Analysis of inhibition by H89 of UCP1 gene expression andthermogenesis indicates protein kinase A mediation of L3-adrenergic

signalling rather than L3-adrenoceptor antagonism by H89

J. Magnus Fredriksson *, Ha®kan Thonberg, Kerstin B.E. Ohlson, Ken-ichi Ohba 1,Barbara Cannon, Jan Nedergaard

Wenner-Gren Institute, Arrhenius Laboratories F3, Stockholm University, SE-106 91 Stockholm, Sweden

Received 20 July 2000; received in revised form 13 December 2000; accepted 8 January 2001

Abstract

Although it has generally been assumed that protein kinase A (PKA) is essential for brown adipose tissue function, this hasnot as yet been clearly demonstrated. H89, an inhibitor of PKA, was used here to inhibit PKA activity. In cell extracts, it wasconfirmed that norepinephrine stimulated PKA activity, which was abolished by H89 treatment. In isolated brownadipocytes, H89 inhibited adrenergically induced thermogenesis (with an IC50 of approx. 40 WM), and in cultured cells,adrenergically stimulated expression of the uncoupling protein-1 (UCP1) gene was abolished by H89 (full inhibition with 50WM). However, H89 has been reported to be an adrenergic antagonist on L1/L2-adrenoceptors (AR). Although adrenergicstimulation of thermogenesis and UCP1 gene expression are mediated via L3-ARs, it was deemed necessary to investigatewhether H89 also had antagonistic potency on L3-ARs. It was found that EC50 values for L3-AR-selective stimulation ofcAMP production (with BRL-37344) in brown adipose tissue membrane fractions and in intact cells were not affected byH89. Similarly, the EC50 of adrenergically stimulated oxygen consumption was not affected by H89. As H89 also abolishedforskolin-induced UCP1 gene expression, and potentiated selective L3-AR-induced cAMP production, H89 must be activedownstream of cAMP. Thus, no antagonism of H89 on L3-ARs could be detected. We conclude that H89 can be used as apharmacological tool for elucidation of the involvement of PKA in cellular signalling processes regulated via L3-ARs, andthat the results are concordant with adrenergic stimulation of thermogenesis and UCP1 gene expression in brown adipocytesbeing mediated via a PKA-dependent pathway. ß 2001 Elsevier Science B.V. All rights reserved.

Keywords: Protein kinase A; Uncoupling protein-1; Thermogenesis ; L3-Adrenoceptor; N-[2-(p-Bromocinnamylamino)ethyl]-5-isoquino-linesulfonamide; Brown adipocyte

0167-4889 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 7 - 4 8 8 9 ( 0 1 ) 0 0 0 7 0 - 2

Abbreviations: AR, adrenoceptor (adrenergic receptor) ; CREB, cAMP-responsive element-binding protein; Erk1/2, extracellular sig-nal-regulated kinases 1 and 2; H89, N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulphonamide; MAP kinase, mitogen-activatedprotein kinase; NE, norepinephrine; PKA, protein kinase A (cAMP-dependent protein kinase) ; UCP1, uncoupling protein-1

* Corresponding author. Fax: +46-8-156756; E-mail : mf@zoofys.su.se1 Present address: Department of Pharmacology, Tokyo Women's Medical University, School of Medicine, 8-1 Kawada-cho, Shinju-

ku-ku, Tokyo 162-8666, Japan.

BBAMCR 14730 17-4-01

Biochimica et Biophysica Acta 1538 (2001) 206^217www.bba-direct.com

1. Introduction

L3-Adrenergic e¡ects are highly important forbrown adipose tissue function, both acutely (thermo-genesis) [1,2] and chronically (expression of the un-coupling protein-1 (UCP1) gene) [3]. Activation ofL3-adrenoceptors (L-ARs) on brown adipocytes in-creases intracellular cAMP levels [4], which is be-lieved to activate protein kinase A (PKA), whichthen presumably activates lipolysis and thermogene-sis [5,6] and Ucp1 expression [7]. However, PKA in-volvement has never been directly demonstrated. Inreality, the only evidence for involvement of PKA inthe mediation of the L3-adrenergic signal is that ad-renergic stimulation of thermogenesis and Ucp1 ex-pression can be mimicked by increases in cAMP lev-els [2,3]. Thus, the assumption of an involvement ofPKA may be questioned on two grounds. One is thatthe ability of a substance to mimic a function is notevidence that it is the physiological mediating agent.Indeed, L3-adrenergic pathways not involving in-creases in cAMP levels have been described [8,9],and thermogenesis and Ucp1 expression could bemediated via such pathways, even though cAMPcan mimic the adrenergic stimulation. The second isthat the tenet that cellular responses stimulated viaincreased cAMP levels are exclusively mediated viaPKA activation is no longer valid. Recent ¢ndingshave demonstrated that cAMP can also signal viaother cellular mediators [10^19]. Therefore, to inves-tigate whether PKA actually mediates the L3-adren-ergic stimulation of Ucp1 expression and thermogen-esis, we attempted to inhibit PKA activity innorepinephrine (NE)-stimulated brown adipocytes.

One established inhibitor of PKA is N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulphona-mide (H89) [20,21]. However, unspeci¢c e¡ects ofcellular inhibitors can compromise the interpretationof results obtained. In this respect, it has recentlybeen reported that H89 also functions as a directcompetitive adrenergic antagonist on L1- and L2-ARs [22]. While this clearly severely reduces its use-fulness as a tool in investigations of L1- and L2-ad-renergic signalling mechanisms, it also leads to ap-prehension that H89 could be an antagonist also forthe third L-adrenoceptor, the L3-AR. This possibilityhas not as yet been investigated, but an examinationof this issue is evidently essential for further use of

H89 in investigations of L3-AR signalling in brownadipocytes and in other cells and tissues possessingL3-ARs.

In the present study, we have therefore also inves-tigated the possible antagonistic e¡ect of H89 on L3-ARs. We conclude that H89 has no antagonistic po-tency for L3-ARs. Rather, H89 appears to be a sat-isfactory inhibitor of the adrenergically inducedpathways earlier anticipated, and herein shown to,progress through PKA activation in these cells,such as stimulation of thermogenesis and Ucp1 ex-pression.

