JOURNAL OF EXPERIMENTAL ZOOLOGY 287:199–212 (2000)

© 2000 WILEY-LISS, INC.

JEZ 0788

Cloning and Characterization of Salmon hsp90cDNA: Upregulation by Thermal andHyperosmotic Stress

FENG PAN,2 JACQUES M. ZARATE,1 GEORGE C. TREMBLAY,2 ANDTERENCE M. BRADLEY1*1Department of Fisheries, Animal and Veterinary Science, University ofRhode Island, Kingston, Rhode Island 02881

2Department of Biochemistry, Microbiology, and Molecular Genetics,University of Rhode Island, Kingston, Rhode Island 02881

ABSTRACT Accumulating evidence suggests that glucocorticoids are essential for developmentof hypoosmoregulatory capacity in salmon during adaptation to seawater. Heat shock protein (hsp)90has been reported to function in signal transduction and the maturation and affinity of glucocorti-coid receptors. We sought to determine whether this hsp might be upregulated by thermal andhyperosmotic stress in salmon, a species that migrates between the freshwater and marine envi-ronments. A 2625-bp cDNA cloned from a salmon cDNA library was found to code for a protein of722 amino acids exhibiting a high degree of identity with zebra fish (92%) and human (89%)hsp90β. Accumulation of hsp90 mRNA was observed in isolated branchial lamellae incubatedunder hyperosmotic conditions and in branchial lamellae of salmon exposed to hyperosmotic stressin vivo. In contrast, exposure of kidney to hyperosmotic stress in vitro and in vivo failed to elicitan increase in the quantity of hsp90 mRNA. By way of comparison, accumulation of hsp90 mRNAwas observed in both branchial lamellae and kidney tissue subjected to thermal stress in vitroand in vivo. Western blot analyses of proteins isolated from tissues under identical conditions invitro revealed that the pool of hsp90 increased with thermal stress but not with osmotic stress.The results suggest that accumulation of hsp90 mRNA in response to osmotic stress is unrelatedto cellular protein denaturation and that synthesis of hsp90 may be regulated at both the level oftranscription and translation. J. Exp. Zool. 287:199–212, 2000. © 2000 Wiley-Liss, Inc.

One of the major classes of heat shock proteins(hsps) induced in cells in response to thermalstress is the hsp90 family. Hsp90 has been re-ported in all organisms examined and in eukary-otes may constitute 1–2% of total cytosolic proteins(Parsell and Lindquist, ’93). Accumulating evi-dence suggests that hsp90 interacts with specifictarget proteins (e.g., kinases) and plays an essen-tial role in steroid–receptor fidelity (Parsell andLindquist, ’93; Craig et al., ’94; Nathan et al., ’97).Unlike hsp70, a nonspecific chaperonin that is in-duced by a constellation of compounds, the hsp90response to chemical stressors is limited (Nover,’91; Ali et al., ’96).

Juvenile salmon in the wild are subjected to hy-perosmotic stress during migration to the sea inthe spring of the year following parr–smolt trans-formation, a developmental process encompassingnumerous physiological and biochemical changesthat prepare the fish for life in the marine en-vironment (Hoar, ’88). In commercial salmon

aquaculture, exposure to hyperosmotic stress isexacerbated by the routine practice of directtransfer of juveniles from hatcheries in fresh-water (typically < 100 mOsm) to netpens in fullsalinity seawater (1100 mOsm). Plasma concen-trations of chloride can increase from 110 to ≥200mMol/liter by 12 hr after transfer (Handeland etal., ’96). Salmon incapable of regaining osmotichomeostasis die or grow at a reduced rate (Bjorns-son et al., ’88; Duston, ’94). Previous investiga-tions in our laboratory demonstrated that hsp70is induced up to tenfold in tissues of juvenile At-lantic salmon exposed to hyperosmotic stress

Grant sponsor: United States Department of Agriculture NationalResearch Initiative Competitive Grants Program; Grant number:9404491; Grant sponsor: Rhode Island Agricultural Experiment Station.

*Correspondence to: Terence M. Bradley, Dept. of FAVS, U.R.I.,Building 14, East Farm, Kingston, RI 02881. E-mail: tbradley@uri.edu

Received 16 April 1999; Accepted 7 March 2000

200 F. PAN ET AL.

(Smith et al., ’99b). Furthermore, induction ofhsp70 by mild thermal shock of the living animalconferred protection against subsequent osmoticchallenge (DuBeau et al., ’98). Hsp70 appears tofunction transiently to renature or reduce dena-turation of proteins by hyperosmotic conditions un-til long-term hypoosmoregulatory mechanismsdevelop.

In the present report, we describe the cloningand sequencing of the cDNA coding for hsp90 ofAtlantic salmon (Salmo salar) and characterizethe response of this gene in branchial lamellaeand kidney, the two primary osmoregulatory or-gans of teleost fishes, during exposure to thermaland hyperosmotic stress. The hypothesis that hsp90is involved in adaptation of salmon to hyperosmoticconditions was prompted by evidence demonstrat-ing the role of this chaperonin in glucocorticoid re-ceptor maturation and fidelity (Vamvakopoulos, ’93;Bohen, ’95) and in signal transduction (Pratt, ’97,’98). The glucocorticoid, cortisol, is essential for ad-aptation of salmon to seawater (McCormick, ’95).Indeed, previous investigations have reportedprominent increases in circulating levels of cortisolin salmon at the time of parr–smolt transforma-tion (Hoar, ’88) and demonstrated that administra-tion of exogenous steroid stimulates differentiationof chloride cells and synthesis of the antiport pump,Na+/K+ ATPase, in branchial lamellae (McCormickand Bern, ’89; Madsen, ’90; McCormick, ’95).

MATERIALS AND METHODSAnimals

Experiments were conducted with tissues fromjuvenile Atlantic salmon (Salmo salar) reared at theUniversity of Rhode Island Aquaculture Center in2-m-diameter fiberglass tanks provided with supple-mental aeration and receiving single-pass water atambient temperature (6–18°C). Fluorescent lightssuspended approximately 1 m above the surface ofthe water provided a simulated natural photope-riod, which was adjusted weekly. Fish were fed acommercial formulated feed to satiation (NelsonSterling Silver Cup, Murray, UT) five times daily.

