Fish Physiology and Biochemistry vol. 12 no. 6 pp 513-523 (1994) Kugler Publications, Amsterdam/New York

The effect of dietary vitamin A on the immunocompetence of Atlantic salmon (Salmo salar L.) I. Thompson, T.C. Fletcher, D.F. Houlihan and C.J. Secombes Department of Zoology, University of Aberdeen, Tillydrone Avenue, Aberdeen, AB9 2TN, U.K. Accepted: December 3, 1993 Keywords: vitamin A, Atlantic salmon, immunity, diet

Abstract Atlantic salmon, Salmo salar L. were maintained on diets containing low (0.37 mg kg-' diet), normal (1.95 mg kg-l diet) and high (15 mg kg-' diet) levels of vitamin A fed at 1.5% body weight per day. After 4 months, liver vitamin A levels reflected dietary intake and growth rates of all three groups were similar. Kidney leucocyte migration and serum bactericidal activity were found to be significantly reduced in fish fed low levels of vitamin A. On the other hand, high levels of vitamin A in the diet were found to augment serum antiprotease activity relative to the levels found in the other dietary groups. However, phagocyte respiratory burst activity, bactericidal activity and eicosanoid production were unaffected by the dietary vitamin A regime, as were lymphocyte functions (lymphokine and antibody production) and both serum lysozyme and classical complement activity. That the overall immunomodulatory effect of vitamin A was small was reflected in the resistance to Aeromonas salmonicida. No significant differences were found between the different vitamin A intake groups despite a trend to decreased resistance in the low vitamin A diet group.

Introduction In recent years, a considerable amount of research has been carried out investigating the effects of micronutrients on the immune system of fish, with the aim of providing a means of enhancing disease resistance through dietary modification (reviewed by Blazer 1992). Much of this research has concentrated on the immunomodulatory effects of vitamins, particularly C and E. However, the potential benefits of feeding elevated levels of vitamin A have been largely overlooked. Vitamin A includes all those compounds with the biological activity of retinol. Such compounds are obtained directly from the diet, or indirectly through enzymatic modification of carotenoid (provitamin A) precursors (Schiedt et al. 1985). In Atlantic salmon, the

predominant provitamin is astaxanthin which is present in high concentrations in crustaceans which make up a high proportion of the diet of these fish in the wild (Torrissen 1989). Following absorption from the intestine, vitamin A is stored primarily in the liver as retinyl esters. Such stores may be rehydrolysed to yield free retinol which is bound in serum by a specific transport protein (retinol binding protein) and distributed throughout the body. Studies involving mammals and poultry (reviewed by West et al. 1991; Duhr et al. 1991) have produced significant evidence for an important role for vitamin A in regulating both specific and nonspecific defence mechanisms. A single study carried out on channel catfish has shown an effect on the fish immune system, where natural cytotoxic cell activity was shown to be enhanced by high levels of

514 Table 1. Formulation of experimental diets fed to groups of Atlantic salmon for 4 months Ingredient

g kg-' dry diet

Vitamin free casein Pre-cooked starch a-cellulose Tocopherol stripped oil Additive free lard Canning oil Vitamin A free vitamin mix' Mineral mix2 Cystine BHA mix 3 Choline chloride (40%) Edicol sunset yellow (colourant) Taste attractant 4

494 150 155 80 40 20 10 40 10 1 1 1 90 ml

Ig ingredient kg-' mix: Vitamin A (500,000 IU/g) variable; Vitamin D (500,000 IU/g) 0.96 g; Vitamin E (40%) 25 g; Menadione 1 g; Ascorbic acid 100 g; Thiamine HCI 1 g; Riboflavin 2 g; Pyridoxine 1.2 g; Pantothenic acid 4.4 g; Nicotinic acid 15 g; Biotin 0.1 g; Folic acid 0.5 g; Vitamin Bi 2 2 mg; myoinositol 40 g; a-cellulose 808 g; 2g ingredient kg- l mix: Calcium carbonate 17.75 g; Calcium tetrahydrogen di-orthophosphate 416.6 g; Potassium phosphate (dibasic) 206 g; Sodium phosphate monobasic 130 g; Sodium chloride 66.4 g; Potassium chloride 50 g; Magnesium carbonate 91 g; Ferrous sulphate 30 g; Zinc sulphate 4 g; Copper sulphate I g; Manganous sulphate 3.6 g; Potassium iodide 0.2 g; Copper sulphate I g; 3g ingredient 100 g-l mix: Butylated hydroxyanisole 20 g; Propyl gallate 6 g; Citric acid 4 g; Propylene glycol 70 g; 4Synthetic squid mix dissolved in 90 ml water: L-Aspartic acid 0.18 g; L-Threonine 0.44 g; L-Serine 0.33 g; L-Glutamic acid 0.53 g; L-Valine 0.36 g; LMethionine 0.36 g; L-Isoleucine 0.29 g; L-Leucine 0.55 g; LTyrosine 0.22 g; L-Phenylalanine 0.29 g; L-Lysine HCI 0.29 g; L-Histidine HCI 0.15 g; Taurine 3.37 g; L-Proline 14.56 g; Glycine 8.92 g; L-Alanine 2.73 g; L-Arginine 2.28 g; Betaine free base 9.1 g; Trimethylamine oxide HCI 11.38 g; Trimethylamine HCI 0.91 g; Hypoxanthine 0.47 g; Inosine 0.25 g; Adenosine-5'-monophosphoric acid (Na salt) 0.4 g; L-(+)Lactic acid 0.91 g.

