Differential expression of prostanoid receptors in hepatocytes, Kupffer cells, sinusoidal endothelial cells and stellate cells of rat liver

Journal of Hepatology 1999; 30:38~-7 Printed in Denmark . All rights reserved Munksgaard. Copenhagen

Copyright © European Association /br the Stud)' of the Liver 1999 Journal of Hepatology ISSN 0168-8278

Differential expression of prostanoid receptors in hepatocytes, Kupffer cells, sinusoidal endothelial cells and stellate cells of rat liver Alexandra Fennekohl, Henrike L. Schieferdecker, Kurt Jungermann and Gerhard P. P~schel hlstitut ffir Biochemie und Molekulare Zellbiologie, G6ttingen, Germany

Background~Aims: Prostanoids produced by nonparenchymal cells modulate the function of parenchymal and nonparenchymal liver cells during homeostasis and inflammation via eight classes of prostanoid receptors coupled to different G-proteins. Prostanoid receptor expression in parenchymal and nonparenchymal cells was studied in order to get a better insight into the complex prostanoid-mediated intrahepatic signaling network. Methods: RNA was isolated from freshly purified parenchymal and nonparenchymal rat liver cells and the mRNA level of all eight prostanoid receptor classes was determined by newly developed semiquantitative reverse transcription-polymerase chain reaction protocols. Results: The mRNAs for the prostanoid receptors were differentially expressed. Hepatocytes were the only cell type which contained the mRNA of the Gqlinked prostaglandin F2a receptor; they were devoid of any mRNA for the Gs-linked prostanoid receptors. Kupffer cells possessed the largest amount of mRNA for the Gs-linked prostaglandin Ez receptor subtype 2. Endothelial cells expressed high levels of mRNA for the Gq-linked thromboxane receptor and medium levels of mRNA for the G~-linked prostacyclin recep-

RAT LIVERthe five major prostanoids, prostaglandin I N(PG) D2, E2, F2~, PGI2 (prostacyclin) and thromboxane A2, are produced by nonparenchymal sinusoidal cells which comprise Kupffer cells, endothelial cells and stellate cells. They are synthesized and released in response to a variety of physiological and pathophysiological stimuli, i.e. glucagon (1), noradrenaline (2), nucleoReceived 17 March; revised 28 July; accepted 11 August 1998

Correspondence: Gerhard P. POschel, Institut for Biochemic und Molekulare Zellbiologie, Humboldtallee 23, D-37073 GOttingen, Germany. Fax: +49 551 395960. E-mail: [email protected] 38

tot, while stellate cells had the highest levels of mRNA for the prostacyclin receptor. The mRNAs for the Gqlinked prostaglandin E2 receptor subtype 1 and the Gi-linked prostaglandin Ea receptor subtype 3 were expressed in hepatocytes and all nonparenchymai cell types at similar high levels, whereas the mRNA of the Gs-linked prostaglandin Dz receptor was expressed in all nonparenchymal cells at very low levels. Conclusions: In hepatocytes the prostaglandin F2a receptor can mediate an increase in glucose output via an increase of intracellular InsP3 while cAMP-dependent glucose output can be inhibited via the subtype 3 prostaglandin Ez receptor. The subtype 2 prostaglandin Ez receptor can restrain the inflammatory response of Kupffer cells via an increase in intracellular cAME The thromboxane receptor and the prostacyclin receptor in sinusoidal endothelial and the prostacyclin receptor in stellate cells may be involved in the regulation of sinusoidai blood flow and filtration.

Key words: Gene expression; Hepatic stellate cell; Hepatocyte; Kupffer cell; Prostacyclin receptor; Prostaglandin receptor; Sinusoidal endothelial cell; Thromboxane receptor.

tides and nucleosides (3-5), endotoxin (6), immune complexes (7), complement peptides (8-10) and cytokines (11,12). Prostaglandins E2 and Fza can modulate hepatocyte metabolism by paracrine intercellular communication between nonparenchymal and parenchymal cells (13). The prostaglandins, prostacyclin and thromboxane also control the function of nonparenchymal cells in both an autocrine (14) and a paracrine (15) mode. Prostanoids transmit their signal via G-proteincoupled receptors (GPCR). Eight prostanoid receptors, one each for PGD2, PGF2~, PGI2 and TXA2 and four for PGE2, have been identified by pharmacological approaches. These have been classified according to their

Differential expression of hepatic prostanoid receptors TABLE 1 Classification and G protein coupling of prostanoid receptors Name

Natural ligand

G-Protein

Second messenger

DP-R EP1-R EP2-R EP3-R EP4-R FP-R IP-R TP-R

PGD 2 PGEz PGE 2 PGE2 PGE2 PGF2~ prostacyclin thromboxane A2

Gs Gq Gs Gi Gs Gq G~ Gq

cAMP']" InsP3"[" cAMPS cAMP,I, cAMP1" InsP3"l" cAMPq" InsP31"

affinity to natural and synthetic ligands and according to their coupling to different G-proteins (Table 1) (16). Understanding the complex prostanoid-mediated intrahepatic intercellular communication depends on knowing the distribution of the different prostanoid receptors on different types of liver cells. The prostaglandin E2 and F2a receptors that mediate the increase in glucose output from hepatocytes (17) and the decrease in glucose output from hormone-stimulated hepatocytes (18-20) have been characterized by binding studies (21), functional studies involving receptor specific synthetic ligands (22) and recently by molecular cloning (23,24). By contrast, prostanoid receptors on nonparenchymal cells have been detected in functional studies only. For example, prostaglandin E 2 receptors have been shown to mediate a decrease of endotoxinstimulated TNFa formation in Kupffer cells (25), thromboxane A2 receptors on sinusoidal endothelial cells have been implicated in the development of hepatic

reperfusion injury (26), while IP receptors on precontracted hepatic stellate cells seem to mediate their relaxation in response to prostacyclin (15). However, so far, prostanoid receptors on nonparenchymal liver cells have not been characterized by binding studies or on the molecular level. Despite much research worldwide, antibodies that would allow receptor distribution to be studied on the protein level are not available for most prostanoid receptors. Binding studies with receptor specific ligands are not sufficiently selective to discriminate between different prostanoid receptors that are frequently present on the same cell. However, the cloning of all eight pharmacologically defined prostanoid receptors from different organs and species (27,28) opened up the possibility of studying their distribution in different cell types of the hepatic sinusoid at the mRNA level. In the current study the differential expression of prostanoid receptors in hepatocytes and nonparenchymal liver cells was analyzed by semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR).

