INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

A COMMON RESPONSE OF poikilothermic organisms to altered temperature is a restructuring of membrane lipid composition. This restructuring occurs in diverse taxa, including bacteria (43), plants (51), and animals (17), and encompasses virtually all lipid constituents of the membrane. Temperature-induced modifications in membrane lipid composition include 1) the type and quantity of unsaturated fatty acids (14); 2) the mix of molecular species comprising a given phospholipid class (18, 56); 3) the size, hydrophobicity, and charge of phospholipid head groups (38); 4) the balance between bilayer-stabilizing and -destabilizing lipids (27, 33, 40); 5) the proportion of plasmalogen relative to diacyl phospholipids (29); and 6) the cholesterol-to-polar lipid ratio (42). Such fine tuning of membrane lipid composition is perhaps the most pervasive cellular response to temperature change and is presumed to reflect a homeostatic mechanism of fundamental adaptive significance.

One paradigm widely invoked to explain the temperature-induced remodeling of membrane lipid composition is preservation of the bulk-phase physical properties, or "fluidity" [more correctly (on the basis of the probes most commonly used to assess fluidity), membrane order], of membrane lipids(6), and is termed homeoviscous adaptation (HVA) (45). The seminal discovery by Cossins and colleagues (4, 7) that membrane lipid order is conserved in synaptosomal membranes of various vertebrates when assessed at the respective cell or body temperatures provides the strongest support for HVA. However, the observations 1) that the extent and occurrence of HVA are quite variable (8), 2) that HVA can occur without compensation of membrane function and vice versa (39, 44), 3) that some aspects of membrane remodeling (e.g., low-temperature accumulation of polyunsaturated fatty acids and bilayer-destabilizing lipids) are not consistent with HVA (discussed in Ref. 15), and 4) that membrane function is frequently poorly correlated with acyl chain order (48) all argue that factors other than, or in addition to, membrane order are important for the conservation of membrane function in a variable thermal environment. An alternative view of membrane adaptation emphasizes the dynamic phase behavior of membrane lipids (15). According to this view, lipid composition is regulated so that membranes remain within a lamellar fluid phase "window" that extends from the thermotropic transition to the gel phase at low temperature to the formation of nonlamellar lipid phases at high temperatures (32). Moreover, the proximity of the growth temperature to the liquid crystalline/reversed hexagonal (H_II) phase transition temperature (T_h) may be particularly important in the regulation of dynamic membrane properties essential for cellular processes dependent on membrane fusion, such as cell division in prokaryotes and intracellular membrane traffic in eukaryotes (10, 34). In support of this view, both Escherichia coli (40, 41) and Acholeplasma laidlawii (27) regulate the lipid composition of their membranes so that the transition to nonbilayer phases occurs ~10°C above the growth temperature (40). Collectively, these results suggest that the dynamics of membrane lipid phase behavior are conserved in response to thermal challenge.

However, lipid phase behavior has not been extensively studied in membranes of temperature-acclimated eukaryotes. Consequently, the present experiments were undertaken to examine the phase behavior and physical properties of plasma membranes isolated from hepatocytes of thermally acclimated (to 5 and 20°C) rainbow trout (Oncorhynchus mykiss). Fluorescent probes particularly sensitive to the formation of either gel (laurdan) or nonlamellar [N-[7-nitrobenz-2-oxa-1,3-diazol-4-yl] dipalmitoyl-L--phosphatidylethanolamine (NBD-PE)] lipid phases were employed along with Fourier transform infrared spectroscopy (FTIR) to determine 1) whether lipid phase transition boundaries (gel-fluid and lamellar-hexagonal) vary with acclimation temperature and 2) whether the proximity of the acclimation temperature to these phase boundaries is conserved as a consequence of temperature acclimation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. Rainbow trout (O. mykiss) ranging in body weight from 150 to 300 g were obtained from the Alchesay Federal Trout Hatchery in Whiteriver, AZ. After transport to the laboratory, fish were maintained in circular, fiberglass aquariums at temperatures of either 5 or 20°C under recirculating conditions (with the total water volume of the aquariums being replaced by fresh water every 24 h). Fish were fed a commercial diet (Glencoe Mills, Minneapolis, MN) to satiation once daily and were held on a 12:12-h light-dark photocycle. Trout were maintained at their respective acclimation temperatures for at least 6 wk before experimentation.

