Seasonality of the red blood cell stress response in rainbow trout(Oncorhynchus mykiss)

The red blood cell β-adrenergic response is of crucial importance for oxygen delivery to the tissues in numerous fish species during stressful events such as burst swimming or environmental hypoxia(Primmett et al., 1986; Nikinmaa, 1992). In salmonid fish, the magnitude of the β-adrenergic response appears to vary seasonally. Thus, Cossins and Kilbey(1989) showed that in vitro the adrenergically stimulated H⁺ efflux in rainbow trout(Oncorhynchus mykiss) red blood cells (RBCs) was gradually reduced during winter and increased abruptly early in the following spring. Nikinmaa and Jensen (1986) found that neither exercise nor adrenaline injection of rainbow trout in winter affected RBC volume or the pH gradient across the RBC membrane, in contrast to summer fish. These authors suggested that the number or affinity of theβ-adrenergic receptors (β-ARs) might be seasonally reduced or that the activity of some other step in the cascade leading to Na⁺/H⁺ exchanger (βNHE) activation might be altered. Seasonality in magnitude and temperature sensitivity of theβ-adrenergic response has also been shown in vitro in RBCs of Arctic charr (Salvelinus alpinus; Lecklin and Nikinmaa, 1999). By contrast, Tetens et al.(1988) found no difference in the magnitude and affinity of the β-adrenergic response of rainbow trout RBCs in vitro in winter and summer fish.

The seasonal loss of βNHE activity is likely to impair the normal respiratory response to stress and this might negatively affect stress tolerance. Understanding the underlying mechanisms of this seasonally reducedβ-adrenergic responsiveness and its physiological significance requires an intimate understanding of how RBC biology varies over the seasonal timescale and how this relates to circulating hormones. Thus, a reduction in rainbow trout RBC adrenergic responsiveness and flounder (Platichthys flesus) Na⁺/H⁺ exchanger activity has been circumstantially linked to the reproductive cycle(Lecklin and Nikinmaa, 1999;Weaver et al., 1999, respectively). Furthermore, plasma cortisol levels also show seasonal variation in salmonid fish(Pickering and Christie,1981), and Reid and Perry(1991) have shown an upregulation of RBC β-AR numbers in response to highly elevated plasma cortisol level. Also, Perry et al.(1996) observed an increased RBC βNHE responsiveness when cortisol was elevated through repeated physical stress.

Thus, in order to throw light on the occurrence, duration and magnitude of seasonal downregulation of the β-adrenergic response and to identify the underlying mechanisms, we have undertaken an extensive monthly monitoring programme over 27 months. We have sought to relate changes in RBC adrenergic responsiveness to the number and affinity of RBC β-ARs and also to link changes in both to a range of other physiological parameters, including reproductive condition and circulating levels of cortisol and reproductive hormones. In particular, we have sought measures of erythropoietic activity,since Lecklin et al. (2000)have shown that immature RBCs have a higher adrenergic responsiveness than mature RBCs. Thus, we have determined the cellular amounts of transcripts for both the β-AR and the βNHE through the seasons. Finally, we employed known pharmacological activators of the βNHE response to probe the effectiveness during winter and summer of the transduction pathway linkingβ-AR with adrenergically activated NHE activity.

Rainbow trout (Oncorhynchus mykiss Walbaum; Stirling strain;hatched March 1997; mean mass 728 g at the start of the study) were transferred to eight 1000 litre outdoor holding tanks in October 1999, 50 fish per tank, with a constant flow (35 l min^-1) of Windermere lake water. The fish were fed commercial feed three times per week according to the manufacturer's recommendations, and their mean mass increased to 1642 g and 1882 g in December 2000 and 2001, respectively. Water temperature was electronically logged at 6-h intervals. At approximately monthly intervals,for 27 months, eight female and eight male fish (until April 2000), two fish from each holding tank, were netted into a bucket containing anaesthetic(2-phenoxyethanol; 1:2000; v/v). Blood (∼8-10 ml) was sampled from the caudal vessels into heparinised syringes, transferred to polypropylene tubes and kept on ice. Between May 2000 and May 2001, a second blood sample (∼1 ml), for determination of β-AR and βNHE mRNA levels, was collected into sterile syringes and transferred to sterile micro-centrifuge tubes before being stored, unwashed, at -80°C. The fish were then killed by a sharp blow to the head, length and weight measured, after which the fish were pithed and sexed. The gonads were weighed in male and unovulated female fish.

