Suppressive Effects of Ansamycins onInducible Nitric Oxide Synthase Expressionand the Development of ExperimentalAutoimmune Encephalomyelitis

Patricia Murphy,1 Anthony Sharp,1 Joseph Shin,1 Vitaliy Gavrilyuk,1

Cinzia Dello Russo,1 Guy Weinberg,1 Frank R. Sharp,2 Aigang Lu,2

Michael T. Heneka,1 and Douglas L. Feinstein1*1Department of Anesthesiology, University of Illinois, Chicago, Illinois2Department of Neurology, University of Cincinnati, Ohio

The production of nitric oxide by the inflammatory iso-form of nitric oxide synthase (NOS2) in brain glial cells isthought to contribute to the causes and development ofneurological diseases and trauma. We previously dem-onstrated that activation of a heat shock response (HSR)by hyperthermia reduced NOS2 expression in vitro, andin vivo attenuated the clinical and histological symptomsof the demyelinating disease experimental autoimmuneencephalomyelitis (EAE; Heneka et al. [2001] J. Neuro-chem. 77:568–579). Benzoquinoid ansamycins arefungal-derived antibiotics with tyrosine kinase inhibitoryproperties, and which also induce a HSR by allowingactivation of HS transcription factor HSF1. We now showthat two members of this class of drugs (geldanamycinand 17-allylamino-17-demethoxygeldanamycin) also in-duce a HSR in primary rat astrocytes and rat C6 gliomacells. Both drugs dose-dependently reduced nitrite ac-cumulation, NOS2 steady-state mRNA levels, and thecytokine-dependent activation of a rat 2.2-kB NOS2 pro-moter construct stably expressed in C6 cells. These in-hibitory effects were partially reversed by quercetin, abioflavonoid which prevents HSF1 binding to DNA andthus attenuates the HSR. Ansamycins increased mRNAlevels of the inhibitory I�B� protein, suggesting that in-hibition of NF�B activation could contribute to their sup-pressive effects. Finally, in C57BL/6 mice actively immu-nized to develop EAE, a single injection of geldanamycinat 3 days after immunization reduced disease onset byover 50%. These results indicate that ansamycins canexert potent anti-inflammatory effects on brain glial cellswhich may provide therapeutic benefit in neuroinflamma-tory diseases. © 2002 Wiley-Liss, Inc.

Key words: cytokines; heat shock response; multiplesclerosis; astrocytes

Expression of pro-inflammatory cytokines and theinducible form of nitric oxide synthase (NOS2) in brain isthought to contribute to the pathogenesis of neurodegen-

erative diseases such as multiple sclerosis (MS) and Alzhei-mer’s disease (AD; for recent reviews, see Akiyama et al.,2000; Rothwell and Luheshi, 2000; Benveniste et al.,2001). Anti-inflammatory treatments may provide thera-peutic benefit, at least in part, by reducing pro-inflammatory gene expression. A key mediator of inflam-matory gene expression is activation of transcription factorNF�B (for recent reviews, see Baldwin, 2001; Cachetto,2001; Mattson and Camandola, 2001), and canonicalbinding sites for NF�B have been located in the promoterregions of many such genes. NF�B activation is regulatedto a large degree by association with inhibitory IkB pro-teins, which form complexes with NF�B, thus masking itsnuclear localization sites and thereby maintaining it withinthe cytoplasmic compartment (Schmitz et al., 2001). Themechanism of action of anti-inflammatory treatments inmany cases involves an increase and/or stabilization of IkBexpression.

Several groups have demonstrated that inflammatorygene expression, including that of cytokines, chemokines,as well as of NOS2, is potently suppressed by the stressresponse (more generally referred to as the heat shockresponse, HSR), induced either by hyperthermia or bychemical means. The HSR reduced NOS2 expression in avariety of cell types and tissues, including that due toinduction by interleukin 1� (IL-1�) in rat pulmonaryartery smooth muscle cells (Wong et al., 1995); induced bycytokines plus lipopolysaccharide (LPS) in rat primaryastrocytes, rat C6 glioma cells, and Rat-1 fibroblasts (Fein-stein et al., 1996, 1997); induced by cytokines in rat

Contract grant sponsor: NIH; Contract grant number: NS31556; Contractgrant sponsor: National Multiple Sclerosis Society.

