Research Paper Mediators of Inflammation, 8, 295–303 (1999)

LPS and MTP-PE (liposome-encapsulated N-acetyl-muramyl-L-alanyl-D-isoglutaminyl-L-alanine-2-:[1',2'dipalmitoyl -sni-glycero-3-(hydroxy-phosphoryl-oxyl)] etylamide) induce in liver macrophages a synthesis and release of TNF-alpha, nitric oxide and prostanoids. Both agents induce an expression of mRNA's encoding TNF-alpha, inducible nitric oxide synthase (iNOS) and cyclooxygenase (COX)-2, and of corresponding proteins. LPS and MTP-PE induce a rapid activation of the extracellular regulated kinase (ERK) isoenzymes-1 and -2. Inhibition of map kinase isoenzymes leads to a decreased release of TNF-alpha, nitric oxide and prostaglandin (PG) E2 after both agents. The transcription factors NF-kappaB and AP-1 are strongly activated by LPS within 30 minutes. MTP-PE induces a weak activation of both transcription factors only after 5 hours. Inhibition of NF-kappaB inhibits the LPS- but not the MTP-PE-induced release of TNF-alpha, nitric oxide and PGE2. PGE2 release after LPS is higher than after MTP-PE. Exogenously added PGE2 inhibits the activation of map kinase and TNF-alpha release by LPS, but not by MTP-PE. Release of nitric oxide after LPS and MTP-PE is enhanced after prior addition of PGE2. PGD2 is without any effect. MTP-PE, but not LPS, induces a cytotoxicity of Kupffer cells against P815 tumor target cells. The MTP-PE-induced cytotoxicity is reduced by TNF-alpha neutralizing antibodies, indicating the involvement of TNF-alpha. Thus our results suggest that the different potencies of LPS and MTP-PE as immunomodulators probably result from different actions on Kupffer cells, resulting in differences in the amounts and kinetics of released TNF-alpha and PGE2, and that PGE2 plays an important regulatory role in the action of LPS, but not in the actions of MTP-PE.


Introduction
Live r mac rophages (Kup ffer c ells) are macrophages residing in the sinusoids of the liver. This strate gic loc ation makes them the first macrophage s to c ome into c ontact w ith nox ious mate rials that enter circulation via the portal ve in. This loc ation and the fact that the y c onstitute the largest pool of mac rophages in the body attributes to the m an important func tion in the clearanc e of nox ious material, the removal of migrating tumor ce lls, and in the pathophysiology of se ptic shock. 1 Liver mac rophages have be en re ported to secre te a w ide array of biologically active compounds including cytokines, prostanoids, nitric ox ide and ox ygen radicals, w hich have be en show n to be involved in the pathogenesis of septic shock, and in the cytotoxicity of Kupffer c ells against tumor target c ells. 1 -7 LPS and muramyl tripep tide s are c omp one nts of the outer c ell membrane of most bacte ria and display most of the immunologic al activitie s c omp ared w ith an infec tion of w hole bacte ria. 8 -1 1 Whe n administrate d i.v., LPS and liposome-enc apsulated N-ace tyl-muramyl-L-alanyl-D-isoglutaminyl-L-alanine -2-[19 , 29 dipalmitoyls n-glyce ro-39 -(hydrox y-phosphoryl-ox yl)] e thyl-amide (MTP-PE) 3 are delive red predominantly to mac rophage s, e spec ially liver mac rophages. 1 2,13 While LPS has bee n show n to p lay a significant role in the pathogenesis of gram-ne gative se ptic shock, muramyl tripe ptides have been demonstrated to be prote c tive against LPS-induc ed se ptic emia. 14 ,15 Furthe rmore , i.v. administe re d MTP-PE has bee n show n to induc e antitumor reactivity, p robably as a re sult of mac rophage ac tivation, 16 -1 9 and to inhibit HIV replication in macrophage s. 20 In contrast to LPS, ve ry little is know n about the intrace llular signal transduc tion pathw ays of muramyl tripe ptides. Re cently, w e show e d that MTP-PE and LPS induc e similar re ac tions in liver mac rophages, like the formation of TNF-a , nitric ox ide and prostanoids. 2 1 Furthe rmore , w e demonstrated, that c ellular c alcium and protein kinase C isoenzyme s are not involved in the signal pathw ays of LPS and MTP-PE. 2 1

Cell culture
Live rs of male Wistar rats (Charles River, Sulzfeld, Ge rmany) are remove d ase ptic ally under Ne mbutal anae sthesia, and the live r macrophage s are is olated by a ce ntrifugal elutriation p roc edure . 22 Liver mac rophage s are maintained in primary culture w ith RPMI 1640 me dium containing 30% new born calf serum. All ex pe riments are performed w ith ce lls kept in primary culture for 48 -72 h. P815 mouse mastocytoma cells and L929 c ells are grow n in RPMI 1640 me dium containing 10% FCS.

