BASIC STUDIES Identi ...Caffeine, i.e. 1,3,7-trimethylxanthine, is the most widely consumed pharmacologically active substance in the world. In North America, 90% of adults consume

BAS IC STUDIES

Identi¢cation of paraxanthine as themost potent ca¡eine-derivedinhibitor of connective tissue growth factor expression in liverparenchymal cellsOlav A. Gressner, Birgit Lahme, Monika Siluschek and Axel M. Gressner

Institute of Clinical Chemistry and Pathobiochemistry, Central Laboratory, RWTH-University Hospital, Aachen, Germany

Keywords

caffeine – connective tissue growth factor –

CTGF – liver fibrosis – paraxanthine – TGF-b –theobromine – theophylline

Abbreviations

15-PGJ2, 15-deoxy-d(12,14)-prostaglandin J2;cAMP, cyclic adenosine monophosphate;

cAMP-PDE, cAMP-specific

phosphodiesterase; CTGF, connective tissue

growth factor; HMIS, hazardous materials

identification system; HSC, hepatic stellate

cells; hydrate (Rp-diastereomer of adenosine-

30,50-cyclic monophosphothioate); PC,

hepatocytes (parenchymal cells); PPARg,peroxisome proliferator-activated receptor g;Rp-cAMP, Rp-adenosine 30,5-cyclic

monophosphorothioate triethylammonium

salt; siRNA, small interfering RNA; TGF-b,transforming growth factor b.

Correspondence

Priv. Doz. Dr med. Olav A. Gressner, Institute of

Clinical Chemistry and Pathobiochemistry,

Central Laboratory, RWTH University Hospital,

Pauwelsstraße 30, Aachen 52074, Germany.

Tel: 10049 241 808 8671

Fax: 10049 241 808 2512

e-mail: [email protected]

Received 14 November 2008

Accepted 21 December 2008

DOI:10.1111/j.1478-3231.2009.01987.x

AbstractBackground: Recently, we identified hepatocytes as the major cellular sourceof profibrogenic connective tissue growth factor (CTGF/CCN2) in the liver.Based on reports of a hepatoprotective effect of coffee consumption, we werethe first to provide evidence that caffeine suppresses transforming growthfactor (TGF)-b dependent and -independent CTGF expression in hepatocytesin vitro and in vivo, thus suggesting this xanthine-alkaloid as a potentialtherapeutic agent. Aim: This study aims at comparing the inhibitory capa-cities of caffeine and its three demethylated derivates paraxanthine, theophyl-line and theobromine on CTGF expression in hepatocytes and hepatic stellatecells (HSC). Results: Our data suggest paraxanthine as the most importantpharmacological repressor of hepatocellular CTGF expression among thecaffeine-derived metabolic methylxanthines with an inhibitory dosage (ID)50of 1.15 mM, i.e. 3.84-fold lower than what is observed for caffeine. In addition,paraxanthine displayed the least cell toxicity as proven by the water-solubletetrazolium-1 cell vitality assay. However, caffeine or any of the metabolitesdid not inhibit CTGF expression in HSC. At the toxicological thresholdconcentration of 1 mM for paraxanthine, we observed an inhibition ofhepatocellular CTGF synthesis by 44%, which was strongly reverted in thepresence of the specific competitive cyclic adenosine monophosphate inhibi-tor Rp-adenosine 30,5-cyclic monophosphorothioate triethylammonium salt.Furthermore, CTGF protein expression induced by various concentrations ofTGF-b (0.13–1 ng/ml) is still reduced by, on average, 27%/45% in the presenceof paraxanthine (1.25 mM/2.5 mM). Conclusion: Our data provide an evi-dence-based suggestion of the caffeine-derived primary metabolite para-xanthine as a potentially powerful antifibrotic drug by its inhibitory effect on(hepatocellular) CTGF synthesis.

Caffeine, i.e. 1,3,7-trimethylxanthine, is the most widelyconsumed pharmacologically active substance in theworld. In North America, 90% of adults consumecaffeine daily (1). Caffeine is metabolized in the liver bythe cytochrome P450 oxidase enzyme system (CYP1A2)into the three metabolic dimethylxanthines, i.e. para-xanthine (1,7-dimethylxanthine; 84%), theobromine(3,7-dimethylxanthine; 12%) and theophylline (1,3-di-methylxanthine; 4%) (1–4). Further, demethylation andoxidation forms urates and uracil derivatives. About adozen caffeine metabolites can be recovered in the urine,but they comprise o 3% of the ingested caffeine (1, 3).

These major caffeine derivatives have common me-chanisms of action, i.e. competitive inhibition of thecyclic adenosine monophosphate-specific phosphodies-terase (cAMP-PDE) and of adenosine receptors (parti-cularly to subtypes 1 and A2). However, the fraction bywhich any of the pathways is affected differs betweenthem.

Two large population-based surveys (National Healthand Nutrition Examination Survey I and III) as well as arecent publication by the National Institute of Diabetesand Digestive and Kidney/National Institute of Healthhave shown that higher caffeine consumption by patients

Liver International (2009)886 c� 2009 John Wiley & Sons A/S

Liver International ISSN 1478-3223

mailto:[email protected]

with chronic liver diseases is associated with a mildercourse of fibrosis (5, 6), especially in alcoholic cirrhosis(6–8), and with a decrease in alanine aminotransferase(7, 8) and g-glutamyltransferase (8) activities. However,the underlying pathophysiological mechanisms re-mained obscure.

Connective tissue growth factor (CTGF/CCN2) iswidely regarded as a central modulator of the activitiesof the profibrogenic master cytokine transforminggrowth factor (TGF)-b. Previously, we could show thathepatocytes [parenchymal cells (PC)] are the majorcellular source of CTGF in the liver, that CTGF isspontaneously upregulated in the hepatocyte under con-ditions of cellular stress and that it strongly responds toTGF-b stimulation (9, 10). Even though CTGF is foundto be overexpressed in fibrotic lesions, and in vivointoxication of the liver in the rat leads to an increase inhepatocellular CTGF in immunohistochemical staining(9), its exact pathophysiological role in regard to fibro-genic processes of the liver still needs to be defined (11).However, it was shown that silencing of CTGF by smallinterfering RNA (siRNA) almost entirely inhibits fibroticremodelling of livers in mice previously subjected tohepatotoxic agents (12, 13).