Synthesis of salmon hsp90 and actincDNA using PCR

A 0.3 kilobase (kb) fragment of hsp90 was syn-thesized by polymerase chain reaction (PCR) foruse in screening a cDNA library and Northern blotanalyses. Total RNA was isolated from Atlanticsalmon branchial lamellae incubated for 12 hr inculture medium supplemented with 125 mM NaCland maintained at ambient water temperature

(see In vitro experiments for details). mRNA wasreverse transcribed to cDNA using oligo dT18 prim-ers and Superscript II reverse transcriptase (LifeTechnologies, Rockville, MD). The target cDNAwas amplified by PCR using the following degen-erate primers with a NotI restriction site incor-porated at the 5′ end: 5′CGGGGCGGCCGCTA(C/T)TCNAA(C/T)AA(A/G)GA(A/G)AT(A/C/T)3′ and5′GCCCGCGGCCGCTA(A/G)AANCCNACNCC(A/G)AA(C/T)TG3′ where N = A,C,G,T (Krone andSass, ’94). A 100-µl reaction mixture containing 1µl of the RT reaction, 50 pmols of each primer,0.3 mM dNTPs, 2.5 mM magnesium chloride, and5 U Taq DNA polymerase (Promega, Madison, WI)was subjected to the following thermal cycle regi-men: 2 cycles of 94°C for 1 min, 50°C for 1 minand 72°C for 2 min followed by 30 cycles of 94°Cfor 1 min, 55°C for 1 min, and 72°C for 2 minwith a final elongation step at 72°C for 5 min. A0.5-kb DNA fragment of salmon actin also wasgenerated by PCR using the primers 5′GAGAA-GATGACCCAGATTATG3′ and 5′GTTGTATGTG-CTCTCGTGGAT3′ (a generous gift from Dr. MartaGomez-Chiarri, URI) and the reaction mixture de-scribed above for hsp90. The target DNA was am-plified using 30 cycles of 94°C for 1 min, 46°C for1 min, and 72°C for 2 min followed by a final elon-gation step of 72°C for 5 min. PCR products wereanalyzed by electrophoresis in a gel of 1.5% aga-rose in TAE buffer (40 mM Tris-acetate, pH 8.0, 2mM EDTA) and 0.1% ethidium bromide. Singleproduct bands of 0.3 kb and 0.5 kb for hsp90 andactin, respectively, were isolated from the gel byfiltration through a spin column (Supelco, Belle-fonte, PA).

Cloning hsp90 cDNAA cDNA library was constructed in a λ Uni-ZAP

XR vector (Stratagene, La Jolla, CA) using DNAprepared from salmon branchial lamellae as de-scribed above. Approximately 10 µg poly (A)+ RNAwas isolated from the total pool of RNA using mag-netic separation (Promega, Madison, WI) and uti-lized as template for first and second strand cDNAsynthesis. The double stranded cDNAs were frac-tionated by passage through a column comprisedof Sepharose CL-2B gel and cDNAs ranging insize from 1.2 to 6.0 kb were ligated into the vec-tor and packaged in phage as per the manu-facturer’s recommendations. The 0.3-kb ampliconof hsp90 was radiolabeled with α-32P dATP (ICN,Costa Mesa, CA) using DNA polymerase I (Klenowfragment; New England Biolabs, Beverly, MA) andrandom dodecamer primers (Ambion, Austin, TX)

CLONING AND CHARACTERIZATION OF SALMON hsp90 cDNA 201

and used to screen the cDNA library. Followingtwo rounds of purification, nine plaques were se-lected and transformed into a SolR strain of E.coli to generate the double stranded phagemidpBluescript by in vivo excision. Cells were inocu-lated onto Larin Bertani agar plates containing50 µg/ml carbenicillin, and individual colonieswere screened for the presence of an hsp90 insertusing PCR with the primers described above andthe following thermal cycle: 96°C for 10 min, 30cycles of 96°C for 20 sec, 37°C for 1 min, 72°C for2 min, followed by a final step at 72°C for 10 min.The size of selected inserts was determined by PCRusing primers to the T3 and T7 promoters flank-ing the multiple cloning site followed by electro-phoresis of the amplicons in a gel of 1.0% agaroseas described above. Plasmid DNA was isolated(Qiagen, Chatsworth, CA) from a clone containinga 2.6-kb insert and sequenced in both directionsusing a primer walking strategy with Taq FS DNApolymerase and fluorescent-dideoxy terminators(W.M. Keck Foundation Biotechnology ResourceLaboratory, Yale University, New Haven, CT). Iden-tity of the contiguous hsp90 cDNA with nucleic acidand deduced amino acid sequences was determinedusing BLASTN (Altschul et al., ’97).

In vitro experimentsThermal stress

Fish were killed by a sharp blow to the headand exsanguinated by excision of the caudal finto reduce contamination of tissues with erythro-cytes. Three to four hemibranchs of the gills wererapidly excised, and the posterior kidney was re-moved through a ventral incision. Hemibranchswere cut just above the septa to separate thelamellae and the posterior kidney was diced intosmall cubes (3–5 mm) using methanol-cleaned ra-zor blades. Approximately 0.8 g of lamellae or kid-ney from each individual was rinsed in 30 ml ofminimum essential medium with Earle’s salts(MEM; Sigma, St. Louis, MO) and 0.4 g of eachtissue transferred to a 25 ml Erlenmeyer flaskcontaining 8 ml MEM supplemented with 100 U/ml penicillin and 100 µg/ml streptomycin and ad-justed to pH 7.3 with a 7.5% solution of sodiumbicarbonate. Flasks containing the tissues weregassed with O2:CO2 (99:1), capped with a rubberseptum and placed on an orbital shaker (80–100rpm) and incubated at either ambient water tem-perature (10°C) or 26°C for 3 hr. Culture mediumbathing the tissues was overlaid with O2:CO2 toreduce the potential for anoxia; molecular oxygendoes not stimulate induction of the major hsps

(Visner et al., ’96). Upon termination of the expo-sure period, RNA and protein were isolated fromtissues as described below. The thermal regimenemployed in these investigations has been dem-onstrated to induce the major hsps of salmon(Smith et al., ’99a).