2. Materials and methods

2.1. Cell culture

Brown adipocyte precursors were isolated from 3^4-week old mice of the NMRI strain, obtained froma local supplier (Eklunds), principally as described byNechad et al. [23] with modi¢cations as describedearlier [24,25].

The cells were cultured in Costar 6-well or 12-wellplates, as detailed in Fredriksson et al. [25]. The cul-ture medium was that described earlier [24,25] (prin-cipally Dulbecco's modi¢ed Eagle's medium(DMEM) with 10% newborn calf serum), and thecells were cultured at 37³C in a water-saturated at-mosphere of 8% CO2 in air in a Heraeus CO2-auto-zero B5061 incubator. On day 1, the cultures werewashed with prewarmed DMEM, and fresh pre-warmed medium was then added. Medium wasthen routinely changed every other day until theday of experimentation, which was day 5^6.

2.2. Analysis of PKA activity

After stimulation of cell cultures, protein extractswere isolated. Brie£y, each culture dish was rinsedwith PBS and cells were collected by scraping in0.5 ml of extraction bu¡er (25 mM Tris-HCl,0.5 mM EDTA, 0.5 mM EGTA, 10 mM L-mercap-toethanol, 1 Wg/ml leupeptin, 1 Wg/ml aprotinin and0.5 mM PMSF). The suspension was homogenisedby sucking through a 23G needle ten times. It wasthen centrifuged 5 min at 14 000Ug, after which thesupernatant was collected and 5 Wl aliquots were

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217 207

used to quantify PKA activity with the SignaTECTcAMP-Dependent PKA Assay System (Promega) ac-cording to the manufacturer's instructions. Brie£y, inthis assay, incorporation of [Q-32P]ATP into a bio-tinylated peptide (LRRASLG), a highly speci¢c sub-strate for PKA, was analysed by scintillation count-ing after incubation of the sample and peptide in thepresence of 100 WM ATP (with the labelled ATPincluded). Some of these incubations were performedin the presence of 50 WM H89. The activity of eachextract was normalised to the total activity of thatextract induced by addition of 5 WM cAMP to aseparate incubation.

2.3. Analysis of cAMP levels in cultured cells

After stimulation of cell cultures for 15 min, theculture medium was aspirated and the cells weretreated principally as described earlier [4]. Brie£y,0.8 ml 75% ethanol with 1 mM EDTA was addedto each well, and after 10 min, the cells were har-vested by scraping. Ethanol was removed by Speed-Vac centrifugation and the pellets were suspended in0.5 ml 4UTE bu¡er and sonicated 5 s. After centri-fugation, a 25 Wl supernatant aliquot from each sam-ple was used for determination of cAMP levels withthe Cyclic AMP Assay System (Amersham), accord-ing to the manufacturer's instructions.

2.4. RNA isolation and analysis of mRNA levels

After stimulation of cell cultures, the culture me-dium was discarded and the cells were dissolved in0.8 ml of an Ultraspec solution (Biotecx), and themanufacturer's procedure for RNA isolation was fol-lowed. 10 Wg of total RNA was separated by electro-phoresis in an ethidium bromide-containing agarose-formaldehyde gel, as described earlier [4,25]. The in-tensity of the 18S and 28S rRNA bands under UVlight was checked to verify that all samples wereequally loaded and that no RNA degradation hadoccurred. mRNA levels were analysed by Northernblotting as described earlier [4,25].

The UCP1 and L-actin cDNAs were those usedearlier [3]. The probes were labelled with[K-32P]dCTP with Ready To Go DNA labellingbeads (Amersham Pharmacia Biotech), according tothe manufacturer's instructions.

2.5. Preparation of membrane fractions

A membrane fraction of brown adipose tissue wasprepared from adult Syrian hamsters living at roomtemperature. The interscapular and cervical depots ofbrown adipose tissue were minced in bu¡er (Tris-HCl/Tris-Base 50 mM, MgCl2 10 mM, ascorbicacid 0.5 mM, EDTA 1 mM, PMSF 0.1 mM; pHwas adjusted to 7.4 with NaOH) supplemented with0.25 M sucrose, and homogenised in a Potter-Elve-hjem homogeniser; this and the following procedureswere performed at about 4³C. The homogenate wascentrifuged for 10 min at 9200Ug, the fat layer wasremoved and the supernatant was ¢ltered throughone layer of silk cloth. The supernatant was thencentrifuged in sucrose bu¡er for 30 min at100 000Ug, the pellet was homogenised again andcentrifuged at the same speed once more, in bu¡erwithout sucrose. After this centrifugation, the pelletwas rehomogenised in bu¡er and ¢ltered againthrough silk cloth. The membrane homogenate wasresuspended to about 3 mg protein/ml and thenstored at 380³C. Protein concentration was deter-mined by the method of Bradford, with bovine se-rum albumin as standard.

2.6. Analysis of cAMP production in membranefractions

The cAMP assay was performed in an assay mixof 90 Wl containing: ATP 0.5 mM, MgCl2 10 mM,creatine phosphate 10 mM, creatine kinase 50 units/ml, theophylline 2 mM, dithiothreitol 1 mM, in Tris-Base 25 mM; pH 7.4, adjusted with HCl. To eachassay tube, 60 Wl membrane fraction (1 mg protein/ml) was added together with agents as described.After incubation at 37³C for 10 min, the assay tubeswere boiled for 5 min, put on ice and centrifuged. Asupernatant aliquot (50 Wl) was taken from each in-cubation tube and cAMP levels were determinedwith the Cyclic AMP Assay System (Amersham), ac-cording to the manufacturer's instructions.

2.7. Measurement of oxygen consumption in freshlyisolated cells

Brown adipocytes were isolated from Syrian ham-sters of both sexes (about 35 weeks old), principally

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217208

as described earlier, by collagenase digestion ofpooled brown adipose tissue depots [26]. The cellswere counted in a Bu«rker chamber. Samples withthe intended amount of cells for each experimentwere taken from a concentrated cell suspension. Ex-periments were also performed with cells fromNMRI mice (about 10 week old males), as indicatedin the text.