Freshly drawn blood was washed three times in isotonic saline (145 mmol l^-1 NaCl, 6 mmol l^-1 KCl, 1 mmol l^-1MgSO₄, 5 mmol l^-1 CaCl₂, 5 mmol l^-1 D-glucose and 10 mmol l^-1 Hepes, pH 7.9 and an osmolality of 319±1 mosmol kg^-1; mean ± s.e.m.; N=4) to remove the buffy coat. The red cells were re-suspended in fresh saline at a haematocrit (Hct) of ∼30%. Plasma removed after the first spin was frozen (-20°C) in aliquots (200 μl)for subsequent hormone and metabolite assays. High-affinity β-ARs were measured using CGP 12177 as a β-adrenergic agonist according to Marttila and Nikinmaa (1988). Briefly,duplicate samples (50 μl) of the RBC suspension were added to a series of tubes containing 450 μl saline with final concentrations of 0.3-4.3 nmol l^-1 of [³H]CGP 12177{[(-)-4-(3-t-butylamino-2-hydroxypropoxy)-(5,7-[³H]benzimidazol-2-one)];specific activity 1.85 TBq mmol^-1; Amersham, Hertfordshire, UK}either alone (total binding, B_t) or in the presence of excess unlabelled CGP 12177 (gift from Novartis Pharma AG, Basel, Switzerland;non-specific binding, B_ns). The samples were mixed and incubated for 6 h at 4°C, during which time they were regularly agitated. The RBCs were counted using a haemocytometer (Weber Scientific International Ltd, Teddington, UK). At the end of the incubation period, 400 μl of each individual sample was filtered through glass microfibre filters (2.5 cmØ, retention ≥1.0 μm; Whatman, Maidstone, UK) held on a vacuum filtration manifold (Model 1225; Millipore, Watford, UK) attached to a vacuum pump. The filters were washed three times with 2 ml ice-cold saline to remove unbound ligand. The filters were placed in vials containing 5 ml of scintillation cocktail (Eco-Safe; Meridian, Epsom, UK), shaken vigorously and stored in the dark at 4°C for 72 h before liquid scintillation counting(Tri-Carb 2100 TR; Packard, Perkin Elmer, Seer Green, UK). The activity present in the incubation media was determined by counting of the stock standard solution activities. Counts for B_t and B_ns were multiplied by 1.25 to account for the fact that only 80% of the total volume of the samples was counted. Values for specifically bound ligand (B_s) were determined by subtraction of B_ns from B_t and subsequently converted to fmol tube^-1 by division by the specific activity. Receptor density (B_max) and affinity(K_d) were determined by use of non-linear regression analysis (Sigmaplot 4; Jandel Scientific, Woking, UK) of B_s as a function of free [³H]CGP 12177 [total([³H]-CGP)-B_t]. B_max values were then converted from fmol tube^-1 to receptors cell^-1by multiplication by Avogadro's number fmol^-1 cells tube^-1.

Cortisol may influence the number of RBC membrane-bound β-ARs upon exposure to a stressor (Reid and Perry,1991). To establish whether seasonal fluctuations in plasma cortisol concentrations influence the number of β-ARs, we incubated sub-samples of RBC suspensions under anoxic conditions at 5°C for 1 h prior to incubation with labelled and unlabelled ligand (July 2000-March 2001). The subsequent procedures were as described above.

Samples of 1.2 ml washed RBC suspension were placed in rotating Eschweiler tonometers maintained thermostatically at 15°C and equilibrated with humidified N₂ (anoxic condition) for 60 min. After 50 min of equilibration, ouabain (5×10^-4 mol l^-1 final concentration) was added and duplicate samples of the suspension were taken for Hct determination. The RBC suspension was transferred from the tonometer flask into polypropylene test tubes containing saline with ouabain(5×10^-4 mol l^-1 final concentration; 0.5% DMSO v/v vehicle) and ²²Na (∼0.05 MBq ml^-1) equilibrated to the same conditions as the blood cells, resulting in a 10-fold RBC dilution(this was defined as time zero). The ²²Na influx was measured in unstimulated cells (basal flux) and in cells stimulated by 10^-5 mol l^-1 (final concentration) isoproterenol added at 5 min.

