Anal. Chem. WSO, 62,2720-2735
2728
Optical Method for Monitoring the Concentration of General Anesthetics and Other Small Organic Molecules. An Example of Phase Transition Sensing Sabina Merlol a n d P a u l Yager*
Center for Bioengineering, FL-20, University of Washington, Seattle, Washington 98195, and T h e Washington Technology Center, Seattle, Washington 98195
As an exampte of chemlcal sensing based on perturbations of thermal phaw, trandtions, we have shown that phoephoIlpkk labded wlth a fluorescent dye may be used to measure the concemtrafh of general anesthetics and other small organ& md.cuks. The emisskn maximun of the hydrophoMc fluorescent probe Laurdan In pbqh&@d bilayers SMttS from a wavelength of 445 nm below the main phase transition of the llpld to 480 nm above It, w%han lsosbestlc point at approxhnatdy 475 nm. The greatert dranges in Intensky at the transitton occur at 440 and 500 nm,so the ratio of the intensltles at these two points was used as an "order parameter". The effects of varlatlon of the liposomal preparation method on the order parameter were explored, and It was found that in mtxed lipids the parameter varied nearly linearly over the physiological temperature range. Fluorometry detected changes In the order of Mayers caused by solublkatlon of the anesthetic lsoflurane (Forane) and of ethanol. At a deflned temperature, the Intenslty ratio measured In the presence of anesthetic decreases In a concentratlon-dependent manner. ImmoblHzlng the Uposomes In a hydrogel dkl not perturb the response of the system. This work demondrates the potentlal for wing upld phase transltlons In an optrcal sensor for monltorlng anesthetks and other small nonpolar molecules.
INTRODUCTION While there are many techniques available for measuring the concentration of anesthetics in the gas phase, there is no corresponding technique currently in use to directly assay their concentrations in blood or tissue. To measure anesthetics in these tissues and fluids, a substantial sample of biological material must be removed and subjected to an extraction, after which the release gas is analyzed by chromatography or spectroscopy. Development of a rapid method for detection of anesthetic levels in vivo would allow clinically significant measurements to be made that are not now possible. As a step in developing such a sensor, this paper describes a system for measuring the concentration of general anesthetics and other small nonpolar molecules in condensed phases. This method embodies a novel principle of sensing using phase transitions. The strong correlation between the potency of general anesthetics and their solubility in lipids has long suggested that the site of anesthetic action is the hydrophobic region of cellular membranes (I). Most small nonpolar molecules, including general anesthetics and a wide variety of drugs, partition into lipid bilayers. Since phospholipids are the most
* To whom correspondence is to be addressed.
Current address: Universit4 di Pavia, Facolti di Ingegneria, Dip. di Elettronica, Laboratorio di Elettroottica, Via Abbiategrasso 209, 27100 Pavia, Italy.
common lipids found in biological membranes, bilayers formed by dispersing such lipids in excess aqueous solution (called liposomes or vesicles) have been extensively used as model systems for studies on membranes, including modeling anesthetielipid interactions (1-7). mid simple phospholipids used for such studies include dipalmitoyl phosphatidylcholine (DPPC) and dimyristoyl phosphatidylcholine (DMPC), which have chain-melting phase transitions a t 41.4 and 23 "C, respectively. The width of the transition is fiiite (occurring over no less than about 0.2 "C) and increases as the liposomal diameter decreases, which is consistent with the cooperative nature of the chain melting process (8). The transition temperature T, depends not only on the kind of lipid but also on hydration, pressure, pH, and presence of ions or organic molecules. This sensitivity to perturbations is the operating principle of our sensor. It is well-known that the freezing point of a material Y (T, in the aqueous phospholipid system) is depressed by incorporation of small amounts of a material Z that is freely soluble in the fluid phase of Y but insoluble in the crystal. In fact, for most systems the change in the transition temperature depends simply on the molality mZ of the material Z in the YZ mixture and on a constant K, that is characteristic of the material Y, as
AT,
x
Kmmz