Determination of PGE 2 and PGD 2
Macrophages are incubated in RPMI medium containing 10% new born calf serum w ith or w ithout LPS or MTP-PE, and afte r the indic ate d times ce ll media are remove d and centrifuged. The amount of PGE 2 and PGD 2 in cell me dia is measured by ELISA and RIA, respective ly, as desc ribed e lsew here. 2 3 Determination of TNF-a activity The re lease of TNF-a activity is me asure d using the L929 ce ll cytotox ic ity assay in w hich lysis of actinomycin-D-treated L929 cells by TNF-a is measured by crystal viole t staining of survival c ells in monolayer culture. 2 4 L929 ce lls are grow n on 96-w ell microte st plates to the de nsity of ~4´10 4 c ells /w e ll. Culture me dia to be tested are added to the w ells together w ith actinomycin (400 ng /w ell). Afte r 24 h, the supernatants are discharge d and the re maining c ells are staine d w ith 0.5% c rystal violet in 25% aque ous me thanol for 10 min. The dye is re moved by w ashing three times w ith 25% me thanol, and 5% SDS is adde d to solubilize the adherent ce lls. Then the aborbance is read w ith a mic roplate reade r (Tite rtek Multiskan Plus) at 550 nm. The absorbance is c ompared w ith that of a standard solution of murine rTNF-a . Units of TNF-a activity are the rec ip roc al dilution fac tor of a sample c ausing 50% lysis of L929 c ells. The sp ecificity is c ontrolled by ne utralization w ith rat TNF-a -neutralizing antibodie s (dilution of 1:50).

Determination of nitric oxide
Macrophages are incubated in RPMI medium containing 10% new born calf serum w ith or w ithout LPS or MTP-PE, and afte r the indic ate d times ce ll media are remove d and ce ntrifuge d. Nitric ox ide formation is me asured as NO 2 ac cumulate d in the c ell media, w hich reflec ts the re le ase of nitric ox ide from macrophage s. 3 The Grie ss reaction using 20 m l of sulfanilamide (10 mM), 10 m l of HCl (2 M), 10 m l of naphtyle thyle nediamine dihydrochloride (1 mM), and 150 m l of supernatant is applie d to e ach w ell of a mic rotiter plate; the pink azo dye is quantitated by a mic roplate reade r (Tite rtek Multiskan Plus) at 550 nm. Standards are pre pare d using NaNO 2 -.

In vitro cytotoxicity assay
Macrophages are c ulture d on 96-w ell plates at a density of ~10 5 ce lls /w ell. P815 target cells in the ex ponential grow th phase are radiolabe led by a 24 h-inc ubation in RPMI medium w ith 10% FCS containing 20 m Ci of [ 3 H]thymidine /10 6 ce lls in 5 ml.
Then the ce lls are w ashed thre e times to re move the radiois otope and re suspe nde d in culture me dium to a conc entration of 10 5 ce lls /ml. 100 m l (10 4 [ 3 H]thymidine-labeled target cells) is added to c ultured macrophage s in a total volume of 200 m l of medium /w e ll. LPS or MTP-PE is added to mac rop hages 6 h before the addition of target c ells. Radiolabeled targe t c ells are also plated alone as an additional c ontrol. 48 h after the addition of target c ells, the supernatants are collected and the radioactivity is measured in a liquid sc intillation counte r. Cytolysis is c alculated as follow s: supe rnatants of target c ells cocultured w ith mac rophage s, b= cpm in supernatants of target cells c ultured alone, c= c pm in the total amount of targe t ce lls added per w ell.