Materials and Methods Materials All chemicals were analytical grade and from commercial sources.

Primers Reverse transcription was primed with oligo(dT)12_ls (Pharmacia, Freiburg, Germany). The PCR-primers were custom synthesized by Pharmacia (Freiburg, Germany) or NAPS (G6ttingen, Germany). The sequences are shown in Table 2. Primer positions which were located at similar relative positions to the start codon of the respective

TABLE 2 Sequence of the PCR-primers employed to amplify prostanoid receptor and B-actin cDNAs Name

Sequence

Position

Source

Acc. No.

flactinf flactinr DPf DPr EPlf EP 1r EP2f EP2r EP3f EP3r EP4f EP4r FPf FPr IPf IPr TPf TPr

GATATCGCTGCGCTCGTCGTC CCTCGGGGCATCGGAACC

1251-1271 2553-2570 467485 824-847 663-692 1305-1324 498-517 1095-1117 2747 834-855 731-753 1366-1389 506-532 1055-1081 306-327 939-960 41&437 1248-1270

rfl-actin rfl-actin mDP-R gene mDP-R gene rEP1-R cDNA rEP1-R cDNA rEP2-R cDNA rEP2-R cDNA rEP3-R cDNA rEP3-R cDNA rEP4-R cDNA rEP4-R cDNA mFP-R cDNA mFP-R cDNA rlP-R cDNA rlP-R cDNA rTP-R cDNA rTP-R cDNA

J00691 J00691 D29764 D29764 D88751 D88751 U48858 U48858 X83855 X83855 D28860 D28860 D 17432 D 17432 D28966 D28966 D21158 D21158

ACCCGCCGCCCTCGGTCTT AGCAGCGCCATGAGGCTGGAGTAG GTGCTGCCAACAGGCGATAATGGCACAGTG GGGACCTGCGGTCTTTCGGAATCGTGGAGA CATGGCCCTGGAACGCTACC TCAGTGAAGTCCGACAACAGAGG GCGGCGGGCGATGGAGGAGAG GCGGGACACCAGGGCTTTGATG TCTCTTACATGTACGCGGGCTTCA GTCTGGCAGGTATAGGAGGGTGTG GATAAAGACTGGATCGCGTTTGATCAG TGTCGTTTCACAGGTCACTGGGGAATT CATCGGCGTTTGCACTGTTGGT ATGGCCTGCGTGAATCCTCTGA CAACCTGCTGGCGCTGAGTGTG GCCTGGAGCTGGGAAGTGAACCT

Primer names are the name of the receptor subtype followed by "f"=forward (sense) or "r"=reverse (antisense), All sequences are given in the 5' to 3' direction of the primer. The positions indicated refer to the rat (r) or mouse (m) cDNA or gene registered under GenBank .Accession number given in the last column.

39

A. Fennekohl et al. receptors and which would yield PCR-products between about 550 and 850 bp were chosen.

Cell preparation Hepatocytes, Kupffer cells, sinusoidal endothelial cells and stellate cells were isolated from livers of male Wistar rats by established methods (29-31). Hepatocytes were purified by Percoll gradient centrifugation. Cell purity and viability were >98% judged by light microscopy and trypan blue exclusion. Kupffer cells, sinusoidal endothelial cells and stellate cells were purified by Nycodenz density centrifugation followed by centrifugal elutriation for Kupffer and endothelial cells. All nonparenchymal cell fractions were free of contamination with hepatocytes. Yet, as determined by light microscopy, there was always a slight cross-contamination of one nonparenchymal cell fraction with the other nonparenchymal cell types. The purity of the nonparenchymal cell fractions was >90%. All cell preparations were performed in accordance with the German Law on the Protection of Animals.

cDNA preparation Immediately after the cell preparation, total RNA was isolated from the purified cell preparations by CsC1 gradient centrifugation (32) or RNeasy (Qiagen, Hilden, Germany). For all PCR reactions shown in Fig. 1, the RNA preparation was digested with 10 U RNase-free DNase for 20 min at 37°C to remove contamination with genomic DNAs. The DNase was inactivated by heating to 64°C for 10 min and cDNA was synthesized from total RNA by oligo(dT)-primed reverse transcription with Superscript II (Stratagene, Heidelberg, Germany) according to the instructions of the manufacturer; 1 #Ci ct[35S]-dATP was included in the reaction mix to quantify the efficacy of reverse transcription. Incorporated label was separated from free label by precipitation with 0.5 volumes 7.5 M ammonium acetate and 2.5 volumes ethanol, followed by three washes with 80% (v/v) ethanol. The radioactivity incorporated into the cDNAs was assumed to be proportional to the amount of newly synthesized DNA. Thus, equal amounts of radioactivity correspond to equal amounts of cDNA in different reverse transcriptase reactions.