This principle has long been exploited for determination of purity of compounds and for determination of molecular weights of unknowns. Determination of the concentration of Z in Y is limited only by the precision of the method of determining the phase transition temperature T , and one's knowledge of K,. We propose that a variant of this simple technique can be employed in the design of practical chemical sensors. If a sample of material Y is placed in an external environment E that contains Z, and Z is free to diffuse into Y, the equilibrium concentration of Z in Y will be determined by the partition coefficient of Z between Y and E. Consequently a t small loadings Z in Y the phase transition temperature of Y will again be linearly dependent on the concentration of Z in the environment. Any technique that allows the measurement of the transition temperature of Y can be used to measure the concentration of Z in the environment. Candidate techniques include calorimetry, spectroscopy, and monitoring of any of the numerous physical properties of a sample of Y that change at T,. The advantage of this approach over measurement of the changes in physical properties of Y that occur in a single phase on absorption of Z is the large change in the structure of Y that occurs at the phase transition. However a sensor based on this principle has only limited intrinsic selectivity. What selectivity exists derives from variations in the partition coefficients of the Z solutes. However, identification of unknowns would be possible by varying the nature of material Y, covering Y with semipermeable membranes, and by ap-
0003-2700/90/0362-2728$02.50/0 0 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 24, DECEMBER 15,
plying chemometric principles to the analysis of data with overlapping selectivities. The most straightforward method for carrying out such measurements would be to directly measure AT,,, by ramping the temperatures of a sample of Y through an appropriate range below T,. While such measurements on bulk samples are intrinsically slow, the times for such measurements can be reduced by miniaturization of both samples and measurement apparatus. However, if the phase transition has a finite width and a defined shape, it is possible to make a virtually instantaneous determination of AT,,, while holding the experimental T constant near T,,,. This is true as long as an "order parameter" may be extracted from the sample by a physical measurement that allows one to determine the degree of melting and, from that, the true temperature of the midpoint of the transition at the current concentration of Z. It has been shown by a wide variety of techniques including monitoring of fluorescence that all general anesthetics disorder lipid bilayers (1-5).The small anesthetic molecules do not fit well into the tight phospholipid hydrocarbon chain crystal lattice found in some synthetic lipid bilayers, whereas they are easily accommodated in the disordered hydrocarbon chains found above T,. Because of this, general anesthetics are generally found to lower the T,,, values of pure lipids in a concentration-dependent manner (3-5),although the shape of the phase transition may be altered depending on the partition coefficients of the small molecule between the gel and liquid crystalline phases (9). Monitoring the order of a lipid bilayer system relative to the order a t the same temperature in the absence of the anesthetic should allow the determination of the change in the transition temperature, AT,,,, and hence of the anesthetic concentration in that bilayer. A very sensitive technique for monitoring membrane order or fluidity is to observe the fluorescence of a hydrophobic probe that has been introduced into the bilayer. As fluorescence is an efficient process, use of small samples and simple optical systems is possible. A variety of fluorescent probes have been used to determine the fluidity of a lipid membrane (10-20).The fluorescence emission maxima of one particular class of fluorescent probes, including 6-lauryl-2(dimethy1amino)naphthalene(Laurdan) (13,16-18),have been shown to shift substantially with changes in membrane fluidity induced by alterations in temperature (19, 20). Recently, Parasassi et al. (18) studied the spectral properties of Laurdan-labeled DPPC vesicles at 37,42, and 60 "C. They reported a shift of the emission peak toward longer wavelengths as the temperature was increased; however, they explicitly mentioned that no isosbestic point was observed. Our work confirms the sensitivity of Laurdan to DPPC order-the emitted intensity peak shifts from a wavelength of 445 nm below the main phase transition of the lipid to 480 nm just above T,-but shows that it is possible to prepare samples with an isosbestic point a t approximately 475 nm. If baseline intensities are known, determination of membrane fluidity can be made using Laurdan by a simple ratio of emission at two frequencies, which behaves as an order parameter and may be employed to determine the AT,,, caused by the presence of general anesthetics. This feature makes the use of this and similar dyes particularly attractive for monitoring phase transitions in sensors. This fluorometric method was then used to detect changes in the order of lipid bilayers in the form of large vesicles caused by solubilization of the general anesthetic isoflurane (Forane) in clinically relevant concentrations. At a defined temperature, the intensity ratio measured in the presence of anesthetic decreases in a concentration-dependent manner. A technique for immobilization of liposomes that does not significantly affect their themodynamic properties is also demonstrated. In a