The oxygen consumption rates of the isolatedbrown adipocytes were monitored with a Clark-type oxygen electrode, as described previously [27].80 000^100 000 cells were added to a magneticallystirred oxygen electrode chamber thermostatted to37³C, containing Krebs-Ringer bicarbonate bu¡er[26]. The cells were incubated with the indicated con-centrations of H89, or with water addition for thecontrols, for 10 min before the chamber was closed.After registration of the basal rate of oxygen con-sumption for 5 min, norepinephrine was added cu-mulatively in increasing concentrations. Additionswere made with a Hamilton syringe through a smallhole in the cover of the chamber. The oxygen con-sumption rate was then calculated as described ear-lier [26,28].

2.8. Chemicals

H89 (Calbiochem), BRL-37344 (SmithKline Bee-cham), norepinephrine ((3)-Arterenol bitartrate;Sigma) and CGP-12177 (CGP-12177A; Ciba-Geigy)were dissolved in water. Forskolin (Sigma) and lauricacid (Schuchardt, Munich) were dissolved in dimeth-yl sulphoxide (Sigma); ¢nal concentration in assaywas maximally 0.1%.

3. Results

3.1. PKA is activated by norepinephrine in primarybrown adipocyte cultures

The increase in cAMP levels observed in brownadipocytes after adrenergic stimulation [4,25,29] isexpected to induce an activation of PKA. To verifythat PKA is activated in the cell system utilised here,we determined the activity of PKA in cell extractsfrom cultured brown adipocytes stimulated withNE. PKA was activated by NE in a dose-dependent

manner (Fig. 1A). This observation in primarybrown adipocytes in culture is principally in agree-ment with what has been demonstrated in slices ofbrown adipose tissue [30] and in a human brownadipocyte-like cell line [31], but it is in contrast toan observation in immortalised di¡erentiated brownadipocytes, in which adrenergic stimulation unex-pectedly slightly decreased basal PKA activity [32].

3.2. E¡ect of H89 on activity of PKA from brownadipocytes

To con¢rm that H89 was indeed able to inhibitPKA extracted from brown adipocytes, the cellswere stimulated with NE, and PKA activity was an-alysed in extracts from these cells. When added tothe cell extracts, H89 reduced the basal PKA activityand abolished the NE-induced PKA activity (Fig.1B). The maximal PKA activity, i.e. that inducedby addition of an excess of cAMP to the cell extracts,was also fully abolished by H89; only 7% of theactivity remained (not shown; P6 0.0001, Student'spaired t-test with six experiments). Thus, as expected,H89 e¡ectively inhibited NE-induced PKA activity inextracts of brown adipocytes.

3.3. E¡ect of H89 on the kinetics of cAMPproduction stimulated via L3-adrenoceptors

To ascertain that possible inhibiting e¡ects of H89on cellular functions were not caused by L3-adreno-receptor antagonism, this issue was investigated inmembrane fractions.

The method of choice to investigate whether theadrenergic antagonistic property of H89 reported forL1- and L2-ARs [22] also applied to L3-ARs wouldseem to be receptor binding studies. However, twoproblems exist: ¢rst, no established, high a¤nity ra-dioactively labelled ligand exists for L3-ARs. Second,and more importantly, receptor binding studies donot address the signalling function of the receptor.Therefore, we analysed the e¡ect of H89 on the ki-netics of L3-AR-mediated induction of cAMP pro-duction in membrane fractions isolated from brownadipose tissue. In such crude membrane fractions, L3-ARs are potent, although probably not exclusive,stimulators of adenylyl cyclase activity [33,34], butthe use of L3-AR-selective agonists ensured that it

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217 209

was only L3-AR responses that were investigated.The L3-AR-selective agonist BRL-37344 induced a3.3-fold increase in cAMP levels in brown adiposetissue membranes, with an EC50 of 141 þ 24 nM(Fig. 2A).

A competitive adrenergic antagonistic e¡ect ofH89 on L3-ARs would be expected to lead to a shiftin the kinetics to a markedly higher EC50 value. Re-ported Ki values for L1- and L2-AR antagonism by

H89 are 350 nM and 180 nM, respectively [22]. If asimilar antagonistic potency of H89 were to exist forL3-ARs (with a Ki of around 0.2 WM), an H89 con-centration of 50 WM, as used here, would be expectedto lead to an increase in the EC50 value by approx.250-fold (EC50 (with H89) = EC50U(1+(50 WM/0.2WM))), i.e. to an apparent EC50 of about 35 WM.

However, as seen in Fig. 2B, in the presence of50 WM H89 there was no increase in the EC50 value.Rather, a slight but insigni¢cant decrease was ob-served: the EC50 was 116 þ 8 nM. Thus, no adrener-gic antagonistic e¡ect of H89 could be detected inmembrane fractions of brown adipose tissue stimu-lated via L3-ARs.

To substantiate this conclusion, we performed sim-ilar experiments in intact cultured brown adipocytes.In these di¡erentiated brown adipocytes, the L3-ARis the only L-AR coupled to cAMP production [4]and thus this model system could be considered apure L3 system, in contrast to the membrane frac-tions. In these cells, BRL-37344 potently inducedcAMP production to 40-fold that of basal levels,with an EC50 of 17 þ 7 nM (Fig. 3A), in agreementwith earlier reports [4]. Calculated as above, an an-tagonistic potency of H89 for L3-ARs similar to thatfor L1- and L2-ARs would shift the EC50 to an ap-parent EC50 of about 4 WM. However, again, treat-ment with 50 WM H89 did not change the kinetics ofthe L3-AR-mediated cAMP production: the EC50

Fig. 1. E¡ect of H89 on PKA activity from brown adipocytes.Brown adipocytes were isolated and cultured as described inSection 2. They were stimulated with norepinephrine (NE) orwater (C), after which cell extracts were used for analysingPKA activity. In each extract, 100% was de¢ned as the PKAactivity induced by addition of an excess of cAMP (5 WM).(A) PKA activity after 5 min stimulation with the indicated NEconcentrations. The values are mean points from two experi-ments with single wells. Curve is drawn according to simple Mi-chaelis-Menten kinetics. (B) PKA activity after 5 min stimula-tion with 1 WM NE. The resulting cell extracts were analysedfor PKA activity with or without addition of 50 WM H89. Thevalues are means þ S.E. from nine experiments for extracts with-out H89 and four experiments for extracts with H89, all fromsingle wells. In the nontreated extracts, Student's paired t-testof the logarithmetised values, comparing control to NE, yieldeda signi¢cant e¡ect (P6 0.01) of NE treatment. Student's pairedt-test of the logarithmetised values from parallel experiments,comparing NE to H89+NE, yielded a signi¢cant inhibition(P6 0.01) by H89 treatment.6

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217210

was 15 þ 7 nM (Fig. 3B). Thus, neither in the mem-brane fractions nor in the cultured intact cells couldany L3-AR antagonism by H89 be detected.

