Effects of three anti-TNF-α drugs: Etanercept, infliximab and pirfenidone on release of TNF-α in medium and TNF-α associated with the cell in vitro

www.e l sev i e r. com/ l oca te / i n t imp

International Immunopharmacology (2008) 8, 679—687

Effects of three anti-TNF-α drugs: Etanercept,infliximab and pirfenidone on release of TNF-α inmedium and TNF-α associated with the cell in vitroK.J. Grattendick, J.M. Nakashima, L. Feng, S.N. Giri ⁎, S.B. Margolin

Solanan, Inc., Dallas, TX 75229, United States

Received 31 October 2007; received in revised form 14 January 2008; accepted 14 January 2008

⁎ Corresponding author. 11034 Shady75229. Tel.: +1 214 350 3313; fax: +1 2

E-mail address: [email protected]

1567-5769/$ - see front matter © 200doi:10.1016/j.intimp.2008.01.013

Abstract

Tumor necrosis factor-alpha (TNF-α) is a vital component of the inflammatory process and itsaberrant over-expression has been linked to numerous disease states. New treatment strategieshave sought to reduce circulating TNF-α, either with neutralizing anti-TNF-α binding proteinssuch as etanercept or via drugs that inhibit de novo TNF-α synthesis like pirfenidone. In thepresent study, we examined the effects of both classes of drugs on secreted and cell-associatedTNF-α produced by THP-1 cells. All of the tested drugs significantly reduced secreted levels ofbioactive TNF-α following stimulation with LPS as measured by bioassay. However, etanercept-treated cells had approximately six-fold higher levels of cell-associated TNF-α compared withthat of the LPS-alone treatment group. Surprisingly, LPS+ infliximab treated cells did notincrease cell-associated TNF-α relative to the LPS-alone treatment. Pirfenidone significantlyreduced both secreted and cell-associated TNF-α levels. These drug-related differences in cell-associated TNF-αmay have broad implications in the future for the therapeutic uses of anti-TNF-α drugs in the management of TNF-α diseases.© 2008 Elsevier B.V. All rights reserved.

KEYWORDSTumor necrosisfactor-alpha;Pirfenidone;Etanercept;Infliximab;Cytokines

1. Introduction

Tumor necrosis factor-alpha (TNF-α), a proinflammatorycytokine, orchestrates a number of inflammatory responses.The precursor form of TNF-α, transmembrane TNF-α (mTNF),is expressed as a 26-kilodalton type II polypeptide on the cellsurface of activated macrophages and lymphocytes as well asother cell types (endothelium). Membrane-bound mTNF is

Trail, Suite 108, Dallas, TX14 3350 3080.(S.N. Giri).

8 Elsevier B.V. All rights reserved.

subsequently cleaved by a metalloproteinase, TNF-α convert-ing enzyme (TACE) [1], which releases the secreted solubleform of TNF-α, a 17kD polypeptide. Homotrimerization isrequired for the biological activities of both forms of TNF-α.

Biological responses to TNF-α are mediated through twostructurally distinct receptors: type I and type II. Bothreceptors are transmembrane glycoproteins with multiplecysteine-rich repeats in the extracellular N-terminaldomains. Although their extracellular domains share bothstructural and functional homology, the intracellular domainsare distinct and transduce their signals through both over-lapping and distinct pathways. Secreted TNF-α binds to bothtype 1 receptor (TNFR1) and type II receptor (TNFR2), while

680 K.J. Grattendick et al.

the membrane-bound TNF-α binds mainly to type TNFR2[2—5]. Like TNF-α, these receptors exist both as a membrane-anchored form, where they mediate the pleiotropic patho-physiological effects of TNF-α, and in a soluble form, wherethey can bind and neutralize bioactive TNF-α [6—8].

While the biological effects of soluble TNF-α have beenwell studied, research continues with a focus on mTNF inter-actions. Transmembrane mTNF mediates various cytotoxicand inflammatory leukocytic functions via direct cell-cellcontact and binding of mTNF to TNFRs on target cells [9—11].Recently, the binding of TNFRs to mTNF was found to initiatereverse signaling (i.e., receptor-mediated ligand signaltransduction), which can alter the physiology of the mTNF-expressing cell. Binding of anti-TNF-α antibodies or solubleTNFRs to mTNF can phosphorylate an intracellular signalingdomain and may cause intracellular calcium changes [11,12],initiate synthesis and release of various cytokines [11,13,14],increase adhesion molecule expression [15], and alter thecellular response to inflammatory stimuli [16].

The central role of TNF-α in the pathogenesis of severalchronic inflammatory diseases has led to widespread use ofseveral anti-TNF agents to treat rheumatoid arthritis [17,18],Crohn's disease [19], psoriasis [20], ankylosing spondylitis[21,22], and Behcet's disease [23]. Three clinical trials havetreated multiple sclerosis with pirfenidone [24—26].

Three TNF-α binding proteins have been approved by theFDA for human use. Etanercept (Enbrel™) is a recombinantprotein of human soluble TNFR2 coupled to Fc portion ofhuman IgG. Infliximab (Remicade™) is a mouse-humanchimeric anti-human TNF-α antibody. Adalimumab (Humira™)is a human anti-human TNF-α antibody. All three either act asneutralizing antibodies of secreted TNF-α or absorb TNF-α,thus preventing further interactions with cell surface re-ceptors. However, none of these anti-TNF-α agents have anyreported effect on the synthesis of TNF-α. All are adminis-tered either intravenously or subcutaneously to be effective.

Pirfenidone [5-methyl-1-phenyl-2-(1H)-pyridone] is anovel, orally-effective anti-fibrotic drug and is currentlyundergoing Phase III clinical trial for idiopathic pulmonaryfibrosis (IPF). It has been demonstrated to inhibit TNF-αsynthesis at the translational level [27]. Oku, et al. found thattreatment with pirfenidone significantly reduced the LPS-induced elevation of TNF-α, IL-12, and IFN-γ in mice [28].Also, treatment with pirfenidone protected mice againstboth endotoxin-induced and Staphylococcus aureus enter-otoxin B-induced endotoxic shock as well as the concurrentelevation of serum TNF-α levels [29,30].