Osmotic stressTo investigate whether the hsp90 gene is up-

regulated in response to hyperosmotic stress, tis-sues were prepared and treated as described forthermal stress above with the following exceptions:1) tissues were maintained at ambient water tem-perature throughout, 2) the MEM was supple-mented with 125 mM NaCl to create a hyperosmoticcondition, and 3) tissues were incubated for 12 hr.Previous investigations in our laboratory demon-strated that hsp70 and a 54-kDa protein (osmoticshock protein 54, osp54) are induced in isolatedbranchial lamellae incubated using these condi-tions (Smith et al., ’99b).

In vivo experimentsThermal stress

Thirty juvenile salmon (16–18 cm in length)reared at 10°C were transferred directly to a tankcontaining 190 liters of water maintained at 26°Cand provided with supplemental aeration. After30 min, the fish were transferred to a 250 litertank supplied with single-pass ambient tempera-ture water. Branchial lamellae and posterior kid-ney were excised from fish immediately followingcessation of thermal shock and after 3, 6, and 9hr recovery. Control fish were handled in an iden-tical manner but maintained at ambient tempera-ture throughout. RNA was isolated and assayedby Northern blot analysis as described below.

Osmotic stressTo reduce the confounding effects of biochemi-

cal and endocrine changes associated with smolti-fication, experiments to investigate the responseof hsp90 to hyperosmotic stress in vivo were con-ducted approximately 2 months after parr–smolttransformation as assessed by a decline in bran-chial Na+/K+ ATPase activity and increase in con-dition factor (Robertson and Bradley, ’91; data notshown). Fifty juvenile salmon (16–18 cm) weretransferred directly from freshwater to tanks con-taining filtered Narragansett Bay seawater (32parts per thousand salinity). The 190-liter tankswere equipped with charcoal filters and aerationand surrounded with a jacket of flowing freshwa-

202 F. PAN ET AL.

ter to maintain ambient water temperature. Anadditional group of 10 fish was retained underidentical conditions in freshwater to serve as acontrol. Ten fish were randomly sampled 6, 12,24, 48, and 96 hr post transfer and killed by asharp blow to the head. Approximately 1 ml bloodwas collected into a heparinized syringe by cau-dal venipuncture and a 200 µl aliquot of blood wascentrifuged at 5000g for 5 min for determinationof hematocrit. The remaining blood was centri-fuged at 8000g for 5 min and the plasma collectedand frozen at –20°C. Chloride concentration of theplasma was determined by colorimetric assay ofthiocyanate displacement from mercuric thiocyan-ate (Sigma procedure no. 461-M; Cl– standards 955-11). Branchial lamellae and kidney tissue werecollected from four control fish and four individualssampled at 24 and 96 hr post transfer and processedfor isolation of RNA as described below.

RNA isolation and Northern blot analysisTotal RNA was isolated from branchial lamel-

lae and kidney tissue using a guanidinium-isothiocyanate method (Chirgwin et al., ’79). TheRNA concentration of each sample was quantifiedspectrophotometrically (A260) and those with an A260/A280 ratio > 1.6 and exhibiting no evidence of deg-radation, as indicated by clearly defined 28s and18s rRNA bands on ethidium bromide–stainedagarose gels, were used for subsequent analyses.Aliquots of 30 µg total RNA were resolved on athree-(n-morpholino) propanesulfonic acid (MOPS)-formaldehyde 1% agarose gel (Sambrook et al., ’89)and transferred to a charged nylon membrane(Magnacharge, MSI, Westboro, MA) over 12 hr us-ing a turboblotter (Schleicher & Schuell, Keene,NH). RNA was linked to the membrane by bakingat 80°C for 1 hr.

The 0.3-kb hsp90 amplicon and a 2.6-kb cDNAclone that hybridized with this fragment wereused to synthesize probes for Northern blot analy-ses and for screening the library. The cDNAs werelabeled with α-32P dATP using DNA polymerase Iand random dodecamer or T3/T7 primers, respec-tively (Sambrook et al., ’89). Membranes contain-ing RNA were incubated for 2 hr at 42°C inprehybridization buffer (5× SSC, 5× Denharts,0.5% SDS, 100 µg/ml salmon sperm DNA) as pre-viously described (Smith et al., ’99b). Hybridiza-tion was conducted in fresh buffer (5× SSC, 5×Denharts, 50% formamide, 10% dextran sulfate,0.2% sodium dodecyl sulfate (SDS), 100 µg/mlsalmon sperm DNA) containing radiolabeled probe(≥1.5 × 106 cpm/ml) for 18–20 hr at 42°C. Strin-

gency washes were conducted in 2× SSC contain-ing 0.5% SDS at 65°C for 30 min followed by 2×SSC lacking SDS at room temperature for 5 min.Blots probed with hsp90 cDNA were stripped in asolution of 50% formamide and 3× SSC at 65°Cfor 30 min and reprobed with the 0.5-kb cDNAfragment of salmon actin to allow for determina-tion of the relative quantity of hsp90. Radiolabel-ing and hybridization conditions for the actinprobe were as described above for hsp90.

Western blottingUpon termination of exposure of isolated bran-

chial lamellae and kidney tissues to thermal orosmotic stress, flasks containing tissues werechilled on ice for 2 min and the medium was as-pirated. Tissues were rinsed with 5 volumes ofice-cold 0.1 M TRIS (pH 7.6) containing a cocktailof protease inhibitors (AEBSF, pepstatin A, E-64,bestatin, leupeptin, and aprotinin; Sigma, No. P8340) and homogenized in 5 volumes of freshbuffer with three 10-sec pulses of a tissuemizer(Tekmar, Cincinnati, OH). Homogenates were cen-trifuged at 1000g for 10 min at 4°C to removeunlysed cells and debris. Aliquots of the resultingsupernatant (clarified homogenate) were frozenand stored at –70°C for up to 2 weeks prior toanalysis of proteins.

