Thermal Thresholds of Lipid Restructuring and Δ⁹-Desaturase Expression in the Liver of Carp (Cyprinus Carpio L.)

Poikilotherms respond to fluctuating environmental temperatures by activating a suite of compensatory mechanisms operating mainly at the cellular and subcellular levels of organisation (Hochachka and Somero, 1984). Perhaps the most consistent cellular response to cold is the restructuring of cellular membranes, with increasing proportions of unsaturated fatty acids causing a disordering of the membrane hydrocarbon interior that offsets the cold-induced ordering (Cossins, 1994; Hazel, 1995).

The enzymatic basis of this complex response is incompletely described, although it is clear that increased activities of desaturases are a widespread feature of membrane restructuring in microbes and plants as well as in animals. This enzyme is the terminal component of a short electron transport chain composed of cytochrome b₅ and NADH-dependent cytochrome b₅ reductase (Strittmatter et al., 1974), which in vertebrates is expressed most powerfully in the endoplasmic reticulum of the liver. The Δ9-desaturase incorporates the first double bond into saturated fatty acids at the C9–C10 position within the fatty acyl-CoA pool. This and other desaturases, together with elongases, modify the CoA-fatty acids, which are then selectively incorporated into the more complex membrane lipids of tissues by positionally specific acyltransferases. The restructuring of membrane phospholipids in the cold is thus regulated by controlling the activities of all these enzymes (Hazel, 1989; Hazel and Livermore, 1990). The Δ⁹-desaturase is of key interest since fatty acids containing a double bond at the central C9–C10 position have a maximal disordering effect upon membrane physical properties (Barton and Gunstone, 1975).

We have previously linked the cold-induction of Δ⁹-desaturase activity in carp liver to two underlying mechanisms (Tiku et al., 1996). First, desaturase activity was increased shortly after the onset of a progressive cooling regime without any change in the amount of desaturase protein. This has been interpreted as an activation of latent enzyme, possibly by a post-translational mechanism, although at present there is no direct evidence for this. The second mechanism was an increase in the amount of desaturase transcript from day 2 onwards as water temperature was reduced to 10 °C. This was associated with an increase in desaturase protein levels and a further increase in activity, consistent with induced net synthesis.

Tiku et al. (1996) suggested that these two mechanisms possessed different thermal thresholds, the initial activation response occurring with modest cooling whilst the transcriptional up-regulation required extensive cooling. However, the cooling regime employed was continuous from 30 °C down to 10 °C over 3 days and was unable to exclude other possibilities, including the possibility that the two mechanisms were temporally regulated rather than thermally regulated, so that even modest reductions in temperature would elicit a sequential expression of the two mechanisms. We have tested these ideas by modifying the cooling regime used previously to allow a detailed analysis of animals held continuously at 30, 23, 17 and 10 °C for 5 days following the onset of cooling. The livers were analysed for fatty acid composition, desaturase activity, protein and transcript expression.

Carp (Cyprinus carpio L.) weighing 400–500 g were obtained from a local farm (Westlake Fisheries, Fiddlers Ferry, Widnes, UK). Fish were maintained in 2000 l aquaria at 30±0.1 °C for at least 3 months under a constant 12 h:12 h L:D photoperiod and fed to satiation once daily on a commercial trout diet [size 50 pellet; protein 45 %, oil 21 %, ash 10 %, fibre 1 %; Trouw (UK) Ltd, Longridge, Preston, UK]. Rats (Wistar strain, 150–200 g), bred and reared in the Animal House at the University of Liverpool, were fed ad libitum on standard rat chow (Special Diet Service, BP Nutrition, Northwich, UK).

The experimental cooling regime consisted of a stepwise cooling programme, in which animals were cooled to specific temperatures and maintained at those temperatures over a 5 day period (Fig. 1). On day 1, a group of 56 carp (30 °C-acclimated) was cooled to 23 °C and held at this temperature. On day 2, a subgroup of these 23 °C fish was then further cooled to 17 °C before also being maintained at this temperature. Finally, on day 3, a subgroup of the 17 °C carp was further cooled to 10 °C. All temperature changes were at a rate of 1 °C h⁻¹ for a maximum rate of change of 7 °C day⁻¹ (Schünke and Wodtke, 1983). A group of warm-acclimated control carp was maintained at 30 °C throughout the experiment and sampled on day 0, prior to cooling, and on day 6. For each sampling time, replicate fish were killed by pithing, and their livers were rapidly removed prior to the isolation of liver microsomes.

