Use of beta-parinaric acid, a novel fouorimetric probe, to determine characteristic temperatures of membranes and membrane lipids from cultured animal cells.

A naturally occurring fluorescent compound, beta-parinaric acid, was employed as a probe to measure the effects of temperature changes on plasma membrenes, microsomes, and mitochondria and on their respective lipids after isolation form LM cells grown in suspension culture. A computer-centered spectrofluorimenter simultaneously measured the absorbance, absorbance-corrected fluorescence, and relative fluorescence efficiency of beta-parinaric acid incorporated into the membranes or isolated membrane lipids. These parameters were measured as a function of temperature. The probe revealed five characteristic breaks or changes in slope with both the plasma membranes as well as their extracted lipids. These discontinuities occurred at approximately 18, 23, 31, 38, and 43 degrees. The other isolated subcellular organelles, microsomes, and mitochondria, as well as their respective isolated lipids, exhibited approximately the same characteristic temperatures (+/- 1 degree) as plasma membranes. Thus, these data negate one criterion of the theory that an asymmetric distribution of characteristic temperatures exist across the membranes of LM cells.

A naturally occurring fluorescent compound, Bparinaric acid, was employed as a probe to measure the effects of temperature changes on plasma membranes, microsomes, and mitochondria and on their respective lipids after isolation from LM cells grown in suspension culture. A computer-centered spectrofluorimeter simultaneously measured the absorbance, absorbance-corrected fluorescence, and relative fluorescence efficiency of /3-parinaric acid incorporated into the membranes or isolated membrane lipids. These parameters were measured as a function of temperature. The probe revealed five characteristic breaks or changes in slope with both the plasma membranes as well as their extracted lipids. These discontinuities occurred at approximately 18, 23, 31, 38, and 43". The other isolated subcellular organelles, microsomes, and mitochondria, as well as their respective isolated lipids, exhibited approximately the same characteristic temperatures (*lo) as plasma membranes. Thus, these data negate one criterion of the theory that an asymmetric distribution of characteristic temperatures exists across the membranes of LM cells.

Investigators attempting
to ascertain the physical states of lipids or the conformation of proteins in membranes have utilized probe molecules (electron paramagnetic resonance, nuclear magnetic resonance, or fluorescence) that report on the membrane environment through their spectroscopic properties (l-3). Any external probe introduced into a biological or artificial membrane system should satisfy at least two major criteria: (a) the probe should produce a minimum perturbation of its environment, and (b) the nature of the environment in which the probe is located should be ascertained. Many probes in use today are synthetic molecules that often do not satisfy the first criterion (1,(3)(4)(5)(6). Commonly, the second requirement proscribing a knowledge of the probe's location has been indirectly inferred from solvent or model membrane studies, or both (3). Such data can be taken as qualitative evidence only. Recently, Sklar et al. (Ref. 7  discovery of a naturally occurring fluorescent compound, P-parinaric acid, which can readily be used as a probe that can satisfy both of the above requirements. P-Parinaric acid is a linear transconjugated polyene fatty acid with four double bonds and 18 carbon atoms. These investigators characterized the behavior of the molecule in artificial lipid systems and found that P-parinaric acid (either the free acid, or esterified to the position 2 of phosphatidyl choline) was incorporated into artificial lipid membranes. Factors favoring the use of P-parinaric acid as a natural fluorescent probe are: (a) the hydrophobic nature of fi-parinaric acid is typical of other lipid membrane constituents; (b) the molecule is approximately the same size and shape as other fatty acids and would not be expected to alter its environment in a manner differing from that of other fatty acids (For example, it has been observed to interact with serum albumin as do other fatty acids' (7).); (c) P-parinaric acid can be biosynthetically incorporated into Escherichia coli and rat liver lipids* (7); (cl) the fluorescence of the molecule is sensitive to the chain length of membrane fatty acids as well as the type of polar head group of phospholipids3 (7); (e) characteristic temperatures of phospholipids reported by P-parinaric acid are identical to those reported by differential scanning calorimetry3 (7-9). In the present study we have  Ames (19). The lipid extract was further filtered through Na,SO, and glass wool and dried under nitrogen gas as described by Wisnieski et al. (20). Contamination of the lipid extract by protein was determined by the method of Lowry et al. (21) and by measurement of fluorescence due to aromatic amino acids. Both methods indicated negligible contamination of the lipid extracts by protein.