Fifty micrograms of protein from each samplewere resolved on SDS polyacrylamide gels (SDS-PAG) with a total acrylamide concentration of 8%and transferred to polyvinylidene difluoride mem-branes (Hybond, Amersham, Piscataway, NJ) us-ing a semidry transfer apparatus (AmericanBionetics model SBD-1000, Hayward, CA). Mem-brane-bound proteins were probed with a mono-clonal antibody to hsp90 purified from rat (StressGen Biotechnologies, Victoria, BC, Canada; SPA-835). As with Northern blot analyses, actin wasemployed as an internal control for Western blotsand all membranes were probed with a mousemonoclonal antibody to actin (Sigma, A4700). Themembranes were blocked with 5% nonfat powderedmilk in Tris-buffered saline supplemented withTween 20 (TBS-T; 20 mM Tris-HCl, 137 mM NaCl,0.1% (v/v) Tween 20) for 1 hr at room tempera-ture, rinsed briefly in TBS-T and incubated withprimary antibody diluted in TBS-T (1:5000 hsp90;1:500 actin), for 1 hr at room temperature on anorbital shaker. Membranes were briefly washedthree times with fresh TBS-T, then incubated for1 hr at room temperature with alkaline phos-phatase conjugated, goat anti-rat IgG antibody(StressGen; SAB-201; 1:30,000) for hsp90 or anti-

CLONING AND CHARACTERIZATION OF SALMON hsp90 cDNA 203

mouse IgG antibody (Amersham; RPN5781;1:10,000) for actin. Following incubation, mem-branes were washed once more and incubated inenhanced chemifluorescent (ECF) substrate (24 µl/cm2 membrane; Amersham) on a sheet of plasticwrap for 5 min. Immunoreactive proteins were vi-sualized by scanning the membrane using a Mo-lecular Dynamics Storm 840 Unit (Sunnyvale, CA)in fluorescence mode and set to 520 nm with aphotomultiplier tube voltage of 450. Negative an-tigen and antibody controls were used to gaugethe specificity of antibody binding. Supporting evi-dence for the identity of hsp90 and actin was ob-tained by comparison with molecular weightstandards co-electrophoresed on each gel.

Quantification and analysesDigital images of Northern blots were captured

using a Molecular Dynamics Storm 840 phosphorimager and Imagequant software. The size of thetranscripts was determined by comparison withRNA markers co-electrophoresed on each gel andthe integrated optical density (IOD = optical den-sity × area) of hsp90, and actin bands was quan-tified using this instrument. For Western blotanalyses, the IOD of hsp90 and actin bands wasquantified from digital fluorescent images usingImagequant software. The relative concentrationsof hsp90 mRNA and protein were determined bynormalizing the IOD of hsp90 to actin (hsp90:actin). The use of actin provides an effective in-ternal control for comparison of the effects of ther-mal and osmotic stress on expression of a stableconstitutive gene (actin) and an inducible genesuch as hsp90 (Wei and Roepe, ’94).

Values for hematocrit and plasma Cl– were ana-lyzed statistically by repeated measures ANOVA (α= 0.05), followed by a Student’s t-test to identify sig-nificant differences between treatments. Relativehsp90 mRNA and protein concentrations were com-pared using a Student’s t-test. Following statisticalanalysis, the hsp90 values were converted to per-centages and are presented as a percentage of thecontrol value, which was set arbitrarily at 100%.

RESULTSHsp90 cDNA

A 2625-bp clone from the salmon cDNA librarythat hybridized with a 3.1-kb transcript on North-ern blots was sequenced and found to contain anopen reading frame of 2169 nt coding for a pro-tein of 722 amino acids (GenBank accession no.AF135117; Fig. 1). The coding region extended

from nucleotides 86 through 2254, and the PCRamplicon spanned the region of nucleotides 182through 488 and amino acids 32 through 134. Thesalmon cDNA exhibited extensive identity at thededuced amino acid level with sequences of hsp90βof zebra fish (92% identity) and human (89% iden-tity; Fig. 1). Further analysis of the deduced aminoacid sequence using the ExPASy Molecular Biol-ogy Server (www.expasy.ch; Swiss Institute ofBioinformatics) suggests that the protein has amolecular weight of 83.3 kDa and a theoretical pIof 4.89. Dipeptide composition (Guruprasad et al.,’90) and the N-terminal residue (Gonda et al., ’89)suggest the protein to be relatively unstable, witha projected half-life of circa 30 hr.

Thermal stress

To determine the magnitude of hsp90 upregu-lation in response to thermal stress and to allowfor comparison with the response elicited by os-motic stress, isolated branchial lamellae and kid-ney tissue were incubated at 26°C (∆T = 16°C)for 3 hr and assayed for hsp90 mRNA. This ther-mal regimen has been demonstrated to stimulateinduction of the major species of hsps of salmon(Smith et al., ’99a). Northern blot analysis re-vealed a 132% increase in accumulation of hsp90mRNA in branchial lamellae (Fig. 2A,B) and a117% increase in kidney tissue (Fig. 2C,D) exposedto thermal stress. Corresponding with this rise inhsp90 mRNA, was a similar increase in the poolof hsp90 in branchial lamellae (155%, Fig. 3A,B)and kidney tissue (155%; Fig. 3C,D).

The response observed in isolated tissues wascompared with that of the living animal exposedto the same thermal stress of 26°C but of shorterduration (30 min). The quantity of hsp90 mRNAin branchial lamellae did not increase during the30 min thermal shock, but a marginal and statisti-cally significant increase (18%) in hsp90 mRNA wasobserved following 6 hr recovery from thermal shock(Fig. 4A,C). By hour 9 of recovery, the level of hsp90mRNA was no longer statistically different fromthat of control fish. Unlike branchial lamellae, amarginal, but statistically significant increase(18–23%) in hsp90 mRNA in kidney tissue (Fig.4B,D) was observed during thermal stress. Thequantity of hsp90 mRNA remained elevated forthe duration of the 9-hr recovery period.

Osmotic stressWe were most interested in determining whether

hsp90 is upregulated in response to hyperosmotic

204 F. PAN ET AL.

Fig. 1. Comparison of the deduced amino acid se-quences of hsp90β of Atlantic salmon (Ats), zebrafish (Zeb),

and human (Hum). * indicates identity with Atlanticsalmon sequence.