PCR-protocols All PCRs were repeated at least three times with cDNAs from different cell preparations, cDNAs of hepatocytes, Kupffer cells, sinusoidal endothelial cells and stellate cells were diluted to the same apparent concentration according to the radioactivity incorporated. This stock was further diluted 1:4 and 1:16. PCRs were carried out in a 50/~1 reaction mix containing 30 pmol of both forward and reverse primer, 200/~M of each deoxyribonucleotide triphosphate (dNTPs), 3% (v/v) dimethyl sulfoxide, serial dilutions of the cDNA preparation, 5/tl of PCR 10×buffer and 0.25 U Goldstar-Red polymerase (Eurogentec, Seraing, Belgium). The mix was subjected to the following PCR programs: DP-R; 3 min 95°C, 40×(1 min 95°C, 1 min 60°C, 2 min 72°C), 10 min 72°C. EP1-R; 3 min 95°C, 35×1 min 95°C, 1 min 55°C, 1 rain 72°C), 10 min 72°C. EP2-R; 3 min 95°C, 35×(1 rain 95°C, 1 min 58°C, 2 min 72°C), 10 min 72°C. EP3-R; 3 min 95°C, 35×(1 min 95°C, 1 min 65°C, 1 min 72°C), 10 rain 72°C. EP4-R; 3 min 95°C, 35×(1 min 95°C, 1 min 61°C, 2 min 72°C), 10 min 72°C. FP-R; 3 min 95°C, 35×(1 min 95°C, 1 min 55°C, 2 min 72°C), 10 rain 72°C. IP-R; 3 min 95°C, 35×(1 min 95°C, 1 min 63°C, 1 min 72°C), 10 min 72°C. TPR: 3 min 95°C, 35×(90 s 95°C, 1 rain 58°C, 1 min 72°C), 10 rain 72°C. The quantity and integrity of the cDNA were checked by the PCR for fl-actin. The same cDNA preparations were compared in subsequent PCR reactions for all receptors. In order to be in the "linear range" of the PCR, cDNA-stocks had to be less diluted for the prostanoid receptor PCRs than for the B-actin PCR. Relative dilutions were as follows (prostanoid receptor/fl-actin): DP-R, 2:1; EP1-R, 4:1; EP2-R, 2.5:1; EP3-R, 2.5:1; EP4-R, 2.5:1; FP-R, 2:1; IP-R, 2:1; TPR, 2.5:1. Products were separated electrophoretically on 1% to 3% agarose gels and visualized by staining with ethidium bromide. PCR with RNA samples without prior reverse transcription was performed to verify that the DNase treatment of the RNA preparation (see above) had abolished any contamination with genomic DNA.

40

Receptor G.prot

HC

KC

SEC

HSC

DP-R

Gs

[-]

[++]

[+]

[+++]

EP1-R

Gq

+++

+++

+++

++

EP2-R

Gs

+++

++

(+)

+++

+++

++

(+)

+++

(+)

EP3-R

Gi

EP4-R

Gs

FP-R

Gq

IP-R

Gs

(+)

++

+++

TP-R

Gq

++

+++

(+)

+++

+++

+++

~At:~I

++

+++

+++

DP-R

Gs

EP1-R

Gq

EP2-R

Gs

EP3-R

Gi

EP4-R

Gs

FP-R

Gq

IP-R

Gs

TP-R

Gq

~Ac~n

Dilution

1:1 1:4 1:16 1:1 1:4 1:16 1:1 1:4 1:16 1:1 1:4 1:16

Fig. 1. Determination by semiquantitative RT-PCR of the d~ferential expression of prostanoid receptors in hepatoo'tes (HC), Kupffer cells (KC), sinusoidal endothelial cells (SEC) and stellate cells (HSC) of rat liver. Cells were isolated and purified as detailed in the Methods section. Total RNA was purffed from the freshly isolated cells and served as template for an oligo-dT-primed radioactive reverse transcription. 1:1, 1:4 and 1:16 dilutions of equal amounts of cDNA Of the different cell types were amplified with specific primers .~Col"all prostanoid receptor cDNAs and, Jor comparison, the fl-actin eDNA. To compensate for the different amplification ~flficao' off the PCR protocols and the different abundance of the receptor m R N A s of the different prostanoid receptors the first dilution was adjusted relative to fl-actin as follows: DP-R, 2:1, EP1-R, 4:1; EP2R, 2.5.1; EP3-R, 2.5.'1; EP4-R, 2.5.1; FPR, 2:1; IP-R, 2:1; TP-R, 2.5:1. Therefore, a comparison of the relative expression is feasible only for the same receptor h~ different cell types (horizontal) but n o t f o r different receptors in the same cell O'pe (vertical). Note that 40 o,cles were necessary to amplify the DP-R cDNA in comparison to 35 o,cles for all other cDNAs. The RT-PCR products shown are of one optimized parallel cDNA preparation from all four cell types and are representative of a total of" at least three cDNA preparations pet" cell type. The ranking is the average of all preparations and, thereJore, is not necessarily in all cases identical to the strength (~£the PCR product bands shown. Ranking: + + + strong expression, ++ moderate expression, + and ( + ) weak expression or low m R N A levels, possibly due to impurities h7 cell preparation. G-prot: G-protein to which the respective receptor couples.

Differential expression of hepatic prostanoid receptors Characterization of the PCR-products PCR-products of all receptors were extracted from the gel, reamplifled by the same PCR-protocoland the gel-purifiedamplificateswere characterized by cleavagewith two restriction endonucleases(Fig. 2). The identities of the PCR-products of the DP-R, EP2-R and FP-R were in addition confirmedby partial sequencing.

Results DP-receptor DP-R m R N A was undetectable in whole liver R N A by RT-PCR (Gerashchenko D, personal communication). An extremely high number of cycles (n=40) was necessary to detect transcripts in a DNase-treated R N A from nonparenchymal cells. The DP-receptor m R N A was most abundant in stellate cells, followed by Kupffer and sinusoidal endothelial cells (Fig. 1). Since the sequence of the rat DP-receptor was not known, the gel-purified PCR-product was not characterized by restriction enzyme digest but was sequenced in both directions by cycle sequencing using the same primers as were used for the PCR. The sequence was 91% homologous to the corresponding region of the mouse DP-receptor cDNA (not shown). EPl-receptor EP1-R m R N A was present in all liver cell types at similar levels (Fig. 1). However, it was completely absent from 1 of 4 hepatocyte preparations. Cleavage of the

kb-ladder standard: EPI

Drall: 251 Smal: 200

EP2

Aval: 343 Smal: 346

EP3

Ncol: 501 b AIwNI: 363 b

EP4

Avail: 431 bp + 228 bp " Alul" 235 bp + 424 bp

FP

Bglll: 439 b AIwNI: 471 b

IP

Smal: 36: Drall: 25.¢

TP

Smal: 578 bl Ncol: 207 b

to

o4

Fig. 2. Identification of the PCR-products by restriction enzyme digest. PCR-products of Fig. 1 were extracted from the agarose gels, reamplified with the same primer combination as in the first PCR, gel purified and cleaved by the restriction enzyme indicated. The size of the expected fragments is given after the enzyme name.