Notably, the maximal e¡ect of BRL-37344 wasmarkedly potentiated by H89 in the intact cells,with an increase of 4-fold over that induced by theL3-AR agonist alone (Fig. 3A,B), an e¡ect which wasnot present in the membrane fractions (Fig. 2A,B).Similarly, cAMP production induced by NE or theL3-AR agonist CGP-12177 [35] was also markedlypotentiated by H89 treatment in the cultured cells(not shown). We have not clari¢ed the mechanismsof this potentiating e¡ect by H89. However, sincethis potentiating e¡ect is not present in the adrener-gically stimulated membrane fractions, post-cyclasemechanisms must be involved. In other systems,PKA has been demonstrated to inhibit the activityof certain adenylyl cyclase isoforms by about 50%[36,37]. Thus, an attenuation by H89 of such an in-hibitory PKA e¡ect may perhaps explain the massiveinduction of cAMP production in the H89-treatedcells.

Taken together, these results demonstrate thatH89 has no antagonistic e¡ect on L3-AR-stimulatedcyclase activity, and H89 may even augment the re-sulting cAMP levels. These results therefore also im-ply that, in brown adipocytes, inhibitory e¡ects byH89 must necessarily involve signalling mechanismsdownstream of cAMP.

3.4. E¡ect of H89 on adrenergically induced oxygenconsumption (thermogenesis)

To investigate whether adrenergically inducedthermogenesis in brown adipocytes is mediated via

PKA, we analysed the e¡ect of H89 on this process.Isolated mature hamster brown adipocytes werestimulated with NE, the endogenous thermogenic ac-tivator, which potently stimulates oxygen consump-tion ^ i.e. thermogenesis ^ in these isolated cells viaL3-ARs [1,2,38]. This stimulation can be mimickedby cAMP-elevating agents [2].

As exempli¢ed in Fig. 4A, successive additions ofNE to isolated brown adipocytes generated, as ex-pected, an approx. 10-fold increase in oxygen con-

Fig. 2. E¡ect of H89 on the kinetics of cAMP productionstimulated via L3-adrenoceptors in membrane fractions. Mem-brane fractions were isolated from brown adipose tissue andtreated as indicated, after which cAMP levels were determinedas described in Section 2. The values are means from duplicateincubation tubes for each treatment; individual values are indi-cated with error bars. Curves were drawn according to simpleMichaelis-Menten kinetics. (A) cAMP levels after treatmentwith BRL-37344 (BRL) for 10 min. EC50 was 141 þ 24 nM. (B)cAMP levels after treatment with BRL as in A, but with 50 WMH89 present. EC50 was 116 þ 8 nM. The arrow indicates the ex-pected EC50 value if H89 had a Ki of 0.2 WM on the L3-AR.

C

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217 211

sumption. The mean EC50 under control conditions,i.e. without H89, was 155 nM (Fig. 4C). When sim-ilar experiments were performed in the presence ofdi¡erent H89 concentrations (as exempli¢ed in Fig.4B), no change in EC50 was observed (Fig. 4C).Again, if H89 exerted a similar antagonism as forthe classical L-ARs, the EC50 should be shifted sev-eral orders of magnitude (as illustrated in Fig. 4C,dotted line).

Thus, in agreement with the results above forcAMP generation in brown adipose tissue membranefractions (Fig. 2) and cultured brown adipocytes(Fig. 3), these results also indicate that no L3-ARantagonism was exerted by H89.

However, as is evident in Fig. 4B, H89 treatmentled to a marked inhibition of thermogenesis withouta¡ecting the basal metabolism of the cells. The oxy-gen consumption was dose-dependently reduced byH89 with an IC50 of about 40 WM (Fig. 4D). Thisindicates that adrenergically induced thermogenesisin brown adipocytes is mediated via PKA, which isin agreement with expectations [5,6], but has notbeen shown earlier. The IC50 of about 40 WM is ofa similar magnitude to that observed for cAMP-de-pendent processes in other whole-cell systems, e.g. byChijiwa et al. [20].

In the presence of a high dose of H89 (160 WM),the NE-induced oxygen consumption was practicallyabolished (Fig. 4B). However, even under these con-ditions, addition of the free fatty acid lauric acid,which activates mitochondrial thermogenic processesdirectly [28], still led to a fully potent increase inoxygen consumption (Fig. 4B). This indicates thatthe mitochondrial thermogenic processes were intactin the cells, even at very high H89 concentrations.

In a similar series of experiments on brown adipo-cytes isolated from mice, we obtained practicallyidentical results: no e¡ect of H89 on EC50 valuesfor NE, but an inhibitory e¡ect on the magnitudeof NE-induced thermogenesis, with an IC50 of about30 WM, i.e. a response very similar to that shown inFig. 4D (not shown).

Thus, these results demonstrate that the pathwayfor adrenergic activation of thermogenesis in brownadipocytes is indeed L3-AR activation, elevation ofcAMP levels and activation of PKA.