vitamin A both in vivo and in vitro (Evans et al. 1984). The present study was carried out to determine the effects of vitamin A on the salmonid immune system, and to examine the potential of vitamin A supplementation as a practical means of enhancing disease resistance in aquaculture. To this end, Atlantic salmon, Salmo salar L. were maintained for 4 months on diets containing low, normal and high levels of vitamin A, after which a wide range of cell mediated and humoral immune parameters were investigated.

Materials and methods Maintenance of salmon Atlantic salmon parr (mean weight 20 g) were maintained at the Scottish Office Agriculture and Fisheries Department's (SOAFD) Marine Laboratory in Aberdeen. The fish were kept in a semi-recirculating culture system incorporating both chemical and biological filters, comprising six 420 1 capacity circular tanks containing /3 strength sea water supplied at 10 1 min- at constant temperature (11 °C), and a light regime of 10 h light:14 h dark. Four hundred and fifty parr were distributed equally between the tanks. The parr were maintained on a casein based diet (Table 1). The basal diet contained 0.37 mg vitamin A kg- 1 dry diet and this was further supplemented to give final concentrations of 1.95 mg kg- (normal) and 15 mg kg-' (high) as determined by HPLC analysis (see below) immediately prior to the start of the experiment. Diets were stored at -200 C, and no loss of vitamin A activity was observed during the course of the experiment. Paired tanks of parr were fed with each test diet at a rate of 1.5°70 body weight day-' for 4 months. The total weight of each group was determined once a month, and feeding rates were adjusted accordingly. After 4 months, specific growth rates (SGR; o body weight day-) for each group were calculated using the equation: SGR = 100(Log n Wf - Logn Wi)/t where W i and Wf were the initial and final group weights respectively after t days (Ricker 1979). Five fish were also sacrificed from each tank at this time for the assessment of liver vitamin A levels. Following confirmation that in vivo vitamin A levels reflected dietary input, immunological studies were performed on the remaining fish. For each immune parameter investigated, fish were sampled at random from the three diet groups, killed by a blow on the head and bled via the caudal vein. The blood was allowed to clot at room temperature for 2h prior to centrifugation at 10,000 x g for 5 min and the resultant serum was stored at -700 C for future analysis. After bleeding, each fish was dissected. The kidney was removed and placed on ice in

515 Leibovitz-15 medium (L15; Gibco) containing 2% foetal calf serum (FCS) and 10 U ml- heparin prior to the preparation of leucocyte suspensions. Serum proteins were determined according to the method of Bradford (1976).

pure water, and analysed for retinol as described above. Results were expressed as mg retinol kg- 1 dry diet.

Cell mediated immune parameters Vitamin A analysis Livers A maximum of 0.5 g of liver was homogenised in 0.4 ml ethanolic pyrogallol (20 mg/ml of 95% ethanol) in 10 ml graduated stoppered test-tubes and heated in a 70°C water bath for 5 min. The tubes were removed, N 2 bubbled through, and 1 ml of 60% potassium hydroxide added. The tubes were returned to the water bath for a further 20 min with occasional shaking, after which time the saponified homogenates were cooled on ice. The volume was then made up to 10 ml with ultrapure water, and 4 ml hexane containing butylated hydroxytoluene (BHT) at 1.25 ptg ml-' was added. The tubes were shaken vigorously for 2 min and left in the dark to settle. One hundred g1 of the hexane extract was injected and vitamin A separated by HPLC on a 250 x 4.6 mm Alphasil C18 reverse phase column (HPLC Technology Ltd) with hexane (HPLC grade) containing 1% isopropanol as the mobile phase at a flow of 1 ml min- l. Detection of retinol was by a Perkin-Elmer fluorescence spectrophotometer at 325 nm coupled with a Shimadzu C-RIB integrator. Hexane extracts were kept in the dark at 40 C prior to their analysis on the day of extraction. During validation of the method, spiking with retinol confirmed that > 95% recovery was achieved during the extraction procedure. Results were expressed as g retinol g' liver. Diets Two g samples of diet (in triplicate) were ground up with pestle and mortar, and refluxed in the dark for 30 min at 70°C in 100 ml ethanolic pyrogallol containing 10 g of potassium hydroxide. The flasks were then cooled on ice and 200 ml of ultra-pure water added. Vitamin A was extracted into 50 ml hexane containing BHT at 1.25 gg ml- l . Five ml of hexane extract was washed twice with 5 ml of ultra-