Plasma cortisol, testosterone and oestradiol-17β concentrations were measured by previously validated radioimmunoassays (cortisol: Pickering et al., 1987;testosterone: Pottinger and Pickering,1985; oestradiol-17β: Pottinger and Pickering,1990). Plasma lactate concentration was determined enzymatically using lactate oxidase and peroxidase followed by spectrophometric analysis(Roche, Basel, Switzerland).

Eighty rainbow trout from a stock population were divided evenly between four 1000 litre outdoor holding tanks (conditions as above) and acclimated for two weeks. At the onset of the experiment, the water level in two of the tanks was dropped to ∼10 cm for ∼5 min once or twice daily for a two-week period. The control tanks were left undisturbed. Upon sampling, three fish from each tank (mean mass ± s.e.m., 1773±92 g; N=12) were anaesthetised and blood was collected and prepared for Na⁺ influx measurements exactly as described above. The abundance of RBC β-ARs was determined for RBC suspensions incubated under normoxic and anoxic conditions as described above.

Six rainbow trout from the stock population (mean mass ± s.e.m., 1847±83 g) were netted into anaesthetic and blood was sampled, washed and stored overnight for Na⁺ influx measurements as described above. The Na⁺ influx was measured as described above 5 min after addition of (1) isoproterenol (10^-5 mol l^-1; non-selective β-AR agonist), (2) forskolin(1.5×10^-4 mol l^-1; adenylate cyclase activator),(3) dibutyryl cAMP (10^-3 mol l^-1; membrane-permeable cAMP analogue) or (4) calyculin A (10^-7 mol l^-1;phosphatase inhibitor). In addition, the Na⁺ influx was measured in unstimulated RBCs.

Blood was sampled, washed and stored as described above from rainbow trout(Shasta strain; mean mass ± s.e.m., 937±68 g; N=6) during October-November 2001. This treatment assured that the RBC membrane transporters and cell volume were in an unstimulated steady-state condition (Bourne and Cossins,1982). After overnight storage, the RBCs were washed once and re-suspended in saline containing 1% bovine serum albumin (BSA) at an Hct of approximately 80%. The RBCs were separated into age fractions by fixed-angle(30°) centrifugation (10 000 g, 4°C, 15 min) in narrow tubes (diameter 4 mm, length 45 mm, volume 0.5 ml; Speckner et al., 1989;optimized for trout by Phillips et al.,2000). The youngest, least dense, cells are located in the top layer, whereas the older, more dense cells are located in the middle and bottom of the tube. The tubes were cut into a top (24±2%), middle(60±3%) and bottom (15±1%) fraction, containing RBCs of increasing age, which were washed in isotonic saline three times to remove BSA. The mean cellular haemoglobin concentration (MCHC), determined as[Hb]/Hct, was used to verify that the cells were separated according to age. MCHC is lower in younger cells than in older cells(Speckner et al., 1989; Lund et al., 2000). Adrenergically stimulated Na⁺ influx was measured in each of the different cell fractions and also in the original unseparated population of RBCs.

Total RNA was isolated from frozen tissue by homogenisation in guanidinium thiocyanate (Chomczynski and Sacchi,1987) using Trizol Reagent (Invitrogen, Ontario, Canada). After treatment with DNase I (5 units per μg RNA; Invitrogen) to remove any remaining genomic DNA, the quality of the RNA was assessed by gel electrophoresis. cDNA was synthesised from 1-2 μg RNA using random hexamer primers and Superscript II reverse transcriptase (Invitrogen). Previous studies (Nickerson, 2003) have demonstrated that the trout RBC β-AR most closely resembles the β₃-AR of mammals or theβ _4c-AR of turkey RBCs (Chen et al., 1994). Because the trout RBC β-AR is found exclusively in the blood, it has been termed β_3b-AR(Nickerson et al., 2003).β _3b-AR or βNHE mRNA levels were assessed by Q-PCR on duplicate samples of cDNA (1 μl) using a Hot StarTaq Master Mix kit(Qiagen, Ontario, Canada) and a Stratagene (West Cedar Creek, TX, USA) MX-4000 multiplex quantitative PCR system. CYBR green (Molecular Probes Inc., Boulder,CO, USA) and ROX (Stratagene) were used as DNA and reference dyes,respectively. The PCR conditions (final reaction volume=20 μl) were as follows: cDNA template=1 μl; forward and reverse [primer]=150 pmol l^-1; [Mg²⁺]=2.0 mmol l^-1; CYBR green=1:50 000 final dilution; ROX=1:30 000 final dilution; dNTP=200 μmol l^-1. The annealing and extension temperatures over 40 cycles were 58°C (45 s)and 72°C (60 s), respectively. The following primer pairs were designed using Primer3 software(http://www-genome.wi.mit.edu/cgibin/primer/primer3_www.cgi):