CLONING AND CHARACTERIZATION OF SALMON hsp90 cDNA 205

stress. To investigate the response in vitro, isolatedbranchial lamellae and posterior kidney, the pri-mary osmoregulatory organs of teleost fishes, wereincubated in MEM supplemented with 125 mMNaCl for 12 hr. Northern blot analysis revealed a133% increase in hsp90 mRNA in branchial lamel-lae (Fig. 5A,B) but the levels of hsp90 mRNA inthe kidney were unaffected by this osmotic stressregimen (Fig. 5C,D). Despite an increase in hsp90mRNA, no increase in the quantity of hsp90 in ei-ther tissue was detected by Western blot analysis(data not shown).

Finally, we sought to determine whether os-motically induced accumulation of hsp90 mRNAmight be observed in the living animal. Juvenilesalmon were transferred directly from freshwa-ter to seawater, and blood, branchial lamellae,and kidney tissue were collected from individu-als at designated intervals for assay of hemat-ocrit, plasma Cl–, and hsp90 mRNA. Transferfrom freshwater to full-salinity seawater resulted

in a rise in the concentration of plasma chloridefrom 115 mM/liter to 155 mM/liter by hour 6,and the levels remained elevated until 96 hrwhen the study was terminated (Fig. 6A). Con-versely, hematocrit levels declined from 48 to 29%during the first 6 hr exposure to seawater andremained depressed for the duration of the ex-periment (96 hr; Fig. 6B). Similar to the resultsobtained with isolated tissues, hsp90 mRNA wasupregulated (73%) by osmotic stress in vivo inbranchial lamellae 24 hr following transfer toseawater, but was unaltered in kidney tissue(Fig. 7A–D). After 96 hr in seawater the quan-tity of hsp90 mRNA in branchial lamellae de-clined to freshwater levels, as did plasma Cl–

indicating successful adaptation to seawater.

DISCUSSIONThe nucleotide sequence of the Atlantic salmon

2625 bp cDNA contained an open reading frameof 2169 nt coding for a protein of 722 amino acids

Fig. 2. Northern blot analyses of hsp90 mRNA in (A) bran-chial lamellae and (C) posterior kidney tissue isolated fromthree individual salmon and incubated at 10°C (ambient) or26° for 3 hr. Values in the histograms of (B) branchial lamel-

lae and (D) kidney are presented as the mean ± SD (n = 3fish per temperature). Values in the same figure marked with* are statistically different (P < 0.05).

206 F. PAN ET AL.

consistent with hsp90β of zebrafish (724 aminoacids, 92% identity; GenBank no. AF068772) andhuman (724 amino acids, 89% identity; Rebbe etal., ’87) suggesting that this cDNA does indeedcode for hsp90β of Atlantic salmon, The Atlanticsalmon hsp90β also exhibited 82% identity withthe deduced amino acid sequence of hsp90α ofchinook salmon (Oncorhynchus tshawytscha;GenBank no. U89945). Most higher vertebratespossess two closely related hsp 90 genes (α andβ) (Nover, ’91; Krone and Sass, ’94).

Expression of hsp90 was upregulated in salmontissues exposed to thermal shock in vitro and invivo. The quantity of hsp90 mRNA and protein ingill and kidney tissues following thermal shockin vitro increased more than 100% above the lev-els in tissues maintained at ambient temperature.In the living animal, hsp90 mRNA also increasedduring exposure to thermal stress, but to a lesserdegree (18–23%) than observed in isolated tissues.

The more prominent response in vitro could berelated to the degree of protein denaturation andthe specific proteins affected by the extended ther-mal shock (3 hr vs. 30 min). A recent report sug-gests that hsp90, in contrast to hsp70, maintainsa restricted role in thermal stress and interactsonly with specific classes of proteins that fail toreadily attain native conformation (Nathan et al.,’97). The relatively high constitutive levels ofhsp90 observed in salmon tissues might be suffi-cient to contend with all but the most extremethermal stress. Krone and Sass (’94) reported asimilar increase (100–200%) in hsp90β mRNAfollowing heat shock of early stage zebrafish em-bryos. The magnitude of response declined dra-matically with age and increased constitutivelevels of hsp90β.

In salmon, the response of hsp90 to thermalshock was less prominent than that of hsp70. Thequantity of hsp70 mRNA increased by ≥300% in

Fig. 3. Western blot analyses of hsp90 in (A) branchiallamellae and (C) posterior kidney tissue isolated from threeindividual salmon and incubated at 10°C or 26°C for 3 hr.Values in the histograms of (B) branchial lamellae and (D)

kidney are presented as the mean ± SD (n = 3 fish per tem-perature). Values in the same figure marked with * are sta-tistically different (P < 0.05).

CLONING AND CHARACTERIZATION OF SALMON hsp90 cDNA 207

branchial lamellae of juvenile salmon exposed invivo to thermal shock conditions (26°C for 15 min)similar to those employed in these investigations(Zarate and Bradley, unpublished data). These re-sults agree with a previous report demonstrating

that the accumulation of hsp90β mRNA in Xeno-pus laevis exposed to thermal shock was muchless pronounced than the increase in hsp70 mRNA(Ali et al., ’96).

Of greater interest, hsp90 was upregulated in

Fig. 4. Northern blot analyses of RNA isolated from (A)branchial lamellae and (B) kidney tissues of salmon subjectedto a 30 min thermal shock at 26°C (∆T = 16°C) and sampledeither immediately after shock (0) or following 3, 6, or 9 hrrecovery in ambient temperature water. Control fish (Con)

were subjected to the same experimental procedures at am-bient temperature. Values in the graphs of (C) gill and (D)kidney are presented as the mean ± SD (n = 2 fish per timepoint). Values in the same figure marked with * are statisti-cally different from the control (P < 0.05).

208 F. PAN ET AL.

Fig. 5. Northern blot analyses of hsp90 mRNA in salmon(A) branchial lamellae and (C) kidney tissue incubated for12 hr in tissue culture medium (0) or medium supplementedwith 125 mM NaCl (125). Values in the graphs of (B) bran-chial lamellae and (D) kidney are presented as the mean ±SD (n = 3 fish per time point). Values in the same figuremarked with * are statistically different (P < 0.05).