purified 672 bp PCR-product with SmaI or DralI yielded the expected products of 200 bp and 472 bp or 251 bp and 421 bp, respectively (Fig. 2). EP2-receptor EP2-R mRNA was most abundant in Kupffer cells, followed by sinusoidal endothelial cells. Extremely low levels were also present in some preparations of stellate cells (Fig. 1). It was absent in all hepatocyte preparations tested. Cleavage of the purified 620 bp PCRproduct with SmaI or AvaI yielded the expected products of 274 bp and 346 bp or 276 bp and 343 bp, respectively (Fig. 2). SmaI and AvaI have different cleavage sites but the same recognition sequence. Therefore, the identity of EP2-R was confirmed in addition by sequencing (not shown). EP3-receptor EP3-R m R N A was present at similar levels in all cell types (Fig. 1). Two isoforms of this receptor are generated by alternative splicing. The PCR-protocol employed did not distinguish between these C-terminal splice variants. Yet, at least in hepatocytes two splice variants, the EP3a and the EP3fl mRNA were detectable (not shown) by an alternative PCR-protocol (33). Cleavage of the purified 829 bp PCR-product with AlwNI or NcoI yielded the expected products of 363 bp and 466 bp or 328 bp and 501 bp, respectively (Fig. 2). EP4-receptor EP4-R m R N A yielded only weak signals with most R N A preparations: it was most abundant in sinusoidal endothelial cells, followed by Kupffer cells (Fig. 1). Extremely low levels were present in some preparations of stellate cells; no EP4-R m R N A was detected in any of the hepatocyte preparations tested. Cleavage of the purified 659 bp PCR-product with AluI or AvalI yielded the expected products of 235 bp and 424 bp or 228 bp and 431 bp, respectively (Fig. 2). FP-receptor FP-R was expressed exclusively in hepatocytes (Fig. 1). It was not detectable in any of the other cell preparations. In one experiment (cDNA-preparation) a weak signal could be detected in the Kupffer cell preparation, which most probably was due to contaminating or phagocytosed hepatocyte detritus. The 576 bp fragment was cleaved with AlwNI and yielded the expected fragments of 105 bp and 471 bp (Fig. 2). Unexpectedly, the FP-R fragment was not cleaved by BgllI. Therefore, the gel-purified fragment was sequenced by cycle sequencing using the PCR primers. The sequence was 41

A. Fennekohlet al. identical with the published sequence of the rat FPreceptor (not shown), except that the BglII site was lost due to an A to G exchange at position 871 (AccNoX83856).

IP-receptor IP-R mRNA was most abundant in stellate cells followed by sinusoidal endothelial cells (Fig. 1). A weak signal was also obtained with cDNA from the Kupffer cell preparation, which, however, could be due to contaminating sinusoidal endothelial cells. No IP-R mRNA was present in the hepatocyte preparation. Cleavage of the purified 655 bp PCR-product with DraII or Sinai yielded the expected products of 259 bp and 396 bp or 293 bp and 362 bp, respectively (Fig. 2). In some of the cDNA preparations an additional band at about 1600 bp appeared after PCR-amplification (not shown). This band was no longer amplified if the RNA was digested with RNase-free DNase prior to reverse transcription and thus most likely represents the amplificate of contaminating genomic DNA. In line with this notion, the primer pair employed in the PCR-reaction flanks the second intron in the homologous region of the human IP-R gene (33). TP-receptor TP-R mRNA was most abundant in sinusoidal endothelial cells (Fig. 1). A weaker signal was detected in Kupffer cells and stellate cells. TP-R mRNA was absent from the hepatocytes. Cleavage of the purified 853 bp PCR-product with NcoI or SmaI yielded the expected products of 207 bp and 646 bp or 275 bp and 578 bp, respectively (Fig. 2).

Discussion Validity of the experimental protocol Semiquantitative RT-PCR has been used to quantify mRNA of prostanoid receptors in hepatic cells. Because of the apparently very low copy number of these mRNAs and the limited quantity, especiaily of nonparenchymal cells, that can be isolated from one liver, other methods like Northern blot or RNase-protection assay were not sensitive enough to quantify or even detect prostanoid receptor mRNAs. The validity of the semiquantitative RT-PCR depends on two crucial prerequisites: the efficacy and quality of the reverse transcription have to be monitored to ensure that equal amounts of mRNA yield equal amounts of cDNA, and the subsequent PCR-reaction must not reach saturation. Here, the efficacy of the reverse transcription was monitored by the incorporation of [35S]dATP into the newly synthesized cDNA. Equal amounts of radioactivity were assumed to correspond to equal amounts 42

of cDNA. Yet, equal amounts of radioactivity could correspond to a small number of full-length transcripts or a large number of shorter fragments. Therefore, the quality of the reverse transcription was checked by amplification of the cDNA of the "house-keeping gene" flactin with a primer pair that is located in the translated region of the mRNA close to the start codon. To ensure that the PCR-reaction was not saturated, three dilutions of all cDNAs were amplified in parallel, yielding decreasing amplificate concentrations. However, semiquantitative PCR does not assess the amplification efficiency, which certainly differs between the PCRprotocols for the different prostanoid receptors studied here. Therefore, a comparison of the apparent expression level is not feasible between different receptors in the same cell type, but only with the same receptor in different cell types. Preparations of freshly isolated nonparenchymal liver cell fraction are never 100% pure. While hepatocytes were essentially free of contamination with nonparenchymal cells, and nonparenchymal cell fractions did not contain hepatocytes, cross-contamination was likely between the different nonparenchymal cell fractions. It is not possible to quantitate such cross-contamination in freshly isolated cells, since antibodies for immunological markers fail to react with freshly isolated cells due to the protease treatment during cell preparation. A "gold standard" for a PCR-product, that would be exclusively amplified from the cDNA of one cell type has not yet been established. However, cross-contamination can be estimated by light microscopical parameters such as cell size or characteristic features such as the fat droplets of hepatic stellate cells. According to these estimates, the nonparenchymal cell fractions were >90% pure. Thus, a PCR signal can be interpreted as false positive if the receptor mRNA level appears to be 10-fold lower in cell type A than in the possibly contaminating cell type B. This is indicated in Fig. 1 as (+).

