INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

NUMEROUS STUDIES HAVE SHOWN that fluidity is an important factor in the function of biological membranes. Changes in fluidity affect the activity of membrane-bound enzymes (16, 21, 25) and the activity of transporters (23), as well as the permeability of membranes to nonelectrolytes (14), water (26), and cations (11). Given that temperature has profound effects on membrane fluidity, it is not surprising that poikilotherms adjust the composition of their membranes in ways that defend fluidity in the face of changes in body temperature. Such compensatory modifications of membrane composition are termed "homeoviscous adaptations," a phenomenon first described for Escherichia coli incubated at different temperatures (22).

Although the ways in which membrane composition is altered in response to temperature are not always consistent among species, tissues, cells, or even organelles (17), a few important trends have emerged. One prominent response to a decrease in the body temperature of poikilotherms is an increase in the percentage of unsaturated fatty acids that make up the phospholipids. In some instances, saturated fatty acids are replaced by monoenes, whereas in others, the percentage of polyunsaturated fatty acids increases. Hazel (17) compiled data from numerous temperature acclimation studies and found that in the cold treatment [change in temperature (T) ranged from 13 to 22°C], the proportion of saturated fatty acids in phospholipids decreased, on average, by 19%, with a corresponding increase in the proportion of unsaturates. Phospholipids with saturated fatty acids pack readily into bilayers, whereas phospholipids with unsaturated (and therefore, kinked) acyl chains tend to disrupt hydrophobic interactions among acyl chains of adjacent phospholipids. An increase in the proportion of unsaturated fatty acids thus results in an increase in membrane disorder and fluidity, which tends to oppose the ordering effect of a drop in temperature. The other common effect of cold acclimation is a shift in the proportions of phospholipid head groups in the membranes. The lipid component of biological membranes in marine fish is comprised mostly of phosphatidylcholine (PC) and phosphatidylethanolamine (PE) (2). Cold acclimation tends to increase the proportion of PE in membranes. A less common response is a decrease in the proportion of PC. From a variety of acclimation studies compiled by Hazel (17), the average increase in the proportion of PE phospholipids with cold acclimation (T ranged from 14 to 25°C) was 21%, and the average decrease of PC levels was 6%. The conical geometry of the PE molecule is such that its presence in a membrane is bilayer destabilizing and thus membrane fluidizing, whereas the more cylindrical PC favors the formation of lamellar bilayers and therefore tends to decrease bilayer viscosity (17). Alterations in the proportions of PE and PC can thus be used to defend membrane fluidity as environmental conditions change.

Most of the work that has been done on homeoviscous adaptation to date has involved the use of whole animal acclimation or acclimatization. Very little work has been done on membrane adaptation within regionally heterothermic animals. Dean and Hilditch (8) demonstrated in pigs that the subcutaneous fat closest to the skin has the lowest melting point, as well as the lowest ratio of saturated to unsaturated fats. Irving (19) interpreted these data in light of the regional heterothermy exhibited by pigs, in which the temperature of the skin on the dorsal surface may be as low as 10°C.

Bluefin tuna maintain an elevated visceral temperature through the use of five main heat exchangers, or retia mirabilia, which thermally isolate the viscera from the cold gills (12). By virtue of their function as heat exchangers, these tissues exhibit rather steep thermal gradients, which have been estimated at ~10°C (13). This heterothermic tissue is a good model for the study of biochemical temperature adaptation because it eliminates the potentially confounding effects of whole animal acclimation or acclimatization (13). We recently demonstrated significant enzyme adaptation along the visceral retia from bluefin tuna (13) and reasoned that rete lipids might also be adapted to the thermal gradient that exists along this tissue. Specifically, we expected to see differences in phospholipid head group class and fatty acid composition along the retia that were consistent with the principles of homeoviscous adaptation established from whole organism studies. Our data are unique in that they are the first reported data for the composition of specific phospholipids and their fatty acids along a heterothermic tissue.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Tissue collection. The visceral retia used for lipid composition analysis were obtained from seven captive bluefin tuna at a tuna ranching operation in St. Margaret's Bay, Nova Scotia. The average weight of the fish used was 357 kg. These fish were held in enclosures for 1-2 mo and were fed ~50 kg of fish (mostly Atlantic mackerel, Scomber scombrus) daily. Fish were killed between 8:30 and 8:45 AM on October 5, 1994. Fish were last fed at 6:00 PM on the previous day. Visceral retia were removed as soon after death as possible and perfused with 0.9% saline solution to remove blood before being divided longitudinally into three equal strips. Bluefin possess five sets of visceral retia, which serve different regions of the viscera (12). We were unable to sample the same rete from every fish; some of the retia sampled were associated with cecum, and some were associated with the stomach. Only one rete from each fish was used for phospholipid and fatty acid analysis. After perfusion and longitudinal sectioning, the strips of rete tissue were lashed to labeled tongue depressors and frozen in liquid nitrogen. Rete samples had a mean length of 52 ± 2 mm (SE). Samples were transported in liquid nitrogen to the University of Guelph, where they were transferred to a 80°C freezer. Strips of frozen rete tissue were partially thawed, and three pieces of tissue (between 200 and 500 mg) were dissected out for lipid extraction: one from the cold end of the rete, one from the warm end, and one from the middle.