Respiratory burst activity A leucocyte suspension was prepared from the kidney as described previously (Chung and Secombes 1988), layered onto 51%/34% Percoll density gradient and centrifuged at 400 x g for 30 min. The phagocyte enriched fraction at the interface was removed and washed twice in L 15 medium containing 10 U ml -- ' heparin and 2% FCS, and adjusted to 106 viable cells ml-'. Respiratory burst activity of phagocytes was quantified by measuring the reduction of ferricytochrome C following membrane stimulation with phorbol 12-myristate 13-acetate (PMA; Sigma) for 15 min using a protocol described previously (Thompson et al. 1993). Results were expressed as nmoles O2 produced per 105 kidney leucocytes in 15 min. Macrophage bactericidal activity A leucocyte suspension was prepared from the kidney and enriched for phagocytes as described above and adjusted to 107 viable cells ml- l in L15 medium containing 0.1% FCS. One hundred Rl aliquots of cells were added to 96 well microtiter plates and left to adhere for 2h at 18C. Non-adherent cells were then washed off with L 15 medium, and the remaining macrophages supplemented with L15 medium containing 5% FCS. After 48h the macrophage monolayers were washed with L15 medium. The capacity of these macrophages to kill an avirulent strain (A-layer negative: MT004) of A. salmonicida (the causative agent of furunculosis), was investigated according to the method of Graham et al. (1988), using a concentration of 2 x 106 bacteria/mi. Results were expressed as % bacteria killed. Lymphokine production A suspension of kidney leucocytes was layered onto 51% Percoll and centrifuged at 400 x g for 30 min to remove erythrocytes. Leucocytes were collected from the Percoll/medium interface, after which the

516 cells were washed as described above and adjusted to 5 x 10 6 /ml in L15 medium containing 5 x 10 - 5 M 2-mercaptoethanol (2ME). One ml aliquots were added to 24 well tissue culture plates, and production of macrophage activating factor (MAF) stimulated according to the protocol described by Graham and Secombes (1990). To test for MAF activity, 1:4 and 1:8 dilutions of the resulting supernatants were added to macrophage monolayers isolated from a single large stock salmon as described earlier. After 48h, the supernatants were washed off and a respiratory burst assay (reduction of ferricytochrome C) was performed to determine the degree of activation. Results were expressed as a stimulation index, by dividing the amount of O2produced by the activated macrophages by the amount produced by the negative control. Leucocyte migration A phagocyte enriched cell suspension was prepared as described earlier, and the migratory response to fresh salmon serum (diluted 1:50 in Hanks Balanced Salt Solution; HBSS) was investigated using a microchemotaxis chamber according to a protocol described previously (Thompson et al. 1993). Results were expressed as the number of cells migrating above background levels (i.e., to HBSS alone) per mm 2 of filter. Eicosanoid production Leucocyte suspensions were layered onto 51% Percoll and purified as described above. Cells were adjusted to 5 x 106 viable cells/ml in L15 medium and treated with 5 gM calcium ionophore (Sigma). The resultant supernatants were analysed for lipoxin (LXA 4 ) and leukotriene (LTB 4 and LTB 5) activity by HPLC according to the protocol described by Rowley (1992). Results were expressed as ng of product per 106 leucocytes. Humoral immune parameters Antiprotease activity Serum antiprotease activity was quantified using a modified assay based on that of Ellis (1990) utilising the ability of salmon antiproteases to inhibit trypsin activity. Test sera were serially diluted in

0.05 M tris buffered saline pH 8.0 (TBS) in a round bottom 96 well microtiter plate. Five gl aliquots were transferred to a flat bottomed 96 well microtiter plate and 15 1ll trypsin (100 gg ml-1 in TBS) was added with mixing at room temperature for 5 min. Two hundred ll of Na-benzoyl-dl-arginine pnitroanilide (BAPNA) substrate solution (10.4 mg BAPNA per 2 ml of dimethylformamide in 20 ml of 0.01 M tris/CaCl 2 buffer pH 7.4) was added to all wells and subsequent catabolism of substrate by uninhibited trypsin was determined during a 15 min kinetic run in a multiscan spectrophotometer (MDC Thermomax) at 450 nm against a TBS/BAPNA blank. Readings were converted to units trypsin/ml, and the volume of serum required to inhibit 85% of trypsin activity was determined for each sample. Results were expressed as trypsin equivalents/ml serum. Lysozyme activity Serum lysozyme activity was determined by measuring the rate of lysis of a suspension of Micrococcus lysodeikticus (75 mg 100 ml-' 0.1 M phosphate/citrate buffer containing 0.09% of NaCl, pH 5.8). One hundred and seventy five 1ll of this suspension was added to 25 1 of each serum sample, and to hen egg white lysozyme (Sigma) standards (0-10 pg/ml phosphate/citrate buffer). The plate was shaken and the rate of lysis determined against a M. lysodeikticus blank during a 5 min kinetic run at 450 nm on a multiscan spectrophotometer (MDC thermomax) at 25C. Serum lysozyme concentrations were calculated from the standard curve, and results expressed as g lysozyme ml-1 serum. Classical complement activity Complement mediated lysis of sheep erythrocytes (SRBC), sensitised with rainbow trout (Oncorhynchus mykiss) anti-SRBC serum, was determined as described previously (Hardie et al. 1991). The volume of serum producing 50% haemolysis (CH 50 value) was determined and the number of CH 50 units ml- 1 serum calculated. Serum bactericidal activity A suspension of A. salmonicida (MT004) containing 108 bacteria/ml was prepared in 3% TSB, and