Rat Δ⁹-desaturase was induced by a fasting–refeeding method (Strittmatter et al., 1974). Animals were starved for a period of 48 h and then refed standard laboratory chow for 24 h before a second starvation period of 48 h. Rats were then fed a fat-free diet (Special Diet Service, BP Nutrition, Northwich, UK) for 24 h, at the end of which the animals were killed and their livers removed prior to the isolation of liver microsomes.

All organic solvents, inorganic salts of analytical grade and silica gel 60, both types G and H, were obtained from BDH/Merck Ltd. All other chemicals were of analytical grade and purchased from Sigma Chemical Co. Ltd (Poole, UK). All aqueous solutions were made up using ultra-pure distilled water (Milli-Q plus, Millipore S.A., Molsheim, France). All gases were supplied by BOC Ltd, Guildford, Surrey.

Liver microsomes from carp and rat were prepared by a modification of the method of Wodtke and Cossins (1991). All procedures were done at 0–4 °C. The liver tissue was weighed, minced and homogenised in 4 vols (w/v) of 250 mmol l⁻¹ sucrose, 20 mmol l⁻¹ Hepes (pH 7.4) with eight passes in a glass/Teflon homogeniser (MSE Scientific, Crawley, Sussex, UK). The resulting homogenate was centrifuged at 10 000 g for 30 min (J2-21 preparative centrifuge, JA.20 rotor, Beckman, High Wycombe, UK) and the supernatant removed. CsCl was added to a final concentration of 15 mmol l⁻¹, and the supernatant was centrifuged at 120 000 g for 80 min (Sorvall Ultra Pro 80 ultracentrifuge, Sorvall T-865 fixed-angle ultracentrifuge rotor). The resulting microsomal pellet was gently resuspended in a saline solution containing 150 mmol l⁻¹ NaCl and 0.1 mmol l⁻¹ EDTA and 20 mmol l⁻¹ Hepes (pH 7.4), taking care not to disturb the glycogen portion of the pellet, and was centrifuged at 120 000 g for 60 min. The resulting pellet was resuspended in a small volume of the saline and used immediately or frozen at −80 °C.

The phospholipid fraction of liver microsomes was extracted according to the method of Bligh and Dyer (1959). Phospholipid headgroup classes were fractionated by single-dimension thin-layer chromatography, and the phosphatidylcholine (PC) and phosphatidylethanolamine (PE) classes were eluted and transmethylated, as described previously (Lee and Cossins, 1990). The fatty acid methyl esters were analysed by gas–liquid chromatography (610 Series FID gas chromatograph, ATI Unicam, Cambridge, UK) on a free fatty acid phase fused silica capillary column, 30 m×0.25 mm (J & W Scientific, Phase Sep., Queensferry, Clywd, UK). The methyl esters were identified by comparing peak retention times with those of known standards whose identity had been confirmed by mass spectrometry (D. R. Tocher, University of Stirling, personal communication).

The hydrocarbon static order of the liver microsomal membranes was determined by steady-state fluorescence polarisation measurements on a T-format polarisation fluorometer (PC1 photon counting spectrofluorometer, ISS Inc., Champaign, Illinois, USA) using the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene (DPH). Sample preparation and anisotropy measurements were performed as described previously (Cossins and Macdonald, 1986).