Multiple sclerosis (MS) is characterized by demyelination,progressive axonal loss in the white matter and subsequentneurological deficits. Several lines of evidence associateelevated TNF-α levels with demyelination and the progres-sive pathogenesis of MS [31—34]. The importance of TNF-α insecondary progressive multiple sclerosis (SPMS) was furthersubstantiated in two open-label studies [24,26] and onedouble-blind, randomized, placebo-control trial [25] where pirfeni-done arrested the progression of the disease and stabilized thecondition of the patients. In marked contrast, the use ofprotein-based anti-TNF-α drugs such as infliximab in patientswith rapidly progressive MS exacerbated the disease activityand forced the discontinuation of the trial [35]. Furthermore,administration of etanercept for treatment of juvenile rheu-matoid arthritis was associated with the onset of MS [36].

Wewere perplexed as to why two different classes of anti-TNF-α therapies appear to cause such different clinical out-comes in SPMS — exacerbation by protein-based anti-TNF-αdrugs versus the stabilization of MS by pirfenidone. With thisapparent paradox inmind and given the paucity of data in thisarea, we chose to study the effects of both drug classes on arelatively simple and well-studied system, LPS-stimulatedsecretion of TNF-α in the medium and TNF-α associated withthe cell in vitro.

2. Materials and methods

2.1. Reagents

All reagents were purchased from Sigma-Aldrich (St. Louis, MO)unless otherwise stated. Etanercept (Enbrel™) was purchased from apharmaceutical supplier in 1 cc syringes at 50 mg/ml and was used inboth native carrier and dialyzed (in PBS) forms with no differences inefficacy or toxicity between the two. Infliximab (Remicade™) wasalso purchased from a pharmaceutical supplier in a 100 mg vialcontaining the lyophilized drug which was reconstituted in sterilePBS prior to use. Heating (95 °C×10 min) of either etanercept orinfliximab stock solutions completely inactivated all TNF-α-neutra-lizing activity as assessed by TNF-α bioassay (data not shown).Pirfenidone (PFD) was a generous gift of Marnac, Inc. (Dallas, TX). Alldrugs and reagents used in our experiments were diluted in RPMI-1640 with 2% FBS and supplemented as described below.

2.2. In vitro cell treatment

THP-1 cells (ATCC, Manassas, VA), a mononuclear cell line which canbe differentiated into macrophage-like cells [37], were collectedfrom culture media by centrifugation at 250 ×g and resuspended at1×106 cells/ml in RPMI-1640 supplemented with 10% FBS (Hyclone,Logan, UT), 50 mg/ml gentamicin, 25 mM HEPES, 1.25 g/L sodiumbicarbonate, 100 μM 2-mercaptoethanol and 50 ng/ml phorbolmyristate acetate (PMA; to facilitate differentiation and celladherence to the culture plates). Five hundred μl of cell suspensionwere added to each well of a 24-well Costar plate (Thomas Scientific,Swedesboro, NJ) and allowed to incubate for 18 hours at 37 °C under5% CO2. After incubation, cells were washed 2× and fresh media(without PMA) were added and cells were allowed to incubate for anadditional 24 h to abate the effects of PMA on cell activation. Mediawere then removed and cells treated with LPS, PFD, etanercept (ET),or infliximab (IN) dissolved in RPMI for 3 h. Following incubation,culture media and cell lysates (to be discussed later) were collectedand analyzed for TNF-α by bioassay or ELISA, respectively.

2.3. TNF-α bioassay

TNF-α was measured in culture media using a modification of theTNF-α bioassay utilizing Wehi-164 var13 cells (ATCC) as describedpreviously [38]. The modification of the protocol involved replacingneutral red staining with the more sensitive and reproducible solubleformazan dye-based viability assay to determine TNF-α mediatedcytotoxicity. Briefly, Wehi-164 cells were plated at 1.5×104 per wellin Falcon Primaria 96-well plates (Thomas Scientific) and allowed toadhere. After 6 h, culture media from drug-treated THP-1 cells (asdescribed previously) were centrifuged to remove cell debris andadded to the Wehi-164 cells. Serial 1:2 dilutions were performed in-well for all samples. Serial dilutions of recombinant human TNF-αstandard (3.9–500 pg/ml; BD Biosciences, San Jose, CA) were alsoincluded to determine the quantity of TNF-α in the experimentalsamples. Finally, 2 μg/ml actinomycin-D were added to each welland the cells were incubated overnight. After 20 h, media wereremoved and cell viability was determined by MTS assay (Promega,

681Effects of three anti-TNF-α drugs

Madison, WI) using manufacturer's instructions. Absorbance at492 nm was measured with a Thermo Multiskan EX (Thermo FisherScientific, Inc, Waltham, MA). Levels of TNF-α in experimentalsamples were calculated based on the formula derived from thestandard curve for the recombinant TNF-α standard.

2.4. Cell lysate analysis for TNF-α by ELISA

After media were removed from THP-1 cells as described in theprevious section, the adherent cells were washed 3× in PBS and thenincubated at 4 °C for 15 min with lysate buffer (10 mM HEPES, 1 mMEDTA, 60 mM KCl, 0.5% NP-40, 1 mM DTT, 1 mM PMSF) containingprotease inhibitor cocktail. After 2 freeze-thaws, lysate mixtureswere collected and centrifuged at 10,000×g at 4 °C and supernatantscollected and stored at −80 °C until analyzed for TNF-α by ELISA.Immediately prior to TNF-α quantification, supernatants from eachsample group were normalized for protein content as determined byBCA kit (Promega). The TNF-α OptEIA ELISA kit was purchased fromBD BioSciences and the protocol used in this study was permanufacturer's instructions. TMB conversion was measured at450 nm and 540 nm on a Thermo Multiskan EX plate reader.