1990 2729
subsequent paper a fiber-optic probe is fabricated based on this measurement technique that responds linearly to physiologically relevant concentrations of general anesthetics (21). A preliminary version of this work has been published elsewhere (22), and patents have been applied for (23). EXPERIMENTAL SECTION Lipids and Other Reagents. 1,2-Bis(palmitoyl)-sn-glycero3-phosphocholine (DPPC) and 1,2-bis(myristoyl)-sn-glycero-3phosphocholine (DMPC)in chloroform solution at a concentration of 20 mg/mL were purchased from Avanti Polar Lipids Inc. (Birmingham AL) (purity >99%) and were used without further purification. Samples were found to give one spot by thin-layer chromatography. Laurdan was obtained from Molecular Probes (Eugene,OR) and was dissolved in chloroform to a concentration of 1mg/mL. Agarose Type I-A No. A-0169 was purchased from Sigma. Pure ethanol (200 proof) was distilled prior to use. Isoflurane (Forane, l-chloro-2,2,2-trifluoroethyldifluoromethyl ether) was donated by the Anesthesiology Department of the University of Washington. Vesicle Preparation. Fluorescent-labeled hydrated lipid was prepared as follows. Laurdan and one or more phospholipids (molar ratio 1:150) were first codissolved in chloroform. The solvent was evaporated in an Evapotec Micro Rotary Film Evaporator (Haake Buchler Instruments, Inc.) while the container was warmed in a water bath at 50 "C. In order to eliminate any trace of solvent, the sample was then stored under vacuum for 8 h. Dried lipid was then rehydrated with buffer (10 mM HEPES/100 mM NaCl in distilled4eionized water, pH 7.5). For fluorescence measurements, lipid suspensions at 1mg/mL were prepared. The samples were usually stored at 4 "C overnight, and vesicles were preared the next day by one of the following methods: ( I ) Sonicated Vesicles. Samples were sonicated for 30 min in a water bath sonicator (Ultrasonic cleaner Model GllBSPlT, Laboratory Supplies Co., Inc.) at a temperature >42 OC until they became nearly transparent. Small unilamellar vesicles (SUVs) are the primary product of such treatment (8). (2)Sonicated and Frozen Vesicles. After sonication, as performed for method 1,samples were frozen and then brought to room temperature. The freezing step was used with the intent of rapidly converting SUVs to larger structures through aggregation and fusion. (3)Sonicated, Frozen, and Aggregated Vesicles. After sonication, as performed for method 1, the samples were frozen, brought to room temperature, and stored in the refrigerator for 3 days (aggregation step). We believe that storage caused more complete aggregation and fusion of the vesicles than was achieved by process 2. ( 4 ) Shaken Vesicles. Hydrated lipids were warmed to above the main transition temperature and gently shaken by hand (24). (5) Vortexed Vesicles. Hydrated lipids were vortexed with a Vortex Genie Mixer (Scientific Products) at setting 7 for a total of 10 min in bursts of 1min interspersed with reheating to above the phase transition temperature (25). Fluorescence Measurements. Fluorescence measurements were carried out by using four-sided far-Uv-spectrophotometric cuvettes with the capability of being hermetically sealed using screw caps with Teflon septa (Spectrocell Corp.). The temperature was monitored inside cuvettes with Teflon-coated thermocouples connected to a Sensorket BAT-10 digital thermometer with a resolution of 0.1 "C. The thermometer was checked against a standard temperature bath in the Oceanography Department of the University of Washington and was found to be accurate to within 0.1 "C. Fluorescence measurements were performed by a Perkin-Elmer LS-5B luminescence spectrometer with a fourposition thermostatically controllable cell holder. A thermostatically controlled water bath and pump (Neslab RTE-11OP) with a programming controller were connected to the cell holder. The luminescence spectrometer was controlled and data were acquired by using a Zenith (IBM compatible) personal computer with an RS-232C interface by means of a modified version of a Perkin-Elmer compiled Basic software package. An excitation wavelength of 350 nm was utilized, slit widths were 5 and 3 nm for excitation and emission, respectively, and a scanning speed of 120 nm/min was chosen.