actin forward 5′-CAC CAT GAA GAT CAA GAT CAT YGC-3′

actin reverse 5′-ATT TRC GGT GGA CGA TGG AG-3′

β_3b forward 5′-CTT GGG CTA TGG TGG CAG TA-3′

β_3b reverse 5′-CCA TGA TAA TGC CCA AGG TC-3′

βNHE forward 5′-GGG TAA TGC GTC AGA CAA CC-3′

βNHE reverse 5′-CCA TGA TAA TGC CCA AGG TC-3′

The specificity of the primers was verified by cloning (TOPO TA cloning kit; Invitrogen) and sequencing of the amplified products. To ensure that CYBR green was not being incorporated into primer dimers or non-specific amplicons during the Q-PCR runs, PCR products were analysed by gel electrophoresis in initial experiments; single bands of expected size were obtained in all instances. Furthermore, the construction of CYBR green dissociation curves after completion of 40 PCR cycles revealed the presence of single amplicons for each primer pair. Relative expression of mRNA levels was determined (using actin as an endogenous standard) by a modification of the delta-delta Ct method (Pfaffl, 2001). Amplification efficiencies were determined from standard curves generated by serial dilution of plasmid DNA.

Throughout, the data are presented as means ± s.e.m. The data sets were analysed using analysis of variance (ANOVA; Genstat 5, Lawes Agricultural Trust, Harpenden, UK) with, for the seasonal study, individual fish, tank and time as factors; for the implant studies, fish, tank and treatment as factors; and for the separated cell study, fish and cell fraction as factors. Significant differences between times, treatment groups or fractions were determined using the estimated standard error of the differences between means. Where mean and variance were found to be interdependent, the data were log-transformed prior to analysis. Sigma Plot 4.0 was used to assess statistical significance of fitted linear and non-linear regressions.

Only data for female fish are presented. Because of differential mortality rates during the previous spawning season (male rainbow trout tend to mature one year earlier than females and are more vulnerable to fungal infections when sexually mature), sufficient numbers of male fish were available only for the initial five samples. Over the five-month period, which included one sample with a reduced β-adrenergic responsiveness (December 1999; Fig. 2), we observed no statistically significant sex-related differences in β-AR characteristics or βNHE activity. Maximum androgen and estrogen levels occurred at the same time in males as in female fish (data not shown). These limited data suggest that our observations of female fish are also valid for male fish.

Fig. 1 presents the time course over 27 months of β-AR numbers together with the measured water temperatures. The number of β-ARs varied seasonally from a minimum of 668±112 receptors cell^-1 to a maximum of 2654±882 receptors cell^-1 (N=8; Fig. 1). During late summer 2000, the β-AR number increased, remained high until late autumn and then decreased into the winter. A similar, though less pronounced, rhythm was evident in 2001 (Fig. 1). Throughout this period, the β-AR number was positively related in a linear fashion to the average ambient water temperature (P=0.002, r²=0.417; Fig. 1 inset). The equilibrium dissociation constant(K_D) for CGP binding to the RBC β-ARs varied between 0.31±0.08 nmol l^-1 and 1.35±0.26 nmol l^-1(N=8; not shown) but did not correlate with the ambient water temperature (linear regression: P=0.77).

Reid and Perry (1991)showed that cortisol applied through an osmotic pump can pre-adapt the RBCs to cope with a stressor by elevating the number of membrane-bound β-ARs, and this potentially increases adrenergic responsiveness. During our routine procedures for measuring β-adrenergic responsiveness we also stressed the RBCs by incubating cells throughout under anoxic conditions. We therefore examined whether seasonal fluctuations in cortisol, with elevated levels during the spawning periods (Fig. 2), influenced the number or affinity of functional β-ARs when incubated under oxygenated or anoxic conditions. There was, however, no statistical difference in β-AR characteristics (number and affinity)between oxygenated and anoxically treated RBC suspensions during the entire period between July 2000 and March 2001 (data not shown).