Fig. 3. E¡ect of H89 on the kinetics of cAMP productionstimulated via L3-adrenoceptors in intact cells. Brown adipo-cytes were isolated and cultured under conditions identical tothose in Fig. 1 and treated as indicated, after which the cellswere harvested and cAMP levels determined as described inSection 2. The values are means þ S.E. from two experimentswith duplicate wells for each treatment. Curves were drawn ac-cording to simple Michaelis-Menten kinetics. (A) cAMP levelsafter stimulation with BRL-37344 (BRL) for 15 min. Maximalincrease was 80 þ 7 pmol/well with EC50 = 17 þ 7 nM. (B) cAMPlevels after stimulation with BRL as in A, but with 50 WM H89added shortly before BRL. The values are from parallel wellsto those in A. Maximal increase was 337 þ 31 pmol/well withEC50 = 15 þ 7 nM. The arrow indicates the expected EC50 valueif H89 had a Ki of 0.2 WM on the L3-AR.

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217212

3.5. E¡ect of H89 on adrenergic stimulation ofUCP1 gene expression

As it was clear from the investigations above thatH89 did not have an antagonistic e¡ect on L3-ARsand did not disturb general cell metabolism (see alsobelow), we found H89 to be a suitable tool also toinvestigate the involvement of PKA in adrenergicstimulation of expression of the brown adipocytespeci¢c mitochondrial UCP1 gene. Adrenergic stim-

ulation of the expression of this gene is mediated viaL3-ARs and this e¡ect can be mimicked by increasedcAMP levels [3]. To examine further mediation viaPKA, cultured brown adipocytes were treated withH89 and the e¡ect on adrenergically induced Ucp1expression was analysed.

As expected, NE, and also the direct activator ofadenylyl cyclase, forskolin, were potent inducers ofUcp1 expression (Fig. 5A,B). However, in H89-treated cells, the stimulatory e¡ects of both NE

Fig. 4. E¡ect of H89 on the kinetics of adrenergically stimulated oxygen consumption. Mature brown adipocytes were freshly isolatedas described in Section 2 and treated with the indicated concentrations of H89 for 15 min. The cells then received cumulative addi-tions of norepinephrine (NE), during which time oxygen consumption was measured as described in Section 2. (A) Recording of oxy-gen consumption in control cells after additions of 1, 10, 100 nM, 1, 10 WM NE as indicated by arrows. (B) Recording of oxygen con-sumption in the presence of 160 WM H89 with additions of 1 nM^100 WM NE and 5 mM lauric acid as indicated by arrows. (C)EC50 values for NE-induced oxygen consumption in the presence of 0^80 WM H89. EC50 values were estimated from experiments asin A and B by linear interpolation based on logarithmic NE concentration data (as oxygen consumption was practically abolished at160 WM H89 (B), a meaningful EC50 value could not be obtained at this concentration). The values are means þ S.E. from two experi-ments. (D) Basal and maximal NE-induced oxygen consumption in the presence of 0^160 WM H89. The values are means þ S.E. from¢ve experiments for basal values and from two experiments for NE-induced values. a, basal oxygen consumption; b, maximal NE-in-duced oxygen consumption. The curve was drawn according to simple Michaelis-Menten kinetics.

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217 213

and forskolin were fully abolished (Fig. 5A,B).Although the inhibition by H89 of the NE e¡ect assuch could be understood either as an inhibition ofPKA or as an antagonism on the L3-AR, it is evidentthat the e¡ect on forskolin-induced expression couldnot be due to receptor antagonism. As H89 did not

decrease forskolin-induced cAMP levels (not shown),the inhibition must be downstream of cAMP.

To con¢rm that no unspeci¢c detrimental e¡ectson mRNA had occurred, the levels of L-actin mRNAwere also analysed. There were no signi¢cant e¡ectsof H89 treatment on L-actin mRNA levels (Fig. 5A),and thus no general detrimental e¡ects on mRNAhad occurred during the time frame investigated(100 min). Further, in these cells, adrenergic activa-tion of extracellular signal-regulated kinases 1 and 2(Erk 1/2) via the K1-AR pathway is una¡ected by 50WM H89 [39], which indicates that other adrenergicsignalling mechanisms are still intact in H89-treatedbrown adipocytes.

Although it has earlier been suggested that adre-nergically induced Ucp1 expression is induced viaPKA activity [40,41], and although it has been dem-onstrated that PKA (catalytic subunit) can mimicadrenergically induced Ucp1 expression [7], this isthe ¢rst demonstration that the NE-induced increasein UCP1 mRNA levels is indeed mediated via PKA.H89 has been reported to inhibit the adrenergicallystimulated increase in UCP1 protein by about 50% intransformed brown adipocytes [32], but in that sys-tem, adrenergic stimulation did not induce PKA ac-tivity [32].

Thus, the present results demonstrate that H89 isfunctional within the primary brown adipocytes usedin this study by inhibiting pathways downstream ofcAMP, which are expected to progress through PKAactivation, such as adrenergically induced Ucp1 ex-pression.

4. Discussion

4.1. H89 is not an antagonist on the L3-adrenoceptor

In the present study, we have investigated whetherPKA is involved in the adrenergically stimulatedpathways which activate thermogenesis and expres-sion of the UCP1 gene in brown adipocytes. Theinvolvement of PKA in stimulation of these process-es has been assumed earlier, but never directly dem-onstrated. To investigate this assumed PKA involve-ment, we used the PKA inhibitor H89. However, inother cell systems, H89 has been demonstrated tohave potent adrenergic antagonistic potency for the

Fig. 5. E¡ect of H89 on adrenergic stimulation of UCP1 geneexpression. Brown adipocytes were cultured under conditionsidentical to those in Fig. 1. On day 6, the cells were treated ornot with 50 WM H89 for 40 min and thereafter stimulated, asindicated, for 1 h, after which the cells were harvested andRNA was isolated for Northern blot analysis as described inSection 2. C, control; NE, 10 WM norepinephrine; Fsk, 10 WMforskolin. (A) Northern blot resulting from treatment as indi-cated. Upper panel shows signals for UCP1 mRNA. Lowerpanel shows signals for L-actin mRNA in the same lanes. (B)UCP1 mRNA levels after stimulation as indicated. The valuesare means þ S.E. from one experiment with quadruplicate wellsfor each treatment. The mean value of NE-treated cells was setto 100%. The full abolishment by H89 of the NE e¡ect was re-produced in two other independent experiments under identicalconditions.