Preparation of plasma membranes. Hepatic plasma membranes were isolated by a combination of differential and density gradient centrifugation according to a modification of the procedure of Armstrong and Newman (3) as previously described (42, 55). Briefly, 4 g of liver representing tissue pooled from two fish in the case of cold-acclimated trout (198.5 ± 25.7 g total body wt) or three fish in the case of warm-acclimated trout (141.5 ± 18.7 g total body wt) were homogenized in four volumes of homogenizing buffer (HB: 0.25 M sucrose, 0.02 M Tris · HCl, pH 7.4 containing 1 mM of freshly prepared benzamidine and 1 mg/ml DNase) by six up-and-down strokes of a Teflon/glass Potter-Elvehjem tissue homogenizer (500 rpm). Homogenates were filtered through 250-µm nylon mesh and diluted to a final volume of nine times the liver weight with HB lacking DNase and benzamidine. The resulting crude homogenate (in 20-ml aliquots) was layered over 15 ml of 41% (wt/vol) buffered sucrose (0.02 M Tris · HCl, pH 7.4) and centrifuged at 22,600 g for 30 min in a Beckman JA-20 rotor. Membrane material that accumulated at the boundary of the 41% sucrose solution was collected by Pasteur pipette, diluted fourfold with fresh HB, and centrifuged at 7,000 g for 15 min in a Beckman JA-20 rotor. The resulting pellet was resuspended in 1 ml of fresh HB, layered over 20 ml of 18% Percoll (prepared in HB), and centrifuged at 33,000 g for 25 min in a Beckman Ti50.2 rotor. The plasma membrane band, located in the upper region of the tube, was collected by displacement with 66% (wt/vol) sucrose and diluted ~10-fold with 0.15 M NaCl, 10 mM Tris · HCl, pH 7.4 before centrifugation at 100,000 g for 2 h in a Beckman Ti50.2 rotor. The final membrane fraction (present as a thin surface film on top of a translucent Percoll pellet) was washed free of the Percoll pellet, resuspended in storage buffer (20 mM Tris · HCl, pH 7.8), and stored frozen at 80°C until used. Measurement of marker enzyme activities established that although this membrane fraction contains both apical (canalicular) and basolateral domains (but no significant contamination from endoplasmic reticulum, mitochondria, peroxisomes, or lysosomes), it is two- to threefold more enriched in canalicular (overall enrichment of canalicular membranes based on specific activities of Na⁺-K⁺-ATPase and 5'-nucleotidase is 10- to 100-fold) than basolateral membranes, as reported in previous studies (42, 55).

Lipid extraction and preparation of lipid vesicles for FTIR and model membrane studies. Total plasma membrane lipid extracts were prepared by the method of Bligh and Dyer (5). Multilamellar lipid vesicles were prepared by the reversed-phase evaporation method as described by Gruner et al. (12). Vesicles were prepared in 20 mM Tris · HCl buffer (pH 7.4) containing both NaCl and KCl, each at 72.5 mM.

Fluorescence spectrometry. The generalized polarization of laurdan (6-dodecanoyl-2-dimethyl aminonapthalene) fluorescence was employed because in model membrane systems it is particularly sensitive to the presence of gel phase lipids (35). Isolated membrane was diluted (in storage buffer) to a concentration of 100 µg membrane protein/ml. The diluted membrane suspension (4 ml) was incubated with laurdan (2 µl of a 2 mM stock solution in methanol) at room temperature (with gentle stirring in an amber bottle) for 1 h before fluorescence measurements. The generalized polarization (GP) was computed from the emission intensities at 440 nm (I_B) and 490 nm (I_R), employing an excitation wavelength of 350 nm according to the following equation as described by Parasassi et al. (35)

(1)

