Research Paper Mediators of Inflammation, 8, 219–227 (1999)

Cystic fibrosis (CF) is caused by mutations in the CF gene, which encodes CF transmembrane conductance regulator protein (CFTR), a transmembrane protein that acts as a cAMP-regulated chloride channel The disease is characterized by inflammation but the relationship between inflammation, abnormal transepithelial ion transport, and the clinical manifestations of CF are uncertain. The present study was undertaken to determine whether three nonsteroidal anti-inflammatory drugs (NSAIDs) (aspirin, ibuprofen, and indomethacin) modulate CFTR gene expression in T-84 cells. Treatment with NSAIDs reduced CFTR transcripts, and decreased cAMP-stimulated anion fluxes, an index of CFTR function. However, the two phenomena occurred at different concentrations of both drugs. The results indicate that NSAIDs can regulate both CFTR gene expression and the function of CFTR-related chloride transport, and suggest that NSAIDs act via multiple transduction pathways.


Introduction
Inflammation in re sponse to an ex te rnal attac k or inte rnal malfunc tion trigge rs the synthe sis of many ex tra-and intrac e llular mediators that can modulate the ac tivity of the ce ll's me mbrane , c ytoplasm and nuc leus. The inhibitor y e ffec ts of aspirin on most inflammator y re spons es have long bee n attrib ute d to its ability to block the synthe sis of prostaglandins . These natural autac oṏds produc ed by cycloox yge nase activatio n provoke inflammation. 1 It is now w e ll establis hed that aspirin and the othe r nonste roidal anti-inflammatory drugs (NSAIDs) also modulate me mbrane [2][3][4][5][6] and nuc lear 7-9 re sponse s inde pendently of the inhibitio n of c ycloox ygenas e. This rais es questions about the mole cular re actions contrib uting to the be ne fic ial e ffec ts of NSAIDs on the symptoms of c ystic fibros is (CF). 10 CF trans me mbrane c onduc tanc e re gulator prote in (CFTR) is know n to ac t as a c AMP-activate d anio nic channe l. Clinic al manife statio ns of CF are c orre late d w ith mutatio ns in the CF gene , w hich cause the prote in it make s to func tion abnormally. 11 How ever, CF invo lves not only hydroelec trolytic abnormalitie s, but also chronic inflammation and infe ction, and antiinflammator y the rapie s improve the clinic al condition of patie nts. The link betw ee n change s in anio nic trans me mbrane trans port and inflammatio n is unc ertain, and some obse rvatio ns suggest that NSAIDs may have a dual effe ct in CF, c ombining inhibition of inflammation w ith the modulation of anion transme mbrane trans port.
Several effe cts of rap id modulatio ns in ionic conduc tanc e s by NSAIDs have been de scribe d. For ex ample, the fenamate s (niflumic and flufenamic acids) block the nonse lective ion channe ls in rat ex ocrine panc re as, 3 or ac tivate the voltage -depe nde nt K c urre nt ex pre sse d in Xenopus oocyte s. 6 Se veral re ports indic ate that NSAIDs alte r chloride conduc tanc e s in epithelial ce lls. Short applicatio ns of NSAIDs inhibit Ca 2+ -and cAMP-re gulate d chloride se cre tion of culture d trac heal epithe lial cells. 4 The fenamate s also modulate chloride conduc tanc e in bovine re tinal pigment epithe lium, but in a more complex w ay. 5 Re ce ntly, it w as re porte d that ibup rofen blocks the CFTR-mediate d chloride sec re tion in T-84 cells and in human and mouse trac heal epithelia. 2 Other data suggest that aspirin and othe r NSAIDs might modulate CFTR gene ex pre ssion by modulating vario us trans c rip tion fac tors w hich bind to specific nuc leotidic sequenc es p re se nt in the CFTR promote r. 12 Aspirin has been show n to inhibit the activatio n of trans cription fac tors such as NFk B 7 and AP 1 8 , and to fac ilitate the binding c ap ac ity of the Heat Shock Factor, HSF1. 9 Indome thac in has also be en show n to inhibit the activatio n of AP-1. 8 We sugge ste d that aspirin may control CFTR gene ex pre ssion. This w ould not be the first 'nuc le ar effe ct' of the drug, sinc e aspirin had alre ady be en show n to suppre ss inflammation-induc e d ex pre ssion of inte rleukin-1 (Il-1), Il-6 and adhesion molecules in Hela c ells 7 and to inhibit the synthesis of IL-1-induc ed PGH synthas e. 13 The pre sent study w as unde rtake n to dete rmine w hether aspirin, ibuprofe n, and indome thac in modify CFTR gene func tion. We did this by measuring CFTR trans c ripts and cAMP-stimulate d anio n flux e s, the se being indic e s of CFTR func tion in T-84 cells.