In a recent study, we provided evidence that caffeinesuppresses TGF-b-dependent and -independent CTGFexpression in hepatocytes in vitro and in livers subjectedto D-galactosamine toxicity in vivo via a mechanism thatinvolves upregulation of the nuclear receptor peroxisomeproliferator-activated receptor (PPAR) g and thus sensi-tization towards the natural PPARg ligand 15-PGJ2, aswell as the enhanced degradation and inhibition ofphosphorylation of the TGF-b effector Smads 2 and 3(14, 15).

Based on these findings, our current study aims atcomparing the inhibitory capacities of caffeine and itsthree hypomethylated derivatives paraxanthine, theo-phylline and theobromine on CTGF expression in hepa-tocytes and hepatic stellate cells (HSC). The results mayhelp initiate pioneer future investigations for the evalua-tion of a caffeine-derived therapeutic agent in the treat-ment of fibrotic liver diseases.

Materials and methods

Materials

Caffeine (1,3,7-trimethylxanthine; #27600), paraxanthine(1,7-dimethylxanthine; #41761), theobromine (3,7-di-methylxanthine; #88304), theophylline (1,3-dimethyl-xanthine; #88308) and Rp-cAMP (Rp-adenosine 30,50-cyclic monophosphorothioate triethylammonium salthydrate; #A165) were from Sigma-Aldrich (St Louis,MO, USA). The three-dimensional shapes and structuresof caffeine and its hypomethylated metabolites used inthis study are shown in Figure 1. Recombinant humanTGFb-1 (240-B) was from R&D Systems (Minneapolis,MN, USA).

A goat polyclonal anti-CTGF/CCN2 (L-20, sc-14939;Santa Cruz, Santa Cruz, CA, USA), epitope mappingwithin an internal region of CTGF of human origin, wasused for Western blotting analysis.

Animals

Adult Sprague–Dawley rats (body weight 200–250 g) hadfree access to a standard laboratory chow diet and normaltap water throughout the experimental period. All ani-mals received care and treatment in compliance with theGerman Animal Protection Act, which is in accordancewith the German Research Council’s criteria.

Isolation and culture of primary rat hepatic stellate cells

Hepatic stellate cells were isolated by the pronase–colla-genase perfusion technique (16), purified by a single-stepdensity gradient centrifugation with Nycodenz (Nye-gaard, Oslo, Norway) as described (17) and characterizedby their typical light microscopic appearance, immunos-taining for desmin and vimentin and vitamin A-specificautofluorescence. Cell viability checked by the trypanblue exclusion test was normally 4 95%, and the meanpurity was 90� 5%. The cells were cultured for 4 days inserum-free Dulbecco’s modified Eagle medium (DMEM;BioWhittaker Europe, Verviers, Belgium), supplementedwith 4 mM L-glutamine (PAA Laboratories, Coelbe,Germany), 100 IU/ml penicillin and 100 mg/ml strepto-mycin (PAA Laboratories) under a humidified atmo-sphere of 5% CO2/95% O2.

Cell culture and preparation of rat hepatocytes

Parenchymal cells were isolated from male Sprague–-Dawley rats (180–320 g body weight) by the two-stepcollagenase method of Seglen (16) modified as describedbefore (18). Cells were isolated and cultured in thecomplete absence of (TGF-b-containing) foetal calf ser-um (FCS). The viability of the final parenchymal cellsuspension, checked by trypan blue exclusion, wasaround 85%, and cell recovery was 20� 50� 107 cells/liver. Contamination with other non-parenchymal cellswas o 2%.

Cells were seeded in 2 ml GIBCOTM HepatoZYME-SFMTM (Invitrogen, Carlsbad, CA, USA) without FCS ontype I collagen (BD Bioscience, Clontech, Palo Alto, CA,USA)-coated plastic dishes (Becton Dickinson, FranklinLakes, NJ, USA) with a density of 5.4� 104 cells/cm2.They were cultured at 37 1C in a humidified atmosphereof 5% CO2 and 95% air supplemented with 4 mML-glutamine, penicillin (100 IU/ml) and streptomycin(100 mg/ml) (all from Cambrex, East Rutherford, NJ,USA). The first change of the medium was 1 h afterseeding and unattached PC were washed off with DMEM(Cambrex). PCs were then cultured for various times inGIBCOTM HepatoZYME-SFMTM, supplemented as de-scribed above.

Liver International (2009)c� 2009 John Wiley & Sons A/S 887

Gressner et al. Paraxanthine as the most potent caffeine-derived inhibitor of CTGF expression

Joseph GeorgeRectangle

Sodium dodecyl sulphate – polyacrylamide gelelectrophoresis and Western blotting

Preparations of cytoplasmatic cell extracts, determina-tion of protein concentrations and Western blot analysiswere performed as described previously (10, 19). Densi-tometric quantification of the blot results was performedusing the Lumi-Imager System (Roche Diagnostics,Mannheim, Germany) and the LUMIANALYST 3.0 software(Roche Diagnostics).

Reverse transcriptase-polymerase chain reaction for ratconnective tissue growth factor

Total cellular RNA was extracted with the Qiagen RNeasypurification kit (Qiagen, Hilden, Germany). cDNA wasreverse transcribed using the First-Strand cDNA synth-esis kit (Invitrogen). Reverse transcriptase-polymerasechain reaction (RT-PCR) was performed using the Bio-metra T3000 Thermocycler PCR System (Biometra,Göttingen, Germany) and the following primers: rCTGF

Fig. 1. Two-dimensional (2D) and 3D shapes and structures of caffeine and its derived metabolic methylxanthines used in this study. In 3Dshapes, atoms are colour coded: C, grey; N, dark blue; O, red; H, white.


Paraxanthine as the most potent caffeine-derived inhibitor of CTGF expression Gressner et al.