*Correspondence to: Douglas L. Feinstein, 1819 West Polk Street, MC519, Rm 544, Chicago, IL 60612. E-mail: dlfeins@uic.edu

Received 27 June 2001; Revised 25 October 2001; Accepted 30 October2001

Published online 7 January 2002

Journal of Neuroscience Research 67:461–470 (2002)DOI 10.1002/jnr.10139

© 2002 Wiley-Liss, Inc.

hepatocytes (de Vera et al., 1996); by cytokines in murinelung epithelial cells (Wong et al., 1999); and by IL-1� incultured rat and human islets (Scarim et al., 1998). Inwhole animals, treatment with arsenite, which induces aHSR, blocked endotoxin-induced NOS2 expression inlung (Hauser et al., 1996), and hyperthermia preventedNOS2 induction in brain following intrastriatal injectionof cytokines and LPS (Heneka et al., 2000). In addition toNOS2, the HSR decreased expression of other inflamma-tory genes which are implicated in the development ofneurodegenerative diseases, including the RANTES (reg-ulated on activation of normal T-cell-expressed and se-creted) chemokine (Ayad et al., 1998), the pro-inflammatory cytokines IL1-� (Simon et al., 1995;Housby et al., 1999) and tumor necrosis factor-� (TNF-�)(Snyder et al., 1992; Meng et al., 1999), and the intercel-lular adhesion molecule-1 (ICAM1) (Housby et al., 1999).The exact mechanisms by which the HSR prevents in-flammatory gene expression are not yet known, but stud-ies demonstrating that the HSR causes upregulation ofIkB� expression (Feinstein et al., 1996, 1997; DeMeesteret al., 1997), and the presence of a consensus binding sitefor the HS transcription factor HSF1 within the IkB�promoter region (Wong et al., 1999) suggests that induc-tion of IkB expression contributes to the ability of theHSR to prevent NF�B activation.

In view of the fact that the HSR could preventcytokine, chemokine, and NOS2 expression, all of whichare implicated in the development of MS, we recentlytested the possibility that induction of a HSR wouldattenuate development of disease in experimental autoim-mune encephalomyelitis (EAE), the most commonly usedmodel to study the pathogenetic events occurring in MS(Linington and Lassman, 1987; Johns et al., 1995). Weshowed that the HSR induced by whole body hyperther-mia in mice significantly reduced the incidence of disease(from 100% to 30%), and decreased disease severity inthose mice which became ill (Heneka et al., 2001). Theeffects of HSR in brain were accompanied by decreasedexpression of the inflammatory genes NOS2 and RANTES,and by reduced activation of transcription factor NF�B.The fact that the HSR can induce expression of theinhibitory protein IkB� in brain (Heneka et al., 2000)suggests that the anti-inflammatory and protective effectsof HSR were due, at least in part, to inhibition of NF�Bactivation.

The HSR is induced by a variety of stresses, includ-ing hyperthermia, oxidative stress, heavy metals, viral in-fection, and UV irradiation, all of which ultimately lead toactivation of HS transcription factors (HSFs) and initiationof HSP gene transcription (Welch, 1992). It was recentlyshown that ansamycins, a family of antibiotic benzoqui-nones originally isolated from actinomycetes (DeBoer etal., 1970; Omura et al., 1979), can induce a HSR in avariety of cells and tissues (Ali et al., 1998; Bagatell et al.,2000). Ansamycins, which include herbimycin-A andgeldanamycin, were originally characterized as agentswhich inhibit protein tyrosine kinase (PTK) catalytic ac-