The fluorescence anisotropy of NBD-PE was employed because of its sensitivity in model membrane systems to the lamellar-to-hexagonal phase transition as described by Han and Gross (13). Plasma membrane diluted (to a final volume of 4 ml in storage buffer) to a concentration of 100 µg protein/ml was incubated with NBD-PE (2 µl of a 1 mg/ml stock solution in ethanol) as previously described for laurdan. Fluorescence anisotropies (r) were determined from the fluorescence intensities measured in directions parallel (I_V) and perpendicular (I_H) to the electric vector of the exciting light (employing excitation and emission wavelengths of 460 and 528 nm, respectively) as a function of temperature according to the following equation (25)

(2)

All fluorescence measurements were made using a Perkin-Elmer LS-50B luminescence spectrophotometer equipped with a thermally jacketed cell holder connected to a constant-temperature circulating water bath. The temperature of the cuvette contents was measured directly employing a fine thermocouple thermometer.

FTIR. Infrared spectroscopy measures the low-energy transitions between vibrational levels generated by characteristic motions (e.g., stretching) of different chemical bonds in the lipid molecule without introducing a potentially perturbing probe (28). Vibrational frequencies attributed to the methylene (CH₂) asymmetric stretch are particularly sensitive to the conformational order of the lipid acyl chain, shifting to lower wave numbers in the gel compared with the fluid lamellar phase. Vesicle samples were placed between two CaF₂ windows separated by a 50.8-µm (0.002 in.) Teflon spacer. Sample temperature was controlled to within 1.0°C by a combination of circulating coolant and microprocessor-controlled heating elements in a custom-designed (CIC Photonics) cell. Data were collected under continuous nitrogen gas purge with a Perkin Elmer Spectrum 2000 FTIR spectrophotometer. Seventy-five spectra were averaged at each temperature and Fourier transformed employing a strong apodization function to yield data every 1 cm¹. Peak frequencies were determined, without baseline adjustment, from second-derivative spectra following subtraction of a background scan (collected under N₂ purge) employing the Spectrum 2000 software.

Analytic and statistical procedures. Protein was assayed colorimetrically employing the bicinchoninic acid (BCA) method as described by Smith et al. (47). Discontinuities in the Arrhenius plots of laurdan GP (T^GP_l) were assigned by using the breakpoint regression routine (employing the quasi-Newton method) of Statistica. Minima in the temperature dependence of NBD-PE fluorescence anisotropy (T_h or T_NBD) were identified by fitting the data to a polynomial regression (employing SigmaPlot); the lowest calculated (theoretical) value of fluorescence anisotropy versus temperature was taken as the minimum value. From the FTIR data, the midpoint temperature (point of inflection) of the gel-fluid transition (T_m) was determined by first-derivative analysis of the temperature dependence of the methylene asymmetric stretching frequencies employing the TableCurve-2D software package; the onset (T^FTIR_l: the temperature above which lipids are in the fluid or liquid phase) and completion temperatures (T_s: the temperature below which lipids are in the gel phase) of the gel-fluid transition were similarly determined from maxima and minima in second-derivative spectra of the primary data. All statistical tests were performed using the Statistica computer program.