Hepato~Ttes Hepatocytes were the only liver cell type that expressed Gq-linked FP-R mRNA. They also expressed Gqlinked EP1-R mRNA. FP-R and EP1-R can mediate an increase in InsP3, Ca 2+ and, as a consequence, activation of glycogen phosphorylase and hence glucose output (17,22,34). The effect of ligand binding to the receptor is "glucogenic". The mRNA of the FP-receptor is most abundant in the periportal zone of the liver acinus (35). This is in line with the larger glucogenic capacity of periportal as compared with perivenous hepatocytes (36) (Fig. 3). In addition to the two Gq-linked "glucogenic" prost-

Differential expression of hepatic prostanoid receptors

Fig. 3. Integration by prostaglandins of hepatic functions in homeostasis. Dotted arrows: antiglucogenic signal chain; solid arrows: glucogenic signal chain. Abbreviations: A CY, adenylate cyclase; SEC, sinusoidal endothelial cell," Gcg, glucagon; HC, hepatocyte; HSC, hepatic stellate cell; KC, Kupffer cell," NT, nerve terminal; PCK, phosphoenolpyruvate-carboxykinase; PLC, phospholipase C," -R, -receptor.

anoid receptors, hepatocytes expressed Gi-linked EP3R mRNA. Ligand occupation of this receptor decreases hormone-stimulated cAMP formation in hepatocytes (18-20) and hence reduces hormonestimulated glycogenolysis, gluconeogenesis and glucose output (22). It also attenuates the glucagon-induced expression of the key glucogenic enzyme phosphoenolpyruvate carboxykinase (37). The overall effect of ligand binding to this receptor is "antiglucogenic" (Fig. 3). No m R N A from any of the G~-linked prostanoid receptors was detected in hepatocytes. The PCRs were negative, even if a 10-fold amount of cDNA was used in the PCR-reaction mixture (not shown). Controversial results have been reported concerning the functional evidence for a Gs-linked prostanoid receptor on hepatocytes. While some groups found no prostanoidinduced cAMP formation in rat hepatocyte cultures (34,38,39), others (40-42), and even the same group in earlier publications (43), reported PGE2-induced cAMP formation in hepatocytes or liver plasma membranes. This discrepancy is most likely due to contaminating nonparenchymal cells (membranes) in those hepatocyte or liver membrane preparations that responded with cAMP formation to PGE> because most

hepatocyte preparations are not purified by density gradient centrifugation and contain between 0.1 and 1% Kupffer cells. PGE2 increased cAMP formation in Kupffer cell cultures to about 100-fold higher values than in hepatocyte cultures (25 vs. 41). In addition, in the experiments with hepatocyte cultures (43), the expression of a G~-linked EP-R may have been induced by the culture conditions (see below: hepatic stellate cells). It is also puzzling that hepatocytes contain no DPR although PGD2 appears to be the prostanoid that is formed in largest quantities in Kupffer cells in response to a variety of different stimuli (6,8). Some groups even reported that PGD2 was the most potent prostanoid in activating glycogen phosphorylase in isolated hepatocytes (6). This was, however, not confirmed in other studies (34), which also showed that in contrast to PGD2, a DP-R selective, metabolically stable agonist, did not stimulate hepatocyte glycogen phosphorylase activity. This study concluded that either PGD2 must act via hepatocyte FP-R, o r P G D 2 must be metabolically converted into another active prostanoid.

Kupffer cells mRNA for the Gs-linked EP2-R was most abundant in Kupffer cells, mRNAs for the three other Gs-linked prostanoid receptors, EP4-R, DP-R and IP-R, were detectable at low levels. PGE2 has been shown to decrease endotoxin-induced T N F a production via an increase in intracellular cAMP (25). This effect could be mediated by either EP2-R or EP4-R (Fig. 4). Pharmacological approaches cannot distinguish between these two possibilities, since receptor subtype specific agonists display insufficient specificity. The prolonged time course of PGEz-mediated inhibition of T N F a formation would favor EP2-R, because EP4-R is a rapidly desensitized receptor (44). This view is supported by the comparatively strong expression of EP2-R in Kupffer cells. In addition to the two G~-linked PGE2-receptors, Kupffer cells express the Gi-linked EP3-R mRNA. This is, at first glance, puzzling because these three receptors have a similar affinity for PGE2 and, thus, the same agonist (PGE2) would act to increase and to decrease cAMP formation in the same cell. Possible explanations are either that the Kupffer cell preparation is heterogeneous and contains cells with only EP2-R/EP4-R or only EP3-R, or that the two receptors are present in the same cell but are located in different regions of the cell that are not simultaneously exposed to PGE2, e.g. on sinusoidal surfaces of Kupffer cells versus Kupffer cell protrusions into the space of Disse. Finally, the receptor proteins 43

A. Fennekohl et al. tion, possibly via the IP-R that at least on the mRNA level was also present on sinusoidal endothelial cells (Fig. 4). Sinusoidal endothelial cells have been shown to act as antigen-presenting cells (46). In a T-cell activation assay the antigen presentation seemed to be reduced by PGE2 (47). It is not clear which of the four EP-R types is involved in this signaling chain.

Fig. 4. Integration by prostanoids of hepatic functions in inflammation. Dotted arrows: anti-inflammatory signal chains; solid arrows: proinflammatory signal chains. Abbreviations: A CY, adenylate cyelase; SEC, sinusoidal endothelial cell, Etx, endotoxin; Gcg, glucagon; HC, hepatocyte; HSC, hepatic stellate cell; IL, interleukin; KC, Kupff er cell; PCK, phosphoenolpyruvate-carboxykinase; PLC, phospholipase C," - R, -receptor; R OI, reactive o.~vgen intermediates; TNF, tumor necrosis factor c~.

might be differentially desensitized or downregulated, and thereby be uncoupled from their signaling pathways. There is currently no functional evidence for the EP3-R on Kupffer cells. The strong expression of EP1-R mRNA in Kupffer cells is at present not substantiated by functional evidence. The moderate levels of TP-R mRNA may well be due to contaminating sinusoidal endothelial cells in the Kupffer cell preparation. Again, there is no functional evidence for a Kupffer cell TP-R. Sinusoidal endothelial cells Endothelial ceils expressed the mRNA of all prostanoid receptors but the FP-R. So far, functional evidence has been provided for the existence of Gq-linked TP-R (26), Gs-linked IP-R and G~-linked EP-R (EP2R or EP4-R) (45,46) (Fig. 4). The TP-R has been detected in sinusoidal endothelial cell plasma membranes in binding studies. The receptor number was reduced by homologous desensitization during endotoxin exposure (26). TP-R on sinusoidal endothelial cells has been implicated in reperfusion injury (45), while the IP-R agonist iloprost seemed partially to prevent reperfusion injuries by antagonizing the thromboxane ac44