The activity of the acyl-CoA-Δ9-desaturase was measured by a modification of the method of Leifkowith (1990), whereby the rate of conversion of 14C-labelled palmitoyl-CoA to palmitoleoyl-CoA was determined. Preparation of the reaction mixture and subsequent assay steps were performed as described previously (Macartney et al., 1996). Desaturase enzyme activity was calculated as nmole of product formed per minute per milligram of microsomal protein. Radiolabelled palmitoyl-CoA, rather than stearoyl-CoA, was used as a substrate because of its commercial availability. The Δ9-desaturase converts both substrates, but previous work using carp liver (Macartney, 1994) and rat liver microsomes (Umeki et al., 1984) has shown a threefold lower specific activity with the C16 substrate, although the kinetics for both substrates was largely identical.

Total microsomal protein was assayed by the method of Bradford (1976). Equal quantities of microsomal protein (20 μg) were subjected to 12 % SDS–polyacrylamide gel electrophoresis, wet-electroblotted onto a polyvinylidene difluoride membrane (Immobilon-P, Millipore, Watford, UK) and probed with a polyclonal antibody to rat desaturase (Strittmatter et al., 1988). The use of this antibody against a 33 kDa carp protein has been described previously (Macartney et al., 1996; Tiku et al., 1996). The secondary antibody was goat antibody to rat immunoglobulin G conjugated to alkaline phosphatase, detected using the bromochloroindolyl phosphate/nitro blue tetrazolium (BCIP/NBT) substrate. Amounts of desaturase protein were determined using a laser densitometer (Molecular Dynamics, Sunnyvale, CA, USA). Band density was proportional to protein loading over the range used here (r2>0.95). Variations in the efficiency of protein transfer between blots were corrected by reference to the corresponding desaturase band for a diet-induced rat included on all blots. Amounts of carp desaturase protein were determined by normalising band intensity to the rat standard, thus overcoming variations between blots in transfer and detection efficiency. This normalised amount of desaturase protein relates to the quantity of microsomal protein used, which was constant for all samples.

Total RNA was isolated from carp livers by a modification of the method described by Chomzynski and Sacchi (1987). Total RNA (5 μg) was hybridised at 45 °C simultaneously to antisense probes for carp Δ⁹-desaturase and human 18S rRNA. The DNA probes were prepared and synthesised according to the method described previously (Tiku et al., 1996). Ribonuclease protection assays were performed as described in the RPA-II kit from Ambion. The radiolabelled probes were hybridised overnight with sample RNA at 45 °C and treated with ribonuclease to degrade single-stranded, unhybridised probe. Labelled probe that hybridised to RNA sequences in the samples was then separated on pre-run 8 mol l⁻¹ urea polyacrylamide gels. The gels were dried (Drygel Sr. SE 1160, Hoefer Scientific Instruments, San Francisco, California, USA) and placed in storage phosphor screens (Eastman Kodak Co., New York, USA) overnight before detection using the Molecular Dynamics ‘Storm’ phosphorimager (Sunnyvale, California, USA). The amounts of desaturase transcript were judged relative to 18S ribosomal RNA.

The use of 18S rRNA as a constitutively expressed control was recently criticised by Vera et al. (1997), who claimed that the level of carp rRNA in liver and pituitary cells varied on a seasonal basis. Our experiments were conducted over a period of a few days rather than over a season, and we have observed no consistent change in 18S rRNA expression in relation to total RNA over this period at any temperature.

Results are presented as means ± S.E.M. for three or four fish. Significant differences between warm-acclimated control and cooled fish were tested for using a one-way analysis of variance (ANOVA). A value of P<0.05 was used to establish statistical significance.

The effects of differential cooling on the fatty acid composition of the phospholipid classes, PC and PE, were examined 5 days after the onset of cooling (Fig. 2). The 5 day time point was chosen as a convenient point at which to compare the effects of temperature on fatty acid composition since, even with extreme cooling, we have previously observed substantial and almost complete responses within this period (Tiku et al., 1996). In the PC fraction, cooling over this period led to a linear decrease in the proportion of total saturates, corresponding to changes in the proportion of polyenes down to 17 °C and of monoenes from 17 °C down to 10 °C. The PE fraction also showed a linear decrease in saturates with cooling, compensated by monoenes down to 17 °C but not at 10 °C. The magnitude of the cooling-induced changes in fatty acid composition was substantially greater in the PE fraction compared with the PC fraction; saturates decreased by 15.8 % for a 20 °C drop in temperature, whilst the corresponding value for PC was 11.3 %.