Materials. Laurdan (6-dodecanoyl-2-dimethylaminonapthalene) and NBD-PE were obtained from Molecular Probes (Eugene, OR). BCA protein assay kits were purchased from Pierce (Rockford, IL). Lipid standards were obtained from Avanti Polar Lipids (Alabaster, AL). Percoll, DNase I (Type II from bovine pancreas), and miscellaneous biochemicals were from Sigma (St. Louis, MO). All other reagents were of analytic grade. Organic solvents were redistilled before use.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Laurdan GP. Laurdan consists of a fluorescent napthalene moiety covalently linked to a 12-carbon lauric acid chain. In the membrane, the lauroyl chain intercalates into the bilayer interior while the napthalene moiety anchors the probe at the lipid-water interface. The emission spectrum of laurdan is exquisitely sensitive to the phase state of lipid membranes. In pure lipid membranes, the emission maximum of laurdan is red shifted by ~50 nm in the liquid crystalline as opposed to the gel phase (35). Figure 1A illustrates that laurdan displays similar behavior when incorporated into trout liver plasma membranes. Whereas the intensity of the emission spectra measured at low temperatures (e.g., 5°C) is dominated by a peak at ~440 nm, the amplitude of this peak declines and a secondary, red-shifted emission peak appears at ~490-500 nm as temperature is increased. This red shift in the emission spectra (Fig. 1A) reflects a change in both the polarity of the probe environment and the rate of relaxation of molecules (most likely water) that can reorient around the fluorescent moiety during its excited-state lifetime (36). The GP of laurdan fluorescence, as measured by the difference in fluorescence intensity of the blue and red emission maxima compared with the total fluorescence intensity (refer to Eq. 1), thus provides a sensitive indicator of the onset of the gel-fluid transition in lipid membranes of simple and defined composition. This is illustrated in Fig. 1B, which shows that Arrhenius plots of laurdan GP in multilamellar vesicles of dimyristoyl phosphatidylcholine (DMPC) display an abrupt discontinuity at the gel-fluid transition temperature (24°C) (36); GP values are relatively high (>0.4) in the gel phase (i.e., at temperatures >24°C) and low (<0.3) in the fluid phase.

View larger version (32K): [in this window] [in a new window]   Fig. 1.   A: effect of assay temperature on emission spectra of laurdan in hepatic plasma membranes from 20°C-acclimated trout. Note that emission spectra are red shifted at elevated temperatures, indicating increased proportions of fluid-phase lipids. B: effect of assay temperature on generalized polarization (GP) of laurdan in multilamellar vesicles of dimyristoyl phosphatidylcholine (DMPC), presented as both an Arrhenius plot and a direct linear plot (inset). Temperature of gel-fluid transition is indicated by arrow.

Arrhenius plots of laurdan GP for probe incorporated into plasma membranes of both 5- and 20°C-acclimated trout exhibited a single discontinuity (Fig. 2); this discontinuity was a consistent feature of the data and was observed in all experiments (n = 5). Most notably, the discontinuity occurred at a significantly lower (P = 0.014) temperature in plasma membranes of 5°C (7.2 ± 0.1°C)- than 20°C (13.2 ± 0.7°C)-acclimated fish (Table 1). Fitting the data to a model that assumed a single breakpoint and two linear segments of dissimilar slope explained more than 99.5% of the variance in the data. Similar discontinuities were present at the same temperatures in both direct-linear [laurdan GP vs. temperature (°C)] and log-linear [log laurdan GP vs. temperature (°C)] plots (data not shown), indicating that the discontinuities are not an artifact of the mode of data presentation. Slopes at temperatures below the discontinuity were significantly lower than those at higher temperatures (by a factor of ~1.6-fold).

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Arrhenius plot demonstrating effect of temperature on generalized fluorescence polarization of laurdan in plasma membranes isolated from hepatocytes of 5- and 20°C-acclimated rainbow trout. Typical data from a single experiment are illustrated.

CH₂ asymmetric stretch: FTIR. The temperature dependence of the methylene stretching vibrations derived from the acyl chains of membrane phospholipids has been extensively employed to define the thermal range of the gel-fluid phase transition (2, 28, 49). As illustrated in Fig. 3 for DMPC (inset), the wave number of this vibration increases dramatically with warming as the sample passes through the gel-fluid transition (at 24°C). In phospholipid vesicles from trout hepatocyte plasma membranes (Fig. 3), the gel-fluid transition occurs over a broader temperature range (Table 1; ~11 to 16°C, for 5- and 20°C-acclimated fish, respectively), indicating a less cooperative melting process, most likely reflecting the complex and heterogeneous nature of the lipids involved. Although neither T_m (1.6 to 2.3°C) nor T_s (approximately 9°C) of the gel-fluid transition differed significantly between acclimation groups (Table 1), T_l (defined from a cooling perspective) was significantly lower in cold (approximately 2.0°C)- than warm-acclimated (7.2°C) fish. Accordingly, the width of the gel-fluid transition was significantly broader in warm (~16°C)- than cold-acclimated (~11°C) trout (Table 1). In contrast, although membranes of cold-acclimated fish were consistently less ordered than those of warm-acclimated fish (i.e., characterized by higher wavenumbers) at temperatures above 0°C, differences in membrane order between acclimation groups were not significant in the physiologically relevant temperature range (0-25°C).