Western blot analysis
Total prote in (10 m g) is separate d on 10% polyacrylamide gels unde r re ducing c onditions. Prote ins are transfe rre d to nitrocellulose me mbranes, and nonspe cific binding is blocked by inc ubation in PBS containing 0.1% Tw e en 20 and 5% milk pow de r. The plots are probed w ith antibodie s against COX-2 and iNOS. After ex te nsive w ashing, the blots are incubate d w ith the secondary antibodies, w ashe d ex te nsively and de ve lope d using the ECL We ste rn blotting dete ction syste m (Ame rsham). For quantitativ e analysis, the bands are sc anned densitometric ally.

In situ map-kinase assay
Map kinase ac tivity is performe d by an in vitro renaturation assay. 25

Electrophoretic mobility shift assay
De te rmination of the DNA-binding c apacity of NF-k B and AP-1 is p erformed ex actly as de sc ribe d previously. 2 7 RT-PCR studies RNA from macrophage s is prepared ac cording to Chomzynski and Sac chi. 28 In brief, 1-

Results
Release of TNF-a , nitric oxide and prostaglandins LPS and MTP-PE induce a synthe sis and release of TNF-a , nitric ox ide and prostanoids in mac rophages (Fig. 1). LPS induce s a rapid and transie nt ac cumulation of TNF-a , MTP-PE-induc ed TNF-a rele ase show s a lag phase of about 6 h, and inc reases the re afte r up to 48 h (Fig. 1A). The release of nitric ox ide (Fig. 1B) and PGD 2 (Fig. 1C) is almost identical for both agents. The re lease of PGE 2 after LPS is highe r than after MTP-PE (Fig. 1C).

mRNA and protein levels
In order to investigate at w hich levels LPS and MTP-PE induc e the formation of TNF-a , nitric ox ide and prostanoids, c orre sponding mRNA and prote in levels are de te rmined by RT-PCR and Weste rn blot analysis, respective ly. Figure 2 show s that unstimulate d mac rophage s contain ve ry small amounts of mRNAs encoding TNF-a , iNOS and COX-2. mRNAs encoding b -ac tin and COX-1 are ex pressed at much higher le vels (Fig.  2). LPS induce s a rapid and transie nt acc umulation of TNF-a mRNA w ith a max imum at about 1.5 h (Fig.  3A). In c ontrast, TNF-a mRNA after MTP-PE bec omes dete ctable only after a lag p hase of about 4 h. Both, LPS and MTP-PE induce a rapid acc umulation of iNOS mRNA (Fig. 3B). The leve l of iNOS mRNA afte r LPS peaks at about 4 h, and decline s thereafter. In contrast, MTP-PE induc es an incre ase of iNOS mRNA up to 12 h, w hich re mains thereafter elevate d at this level. COX-2 mRNA inc re ases rapidly to similar levels after the addition of LPS and MTP-PE (Fig. 3C). Re sting macrophage s show no dete ctable amounts of iNOS and COX-2 protein (Fig. 4). Addition of LPS and MTP-PE induc e an ex p re ssion of both p rote ins, w hich bec omes detectable at first after 6 -8 h. LPS induce s a higher ex pre ssion of iNOS protein (24 h), COX-2 prote in is higher ex pressed after MTP-PE.

Activation of ERK-1 and -2
Rec ently w e demonstrate d that c alcium and prote in kinase C isoenzymes are not involved in the actions of LPS and MTP-PE in mac rophages. 2 1 He re w e show (Fig. 5), that both agents induc e a rapid ac tivation of the map kinase isoenzymes ex trac ellular regulate d kinase (ERK)-1 and ERK-2.

Activation of NF-k B and AP-1
LPS induc es a rapid activation of the transcription factors NF-k B and AP-1 (Fig. 6). No ac tivation of both transc ription factors at this e arly time-point is se en w ith MTP-PE. Activation of NF-k B and AP-1 by MTP-PE bec omes dete ctable only after a lag phase of about 5 h, and is w eake r pronounc ed than w ith LPS.