tivity, exhibited antitumor activity by virtue of inhibitingactivity of c-src (Uehara et al., 1989), and could reverttransformed cells to their normal phenotype (Uehara et al.,1986). However, further studies revealed that c-src inhi-bition was not due to direct suppression of its catalyticactivity by geldanamycin, but instead was mediated viabinding of geldanamycin to HSP90 (Whitesell et al.,1994). Upon binding, geldanamycin prevents binding ofATP to the N-terminal area of HSP90, and thus preventsstabilization of HSP90 client proteins, which includec-src, leading to their proteolytic degradation by the 26Sproteasome (Mimnaugh et al., 1996; Neckers et al., 1999).Subsequent studies identified other HSP90 client proteins,amongst which is included heat shock transcription factor(HSF)-1. Since association with HSP90 keeps HSF1 in-active, disruption of that association by geldanamycin alsoleads to initiation of HSF1 activation and induction of theHSR (Ali et al., 1998; Bagatell et al., 2000).

Since whole body hyperthermia is not a practicalapproach to the treatment of neurological disorders such asMS, particularly since elevated temperatures aggravate thesymptoms of MS, we sought an alternative means toinduce a HSR. We therefore tested the possibility thatgeldanamycin, or a related analogue 17-(allylamino)-17-demethoxygeldanamycin (17-AAG; Schulte and Neckers,1998), could induce a HSR in glial cells and whether theywould reduce NOS2 expression. Our results confirm thishypothesis in both primary rat astrocytes and C6 gliomacells, and furthermore demonstrate that these drugs canreduce the development of clinical symptoms of EAE.

MATERIALS AND METHODS

Reagents

Dulbecco’s modified eagle medium (DMEM), antibiotics,LPS (Salmonella typhirium), Trizol reagent, and quercetin werefrom Sigma (St. Louis, MO). Fetal calf serum (FCS) was fromAtlanta Biological (Norcross, GA). Human recombinant IL-1�(4 � 106 unit/mg) was obtained from the NIH AIDs reagentsprogram. Recombinant rat interferon-� (IFN-�) (4 � 106 unit/mg), synthetic oligonucleotides, and Taq polymerase were fromGIBCO (Gaithersburg, MD). Bright Glo luciferase assay re-agents and Cytotox-96 kits were from Promega (Madison, WI).Enchanted chemiluminescence reagents were from Pierce(Rockford, IL). Antibody to HSP70 (SPA-810) was from Stress-gen (Vancouver, BC). Geldanamycin was from Calbiochem (LaJolla, CA), and 17-AAG was from Dr. Schultz at the DrugSynthesis and Chemistry Branch, NCI, Bethesda, MD. Thechemical structures are illustrated in Figure 1.

Cells

C6 glioma parental cells and C6-2.2 cells, which are stablytransfected with a 2.2-�B region of the rat NOS2 promoterdriving expression of the luciferase report gene (Stasiolek et al.,2000), were grown in DMEM containing 10% FCS and anti-biotics (penicillin and streptomycin). Cells were passaged once aweek, and used after 3 to 4 days, at which point they were90–95% confluent. Primary astrocytes were prepared from ce-rebral cortices of postnatal 1-day-old Sprague-Dawley rats as

462 Murphy et al.

previously described (Galea et al., 1992). Media was changedevery 3 days. After 2 weeks growth in complete media (DMEMwith 10% FCS), the cultures consisted of 95–98% astrocytes, and2–3% microglia.

NOS2 Induction and Activity Measurements

Cells were grown to 80–90% confluency, the growthmedia removed, the cells were washed in serum-free media, andincubations were carried out in fresh DMEM containing anti-biotics and 1% FCS. NOS2 was induced by incubation withbacterial endotoxin lipopolysaccharide together with recombi-nant rat IFN-� (20 U/ml). Induction of NOS2 was assessedindirectly by nitrite production in the cell culture media after 18to 24 hr. An aliquot of the cell culture media (80–100 �l wasmixed with half volume of Griess reagent (Green et al., 1982)and the absorption measured at 550 nm. Solutions of NaNO2dissolved in DMEM/1% FCS served as standards, and the ab-sorption due to cells incubated in media alone (no NOS2inducers) was subtracted from experimental values.