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Temperature dependence of the asymmetric methylene stretching vibration frequency in multilamellar lipid vesicles prepared from hepatic plasma membranes of 5- and 20°C-acclimated rainbow trout. Typical data from a single experiment are illustrated. Inset: temperature dependence of methylene asymmetric stretching vibration for DMPC.

Anisotropy of NBD-PE. NBD-PE possesses a fluorescent reporter moiety covalently attached to the head group of dipalmitoyl phosphatidylethanolamine. Because the motional freedom of the polar head group is reduced in the hexagonal compared with the lamellar phase (13), a transition to the hexagonal phase is expected to result in a substantial increase in the fluorescence anisotropy of fluorophores in the polar head group region. Conversely, within the lamellar phase, head group motion is positively correlated with temperature, and fluorescence anisotropy thus decreases with rising temperature until T_h is attained. Consequently, minima in the temperature dependence of the fluorescence anisotropy of head group probes such as NBD-PE define T_h. As illustrated in Fig. 4, this technique accurately reflects the differences in T_h for both 1-palmitoyl-2-oleoyl phosphatidylethanolamine (T_h = 67°C; Ref. 13) and binary mixtures of dioleoyl phosphatidylethanolamine-dioleoyl phosphatidylcholine (T_h = 36°C for a 5:1 molar ratio; Ref. 13) employed as reference standards.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Effect of temperature on fluorescence anisotropy of NBD-PE incorporated into multilamellar vesicles of 1-palmitoyl-2-oleoyl phosphatidylethanolamine (POPE) and a binary mixture of dioleoyl phosphatidylethanolamine (DOPE)-dioleoyl phosphatidylcholine (DOPC) (80:20). Observed minima correspond well with reported hexagonal phase transition temperatures of these vesicles (11).

The temperature dependence of NBD-PE anisotropy was more complex in trout liver plasma membranes (Fig. 5). Anisotropies for both 5- and 20°C-acclimated trout displayed a clear minimum between 56 and 58°C (Table 1). However, plasma membranes of 5°C-acclimated trout differed significantly from those of 20°C-acclimated trout in displaying an additional, clearly defined minimum at ~25°C (Table 1), resulting in higher NBD-PE anisotropies (i.e., lower head group mobilities) in cold- than warm-acclimated trout at temperatures above 25°C (Fig. 5). The similarity of NBD-PE anisotropy values between acclimation groups at temperatures below 25°C indicates a lack of significant differences in the membrane properties sensed by this probe at physiological temperatures.

View larger version (27K): [in this window] [in a new window]   Fig. 5.   Effect of assay temperature on fluorescence anisotropy of N-[7-nitrobenz-2-oxa-1,3-diazol-4-yl] dipalmitoyl-L--phosphatidylethanolamine (NBD-PE) incorporated into hepatic plasma membranes of 5- and 20°C-acclimated trout. Typical data from a single experiment are illustrated. Consistently identified minima are indicated by arrows.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Because changes in temperature may shift the thermodynamic balance of forces within a membrane toward lipid phases that are biologically nonfunctional, the objective of the present experiments was to determine the extent to which the boundaries (i.e., transition temperatures) between the phases assumed by membrane lipids are adjusted as growth temperature changes. As illustrated in Fig. 6, the fluid lamellar phase (indicated graphically in Fig. 6 by the total length of the rectangle) is delimited at low temperatures by transition to the gel phase (at T_m) and at higher temperatures by the transition to nonlamellar (e.g., hexagonal, cubic, or micellar, at the lamellar-nonlamellar transition temperature, T_NL) phases. Excursions in temperature either below T_m or above T_h significantly perturb membrane structure and function (20, 46, 53, 54). To avoid such problems, membrane lipid composition is adjusted so that the growth temperature of the organism (Fig. 6) lies within the boundaries of the fluid lamellar phase (i.e., within the temperature interval between T_m and T_NL). Although thermal compensation of membrane lipid composition is well established, the extent to which the lipid packing forces that determine the phase behavior of isolated membrane lipids are conserved at different growth temperatures is not currently clear, especially in eukaryotic poikilotherms. Accordingly, the spectroscopic approaches employed in the present experiments were selected because they are well suited for detecting changes in the phase behavior of membrane lipids both within the lamellar phase (laurdan GP and FTIR) and between lamellar and nonlamellar lipid phases (NBD-PE). It is worth noting, however, that in most instances the methods employed sense primarily average membrane properties and are not expected to reflect the properties of specific microdomains that may be important for some membrane functions.