It was necessary to measure secreted TNF-α and cell-associatedTNF-α by different assays. The current investigators measuredsecreted TNF-α using a bioassay because etanercept is a neutralizingbiologic that does not remove the TNF-α molecules from thesolution. The presence of the etanercept-bound TNF-α was still ableto bind to the antibodies utilized in the ELISA and so thereforeresulted in aberrantly high measurements. However, infliximab wasable to block the epitopes on TNF-α that were specific to the ELISAantibodies. These two findings are consistent with reports byScallon, et al. on the binding properties of the two TNF-α antagonists[39]. The bioassay measures biologically available TNF-α only and isbetter suited for measuring cytokines in the presence of neutralizingagents. The cell lysate samples were also tested for TNF-α bybioassay, but it was not possible to determine cytotoxicity effectsattributed to the presence of TNF-α versus that attributed to thepresences of residual lysis buffer. Attempts were made to dialyze thelysis buffer against an isotonic saline solution, but this introducedmore variables that further confounded the results.

2.5. Apoptosis analysis

THP-1 cells were incubated in 6-well plates with RPMI supplementedwith PMA at 1×106 cells/ml for 12 h. Fresh media without PMA werethen added for 24 h. Next, cells were incubated for 18 h with controlmedia, 1 ng/ml LPS, LPS plus 200 μg/ml pirfenidone, LPS plus 0.1 μg/ml or 1 μg/ml etanercept, 200 μg/ml pirfenidone alone, 0.1 μg/mletanercept alone, or anti-FAS mAb as a positive control. Cell lysateswere then collected and analyzed for caspase-3 activity using

Figure 1 Effects of pirfenidone or etanercept on THP-1 cell viabilitwith 1 ng/ml LPS alone, 1 ng/ml LPS plus the indicated concentrationof etanercept for 6 h. A MTS assay was used to determine cell viabsimultaneously compared via one-way ANOVA and Tukey multiple co

CaspACE Colorimetric Assay kit (Promega) per manufacturer'sinstructions. The absorbance was measured at 405 nm on a ThermoMultiskan EX plate reader.

2.6. Immunocytochemistry

THP-1 cells were cultured at 5×104 cells/well on 16-well glasschamber slides with PMA as described above. After an incubation of24 h with fresh media, cells were treated with control media, LPS at1 ng/ml, LPS and pirfenidone, LPS and etanercept, or etanerceptalone. The media were collected 3 h later and cells were fixed usingBouin's solution. The slides were treated to quench endogenousperoxidases, blocked with normal serum and then stained witheither monoclonal mouse anti-human TNF-α (R&D Systems, Minnea-polis, MN) diluted 1:50 or a positive control mouse anti-humanpolyclonal IgG for one hour at room temperature. Next, the boundantibodies were detected via a VectaStain ABC anti-mouse IgG-peroxidase kit and diaminobenzidine (DAB) followed by a hematox-ylin counterstain (Vector Labs, Burlingame, CA). After cover slipswere mounted, the slides were photographed using a computer-controlled Zeiss Axiovert 100 M digital light microscope and 40X oil-emersion magnification.

2.7. Statistical analysis

Data are expressed as the mean±SEM for at least three replicates.Statistical differences between LPS treatment alone and variousother treatment groups were analyzed using one-way ANOVA withTukey's multiple comparison post-test and a value of Pb0.05 wasconsidered to be the minimum level of statistical significance. ForFig. 4, the same statistical comparison was also performed betweenLPS+ET and LPS+ET+PFD. The statistical software utilized in thisstudy was Graph Pad's Prism for Apple Macintosh, version 4.0c.

3. Results

3.1. Effects of pirfenidone or anti-TNF-α proteins oncellular cytotoxicity in vitro

The highest concentration of pirfenidone, etanercept orinfliximab that did not affect cellular viability was determinedby incubating THP-1 cells with or without 1 ng/ml LPS combinedwith 1 μg/ml to 500 μg/ml pirfenidone, 0.001 μg/ml to 100 μg/ml etanercept, or 0.01 μg/ml to 100 μg/ml infliximab in 96-wellplates for 3, 6, 9, or 24 h. Cellular viability was determined byMTS assay (Fig. 1) and trypan blue exclusion (data not shown). InFig. 1, the results for 6 h incubation are shown as all experiments

y. PMA-transformed THP-1 cells were incubated on 96-well platesof pirfenidone, or 1 ng/ml LPS plus the indicated concentrationility. Values represent mean±SEM. All treatments groups weremparison test. ⁎Pb0.05, ⁎⁎⁎Pb0.001 versus LPS alone.

Figure 2 Dose response of pirfenidone, etanercept, orinfliximab on secreted (A) or cell-associated (B) TNF-α levels.PMA-transformed THP-1 cells were incubated for 3 h on 96-wellplates with culture media, 1 ng/ml LPS, or 1 ng/ml LPS plusvarious concentrations of either pirfenidone, etanercept, orinfliximab. Pirfenidone: Conc 1=10 μg/ml, Conc 2=33 μg/ml,Conc 3=100 μg/ml, Conc 4=200 μg/ml, and Conc 5=300 μg/ml.Etanercept: Conc 1=0.001 μg/ml, Conc 2=0.01 μg/ml, Conc3=0.1 μg/ml, Conc 4=1.0 μg/ml, and Conc 5=10 μg/ml.Infliximab: Conc 1=0.01 μg/ml, Conc 2=0.1 μg/ml, Conc3=1.0 μg/ml, Conc 4=10 μg/ml, and Conc 5=100 μg/ml. Culturemedia (A) were analyzed for secreted TNF-α via bioassay andlysates (B) were analyzed for cell-associated TNF-α via ELISA.Values represent mean±SEM. All treatments groups weresimultaneously compared via one-way ANOVA and Tukey multiplecomparison test. ⁎⁎⁎Pb0.001 versus LPS alone. ND=not detected.

682 K.J. Grattendick et al.

were conducted within this time-frame; however, earlier andlater time-points showed similar results. For pirfenidone,concentrations below 300 μg/ml showed no effects on cellularviability for any incubation period. Two hundred fifty ng/ml ofetanercept was the highest concentration found to not affectTHP-1 cell viability. Infliximab had no significant (PN0.05) effecton cell viability at any tested concentrations when compared tocontrols (data not shown).