Incorporation of &Parinaric Acid into LM Subcellular Membranes and Lipids-The fluorescence probe, P-parinaric acid, was stored in hexane (3 mg/ml) as described by Sklar et al. (7). Working solutions of the probe were prepared fresh daily by dilution in ethanol (1:lOO). Aliquots of the working solution were placed in acid-washed, Tefloncapped, Pyrex test tubes and the solvent was evaporated with a gentle stream of nitrogen. A 1.5ml aliquot of membrane suspension (50 pg of protein/ml of PBS, pH 7.4) was added to the tube. The sample tube was flushed with nitrogen, capped, and blended at maximum speed on a Vortex Genie mixer for 3 min. Membrane lipid dispersions were treated similarly except that the isolated lipids were redissolved in chloroform:methanol (2:l) and an aliquot was added to the sample tube. The solvent was evaporated with nitrogen and 1.5 ml of PBS was added. The sample was then blended on a Vortex mixer as above. The final concentration of lipid in the sample tube was equivalent to that extracted from 50 Kg of membrane protein/ml of PBS. Unless otherwise specified, the molar ratio of P-parinaric acid probe to lipid was in all cases between 1:500 to 1:lOO. All of the above procedures with /3-parinaric acid were carried out under N, and reduced light. Instrumentation and Spectroscopy-The computer-centered spectrofluorimeter was used in the excitation mode to measure absorption (A), absorption-corrected fluorescence (CO) which corrects for the inner filter effect as well as instrumental variables, and relative fluorescence efficiency (RFE) spectra. RFE is proportional to the quantum efficiency of fluorescence.
In the emission mode the instrument was utilized to determine corrected fluorescence emission spectra (10, 11). These parameters were defined in more detail elsewhere (10)(11)(12) and RFE is identical with the quantity, partial quantum 'Schroeder, F., Perlmutter, J. F., Glaser, M., and Vagelos, P. R. In addition, during excitation scans it outputs RFE, a quantity which is directly related to the total fluorescence efficiency of a fluorophore, thereby eliminating the necessity for using two scans on different instruments to measure fluorescence efficiency.
The recorded values along the scan axis of each of the above quantities (A, CO, corrected fluorescence emission, and RFE) represent the average of 40 measurements taken by the computerized system in several milliseconds. Each value deviated less than 0.2% in repeat scans. In each repeat scan an additional 40 measurements were taken with the same sample at a given temperature.
Data were taken every 0.25 nm during a scan (10, 11). The sample temperature was continuously monitored with a thermocouple.
Data were automatically obtained by the computer every 1" change during temperature scans over the range 13 to 50". The sample and reference temperatures were controlled by a water-jacketed cuvette holder and the temperature was varied at a rate of 2"/min.
Unless otherwise stated, samples were equilibrated at the lowest temperature for at least 30 min before increasing temperature scanning.
Plots of CO,,, of P-parinaric acid versus temperature in solvents, such as ethanol, indicated an exponential decay with increasing temperatures, but no discontinuities or characteristic temperatures were found. Thus the characteristic temperatures determined under "Results" do not appear to be a systematic instrumental artifact. In addition, others (7) have demonstrated similar exponential decreases in fluorescence of a-parinaric acid in decane; also, no discontinuities were apparent. Quantum yields were determined relative to ANS (Pierce Chemical Co.) in ethanol (quantum yield, 0.37 for ANS according to Stryer (22)).