517 75 l added to 25 l of serum in triplicate in a 96 well microtiter plate. Bacteria were also added to wells containing 25 1 of 3% TSB as controls. The plate was shaken and incubated at 18C for 3h, shaking hourly. The plate was then centrifuged at 150 x g for 10 min and the supernatants removed. MTT (5 mg ml- l H 2 0) was diluted 10 fold in 3% TSB and 100 gl were added to all wells. The plate was then incubated in the dark at 18 0C for 15 min and read in a multiscan spectrophotometer (MDC Thermomax) at 600 nm. Killed bacteria were quantified by substracting the O.D. measured for each serum sample from the mean control O.D. This difference, divided by the control O.D. x 100, gave the % bacteria killed. Specific antibody response Ten fish from each ration group were anaesthetised with benzocaine (25 plg/ml) and immunised with 0.5 ml formalin killed A. salmonicida (10 mg wet weight bacteria ml- 1 in 0.15 M phosphate buffered saline pH 7.4; PBS) by intraperitoneal injection. Ten further fish from each vitamin A status group were bled via the caudal vein at the time of immunisation to provide a basal antibody titre for each group. Seven immunised fish from each vitamin A status group were subsequently bled 7 weeks postimmunisation. Serum antibody titres were determined by ELISA. Briefly, plates were coated with 100 1I whole formalin killed A. salmonicida at 10 mg ml- l in carbonate buffer overnight at 4°C. Plates were washed three times with PBS containing 0.005% Tween-20 (PBS-Tween) and incubated with 100 l carbonate buffer containing 10 mg/ml of bovine serum albumin (BSA) for 2h at room temperature to block any non-specific binding. Plates were then washed three times in PBS-Tween, and 100 gl test sera added (serially diluted from 1:25 in PBS-Tween containing 0.5 mg ml- BSA) to each row of wells for 2h. Plates were washed three times in PBS-Tween, and 100 pl mouse monoclonal antitrout Ig/HRP conjugate (1:8000 in PBS-Tween/ BSA) was added to all wells for 2h. Finally, plates were washed three times in PBS-Tween, and 200 1 1,2-phenylenediamine dihydrochloride (Aldrich) substrate solution (0.4 mg/ml in 0.1 M citrate buffer, pH 5.0 containing 0.015% hydrogen perox-

ide) was added to all wells. After a 30 min incubation in the dark, O.D. readings were taken at 450 nm on a multiscan spectrophotometer (MDC Thermomax). Results were expressed as log 2 endpoint dilution (last dilution to have an O.D. of > 0.1, the highest value obtained using control serum on antigen coated wells or immunised serum on uncoated wells).

Disease resistance Twenty five fish from each dietary group were anaesthetized and marked with Alcian blue to allow identification of vitamin A status, and placed into a separate identical tank equipped with a sterilized outflow suitable for a bacterial challenge. The following day, the water supply to the challenge tank was shut off and a pathogenic strain of A. salmonicida(MT423) was added to give a final concentration of 105 cells/ml. The water supply remained shut off overnight, and strong aeration was provided during this period. The following day, the water supply was restored to the tank. Dead fish were removed at a fixed time each day for a 16 day period (by which time the mortality rates had declined to zero) and the numbers and vitamin A status were recorded. Kidney swabs were taken from the first few mortalities, inoculated onto tryptic soy agar plates and incubated for 48h at 18 0C to confirm the presence of A. salmonicida.

Statistical methods Intergroup differences were investigated by analysis of variance followed by Tukey's multiple range test where appropriate. Mortality rates measured during the disease challenge were compared using the log-rank test (Peto et al. 1977).

Results After four months, liver retinol levels reflected dietary vitamin A intake (Table 2). Significantly higher levels (p < 0.01) were found in the high

518 Table 2. Liver vitamin A levels, specific growth rates and serum protein concentrations in Atlantic salmon fed low, normal and high levels of vitamin A for 4 months Vitamin A ration group Parameter

Low

Normal

High

Liver retinol concentration (pxg g-1 tissue) n= 10 Specific growth rate (°7 body weight day- ') Serum protein concentration (mg m- l) n= 10

< 1

16.4 + 1.8

47.6

0.41

0.46

40.3 + 4.1

44.6

+ 3.0

0.42 4.1

46.76 + 3.0

Data are expressed as mean + SEM where appropriate. Table 3. Humoral immune parameters measured in Atlantic salmon fed low, normal and high levels of vitamin A for 4 months Vitamin A ration group Immune parameter

Low

Antiprotease activity (Trypsin equiv ml- l) n=8 Lysozyme (fig ml-1) n=8 Classical complement activity