To further confirm these results, caspase-3 activity (anindicator of apoptosis) was investigated with THP-1 cellsincubated with control media, 1 ng/ml LPS, 1 ng/ml LPS plus100 μg/ml or 200 μg/ml pirfenidone, 1 ng/ml LPS plus 0.1 μg/mlor 1 μg/ml etanercept, 100 μg/ml or 200 μg/ml pirfenidonealone, 1 μg/ml etanercept alone, or anti-FAS mAb as a positivecontrol for 18 h. The caspase-3 activity was determined using acommercially available colorimetric assay. There were nosignificant (PN0.05) differences in caspase-3 activity betweenLPS-treated cells and other treatment groups, which paralleledthe results from an MTS assay (data not shown). There was,however, a slight increase observed when cells were exposed toLPS versus those exposed to control media or drug without LPSchallenge. This, again, mirrored results obtained when examin-ing cell viability using the MTS assay.

3.2. Effects of pirfenidone or anti-TNF-α proteins onsecreted and cell-associated TNF-α In vitro

The effects of each drug at various concentrations wereexamined on 1 ng/ml LPS-induced TNF-α secretion and cell-associated TNF-α. The results of this study revealed thatpirfenidone at 200 μg/ml had a maximal inhibitory effect onboth secreted and cell-associated TNF-α without compromisingcell viability (Fig. 2A and B). Secreted TNF-α levels from LPS-treated and etanercept- or in infliximab-exposed THP-1 cellswere maximally neutralized at concentrations of 0.01 μg/ml and0.1 μg/ml, respectively (Fig. 2A). With respect to cell-associatedTNF-α, etanercept maximally increased TNF-α levels at 0.1 μg/ml and gradually declined at concentrations above that(Fig. 2B). However, it was consistently found that cells treatedwith any concentration of etanercept plus LPS resulted in higherlevels of TNF-α than LPS only treated cells. Infliximab did notshow any effects on cell-associated TNF-α levels at any dose onLPS-treated cells as compared to LPS treatment alone (Fig. 2B).

To compare the effects of pirfenidone or etanercept on LPS-induced TNF-α levels in media from THP-1 cells, cells werestimulated in 24-well plates with media alone, 1 ng/ml LPS, 1 ng/ml LPS plus 200 μg/ml pirfenidone, 1 ng/ml LPS plus 0.1 μg/mletanercept, 1 ng/ml LPS plus 0.1 μg/ml heat-inactivatedetanercept, 200μg/ml pirfenidonealone, or 0.1μg/ml etanerceptalone. At 200 μg/ml, pirfenidone reduced LPS-induced secretedTNF-α as measured by bioassay by a factor of approximately 2(Fig. 3A). The TNF-α detected in the culture media from cellsincubated with etanercept and LPS was not significantly (PN0.05)different from that measured in media from untreated cells,indicating complete neutralization of secreted TNF-α (Fig. 3A).Heat-inactivated etanercept did not reduce LPS-induced TNF-αwhen compared to cells treated with LPS alone. Infliximab, achimeric anti-TNF-α Ab, produced results similar to etanercept inits ability to neutralize TNF-α (Fig. 3B).

Cell-associated TNF-α was measured by ELISA. Lysates fromcells exposed to 1 ng/ml LPS and 200 μg/ml pirfenidonecontained approximately 33% of the TNF-α compared to cells

treated with LPS alone (Fig. 4A). However, there was anapproximate 650% increase in the amount of TNF-α measuredin lysates from cells treated with 0.1 μg/ml etanercept and 1 ng/ml LPS compared to lysates from LPS-only treated cells (Fig. 4A).The addition of pirfenidone to LPS and etanercept treated cellsreduced the cell-associated TNF-α by approximately 20%compared to LPS plus etanercept. Heat-inactivation of etaner-cept returned TNF-α levels to that seen in lysates derived fromcells treated with LPS alone. Incubating cells in the presence of

Figure 4 Comparison of pirfenidone and etanercept (A) orinfliximab (B) on cell-associated TNF-α. PMA-transformed THP-1cells were incubated for 3 h on 96-well plates with culturemedia, 1 ng/ml LPS, 1 ng/ml LPS plus 200 μg/ml PFD, and one ofthe following treatments: (A) 1 ng/ml LPS plus 0.1 µg/ml ET,1 ng/ml LPS plus 0.1 µg/ml ET and 200 µg/ml PFD, 1 ng/ml LPSplus 0.1 µg/ml heat-inactivated (HI) ET; (B) 1 ng/ml LPS plus1.0 µg/ml IN, or 1 ng/ml LPS plus 1.0 µg/ml heat-inactivated (HI)IN. Media were then removed, cells were washed and treatedwith lysate buffer. Lysates were analyzed for cell-associatedTNF-α via ELISA. Values represent mean±SEM. All treatmentsgroups were simultaneously compared via one-way ANOVA andTukey multiple comparison test. ⁎Pb0.05, ⁎⁎⁎Pb0.001 versus LPSalone. †††Pb0.001 between LPS+ET and LPS+ET+PFD.

Figure 3 Comparison of pirfenidone and etanercept (A) orinfliximab (B) on secreted TNF-α. PMA-transformed THP-1 cellswere incubated for 3 h on 96-well plates with culture media,1 ng/ml LPS, 1 ng/ml LPS plus 200 μg/ml PFD, 1 ng/ml LPS plus0.1 μg/ml etanercept (ET) (A), 1 ng/ml LPS plus 0.1 μg/ml heat-inactivated (HI) ET (A), 1 ng/ml LPS plus 1.0 μg/ml infliximab(IN) (B), or 1 ng/ml LPS plus 1.0 μg/ml heat-inactivated IN (B).Culture media were analyzed for secreted TNF-α via bioassay.Values represent mean±SEM. All treatments groups weresimultaneously compared via one-way ANOVA and Tukey multiplecomparison test. ⁎⁎⁎Pb0.001 versus LPS alone.

683Effects of three anti-TNF-α drugs

LPS and infliximab did not change cell-associated TNF-αcompared to cells exposed to LPS alone (Fig. 4B).