Cell culture and drug treatment
Human T-84 c arc inoma ce lls w ere obtaine d from the Americ an Type Culture Colle ction (Rockville , MD) and propagate d in a mix ture (1:1) of Dube lc co's modifie d Eagle 's me dium (DMEM) and Ham's F-12 me dium containing 15 mM Hepe s and 10% fetal calf se rum (FCS). LLCPK c ells stably trans fec te d w ith w ild type CFTR, D F 508 -CFTR, and nontrans fec te d c ells w ere both c ulture d in DMEM medium.
The cells w ere seeded at a dens ity of 23 10 4 /cm 2 . Stock solutions of aspirin (1 mol/l), indo methac in (53 10 -3 mol/l), and ibuprofen (10 -2 mol/l) w ere made up in e thano l. On the third day afte r p assage, the ce lls w e re tre ate d w ith as pirin or the other NSAIDs in FCS-e nriched medium. RNA w as analyze d afte r 24 h of tre atme nt (ex cep t for kine tic studie s). Functional studie s w ere pe rformed on T-84 cells inc ubate d w ith the drugs for 48 h. Control cells re c eive d the same volume of solve nt (<1%). All the culture mate rials w ere obtaine d from Life Te chnologie s (Les Ulis, Franc e); the drugs w e re from Sigma (Franc e ).

Analysis of cell viability
In a pre liminary se rie s of ex p erime nts w e verifie d that, even at the highe st conce ntrations used in this study, the drugs w e re not tox ic. Ce lls maintaine d unde r contro l conditions , or afte r a 48-h tre atme nt w ith asp irin (10 -3 mol/l), indo me thac in (23 10 -5 mol/ l), or ibuprofen (53 10 -4 mol/l) w e re w ashed w ith phosp hate -buffe re d saline (PBS), trypsinize d, suspende d in serum-fre e medium, and ce ntrifuge d. Ce lls w ere w ashed tw ice more w ith PBS and suspende d in PBS. Parts of this suspension w ere used to te st ce ll viability w ith trypan blue, for counting ce lls in a Malas sez cell, and to measure prote in c onc entratio n according to Low ry e t a l. 14 Ce ll counting and tryp an blue ex clusion te st show ed that w he n 10 -3 mol/l aspirin w as adde d to T-84 c ells for a 48-h inc ubation, it ne ithe r kille d them nor alte re d their p rolife ration rate . The same tre atme nt also had no effe ct on the prote in conte nt of the culture . The same re sults w ere observe d afte r 48-h inc ubatio n w ith 23 10 -5 mol/l indomethac in and 53 10 -4 mol/l ibup rofen (Table 1).

RNA extraction and analysis
Total RNA w as isolate d w ith phe nol/chloroform 15 using the Triz ol re age nt (Life Te chnologie s), ac cording to the manufac ture r's instruc tions . The RNA w as then frac tionate d on 0.9% agaros e ge ls (15 m g/w ell), transferre d to nylon me mbrane s (Prome ga Charbonniè re s, Franc e ) and fix ed by heating . The filte rs w ere hybridize d to 32 P-labe le d c DNA probes (specific activ ity >10 9 cpm/mg) w ith the Quik Hyb protoc ol provide d by Stratage ne (Ozyme, Le s Ulis, Franc e), w ashe d unde r stringe nt c onditio ns (0.1 SSC, 0.1% sodium docdecyl sulfate at 52°C for 20 min) and autoradio graphed at -80°C. The CFTR probe w as the 1.5-kb Eco R1-Eco R1 fragment of human CFTR-cDNA probe labeled by random priming . The me mbrane s w ere re hybridiz ed w ith a human b -ac tin cDNA probe from Onc ogene Scie nce (Franc e Biochem, Meudon, Franc e ). The mRNAs w ere quantifie d by de nsitome tric sc anning of the autoradio grams on an ImageMaste r D. To n de lie r et al.
Mediators of Inflammation · Vol 8 · 1999 VSD (Pharmac ia-Biote ch-Amersham, Orsay, Franc e), and CFTR mRNA amounts w ere normalize d to those of b -actin. All ex pe riments w ere re p eate d at least four time s.