32.03

(CH 50 ml- l) n=8 Bactericidal activity (%70 killing) n=8 Specific antibody level (Log 2 ELISA endpoint titre) 7 weeks post-immunization n=7

2.59a

1.98 + 0.32a

Normal

High

34.15 + 2.01ab

42.54 + 4.02b

2.23 + 0.64a

1.95 + 0.41a

62.31 + 3.la

69.20 + 1.2a

35.50 + 4.13a

56.87 ± 2.47b

50.38 ± 6.11 b

11.64 + 0.38a

11.07 + 0.37a

10.31 + 0.42a

67.3

+ 0.8a

Data are expressed as mean ± SEM; Means with common superscript are not significantly different (p > 0.05).

vitamin A intake group compared with those fish fed normal levels. No retinol was detectable in the livers of the low vitamin A intake group. No significant differences were observed in the growth rates or plasma protein levels of the groups fed the low, normal and high vitamin A diets (Table 2). All three groups displayed normal feeding behaviour throughout the experiment.

Humoral immune parameters Values recorded for these immune parameters are given in Table 3. Antiprotease activity was found to be significantly influenced by dietary vitamin A intake (p < 0.05), with a 25% increase in activity in the group fed high levels of vitamin A compared with those fed low levels. The difference in anti-

519 Table 4. Cell mediated immune responses in Atlantic salmon fed low, normal and high levels of vitamin A for 4 months Vitamin A ration group Immune parameter Respiratory burst (nmoles O0 per 105 leucocytes) n=8 Bactericidal activity (% killing) n=5 MAF activity (Stimulation index) A = 1:4 dilution B= 1:8 dilution n=5 Migration response (Cells per mm 2 filter) n=8 Eicosanoid production (ng per 106 leucocytes) A=LTB4 B = LTB 5 C=LXA 4 D = Total n=5

Low

Normal

6.30

1.03a

5.40

0.85a

31.40

2.54a

32.40

3.25a

2.08 1.48

0.80 a 0.26a

114.12

6 .98b

A= 1.99 + 0.34a B= 1.59 + 0.43a

84.62 + 7.12a

A= B=

High 6.68

0.89a

30.00 + 3.33a

A= B=

2.40 0.50a 2.11 + 0.37a

105.5 + 8.15ab

A= 13.6 + 3.9a

A=

7.4

0.9 a

A=

8.6

0.8 a

2.0 + 0.7 a

B=

0.9 + 0.l a

B=

1.2

0.la

C= 2.2 + 0.3a

C=

2.2 + 0.3 a

C=

1.7 + 0.3 a

D= 17.8

D= 10.5 + I.la

B=

4.3a

D= 11.5 + .l a

Data are expressed as mean + SEM; Means with common superscript are not significantly different (p > 0.05).

protease levels measured in the normal and high vitamin A intake groups was found to be marginally non-significant (p = 0.08). No significant differences were observed in either lysozyme or complement activity between the different groups. Similarly, levels of specific antibody seven weeks post-immunisation with A. salmonicidawere found to be unaffected by dietary vitamin A intake. However, serum bactericidal activity was found to be significantly affected by dietary vitamin A intake, with around 40% (p < 0.01) higher activity in the normal, and around 30%o (p < 0.05) higher activity in the high vitamin A intake group compared with the group fed low levels of vitamin A.

Cell mediated immune parameters Values recorded for these immune parameters are given in Table 4. Leucocyte respiratory burst activity was unaffected by dietary vitamin A intake, as were macrophage bactericidal activity and production of MAF, though the latter tended to be higher in the high vitamin A intake group compared with the low intake group. Kidney leucocyte migration was found to be significantly influenced (p < 0.05) by dietary vitamin A intake, being reduced by around 26% in the low vitamin A intake group compared with those fed normal levels. However, the high vitamin A intake group did not have an enhanced response. Production of leukotrienes (LTB 4 and LTB 5 ) and lipoxins (LXA 4) were not significantly affected by dietary vitamin A intake whether the data were analysed separately or

520 nificant difference in mortality rates between the different vitamin A intake groups.

100 90

80

Discussion

70 60 o

50 W

40

E )

30 20 10 0 0

2

4

6

8

10

12

14

16

Days Post Challenge Fig. 1. Morrtalities observed in Atlantic salmon fed low ( ), normal ( ), aind high ( o ) levels of vitamin A for 4 months and then challenged with a virulent strain of A. salmonicida (MT423).

pooled. However, there was a clear trend for increased Ievels of eicosanoids in the fish fed the low vitamin )Adiet compared with the other two groups.