(forward: 50-CGT CCA TGC TGC ACA GG-30; reverse:50-CAA GTT TGA GCT TTC TGG-30). The size of theamplified fragment was 144 bp.

In general, 1ml cDNA in 14.5 ml of RNAse-free water(Roche Diagnostics) and 5ml 5� Taq Master PCREnhancer (Eppendorf, Hamburg, Germany) were mixedwith 0.5 ml dNTPs (Qiagen), 2.5 ml of 10� PCR buffer/15 mM MgCl2 (Qiagen), 0.5 ml of 10 mM forward primer,0.5 ml of 10 mM reverse primer and 0.5 ml of Taq poly-merase (1 U/ml; Roche Diagnostics). The following PCRconditions were used: 95 1C/4 min, 95 1C/1 min, followedby 40 cycles at 95 1C/15 s and 58 1C/1 min, 72 1C/1.5 minand 72 1C/10 min, and stopped at 4 1C. PCR productswere separated on 4% agarose gels and visualized underUV light (MWG Biotech AG, Ebersberg, Germany).

Generation of recombinant connective tissue growthfactor reporter adenovirus (Ad-hCTGF-Luc)

For generation of the reporter adenovirus Ad-hCTGF-Luc, the �2.5-kbp ClaI fragment of vector pGL3-Basic-hCTGF-Luc (10) harbouring a fusion of human CTGFgene promoter and the luciferase reporter gene wascloned into the ClaI site of vector pDE1sp1A (20),resulting in the generation of pDE1sp1A-hCTGF-Luc.The integrity of cloning boundaries was verified byrestriction analysis and sequencing with the flankingprimers 50-GCG TAA CCG AGT AAG AAT TTG-30 and50–GGC GAC CAT CAA TGC TGG AG-30 that wereobtained from MWG-Biotech AG.

The integration of the reporter cassette from pDE1-sp1A-hCTGF-Luc into the adenoviral backbone vectorpJM17 (21) was performed by in vitro homologousrecombination in the human embryo kidney cell line293 using a protocol described before (22). Successfulgeneration of recombinant viral particles was visualizedby viral foci formation. After total infection, the viralparticles were released from cells by three rounds of afreeze–thaw cycle and separated from cell debris bycentrifugation at 600� g for 10 min. To generate high-titre viral stocks, 293 cells were reinfected at a multi-plicity of infection of 1 and grown for 3–4 days. Theamplified viruses were harvested, concentrated throughstandard CsCl gradient centrifugation and subsequentlypurified using the BD Adeno-XTM Purification Filtersystem (BD Biosciences) according to the manufacturer’sinstructions.

Luciferase gene reporter assay

Cells were cultured in black 96-well plates and infectedwith 1� 108 virons/ml of Ad-hCTGF-luc reporter virus.After specific treatment, the Luciferase activity wasmeasured as described previously (1).

Water-soluble tetrazolium-1 cell vitality assay

To quantify cell proliferation, the water-soluble tetrazo-lium (WST)-1 assay using a sulphonated tetrazolium

salt, 4-(3-[4-ioophenyl]-2-[phenyl-2H-5-tetrazolio-1,3-benzene disulphonate]) (Roche Diagnostics), was per-formed as recommended by the manufacturer.

Analysis of data

Where applicable, SPSS 16.0 (SPSS, Chicago, IL, USA)was used for statistical analysis, applying two-tailedunpaired Student’s t tests for comparison of means witha P-value for significance set at least at 0.05. Correlationsbetween variables were analysed with the Pearson corre-lation test. Again, values of Po 0.05 were consideredstatistically significant. Furthermore, we performed re-gression analysis to assess associations between CTGFprotein expression, promoter activation or cAMP values,and methylxanthine or TGF-b concentrations.

Molecular visualization of caffeine and its metabolicdimethylxanthine derivates was performed using themolecular visualization system ChemDraw Ultra 8.0 andChem3D Ultra 8.0 (CambridgeSoft, Cambridge, MA,USA).

Results

Dose-dependent effects of caffeine and its derivedmetabolites paraxanthine, theophylline and theobromineon connective tissue growth factor protein expression andon the transcriptional activation of the connective tissuegrowth factor promotor in rat hepatocytes and hepaticstellate cells

Treatment of primary rat PC with caffeine confirmed thepreviously observed inhibitory effect of the CTGF pro-moter and CTGF protein accumulation on transcrip-tional activation (Fig. 2) (14). The estimated inhibitorydosage (ID)50 was set at 6.40 mM for transcriptionalactivation (Fig. 2A) and at 4.42 mM for protein expres-sion (Fig. 2B).

All primary metabolic caffeine derivates, para-xanthine, theophylline and theobromine, also exerted aninhibitory effect on CTGF protein expression (Fig. 3A)and promoter activity (Fig. 3B), although dose depen-dency and degree of inhibition differed among them. Thecalculated ID50 was set at 1.15 mM for paraxanthine,2.46 mM for theophylline and 1.48 mM for theobrominefor CTGF protein expression (Fig. 3A) and at 0.82 mMfor paraxanthine, 4.72 mM for theophylline and 1.95 mMfor theobromine for CTGF promoter activity (Fig. 3B).

Using Pearson’s correlation and simple regressionanalysis for the association of mean CTGF proteinexpression and dosage of each methylxanthine, we calcu-lated correlation coefficients of � 0.78 for caffeine,� 0.62 for paraxanthine, � 0.77 for theophylline and� 0.87 for theobromine. Coefficients of determination(R2) were 0.61 for caffeine, 0.39 for paraxanthine, 0.59for theophylline and 0.76 for theobromine. In general,the metabolic derivatives proved to be more powerfulrepressors of hepatocellular CTGF expression than caf-feine itself. Paraxanthine, in particular, exerted a highly



suppressive effect on CTGF synthesis. Table 1 sum-marizes the concentrations of the compounds necessaryto inhibit CTGF protein expression and transcriptionalactivity by 25% (ID25), 50% (ID50) and 75% (ID75).