Spectral
Characteristics and Probe Environment-The probe, P-parinaric acid, is nonfluorescent in aqueous solution (7). The spectral parameters of P-parinaric acid in ethanol are shown in Fig. 1, illustrating the type of on-line data obtained with the computerized spectrofluorimeter. Fig. 1 (7). The shape of the relative fluorescence efficiency curve indicated that the chromophore was also the major fluorophore and little, if any, absorbance due to impurities was present (10, 11). Fig. 1D illustrates a corrected fluorescence emission scan and shows that /3-parinaric acid had maximal emission at 415 nm, also as predicted (7). The total quantum efficiency of fi-parinaric acid fluorescence in ethanol at 25", using ANS as a standard (22), was 0.083. The absorbance and corrected fluorescence emission spectra needed for this determination were obtained simultaneously with the same instrument. In order for a probe to be informative, its spectral characteristics must be influenced by the environment (3,23). By comparing the spectral characteristics of a fluorescence probe in a series of solvents of differing hydrophobicity, polarity, dielectric constant, or hydrogen bonding ability with the same parameters measured in membranes or membrane lipids, it is possible to qualitatively ascertain the type of environment in which the probe may be located in the more complex membrane or membrane lipid systems. Both the wavelength of fluorescence emission and the quantum efficiency of fluorescence may be sensitive to the dielectric constant, t, as is the case for ANS and iV-phenyl-l-naphthylamine (3,23,24). Such data have been interpreted to correlate with the "polarity" of a probe's environment or binding site. However, such interpretations may be oversimplified since polarity is dependent on at least three factors: dielectric constant of the solvent, dipole-induced dipole interactions between the probe and its solvent environment, and the polarization of solvent molecules induced by the fluorescence probes. As shown in Fig. 2, the spectral parame- were measured at 313 nm with emission maintained at 415 nm as described under "Materials and Methods." The emission wavelength maxima (04) were measured from corrected fluorescence emission spectra with excitation at 313 nm as described in Fig. 1. The values of these parameters for ,Kparinaric acid dissolved in methanol or ethanol were indicated on each curue by X and A, respectively. 6 values were taken from previously published data for methanol, ethanol, and 0 to 100% dioxane:H,O (24). The concentration of /3-parinaric acid was approximately 2.0 mg/ml. ters of P-parinaric acid were measured as a function of dielectric constant, e. The dielectric constant was varied by dissolving the acid in a series of dioxane:water mixtures as previously described by ANS (24). In contrast to results obtained with ANS and N-phenyl-l-naphthylamine (24), the wavelength of maximal fluorescence emission of P-parinaric acid was relatively constant over a wide range of t rather than continuously varying as a function of c. Similar results have been obtained by Simoni and co-workers (Ref. 7 and Footnotes 1 to 4). The fluorescence of the probe in methanol and ethanol was used to confirm this relationship.
The large decrease in -Ls and CO,,, with increasing t (Fig. 2B) indicated that P-parinaric acid became increasingly insoluble in polar environments such as H,O. Water has an e of 80 (24). Similar solubility problems in aqueous media have been encountered with other polyenes (12, 13). These data indicate that the behavior of P-parinaric acid in aqueous solvents is probably not ideal and that micelles and aggregates may form at high concentrations. It is important to note that the absorbance of P-parinaric acid in water, when added to the water as described under "Materials and Methods" by first coating the sides of the test tube before addition of aqueous solvent, is low. The probe is so poorly soluble that the absorbance is almost negligible (see Fig. 2). However, the large decrease in RFE,,,, which is independent of fluorophore concentration, with increasing c, indicates that other factors, e.g. solvent polarizability (3,25,26) are affecting the fluorescence efficiency of this fluorophore.
It was previously shown that the ratio of absorbance peaks may be a sensitive indicator of polyene chromophore conformation and the ratio of fluorescence peaks was a measure of noncovalent interaction with the fluorophore (12, 13). Table I illustrates that the ratio of absorbance maxima (A,,,/A,,,) of P-parinaric acid, in various solvents of varying hydrophobicity and t, was relatively constant. However, the ratio of the fluorescence excitation maxima, especially CO,,,/CO,,, varied as a sensitive function of the fluorophore environment. These data were consistent with the known sensitivity of polyene fluorophores to polarizability and other factors affecting the excited state (25,26) rather than conformational changes of the chromophore. /3-Parinaric acid located in plasma membrane or plasma membrane lipids had low values of CO,,,/CO,,, Lipids-The intensity of fluorescence of a probe located in membranes or lipids is a concentration-dependent parameter and will also be dependent on factors, such as binding of the probe to the membrane, affecting the fluorescence efficiency (24,27).  COsl, or RFE,,, measured at 24". The wavelength of maximum emission was constant with time and independent of the temperature at which CO and RFE were measured. Since the absorbance of &parinaric acid in water was very low (Fig. 2), the absorbance of unbound probe would not be expected to contribute significantly to RFE,,, of probe incorporated into membranes or lipids. The binding characteristics of p-parinaric acid with plasma membranes and plasma membrane lipids of LM cells are shown in Table III. The dissociation constant, K,, increased only slightly (10%) with increasing temperature. Concomitantly the number of lipid molecules/p-parinaric acid binding site decreased by approximately 10 to 15%. In addition the Coal, of @-parinaric acid in the plasma membrane or plasma membrane lipids was found to increase in hyperbolic fashion reflecting an increase in unbound probe as saturation is approached while RFE,,, remained independent of probe concentration. Because of the low solubility and probability of micelle formation of fatty acids in water, the accuracy of the K, values may be suspect. However, we have attempted to partially circumvent this problem by coating the probe fatty acid on the side of the tube and then blending on a Vortex mixer with membranes or lipids in buffer (see "Materials and Methods") and this would not require a high concentration of probe in the aqueous medium once equilibrium of P-parinaric acid was established between the glass wall, the aqueous buffer, and the membrane vesicles. The membrane or lipid vesicles would absorb the probe from the side of the reaction vessel in a binding process that may reflect saturation better than if the fatty acid probe were simply added in ethanol solution at concentrations possibly much higher than the critical micelle concentration.