The mean isoproterenol-stimulated Na⁺ influx was 110.4±2.3 mmol l^-1 RBCs h^-1 (N=128)during the months where the β-adrenergic response was not reduced(Fig. 2). A downregulated response was arbitrarily defined as any flux lower than 85 mmol l^-1RBCs h^-1. In December 1999 and February 2001, the βNHE activity was reduced by 57% and 34%, respectively, compared with the average value, whereas during the third winter of the study there was no statistically significant reduction in the βNHE activity(Fig. 2). Surprisingly, there was no obvious dependence of the magnitude of the Na⁺ influx on the number (linear regression: P=0.83; N=163) or affinity(linear regression: P=0.68; N=163) of the β-ARs. These data suggest that the seasonal changes in βNHE activity do not reflect changes in the binding characteristics of the hormone to the receptors. Also,the Na⁺ influx did not correlate with ambient water temperature(linear regression: P=0.73; non-linear regression: P=0.72),i.e. the temperature at which the fish were acclimated, which agrees with the results obtained by Cossins and Kilbey(1989).

RBC β-AR density has been shown to decrease in response to chronically elevated plasma catecholamine levels(Gilmour et al., 1994). We exposed fish to repeated daily stress over a 2-week period in January 2001,when the βNHE responsiveness was normal, to elevate plasma catecholamine levels and test if unidentified stressors in the holding facilities would affect the β-adrenergic responsiveness. The βNHE activity was not affected by repeated physical stress, with Na⁺ influx values of 129.7±4.7 mmol l^-1 RBCs h^-1 and 128.8±7.4 mmol l^-1 RBCs h^-1 in control and disturbed fish,respectively.

Having established that the seasonal reduction in βNHE activity was not linked to seasonal changes in β-AR number or affinity, we looked for other causal factors. Binding of catecholamine to the membrane-bound β-AR triggers an adenylate cyclase-catalysed synthesis of cAMP(Mahe et al., 1985), which in turn stimulates protein kinase A to activate the βNHE(Guizouarn et al., 1993). Accordingly, we tested whether the cause of the seasonally reduced adrenergic responsiveness lay in the transduction pathway linking the receptor with the transporting effector using chemical compounds known to stimulate theβNHE by intervening at different points on a possibly long transduction pathway. None of the compounds stimulated the Na⁺ influx to a greater extent than isoproterenol (data not shown). More significantly in the present context, the extent to which these compounds stimulated the βNHE relative to stimulation by isoproterenol was similar in RBCs with a seasonally reduced response compared with RBCs with a normal response(Table 1). Thus, the reduced isoproterenol-induced βNHE activity could not be rescued by any of the transduction activators, indicating that the seasonal diminution of βNHE activity was not linked to any of the steps lying between the β-AR and the βNHE. This together with the absence of any changes in the binding affinity and number of β-ARs, suggests that the winter loss of βNHE activity was linked to changes in the βNHE itself.

The seasonal variation in β3b mRNA levels is shown in Fig. 3. The relative quantity of β3b mRNA was more or less constant between May and October, decreased markedly by ∼90% between October and December and remained at low levels between December and May. The data showed that the level of β3b mRNA,like the receptor number per cell, was correlated in a linear fashion with the ambient water temperature (P=0.01, r²=0.632) and also that the number of receptors depended in a linear fashion(P=0.005, r²=0.755) on the level of β3b mRNA present in the RBCs (Fig. 3inset).

Fig. 4 illustrates the corresponding changes in βNHE mRNA. The relative quantity increased between May and November, decreased markedly between November and December and increased again progressively between December and May(Fig. 4). Examination of the levels of βNHE mRNA at different times of the year showed that theβNHE activity had a one-sampling time delayed hyperbolic dependency on the mRNA level (P=0.038; r²=0.538) present in the cell (Fig. 4 inset).