Fig. 6. Plasma Cl– concentrations and hematocrit (Hct)of salmon maintained in freshwater (0) or 6, 12, 24, 48, and96 hr following transfer to seawater (32 ppt). Values are pre-sented as the mean ± SD (n = 10 fish per time point). Timepoints marked with different superscripts are statistically dif-ferent (P < 0.05).

Atlantic salmon branchial lamellae in response tohyperosmotic stress in vitro and in vivo. Up-regulation in vivo coincided with cellular dehy-dration as indicated by a prominent increase inplasma chloride and a sharp decline in hematocrit.Exposure of cells to hyperosmotic conditions andconsequent dehydration concentrates ions andmacromolecules and disrupts the internal milieu.Such changes can denature proteins, alter enzymekinetics, and disrupt ionic bonds (Yancey et al.,’82; Somero, ’86; Burg et al., ’97). Accumulatingevidence suggests that hsps provide a rapid re-sponse to this cellular perturbation, limiting pro-tein denaturation until such time that mechanismsfor long-term regulation of cell volume (e.g., osmo-lyte transporters and antiport pumps) develop.Upregulation of hsp70 in response to osmotic stresshas been observed in mammalian renal cells (Cohenet al., ’91; Petronini et al., ’93; Sheikh-Hamad etal., ’94; Rauchman et al., ’97) and more recently infish tissues (Smith et al., ’99b). Additionally,upregulation of osp94, a member of the hsp110/SSE gene family, has been reported during hy-

perosmotic stress of renal cells (Santos et al., ’98).The present study is the first report of upregu-lation of hsp90 in response to hyperosmotic stress.

The absence of an increase in hsp90 accompa-nying the marked elevation of hsp90 mRNA inbranchial lamellae subjected to osmotic stress wasunexpected. Previous investigations suggestedthat hyperosmotic conditions can disrupt the pro-tein synthetic machinery and reduce the rate ofsynthesis of all but a limited number of proteins(Harrington and Alm, ’88; Cohen et al., ’91; Kultz,’96; Kurz et al., ’98; Smith et al., ’99b). Preferen-

CLONING AND CHARACTERIZATION OF SALMON hsp90 cDNA 209

tial synthesis of hsp70 during osmotic stress(Cohen et al., ’91; Petronini et al., ’93; Rauchmanet al., ’97; Santos et al., ’98; Smith et al., ’99b),but not hsp90, might be related to the differentfunctions of these chaperones. Translation of hsp90mRNA may lag behind transcription because of cel-lular perturbation caused by elevation in the con-centrations of intracellular Na+ and Cl–. Thepossibility also remains that synthesis of hsp90 isregulated at both the levels of transcription andtranslation. Although osmotic stress markedlystimulated gene expression, translation of themessage may require additional factors for ini-tiation or elongation.

We had anticipated increased transcription ofhsp90 in both branchial lamellae and kidney tis-sue subjected to osmotic stress because of the roleof these tissues in osmoregulation. Although invitro and in vivo exposure of salmon to thermalstress stimulated an increase in hsp90 mRNA in

both tissues, exposure to osmotic stress elicitedan hsp90 response only in branchial lamellae. Sev-eral plausible explanations might account for thistissue specific response to osmotic stress. It couldbe argued that the anatomical location and physi-ology of the kidney reduce exposure to osmoticstress. The kidney of teleost fishes is incapable ofconcentrating ions to produce the hypertonic urinecharacteristic of the mammalian renal system,limiting exposure to ion concentrations no greaterthan those of the plasma (Evans, ’98). In contrast,the mucosal cells of the branchial lamellae areexposed directly to full salinity seawater, increas-ing the potential for denaturation of cellularproteins. Consistent with this hypothesis, tissue-specific responses to thermal shock of salmon andother species of fish indicate that the hsp responseis more rapid and prominent in the gill than ininternal organs (Koban et al., ’91; Smith et al.,99a). However, the absence of an hsp90 response

Fig. 7. Northern blot analyses of hsp90 expression in (A)branchial lamellae and (C) kidney of salmon from the ex-periment described in Figure 6. RNA was collected from fishmaintained in freshwater (0) or following 24 or 96 hr expo-

sure to seawater. Values for (B) branchial lamellae and (D)kidney are presented as the mean ± SD (n = 4 fish per timepoint). Time points in the same figure are marked with * arestatisitically different (P < 0.05).

210 F. PAN ET AL.

in isolated kidney tissue exposed to osmotic stresssuggests a response unrelated to protein denatur-ation. Interestingly, hsp70 but not hsp90, is in-duced in the mammalian renal medulla where theconcentrations of ions and urea can exceed molallevels (Guyton, ’86; Burg et al., ’97).

Upregulation of hsp90 in branchial lamellaeduring hyperosmotic stress could facilitate the roleof cortisol in ion transport in the fish gill, the pri-mary site of monovalent ion excretion. Cortisolstimulates differentiation of branchial chloridecells and an increase in the activity of branchialNa+/K+ ATPase in several species of fish, includ-ing salmonids (McCormick, ’95). Consistent withthis role, circulating levels of cortisol increase dur-ing the parr–smolt transformation of salmonids,leading to increased tolerance of seawater (Hoar,’88). The capacity of the gill to respond to endog-enous cortisol appears related to the populationof high-affinity cortisol receptors present (Shrimp-ton and Randall, ’94). Investigations with mam-malian cell lines indicate that hsp90 is essentialfor effective glucocorticoid action (Dalman et al.,’89; Vamvakopoulos, ’93; Bohen, ’95; Kang et al.,’99). It has been hypothesized that hsp90 bindsto the nascent glucocorticoid receptor and forms amultimeric complex that conveys affinity for ligandto the aporeceptor and reduces non-specific inter-action. Indeed, glucocorticoid receptors synthesizedin vitro in the absence of hsp90 or in vivo in yeastproducing mutant hsp90 fail to bind ligand (Dal-man et al., ’89; Bohen, ’95). Stimulation of gill hy-poosmoregulatory capacity by cortisol suggests anincreased requirement for cortisol receptors andhsp90 during adaptation to seawater.