Diseaserresistance Followin g challenge with A. salmonicida, the first mortalityi was recorded in the low vitamin A intake group onLday 5 (Fig. 1). This was 1 day earlier than the first lmortalities in the normal vitamin A intake group, a:nd 2 days earlier than in the high vitamin A intake group. Mortalities continued in all groups until day 9, after which only further low and high vitamin )Aintake fish died. The last mortalities were recorded on day 13 post challenge, when 96%, 76% and 80% of the fish had died in the low, normal and high vita min A intake groups respectively. By day 16, all surviving fish had recommenced feeding and appearedI healthy, at which point the experiment was termiinated. The log-rank test revealed no sig-

In the present study, vitamin A was found to have a significant effect on several aspects of both cell mediated and humoral immunity in Atlantic salmon. This was elucidated using fish varying in their in vivo levels of vitamin A, obtained by feeding three diets differing in vitamin A content for 4 months. Since teleosts are incapable of de novo synthesis of vitamin A, the entire physiological requirement must be provided in the diet. Vitamin A is stored principally in the liver, which is usually used to determine in vivo levels. Due to a paucity of previous research, it is difficult to ascertain whether the liver vitamin A concentrations obtained in the present study are typical for fish fed these particular levels, but they are in good agreement with those measured by Hilton (1983) once allowance for the contribution of the natural vitamin A content of the fish meal and fish oil contained in the diets used in that study (as estimated by our own analyses) was made. The specific growth rates and serum protein concentrations of the three groups in the present study were similar during the 4 month experimental period, despite the apparent absence of vitamin A in the liver of the low intake group, suggesting that gross protein intake and synthesis were unaffected. Furthermore, fish in the low vitamin A diet group fed normally throughout the experiment, indicating that no visual impairment had occurred as a result of vitamin A deficiency. Of the humoral immune parameters investigated, increased levels of dietary vitamin A resulted in significantly increased antiprotease activity and bactericidal activity. Antiproteases are essential for the neutralisation of bacterial extra-cellular proteases (ECP's) released during infection and the a 2-macroglobulin antiproteases of Atlantic salmon have been shown to play an important role in neutralising the ECP of A. salmonicida (Ellis 1987). An increase in antiprotease activity might therefore be expected to increase disease resistance in the high vitamin A intake fish. The lack of a difference

521 between the low and normal vitamin A intake groups suggests that deficiency does not compromise fish immunologically with respect to this parameter. Studies in the mammalian literature indicate a role for vitamin A in the synthesis of a,-macroglobulin (Kiorpes et al. 1976), though a 2-macroglobulin levels were found to be unaffected by vitamin A deficiency in another study (Bohannon et al. 1979). Bactericidal activity from serum of fish fed normal and high levels of vitamin A was significantly greater than that of fish fed low levels, which does suggest that deficiency compromises this activity. B, ctericidal activity of serum is mediated by lysozyme (Grinde 1989) and complement (Ourth and Bachinsky 1987), yet lysozyme and classical (antibody-mediated) complement activity were not significantly affected by vitamin A intake in the present study. However, the decreased serum bactericidal activity in the low vitamin A intake fish in the present study may be attributable to an effect on the alternative complement pathway which is activated directly by bacterial surfaces. Whilst potentially this might correlate with reduced disease resistance in the low vitamin A intake group, the issue is complicated by the fact that with A. salmonicidathe most resistant (A-layer positive) strains are resistant to attack by this pathway (Munn et al. 1982). The relationship between dietary vitamin A intake and specific humoral responses is also complex, and often apparently contradictory in other animal groups. In humans, for example, there is little effect of vitamin A status on production of specific antibody (reviewed by Dhur et al. 1991), yet in vitamin A deficient chickens suppressed primary and secondary responses to T-dependent antigens have been observed, suggesting a lesion in T-cell function (West et al. 1991). However, studies of the proliferative response of adult rats have shown B-cell but not T-cell mitogen responses to be reduced by vitamin A deficiency (Van Bennekum et al. 1991). Whilst no effect of vitamin A status was seen on antibody production in the present study, A. salmonicida is likely to be a T-independent antigen and so further research is clearly necessary to confirm that there is no interaction between vitamin A and antibody production in Atlantic salmon.

Cell mediated immune responses were relatively insensitive to dietary vitamin A in the present study. MAF production tended to be elevated by a high level of vitamin A intake, but phagocyte respiratory burst activity and bactericidal activity were unaffected by dietary vitamin A intake. The lack of an effect on the phagocyte respiratory burst may account for the lack of a significant effect of vitamin A intake on phagocyte bactericidal activity in the present study, since the respiratory burst is the principal mechanism by which macrophages kill the strain of A. salmonicidaused (Sharp and Secombes 1993). However, this result is in contrast to the reduced respiratory burst activity seen in vitamin A deficient chickens (West et al. 1991), and the reduced leucocyte bactericidal activity seen in mammals fed low levels of vitamin A (reviewed by Dhur et al. 1991). Vitamin A deficiency did result in a significant reduction in the ability of salmon leucocytes to migrate in vitro. The migration of phagocytes towards a site of infection is an essential part of the inflammatory response, and any reduction in the capacity for leucocyte migration might be expected to result in compromised disease resistance. Other important components in the inflammatory response are the phagocyte derived products known as eicosanoids (Secombes and Fletcher 1992). Eicosanoid production was not found to be affected significantly by dietary vitamin A intake, though levels of LTB 4 in particular tended to be higher in the low vitamin A intake group. The observation of generally higher levels of eicosanoid production in the low vitamin A intake group may be attributable to a reduction in synthesis of hydroxyeicosatetraenoic acids (HETE) which are intermediates in the production of some eicosanoids, through vitamin A mediated inhibition of the enzymatic oxidation of arachidonic acid (Halevy and Sklan 1987). However, since LTB 4 has been shown to be a potent chemoattractant for fish leucocytes (Hunt and Rowley 1986) it is difficult to predict the in vivo consequences of reduced migration plus increased LTB 4 production. The overall consequences of low and high dietary vitamin A intake on Atlantic salmon immunity are presumably reflected in their ability to resist infection. Analysis of mortality rates following chal-