After separation, the top layer contained 23.9±2.9%, the middle layer 61.5±3.0% and the bottom layer 14.7±1.1% of the cells(N=6). The values for MCHC are presented in Table 2. MCHC increased from the top to the bottom fractions of RBCs, showing that the cells had been separated according to age. Based on measurements for the individual fractions, MCHC of the unseparated RBCs suspension was calculated to be 4.19 mmol l^-1, which is within the 0.05 significance limits of the measured value. The activity of the βNHE of the top fraction RBCs was 23.6 mmol l^-1 RBCs h^-1 and 35.5 mmol l^-1 RBCs h^-1 higher than that of the middle and bottom fractions,respectively (P<0.01 and P<0.05, respectively; Table 2).

The seasonal fluctuations in plasma sex steroids in female fish within the experimental population are illustrated together with the βNHE activity in Fig. 5. Plasma testosterone levels in the female fish increased steeply just prior to or concomitantly with the observed major changes in the RBC βNHE activity(Fig. 5). It is noticeable that the maximal levels of plasma testosterone observed in successive years decreased gradually and in a statistically significant manner; i.e. in the third year of the study, the plasma testosterone level was reduced compared with the level in the first (P<0.01) and second(P<0.001) year of the study. Plasma oestradiol levels varied in a very predictable manner, i.e. the maximal levels and the timing of the increases and decreases were similar each year(Fig. 5).

Plasma lactate varied seasonally throughout the study, with minimal values of 1.13±0.11 mmol l^-1 (N=16) in February and maximal values of 2.54±0.24 mmol l^-1 (N=16) in August. The response to β-adrenergic stimulation, which lowers the risk of anoxic metabolism and lactic acid production, was also lowest in the winter(see above). The lactate data, therefore, indicate that, despite the reduction in β-adrenergic responsiveness, the oxygen uptake from the water was entirely sufficient to prevent activation of anaerobic metabolism under conditions of routine activity in the well-aerated holding facilities.

In agreement with previous studies(Cossins and Kilbey, 1989), the magnitude of the RBC β-adrenergic response, as indicated by βNHE activity measured under standard conditions, varied seasonally. Responses declined significantly below normal levels during some winter months and recovered during the early spring. As before, the spring recovery occurred prior to any increase in ambient temperature(Cossins and Kilbey, 1989). However, the timing of the downregulation was not consistent over the life of the experimental trout population; during the first winter the response was minimal in December, 2-3 months prior to spawning, whereas during the following winter the maximal reduction coincided exactly with the spawning period in February. The inconsistency in timing of the reduced βNHE activity was somewhat unexpected but is in accordance with the timing of peaks in plasma testosterone levels (see below). One of the factors that varied during our study was the age of the fish. To our knowledge, no studies have previously looked into possible age-related variations in timing of parameters involved in the reproductive cycle or the stress response. Also, the magnitude of reduction varied during the study; thus, it was larger (57% of the meanβNHE activity) during the first winter than during the following winter(34% of the mean βNHE activity) whilst there was no significant reduction in the third winter. In part, this may be explained by our sampling point not coinciding exactly with time of the maximal reduction but might represent a trend that is associated with increasing age or size.

In considering the magnitude of the winter reduction in β-adrenergic responsiveness, it is worth pointing out that the trout RBC βNHE activity is markedly temperature dependent, with a Q₁₀ of 7.9 over the temperature range 0-19°C (Cossins and Kilbey, 1990). Thus, assay of βNHE activity in vitroat a standard temperature of 15°C irrespective of seasonally varying environmental temperature must, therefore, profoundly underestimate the seasonal fluctuation in the in vivo RBC β-adrenergic response. Thus, the flux of 72 mmol l^-1 RBCs h^-1 measured in vitro in February 2001 at 15°C would be reduced in vivo to∼9 mmol l^-1 RBCs h^-1 at the prevailing water temperature at that time.

Repeated physical stress and confinement both elevate plasma catecholamine and cortisol levels, with opposite effects on the β-adrenergic response(Perry et al., 1996). Thus,exposure of rainbow trout to daily physical stress reduced the number of membrane-bound β-ARs determined both under normoxic and hypoxic conditions but elevated adrenergic responsiveness due to increased β-AR affinity (Perry et al., 1996). We therefore considered the possibility that (unidentified) seasonal differences in the degree of stress to which the experimental fish were exposed could have influenced RBC adrenergic responsiveness and contributed to the seasonal variation in RBC function observed during this study. However, at a time of the year when the β-adrenergic response was normal, exposure of the fish to a regime of repeated daily disturbance stress did not change theβNHE responsiveness. The seasonal reductions in adrenergic responsiveness, therefore, seem to occur independently of increased stress levels.