Effect of inhibitors of ERK-1, -2 and NF-k B on the release of TNF-a , nitric oxide and PGE 2
Inhibition of the map kinase is oe nzyme s ERK-1 and ERK-2 by PD 98059 29 le ads to a dec re ased re le ase of TNF-a , nitric ox ide and PGE 2 ( Table 1). TNF-a re le ase after LPS and MTP-PE, and PGE 2 release afte r MTP-PE is   comple te ly inhibite d by PD 98059. The release of nitric ox ide afte r LPS and MTP-PE, and of PGE 2 afte r MTP-PE, is inhibited by PD 98059 by about 70%, 30% and 60%, respective ly ( Table 1). Inhibition of the transcription factor NF-k B by BAY 11-7082 3 0 has no e ffec t on the MTP-PE-induce d re le ase of TNF-a , nitric ox ide and PGE 2 (Table 1). How e ve r, the LPS-induced re lease of TNF-a and nitric ox ide is completely inhibited by BAY 11-7082. The release of PGE 2 after LPS is inhibited by BAY 11-7082 by about 60% ( Table 1).
Effect of PGE 2 and PGD 2 PGE 2 , ex ogenously adde d to macrophage s, inhibits TNF-a re le ase by LPS but not by MTP-PE (Table 2). In contrast, the release of nitric ox ide after LPS and MTP-PE is e nhanc ed by p rior addition of PGE 2 . PGD 2 is w ithout any effec t ( Table 2). In orde r to investigate at w hich level PGE 2 ex erts its ac tion, the effect of ex ogenously added PGE 2 on the ac tivation of map kinase isoenzyme s ERK-1 and ERK-2 is determine d. PGE 2 has no effec t on the MTP-PE-induc ed activation of ERK-1 and ERK-2 (Fig. 6). How e ve r, map kinase activation by LPS is completely suppressed.

Effect of LPS and MTP-PE on the cytotoxicity of macrophages
MTP-PE but not LPS activate s mac rop hages to cytotox ic ity against P815 tumor targe t ce lls (Fig. 7). Addition of neutralizing antibodie s against TNF-a reduced the c ytotox ic e ffec t of MTP-PE by about 75%, indicating that TNF-a is involved in the c ytotox ic reaction.