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Dynamic phase behavior model of thermal adaptation in biological membranes. Thermal range over which a fluid lamellar phase is maintained is indicated by the total length of the rectangle [i.e., interval from midpoint temperature (Tm) to nonlamellar temperature (TNL)]. An acute change in ambient temperature alters the relationship between acclimation (i.e., body) temperature (TA, i.e., the temperature at which the membrane is functioning) and the transitions to the gel Tm and nonlamellar TNL phases: a drop in temperature from the warm acclimation temperature (TAW) to T'A (indicated by "transitional" diagram in the center of the figure) increases the interval between TA and TNL (indicated by increased length of shaded area within the larger rectangle), while decreasing the interval between Tm and TA; a rise in temperature would have the opposite effects. Acclimation or adaptation to the lowered temperature (bottom) restores the proximity of TNL and Tm to the cold-acclimation temperature (TAC).

Membrane behavior at low temperature: the GP of laurdan fluorescence and the asymmetric methylene stretching vibration. Hepatic plasma membranes of both warm- and cold-acclimated trout display distinct and reproducible discontinuities in Arrhenius plots of laurdan emission GP values (Fig. 2). The mathematical treatment applied to these data, which assumes a biphasic temperature dependency, accounts for the vast majority (>99%) of the variance in each experiment. Furthermore, even if one were to assume the relationship to be curvilinear, it is clear from the data in Fig. 2 that the region of maximal curvature would occur at different temperatures in 5°C- and 20°C-acclimated trout. We therefore consider a discontinuous regression model to be a valid approach for identifying the temperature(s) at which the laurdan GP/temperature relationship changes. Such an analysis reveals that acclimation from 20 to 5°C depresses the temperature of this transition by ~6°C (Table 1). One possible interpretation of these data is that the Arrhenius discontinuities reflect the initial formation of gel-phase lipid domains, i.e. the onset of the gel-fluid phase transition (T_l). Complete transition to the gel phase is presumed to occur at temperatures below those assayed in these experiments (i.e., temperatures <0°C). This interpretation is supported by the FTIR data in Fig. 3, which indicate that completion of the fluid-to-gel transition in lipid vesicles prepared from trout hepatocyte plasma membranes occurs at approximately 9°C in both groups (Table 1). Similar to the transition sensed by laurdan (T^GP_l), the onset of the fluid/gel transition detected by FTIR (T^FTIR_l) is reduced ~5°C in cold (1.9°C)- compared with warm-acclimated (7.2°C) trout. However, T^FTIR_l is ~5°C lower than T^GP_l (1.9 vs. 7.2°C for cold-acclimated trout; 7.2 vs. 13.2°C for warm-acclimated trout). Estimates of T^FTIR_l, however, are likely to be conservative and underestimate the true onset temperature of the transition by 3-5°C because they were obtained from a second-derivative function that estimates the temperature of maximal (rather than initial) change in slope of the initial function (i.e., the wave number/temperature plot). Taking this into consideration, the laurdan and FTIR estimates of T_l are in excellent agreement.