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217214

classical L-ARs, L1 and L2 [22]. In order to evaluatethe usefulness of H89 for investigations on L3-ARsignalling pathways, it was therefore necessary to in-vestigate whether H89 had antagonistic potency alsofor the L3-ARs. Our investigation demonstrates thatthe kinetics (EC50) of L3-AR-mediated stimulation ofcAMP production in brown adipose tissue mem-brane fractions and intact brown adipocytes wereuna¡ected by high concentrations of H89. Similarly,the EC50 of adrenergically induced oxygen consump-tion in intact cells was not signi¢cantly a¡ected byH89. It is clear that if an antagonistic activity of H89on L3-ARs were to exist, with a reasonably low Ki

(high a¤nity), as has been shown for the classical L-ARs, the kinetics of the L3-AR-mediated e¡ects in-vestigated here would have been markedly a¡ected,with increases in the apparent EC50 values of severalorders of magnitude (as illustrated in Fig. 2B, 3B and4C). As no such e¡ect was seen, our results demon-strate that no antagonism was exerted by H89 on L3-ARs.

Apparently, the agonist binding site on L1- and L2-ARs is su¤ciently di¡erent from that on L3-ARs toallow high a¤nity binding of H89 to the former, butnot to the latter. Of course, the marked pharmaco-logical di¡erences between L3-ARs and L1/L2-ARs,that allowed for their pharmacological distinction[42], are congruent with such a di¡erence in the bind-ing a¤nities for H89. Thus, the restricted usefulnessof H89 for the classical L-ARs does not apply to L3-ARs.

4.2. Adrenergic induction of thermogenesis andUcp1 expression is PKA-mediated

In the present study, we demonstrate that certainadrenergically induced processes (thermogenesis andUCP1 gene expression) in brown adipocytes wereinhibited by H89 and therefore conclude that theseprocesses are mediated via PKA activation.

Rather high concentrations of H89 (about 50 WM)were used in the experiments to observe this inhibi-tion. A lower dose (10 WM) was also able to inhibite.g. adrenergically induced Ucp1 expression, but theinhibitory e¡ect was not complete (not shown). Asthe mechanism of action of H89 (and other isoqui-nolinesulphonamides) is as a competitive inhibitorwith ATP on the active site of PKA [20,43], the ne-

cessity for a high concentration of H89 is what couldbe expected. The a¤nity of ATP for the catalytic siteis about 5 WM, and the Ki of H89 for this site isabout 50 nM [20,43]. Thus, even with an H89 con-centration of 50 WM, the apparent a¤nity of PKAfor ATP would only be decreased to about 5 mM(5 WMU(1+(50 WM/50 nM))). With an intracellularATP concentration in the millimolar range even thisincrease in apparent EC50 would theoretically hardlybe su¤cient to fully inhibit endogenous PKA activa-tion. Thus, it is clear that H89 concentrations in theorder of magnitude used here are necessary for PKAinhibition within intact cells (in agreement with ob-servations in other cell types [20]).

Many adrenergic e¡ects in brown adipocytes canbe mimicked by cAMP and have therefore been im-plied to be mediated by PKA, but evidence for thishas been only circumstantial. That an adrenergic ef-fect is mimicked by an agent does not necessarilyindicate that this agent is the physiological mediatorof the adrenergic stimulation, and it is now recog-nised that not all cAMP e¡ects are PKA mediated[10^19]. No evidence has been presented until nowthat establishes endogenous cAMP/PKA involve-ment in the mediation of these L3-AR e¡ects inbrown adipocytes. The demonstration here that inhi-bition of adrenergically induced PKA activity abol-ishes adrenergically induced thermogenesis and ex-pression of Ucp1 is thus the ¢rst time it isestablished that the sequential signalling pathwayfor these processes is indeed L3-AR activation,cAMP elevation and PKA activation. However, aswith any inhibitor, H89 may not be fully speci¢cfor its intended target. In this context, it was recentlydemonstrated that H89 is also an inhibitor of kinasesother that PKA, such as MSK1, ROCK-II and S6K1[44], but based on the activation pathways demon-strated for these kinases, none of these other kinasesare likely to be associated with adrenergic stimula-tion of Ucp1 expression or thermogenesis. The mostlikely conclusion of the results in this study, and fullycompatible with these results, is mediation of the ad-renergic stimulation via PKA.

The downstream targets of the activated PKA inthese cells may be of diverse nature. There may bedirect activation of enzymes, such as hormone-sensi-tive lipase, the activation of which probably results instimulation of lipolysis and thereby induces thermo-

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217 215

genesis [5,28]. There may be direct activation of tran-scription factors such as cAMP-responsive element-binding protein (CREB) [41] (Thonberg et al., un-published data) which is probably the mediator ofadrenergically induced Ucp1 expression [7,40,41].Also, further signalling cascades may be adrenergi-cally activated via PKA, such as the Src tyrosine ki-nase-Erk1/2 mitogen-activated protein (MAP) kinasecascade [39] with its pronounced e¡ects on cellulardestiny.

4.3. Conclusions

We have demonstrated here that H89 inhibits ad-renergic activation of thermogenesis and Ucp1 ex-pression in brown adipocytes and that H89 doesnot inhibit these physiologically relevant e¡ects ofadrenergic agonists via L3-AR antagonism. Rather,all data indicate that H89 functions via an inhibitorye¡ect downstream of cAMP, i.e. by inhibition ofinduced PKA activity. We thus conclude that H89may be used for the elucidation of PKA-dependentsignalling mechanisms in cell systems which entopi-cally express L3-ARs, and that PKA has indeed anessential role in brown adipose tissue function bymediating adrenergic stimulation of thermogenesisand Ucp1 expression.

Acknowledgements

This investigation was supported by a grant fromthe Swedish Natural Science Research Council.

References

[1] J. Zhao, L. Unelius, T. Bengtsson, B. Cannon, J. Neder-gaard, Coexisting L-adrenoceptor subtypes: signi¢cance forthermogenic process in brown fat cells, Am. J. Physiol. 267(1994) C969^C979.