6-Methoxy-N-ethylquinolinium fluorescence assay
The 6-methox y-N-ethylquinolinium (MEQ) w as synthesized in the laborato ry according to the method desc ribed by Biw e rsi and Verkman. 16 (Fig. 1). The p lots obtaine d using the same solutions suppleme nte d w ith the diffe re nt NSAIDs (asp irin, indome thac in, ibuprofe n) did not modify the Ste rn -Volmer c onstant (Fig. 1).
The me asure ments of intrac ellular MEQ fluore sce nce change w e re carrie d out as p re viously de scribed. 17 Brie fly, T-84 ce lls subc ulture d on glass slide s (contro l c ells, and those tre ate d for 48 h w ith NSAIDs) w ere w ashed from c ulture medium, then loade d w ith MEQ and Ifor 10 min in hyp otonic iodine solution (1:2 dilutio n of the is otonic solution, pH 7.4, adjuste d w ith NaOH, containing the follow ing re agents: 138 mmol/l NaI, 2.4 mmol/l K 2 HPO 4 , 10 mmol/l He pes, 1 mmol/l CaCl 2 , 10 mmol/l glucose ) and allow e d to re c ove r for 10 min in isotonic Isolution before be ing plac ed in a p erfusion chamber on the stage of an inve rte d mic ros cope (Diaphot, Nikon, Franc e ), w he re the y w ere c ontinuous ly perfuse d at 37°C w ith isotonic iodide solution. Afte r a 2 min perfusion pe riod, Iions w ere re p lace d by NO 3 -ions. Bec ause nitrate does not inte rac t w ith MEQ, fluore sce nce inc re as es as ce ll iodide flow s from the ce ll through anio n pathw ays in the plasma membrane , unmasking possible basal anio n conduc tanc e . Changes in fluore sce nc e unde r bas al c onditio ns, and w ith the cAMP c ocktail (53 10 -4 mol/l 8-(4-chlorophenylthio)-c AMP (cpt-cAMP) and 10 -4 mol/l IBMX) in the nitrate perfusion solution w e re re corde d. The w hole ex pe rime nt w as performed w ithout NSAIDs in the perfusion solutions . Fluore sce nce from the MEQ w as measure d in single c ells w ith a digital imaging syste m and a CDD c ame ra (Photonic s Sc ie nce , UK), and the re sults w ere analyze d using Imstar softw are (Paris , Franc e ). The initial rate of inc re as e of MEQ fluore sce nce w as measure d in basal (D F basal /D t ) and stimulato ry (D F c AMP /D t) conditio ns. 17 To validate the MEQ te chnology as a me thod for inve stigating CFTR gene func tion, w e use d LLCPK ce lls stably trans fe cte d w ith the w ild-typ e or the D F508-CFTR-mutate d human cDNA CFTR gene . 18 The ce lls w ere the n assaye d w ith the MEQ te chnology alre ady described. Figure 2 show s that the func tion of CFTR gene induc es an inc re as e in the rate of fluore sce nce afte r the additio n to the p erfusate of the cAMP c ocktail in the CFTR-LLCPK c ells. No cAMPdepende nt fluore sce nce change s w ere de te c te d in LLCPK nontrans fe cte d ce lls or D F508-CFTR-LLCPK ce lls. Thus, the MEQ as say c an be take n as an index of CFTR ac tivity.

Statistical analysis
Statis tic al analys is w as by unpaire d Stude nt's t-te st, and, w he re appropriate , by analys is of varianc e , w ith P < 0.05 c onside re d statis tic ally signific ant.