Within the monoenoic fraction there were complex changes. Fig. 3 shows the changes in the proportions of 16:1, 18:1 and 20:1, all of which are n-9 fatty acids and which therefore may have been products of the Δ9-desaturase. In the PC fraction, there was a significant increase only in 16:1 from 2.71±0.29 % in the 30 °C-acclimated carp to 4.38±0.72 % on day 1 after cooling to 23 °C. This increase was sustained following transfer to lower temperatures and at 10 °C showed a large increase between days 4 and 5. The fluctuations in 18:1 were not significant.

In the PE fraction, the changes were much larger and more complex in time course. On cooling to 23 °C, there was a transient increase in 16:1 and a corresponding transient decrease in 18:1. The proportion of 20:1 increased significantly on days 4 and 5. On further cooling to 17 °C on day 2, the proportions of 18:1 and 20:1 were significantly increased. The increase in the proportion of 20:1 was substantially greater on further cooling to 10 °C. Thus, despite transient changes in 16:1 and 18:1, the end result after 5 days was a significant increase in 20:1 from 2.17±0.74 to 10.19±2.29 % (P<0.05). For this reason, and because of the greater change in levels of saturates, we have related the time course of changes in desaturase expression to the variations in the fatty acid composition of the PE fraction.

DPH anisotropy represents the degree of hindrance to free probe rotation and is therefore a measure of structural order within the parallel array of hydrocarbon chains (Cossins and Lee, 1985). Decreasing anisotropy values represent a less ordered membrane interior. Fig. 4 shows the DPH anisotropy determined at 20 °C for liver endoplasmic reticulum membranes isolated on day 5 for each temperature. Cooling from 30 °C to 23 and to 17 °C led to a linear decrease in anisotropy (r²=0.935; P<0.05). Further cooling to 10 °C failed to cause any change beyond that seen at 17 °C, although there was substantially greater variation in anisotropies at 10 °C.

The full time course of changes in fatty acid composition, in desaturase enzymatic activity and in amounts of desaturase protein and transcript, has been determined for animals cooled and held at 23 °C (Fig. 5) and compared with control animals held at 30 °C. The proportion of monoenes displayed a biphasic response, an initial significant increase on day 1 followed by a return to control levels and a subsequent increase on days 4 and 5. The proportion of saturates remained relatively constant over the first 3 days before decreasing significantly on days 4 and 5, concomitant with the later increase in the proportion of monoenes (Fig. 5A). There was no significant change in the polyene fraction. Desaturase activity displayed a significant 2.5-fold increase on day 1 that subsequently returned to control levels on day 3 (Fig. 5B). A substantial amount of desaturase protein was found in control animals, and this remained constant over the first 2 days following the onset of cooling (Fig. 5C). However, there was a significant and substantial increase in the amount of protein on days 4 and 5, although this did not correspond to any increase in activity; presumably, this additional desaturase was inactive. Throughout most of the time course, the amount of desaturase mRNA remained very low, although there was a small but significant increase on day 5 (Fig. 5D).

Further cooling to 17 °C led to a greater increase in the proportion of monoenes in the PE fraction than observed at 23 °C, an increase that was sustained throughout the 5 day study period (Fig. 6A). There was no significant change in the polyene fraction. Desaturase enzyme activity increased progressively over the time course (Fig. 6B), reaching a sevenfold greater activity compared with 30 °C-acclimated fish on day 5. The amount of desaturase protein remained at a relatively constant level before increasing by approximately 60 % on day 4 (Fig. 6C). Amounts of desaturase mRNA increased five- to tenfold on day 2, and there was a modest but significant increase over the remainder of the time course reaching 10- to 20-fold by day 5 (Fig. 6D).