Implication of hsp90 in signal transduction (forreviews see Pratt, ’97, ’98) lends additional supportto a role in branchial osmoregulation. Of specificrelevance is the demonstration that the families ofmitogen-activated protein kinases (MAPK) andstress-activated protein kinases (SAPK) are mark-edly activated by osmotic shock (Matsuda et al.,’95, Kultz et al., ’98). Hsp90 binds to specific ki-nases of the MAPK system and may be involvedin activation of this system (Pratt, ’97). Althoughinvestigation of these osmosensing cascades in fishis lacking, Kultz (’96) reported modification ofcJun, an AP-1 transduction factor of the SAPKsystem, in the gills of a euryhaline teleost (Gilli-chthys mirabilis) exposed to osmotic shock.

The results of these studies and a previousreport from our laboratory (Smith et al., ’99b)suggest that hsps are essential elements for ad-aptation of salmon to hyperosmotic stress. In the

previous investigation, hsp70 was upregulated inisolated branchial lamellae, erythrocytes, and liverof salmon incubated in hyperosmotic medium. Therelative quantity of hsp70 mRNA in branchiallamellae increased as much as 900% above thelevels in tissue maintained under isoosmotic con-ditions. In contrast, hyperosmotic stress of salmonin vitro and in vivo elicited a 73–133% increasein hsp90 mRNA in branchial lamellae and failedto stimulate upregulation of hsp90 in kidney. Thetissue specificity and differences is not the mag-nitude of response suggest distinct but comple-mentary roles of hsp70 and hsp90 in the initialstages of adaptation of salmon to seawater. Con-sistent with its role in protein renaturation andnonspecific chaperonin function, hsp70 likely de-creases the lability of or renatures proteins sus-ceptible to denaturation upon an increase in theconcentration of intracellular ions (i.e., Na+ andCl–) following transfer of salmon to seawater. Theless prominent response of hsp90 in salmon tis-sue suggests a more specific role, perhaps in sig-nal transduction and subsequent development ofhypoosmoregulatory capacity in the gill (e.g.,chloride cell differentiation and increased Na+/K+ ATPase activity). Investigations are in progressto determine the role of other hsps and novel pro-teins in adaptation of salmon to seawater.

ACKNOWLEDGMENTSThe authors thank Larry Lofton of the North

Attleboro National Fish Hatchery, North Attle-boro, MA, and Peter Angelone of Lafayette StateHatchery, North Kingstown, RI, for supplying theAtlantic salmon used in these investigations (RIAgriculture Experiment Station contribution no.3710).

LITERATURE CITEDAli A, Krone PH, Pearson DS, Heikkila JJ. 1996. Evalua-

tion of stress-inducible hsp90 gene expression as a potentialbiomarker in Xenopus laevis. Cell Stress Chaperones1:62–65.

Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z,Miller W, Lipman DJ. 1997. Gapped blast and psi-blast: anew generation of protein database search programs.Nucleic Acids Res 25:3389–3402.

Bjornsson BT, Ogasawara T, Hirano T, Bolton JP, Bern HA.1988. Elevated growth hormone levels in stunted Atlanticsalmon, Salmo salar. Aquaculture 73:275–281.

Bohen SP. 1995. Hsp90 mutants disrupt glucocorticoid recep-tor ligand binding and destabilize aporeceptor complexes.J Biol Chem 270:29433–29438.

Burg MB, Kwon ED, Kultz D. 1997. Regulation of gene ex-pression by hypertonicity. Ann Rev Physiol 59:437–455.

Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ. 1979.

CLONING AND CHARACTERIZATION OF SALMON hsp90 cDNA 211

Isolation of biologically active ribonucleic acid from sourcesenriched in ribonuclease. Biochemistry 18:5294–5299.

Cohen DM, Wasserman JC, Gullans SR. 1991. Immediateearly gene and hsp70 expression in hyperosmotic stress inMDCK cells. Am J Physiol 261(Cell Physiol 30):C594–C601.

Craig EA, Weismann JS, Horwich AL. 1994. Heat shock pro-teins and molecular chaperones: mediators of protein con-formation turnover in the cell. Cell 78:365–372.

Dalman FC, Bresnick EH, Patel PD, Perdew GH, Watson SJ,Pratt WB. 1989. Direct evidence that the glucocorticoid re-ceptor binds to hsp90 at or near the termination of recep-tor translation in vitro. J Biol Chem 264:19815–19821.

DuBeau SF, Pan F, Tremblay GC, Bradley TM. 1998. Ther-mal shock of salmon in vivo induces the heat shock proteinhsp 70 and confers protection against osmotic shock. Aqua-culture 168:311–323.

Duston J. 1994. Effect of salinity on survival and growth ofAtlantic salmon (Salmo salar) parr and smolts. Aquacul-ture 121:115–124.

Evans DH, editor. 1998. The physiology of fishes, 2nd ed. BocaRaton, FL: CRC Press. 519 p.

Gonda DK, Bachmair A, Wunning I, Tobias JW, Lane WS,Varshavsky A. 1989. Universality and structure of the N-end rule. J Biol Chem 264:16700–16712.

Guruprasad K, Reddy BV, Pandit MW. 1990. Correlation be-tween stability of a protein and its dipeptide composition: anovel approach for predicting in vivo stability of a proteinfrom its primary sequence. Prot Engineer 4:155–161.

Guyton AC. 1986. Textbook of medical physiology, 7th ed.Philadelphia: Saunders. 1057 p.

Handeland SO, Jarvi T, Ferno A, Stefansson SO. 1996. Os-motic stress, antipredator behaviour and mortality of At-lantic salmon smolts. Can J Fish Aquat Sci 53:2673–2680.

Harrington HM, Alm DM. 1988. Interaction of heat and saltshock in cultured tobacco cells. Plant Physiol 88:618–625.

Hoar WS. 1988. The physiology of smolting salmonids. In:Hoar WS, Randall DJ, editors. Fish physiology, vol. 11B.San Diego: Academic Press. p 275–343.

Kang KI, Meng X, Devin-Leclerc J, Bouhouche I, Chadli A,Cadepond F, Baulieu EE, Catelli MG. 1999. The molecularchaperone hsp90 can negatively regulate the activity of aglucocorticosteroid-dependent promoter. Proc Natl Acad SciUSA 96:1439–1444.