[2] J. Zhao, B. Cannon, J. Nedergaard, K1-Adrenergic stimula-tion potentiates the thermogenic action of L3-adrenoceptor-generated cAMP in brown fat cells, J. Biol. Chem. 272(1997) 32847^32856.

[3] S. Rehnmark, M. Nechad, D. Herron, B. Cannon, J. Neder-gaard, K- and L-adrenergic induction of the expression ofthe uncoupling protein thermogenin in brown adipocytesdi¡erentiated in culture, J. Biol. Chem. 265 (1990) 16464^16471.

[4] G. Bronnikov, T. Bengtsson, L. Kramarova, V. Golozoubo-va, B. Cannon, J. Nedergaard, L1 to L3 switch in control of

cAMP during brown adipocyte development explains dis-tinct L-adrenoceptor subtype mediation of proliferationand di¡erentiation, Endocrinology 140 (1999) 4185^4197.

[5] J. Nedergaard, O. Lindberg, The brown fat cell, Int. Rev.Cytol. 74 (1982) 187^286.

[6] J.A. MacDonald, K.B. Storey, cAMP-dependent protein ki-nase from brown adipose tissue: temperature e¡ects on ki-netic properties and enzyme role in hibernating groundsquirrels, J. Comp. Physiol. 168 (1998) 513^525.

[7] P. Yubero, M.J. Barbera, R. Alvarez, O. Vinas, T. Mampel,R. Iglesias, F. Villarroya, M. Giralt, Dominant negative reg-ulation by c-Jun of transcription of the uncoupling protein-1gene through a proximal cAMP-regulatory element: a mech-anism for repressing basal and norepinephrine-induced ex-pression of the gene before brown adipocyte di¡erentiation,Mol. Endocrinol. 12 (1998) 1023^1037.

[8] C.C. Gerhardt, J. Gros, A.D. Strosberg, T. Issad, Stimula-tion of the extracellular signal-regulated kinase 1/2 pathwayby human beta-3 adrenergic receptor: new pharmacologicalpro¢le and mechanism of activation, Mol. Pharmacol. 55(1999) 255^262.

[9] K.J. Soeder, S.K. Snedden, W. Cao, G.J. Della Rocca, K.W.Daniel, L.M. Luttrell, S. Collins, The L3-adrenergic receptoractivates mitogen-activated protein kinase in adipocytesthrough a Gi-dependent mechanism, J. Biol. Chem. 274(1999) 12017^12022.

[10] Y. Shintani, Y. Marunaka, Regulation of chloride channeltra¤cking by cyclic AMP via protein kinase A-independentpathway in A6 renal epithelial cells, Biochem. Biophys. Res.Commun. 223 (1996) 234^239.

[11] X.Y. Du, S. Sorota, Modulation of dog atrial swelling-in-duced chloride current by cAMP: protein kinase A-depen-dent and -independent pathways, J. Physiol. 500 (1997) 111^122.

[12] J. de Rooij, F.J.T. Zwartcruis, M.H.G. Verheijen, R.H.Cool, S.M.B. Nijman, A. Wittinghofer, J.L. Bos, Epac is aRap1 guanine nucleotide-exchange factor directly activatedby cyclic AMP, Nature 396 (1998) 474^477.

[13] H. Kawasaki, G.M. Springett, N. Mochizuki, S. Toki, M.Nakaya, M. Matsuda, D.E. Housman, A.M. Graybiel, Afamily of cAMP-binding proteins that directly activateRap1, Science 282 (1998) 2275^2279.

[14] M.C. Broillet, S. Firestein, Cyclic nucleotide-gated channels.Molecular mechanisms of activation, Ann. NY Acad. Sci.868 (1999) 730^740.

[15] L.A. Cass, S.A. Summers, G.V. Prendergast, J.M. Backer,M.J. Birnbaum, J.L. Meinkoth, Protein kinase A-dependentand -independent signaling pathways contribute to cyclicAMP-stimulated proliferation, Mol. Cell. Biol. 19 (1999)5882^5891.

[16] N. Niisato, Y. Ito, Y. Marunaka, cAMP stimulates Na(+)transport in rat fetal pneumocyte: involvement of PTK- butnot a PKA-dependent pathway, Am. J. Physiol. 277 (1999)L727^L736.

[17] J. de Rooij, H. Rehmann, M. van Triest, R.H. Cool, A.Wittinghofer, J.L. Bos, Mechanism of regulation of the

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217216

Epac family of cAMP-dependent RapGEFs, J. Biol. Chem.275 (2000) 20829^20836.

[18] N. Pham, I. Cheglakov, C.A. Koch, C.L. de Hoog, M.F.Moran, D. Rotin, The guanine nucleotide exchange factorCNrasGEF activates ras in response to cAMP and cGMP,Curr. Biol. 10 (2000) 555^558.

[19] M.S. Shapiro, W.N. Zagotta, Structural basis for ligand se-lectivity of heteromeric olfactory cyclic nucleotide-gatedchannels, Biophys. J. 78 (2000) 2307^2320.

[20] T. Chijiwa, A. Mishima, M. Hagiwara, M. Sano, K. Hay-ashi, T. Inoue, K. Naito, T. Toshioka, H. Hidaka, Inhibitionof forskolin-induced neurite outgrowth and protein phos-phorylation by a newly synthesized selective inhibitor of cy-clic AMP-dependent protein kinase, N-[2-(p-bromocinnamy-lamino)ethyl]-5-isoquinolinesulfonamide (H-89), of PC12Dpheochromocytoma cells, J. Biol. Chem. 265 (1990) 5267^5272.

[21] R.A. Engh, A. Girod, V. Kinzel, R. Huber, D. Bossemeyer,Crystal structures of catalytic subunit of cAMP-dependentprotein kinase in complex with isoquinolinesulfonyl proteinkinase inhibitors H7, H8 and H89. Structural implications,J. Biol. Chem. 271 (1996) 26157^26164.

[22] R.B. Penn, J.L. Parent, A.N. Pronin, R.A. Panettieri Jr., J.L.Benovic, Pharmacological inhibition of protein kinases inintact cells : antagonism of L-adrenergic receptor ligandbinding by H-89 reveals limitations of usefulness, J. Pharma-col. Exp. Ther. 288 (1999) 428^437.