Alternatively, the discontinuities in the temperature dependence of laurdan GP and the onset of the gel-fluid transition as detected by FTIR may reflect fundamentally different events. It is possible that T^GP_l reflects a transition from the liquid-disordered to the liquid-ordered phase, which has been reported in model phospholipid membranes containing >10-15 mol% cholesterol (23, 52). The levels of cholesterol present in trout liver plasma membranes (10-20 mol% relative to phospholipid), although not sufficiently high to eliminate the gel-fluid transition (50 mol% or higher), are consistent with this possibility (37). In addition, the relatively high values of laurdan emission GP (0.4-0.5) observed at temperatures above the discontinuity in trout liver plasma membranes compared with DMPC (Figs. 1 and 2) and the relatively low intensity of the emission at 490 nm are both indicative of the presence of cholesterol in trout liver plasma membranes (36). Even though the phase behavior responsible for T^GP_l cannot be unambiguously assigned based on the present work, this discontinuity most likely reflects either a phase separation resulting in the formation of cholesterol-rich microdomains or the initiation of the gel-fluid transition.

On the basis of the foregoing results, most, if not all, of the lipids in plasma membranes of warm-acclimated trout are expected to be in a fluid state because the acclimation temperature (20°C) is well above both T^GP_l (13°C) and T^FTIR_l (7.2°C). In contrast, cold-acclimated fish, for which the acclimation temperature (5°C) lies between T^FTIR_l (1.9°C) and T^GP_l (7°C), operate much closer to the low-temperature boundary of the fluid phase and may contain some gel phase or phase-separated lipids at normal physiological temperatures. This suggestion is consistent with reports of a small but significant fraction of gel phase lipids in a variety of membranes at physiological temperatures (21, 22, 26, 57). The failure of cold-acclimated fish to depress the onset of the gel-fluid transition to an extent (relative to acclimation temperature) comparable to that in warm-acclimated fish most likely reflects the closer proximity of cold-acclimated fish to the physiological limitations imposed by freezing temperatures; there is no apparent advantage to be derived from shifting the gel-fluid transition to sub-zero temperatures that the fish cannot survive. Nevertheless, the significant reduction (of ~5-6°C) in both T^GP_l and T^FTIR_l with cold acclimation is expected to permit the maintenance of a predominantly fluid membrane to lower temperatures (2-7°C) in cold- than warm-acclimated trout (7-13°C; Table 1). Interestingly, neither T_m nor T_s are significantly influenced by growth temperature. Thus the primary locus of the acclimatory response in rainbow trout is the temperature at which gel phase or cholesterol-rich domains begin to form with cooling. These results are generally similar to those previously reported for the simpler membranes of a variety of microorganisms (reviewed in Ref. 17) and are consistent with the principle of homeophasic adaptation originally proposed by McElhaney (31), with the exceptions that, in trout, the width of the gel-fluid transition is broader and neither T_m nor T_s is subject to acclimatory adjustment.

The gel-fluid transition observed for trout hepatocyte plasma membranes is considerably broader (11-16°C; Table 1) or less cooperative than that reported for membranes of human platelets (49) and is more consistent with that previously reported for trout spermatozoa (24), most likely reflecting either 1) the heterogeneous nature of the lipid composition or 2) the presence of considerable amounts of polyunsaturated fatty acids and/or cholesterol.

In contrast to phase behavior, estimates of membrane physical properties for trout hepatocyte plasma membranes (to the extent that they can be inferred from the absolute values of either laurdan emission GP or the wave number of the CH₂ asymmetric stretching vibration) depend on the assessment method. Values for laurdan emission GP in the fluid phase did not vary significantly between acclimation groups (Fig. 2), whereas FTIR measurements indicated a less ordered membrane in cold- than warm-acclimated trout (corresponding to a regulatory efficacy of ~95%), particularly at elevated temperatures (Fig. 3). Previous studies of trout hepatocyte plasma membranes indicated nearly perfect (efficacy ~91%) compensation of membrane order in the bilayer interior [as assessed by 1,6-diphenyl-1,3,5-hexatriene (DPH) fluorescence polarization], but the same study found no significant compensation employing a probe 1-(4-[trimethylamino]-phenyl)-6-phenyl-1,3,5-hexatriene anchored at the bilayer-water interface (16). The present results with laurdan confirm the lack of compensation of membrane physical properties in the interfacial region of the bilayer, but the FTIR data in combination with the laurdan results and previous work [both on these membranes employing DPH (16) and on other membrane systems (11, 19)] support the concept of a gradient of compensation in membrane order across the bilayer, with the greatest degree of compensation being observed in the bilayer interior.