Phospholipid extractions. Total membrane lipids were extracted using a modification of the method of Bligh and Dyer (1) as described by Glemet and Ballantyne (15). Phospholipids were separated using one-dimensional thin-layer chromatography (TLC) (15) and identified with standards.

Methylation of fatty acids. Individual phospholipid bands from the TLC plate were scraped into glass Kimax tubes, after which 2 ml of 6% H₂SO₄ in methanol was added for the methylation reaction. Heptadecanoic acid (17:0) (10 µg) was added as an internal standard, and tubes were tightly sealed with Teflon-lined caps and vortexed for 10 s. Tubes were then placed in an oven at 80°C for 2 h, after which they were removed from the oven and allowed to cool to room temperature. After the addition of 2 ml of petroleum ether, each tube was vortexed for 20 s. Tubes were centrifuged at 660 g for 6 min, and the upper petroleum ether phase was removed and transferred to a gas chromatography (GC) vial. Individual fatty acids from each phospholipid and neutral lipid fraction were then separated by GC as described below.

Gas chromatographic separation of fatty acids. Petroleum ether samples containing fatty acid methyl esters were dried down under oxygen-free nitrogen gas in GC vials and redissolved in 25 µl of CS₂. CS₂ samples were injected into a gas chromatograph (Hewlett-Packard, HP5890 series II) fitted with a flame ionization detector, an automatic injector (Hewlett-Packard, 7673A), and an electronic pressure control unit. Fatty acid methyl esters were analyzed on a DB 225 megabore fused silica column (Chromatographic Specialties, Brockville, ON) at 200°C for 30 min, which included an initial ramping from 80 to 200°C over the first minute. This method allowed for optimal peak separation. Fatty acids were identified by comparing retention times from a standard containing all fatty acids of interest. Chain lengths shorter than C:14 were not resolved and thus were not reported. Absolute amounts of each fatty acid were determined by comparison with a known concentration of internal standard, heptadecanoic acid (17:0), added to the samples before the methylation process.

Statistics and data analysis. Statistical analysis was carried out using Statistical Analysis Software (SAS Institute, Cary, NC). The significance of the main effect of position on percent phospholipid, percent fatty acid, and all of the summary statistics (percent saturates, percent monoenes, etc.) were determined using a mixed model in which each fish constituted a block. The data for some tests were transformed to satisfy assumptions of the model. Multiple comparisons were performed using Bonferroni's multiple-comparison test (20). To compensate for the large number of multiple comparisons performed within each set of phospholipid data, an adjusted, comparisonwise P value (_c) was calculated for each set of phospholipid data based on an experimentwise P value (_e) of 0.05. The adjusted comparisonwise P value was calculated according to the equation that relates the comparisonwise type I error rate to the experimentwise type I error rate as a function of the number of tests being performed (20) where n is the number of tests performed. The adjusted comparisonwise type I error rate for the phospholipid class data was 0.0085, and for the cumulative phospholipid fatty acid comparisons it was 0.0030.