Variation in CFTR mRNA levels
The modulatio n of CFTR gene ex pre ssion by NSAIDs w as inve stigate d by Northe rn blot analys is, pe rforme d on prolife rating ce lls.
The re lative amount of CFTR gene products (normaliz ed to b -actin products) w as firs t dete rmine d in ce lls tre ate d for 24 h w ith as pirin, ibup rofe n and indomethac in. Figure 3A and B show s that, unde r our ex pe rime ntal conditions , aspirin (53 10 -4 mol/l), ibuprofe n (53 10 -4 mol/l), and indo methac in (23 10 -5 mol/l) re duce d the amount of CFTR mRNA in T-84 cells.
Afte r tre ating the c ells for 24 h w ith aspirin, the amount of CFTR mRNA w as unalte re d by low conc entratio ns of the drug, and w as de cre ased by the conc entratio ns large r than 10 -4 mol/l (Fig. 4A). A half-D. To n de lie r et al. max imal inhibitio n w as obse rve d w ith asp irin conce ntratio ns close to 10 -3 mol/l. b -ac tin trans c rip ts w ere unc hange d. Figure 4B show s the ac tion of aspirin in re latio n to time , w ith the c onc entratio n 23 10 -3 mol/l (corre sp onding to an 80% inhibitio n of CFTR trans cripts afte r 24 h of tre atme nt) chose n for these ex pe rime nts . Afte r 6 h of tre atme nt, a 50% inhibition w as observe d, and the effect w as greate st afte r 9 h. Togethe r, the data show that the CFTR dow nre gulatio n is conc entratio n and time depe nde nt. Indomethac in (10 -5 and 23 10 -5 mol/l) and ibuprofen (10 -4 and 53 10 -4 mol/l) also signific antly de cre ase d CFTR mRNA le vels in a c onc entratio n-depe nde nt manne r (Fig. 5). The ce ll conte nt in b -actin mRNA w as not alte re d by indo me thac in or ibuprofe n at any conc entratio n (re sult not show n). A re latio nship be tw e en the anti-inflammatory effects of aspirin, indo methac in, and ibuprofen, and their ability to dec re as e CFTR mRNA might proc ee d from the ir capacity to inhibit c ycloox ygenase ac tivity. To te st this hypothe sis, ex oge nous prostaglandin E 2 (PGE 2 ) w as adde d to the ce lls at the same time as the NSAIDs. As show n in Fig. 6, 10 -6 mol/l ex ogenous PGE 2 by itse lf de cre ase d ce ll CFTR mRNA conte nt, and w hen adde d w ith as pirin, indo methac in, and ibuprofe n, it did not pre vent their dow n-re gulating effect. The re fore , the de cre ase in CFTR gene ex p re ssion induc e d by the NSAIDs canno t be linke d to the inhibition of c ycloox ygenase ac tivity.

Functional studies: MEQ assay
To dete rmine w hethe r the NSAIDs modulate d the func tion of CFTR prote in, w e pe rforme d a MEQ assay, in the absenc e of the drugs , on T-84 c ells tre ate d for 24 and 48 h w ith 53 10 -4 mol/l aspirin, 53 10 -4 mol/l ibuprofe n, and 23 10 -5 mol/l indo methac in ( Table 2). The re sults w ere c ompare d w ith  Table 2). This points to activation of the c AMP-re gulate d anio n pathw ay(s) (Fig. 7).  Table 2, row 'aspirin', comparing columns 2 and 5, or columns 3 and 6) and dec re as ed in the c ells tre ate d for 48 h w ith 53 10 -4 mol/l ibuprofe n ( Fig. 7B and Table 2, row 'ibuprofe n', c omparing columns 2 and 5, or 3 and 6) or 23 10 -5 mol/l indomethac in (Table 2, row 'indo methac in', comparing column 2 and 5, or 3 and 6).
Using the MEQ methodology, w e have thus show n a de cre ase in c AMP-induc e d anio nic flux p rovoke d by the tre atme nt of the c ells w ith aspirin, indo me thac in, and ibuprofen, and also de monstrate d a partic ular action of ibuprofe n and aspirin on T-84 c ell basal anion conduc tanc e . Unde r our ex p erime ntal protocol, this basal anio n c onduc tanc e (e vide nc ed by the light emission provoke d by re placing Iby NO 3 in the supe rfusing solution) w as small (as Fig. 7A and Table   D. To n de lie r et al. 2, row 'aspirin' , column 1). Tre atme nt of the c ells w ith the thre e te ste d NSAIDs modulate d this parame te r in vario us w ays (c olumn 4, Table 2). Aspirin dec re as ed it tw ofold (columns 4 and 7, Table 2), indomethac in w as w ithout effe ct ( Table 2, row 'indo methac in', columns  4 and 7), and ibuprofen inc re as ed it tw ofold ( Figure  7B and Table 2, row 'ibuprofe n', columns 4 and 7).