Membrane behavior at high temperature: fluorescence anisotropy of NBD-PE. Plasma membranes of both warm- and cold-acclimated trout display a well-defined minimum in NBD-PE anisotropy at relatively high temperatures (56-58°C) similar to that seen in model lipid membranes undergoing the lamellar-to-H_II phase transition (Fig. 5). This high-temperature minimum most likely reflects changes in packing constraints originating from within the hydrophobic region of the bilayer that reflect the tendency of these lipids to adopt some type of nonlamellar configuration. Unambiguous confirmation of this hypothesis requires the demonstration of nonlamellar lipid phases by more rigorous biophysical methods. The temperatures of these minima are clearly well beyond the physiological temperature range of rainbow trout. Nevertheless, in situ processes believed to be dependent on transient formation of nonlamellar lipid intermediates, including membrane fusion and intracellular membrane trafficking, are normally controlled by a suite of regulatory proteins (30, 59). One function of such proteins may be to lower the temperature at which localized nonlamellar structures are formed to within the physiological temperature range. We thus believe that the measured transition temperatures accurately reflect the intrinsic properties of membrane lipids and reliably report the relative propensities of these lipids to form nonlamellar structures.

Remarkably, membranes of cold-acclimated trout display an additional minimum at significantly lower temperatures (~25°C). It is unlikely that this low-temperature minimum reflects the formation of a nonlamellar lipid phase. Nevertheless, this minimum, present only in membranes of cold-acclimated fish, must reflect a temperature-induced change in membrane lipid packing that constrains, to some degree, molecular motions in the head group region (9); this constraint is most likely due to the thermal expansion of the hydrophobic volume with rising temperature. Increased proportions of bilayer-destabilizing lipids (such as phosphatidylethanolamine) are a common adaptation of membrane lipid composition to growth at low temperature in poikilotherms (14, 50). Because bilayer-destabilizing lipids have a greater propensity than bilayer-stabilizing lipids to form the H_II phase, the presence of elevated proportions of the former in plasma membranes of cold-acclimated trout may cause significant changes in lipid packing forces to arise at lower temperatures (~25°C) in membranes of cold- than warm-acclimated trout (57°C). The relatively large acclimatory shift (of ~32°C) in the temperature at which this change in lipid packing is first evident points to the potential significance of regulating the propensity of lipids to form nonlamellar phases, possibly because these phases have been implicated in dynamic attributes of membrane function such as vesicle fusion and trafficking. This hypothesis is supported by recent findings in E. coli (41) and A. laidlawii (1), demonstrating that organisms grown at different temperatures modulate membrane lipid composition so that the transition to the H_II phase occurs ~10°C above the growth temperature.

Perspectives

View larger version (25K): [in this window] [in a new window]   Fig. 7.   Summary of effects of acclimation temperature on phase behavior of membrane lipids in hepatic plasma membranes of rainbow trout acclimated to 5 and 20°C. Tl, temperature of the onset of the gel-fluid transition; Ts, completion temperature of the gel-fluid transition; TNBD', temperature of the minimum in NBD-PE fluorescence anisotropy unique to cold-acclimated trout. Unshaded polygons define the minimal expanse of the fluid lamellar phase; hatched regions indicate presence of gel-phase lipids; crosshatched regions indicate the presence of nonlamellar phase lipids; shaded regions indicate minimal range of the gel-fluid transition: diagonal boundary runs between the onset of the gel fluid transition as detected by laurdan GP (TGPl) and FTIR (TFTIRl), respectively; stippled region in cold-acclimated fish represents interval between first indications of altered lipid packing forces (TNBD') and TNL. Dashed arrow indicates uncertainty in the molecular events giving rise to TNBD'. Vertical arrows indicate direction of change in phase boundaries when acclimation temperature is altered. * Statistically significant change in phase boundaries.