Cooling to 10 °C induced a similar increase in the proportion of PE monoenes to that observed following cooling to 17 °C (Fig. 7A). Again, there was no significant change in the polyene fraction. The desaturase activity increased progressively following cooling to 10 °C before peaking on day 4, but the response was reduced compared with that observed at 17 °C (Fig. 7B, compare with Fig. 6B). There was a significant increase in the amount of immunodetectable protein, which increased by approximately 40–50 % on day 4 (Fig. 7C). However, in contrast to the situation at 17 °C, the amount of desaturase mRNA displayed a very large 40- to 50-fold increase, peaking on day 4 before quickly subsiding on day 5 (Fig. 7D).

The changes in lipid composition during temperature-induced membrane restructuring are undoubtedly complex.

This is mainly because the fatty acid composition of membrane phospholipids is itself complex, with varied chain lengths, unsaturation and positional distribution in the two substitution positions of the phospholipid (Stubbs and Smith, 1984; Hazel and Williams, 1990). The physical properties of the fatty acids depend upon all these features, although unsaturation perhaps dominates. Inclusion of the first unsaturated bond causes the greatest change in physical properties of any desaturation, with subsequent unsaturation bonds leading to progressively smaller effects (Coolbear et al., 1983; Stubbs and Smith, 1984). Thus, a frequently used index of fatty acid composition is the percentage saturation (Lee and Cossins, 1990).

We show here that carp liver microsomes display a linear reduction in the percentage saturation with reducing exposure temperature over the full temperature range from 30 °C to 10 °C. For the PE fraction, the percentage saturation decreased by approximately 8 % for every 10 °C reduction in temperature; the corresponding value for the PC fraction was 5.6 %. These combined effects were linked to changes in physical properties of the membrane, as indicated by measurements of DPH anisotropy, at least down to 17 °C. Over this temperature range, the changes in percentage saturation in the PE fraction were correlated with changes in levels of monounsaturated fatty acids.

We also show complex changes in the chain length distribution of the monoene fraction over the 5 days. This was clearly evident from the substantial and sustained increase in levels of 20:1 in the PE fraction with only transient changes in 16:1 and 18:1. 20:1(n-9) is the elongation product of 18:1, which is itself the desaturation product of 18:0 (stearic acid). This effect was also evident in an earlier analysis (Gracey, 1996), and Farkas et al. (1980) have demonstrated a two-to fourfold increase in the formation of 20:1(n-9) in liver fatty acids from carp exposed to cooling.

Thus, cooling treatment caused significant changes to the activity of monoene-specific elongases (Alegret et al., 1995; Prasad et al., 1986) as well as to the Δ⁹-desaturase. We have previously detected a novel gene in carp containing an enoyl-CoA hydratase domain linked to a fatty acid–CoA binding domain (A. Y. Gracey, S. Colonno-Romano, B. Maresca and A. R. Cossins, unpublished observations), and this might form part of the elongase system. Significantly in the present context, the transcript levels for this gene were increased two-to fourfold in the 2–5 days following cold treatment. The fact that changes in levels of monoenes were evident mainly in the PE fraction rather than in the PC fraction points to a considerable degree of phospholipid specificity in the acyltransferase reaction. Furthermore, the cold-induced increase in monoenes is known to be restricted to the sn-1 position of the PE fraction of carp liver endoplasmic reticulum (Tiku et al., 1996), indicating a precise positional specificity to the acyltransferase activity. Thus, lipid adjustments in carp liver in response to cold are clearly the product of coordinated changes in the activities of several classes of lipid biosynthetic enzymes.

Cooling of carp from 17 °C to 10 °C led to no further change in DPH anisotropy, and the changes in indices of unsaturated PE fatty acids were non-significant. It is worth pointing out that the variation among animals on day 5 was significantly greater than at all other phases of the cooling exposure, with respect to both DPH anisotropy and fatty acid composition. This might indicate that some individuals show further changes in these parameters at day 5 and beyond, both at 17 °C and at 10 °C. Nevertheless, the changes in DPH anisotropy were most closely correlated with the changing proportions of saturated and monounsaturated fatty acids in the PE fraction. Moreover, the change in DPH anisotropy over the full range of temperatures was more closely related to changes in the percentage of monoenes (r²=0.961) in the PE fraction than to the percentage of saturates in either the PE (^r2=0.853) or PC (r²=0.813) fraction.