[23] M. Nechad, P. Kuusela, C. Carneheim, P. Bjo«rntorp, J. Ne-dergaard, B. Cannon, Development of brown fat cells inmonolayer culture. I. Morphological and biochemical dis-tinction from white fat cells in culture, Exp. Cell Res. 149(1983) 105^118.

[24] G. Bronnikov, J. Houstek, J. Nedergaard, L-Adrenergic,cAMP-mediated stimulation of proliferation of brown fatcells in primary culture. Mediation via L1 but not via L3

receptors, J. Biol. Chem. 267 (1992) 2006^2013.[25] J.M. Fredriksson, J.M. Lindquist, G.E. Bronnikov, J. Ne-

dergaard, Norepinephrine induces vascular endothelialgrowth factor gene expression in brown adipocytes througha L-adrenoreceptor/cAMP/protein kinase A pathway involv-ing Src but independently of Erk1/2, J. Biol. Chem. 275(2000) 13802^13811.

[26] K.B.E. Ohlson, N. Mohell, B. Cannon, S.G.E. Lindahl, J.Nedergaard, Thermogenesis in brown adipocytes is inhibitedby volatile anesthetic agents. A factor contributing to hypo-thermia in infants?, Anesthesiology 81 (1994) 176^183.

[27] N. Mohell, J. Nedergaard, B. Cannon, Quantitative di¡er-entiation of K- and L-adrenergic respiratory responses inisolated hamster brown fat cells : evidence for the presenceof an K1-adrenergic component, Eur. J. Pharmacol. 93(1983) 183^193.

[28] A. Matthias, K.E.B. Ohlson, J.M. Fredriksson, A. Jacobs-son, J. Nedergaard, B. Cannon, Thermogenic responses inbrown-fat cells are fully UCP1-dependent: UCP2 or UCP3do not substitute for UCP1 in adrenergically or fatty-acidinduced thermogenesis, J. Biol. Chem. 275 (2000) 25073^25081.

[29] H. Thonberg, S.-J. Zhang, P. Tvrdik, A. Jacobsson, J. Ne-dergaard, Norepinephrine utilizes K1- and L-adrenoreceptorssynergistically to maximally induce c-fos expression in brownadipocytes, J. Biol. Chem. 269 (1994) 33179^33186.

[30] J.P. Skala, B.L. Knight, Protein kinases in brown adiposetissue of developing rats, J. Biol. Chem. 252 (1977) 1064^1070.

[31] R. Jockers, T. Issad, V. Zilberfarb, P. de Coppet, S. Mar-ullo, A.D. Strosberg, Desensitization of the L-adrenergic re-sponse in human brown adipocytes, Endocrinology 139(1998) 2676^2684.

[32] J. Klein, M. Fasshauer, M. Benito, C.R. Kahn, Insulin andthe L3-adrenoceptor di¡erentially regulate uncoupling pro-tein-1 expression, Mol. Endocrinol. 14 (2000) 764^773.

[33] J.G. Granneman, Norepinephrine and BRL 37344 stimulateadenylate cyclase by di¡erent receptors in rat brown adiposetissue, J. Pharmacol. Exp. Ther. 254 (1990) 508^513.

[34] A. Chaudhry, K.N. Lahners, J.G. Granneman, Perinatalchanges in the coupling of L1- and L3-adrenergic receptorsto brown fat adenylyl cyclase, J. Pharmacol. Exp. Ther. 261(1992) 633^637.

[35] N. Mohell, A. Dicker, The L-adrenergic radioligand[3H]CGP-12177, generally classi¢ed as an antagonist, is athermogenic agonist in brown adipose tissue, Biochem. J.261 (1989) 401^405.

[36] G. Iwami, J. Kawabe, T. Ebina, P.J. Cannon, C.J. Homcy,Y. Ishikawa, Regulation of adenylyl cyclase by protein ki-nase A, J. Biol. Chem. 270 (1995) 12481^12484.

[37] Y. Chen, A. Harry, J. Li, M.J. Smit, X. Bai, R. Magnusson,J.P. Pieroni, G. Weng, R. Iyengar, Adenylyl cyclase 6 isselectively regulated by protein kinase A phosphorylationin a region involved in GKs stimulation, Proc. Natl. Acad.Sci. USA 94 (1997) 14100^14104.

[38] J. Zhao, B. Cannon, J. Nedergaard, Thermogenesis is L3-but not L1-adrenergically mediated in rat brown fat cells,even after cold acclimation, Am. J. Physiol. 275 (1998)R2002^R2011.

[39] J.M. Lindquist, J.M. Fredriksson, S. Rehnmark, B. Cannon,J. Nedergaard, L3- and K1-adrenergic Erk1/2 activation is Srcbut not Gi-mediated in brown adipocytes, J. Biol. Chem. 275(2000) 22670^22677.

[40] U.C. Kozak, J. Kopecky, J. Teisinger, S. Enerba«ck, B. Boy-er, L.P. Kozak, An upstream enhancer regulating brown-fat-speci¢c expression of the mitochondrial uncoupling proteingene, Mol. Cell. Biol. 14 (1994) 59^67.

[41] A. Chaudhry, J.G. Granneman, Di¡erential regulation offunctional responses by L-adrenergic receptor subtypes inbrown adipocytes, Am. J. Physiol. 277 (1999) R147^R153.

[42] J.R.S. Arch, The brown adipocyte L-adrenoceptor, Proc.Nutr. Soc. 48 (1989) 215^223.

[43] H. Hidaka, M. Inagaki, S. Kawamoto, Y. Sasaki, Isoquino-linesulfonamides, novel and potent inhibitors of cyclic nucle-otide dependent protein kinase and protein kinase C, Bio-chemistry 23 (1984) 5036^5041.

[44] S.P. Davies, H. Reddy, M. Caivano, P. Cohen, Speci¢cityand mechanism of action of some commonly used proteinkinase inhibitors, Biochem. J. 351 (2000) 95^105.

BBAMCR 14730 17-4-01

J.M. Fredriksson et al. / Biochimica et Biophysica Acta 1538 (2001) 206^217 217