2730
ANALYTICAL CHEMISTRY, VOL. 62, NO. 24, DECEMBER 15, 1990
0 0 0 0
0
e e
e
e
0
e 375
425
475
525
575
Wavelength
Figure 1. A typical set of fluorescent spectra of Laurdan-labeled "sonicated and frozen" pure WPC vesicles: (A) 37.4 O C , (B) 38.4 O C , (C) 39.5 O C , (D) 41.0 O C , (E) 42.2OC. Unlabeled spectra collected at temperatures lower than 37.4 OC and higher than 42.2 O C are also
shown. When sensitivity to ethanol was tested, ethanol was first mixed with the buffer that was then used to dilute the lipid suspension. Four cuvettes containing a blank and three samples were usually placed in the cuvette holder. When sensitivityto anesthetics was tested, the experiments were performed by equilibratingthe lipid solution with anesthetic in the gas phase in a closed system (26) described below. Two cuvettes containing a blank and such a sample were placed in the cuvette holder. Equilibration of Lipid Suspensions at Clinical Concentrations of Volatile Anesthetics. A 1087-mLPyrex glass flask was modified in order to be able to dose it with a Mininert Valve (Pierce). Liquid anesthetic was injected through the Mininert Valve into the flask using a Hamilton gastight syringe and immediately vaporized. The flask contained 30 mL of fluorescent-labeled pure DPPC vesicle suspension. When volatile anesthetic in the liquid phase is added to the dry empty flask, it is possible to calculate the vlue of anesthetic partial pressure with the ideal gas law. When an aqueous phase exists, at equilibrium we would expect equal partial pressure of anesthetic in the gas and in the liquid phase. Unexpected differences between the partial pressures in the gas and liquid phase were detected with a Varian Aerograph Series 1200 gas chromatograph using a flame ionization detector. These differences were probably due to the low accuracy and reproducibility of the method used for determining the anesthetic concentration in the liquid phase. It is somewhat disturbing to note that this is the technique used in clinical medical practice to determine the concentration of general anesthetics in blood samples. On the other hand, the closed system based on the hermetically sealed flask does allow one to reproducibly generate (and maintain) clinically relevant anesthetic concentrations. Entrapment of Liposomes in Agarose Gel. Agarose gel was prepared by hydrating agarose powder with aqueous buffer. The aqueous agarose solution was then held above the gelling temperature for 10 min. Next, the suspension of lipid vesicles in buffer, which had been kept above the phase transition temperature after vortexing, was mixed into the warm agarose at the desired concentration. The sample was then poured into a cuvette and allowed to gel by cooling to room temperature. RESULTS AND DISCUSSION Monitoring of the Temperktture-Induced Main Chain Melting Phase Transition of Phosphatidylcholines Using Laurdan. The first goal of the work was to determine if the fluorescent probe Laurdan could be used to derive an order parameter that would be useful for monitoring the phase state of DPPC and similar lipids in the vicinity of T,. Laurdan and similar probes have been shown to be sensitive to the phase state of the bilayer in a manner that results in a substantial shift in emission frequency (18), but no previous attempt has been made to use them to quantify bilayer order. In this study, fluorescent spectra were first collected from Laurdan-labeled "sonicated and frozen" pure DPPC vesicles
"
I
34
36
38
40
42
44
46
Temperature in "C = 1(440)/1(500) and R' = 1(448)/1(480) as a function of temperature for a sample of Laurdan-labeled "sonicated and frozen" pure DPPC vesicles: R , 0; R', 0 . Flgure 2. Plot of the ratio R
at different temperatures. It is well known that both T , and the width of a lipid phase transition depend strongly on the diameter of the liposomes, and it was thought that this preparation would be a simple reproducible one for testing the response to Laurdan. A typical set of spectra is reported in Figure 1. The emitted intensity peak of Laurdan incorporated in the vesicles shifts from 445 nm, when the temperature is around 30 "C (lipids in the gel phase) 11.4 "C below T,, to a wavelength of 480 nm a t about 45 "C, when the transition is completed and the lipid is in the liquid crystal phase. The shift in Laurdan's emission allows phase transition to be most simply monitored by taking the ratio of emission intensities at two wavelengths. Such a ratio is independent of the concentration of dye, and, therefore, of the degree of photobleaching (and, possibly, of quenching) of the dye. The fact that such a ratio might directly reflect the degree of conversion from gel to liquid crystal states was supported by the presence of an isosbestic point (i.e. a wavelength where the fluorescence intensity does not depend on the phase state of the lipid) at approximately 475 nm. Its presence suggests that, within the finite width of the phase transition, dye molecules experience only one of two possible chemical environments during their excited-state lifetime (