Reporter gene activation in C6-2.2 cells was measured aspreviously described (Stasiolek et al., 2000). In brief, cells wereincubated with LPS and IFN� and indicated concentrations ofdrugs for 4 to 6 hr, whole cell lysates were prepared, andluciferase activity determined by mixing 20 �L of lysate with20 �L of a luciferase substrate (Bright Glo, Promega). Theresulting luminescence was measured in a single-tube luminometer.

RT-PCR Analysis

Total cytoplasmic RNA was prepared from cells usingTrizol reagent (Sigma); aliquots were converted to cDNA usingrandom hexamer primers, and mRNA levels were estimated byreverse transcriptase-polymerase chain reaction (RT-PCR) as-say (Galea et al., 1994). The primers used for NOS2 detectionwere 1704F (5� CTG CAT GGA ACA GTA TAA GGC AAAC-3�), corresponding to bases 1704–1728, and 1933R (5� CAGACA GTT TCT GGT CGA TGT CAT GA-3�) which yield a230-bp product. The primers used for IkB� detection were

110F (5�-AAG AAG GAG CGG TTG GTG GAC-3�) and504R (5�-CAG GCA AGA TGG AGA GGG GTA TT-3�) fora 395-bp product, and for glyceraldehyde 3-phosphate dehy-drogenase (GDH) were 796F (5�-GCC AAG TAT GAT GACATC AAG AAG-3�) and 1059R (5�-TCC AGG GGT TTCTTA CTC CTT GGA-3�) for a 264-bp product. For NOS2,the PCRs were done in the presence of a 20-fg amount of asmaller (by 50 bp) competitive internal standard (CIS) that isamplified with the same efficiency as the cDNA template. PCRswere initiated by a hot start method, and conditions were35 cycles of denaturation at 93°C for 20 sec; annealing at 63°Cfor 20 sec; and extension at 72°C for 20 sec; followed by 10 minat 72°C. PCR products were separated by electrophoresisthrough 2% agarose gels containing 0.1 �g/ml ethidium bro-mide, and quantification of band intensity obtained using anAlphaInotech (Temeculah, CA) gel imaging system.

Preparation of Protein Lysates and Western BlotAnalysis

Cells were washed twice with ice-cold PBS, scraped fromthe dishes, and collected by centrifugation (3,000 g for 5 min).The cells were resuspended directly into five volumes of 8 Murea. Aliquots were mixed with an equal volume of 2 � buffercontaining 124 mM Tris-Cl pH 6.8, 0.2% sodium dodecylsul-fate (SDS), 10% �-ME, 10 mM ethylenediaminetetraacetic acid(EDTA), 50% glycerol, and boiled for 5 mins. Twenty �g wereseparated through 10% poly acrylamide gel electrophoresis(PAGE)-SDS gels, transferred to polyvinylidenedifloride (PVDF)membranes, blocked in Tris-buffered saline (TBS) containing0.1% Tween-20 (TBST) and 5% dry milk (1 hr), rinsed, andincubated with primary antibody in TBST containing 0.5%bovine serum albumin (BSA) overnight with gentle shaking at4°C. Membranes were washed in TBST, and 0.1 �g/mlperoxidase-labeled goat secondary antibodies added for 1 hr.Following four washes in TBST, bands were visualized byincubation in ECL reagents, and exposure to X-ray film. Poly-clonal antibody (SPA-810) directed against HSP70 was used at1:1,000 dilution and was from Stressgen.

Induction of EAE

EAE was actively induced in 6- to 8-week-old femaleC57BL/6 mice using synthetic myelin oligodendrocyte glyco-protein peptide 35–55 (MOG35–55) as described (Heneka et al.,2001). In brief, �98% purified MOG35–55 peptide (MEVGW-YRSPFSRVVHLYRNGK) was injected subcutaneously (two100-�L injections into adjacent areas in one hindlimb) with anemulsion of 300 �g MOG35–55 dissolved in 100 �L phosphate-buffered saline (PBS), mixed with 100 �L complete Freund’sadjuvant (CFA) containing 500 �g of Mycobacterium tubercu-losis (Difco, Detroit, MI). Immediately after MOG35–55 injec-tion, the animals received an i.p. injection of pertussis toxin (PT,200 ng in 200 �L PBS). Two days later, the mice received asecond PT injection, and 1 week later, they received a boosterinjection of MOG35–55 (referred to as day 0). Clinical signs werescored as: 0, no clinical signs; 1, limp tail and/or impairedrighting; 2, paresis of one hind limb; 3, paresis of two hind limbs;4, moribund; 5, death. Scores of 5 are registered only for the dayof death, after which the group size is reduced by 1. Scoring wasdone at the same time each day by a blinded investigator.