In order to exclude hepatocyte toxicity as a possiblecause of the observed reduction in hepatocellular CTGFexpression in the presence of caffeine or its metabolites, aWST-1 assay was performed (Fig. 3C), and the

Fig. 2. Dose dependency of the inhibitory activities of caffeine against hepatocellular connective tissue growth factor (CTGF) promoter activityand protein expression. (A) CTGF/CCN2 reporter gene activation. Parenchymal cells (PC) cultured under serum-free conditions for 24 h weretransfected with Ad-hCTGF-Luc and subjected to different concentrations of caffeine. Results are given as Ad-hCTGF-Luc in per cent of theblank value (% of 0). Given are mean values of triplicate determinations from three different cultures. The above diagram displays mean valuesof triplicate determinations from three different cultures� standard deviation. Below, a range diagram, regression line and individual 95%confidence intervals for the correlation of Ad-hCTGF-Luc activity and caffeine concentrations are given. Circles represent mean values oftriplicate determinations from one culture. Green, red and blue lines indicate the concentrations of caffeine necessary to achieve a 25 (ID25), 50(ID50) or 75% (ID75) inhibition of CTGF promoter activation. (B) CTGF/CCN2 protein expression. Western blots of CTGF of PC cultured underserum-free conditions with indicated concentrations of caffeine. Analysis was as described in ‘‘Materials and methods’’. Blots were quantifiedrelative to b-actin using the Lumi Imager System. The above diagram displays mean CTGF protein expression [CTGF/b-actin (BLU)] of triplicatedeterminations from at least four different cultures. A representative blot is demonstrated. In the center of the figure, a range diagram,regression line and individual 95% confidence interval for the correlation of CTGF protein expression and caffeine concentrations are given.Circles represent mean values of triplicate determinations from one culture. Green, red and blue lines indicate the concentrations of caffeinenecessary to achieve a 25 (ID25), 50 (ID50) or 75% (ID75) inhibition of CTGF protein expression. (C) Effects of caffeine on CTGF/CCN2 proteinexpression and mRNA synthesis in primary rat hepatic stellate cells. Left: CTGF/CCN2 protein expression. Hepatic stellate cells (HSC) werecultured under serum-free conditions for 4 days. Some cells were treated with caffeine (5 mM) in the absence or the presence of transforminggrowth factor-b (1 ng/ml) and investigated for the expression of CTGF using Western blot analysis. Analysis was as described in ‘‘Materials andmethods’’. b-actin served as the loading control. One representative experiment out of three is shown. Right: reverse transcriptase-polymerasechain reaction (RT-PCR) for CTGF/CCN2 in HSC. RT-PCR was performed using primers for rat CTGF as described in ‘‘Materials and methods’’.The size of the amplified CTGF fragment was 144 bp. One representative result out of three is shown.



morphological development of PC was controlled underthe microscope (Fig. 3D). Of all the metabolites tested(including caffeine), paraxanthine displayed the leasttoxicity towards cultured PC.

Next, we tested the possibility of transferring theresults of a repressive effect of caffeine on CTGF synthesisobtained in PC to a primary rat HSC. However, caffeinedid not inhibit CTGF expression in HSC, either at thetranscriptional level, as demonstrated by RT-PCR(Fig. 2C, right), or at the protein level (Fig. 2C, left).Similar results were found for the primary metabolitesof caffeine (data not shown). These findings are notsurprising as we could show previously that caffeineexerts its inhibitory effect on CTGF expression primarilythrough an interruption of TGF-b-induced Smad2/3signalling to the CTGF promoter (14), whereas theregulation of CTGF expression in HSC is largely TGF-bindependent (10).

Definition of the inhibitory potential of caffeine,paraxanthine, theophylline and theobromine onconnective tissue growth factor expression induced byvarious concentrations of transforming growth factor-b

Transforming growth factor-b is widely regarded as theprofibrogenic master cytokine and major inducer ofhepatocellular CTGF expression. We therefore investi-gated the effect of caffeine, paraxanthine, theophyllineand theobromine on CTGF expression induced by var-ious concentrations of recombinant TGF-b (Fig. 4A–D).On average, 1 ng/ml TGF-b increased CTGF expressionby about 2.3-fold. Overall, the means of CTGF proteinexpression induced by various concentrations of TGF-b(0.13–1 ng/ml) were reduced, on average, by 30%/46% inthe presence of caffeine (2.5 mM/5 mM) (Fig. 4A), by27%/45% in the presence of paraxanthine (1.25 mM/2.5 mM) (Fig. 4B), by 11%/36% in the presence oftheophylline (1.25 mM/2.5 mM) (Fig. 4C) and by 16%/30% in the presence of theobromine (1.25 mM/2.5 mM)(Fig. 4D). Therefore, unmetabolized caffeine was identi-fied as the most potent methylxanthine in terms ofinhibiting TGF-b-dependent CTGF expression, closelyfollowed by paraxanthine. The reducing effect of theo-phylline or theobromine on TGF-b-induced stimulationof CTGF expression in hepatocytes was not consideredsignificant, as the relative magnitude of increase causedby TGF-b remained unchanged between untreated cellsand those treated with theophylline or theobromine.

Efficiency of the specific protein kinase A inhibitor,Rp-adenosine 30,5-cyclic monophosphorothioatetriethylammonium salt, in reverting the repressive effectsof caffeine, paraxanthine, theophylline and theobromineon connective tissue growth factor synthesis

As described above and shown in Figures 2–4, all threemetabolites and caffeine itself reduced hepatocellularCTGF production with varying efficacies despite their

similar mechanisms of action. This suggests that compe-titive inhibition of the cAMP-PDE may be unequallyinvolved in mediating the repressive signals of thesemethylxanthines to the CTGF promoter.

Therefore, it was tempting to define the role of cAMPin mediating the metabolite-specific inhibition of hepa-tocellular CTGF expression by using the specific compe-titive protein kinase A inhibitor Rp-cAMP. Rp-cAMPhad no significant effect on the basal level of CTGFsynthesis (Fig. 5A). However, rat PC subjected to caffeinedisplayed a striking and highly significant reversal ofcaffeine effects on CTGF expression when previouslytreated with Rp-cAMP (Po 0.0001; average 140% in-hibition of caffeine effects). In paraxanthine-treatedcells, Rp-cAMP exerted its inhibitory effect even moreprofoundly (Po 0.0001; 229%). In theophylline-treatedcells, Rp-cAMP caused an average 108% inhibition of themaximal inhibitory effect of theophylline (Po 0.0001),whereas the Rp-cAMP effect was again much morepronounced in theobromine-treated PC, with an average207% inhibition of theobromine-repressed CTGF synth-esis (Po 0.0001). Results were confirmed by the CTGFluciferase reporter assay, suggesting that the cAMP path-way acts at the transcriptional level (Fig. 5B).