522 lenge with virulent A. salmonicida showed no significant differences between groups. However, the mildly immunosuppressive effect of low vitamin A intake was apparent since fish in that group started to die earlier, and ultimately suffered higher mortalities than the other two groups. In a previous study involving salmonids, feeding increased levels of vitamin A was found to enhance disease resistance (Malikova et al. 1961), and there are numerous examples of decreased disease resistance to a variety of pathogens in both human and animal studies (reviewed by Dhur et al. 1990). It is possible that the challenge dose of bacteria used in the present study was slightly too high, since all but one of the low vitamin A intake group died. A lower dose may still have been sufficient to kill most of these fish whilst inducing less mortalities in the other two groups, thereby increasing the difference seen between the low and normal/high intake groups. However, on the basis of the existing results, there is no evidence to suggest that feeding a higher than normal level of vitamin A enhances disease resistance in vivo. In light of the results discussed above, it is important to consider reasons for the relatively small number of significant effects of dietary vitamin A intake on those aspects of immunocompetence investigated in the present study. Firstly, one must consider the duration of the study, which in the present case was 4 months. A longer period of vitamin deficiency or supplementation may have resulted in more significant effects. However, there are problems with extending periods of deficiency, since eventually food intake will be affected, at which point it becomes harder to differentiate between direct vitamin effects, and more general symptoms of malnutrition. Secondly, when considering vitamin A as a potential immunostimulatory factor, there might be some scope for using a higher level of dietary vitamin A supplementation up to a theoretical maximum of around 300 mg/kg diet (based on Hilton 1983). However, Hilton acknowledged that long term maintenance on such levels may be detrimental to fish health. Indeed, the addition of only a small vitamin A supplement to practical salmon diets resulted in significantly increased mortalities in the latter stages of a 28 week

diet trial (Grisdale-Helland et al. 1991). On the basis of the available information, the authors believe that the upper level of vitamin A supplementation chosen in the present study was appropriate. In conclusion, the present study has provided the first detailed report on the role of vitamin A in the immune response of Atlantic salmon. Fish deficient in vitamin A were shown to be immunocompromised compared with those receiving adequate amounts of this essential micronutrient. However, there was little evidence that a high intake of vitamin A was immunostimulatory compared with normal intake. In light of this, and in view of the restricted scope for using very high vitamin A diet supplements, the potential for using high vitamin A diets to enhance disease resistance of farmed Atlantic salmon appears to be limited.

Acknowledgements This study was funded by a MAFF contract (CSA1522), and a SERC studentship to Ian Thompson. Special thanks go to staff at SOAFD Marine Laboratory for helping to maintain the fish, to Mrs. A. White for formulating the test diets, and to Dr. A.F. Rowley (School of Biological Sciences, University of Swansea) for the eicosanoid analyses.

References cited Blazer, V.S. 1992. Nutrition and disease resistance in fish. Ann. Rev. Fish Diseases 2: 309-323. Bohannon, D.E., Kiorpes, T.C. and Wolf, G. 1979. The response of the acute-phase plasma protein a2 -macroglobulin to vitamin A deficiency in the rat. J. Nutr. 109: 1189-1194. Bradford, M.M. 1976. A rapid sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein dye binding. Anal. Biochem. 72: 248-254. Chung, S. and Secombes, C.J. 1988. Analysis of events occurring within teleost macrophages during the respiratory burst. Comp. Biochem. Physiol. 89B: 539-544. Dhur, A., Galan, P. and Hercberg, S. 1991. Vitamin A deficiency and immunity. J. Clin. Biochem. Nutr. 11: 1-19. Ellis, A.E. 1987. Inhibition of the Aeromonas salmonicida extracellular protease by a 2-macroglobulin in the serum of rainbow trout. Microbial Path. 3: 167-177.