Koban M, Yup AA, Agellon LB, Powers DA. 1991. Molecularadaptation to enviornmental temperature: heat-shock re-sponse of the eurythermal teleost Fundulus heteroclitus. MolMar Biol Biotech 1:1–17.

Krone P, Sass JB. 1994. Hsp90α and hsp90β genes are presentin the zebrafish and are differentially regulated in develop-ing embryos. Biochem Biophys Res Comm 204:746–752.

Kultz D. 1996. Plasticity and stressor specificity of osmoticshock and heat shock responses of Gillichthys mirabilis gillcells. Am J Physiol 271(Cell Physiol 40):C1181–C1193.

Kultz D, Madhany S, Burg MB. 1998. Hyperosmolality causesgrowth arrest of murine kidney cells: induction of GADD45and GADD153 by osmosensing via stress-activated proteinkinase 2. J Biol Chem 273:13645–13651.

Kurz AK, Schliess F, Haussinger D. 1998. Osmotic regula-tion of the heat shock response in primary rat hepatocytes.Hepatology 28:774–781.

Madsen SS. 1990. The role of cortisol and growth hormone inseawater adaptation and development of hypoosmoregula-tory mechanisms in sea trout parr (Salmo trutta trutta).Gen Comp Endocrinol 79:1–11.

Matsuda S, Kawasaki H, Moirguchi T, Gotoh Y, Nishida E.1995. Activation of protein kinase cascades by osmotic shock.J Biol Chem 270:12781–12786.

McCormick S. 1995. Hormonal control of gill Na+,K+-ATPaseand chloride cell function: In: Wood CM, Shuttleworth TJ,editors. Cellular and molecular approaches to fish ionic regu-lation, fish physiology vol. 14. New York: Academic Press. p285–313.

McCormick SD, Bern HA. 1989. In vitro stimulation of Na+-K+-ATPase activity and ouabain binding by cortisol in cohosalmon gill. Am J Physiol 256:R707–R715.

Nathan DF, Vos MH, Lindquist S. 1997. In vivo functions ofSaccharomyces cerevisiae Hsp90 chaperone. Proc Natl AcadSci USA 94:12949–12956.

Nover L. 1991. Heat shock response. Boca Raton, FL: CRCPress. 509 p.

Parsell DA, Lindquist S. 1993. The function of heat-shockproteins in stress tolerance: degradation and reactivationof damaged proteins. Ann Rev Genet 27:437–496.

Petronini R, Alfieri R, De Angelis E, Campanini C, BorghettiAF, Wheeler KP. 1993. Different HSP70 expression and cellsurvival during adaptive response of 3T3 and transformed3T3 cells to osmotic stress. Br J Cancer 67:493–499.

Pratt WB. 1997. The role of the hsp90-based chaperone sys-tem in signal transduction by nuclear receptors and recep-tors signaling via MAP kinase. Ann Rev Pharmacol Toxicol37:297–326.

Pratt WB. 1998. The hsp90-based chaperone system: involve-ment in signal transduction from a variety of hormone andgrowth factor receptors. Proc Soc Exp Biol Med 217:420–434.

Rauchman MI, Pullman J, Gullans SR. 1997. Induction ofmolecular chaperones by hyperosmotic stress in mouse in-ner medullary collecting duct cells. Am J Physiol 273(Re-nal Physiol 42):F9–17.

Rebbe NF, Ware J, Bertina RM, Modrich P, Stafford DW. 1987.Nucleotide sequence of a cDNA for a member of the human90-kDa heat shock protein family. Gene 53:235–247.

Robertson JC, Bradley TM. 1991. Hepatic ultrastructurechanges associated with the parr–smolt transformation ofAtlantic salmon (Salmo salar). J Exp Zool 260:135–148.

Sambrook J, Fritsch EF, Maniatis T. 1989. Molecular clon-ing: a laboratory manual, 2nd ed. Plainview, NY: Cold SpringHarbor Laboratory Press.

Santos BC, Chevaile A, Kojima R, Gullans SR. 1998. Charac-terization of the Hsp110/SSE gene family response tohyperosmolality and other stresses. Am J Physiol 274:F1054–1061.

Sheikh-Hamad D, Garcia-Perez A, Ferraris JD, Peters EM,Burg MB. 1994. Induction of gene expression by heat shockversus osmotic shock. Am J Physiol 267(Renal Fluid Elec-trolyte Physiol 36):F28–F34.

Shrimpton MJ, Randall DJ. 1994. Downregulation of corti-costeroid receptors in gills of coho salmon due to stress andcortisol treatment. Am J Physiol 267(Reg Integr CompPhysiol 36):R432–R438.

Smith TR, Tremblay GC, Bradley TM. 1999a. Characteriza-tion of the heat shock protein response of Atlantic salmon(Salmo salar). Fish Physiol Biochem 20:279–292.

Smith TR, Tremblay GC, Bradley TM. 1999b. Hsp70 and a 54kDa protein (osp54) are induced in salmon (Salmo salar) inresponse to hyperosmotic stress. J Exp Zool 284:286–298.

Somero G. 1986. Protons, osmolytes and fitness of internalmilieu for protien function. Am J Physiol 251(Reg IntegrComp Physiol 20):R197–R213.

Vamvakopoulos NO. 1993. Tissue-specific expression of heat

212 F. PAN ET AL.

shock proteins 70 and 90: implication for differential sensi-tivity of tissues to glucocorticoids. Mol Cell Endocrinol98:49–54.

Visner GA, Fogg S, Nick HS. 1996. Hyperoxia-responsive pro-teins in rat pulmonary microvascular endothelial cells. AmJ Physiol 270(Lung Cell Mol Physiol 14):L517–L525.

Wei L-Y, Roepe PD. 1994. Low external pH and osmotic shockincrease the expression of human MDR protein. Biochem-istry 33:7229–7238.

Yancey PH, Clark ME, Hand SC, Bowlus RD, Somero GN.1982. Living with water stress: evolution of osmolyte sys-tems. Science 217:1214–1222.