Effects of caffeine, theophylline and theobromine onparaxanthine-induced repression of hepatocellularconnective tissue growth factor expression

Comparative analysis of caffeine and its derived metabo-lites identified paraxanthine as the most potent inhibitorof CTGF expression in PCs (Figs 2 and 3).

In order to enhance the efficiency of this methyl-xanthine, we looked for a possible synergism betweenparaxanthine and the other metabolites as well as caffeinein terms of CTGF repression at concentrations close tothe respective ID50 determined previously. As seen inFigure 6, the effects of paraxanthine and the othermethylxanthines on CTGF production were additive,and additional inhibition was calculated at, on average,37% for caffeine, 41% for theophylline and 39% fortheobromine. Thus, even though the combination ofparaxanthine and theophylline exerted a highly suppres-sive effect on CTGF synthesis, the degree of inhibitiondid not differ significantly and to a pathobiochemicallyrelevant amount from the other combinations.

Discussion

The present investigation is based on our previousfinding that caffeine, similar to 8-Br-cAMP, stronglydownmodulates TGF-b-induced CTGF expressionin hepatocytes in vitro and in livers subjected toD-galactosamine toxicity in vivo by stimulation of degra-dation of the TGF-b effector Smad2, inhibition of Smad3phosphorylation and upregulation of the PPARg-recep-tor (14, 15). This is of note, as experimental silencing of



CTGF by siRNA was shown to inhibit experimental liverfibrosis in mice (12, 13). Here, we provide evidence thatnot just caffeine itself but also its metabolic methyl-xanthines suppress CTGF expression, and performedintensive studies for comparing caffeine and its metabo-lites in terms of their pharmacological potency andmechanisms of action leading to CTGF suppression.

However, when discussing the data demonstrated inthis study, it should be considered that caffeine applied athigher concentrations displays significant cell toxicity, so

that the observed effect of caffeine on hepatocellularCTGF expression might, at least in part, be attributed tothis cytotoxic effect. This also holds for theophylline.

More interestingly, our data suggest that paraxanthineis the most powerful pharmacological repressor of hepa-tocellular CTGF expression among the drug family ofmethylxanthine derivatives that does not display signifi-cant cytotoxicity at the tested concentrations.

The members of this caffeine-derived family ofxanthine alkaloids exert their effects primarily through



two distinct mechanisms, i.e. (i) inhibition of the cAMP-PDE and/or (ii) competitive antagonism of adenosinereceptors (particularly subtypes 1 and 2A) (23). A stronginhibition of cAMP-PDE has been proven for all meta-bolites, apart from theophylline, which in contrast,compared with its family members, displays a higheraffinity to adenosine receptors (24–33).

Because methylxanthine derivatives are antagonists,not agonists, at adenosine receptors, one would expectno pharmacological effects unless the receptor is beingactivated by endogenous adenosine. It was reported thatendogenous adenosine release from damaged liver cellsplays an important role in the pathogenesis of toxic liverfibrosis, but the intracellular events responsible were notidentified (34). Other reports, however, provided evi-dence that adenosine is able to reverse a pre-establishedCCl4-induced micronodular cirrhosis in rats (35). Thisdiverse effect of adenosine on hepatic fibrogenesis may inpart be explained by the differential effect of activated

adenosine receptors on intracellular cAMP levels. Studiesin isolated rat hepatocytes demonstrated that the con-centration of cAMP increased in response to stimulationof type 2A adenosine receptors, while it decreased inresponse to stimulation of type 1 receptors (36, 37). Thisis explained by the fact that the type 2A adenosinereceptor is linked to a Gs-protein, which stimulatesadenylatecyclase activity whereas the type 1 adenosinereceptor is coupled to a Gi-protein, which inhibits it (36,37). Therefore, binding of caffeine to the type 2Areceptor may result in a negative effect of this methyl-xanthine on cAMP levels by stabilizing an ‘antagonistconformation’ of adenosine receptors (28). As may bededucted from this functional diversity, blockade of type1 and 2A receptors appears to be the less likely mechan-ism for the CTGF inhibitory action of methylxanthines,compared with competitive inhibition of the cAMP-PDE. This is also supported by the fact that theophylline,a pan-specific adenosine receptor antagonist, but only aslight inhibitor of cAMP-PDE (38), has the weakesteffects on CTGF expression in rat PC, which are further-more hardly reverted by Rp-cAMP.

The apparent beneficial effect of methylxanthines inpreventing fibrotic remodelling of the liver should beconsidered as a ray of hope in terms of developing drug-based antifibrotic therapy approaches. After all, the ID50value for caffeine was 4.42 mM. Initially, this findingsuggests that habitual coffee consumption might besufficient in order to gain a healthful effect. However,one cup of coffee at 120 mg caffeine results in bloodconcentrations of 0.015 mM (1). Therefore, largeamounts of coffee would be required to be consumed inorder to achieve plasma concentrations of 4.42 mM.