We have not, however, determined the critical micelle concentration of @-parinaric acid in aqueous buffer solution. Extensive binding studies with model lipids indicated similar characteristics and K, values of the same order of magnitude with 8-parinaric acid as noted here. 3 Characteristic Temperatures of Membranes hnd Membrane Lipids Indicated by @-Parinaric Acid-Fluorescent molecules have been used as probes of the physical state of lipids in membranes, and alterations or transitions in the behavior of the probes have been correlated to physiologically important parameters (3,23,24,27). Sklar et al. (7) have shown that fluorescence measurements of @-parinaric acid can be used to varied at 2"/min. FIG. 5 (center). Characteristic temperatures of microsomes and isolated microsomal lipids. All methods were as described in Fig. 4. FIG. 6 (right). Characteristic temperatures of mitochondria and isolated mitochondrial lipids. All methods were as described in Fig. 4.
monitor the phase transitions of artificial bilayer lipid membranes. We have extended the studies to animal (LM) cell membranes. Figs. 4 to 6 show plots of P-parinaric acid COsls, a concentration-dependent parameter, and RFE,,,, a quantum yield-dependent parameter, uersus temperature in plasma membranes, microsomes, mitochondria, and their isolated lipids. Both parameters indicate characteristic temperatures of plasma membranes (Fig. 4), microsomes (Fig. 5), and mitochondria (Fig. 6) at approximately 18, 23, 31, 38, and 43". The characteristic temperatures of the isolated lipids were in agreement &lo. These same characteristic temperatures (+l') were noted at four different probe concentrations (20-fold range) and in descending as well as ascending temperature scans (data not shown) for LM plasma membranes, microsomes, mitochondria, and their isolated lipids. Very little probe decomposition appears to have occurred since almost full fluorescence intensity returned after the second scans.
As indicated in Figs. 4 to 6, CO,,, and RFE,,, decreased by 70 to 80% with increasing temperature for both membranes as well as lipids. Part of this decrease is due to a decrease in fluorescence efficiency at elevated temperatures (3). A second possibility was that large alterations in probe binding ability as a function of temperature may have occurred. Similar behavior has been noted with ANS (27). As indicated by CO,,, and RFE,,, in the previous section, large alterations in binding ability of P-parinaric acid as a function of temperature did not occur at the probe concentrations employed. Accessibility of Plasma Membrane Lipids to P-Parinaric Acid-It is possible that P-parinaric acid may interact with only a small group of lipids in the plasma membrane or with membrane proteins. It can interact with proteins such as serum albumin (7).' As shown by Table II, the maximal relative fluorescence efficiencies (RFE) of /3-parinaric acid in plasma membranes and plasma membrane lipids were very similar in value (within approximately 10%). The increase of approximately 10% in the isolated lipids could indicate that the extent of the probe-lipid interaction may be obstructed slightly by protein or other membrane components, or both. Thus removal of the protein appears to have slightly increased the ability of the probe to interact with the plasma membrane lipids. The protein may, therefore, sequester some lipid and prevent its interaction with &parinaric acid. In these studies, the probe near two possible transitions for plasma membranes and plasma membrane lipids as previously described (24). The accessible fraction is defined (24) as the value of ACO',: in the plasma membrane/the value in the plasma membrane lipids. concentration was very low and nonsaturating.