Discussion
The pre sent study show s that aspirin, indo methac in, and ibuprofe n, thre e nonste roidal anti-inflammatory drugs w ide ly used in clinic al medic ine , de cre ase both the amount of CFTR trans c ripts and the func tion of CFTR prote in in human T-84 ce lls.
The dec re ase in CFTR mRNA does not app ear to be tox ic , be cause there w as no sign of death in c ells inc ubate d for 48 h w ith the highest conc entratio ns of the drugs use d, and be caus e the amount of b -ac tin mRNA did not change in paralle l w ith that of CFTR mRNA. The high conc entratio ns of aspirin and other drugs re quire d to dow n-re gulate CFTR ge ne ex p re ssion raise tw o is sues: first, the possible ex is te nc e of such an effect during anti-inflammator y tre atme nt; se cond, the re latio nship be tw ee n the modulatio n of CFTR ge ne ex pre ssion and cycloox yge nase inhibitio n. Plasma c onc entratio ns of as pirin as high as those re quire d to modulate CFTR gene ex pre ssion in our ex pe rimental mode l (ove r 10 -4 mol/l) have be en re porte d w he n the drug is use d for sustaine d periods, 19 so our re sults may have clinic al re le vanc e. How e ver, the c onc entrations of aspirin, indo methac in, or ibuprofen re quire d to re duc e CFTR gene ex pre ssion in T-84 c ells are not c ons iste nt w ith the dire ct implicatio n of cycloox yge nase (COX) inhibitio n, w hether of the c onstitutive COX-1 or the induc ible COX-2. Aspirin, indome thac in, and ibuprofen are bette r inhibitors of COX-1 than of COX-2, and thus have a highe r ID 50 for COX-2 than for COX-1; but e ven their re porte d ID 50 for COX-2 20 is at least 10-fold low e r than the ir active c once ntrations in the p re sent study. Furthe rmore , the failure of ex oge nous PGE 2 to pre ve nt the de cre ase in CFTR gene ex pre ssion c ause d by aspirin, indome thac in, and ibuprofen sugge sts that this effect of the NSAIDs doe s not invo lve the inhibitio n of prostaglandin synthe sis. The NSAID conc entratio ns used in the p re sent study are c omparable w ith those used in re ce nt studie s of the effects of these drugs on the nuc leus. This is the c as e for indo methac in binding to the perox ysome prolife rato r-activate d re ce ptor g (PPARg ). 21 High conce ntrations of as pirin are also re quire d to modulate the effect of the trans cription fac tors such as NFk B, AP1, or HSF, 7-9 w hich posse ss spe cific binding site s in the CFTR promote r, 12 and may thus be invo lve d in the NSAID-induc ed dow n-re gulation of CFTR ge ne ex pre ssion. Further studie s w ill be re quire d to de monstrate the ex iste nce of any such re gulatio n of CFTR gene ex pre ssion by aspirin and its derivative s. Sinc e the trans cription fac tors affe cte d by aspirin are activate d during inflammation, this putative me chanis m of ne gative modulation of CFTR ex p re ssion w ould also re sult from an anti-inflammatory effe ct of the NSAIDs.
The c AMP-stimulate d anionic e fflux measure d by the MEQ assay w as also decre as ed by tre ating the T-84 ce lls w ith aspirin, indomethac in, or ibuprofen. Sinc e the ex perime nts w e re pe rforme d w ithout NSAIDs, this effect of the NSAIDs diffe re d from the immediate inhibition of the ionic trans port desc ribed in other studie s. [2][3][4][5][6] The pre se nt data demons trate that the NSAIDs also ex ert a de laye d action on anio nic flux es. The re sults show ing diffe re nc es in the c onc entratio n depende nce of c AMP-re gulate d anion flux es and in the modulation of CFTR mRNA produc tion argue agains t a dire ct c ausal re latio nship be tw een the tw o e ffec ts of NSAIDs. Tw o points may be made . First, the re lative inhibition of c AMP-trigge re d anio n e fflux is large r than the re lative de cre ase in CFTR trans cripts. This w as partic ularly obvious in ce lls tre ate d w ith 53 10 -4 mol/l aspirin, w hich no longe r re sponde d to c AMP in te rm of anionic efflux , despite a dec re ase of only about 30% in their CFTR mRNA conte nt. Se cond, very few CFTR channe ls see m to be ne ce ssary for this func tion. 11 In normal tis sues, this small number of channe ls is corre late d w ith a quantity of CFTR ge ne produc ts too small to be visualiz ed by Northern blotting . Our re sults there fore sugge st that aspirin, ibuprofen and indo methac in have long-acting effects on CFTR func tion, in additio n to their action on CFTR gene ex pre ssion. The NSAIDs may alte r the cytoplasmic proc essing and/or turn-ove r of the CFTR prote in, or change its pote ntial for ac tivation. The se proc esses are modulate d by variatio ns in te mpe rature and by vario us c ompounds (butyrate , glyc erol) (a re vie w 22 ), and the demonstration of the ir sensitivity to NSAIDs w ould be partic ularly inte re sting in the cystic fibrosis c ontex t. Immunoblots of CFTR prote in in c ontrol cells and c ells ex p ose d for 48 h to 0.5 mmol/l as pirin, inc ubate d w ith radiolabeled methionine the n inc ubate d for 2-30 h show e d no clear alte ration in the time-course of the CFTR prote in glycosylation or de gradatio n, but only a gene ral de cre ase in the inte nsitie s of immature and D. To n de lie r et al.