Alternatively, the lethal dose of caffeine for humans,when applied intravenously, is set at about 13 g (27).With a volume of distribution in humans of 42 L, thisresults in a blood concentration of 1.6 mM; therefore, a

Fig. 3. Dose dependency of the inhibitory activities of the caffeine-derived metabolic methylxanthines paraxanthine (left column),theophylline (middle column) and theobromine (right column) against hepatocellular connective tissue growth factor (CTGF) proteinexpression and promoter activity. (A) Parenchymal cells (PC) were cultured under serum-free conditions with indicated concentrations ofparaxanthine, theophylline and theobromine, and Western blot analysis was performed as described in ‘‘Materials and methods’’. Blots werequantified relative to b-actin using the Lumi Imager System. The diagrams display the mean CTGF protein expression [CTGF/b-actin (BLU)] oftriplicate determinations from four to five different cultures and are described as a fraction of the untreated control (%). A representative blotfor each is demonstrated. Below, range diagrams, regression lines and individual 95% confidence intervals for the correlation of CTGF proteinexpression and paraxanthine, theophylline or theobromine concentrations are given. Circles represent mean values of three independentexperiments from one culture. Green, red and blue lines indicate the concentrations of the metabolites necessary to achieve a 25 (ID25), 50(ID50) or 75% (ID75) inhibition of CTGF protein expression. (B) CTGF/CCN2 reporter gene activation. PC cultured under serum-free conditionsfor 24 h were transfected with Ad-hCTGF-Luc and subjected to paraxanthine, theophylline or theobromine at indicated concentrations. Resultsare given as Ad-hCTGF-Luc activity in per cent of the blank value (% of 0). Given are mean values of triplicate determinations from threedifferent cultures. The above diagram displays mean values of triplicate determinations from three different cultures. Below, a range diagram,regression line and individual 95% confidence interval for the correlation of Ad-hCTGF-Luc activity and caffeine concentrations are given.Circles represent mean values of triplicate determinations from one culture. Green, red and blue lines indicate the concentrations of caffeinenecessary to achieve a 25 (ID25), 50 (ID50) or 75% (ID75) inhibition of CTGF promoter activation. (C) Water-soluble tetrazolium (WST)-1 cellviability assay of PCs treated with indicated concentrations of caffeine or its metabolites. PCs were plated in six-well plates, treated withcaffeine or its respective primary metabolites at indicated concentrations and allowed to grow for an additional 24 h before performing theWST-1 assay. Measurements are the averages of n = 4 independent samples� standard deviation. (D) Morphological development of PCstreated with indicated concentrations of caffeine or its metabolites. Shown are light microscopic images (magnification: 40-fold) of PCs treatedwith caffeine (5 mM), theophylline (5 mM), paraxanthine (2.5 mM) or theobromine (2.5 mM) respectively.

Table 1. Estimated inhibitory dosages (ID) of caffeine and its derivedmetabolic methylxanthines against hepatocellular connective tissuegrowth factor protein expression (BLU/b-actin) and promoter activity(hCTGF-Luc)

ID25 ID50 ID75

Caffeine (mM) BLU/b-actin 1.80 4.42 8.00hCTGF-Luc 3.19 6.40 10.15

Paraxanthine (mM) BLU/b-actin 0.34 1.15 2.00hCTGF-Luc 0.23 0.82 1.65

Theophylline (mM) BLU/b-actin o0.01 2.46 4.66hCTGF-Luc 0.54 4.72 9.18

Theobromine (mM) BLU/b-actin 0.53 1.48 2.41hCTGF-Luc 0.45 1.95 4 2.5

Incubation of PCs with the methylxanthines at 37 1C and 5% CO2 for 24 hwas performed as described in ‘‘Materials and methods’’. Means of

triplicate determinations from at least four independent cultures are given.



concentration of 4.42 mM would be a lethal dose. How-ever, caffeine is efficiently metabolized in PCs in vivo.Thus, absorbed caffeine is concentrated in the liver, inparticular in PCs, resulting in increased liver-specific

concentrations that do not necessarily reflect those levelsobserved in the circulation. Still, the lower the ID50needed for an inhibition of hepatocellular CTGF synth-esis, the better.

Fig. 4. Inhibitory potential of caffeine, paraxanthine, theophylline and theobromine on connective tissue growth factor (CTGF) expressioninduced by various concentrations of transforming growth factor (TGF)-b. Rat parenchymal cells were treated for 24 h with recombinanthuman TGF-b1 at the indicated doses added to the culture medium in the absence or the presence of 2.5 mM/5 mM caffeine (A), 1.5 mM/2.5 mM paraxanthine (B), 1.5 mM/2.5 mM theophylline (C) or 1.5 mM/2.5 mM theobromine (D), respectively, and then harvested, and Westernblot analysis was performed as described in ‘‘Materials and methods’’. Blots were quantified relative to b-actin using the Lumi Imager System.The diagrams display mean CTGF protein expression [CTGF/b-actin (BLU)] of triplicate determinations from four different cultures and aredescribed as a fraction of the untreated control (%)� standard deviation. A representative blot for each is demonstrated. CTGF proteinexpression is plotted against TGF-b concentrations. The continuous line indicates TGF-b-treated controls, while the dashed line indicatesTGF-b-stimulated cells treated with 2.5 mM caffeine or 1.5 mM paraxanthine, theophylline or theobromine respectively. The dotted lineindicates TGF-b-stimulated cells treated with 5 mM caffeine or 2.5 mM paraxanthine, theophylline or theobromine respectively.



Paraxanthine, which with an ID50 of only 1.15 mM,was identified as the most potent inhibitor of CTGFsynthesis in liver parenchymal cells, is currently believedto exhibit a lower toxicity than caffeine, and blood levelscommensurating with average intake appear to be fairlyinnocuous. The health hazard class according to the USHazardous Materials Information Systems III (HMISs

III) (39) and US National Fire Protection Associationclassification (40) is set at 1 for paraxanthine andtheobromine whereas theophylline and caffeine are clas-sified as 2. However, among paraxanthine and theobro-mine, the first displays much more attenuated symptomsfollowing exposure. Chronic exposure to paraxanthine atconcentrations of 1–4 mM has so far been only weaklyassociated with increased mutagenicity in humans andteratogenicity in mice (39–41), whereas chronic exposureto the same concentrations of theobromine led to muchstronger mutagenicity in humans, mice and hamsters, toreproductive disabilities in rats and mice as well as toteratogenicity in mice following both oral and intraper-itoneal application (39, 40). Also, theobromine is rated asa Group 3 carcinogen of the International Agency forResearch of Cancer classification (42). Of note, ourstudies also confirmed paraxanthine as the least harmfulcaffeine-derived metabolite in terms of hepatocyte toxi-city in vitro.