Since the concentration of lipid was much greater than the concentration of the probe, not only steric or lipid sequestering effects but also competition between lipid and protein for binding of the probe must be considered. Resolution of these possibilities was further tested in another way. The fraction of membrane lipids accessible to /3-parinaric acid was determined by the method of Trauble and Overath (24). This method utilizes saturating concentrations of probe as well as nonsaturating levels. The [log CO,,,] of P-parinaric acid incorporated into plasma membranes or plasma membrane lipids was plotted as a function of temperature at each of four concentrations (0.041, 0.138, 0.414, and 0.828 fig/ml). Since it was not possible to determine which characteristic temperatures should be chosen as the beginning and end of a "transition," two sets of temperatures were arbitrarily chosen between 24 and 30" and between 37 and 43". The width of such transitions are similar to those of /3-parinaric acid in model systems (7). The ACO,,, was measured at each of the above concentrations for plasma membranes and plasma membrane lipids. The reciprocal of ACO,,, was then plotted versus the reciprocal of the concentration of P-parinaric acid. These plots, when extrapolated to where p-parinaric acid]-' equals zero, gave a limiting ACO,,,. The latter data are summarized in Table IV. The ratio of the limiting ACO,,, of the probe in the membrane lipid/limiting ACO,,, of the probe in the intact plasma membranes is defined as the fraction of membrane lipid accessible to the probe (24). Between 88 and 94% of the lipids available in the lipid extract of the LM cell plasma membrane appeared to be accessible to P-parinaric acid in the intact plasma membrane and took part in the phase transitions.
It is not known why the lower temperature transition and the upper temperature transition, arbitrarily assigned here, should be of opposite sign (compare Fig. 4, A and C). Possibly these discontinuities reflect phenomena other than transitions in the physical state of lipids. In addition all of the accessible lipids may not participate in the transition.
Such possibilities are considered under "Discussion." Despite these shortcomings in interpretations, these values of accessible lipid were similar to those reported previously with microbial membranes (24,(28)(29)(30) and mammalian membranes (31) by a variety of methods.

DISCUSSION
The data presented here indicate that fi-parinaric acid may be an ideal fluorescence probe for measuring changes in the physical properties of the isolated membranes and extracted membrane lipids from LM cells grown in suspension culture. fl-Parinaric acid interacted quickly at room temperature with membranes of lipids and maximal interaction occurred in less than 3 to 5 min. Other probes require lengthy incubation times and often elevated temperatures (20,24,32). P-Parinaric acid appeared to satisfy the two major criteria for a good probe molecule (3) set forth in the introduction: (a) it is a natural molecule that is sensitive to the nature of its environment and (b) its environmental location in LM membranes or lipids can be qualitatively ascertained.
In the intact membranes the fi-parinaric acid interacted with approximately 90% of the plasma membrane lipids, ind:-ating that about 10% of the lipids were inaccessible to the probe. This would tend to support the sequestering of lipid by protein in the intact membrane as the major cause for the difference in the relative fluorescence efficiency between the isolated lipids and the intact membrane. However, the magnitude of this trend may be suspect. These results may be interpreted as being consistent with the Singer membrane model (33) which predicts that at least some of the lipids may be tightly bound by proteins and that a microheterogeneity of lipids may exist within membranes of mammalian cells. Such a microheterogeneity is indicated by a multiplicity of characteristic temperatures (20, 34). As shown here, characteristic temperatures in LM cell membranes (plasma membranes, microsomes, and mitochondria) were located at 18, 23, 31, 38, and 43". The precision of the individual data points used to determine these characteristic temperatures was +0.2%. These data may indicate a close similarity in physical properties of the subcellular organelles of an animal cell line (LM suspension cells) despite considerable variation in the composition and subcellular distribution of sterols5 phospholipids,5 and ether-linked glycerides.7 However, the changes in slope were not always in the same direction at one temperature for separate samples (see Fig. 6). This indicates that the "breakpoints" may appear similar, but the behavior of the samples is not the same. Using electron spin resonance (ESR) probes, it was shown that L suspension cell microsomes and mitochondria had the same fluidity in their membranes (35). These workers also showed that considerable differences in fluidity existed between different cell types such as L suspension cells, lymphocytes, and erythrocytes (L suspension cell membranes were the most fluid). The large number (rather than the expected one or two) and location of characteristic temperatures found in the isolated suspension cultured LM cell plasma membranes, microsomes, and mitochondria were very similar to those noted with ESR by Fox and co-workers (20,34) in plasma membranes and microsomes from monolayer LM cells and membranes of New Castle Disease viruses propagated in embryonated chick eggs. However, these investigators found that isolated lipids from LM monolayer cell microsomes and New Castle Disease viruses revealed only two characteristic temperatures. LM Monolayer cell plasma membranes were compared with New Castle Disease virus lipids (20) and the data were interpreted to indicate the existence in the membranes of two hydrocarbon compartments (inner and outer monolayer of a bilayer) with different sets of characteristic temperatures. Thus an asymmetry of characteristic temperatures was proposed. Several assumptions appear to be implicit in this deduction. First, LM monolayer cell plasma membranes, microsomes and New 'Schroeder, F., and Vagelos, P. R. (1976)