Fig. 1. Chemical structures of ansamycins used in this study.

Heat Shock Response Prevents EAE 463

Data Analysis

All enzymatic experiments were done at least in triplicate,and means S.E.M. determined. The IC50 values for inhibitionof nitrite accumulation were determined by fitting the data to asingle exponential decay curve. Statistical significance was as-sessed using Prism 3.0 software (GraphPad, San Diego, CA) byone-way analysis of variance (ANOVA) analysis followed byDunnett’s multiple comparison test, two-way ANOVA fol-lowed by Bonferroni’s multiple comparison analysis, or MannWhitney ranked sums nonparametric analysis for EAE clinicalscore data. P values 0.05 were considered significant.

RESULTSWe first tested whether geldanamycin could induce

a HSR in glial cells by examining expression of HSP70(Fig. 2). C6 cells expressed low amounts of HSP70, whichwere slightly increased upon incubation with LPS/IFN�,

clearly increased by incubation with geldanamycin alone,and further increased by the combination of LPS/IFN�with geldanamycin. Lower concentrations (10 nM) ofansamycins also induced a HSR, either alone or in thepresence of LPS/IFN� (see Fig. 6). In C6 cells, coincu-bation with geldanamycin led to a dose-dependent de-crease in LPS and cytokine-induced nitrite production,with an IC50 of approximately 5.0 nM (Fig. 3A). Therelated ansamycin 17-AAG also suppressed nitrite produc-tion in C6 cells (Fig. 3B), with a slightly higher potency(IC50 approximately 2.5 nM). Incubation with vehicledimethylsulfoxide, (DMSO) alone had no effect on nitriteaccumulation, and incubation with ansamycins did notcause any increase in LDH release from C6 cells at con-centrations of less than 200 nM (not shown).

Ansamycins were also effective in reducing nitriteproduction in primary cultures of rat brain astrocytes (Fig.

Fig. 2. Geldanamycin (Geld) induces a heat shock response (HSR). C6 cells were treated with LPSplus IFN� (“LI”), geldanamycin (300 nM), or both, proteins prepared after 24 hr, and HSP70expression determined by Western blot analysis. Two different samples are shown for each treatment.

Fig. 3. Ansamycins reduce nitrite production from C6 cells. C6 cells were preincubated for 1 hr invarying concentrations of (A) geldanamycin or (B) 17-allylamino-17-demethoxygeldanamycin (17-AAG), then LPS/IFN� added and nitrite production measured 24 hr later. *P .05; **P .01 versuscontrol (n � 3 each, one-way analysis of variance [ANOVA] followed by Dunnett’s multiplecomparison test). The IC50 values calculated by single decay exponential curve fitting are shown. Theopen circles are incubation with vehicle (0.1% dimethylsulfoxide, DMSO) alone.

464 Murphy et al.

4). As the case for C6 cells, both geldanamycin and 17-AAG dose-dependently reduced nitrite accumulation afterovernight incubation with LPS and IFN�, although inthese cells geldanamycin showed a slightly greater potency(IC50 values were approximately 3.3 nM and 9.6 nM forgeldanamycin and 17-AAG, respectively). Measurementsof LDH release revealed no increase up to 33 nM of eithergeldanamycin or 17-AAG, and incubation of astrocyteswith up to 0.1% vehicle (DMSO) had no effect on eithernitrite production or LDH release.