6
Mediators of Inflammation · Vol 8 · 1999 fully glycosylate d CFTR prote in bands . The latte r w ere very faint in NSAID-tre ate d c ells (re sult not show n). These re sults sugge st that the re is an aspirin-induc e d decre as e in the synthesis of native CFTR prote in, but furthe r studie s are re quire d to spe cify the action of aspirin on CFTR prote in proc essing.
A de cre ase in endogenous CFTR activato rs c ause d by NSAIDs canno t be due to alte re d produc tion of intrac e llular e ndogenous cAMP be cause the MEQ assay w as performed w ith ex ogenous perme ant c pt-cAMP. Othe r c ytoplasmic fac tors that contro l CFTR func tion may also be affe c te d by the NSAIDs. For ex ample, indomethac in ac ts as an irre ve rsible inhibitor of c AMP-dep ende nt p rote in kinas e in rabbit ile al mucosa, 23 and inhibits phospholipase A2 in rabbit polymorphonuc lear le ukocyte s. 24 Such e ffec ts could alte r CFTR func tion dire ctly or indire ctly, but are unlike ly sinc e the re w ere no change s in cAMPinduc ed MEQ fluore scenc e in ce lls inc ubate d for only 24 h w ith NSAIDs.
How ever, the use of the MEQ methodology did re veal that ibuprofen also appe ars to induc e a ne w cAMP-inde pe nde nt anio n c onduc tanc e, w hich c an be activate d in T-84 c ells w ithout furthe r stimulatio n. This effect may c orre spond to an inc re ase d numbe r and/or func tion of anothe r chloride channe l, and thus contrib ute to the improve d clinic al status of CF patie nts tre ate d w ith ibuprofe n.
The pre se nt re sults e mphasize the possible dive rsity of the NSAID e ffec ts, sinc e the se drugs, w hich inhibit CFTR ac tivity ac cute ly, 2 -6 also appear to produce a de laye d de cre ase in CFTR ge ne ex pre ssion and cAMP-stimulate d anionic e fflux . The delaye d decre as e in CFTR prote in func tion doe s not ap pear to re sult from a dec re as e in gene ex pre ssion alone . The NSAIDs thus appe ar to have a broad spe ctrum of action on CFTR, w hich may re fle ct a gene ral modific ation of ce ll metabolism. This suggest that CFTR acts as a 'housekeeping ge ne ' that c an adap t to vario us physiological or pharmac ological disturbanc e s. 25