3.3. Immunocytochemical analysis of cell-associatedTNF-α

Immunocytochemical analysis of cell-associated TNF-α wasperformed for a visual confirmation of the ELISA data. Phorbolester-transformed THP-1 cells were cultured on 16-well glasschamber slides with one of the following: media alone, LPS,pirfenidone plus LPS, etanercept plus LPS, heat-inactivatedetanercept plus LPS, pirfenidone alone, or etanercept alone.Infliximab was not included because it showed no effects on cell-associated TNF-α in the previous experiments. After 3 h, thecells were fixed and stained with anti-TNF-α, developed withDAB and counterstained with hematoxylin. For each treatmentgroup, secreted TNF-α levels were measured by bioassay asdescribed previously and showed similar levels of TNF-α to thosepresented in Fig. 3 (data not shown), indicating that the cellswere responding consistently to the treatments.

Control cells showed limited to no staining for TNF-α,whereas LPS-treated cells showed enhanced staining that wasprimarily perinuclear (Fig. 5A and B). There was no discernibledifference in TNF-α staining between LPS-only and LPS pluspirfenidone treated cells (Fig. 5B and C), perhaps due to the lack

of sensitivity inherent in this assay. However, LPS and etanercepttreated cells showed a greater intensity of TNF-α staining with amore diffuse distribution (Fig. 5D), confirming the results fromELISA analysis. There were no differences between controlmedia cells and those exposed to etanercept without concurrentLPS treatment (Fig. 5E). There were also no differences betweencells treated with LPS alone and heat-inactivated etanerceptplus LPS (data not shown).

4. Discussion

Chronic inflammation has been found to be at the heart ofmany diseases, some of which were not initially recognizedas having an inflammatory component [40]. As a key earlymediator of the inflammatory process, TNF-α has become amajor target of numerous pharmaceutical investigations,which have yielded several unique proteins that bind andneutralize TNF-α bioactivity. Because of their relativenovelty and recent development, little is known about thelong-term consequences of anti-TNF-α biologics or whateffects they might have at the cellular level. This study

Figure 5 Localized TNF-α-specific staining in drug-treated THP-1 cells. PMA-transformed THP-1 cells were incubated for 3 h on glasschamber slides with culture media (A), 1 ng/ml LPS (B), 1 ng/ml LPS plus 200 µg/ml PFD (C), 1 ng/ml LPS plus 0.1 µg/ml ET (D), or0.1 µg/ml ETwithout LPS (E). Cells were stained with a mouse anti-human TNF-α antibody, amplified by an avidin-biotin-peroxidasesystem, and the chromogen DAB (brown color). Negative and positive staining controls are shown in 5F and 5G, respectively. Aftercounterstaining with hematoxylin, the cells were photographed at a final magnification of 40×. The bar represents 50 µm.

684 K.J. Grattendick et al.

examined the effects of two anti-TNF-α biologics, etaner-cept and infliximab, on LPS-induced TNF-α and comparesthem to those of pirfenidone, an anti-inflammatory drugwhich inhibits TNF-α synthesis at the translational level [27].

Neither of the two anti-TNF-α biologics tested in our studywere capable of inhibiting cell-associated TNF-α. In fact,exposure of LPS-treated THP-1 cells to etanercept, but notinfliximab, produced a dramatic six-fold increased cell-associated TNF-α compared to cells treated with LPS alone.This finding was subsequently confirmed visually when cellswere stained with anti-TNF-α Ab.

The unique property of etanercept to increase the level ofcell-associated TNF-α following LPS stimulation as demon-strated in the present in vitro study agrees with the in vivofindings of Mohler, et al. [41] who observed that serum TNF-α

levels were elevated in mice treated with monomeric sTNFRor low doses of a synthetic dimeric sTNFR:Fc (similar instructure to that of etanercept) following LPS treatment ascompared to LPS treatment only.

Although it is difficult to extrapolate beyond the etaner-cept-specific in vitro results presented here, this findingsuggests a possible mechanism for the adverse effects ofprotein based anti-TNF-α therapy in TNF-α mediateddiseases. This hypothetical mechanism is further supportedby the in vivo finding of Solorzano, et al. [42] that cell-associated TNF-α plays a critical role in the hepatocellularnecrosis and apoptosis that accompany D-galactosamine/LPS-or Con A-induced hepatitis.

Pirfenidone caused significant reductions in both secretedand cell-associated TNF-α and partially reversed the

685Effects of three anti-TNF-α drugs

increased cell-associated TNF-α levels following treatmentwith both etanercept and LPS as compared to cells treatedwith only etanercept and LPS. The ability of pirfenidone toreduce LPS-induced TNF-α release fromTHP-1 cells is entirelyconsistent with earlier in vitro studies with peritonealmacrophages, human blood monocytes and peripheral bloodlymphocytes as well as its in vivo protective effects againstendotoxin-induced shock [28—30]. These results are alsoconsistent with the inhibitory effects of pirfenidone onsynthesis of TNF-α at the translational level [27].

It remains unclear if the TNF-α present in cell lysates frometanercept-treated cells was complexed with etanercept,which might inhibit the release of TNF-α from the cellmembrane. Compared to infliximab, etanercept has beenreported to form less stable complexes with both membraneand soluble TNF-α, and this may lead to an abnormally highamount of unbound and bio-available ligand [39]. Also, Xinet al. showed that cells, which had been previously pre-treated with soluble TNFR (analogous to etanercept treat-ment), were then primed for subsequent TNF-α activationthrough reverse signalling [43]. These two findings may helpexplain why etanercept elevated cell-associated TNF-αlevels, but infliximab did not. In the present study, whenTHP-1 cells were simultaneously treated with etanercept andLPS, the binding of etanercept to transmembrane TNF-αmayhave primed the cells for additional activation from the LPS-induced release of soluble TNF-α. Furthermore, because ofthe reported instability of the TNF-α-etanercept complexes,more of the soluble TNF-α would be available for furtherstimulation of cell surface receptors. Additional studies areneeded to determine the exact reason for the increase inTNF-α associated with cells exposed to LPS and etanercept.