523 Ellis, A.E. 1990. Serum antiproteases in fish. In Techniques in Fish Immunology. Vol 1, pp 95-99. Edited by J.S. Stolen, T.C. Fletcher, D.P. Anderson, B.S. Roberson and W.B. van Muiswinkel. SOS Publications, Fair Haven. Evans, D.L., Graves, S.S., Blazer, V.S., Dawe, D.L. and Graztek, J.B. 1984. Immunoregulation of fish non-specific cytotoxic cell activity by retinol acetate but not poly I:C. Comp. Immunol. Microbiol. Infect. Dis. 7: 91-100. Graham, S., Jeffries, A.H. and Secombes, C.J. 1988. A novel assay to detect macrophage bactericidal activity in fish: Factors influencing the killing of A eromonas salmonicida. J. Fish Dis. 11: 389-396. Graham, S. and Secombes, C.J. 1990. Cellular requirements for lymphokine secretion by rainbow trout Salmo gairdnerileucocytes. Dev. Comp. Immunol. 14: 59-68. Grinde, B. 1989. Lysozyme from rainbow trout Salmo gairdneri Richardson as an antibacterial agent against fish pathogens. J. Fish Dis. 12: 95-104. Grisdale-Helland, B., Helland, S.J. and Asgard, T. 1991. Problems associated with the present use of menadione sodium bisulphite in diets for Atlantic salmon. Aquaculture 92: 351-358. Halevy, O. and Sklan, D. 1987. Inhibition of arachidonic acid oxidation by 5-carotene, retinol and a-tocopherol. Biochem. Biophys. Acta 918: 304-307. Hardie, L.J., Fletcher, T.C. and Secombes, C.J. 1991. The effect of dietary vitamin C on the immune response of the Atlantic salmon (Salmo salar L.). Aquaculture 95: 201-214. Hilton, J.W. 1983. Hypervitaminosis A in rainbow trout (Salmo gairdneri): toxicity signs and maximum tolerable level. J. Nutr. 113: 1737-1745. Hunt, T.C. and Rowley, A.F. 1986. Leukotriene B4 induces enhanced migration of fish leucocytes in vitro. Immunology 59: 563-568. Kiorpes, T.C., Salvatore, J.M. and Wolf, G. 1976. A plasma glycoprotein depressed by vitamin A deficiency in the rat: al-macroglobulin. J. Nutr. 106: 1659-1667. Malikova, E.M., Apine, S.O. and Shaldaeva, R.E. 1961. Use of vitamins as prophylactic and therapeutic measures in immature Salmo salar in fish hatcheries. Tr. Nauchn. Inst. Rybn. Khoz. Latviisk. SSR. 3: 445-452. Munn, C.B., Ishiguro, E.E., Kay, W.W. and Trust, T.J. 1982. Role of surface components in serum resistance of Aeromo-

nas salmonicida. Infection and Immunity 36: 1069-1075. Ourth, D.D. and Bachinski, L.M. 1987. Bactericidal response of channel catfish Ictaluruspunctatus by the classical and alternative complement pathways against bacterial pathogens. J. Appl. Icthyol. 3: 42-45. Peto, R., Pike, M.C., Armitage, P., Breslow, N.E., Cox, D.R., Howard, S.V., Mantel, N., McPherson, K., Peto, J. and Smith, P.G. 1977. Design and analysis of randomised clinical trials requiring prolonged observation of each patient. Br. J. Cancer 35: 1-39. Ricker, W.E. 1979. Growth rates and models. In Fish Physiology. Vol. 8, pp. 677-743. Edited by W.S. Hoar, P.J. Randall and J.R. Brett. Academic Press, London. Rowley, A.F. 1992. Identification of eicosanoids in fish tissues. In Techniques in Fish Immunology. Vol. 2, pp. 143-156. Edited by J.S. Stolen, T.C. Fletcher, D.P. Anderson, S.L. Kaattari and A.F. Rowley. SOS Publications, Fair Haven. Secombes, C.J. and Fletcher, T.C. 1992. The role of phagocytes in the protective mechanisms of fish. Ann. Rev. Fish Dis. 2: 53-71. Sharp, G.J.E. and Secombes, C.J. 1993. The role of reactive oxygen species in the killing of the bacterial fish pathogen Aeromonas salmonicidaby rainbow trout macrophages. Fish and Shellfish Immunol. 3: 119-129. Schiedt, K., Leuenberger, F.J., Vecchi, M. and Glinz, E. 1985. Absorption, retention and metabolic transformations of carotenoids in rainbow trout, salmon and chicken. Pure and Appl. Chem. 57: 685-692. Thompson, I., White, A., Fletcher, T.C., Houlihan, D.F. and Secombes, C.J. 1993. The effect of stress on the immune response of the Atlantic salmon (Salmo salarL.) fed diets containing different amounts of vitamin C. Aquaculture 114: 1-18. Torrissen, O.J. 1989. Pigmentation of salmonids: interactions of astaxanthin and canthaxanthin on pigment deposition in rainbow trout. Aquaculture 79: 363-374. Van Bennekum, A.M., Kong, N.R.W.Y., Gijbels, M.J.J., Tielen, F.J., Roholl, P.J.M., Brouwer, A. and Hendriks, H.F.J. 1991. Mitogen response of B cells, but not T cells, is impaired in adult vitamin A deficient rats. J. Nutr. 121: 1960-1968. West, C.E., Rombout, J.H.W.M., Van der Zijpp, A.J. and Sijtsma, S.R. 1991. Vitamin A and immune function. Proc. Nutr. Soc. 50: 251-262.