For the fatty acid percent composition data, statistics were performed only on those fatty acids that comprised more than 1% (on average, across all three rete positions) of the total fatty acids detected. This criterion limited the number of statistical tests so that the adjusted comparisonwise P value would not be prohibitively low. This procedure was justified by the fact that significant differences among trace fatty acids would not have a significant effect on membrane fluidity, other than a possible cumulative effect among several trace fatty acids. It was reasoned that such potential cumulative effects would be detected in the summary statistics (total saturates, total monoenes, etc.).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Phospholipid composition of visceral rete tissue. PC comprised almost half of the phospholipids extracted from rete tissue (average of 48.4%). Sphingomyelin and PE comprised 16.5 and 15.7% of retia phospholipids, respectively. Phosphatidylinositol and cardiolipin were minor constituents, representing 3.54 and 3.93% of the total, respectively (Table 1). There was no effect of position (P  0.0085 = _c) along the retia on the percent composition of any of the six phospholipids measured (Table 1). Figure 1 emphasizes the fact that we found no differences along the retia with respect to PC and PE, the two phospholipids most often implicated in homeoviscous responses.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Mol% of phosphatidylcholine (PC) (open bars) and phosphatidylethanolamine (PE) (solid bars) extracted from 3 positions along visceral retia.

Phospholipid fatty acid composition of rete tissue. For the percent composition of cumulative fatty acids (fatty acids from all six phospholipids), there was no effect of position (P  0.0030 = _c) along the retia for any of the 12 fatty acid species tested statistically (Table 2). There was also no effect of position on any of the summary statistics (total saturates, monoenes, polyenes, unsaturation index, and chain length) for total cumulative phospholipid fatty acids (Table 2). No effect of position was detected for any of the fatty acids from any of the individual phospholipid classes or neutral lipids. Because of the lack of any significant effect of position in these data, they are not presented here for the sake of brevity.

From the total phospholipid fatty acid pool, the fatty acids that were most common were (in decreasing order of percent composition) 16:0, 18:0, 18:1(n-9), 20:5(n-3), 24:1, and 22:6(n-3). These species represented an average of 79.4% of the total fatty acid pool. Saturates represented, on average, 42.1% of the total fatty acids, monoenes 32.6%, and polyenes 25.3%. The average unsaturation index was 152 and the average chain length was 18.8 (Table 2). Figure 2 demonstrates the lack of any positional effect for any of the three classes of fatty acids (saturates, monoenes, and unsaturates).

View larger version (33K): [in this window] [in a new window]   Fig. 2.   Cumulative mol% of total saturates (open bars), monoenes (stippled bars), and polyenes (solid bars) from all 6 phospholipid classes.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

No evidence for homeoviscous adaptation along the visceral retia. The two most common homeoviscous responses to a change in temperature are changes in the phospholipid class composition and changes in phospholipid fatty acid composition (17). Neither of these responses were evident in phospholipids extracted from three positions along the visceral retia. PE is generally more unsaturated than PC (17), and this was the case for PE and PC extracted from the retia (unsaturation indexes of 265 and 141, respectively). A greater degree of unsaturation and smaller headgroup give PE a conical geometry that tends to destabilize lamellar bilayers, whereas PC assumes a more cylindrical geometry that tends to stabilize lamellar bilayers. Although it was expected that the proportion of PE would be higher and/or the proportion of PC would be lower at the cold end of the retia, there was no effect of position on the percent composition of either PE or PC. In addition, there was no effect of position on the proportions of fatty acids from any of the phospholipids. It is clear from these results that if rete lipids are adapted to the thermal gradient, the mechanism does not involve differences in phospholipid or fatty acid composition.

Because no differences in phospholipid or fatty acid composition were detected along the retia, it is tempting to conclude that a gradient of membrane fluidity must therefore exist along the thermal gradient. This is not necessarily the case for several reasons. It is important to remember that we did not measure the fluidity of rete membranes, and therefore any conclusions about membrane fluidity are speculative. Dey et al. (9) demonstrated adaptive changes in membrane anisotropy (a measure of membrane order, and therefore stiffness) in erythrocytes subjected to a gradual increase and decrease in temperature (T = 20°C, dT/dt = 0.5°C/h). Analysis of membrane phospholipid class and fatty acid composition revealed no differences that could explain the observed changes in anisotropy. Their results suggest that although significant differences in membrane composition may serve as evidence for homeoviscous adaptation, the absence of differences in membrane composition does not necessarily rule out the possibility of a compensatory change in membrane fluidity as measured by membrane anisotropy.

It is also important to realize that phospholipids and their associated fatty acids are not the only components of biological membranes. Cholesterol has been implicated in the homeoviscous response and it is possible that a gradient of membrane cholesterol content exists along the retia. Although some studies suggest that cholesterol levels do not change with temperature acclimation (6), others show a positive correlation between acclimation temperature and the cholesterol-to-phospholipid ratio (10). Cholesterol is known to spread out the transition between the fluid and gel phases of lipid bilayers, thereby buffering against dramatic changes in membrane fluidity that would otherwise accompany a temperature change (27). Other factors, such as protein content and the concentration of inorganic ions such as Ca²⁺ and Mg²⁺, can also affect membrane fluidity and could potentially be involved in homeoviscous adaptation along the retia.