Based on the results from our cytotoxicity experiments, itdoes not appear that the increased levels of cell-associatedTNF-α following treatment of THP-1 cells with etanerceptand LPS had any demonstrable effects on cell viability.However, it is plausible that in situ cells laden with TNF-α ontheir membranes might be capable of affecting the physiol-ogy of neighboring or even distant cells through cell-cellinteractions [10,11]. Cells that express excess mTNF couldhave enhanced pro-inflammatory responses via reversesignalling beyond elevated TNF-α production. For example,reverse signalling through TNF-α has been shown to inducevarious pro-inflammatory cytokines [11] and cell adhesionmolecules [15], none of which would be reversed by protein-based TNF-α antagonists. However, by down-regulating theproduction of TNF-α, pirfenidone would be capable ofreducing the forward and reverse signaling associated withTNF-α, thus allowing it to short-circuit the multifactorialinflammatory cascade at its origin.

Tumor necrosis factor-α is an important mediator in thedevelopment and exacerbation of both relapsing-remittingand secondary progressive MS[44]. Circulating immune cellswith elevated levels of cell-associated TNF-α due toexposure to etanercept treatment might be involved in theactivation of other immune cells. These pre-primed inflam-matory cells might then cross the blood brain barrier (BBB),whose integrity has been shown to be compromised in SPMS[45], via the peripheral blood vessels of the CNS and furtherexacerbate the lesions and clinical signs of SPMS. Alterna-tively, protein-based anti-TNF-α drugs, which would nor-mally not cross the BBB, may enter the CNS from peripheral

blood or be carried in on infiltrating leukocytes, act directlyupon resident inflammatory cells and subsequently therebyexacerbate SPMS. A higher rate of TNF-α dissociation foretanercept compared to infliximab could also be involved inthe association of etanercept and MS. Reports suggests thatsoluble TNFR could act as a carrier for TNF-α resulting inprotection from degradation and an increase in the serumhalf-life, thus providing a prolonged, low-level exposure ofendothelial cells in the cerebral vasculature to TNF-α[41,46].

In contrast to the apparent adverse effects of protein-based anti-TNF-α therapies, three separate clinical trialshave demonstrated the beneficial effects of pirfenidone inpatients suffering from secondary progressive MS [24—26].Pirfenidone, a small (MW 185), relatively lipophilic molecule,readily crosses the BBB and is uniformly distributed in thebody [47]. In addition, pirfenidone is known to have anti-oxidant activity by scavenging reactive oxygen species[48,49] known to be elevated in a number of acute andchronic inflammatory diseases including MS [50]. It alsoinhibits NF-κB activation, which may further contribute to itsanti-inflammatory activity [51], and thus providing addi-tional benefits to MS patients. However, additional researchis needed to determine if a link between elevated cell-associated TNF-α levels and the infiltration of activatedleukocytes into the central nervous system is in fact presentin patients suffering from secondary progressive MS and whoare on etanercept therapy.

Complete neutralization of TNF-α bioactivity has beendescribed in the literature as one of the most importantbreakthroughs in the treatment of TNF-α-mediated chronicinflammatory diseases such as rheumatoid arthritis [17,18],Crohn's disease [19], psoriasis [20], ankylosing spondylitis[21,22], and Behcet's disease [23]. However, the crucial roleof TNF-α in mobilizing the host immune response to acuteand chronic infectious diseases may be of equal or greaterimportance. Patients receiving TNF-α neutralizing therapiesare at greater risk for the reactivation of latent tuberculosis,which may be due to a reduction in TNF-α-activatedmacrophages at the site of infection [52]. Treatment withetanercept and other anti-TNF-α biologics have also beenassociated with increased levels of anti-dsDNA antibodiesresulting in an increased incidence of systemic lupuserythematosus [53,54]. Thus, one could hypothesize thatlimiting TNF-α release instead of completely neutralizingTNF-α bioactivity might allow for the reduction or elimina-tion of acute or chronic inflammation without compromisinghost defenses against infection.

The present study reveals for the first time thatetanercept increased the level of LPS-stimulated cell-asso-ciated TNF-α by several fold over the LPS treatment alone inthe THP-1 cell line. The cause for this marked increase in cell-associated TNF-α remains unclear as well as the reason whythis phenomenon was not observed when cells were treatedwith another anti-TNF-α biologic, infliximab. The differencesin their chemical composition and properties (etanercept is asoluble TNF-α receptor, while infliximab is a chimericantibody), binding avidity and stability of the bound complexmay contribute to the observed differences. Thus moreresearch is warranted to elucidate this peculiar finding.Further studies are also needed to determine if thephenomena presented here are present in other cell types.

686 K.J. Grattendick et al.

Regardless, it is suggested from the number of adverse effectsfollowing treatment with anti-TNF-α biologics that the reduc-tion, but not complete neutralization, of TNF-α bioactivityafforded by drugs like pirfenidonemay be a better therapeuticstrategy for themanagement of chronic inflammatory diseaseswithout disrupting TNF-α-dependent host immune defensesagainst latent and incipient opportunistic infections.

References

[1] Black RA, Rauch CT, Kozlosky CJ, Peschon JJ, Slack JL, WolfsonMF, et al. A metalloproteinase disintegrin that releases tumour-necrosis factor-alpha from cells. Nature 1997;385:729—33.

[2] Decoster E, Vanhaesebroeck B, Vandenabeele P, Grooten J,Fiers W. Generation and biological characterization of mem-brane-bound, uncleavable murine tumor necrosis factor. J BiolChem 1995;270:18473—8.

[3] Grell M, Douni E,Wajant H, LohdenM,ClaussM,Maxeiner B, et al.The transmembrane form of tumor necrosis factor is the primeactivating ligand of the 80 kDa tumor necrosis factor receptor.Cell 1995;83:793—802.

[4] Grell M, Wajant H, Zimmermann G, Scheurich P. The type 1receptor (CD120a) is the high-affinity receptor for solubletumor necrosis factor. Proc Natl Acad Sci U S A 1998;95:570—5.