We have previously found that the relatively large amount of desaturase protein expressed in 30 °C-acclimated carp had low enzymatic activity and suggested that the two-to fourfold increase in activity observed on day 1 of cooling to 23 °C, which occurred without any increase in the amount of desaturase protein, was due to the activation of the pre-existing inactive enzyme, possibly by a post-translational mechanism (Tiku et al., 1996). We now show that the initial increase in desaturase activity was transient since it disappeared within a few days at 23 °C, even though there was a subsequent increase in levels of unsaturation on days 4 and 5. The initial increase in levels of monoenes in the PE fraction was not associated with a decrease in the proportion of saturates, and only with prolonged exposure to 23 °C did the level of saturation decrease concurrently with the later increase in PE monoenes. The amounts of transcript at 23 °C remained very low for the first 4 days, and desaturase protein levels showed no change over the initial 2 days. Thus, over the short term, there was no indication of protein synthesis, and these data were consistent with our earlier suggestion of enzymic activation. During the latter part of the time course, amounts of desaturase protein increased substantially, but this appeared to precede the elevation in transcript levels shown on day 5 and was not linked to any change in desaturase activity. It therefore represents further production of inactive protein, the significance of which is not clear.

Further cooling of carp to 17 °C caused a greater and more sustained increase in desaturase activity over that observed at 23 °C, which became particularly evident on days 4 and 5. Desaturase transcript amounts increased immediately after transfer to 17 °C and continued to increase over the next few days, and this was linked to increases in the amount of desaturase protein. Exposure to 17 °C therefore appeared to induce the synthesis of additional desaturase following elevation of transcript levels. In the protozoan Tetrahymena thermophila (Nakashima et al., 1996), expression of Δ⁹-desaturase mRNA was initially detected after cooling from 35 to 27 °C and became more evident at lower temperatures, with expression reaching maximal levels at 15 °C. Although the carp system is no doubt more complex, it may be that a similar threshold system exists and that the initial cooling to 23 °C was insufficient to exceed this threshold.

Amounts of desaturase transcript showed a substantial increase on transfer of fish from 17 to 10 °C. Surprisingly, however, desaturase activity and protein levels were not markedly different from the situation described at 17 °C. This corresponds to the absence of any increase in levels of monoenes over that observed at 17 °C. Instead, the change in saturation over this range of temperatures may be linked more to altered polyenes, which is consistent with a qualitatively different response, presumably with a different underlying mechanism. Exposure to 10 °C therefore caused a substantial transcriptional response, but this was not linked with a similar large change in desaturase protein expression. This suggests that the rates of protein translation were severely affected by the reduction in temperature from 17 to 10 °C, whilst transcriptional responses were not. Thus, the close relationship between transcript expression and the subsequent production of functional protein that is frequently assumed in studies relying solely upon the measurement of transcript levels breaks down with extreme cooling; the two processes becoming ‘uncoupled’. Clearly, an understanding of the mechanisms underlying responses to different levels of cooling requires that each of the various elements must be known before an accurate picture can be drawn.

The different responses observed at 23 and 17 °C were consistent with a separation of two response mechanisms involved in the cold-induced Δ⁹-desaturase activation. First, modest cooling led to the activation of inactive latent desaturase without any increase in the previously low transcript levels. The principal evidence for this mechanism was the two-to fourfold increase (present study) and the 3.5-fold (Tiku et al., 1996) increase in desaturase activity without any measurable increase in the level of desaturase protein.

This scenario is complicated by the recent cloning of a second closely related Δ⁹-desaturase isoform (S. D. Polley, P. E. Tiku, R. J. Trueman, M. X. Caddick and A. R. Cossins, in preparation). The coding nucleotide sequences of the two genes were 91 % identical, and the genes were therefore indistinguishable using the coding sequence probe employed here for the RNAase protection assay. We have demonstrated that cooling caused increased transcript levels for only one of the isoforms (Cds2). The other isoform (Cds1) was evident in liver of 30 °C-acclimated carp, but disappeared within 24 h of cooling to 23 °C.