Although homeoviscous responses to temperature are well-established phenomena, the response is not consistent among all membranes. In this study, phospholipids were extracted from whole tissue homogenates, and therefore the results represent a weighted average of phospholipids from all membranes in the retia. It is possible that certain membranes in the retia exhibit differences in lipid composition along the thermal gradient but that these differences were masked by the absence of an effect or an opposing effect in other membranes, but this seems unlikely.

Typically, it is the most metabolically active membranes, such as mitochondrial membranes, that exhibit the strongest homeoviscous responses (17). Although the use of frozen tissue prohibited the isolation of individual membrane fractions, the results for cardiolipin afford some insight into the effect of the thermal gradient on mitochondrial membranes, as cardiolipin occurs almost exclusively in mitochondrial membranes (7). The fact that no effect of position was detected on the proportion of any of the fatty acids in cardiolipin suggests that there was no effect of position on even the metabolically active mitochondrial membranes within the retia.

Telemetry measurements of stomach temperature in bluefin tuna suggest that the magnitude of the gradient is somewhat variable, depending on the digestive process. The temperature elevation of the digestive organs is between 10 and 15°C in this species while a meal is being digested. However, the temperature elevation above ambient can drop to 8 or 9°C when the stomach is empty (4). Although long considered a seasonal phenomenon, evidence now suggests that membrane restructuring in some species may occur in a matter of hours in response to a change in temperature (3, 18, 24). Such short-term changes most often consist of adjustments in the proportion of phospholipid headgroups such as PE and PC. Carey et al. (4) showed that the rate of temperature increase in the stomach of the bluefin tuna after a meal is ~1.0 ± 0.3°C/h. Such a rate of temperature change is slower than the rate encountered by the fish studied by Carey and Hazel (3), in which significant membrane restructuring was detected in muscle microsomes. It is therefore possible that rete membranes are continuously modified as the thermal gradient varies with feeding and digestion.

Membrane adaptation in regional heterotherms. It is also possible that the membranes of rete cells are capable of functioning over a wider range of temperatures than most membranes. Although visceral warming occurs at a rather gradual rate (~1.0°C/h), cooling of the viscera (associated with feeding) may occur at such a rate that even short-term membrane restructuring could not prevent deleterious changes in membrane fluidity.

There are no studies to date that have focused on the adaptation of membranes to a eurythermal existence. Examination of erythrocytes from endotherms could shed some light on what a eurythermal membrane might look like and how it functions in the face of sudden, acute changes in temperature. One of the consequences of endothermy, especially in cold climates, is that a thermal gradient exists from the body core to the extremities. In cold-adapted and marine endotherms, the magnitude of the temperature gradient is often dramatic. Although membranes along the thermal gradient most likely are adapted to a particular temperature range (illustrated by the example of lipids in pig skin), erythrocytes traveling from the warm body core to the extremities must be able to withstand rapid, acute changes in temperature. Surely there is no time for remodeling of the erythrocyte membrane as it travels from the base of a whale's flipper to the tip, and yet erythrocytes must function normally at both ends of this wide temperature range. If any membrane could be considered "eurythermal," it is the erythrocyte membrane in warm fishes and cold-adapted mammals.

Visceral retia are a convenient model for this kind of study because they are easily dissected out of the animal and are readily available through commercial means. However, the temperature gradient that exists in the white muscle of the bluefin tuna (from the warm core to the cooler skin) is larger in magnitude and more stable than that in the visceral retia (5). Although acquisition of bluefin white muscle is difficult these days given its astronomical value in the Japanese fish market, investigations into the membrane composition along this heterothermic tissue would yield further insight into two questions raised by the present study: does homeoviscous theory apply to the case of heterothermic tissues and, if so, what are the mechanisms?

In summary, we found no evidence of adaptive differences in phospholipid head groups or acyl chains along the heterothermic visceral heat exchangers from bluefin tuna. These results contradict basic principles of homeoviscous adaptation and are inconsistent with the results from a previous study, in which we demonstrated compensatory differences in the activity of metabolic enzymes in exactly the same tissues that were used in the present study.