Hepatic stellate cells The highest DP-R and IP-R mRNA levels in all nonparenchymal cells were found in these cells. Both receptors are linked to Gs to increase cAMP. Thereby, PGI2 or PGD2 might antagonize the contractile response elicited by other agonists acting via an increase in InsP3 and intracellular Ca 2+, e.g. endothelin (48). This has been shown for precontracted hepatic stellate cells in culture which relaxed after application of the prostacyclin analog iloprost (15) (Fig. 4). No or very low amounts of FP-R and TP-R mRNA were detected in freshly isolated stellate cells. This contrasts with functional studies that have shown a PGF2~ and thromboxane-dependent contraction of stellate cells in long-term culture (15). This apparent discrepancy may be due to the transformation of hepatic stellate cells to myofibroblast-tike cells in culture which is accompanied by a change in receptor expression. Some receptors have been shown to disappear as culture time increases. For example, noradrenaline increased prostanoid formation in stellate cells cultured for 24 h, but no longer increased prostanoid formation in cells cultured for 72 h (2). Similarly, the glucagon receptor mRNA was detected in freshly isolated stellate cells, while it was almost absent from 72-h hepatic stellate cell cultures (Krones, Schieferdecker, Jungermann, unpublished observation). The mRNAs of EP-receptor subtypes 1 and 3 were present in stellate cells but currently no specific function can be assigned to any particular receptor. PGE2 has been shown to reduce stellate cell collagen gene expression (49) and PDGF-induced proliferation of stellate cells (50). However, the signaling path responsible for these regulations has not yet been elucidated. Prostanoid receptor-mediated control of homeostasis In rat liver, prostaglandins can act as "glucogenic" and "antiglucogenic" signals in the hepatic sinusoid (Fig. 3). Glucogenic signal chain." Prostaglandins F2~ and D2 released from stellate cells in response to the neurotransmitter noradrenaline (2) can increase glycogen phosphorylase activity in hepatocytes via FP-R (34),

Differential expression of hepaticprostanoid receptors thereby enhancing the nerve stimulation-mediated hepatic glucose output, and thus furnishing fuel for the systemic circulation during an emergency response (13). Antiglucogenic signal chain: PGE2 released from Kupffer cells (1) or possibly hepatic stellate cells (Krones, Schieferdecker, Jungermann, unpublished observation) in response to glucagon, can inhibit by feedback the cAMP-mediated glucagon-dependent glucose release and phosphoenolpyruvate carboxykinase gene expression in hepatocytes (37) via the Gi-linked EP3-R.

mediates and nitric oxide (54); however, the opposite PGE2 effect on N O production has also been reported (55,56).

Acknowledgements The excellent technical assistance of Angela Timmerm a n n and Andreas Nolte is gratefully acknowledged. We also wish to thank Dr. D m i t r y Gerashchenko for providing detailed information a b o u t the rat D P - R P C R protocol before publication. Supported by the Deutsche Forschungsgemeinschaft through the Sonderforschungsbereich 402, Teilprojekt B4 and B6.

Prostanoid receptor-mediated control of inflammation In rat liver, prostanoids can elicit proinflammatory and anti-inflammatory signals in the hepatic sinusoid (Fig. 4). Proinflammatory signal chains." PGE2, PGF2~ and PGD2 released from Kupffer cells in response to a large array of inflammatory mediators (3-7,51) can increase glycogenolysis in hepatocytes via EP1-R and FP-R, respectively (22,34), thereby providing fuel substrate for the inflammatory response of nearby Kupffer cells. During an acute phase response the liver needs large amounts of amino acids and nucleotides to produce acute phase proteins and their m R N A s , respectively. While amino acids are supplied mainly by net proteolysis in muscles via the circulation, nucleotides for m R N A production must be furnished by the hepatocytes themselves. PGE2 can reduce hormone-controlled levels of m R N A s for key enzymes of carbohydrate metabolism (37) to provide the nucleotides needed for the synthesis of acute phase protein m R N A s (52). TXA2 can act on endothelial cells via the TP-R. It might reduce sinusoidal blood flow, thereby facilitating leukocyte adhesion, and it might reduce the size o f sinusoidal fenestrae, thus reducing sinusoidal filtration and increasing the concentration of mediators in the space o f Disse. Antiinflammatory signal chains: PGI2 and PGE2 could antagonize the TXA2 elicited reduction of sinusoidal blood flow and filtration via I P - R and EP4-R on endothelial cells. Hepatic stellate cells may be contracted by an as yet undefined mediator X. PGI2 and PGD2 could also counterbalance this contraction via the I P - R and the DP-R, respectively. Finally, PGE2 and possibly PGD2 in a negative feedback inhibition loop via the EP2-R and possibly the D P - R can attenuate the release from Kupffer cells of proinflammatory cytokines such as IL-1, IL-6 (53), and T N F a (11) that by themselves stimulate P G E z - f o r m a t i o n in Kupffer cells. In this feedback inhibition loop, PGE2 could additionally inhibit the release of reactive oxygen inter-