An obvious factor that might underlie the seasonal variations in RBCβ-adrenergic response is the effectiveness of the cellular β-AR system. Previous work has implicated altered β-AR numbers in the enhanced response to hypoxia (Reid and Perry,1991; Marttila and Nikinmaa,1988); however, the fluctuations in β-AR numbers or affinity in the present study were not in any way related to the seasonal changes inβNHE activity. An alternative explanation is that there was some impairment during winter in the effectiveness of the transduction pathway linking the β-AR with the βNHE. We stimulated the βNHE in a number of receptor-independent ways, using forskolin and calyculin A, which activate the βNHE by effects on adenylate cyclase and protein phosphatase, respectively (Seamon et al.,1981; Guizouarn et al.,1995), and the membrane-permeable cAMP analogue dibutyryl cAMP. We compared the effect of these compounds on the βNHE activity of winter fish with reduced β-adrenergic responsiveness with those in summer fish with the normal high β-adrenergic responsiveness. If winter suppression were due to impairment at a specific step in this pathway we would expect a larger effect of pharmacological activation at a downstream step, relative to the effects of the β-adrenergic agonist isoproterenol acting alone. We found that all three compounds stimulated the βNHE to the same extent in RBCs with a reduction in βNHE activity compared with RBCs with a normal response. This indicates no rescue of β-adrenergic response in RBCs of winter fish by downstream activation; the red cells from animals collected at different times of the year behaved identically with respect to transduction manipulation when tested under common experimental conditions. Whilst we cannot exclude the possibility that there is a critical step lying downstream of the calyculin A-sensitive protein phosphatase, we suggest that the limiting step during winter months in the β-adrenergic response was not upstream to the βNHE. Given that it was not due to changes in β-AR numbers,we therefore conclude that it was due to variations in the number or affinity of the exchanger itself or properties of its microenvironment.

The linkage between seasonal changes in β-adrenergic responsiveness and reduced expression of the βNHE is supported by measurements ofβNHE transcript expression. We found that transcript amounts declined in the autumn and increased in the spring, the changes in the mRNA levels preceding by one sampling period the spring increase in βNHE activity. The changes in βNHE mRNA levels could be caused either by seasonal effects on the transcriptional activity in the circulating population of RBCs or by seasonality in erythropoietic activity coupled with age-dependent changes in RBC transcriptional activity. Age-dependent reductions in mRNA levels have been reported for carbonic anhydrase and the Band 3 anion exchanger in rainbow trout RBCs (Lund et al., 2000). At the physiological level, Lecklin et al.(2000) showed that the volume increase following β-adrenergic stimulation was considerably lower in mature than in immature RBCs, suggesting lower βNHE activity in the older cells. Consistent with this observation, age-dependent reductions in enzyme activity levels have been reported for citrate synthase, cytochrome oxidase,lactate dehydrogenase and pyruvate kinase in rainbow trout RBCs(Phillips et al., 2000). Age-dependent decreases in activity levels therefore seem to be widespread at both the transcriptional and functional levels. Erythropoietic activity and the release of newly synthesized RBCs into the circulation is reduced in rainbow trout during the winter (Lane,1979), and in Baltic salmon (Salmo salar) the proportion of immature cells in circulation decreases during the winter(Härdig and Höglund,1984). Seasonal changes in the age profile of the circulating RBCs make the age dependency of the RBC β-adrenergic volume response and enzyme activities of great interest in the interpretation of seasonal changes in RBC function. We show with density-separated RBCs that the βNHE activity of the top (youngest RBCs) of the separated fractions was significantly higher than that of the middle and bottom fractions, containing older RBCs, which is consistent with the changing age profile of circulating RBCs underpinning the seasonally reduced β-adrenergic responsiveness. In the present study, we do not have any measure of the erythropoietic activity. However, previous work in the same fish-holding facilities at CEH Windermere showed that the circulating RBC number decreased during autumn and winter in mature females of the closely related brown trout Salmo trutta(Pottinger and Pickering,1987). The critical factor might be influence of low water temperatures during late autumn and winter upon erythropoietic activity(Hevesy et al., 1964),progressively reducing the proportion of young RBCs with a highβ-adrenergic responsiveness in circulation.