Finally, considering the above-mentioned toxico-logical threshold concentration of 1 mM for these

Fig. 5. Protein kinase A inhibition of parenchymal cells (PC) treatedwith caffeine and its derived metabolic methylxanthinesparaxanthine, theophylline and theobromine by Rp-adenosine 30,5-cyclic monophosphorothioate triethylammonium salt (Rp-cAMP).(A) Connective tissue growth factor (CTGF)/CCN2 proteinexpression. PCs were treated with 10 mM Rp-cAMP, and 2 h later,5 mM caffeine, 2.5 mM paraxanthine, 2.5 mM theophylline or2.5 mM theobromine, respectively, were added to the culturemedium. Cells were cultured for another 24 h, and Western blotanalysis was performed as described in ‘‘Materials and methods’’.Blots were quantified relative to b-actin using the Lumi ImagerSystem. The diagram displays CTGF protein expression [CTGF/b-actin (BLU); described as a fraction of the untreated control (%)]of triplicate determinations from three different cultures� standarddeviation. A representative blot is demonstrated. �Significant tocontrol (Po0.05). (B) CTGF/CCN2 reporter gene activation. PCscultured under serum-free conditions for 24 h were transfected withAd-hCTGF-Luc and subjected to different concentrations of caffeine,or its primary metabolites paraxanthine, theophylline andtheobromine. Results are demonstrated as Ad-hCTGF-Luc in per centof the blank value (% of 0). Given are mean values of triplicatedeterminations from three different cultures� standard deviation.�Significant to control (Po 0.05).

Fig. 6. Inhibitory potential of caffeine, paraxanthine, theophyllineand theobromine and their combination on hepatocellularconnective tissue growth factor (CTGF) expression. Parenchymalcells were cultured under serum-free conditions with indicatedconcentrations of paraxanthine, theophylline and theobromine, andWestern blot analysis was performed as described in ‘‘Materials andmethods’’. Blots were quantified relative to b-actin using the LumiImager System. The diagram displays CTGF protein expression[CTGF/b-actin (BLU)]; described as a fraction of the untreatedcontrol (%)] of triplicate determinations from four different cultures.A representative blot is demonstrated. �Significant to control(Po0.05).



metabolites, we observe a 44% inhibition of hepatocel-lular CTGF synthesis for paraxanthine, whereas theobro-mine achieves only 38% inhibition.

In conclusion, our data are the first to provide anevidence-based suggestion of the caffeine-derived meta-bolite paraxanthine as a potentially powerful antifibroticdrug by its inhibitory effect on (hepatocellular) CTGFsynthesis. This proposal is based on earlier reports thatdemonstrate that silencing of CTGF by siRNA preventsfibrotic remodelling of murine livers previously subjectedto hepatotoxic agents (12, 13). Therefore, future studiesare urgently recommended to evaluate the significance ofa therapeutic paraxanthine application in patients withchronic liver diseases and hepatic fibrogenesis.

Acknowledgements

We gratefully acknowledge the support of this study bythe Deutsche Forschungsgemeinschaft (DFG) throughgrant # GR 463/14-1. We thank Prof. R. Weiskirchen forproviding the reporter gene construct hCTGF-Luc. Wethank K. Rehbein for expert technical assistance.

References

1. Barone JJ, Roberts HR. Caffeine consumption. Food ChemToxicol 1996; 34: 119–29.

2. Gates S, Miners JO. Cytochrome P450 isoform selectivity inhuman hepatic theobromine metabolism. Br J Clin Phar-macol 1999; 47: 299–305.

3. Ullrich D, Compagnone D, Munch B, et al. Urinary caffeinemetabolites in man. Age-dependent changes and pattern invarious clinical situations. Eur J Clin Pharmacol 1992; 43:167–72.

4. Roberts EA, Furuya KN, Tang BK, Kalow W. Caffeinebiotransformation in human hepatocyte lines derived fromnormal liver tissue. Biochem Biophys Res Commun 1994;201: 559–66.

5. Modi AA, Feld JJ, Park Y, et al. Higher caffeine consump-tion is associated with milder fibrosis in patients withchronic liver diseases. Hepatology 2007; 46: 835A.

6. Ruhl CE, Everhart JE. Coffee and tea consumption areassociated with a lower incidence of chronic liver disease inthe United States. Gastroenterology 2005; 129: 1928–36.

7. Ruhl CE, Everhart JE. Coffee and caffeine consumptionreduce the risk of elevated serum alanine aminotransferaseactivity in the United States. Gastroenterology 2005; 128:24–32.

8. Tanaka K, Tokunaga S, Kono S, et al. Coffee consumptionand decreased serum gamma-glutamyltransferase and ami-notransferase activities among male alcohol drinkers. Int JEpidemiol 1998; 27: 438–43.

9. Weng HL, Ciuclan L, Liu Y, et al. Profibrogenic transform-ing growth factor-beta/activin receptor-like kinase 5 signal-ing via connective tissue growth factor expression inhepatocytes. Hepatology 2007; 46: 1257–70.

10. Gressner OA, Lahme B, Demirci I, Gressner AM, Weis-

kirchen R. Differential effects of TGF-beta on connective

tissue growth factor (CTGF/CCN2) expression in hepatic

stellate cells and hepatocytes. J Hepatol 2007; 47: 699–710.11. Moussad EE, Brigstock DR. Connective tissue growth

factor: what’s in a name? Mol Genet Metab 2000; 71:

276–92.12. Li G, Xie Q, Shi Y, et al. Inhibition of connective tissue

growth factor by siRNA prevents liver fibrosis in rats. J Gene

Med 2006; 8: 889–900.13. George J, Tsutsumi M. siRNA-mediated knockdown of

connective tissue growth factor prevents N-nitrosodi-

methylamine-induced hepatic fibrosis in rats. Gene Ther

2007; 14: 790–803.14. Gressner OA, Lahme B, Rehbein K, et al. Pharmacological

application of caffeine inhibits TGF-beta-stimulated con-

nective tissue growth factor expression in hepatocytes via

PPARgamma and Smad2/3-dependent pathways. J Hepatol

2008; 49: 758–67.15. Gressner OA, Siluschek M, Lahme B, Gressner AM. In-

traperitoneal application of caffeine prevents D-galactosa-

mine induced hepatic expression of connective tissue

growth factor (CTGF/CCN2) in the rat. J Hepatol 2009, in

press.16. Seglen P. Preparation of isolated rat liver cells. In: Prescott

DM, ed. Methods in Cell Biology. New York: Academic,

1987; 29–83.17. Schafer S, Zerbe O, Gressner AM. The synthesis of proteo-

glycans in fat-storing cells of rat liver. Hepatology 1987; 7:

680–7.18. Gressner AM, Pfeiffer T. Preventive effects of acute inflam-

mation on liver cell necrosis and inhibition of heparan

sulphate synthesis in hepatocytes. J Clin Chem Clin Bio-

chem 1986; 24: 821–9.19. Gressner AM, Wulbrand U. Variation of immunocyto-

chemical expression of transforming growth factor (TGF)-

beta in hepatocytes in culture and liver slices. Cell Tissue Res

1997; 287: 143–52.20. Bett AJ, Haddara W, Prevec L, Graham FL. An efficient and

flexible system for construction of adenovirus vectors with

insertions or deletions in early regions 1 and 3. Proc Natl

Acad Sci USA 1994; 91: 8802–6.21. Mcgrory WJ, Bautista DS, Graham FL. A simple technique

for the rescue of early region I mutations into infectious

human adenovirus type 5. Virology 1988; 163: 614–7.22. Weiskirchen R, Kneifel J, Weiskirchen S, et al. Comparative

evaluation of gene delivery devices in primary cultures of

rat hepatic stellate cells and rat myofibroblasts. BMC Cell

Biol 2000; 1: 4.23. Klotz KN, Hessling J, Hegler J, et al. Comparative pharma-

cology of human adenosine receptor subtypes – character-

ization of stably transfected receptors in CHO cells.

Naunyn-Schmiedebergs Arch Pharmacol 1998; 357: 1–9.24. Jafari M, Rabbani A. Studies on the mechanism of caffeine

action in alveolar macrophages: caffeine elevates cyclic

adenosine monophosphate level and prostaglandin synth-

esis. Metabolism: Clin Exp 2004; 53: 687–92.



Joseph GeorgeRectangle

25. Fisone G, Borgkvist A, Usiello A. Caffeine as a psychomotorstimulant: mechanism of action. Cell Mol Life Sci 2004; 61:857–72.

26. Tarka SM Jr The toxicology of cocoa and methylxanthines:a review of the literature. Crit Rev Toxicol 1982; 9: 275–312.

27. Stavric B. Methylxanthines: toxicity to humans. 2. Caffeine.Food Chem Toxicol 1988; 26: 645–62.

28. Daly JW. Mechanism of action of caffeine. In: Garattini S,ed. Caffeine, Coffee and Health. New York: Raven Press,1993; 97–150.

29. Lelo A, Birkett DJ, Robson RA, Miners JO. Comparativepharmacokinetics of caffeine and its primary demethylatedmetabolites paraxanthine, theobromine and theophyllinein man. Br J Clin Pharmacol 1986; 22: 177–82.

30. Timson J. Theobromine and theophylline. Mutat Res 1975;32: 169–78.

31. Charytan D, Jansen K. Severe metabolic complicationsfrom theophylline intoxication. Nephrology 2003; 8: 239–42.

32. Stavric B. Methylxanthines: toxicity to humans. 1. Theo-phylline. Food Chem Toxicol 1988; 26: 541–65.

33. Stavric B. Methylxanthines: toxicity to humans. 3. Theo-bromine, paraxanthine and the combined effects of methyl-xanthines. Food Chem Toxicol 1988; 26: 725–33.

34. Chan ES, Montesinos MC, Fernandez P, et al. AdenosineA(2A) receptors play a role in the pathogenesis of hepaticcirrhosis. Br J Pharmacol 2006; 148: 1144–55.

35. Hernandez-Munoz R, Diaz-Munoz M, Suarez-Cuenca JA,et al. Adenosine reverses a preestablished CCl4-induced

micronodular cirrhosis through enhancing collagenolyticactivity and stimulating hepatocyte cell proliferation inrats. Hepatology 2001; 34: 677–87.

36. Guinzberg R, Uribe S, Diaz-Cruz A, Hernandez Cruz A,Pina E. In rat hepatocytes, different adenosine receptorsubtypes use different secondary messengers to increase therate of ureagenesis. Life Sci 2006; 79: 382–90.

37. Gonzalez-Benitez E, Guinzberg R, Diaz-Cruz A, Pina E.Regulation of glycogen metabolism in hepatocytes throughadenosine receptors. Role of Ca21 and cAMP. Eur JPharmacol 2002; 437: 105–11.

38. Ito K, Lim S, Caramori G, et al. A molecular mechanism ofaction of theophylline: induction of histone deacetylaseactivity to decrease inflammatory gene expression. ProcNatl Acad Sci USA 2002; 99: 8921–6.

39. National Paint and Coatings Association. Hazardous Ma-terials Information Systems III (HMISs III). 2008, cited;available from: http://www.paint.org/hmis/index.cfm (lastaccessed 12 January 2008)

40. National Fire Protection Association. Hazardous MaterialsClassification. 2008, cited; available from: http://www.nfpa.org/index.asp (last accessed 12 January 2008)

41. York RG, Randall JL, Scott WJ Jr Teratogenicity of para-xanthine (1,7-dimethylxanthine) in C57BL/6J mice. Tera-tology 1986; 34: 279–82.

42. International Agency for Research on Cancer. Carcinogeni-city Classification. 2008, cited; Available from: http://www.iarc.fr/en (last accessed 12 January 2008)


www.paint.org/hmis/index.cfmwww.paint.org/hmis/index.cfmwww.nfpa.org/index.aspwww.nfpa.org/index.aspwww.iarc.fr/en

Documents

BASIC STUDIES Identi ...Caffeine, i.e. 1,3,7-trimethylxanthine, is the most widely consumed pharmacologically active substance in the world. In North America, 90% of adults consume