[5] Wallach D, Varfolomeev EE, Malinin NL, Goltsev YV, KovalenkoAV, Boldin MP. Tumor necrosis factor receptor and Fas signalingmechanisms. Annu Rev Immunol 1999;17:331—67.

[6] Old LJ. Tumor necrosis factor (TNF). Science 1985;230:630—2.[7] Vandenabeele P, Declercq W, Beyaert R, Fiers W. Two tumour

necrosis factor receptors: structure and function. Trends CellBiol 1995;5:392—9.

[8] Vassalli P. The pathophysiology of tumor necrosis factors. AnnuRev Immunol 1992;10:411—52.

[9] Caron G, Delneste Y, Aubry JP, Magistrelli G, Herbault N, BlaeckeA, et al. Human NK cells constitutively express membrane TNF-alpha (mTNFalpha) and present mTNFalpha-dependent cyto-toxic activity. Eur J Immunol 1999;29:3588—95.

[10] Decker T, Lohmann-Matthes ML, Gifford GE. Cell-associatedtumor necrosis factor (TNF) as a killing mechanism of activatedcytotoxic macrophages. J Immunol 1987;138:957—62.

[11] Higuchi M, Nagasawa K, Horiuchi T, Oike M, Ito Y, Yasukawa M,et al. Membrane tumor necrosis factor-alpha (TNF-alpha)expressed on HTLV-I-infected T cells mediates a costimulatorysignal for B cell activation–characterization of membrane TNF-alpha. Clin Immunol Immunopathol 1997;82:133—40.

[12] Watts AD, Hunt NH, Wanigasekara Y, Bloomfield G, Wallach D,Roufogalis BD, et al. A casein kinase I motif present in thecytoplasmic domain of members of the tumour necrosis factorligand family is implicated in 'reverse signalling'. Embo J 1999;18:2119—26.

[13] Ferran C, Dautry F, Merite S, Sheehan K, Schreiber R, Grau G,et al. Anti-tumor necrosis factor modulates anti-CD3-triggeredT cell cytokine gene expression in vivo. J Clin Invest 1994;93:2189—96.

[14] Waetzig GH, Seegert D, Rosenstiel P, Nikolaus S, Schreiber S.p38 mitogen-activated protein kinase is activated and linked toTNF-alpha signaling in inflammatory bowel disease. J Immunol2002;168:5342—51.

[15] Harashima S, Horiuchi T, Hatta N, Morita C, Higuchi M, SawabeT, et al. Outside-to-inside signal through the membrane TNF-alpha induces E-selectin (CD62E) expression on activatedhuman CD4+ T cells. J Immunol 2001;166:130—6.

[16] Eissner G, Kirchner S, Lindner H, Kolch W, Janosch P, Grell M,et al. Reverse signaling through transmembrane TNF confersresistance to lipopolysaccharide in human monocytes andmacrophages. J Immunol 2000;164:6193—8.

[17] Bathon JM, Martin RW, Fleischmann RM, Tesser JR, Schiff MH,Keystone EC, et al. A comparison of etanercept and metho-trexate in patients with early rheumatoid arthritis. N Engl JMed 2000;343:1586—93.

[18] Lipsky PE, van der Heijde DM, St Clair EW, Furst DE, BreedveldFC, Kalden JR, et al. Infliximab and methotrexate in thetreatment of rheumatoid arthritis. Anti-Tumor Necrosis FactorTrial in Rheumatoid Arthritis with Concomitant Therapy StudyGroup. N Engl J Med 2000;343:1594—602.

[19] Targan SR, Hanauer SB, van Deventer SJ, Mayer L, Present DH,Braakman T, et al. A short-term study of chimeric monoclonalantibody cA2 to tumor necrosis factor alpha for Crohn's disease.Crohn's Disease cA2 Study Group. N Engl J Med 1997;337:1029—35.

[20] Mease PJ, Goffe BS, Metz J, VanderStoep A, Finck B, Burge DJ.Etanercept in the treatment of psoriatic arthritis and psoriasis:a randomised trial. Lancet 2000;356:385—90.

[21] Braun J, Brandt J, Listing J, Zink A, Alten R, Golder W, et al.Treatment of active ankylosing spondylitis with infliximab: a ran-domised controlled multicentre trial. Lancet 2002;359:1187—93.

[22] Gorman JD, Sack KE, Davis Jr JC. Treatment of ankylosingspondylitis by inhibition of tumor necrosis factor alpha. N Engl JMed 2002;346:1349—56.

[23] Munoz-Fernandez S, Hidalgo V, Fernandez-Melon J, SchlinckerA, Martin-Mola E. Effect of infliximab on threatening panuveitisin Behcet's disease. Lancet 2001;358:1644.

[24] Bowen JD, Maravilla K, Margolin SB. Open-label study ofpirfenidone in patients with progressive forms of multiplesclerosis. Mult Scler 2003;9:280—3.

[25] Walker JE, Giri SN, Margolin SB. A double-blind, randomized,controlled study of oral pirfenidone for treatment of secondaryprogressive multiple sclerosis. Mult Scler 2005;11:149—58.

[26] Walker JE, Margolin SB. Pirfenidone for chronic progressivemultiple sclerosis. Mult Scler 2001;7:305—12.

[27] Nakazato H, Oku H, Yamane S, Tsuruta Y, Suzuki R. A novel anti-fibrotic agent pirfenidone suppresses tumor necrosis factor-alpha at the translational level. Eur J Pharmacol 2002;446:177—85.

[28] Oku H, Nakazato H, Horikawa T, Tsuruta Y, Suzuki R. Pirfenidonesuppresses tumor necrosis factor-alpha, enhances interleukin-10 and protects mice from endotoxic shock. Eur J Pharmacol2002;446:167—76.

[29] Cain WC, Stuart RW, Lefkowitz DL, Starnes JD, Margolin S,Lefkowitz SS. Inhibition of tumor necrosis factor and subse-quent endotoxin shock by pirfenidone. Int J Immunopharmacol1998;20:685—95.