Elevated levels of plasma oestradiol were observed prior to and concomitant with the seasonally reduced βNHE activity in December 1999 and January-February 2001. It remains to be investigated whether seasonal elevations of plasma oestradiol play any role in the β-adrenergic response of the circulating RBCs.

Testosterone is known to stimulate erythropoiesis in humans(Barcelo et al., 1999; Snyder et al., 2000) and birds(Jones and Johansen, 1972; Robinzon and Rogers, 1979; Thapliyal et al., 1982) and is suspected to have the same effect in fish(Pottinger and Pickering,1987). Thus, we suggest that the increased testosterone levels observed during the spawning period in the present work may, by stimulating erythropoiesis, have promoted the rapid increase in adrenergic responsiveness during winter or early spring, the increasing proportion of immature RBCs with increased βNHE activity explaining the enhanced β-adrenergic responsiveness. The increases in β-adrenergic responsiveness were observed 1-2 sampling periods after the increases in plasma testosterone and decreases in βNHE activity. This delay is consistent with the lengthy lag time known to precede the entry of immature RBCs into the circulation(Lecklin et al., 2000).

We provide the first evidence linking, in a linear fashion, the abundance of a receptor, i.e. the β-AR, and its cognate transcript with the ambient water temperature. The spring increase in water temperature was closely associated with an increase in β3b-receptor mRNA and subsequently to an increase in the number of functional receptors. This seems appropriate, as oxygen consumption increases with a rise in water temperature, and stimulation of the β-AR optimizes oxygen transport(Nikinmaa, 1992). It is paradoxical, therefore, that the increase in receptor β-AR numbers did not elevate βNHE activity and, consequently, in functional terms, does not seem to improve oxygen transport. No other parameter monitored during the present study co-varied with the β3b mRNA levels and β-AR numbers. However, we cannot exclude the idea that the influence of temperature onβ3b mRNA levels and β-AR numbers was indirect. How would the lower ambient temperatures routinely experienced by fish in Northern USA, Canada and Scandinavia (0-2°C) affect seasonal fluctuations in β-ARs?Extrapolation of the linear relationship between β-AR numbers and ambient temperature to 1°C predicts 549 receptors cell^-1. This might be entirely sufficient to initiate events leading to βNHE activation. Indeed, the Windermere fish included specimens with very low β-AR numbers yet with powerfully expressed βNHEs, suggesting that exceptionally low winter temperatures would not necessarily incur greater reductions inβNHE activity.

Experimental elevation of plasma cortisol is able to increase β-AR density, probably by increasing transcription of the β3b-receptor gene. This enlarges the pool of cytosolic, physiologically inactiveβ-receptors, which can be mobilised to the plasma membrane on exposure to stress (Reid and Perry, 1991). Whilst we found considerable circannual fluctuations in plasma cortisol, with high levels during February-March, we failed to observe changes in membrane-bound β-ARs in anoxically treated RBCs compared with control RBCs. Cortisol therefore appears to play no role in the seasonal variation of RBC adrenergic responsiveness.

We showed that the number of RBC β-ARs is linearly related both to the level of β-AR transcript and the ambient water temperature. Naturally occurring seasonal fluctuation in β-AR numbers was not related to the more inconsistent fluctuations in βNHE activity. On the other hand,βNHE activity was positively related to the βNHE transcript level and depended on the age of the RBC, showing reduced activity in older RBCs. The seasonally reduced βNHE activity could not be rescued by pharmacological intervention in the transduction cascade, suggesting that fluctuations in βNHE activity were a property of the transporter itself rather than any other factor. The inconsistency in timing of the reductions inβNHE activity was matched by corresponding fluctuations in plasma testosterone. Since testosterone stimulates erythropoietic activity in birds and humans, we suggest that the seasonally occurring fluctuations in βNHE activity of trout red cells are caused by changes in the age profile of circulating RBCs. We further hypothesise that this is due to the seasonal influences of testosterone.

The authors would like to thank Mr T. R. Carrick for maintaining the experimental fish and for measurements of plasma steroid levels. This work was funded by a grant from NERC (GR3/12458).