,2,2-trifluorocyclobutane, 2,3-dichlorooctafluorobutane, and 1,2-dichlorohexafluorocyclobutane in egg phosphatidylcholine lipid vesicles is in accordance with the current findings. The anesthetic molecule 1-chloro-1,2,2-trifluorocyclobutane was found to preferentially localize at the membrane–water interface, whereas the nonanesthetic molecules 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane solubilized more deeply in the lipid core. 18Also using 2H and 19F nuclear magnetic resonance in palmitoyl oleoylphosphatidylcholine membranes, 1-chloro-1,2,2-trifluorocyclobutane was shown to localize at the membrane interface, whereas the nonanesthetic 1,2-dichlorohexafluorocyclobutane was distributed evenly in the hydrocarbon region of the lipid acyl chains. 19 

The current results suggest that the nonanesthetic molecules 2,3-dichlorooctafluorobutane and 1,2-dichlorohexafluorocyclobutane interact preferentially with the purely aliphatic phase represented by n -hexane. This solvent serves as a model for the aliphatic amino acid side-chains of alanine, valine, leucine, and isoleucine and also for the saturated portions of the phospholipid acyl chains. 9In contrast, the anesthetics chloroform and 1-chloro-1,2,2-trifluorocyclobutane interact more favorably with the somewhat polar solvents benzene, methanol, and ethyl methyl sulfide, which serve as models for the aromatic amino acids, serine, and methionine, respectively, in agreement with the earlier study on halothane, enflurane, isoflurane, and sevoflurane. 9Anesthetic molecules are therefore predicted to favor several microenvironments present in proteins because of the potential for energetically more favorable interactions. In contrast, nonanesthetic molecules are predicted to prefer more apolar environments on proteins or the saturated portion of the lipid bilayer. Finally, solvation studies of this type may allow differentiation of nonanesthetic and anesthetic molecules without the necessity of animal studies.

1.
Eckenhoff RG, Johansson JS: Molecular interactions between inhaled anesthetics and proteins. Pharmacol Rev 1997; 49: 343–67
2.
Franks NP, Lieb WR: Molecular and cellular mechanisms of general anesthesia. Nature 1994; 367: 607–14
3.
Ratnakumari L, Vysotskaya TN, Duch DS, Hemmings HC: Differential effects of anesthetic and nonanesthetic cyclobutanes on neuronal voltage-gated sodium channels. A nesthesiology 2000; 92: 529–41
4.
Mihic SJ, Ye Q, Wick MJ, Koltchine VV, Krasowski MD, Finn SE, Mascia MP, Valenzuela CF, Hanson KK, Greenblatt EP, Harris RA, Harrison NL: Sites of alcohol and volatile anaesthetic action on GABAAand glycine receptors. Nature 1997; 389: 385–9
5.
Miller KW: The nature of the site of general anesthesia. Int Rev Neurobiol 1985; 27: 1–61
6.
Koblin DD, Chortkoff BS, Laster MJ, Eger EI, Halsey MJ, Ionescu P: Polyhalogenated and perfluorinated compounds that disobey the Meyer-Overton hypothesis. Anesth Analg 1994; 79: 1043–8
7.
Kandel L, Chortkoff BS, Sonner J, Laster MJ, Eger EI: Nonanesthetics can suppress learning. Anesth Analg 1996; 82: 321–6
8.
Eger EI, Koblin DD, Harris RA, Kendig JJ, Pohorille A, Halsey MJ, Trudell JR: Hypothesis: Inhaled anesthetics produce immobility and amnesia by different mechanisms at different sites. Anesth Analg 1997; 84: 915–8
9.
Johansson JS, Zou H: Partitioning of four modern volatile general anesthetics into solvents that model buried amino acid side-chains. Biophys Chem 1999; 79: 107–16
10.
Hansch C, Leo A: Substituent Constants for Correlation Analysis in Chemistry and Biology. New York, Wiley, 1974, pp 13–7
11.
Ben-Naim A: Solvation Thermodynamics. New York, Plenum Press, 1987, p 46
12.
Mihic JS, McQuilkin SJ, Eger EI, Ionescu P, Harris RA: Potentiation of γ-aminobutyric acid type A receptor-mediated chloride currents by novel halogenated compounds correlates with their abilities to induce general anesthesia. Mol Pharmacol 1994; 46: 851–7
13.
Cardoso RA, Yamakura T, Brozowski SJ, Chavez-Noriega LE, Harris RA: Human neuronal nicotinic acetylcholine receptors expressed in Xenopus  oocytes predict efficacy of halogenated compounds that disobey the Meyer-Overton rule. A nesthesiology 1999; 91: 1370–7
14.
Forman SA, Raines DE: Nonanesthetic volatile drugs obey the Meyer-Overton correlation in two molecular protein site models. A nesthesiology 1998; 88: 1535–48
15.
Johansson JS, Eckenhoff RG, Dutton PL: Binding of halothane to serum albumin demonstrated using tryptophan fluorescence. A nesthesiology 1995; 83: 316–24
16.
Johansson JS: Binding of the volatile anesthetic chloroform to albumin demonstrated using tryptophan fluorescence quenching. J Biol Chem 1997; 272: 17961–5
17.
Bhattacharya AA, Curry S, Franks NP: Binding of the general anesthetics propofol and halothane to human serum albumin. J Biol Chem 2000; 49: 38731–8
18.
Tang P, Yan B, Xu Y: Different distribution of fluorinated anesthetics and nonanesthetics in model membranes: A 19F NMR study. Biophys J 1997; 72: 1676–82
19.
North C, Cafiso DS: Contrasting membrane localization and behavior of halogenated cyclobutanes that follow or violate the Meyer-Overton hypothesis of general anesthetic potency. Biophys J 1997; 72: 1754–61
20.
Abraham MH, McGowan JC: The use of characteristic volumes to measure cavity terms in reversed phase liquid chromatography. Chromatographia 1987; 23: 243–6