Discussion
He re w e show that LPS and MTP-PE induce a re le ase of TNF-a , nitric ox ide , PGE 2 and PGD 2 in mac rophage s. Both agents inc rease mRNA's e ncoding TNF-a , iNOS and COX-2 indic ating their ac tion at the transc riptional level. We de monstrated rec ently that macrophage s do not ex p re ss constitutive NOS, 3 1 w hich sugge sts that the observed release of nitric ox ide after LPS and MTP-PE is c atalysed by iNOS. The formation of prostanoids afte r LPS is probably due to the e nhance d ex pre ssion of COX-2 (Fig. 3C) and an enhance d ex pression of cytosolic phospholipase A 2 . 3 1 Sinc e MTP-PE has no e ffec t on c ytosolic phosp holipase A 2 , 31 prostanoid re le ase after MTP-PE see ms to be triggered only by the enhance d ex pression of COX-2. We c ould also de monstrate re ce ntly that LPS and  MTP-PE have no e ffec t on COX-1 le ve ls, and that the constitutiv e nitric ox ide synthase and sec retory phosholipase A 2 s are not ex pressed in these ce lls. 31 While the release of nitric ox ide and PGD 2 is almost identic al for both immunomodulators, the formation of TNF-a and PGE 2 after LPS and MTP-PE is differe nt.
(1) Acc umulation of TNF-a mRNA and TNF-a activity after LPS is transie nt, a sustained re le ase of TNF-a is induc ed by MTP-PE. The finding, that PGE 2 inhibits TNF-a re le ase after LPS, but not after MTP-PE (Table  2), indicate s that the mechanisms underlying TNF-a formation are different for both immunomodulators. It has bee n show n re cently that the inhibition of the LPS-induce d TNF-a re lease by PGE 2 can be mimicke d by cAMP, 4 sugge sting that intrace llular cAMP mediates the e ffec t of PGE 2 . (2) PGE 2 release afte r LPS is higher than after MTP-PE. It has bee n suggested re cently that the high PGE 2 formation afte r LPS might be due to an activation of the PGE 2 synthase. 3 2,3 3 We de monstrated re ce ntly that both immunomodulators do not induce an ac tivation of the 'phosphatidyl inositol c ycle ', of prote in kinase C is oe nzyme s or a change of the intrac ellular calc ium c oncentration. 2 1 He re, w e show that LPS and MTP-PE induc e a rapid activation of the map kinase isoenzyme s ERK-1 and ERK-2. Inhibition of map kinase by PD 98059 comple te ly suppresses TNF-a re lease after LPS and MTP-PE, and PGE 2 rele ase after MTP-PE. Re lease of nitric ox ide and PGE 2 afte r LPS are partially inhibited by PD 98059. These data indic ate that ac tivation of the map kinase isoenzymes ERK-1 and ERK-2 is an essential ste p for TNF-a re le ase by LPS and MTP-PE, and for PGE 2 formation by MTP-PE. Nitric ox ide release , and PGE 2 formation by LPS see m only to be partially me diated by map kinase . In addition, w e demonstrate he re, that the transc ription factors NF-k B and AP-1 are ac tivated by LPS. Ac tivation of both transc ription fac tors by LPS is obse rved already after 30 min, and persists for more than 24 h. 2 7 The activation of NF-k B and AP-1 by MTP-PE is much w eake r pronounce d, and bec ome s de te c table only after 5 h. Sinc e mRNAs encoding TNF-a , iNOS and COX-2 are ex pre sse d at much earlier time-points, it is very unlikely that NF-k B and AP-1 are involve d in these actions of MTP-PE. This c onclusion is supporte d by the finding that BAY 11-7072, an inhibitor for NFk B, has no effect on MTP-PE-induce d re sponse s. BAY 11-7072 completely suppresses the rele ase of TNF-a and nitric ox ide by LPS, and inhibits PGE 2 rele ase by LPS by 60%, indic ating an involve ment of NF-k B in these ac tions of LPS. The LPS-but not the MTP-PE-induc ed re le sase of TNF-a is sp ecifically inhibite d by PGE 2 . In contrast, nitric ox ide release by LPS and MTP-PE is e nhance d by PGE 2 . PGE 2 has been show n earlie r to ex ert its suppressing e ffec t on TNF-a re lease by LPS at the transc riptional le ve l. 4 How e ve r, PGE 2 doe s not affec t the activation of NF-k B or AP-1 by LPS. 27 He re w e show, that PGE 2 inhibits map kinase activation by LPS, but not by MTP-PE. This indicate s that the mechanisms of LPS and MTP-PE leading to ac tivation of map kinase are differe nt. MTP-PE, but not LPS, induc es a c ytotox icity of macrophage s against tumor target c ells. The MTP-PE induc ed cytotox ic ity is re duce d by TNF-a ne utralizing antibodie s, indicating that TNF-a is involve d in the cytotox ic action. The lack of cytotox icity after LPS might be due to the transie nt release of TNF-a , in contrast to the sustaine d ac cumulation of TNF-a afte r MTP-PE. These findings are in line w ith the fact, that MTP-PE, w hen administe re d i.v., induces an antitumor reactivity in animal models. 1 6 -20 Rec ently, it has be en proposed that the c ytotox icity of mac rophages against the adenomacarc inoma line MCA26 is me diated by TNF-a , and not by nitric ox ide . 34 Our data support the se findings, since LPS and MTP-PE induc e an ide ntic al re le ase of nitric ox ide , but differ in their cytotox ic pote nc ie s.
Base d on the results prese nte d here , w e propose the follow ing scheme for the re gulatory role of PGE 2 in the ac tion me chanisms of LPS and MTP-PE in macrophage s (Fig. 8). LPS (Fig. 8A) ac tivates rap idly the map kinase isoenzymes ERK-1 and ERK-2 and the transc ription factors NF-k B and AP-1, w hich results in a synthesis and release of nitric ox ide , TNF-a and PGE 2 . PGE 2 re duc es in a ne gative fee dback loop the activation of ERK-1 and ERK-2, nitric ox ide release is enhance d, TNF-a formation is suppressed, cytotoxicity against tumor targe t c ells is low. MTP-PE activates rapidly map kinase isoenzymes ERK-1 and ERK-2, but not the transcription factors NF-k B and AP-1. Activation of yet unknow n transc rip tion fac tors results in a synthe sis and release of nitric ox ide, TNF-a and PGE 2 . PGE 2 has no inhibitory effect on the map kinase isoenzymes ERK-1 and ERK-2, TNF-a re lease is not suppressed, and the macrophages show a high cytotox icity against tumor targe t cells. Furthe r ex perime nts have to be c arried out to e luc idate in de tail the molecular mechanisms of the re gulatory function of PGE 2 . These ex periments inc lude the inve stigation of othe r kinase s and other transc ription fac tors, 35 -3 8 , w hich have bee n de monstrated in other c ells to be involved in the ac tion of LPS, and in the formation of TNF-a , nitric ox ide and e icosanoids.