[30] Hale ML, Margolin SB, Krakauer T, Roy CJ, Stiles BG. Pirfenidoneblocks the in vitro and in vivo effects of staphylococcalenterotoxin B. Infect Immun 2002;70:2989—94.

[31] Akassoglou K, Bauer J, Kassiotis G, Lassmann H, Kollias G,Probert L. Transgenic models of TNF induced demyelination.Adv Exp Med Biol 1999;468:245—59.

[32] Hofman FM, Hinton DR, Johnson K, Merrill JE. Tumor necrosisfactor identified inmultiple sclerosis brain. J ExpMed 1989;170:607—12.

[33] Merrill JE, Benveniste EN. Cytokines in inflammatory brainlesions: helpful and harmful. Trends Neurosci 1996;19:331—8.

[34] van Oosten BW, Barkhof F, Scholten PE, von Blomberg BM, AderHJ, Polman CH. Increased production of tumor necrosis factoralpha, and not of interferon gamma, preceding disease activityin patients with multiple sclerosis. Arch Neurol 1998;55:793—8.

[35] van Oosten BW, Barkhof F, Truyen L, Boringa JB, BertelsmannFW, von Blomberg BM, et al. Increased MRI activity and immuneactivation in two multiple sclerosis patients treated with themonoclonal anti-tumor necrosis factor antibody cA2. Neurology1996;47:1531—4.

[36] Sicotte NL, Voskuhl RR. Onset of multiple sclerosis associatedwith anti-TNF therapy. Neurology 2001;57:1885—8.

687Effects of three anti-TNF-α drugs

[37] Tsuchiya S, Yamabe M, Yamaguchi Y, Kobayashi Y, Konno T, TadaK. Establishment and characterization of a human acutemonocytic leukemia cell line (THP-1). Int J Cancer 1980;26:171—6.

[38] Morgan CD, Mills KC, Lefkowitz DL, Lefkowitz SS. An improvedcolorimetric assay for tumor necrosis factor using WEHI 164cells cultured on novel microtiter plates. J Immunol Methods1991;145:259—62.

[39] Scallon B, Cai A, Solowski N, Rosenberg A, Song XY, Shealy D,et al. Binding and functional comparisons of two types of tumornecrosis factor antagonists. J Pharmacol Exp Ther 2002;301:418—26.

[40] Clark IA. How TNF was recognized as a key mechanism ofdisease. Cytokine Growth Factor Rev 2007;18:335—43.

[41] Mohler KM, Torrance DS, Smith CA, Goodwin RG, Stremler KE,Fung VP, et al. Soluble tumor necrosis factor (TNF) receptorsare effective therapeutic agents in lethal endotoxemia andfunction simultaneously as both TNF carriers and TNF antago-nists. J Immunol 1993;151:1548—61.

[42] Solorzano CC, Ksontini R, Pruitt JH, Hess PJ, Edwards PD,Kaibara A, et al. Involvement of 26-kDa cell-associated TNF-alpha in experimental hepatitis and exacerbation of liver injurywith a matrix metalloproteinase inhibitor. J Immunol 1997;158:414—9.

[43] Xin L, Wang J, Zhang H, Shi W, Yu M, Li Q, et al. Dual regulationof soluble tumor necrosis factor-alpha induced activation ofhuman monocytic cells via modulating transmembrane TNF-alpha-mediated 'reverse signaling'. Int J Mol Med 2006;18:885—92.

[44] Filion LG, Graziani-Bowering G, Matusevicius D, Freedman MS.Monocyte-derived cytokines in multiple sclerosis. Clin ExpImmunol 2003;131:324—34.

[45] Kappos L, Moeri D, Radue EW, Schoetzau A, Schweikert K,Barkhof F, et al. Predictive value of gadolinium-enhancedmagnetic resonance imaging for relapse rate and changes in

disability or impairment in multiple sclerosis: a meta-analysis.Gadolinium MRI Meta-analysis Group. Lancet 1999;353:964—9.

[46] Aderka D, Engelmann H, Maor Y, Brakebusch C, Wallach D.Stabilization of the bioactivity of tumor necrosis factor by itssoluble receptors. J Exp Med 1992;175:323—9.

[47] Giri SN, Wang Q, Xie Y, Lango J, Morin D, Margolin SB, et al.Pharmacokinetics and metabolism of a novel antifibrotic drugpirfenidone, in mice following intravenous administration.Biopharm Drug Dispos 2002;23:203—11.

[48] Giri SN, Leonard S, Shi X, Margolin SB, Vallyathan V. Effects ofpirfenidone on the generation of reactive oxygen species invitro. J Environ Pathol Toxicol Oncol 1999;18:169—77.

[49] Misra HP, Rabideau C. Pirfenidone inhibits NADPH-dependentmicrosomal lipid peroxidation and scavenges hydroxyl radicals.Mol Cell Biochem 2000;204:119—26.

[50] Schreibelt G, van Horssen J, van Rossum S, Dijkstra CD,Drukarch B, de Vries HE. Therapeutic potential and biologicalrole of endogenous antioxidant enzymes in multiple sclerosispathology. Brain Res Rev 2007;56:322—30.

[51] Tsuchiya H, Kaibori M, Yanagida H, Yokoigawa N, Kwon AH,Okumura T, et al. Pirfenidone prevents endotoxin-induced liverinjury after partial hepatectomy in rats. J Hepatol 2004;40:94-101.

[52] Gomez-Reino JJ, Carmona L, Valverde VR, Mola EM, MonteroMD. Treatment of rheumatoid arthritis with tumor necrosisfactor inhibitors may predispose to significant increase intuberculosis risk: a multicenter active-surveillance report.Arthritis Rheum 2003;48:2122—7.

[53] Aringer M, Smolen JS. SLE — complex cytokine effects in acomplex autoimmune disease: tumor necrosis factor in sys-temic lupus erythematosus. Arthritis Res Ther 2003;5:172—7.

[54] Mageed RA, Isenberg DA. Tumour necrosis factor alpha insystemic lupus erythematosus and anti-DNA autoantibodyproduction. Lupus 2002;11:850—5.