The presence of a second desaturase gene also raises the question of whether the anti-desaturase antibody binds to both isoforms. Given that the antibody is polyclonal, we expect that both proteins would be detected by immunoassay, although it is currently not possible to test this assertion given the high degree of sequence identity. Because Cds1 predominates in the liver of 30 °C-acclimated carp, we predict that the desaturase detected by immunoassay on day 0 is not that coded by the other cold-inducible transcript. Therefore, the initial increase in enzymatic activity observed over days 1 and 2 may be linked to the activation of the product of Cds1, whilst the increase in activity and protein levels observed on days 4 and 5 might be the product of Cds2. Activation of the former might be caused by a post-translational modification and, although this mechanism has not been characterised in any detail, the predicted carp desaturase possesses a number of consensus sites for phosphorylation, myristoylation and N-glycosylation (Tiku et al., 1996).

The enzymatic activity achieved by activation of latent desaturase is clearly limited by the amount of latent desaturase protein available in the liver of 30 °C-acclimated carp. An increased level of lipid restructuring required following cooling below 23 °C might therefore require augmentation of the activated desaturase with additional catalytic units. Exposure to 17 °C caused an increase in the amount of desaturase transcript, which was associated with increased desaturase protein and enzymatic activity. We have previously shown that the increase in transcript levels results, at least in part, from increased gene transcription; nuclear run-on analysis of carp hepatic nuclei demonstrated that desaturase transcription was up-regulated following cooling, and that this was elicited by the in vivo thermal history of the nuclei rather than the in vitro incubation temperature (Tiku et al., 1996). The transcriptional response was clearly related to the degree of cooling, with substantial responses recorded at 17 and 10 °C and with no response at 23 °C.

Desaturase activity was not induced further by transfer from 17 to 10 °C, and the lipid signature indicates that levels of monoenes were not elevated at least in the PE fraction. Nevertheless, the proportion of saturates continued to decrease with cooling, suggesting that another mechanism was operating at this low temperature. The exact nature of this response is unknown, although it may involve the deacylation/reacylation cycle, the activity of which was greater in the PE metabolism of liver isolated from cold-acclimated trout than from warm-acclimated fish (Hazel et al., 1987). The substrate preferences of acyltransferases are also temperature-dependent, favouring the incorporation of long-chain polyenes at low temperatures (Hazel et al., 1983).

The time course of cooling used here was originally described by Schünke and Wodtke (1983). It is a compromise between a rapid transition, producing a simple and immediate response, and the need to ensure that the general level of physiological activity was not so reduced that a torpid state ensued. Although desaturase activity was clearly induced upon cooling (Wodtke, 1983; Wodtke and Cossins, 1991; Tiku et al., 1996), the rate and extent of cooling will be critical factors determining the magnitude and time course of the observed response. Previous work has focused upon the complete cooling sequence without any assessment of responses at higher temperatures. Reducing the temperature from 30 to 10 °C over 3 days may be too severe a challenge for the fish and therefore may lead to a suppressed activity response at 10 °C compared with animals cooled and held at 17 °C. It seems clear that cooling to 17 °C was sufficient to induce a large increase in activity and that further cooling, despite increased levels of desaturase transcript, actually caused a suppression of this response, perhaps because of the rate-depressing effects of extreme cooling to 10 °C.

The results demonstrate that cooling-induced lipid restructuring was the result of the coordinated expression of various lipid biosynthetic enzymes. The activity of Δ9-desaturase is an integral part of this response, but with progressive cooling and time of exposure, the activity of monoene-specific elongases becomes important, as demonstrated by the substantial increase in levels of 20:1(n-9). Since both desaturases and elongases modify fatty acids within the fatty acyl-CoA pool, it is the regulation and specificity of acyltransferases that must also play a significant role in determining the membrane lipid response to cooling.

R.T. was in receipt of a studentship from, and this project was supported by grants from, the Natural Environmental Research Council (UK) and the Wellcome Trust. We thank Simon Fitzherbert-Brockholes for specimens.