References 1. Hespeling U, Jungermann K, Ptischel GR Feedback-Inhibition of glucagon-stimulated glycogenolysisin hepatocyte/Kupffercell cocultures by glucagon-elicited prostaglandin production in Kupffer cells. Hepatology 1995; 22: 1577-83. 2. Athari A, H~inecke K, Jungermann K. Prostaglandin F2~ and D2 release from primary Ito cell cultures after stimulation with noradrenaline and ATP but not adenosine. Hepatology 1992; 20: 142-8. 3. Tran Thi TA, H~iussingerD, Gyufko K, Decker K. Stimulation of prostaglandin release by Ca2+- mobilizingagents from the perfused rat liver. A comparative study on the action of ATR UTR phenylephrine, vasopressin and nerve stimulation. Biol Chem Hoppe Seyler 1988; 369: 65-8. 4. vom Dahl S, Wettstein M, Gerok W, H~iussingerD. Stimulation of release of prostaglandin D 2 and thromboxane B2 from perfused rat liver by extracellular adenosine. Biochem J 1990; 270: 39-44. 5. Nukina S, Fusaoka T, Thurman RG. Glycogenolytic effect of adenosine involvesATP from hepatocytes and eicosanoids from Kupffer cells. Am J Physiol 1994; 266: G99-G102. 6. Casteleijn E, Kuiper J, Van Rooij HC, Kamps JA, Koster JF, Van Berkel TJ. Endotoxin stimulates glycogenolysisin the liver by means of intercellular communication. J Biol Chem 1988; 263: 6953-5. 7. Buxton DB, Fisher RA, Briseno DL, Hanahan DJ, Olson MS. Glycogenolyticand haemodynamic responses to heat-aggregated immunoglobulin G and prostaglandin E2 in the perfused rat liver. Biochem J 1987; 243: 493-8. 8. Ptischel GR Hespeling U, Oppermann M, Dieter R Increase in prostanoid formation in rat liver macrophages (Kupffer cells) by human anaphylatoxin C3a. Hepatology 1993; 18: 1516-21. 9. Hespeling U, Piischel GR Jungermann K, G6tze O, Zwirner J. Stimulation of glycogen phosphorylase in rat hepatocytes via prostanoid release from Kupffer cells by recombinant rat anaphylatoxin C5a but not by native human C5a in hepatocyte Kupffer cell cocultures. FEBS Lett 1995; 372: 108-12. 10. Flynn JT. Inhibition of complement-mediatedhepatic thromboxane production by mepacrine, a phospholipase inhibitor. Prostaglandins 1987; 33: 287-99. 11. Peters T, Karck U, Decker K. Interdependence of tumor necrosis factor, prostaglandin E2, and protein synthesis in lipopolysaccharide-exposed rat Kupffer ceils. Eur J Biochem 1990; 191: 5839. 12. Gandhi CR, Debuysere MS, Olson MS. Platelet-activating factor-mediated synthesis of prostaglandins in rat Kupffer cells. Biochim Biophys Acta 1992:1136: 68-74. 13. Ptischel GR Jungermann K. Integration of function in the hepatic acinus: intercellular communication in neural and humoral control of liver metabolism. Prog Liver Dis 1994; 12: 19-46. 14. Peters T, Gaillard T, Decker K. Tumor necrosis factor alpha stimulates prostaglandin but not superoxide synthesis in rat Kupffer cells. Eicosanoids 1990; 3:115-20. 45

A. Fennekohl et al.

15. Kawada N, Klein H, Decker K. Eicosanoid-mediated contractility of hepatic stellate cells. Biochem J 1992; 285: 367-71. 16. Coleman RA, Eglen RM, Jones RL, Narumiya S, Shimizu T, Smith WL, et al. Prostanoid and leukotriene receptors: a progress report from the IUPHAR working parties on classification and nomenclature. Adv Prostaglandin Thromboxane Leukot Res 1995; 23: 283-5. 17. Athari A, Jungermann K. Direct activation by prostaglandin F2 alpha but not thromboxane A2 of glycogenolysis via an increase in inositol-l,4,5-trisphosphate in rat hepatocytes. Biochem Biophys Res Commun 1989; 163: 1235-42. 18. Brass EP, Ganity MJ. Structural specificity for prostaglandin effects on hepatocyte glycogenolysis. Biochem J 1990; 267: 59-62. 19. Okumura T, Sago T, Saito K. Effect of prostaglandins and their analogues on hormone-stimulated glycogenolysis in primary cultures of rat hepatocytes. Biochim Biophys Acta 1988; 958: 17987. 20. Bronstad GO, Christoffersen "12 Inhibitory effect of prostaglandins on the stimulation by glucagon and adrenaline of formation of cyclic AMP in rat hepatocytes. Eur J Biochem 1981; 117: 369-74. 21. NeuscMfer-Rube E PC~schel GP, Jungermann K. Characterization of prostaglandin F2alpha binding sites on rat hepatocyte plasma membranes. Eur J Biochem 1993; 211: 163-9. 22. Pfischel GP, Kirchner C, Schr6der A, Jungermann K. Glycogenolytic and antiglycogenolytic prostaglandin-E(2) actions in rat hepatocytes are mediated via different signalling pathways. Eur J Biochem 1993; 218: 1083-9. 23. Neusch~,fer-Rube E DeVries C, H~inecke K, Jungermann K, Pt~schel GP Molecular cloning and expression of a prostaglandin E 2 receptor of the EP3 beta subtype from rat hepatocytes. FEBS Lett 1994; 351:119 22. 24. de Vries C. NeuscMfer-Rube E Hfinecke K, Jungermann K, Piischel GR Molecular cloning, expression and functional chararacterization of a hepatocyte prostaglandin F2~ receptor [abstract]. Hepatology 1995; 22: 304A. 25. Karck U, Peters T, Decker K. The release of tumor necrosis factor from endotoxin-stimulated rat Kupffer cells is regulated by prostaglandin E2 and dexamethasone. J Hepatol 1988: 7:352 61. 26. lshiguro S, Arii S, Monden K, Adachi Y, Funaki N, Higashitsuji H, et al. Identification of the thromboxane A2 receptor in hepatic sinusoidal endothelial cells and its role in endotoxin-induced liver injury in rats. Hepatology 1994: 20: 1281-6. 27. Hirata M, Hayashi Y, Ushikubi E Yokota Y, Kageyama R, Nakanishi S, et al. Cloning and expression of cDNA for a human thromboxane A 2 receptor. Nature 1991; 349: 617-20. 28. Ushikubi E Hirata M, Narumiya S. Molecular biology of prostanoid receptors; an overview. J Lipid Mediat Cell Signal 1995; 12: 343-59. 29. Meredith M J. Rat hepatocytes prepared without collagenase: prolonged retention of differentiated characteristics in culture. Cell Biol Toxicol 1988; 4:405 25. 30. Eyhorn S, Schlayer H J, Henninger HE Dieter R Hermann R, Woort Menker M, et al. Rat hepatic sinusoidal endothelial cells in monolayer culture. Biochemical and ultrastructural characteristics. J Hepatol 1988; 6:23 35. 31. Kawada N, Inoue M. Effect of adrenomedullin on hepatic pericytes (stellate cells) of the rat. FEBS Lek 1994; 356: 109-13. 32. Chirgwin JM, Przybyia AK, MacDonald R J, Rutter WJ. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 1979; 18: 5294-9. 33. Ogawa Y, Tanaka I, Inoue M, Yoshitake Y, Isse N, Nakagawa O, et al. Structural organization and chromosomal assignment of the human prostacyclin receptor gene. Genomics 1995; 27: 1428. 34. Paschel GR NeuscMfer-Rube E de Vries C, Hgnecke K, Nolte A, Kirchner C, et al. Characterization and molecular cloning of hepatocyte prostaglandin receptors possibly involved in the nerve stimulation-dependent increase in hepatic glucose output. In:

46

Shimazu T, editor: Liver lnnervation. London: John Libby & Co.; 1996. pp 87 94. 35. De Vries C, Neusch~ifer-Rube F, Kietzmann T, H~necke K, Freimann S, Jungermann K, et al. Klonierung, Expression, funktionelle Charakterisierung und intrahepatische Lokalisation eines Hepatocyten Prostaglandin F2~ Rezeptors. Z Gastroenterol 1996; 34: 47. 36. Jungermann K, Kietzmann "17'.Zonation of parenchymal and nonparenchymal metabolism in liver. Annu Rev Nutr 1996; 16: 179-203. 37. Pfischel GP, Christ B. Inhibition by PGE2 of glucagon-induced increase in phosphoenolpyruvate carboxykinase mRNA and acceleration of mRNA degradation in cultured rat hepatocytes. FEBS Lett 1994; 351: 353-6. 38. Refsnes M, Dajani OE Sandnes D, Thoresen GH, Rottingen JA, Iversen JG, et al. On the mechanisms of the growth-promoting effect of prostaglandins in hepatocytes: the relationship between stimulation of DNA synthesis and signaling mediated by adenylyl cyclase and phosphoinositide-specific phospholipase C. J Cell Physiol 1995; 164:465 73. 39. Mine T, Kojima I, Ogata E. Mechanism of prostaglandin E2induced glucose production in rat hepatocytes. Endocrinology 1990; 126:2831-6. 40. Ganity M J, Westcott KR, Eggerman TL, Andersen NH, Storm DR, Robertson RE Interrelationships between PGEt and PGI2 binding and stimulation of adenylate cyclase. Am J Physiol 1983; 244: E367-72. 41. Okamura N, Terayama H. Prostaglandin receptor-adenylate cyclase system in plasma membranes of rat liver and ascites hepatomas, and the effect of GTP upon it. Biochim Biophys Acta 1977; 465:54 67. 42. Sweat FW, Wincek TJ. The stimulation of hepatic adenylate cyclase by prostaglandin El. Biochem Biophys Res Commun 1973: 55:522 9. 43. Melien O, Winsnes R, Refsnes M, Gladhaug IE Christoffersen 12 Pertussis toxin abolishes the inhibitory effects of prostaglandins El, E2, I2 and F2 alpha on hormone-induced cAMP accumulation in cultured hepatocytes. Eur J Biochem 1988; 172:293 7. 44. Nishigaki N, Negishi M, Ichikawa A. Two Gs-coupled prostaglandin E receptor subtypes, EP 2 and EP4, differ in desensitization and sensitivity to the metabolic inactivation of the agonist. Mol Pharmacol 1996; 50:1031 7. 45. Ishiguro S, Arii S, Monden K, Fujita S, Nakamura T, Niwano M, et al. Involvement of thromboxane A2-thromboxane A2 receptor system of the hepatic sinusoid in pathogenesis of cold preservation/reperfusion injury in the rat liver graft. Transplantation 1995; 59: 957-61. 46. Lohse AW, Knolle PA, Bilo K, Uhrig A, Waldmann C, Ibe M, et al. Antigen-presenting function and B7 expression of murine sinusoidal endothelial cells and Kupffer cells. Gastroenterology 1996; 110: 1175-81. 47. Knolle PA, Germann T, Jin SC, Duchmann R, Meyer zum Bt~schenfelde KH, Greken G, et al. Bedeutung von Zytokinen in der T-Zellaktivierung durch sinusoidale Endothelzellen der Leber. Z Gastroenterol 1997; 35: 59. 48. Kawada N, Harada K, Ikeda K, Kaneda K. Morphological study of endothelin-l-induced contraction of cultured hepatic stellate cells on hydrated collagen gels. Cell Tissue Res 1996; 286: 477-86. 49. Beno DW, Espinat R, Edelstein BM, Davis BH. Administration of prostaglandin El analog reduces rat hepatic and Ito cell collagen gene expression and collagen accumulation after bile duct ligation injury. Hepatology 1993; t7:707 14. 50. Beno DW, Rapp UR, Davis BH. Prostaglandin E suppression of platelet-derived-growth-factor-induced Ito cell mitogenesis occurs independent of raf perinuclear translocation and nuclear proto-oncogene expression. Biochim Biophys Acta 1994; 1222: 292 300. 51. Schieferdecker HL, Rothermel E, Timmermann A, G6tze O, Jungermann K. Anaphylatoxin C5a receptor mRNA is strongly ex-

Differential expression of hepatic prostanoid receptors pressed in Kupffer and stellate cells and weakly in sinusoidal endothelial cells but not in hepatocytes of normal rat liver. FEBS Lett 1997; 406: 305-9. 52. Christ B, Nath A, Heinrich PC, Jungermann K. Inhibition by recombinant human interleukin-6 of the glucagon-dependent induction of phosphoenolpyruvate carboxykinase and of the insulin-dependent induction of glucokinase gene expression in cultured rat hepatocytes: regulation of gene transcription and messenger RNA degradation. Hepatology 1994; 20: 1577-83.

53. Goss JA, Mangino M J, Callery MP, Flye MW. Prostaglandin E 2 downregulates Kupffer cell production of IL-1 and IL-6 during hepatic regeneration. Am J Physiol 1993; 264: G601-G608. 54. Harbrecht BG, McClure EA, Simmons RL, Billiar TR. Prostanoids inhibit Kupffer cell nitric oxide synthesis. J Surg Res 1995; 58: 625-9. 55. Gaillard T, Mulsch A, Klein H, Decker K. Regulation by prostaglandin E 2 of cytokine-elicited nitric oxide synthesis in rat liver macrophages. Biol Chem Hoppe Seyler 1992; 373: 897-902.

47