To determine whether ansamycins reduced NOS2expression, we measured steady-state NOS2 mRNA levels(Fig. 5A). As previously shown (Galea et al., 1994), incu-bation for 4 hr with LPS and IFN� induced NOS2mRNA accumulation in C6 cells, which was reduced bycoincubation with either geldanamycin or 17-AAG.Quantitative analysis obtained by comparison to the in-ternal competitive standard indicated that geldanamycinreduced NOS2 mRNA levels by 70% (from approxi-mately 27.5 pg to 8.5 pg NOS2 mRNA per �g of startingtotal RNA). Interestingly, incubation with geldanamycinalone for 4 hr caused an increase in NOS2 mRNA levelsversus incubation in media alone; however, this effect was

not observed upon incubation with 17-AAG alone. Todetermine whether ansamycins reduced NOS2 transcrip-tion, we measured activation of a 2.2-kB rat NOS2 pro-moter stably expressed in C6 cells (Fig. 5B). Incubationwith LPS and IFN� led to a 2- to 3-fold increase inreporter gene activity that was dose-dependently reducedby coincubation with geldanamycin, suggesting that areduction in NOS2 transcription accounts, at least in part,for the suppressive effects of ansamycins on NOS2 mRNAlevels and nitrite production. However, the fact thathigher doses of geldanamycin were needed to reduceNOS2 promoter activation as compared to reducing ni-trite production or NOS2 mRNA levels suggest thatadditional mechanisms of regulation are evoked by ansa-mycins.

To begin to determine the molecular basis for NOS2suppression by ansamycins, we examined the effects of17-AAG on inhibitory IkB� mRNA levels (Fig. 6). TheIkB� mRNA was detected in control astrocytes, and itslevels were increased after 4 hr incubation with 10 nM17-AAG. As expected, IkB� mRNA levels were alsoincreased following incubation with LPS and IFN� aspreviously shown (Stasiolek et al., 2000). The accumula-

Fig. 4. Ansamycins reduce nitrite production from astrocytes. Primary astrocytes were incubated withLPS and IFN� and indicated concentrations of geldanamycin (closed circles) or 17-AAG (opencircles), and nitrite production measured 24 hr later. **P .01 versus control (n � 3, one-wayANOVA followed by Dunnett’s multiple comparison test). The calculated IC50 values are shown.

Heat Shock Response Prevents EAE 465

tion of NOS2 mRNA after 4 hr incubation with LPS andIFN� was reduced by coincubation with 10 nM 17-AAG,although to a lesser extent than that found when 100 nMansaymycins were used (Fig. 5). Incubation with 17-AAGalone had no effect on NOS2 mRNA levels. GDHmRNA levels were not modified by cytokines or 17-AAG.

The inhibitory actions of ansamycins on NOS2 ex-pression could be due to inhibition of tyrosine kinaseactivities, and/or induction of a HSR. To begin to dis-tinguish between these possibilities, we made use of thebioflavonoid quercetin, which inhibits binding of HSFs toHS elements present in gene promoters, and thereby pre-vents HSP gene transcription (Hosokawa et al., 1990; Leeet al., 1994; Nagai et al., 1995; Wu and Yu, 2000). Theinduction of nitrite accumulation in astrocytes was notinfluenced by coincubation with up to 5 �M quercetin,

although some inhibition (22 2% inhibition) was ob-served in the presence of 10 �M quercetin (Fig. 7). Thisis consistent with previous studies showing that quercetincan inhibit NOS2 expression at higher concentrations(Kim et al., 1999; Manjeet and Ghosh, 1999; Raso et al.,2001; Chen et al., 2001). As expected, nitrite accumula-tion was significantly (16 5% of control) inhibited byincubation with 50 nM 17-AAG, and this suppression wasdose-dependently prevented by cotreatment with up to5 �M quercetin. No further reversal of 17-AAG effectswere seen at higher concentrations of quercetin tested(10 �M, or higher doses not shown), most likely due tothe fact that at concentrations greater than 5 �M, thequercetin itself is exerting inhibitory effects. However,these results suggest that the inhibitory actions of 17-AAGcan be prevented at concentrations of quercetin which bythemselves have neither stimulatory nor inhibitory effects.

Fig. 5. Ansamycins reduce nitric oxide synthase (NOS2) expression.A: mRNA analysis. C6 cells were incubated in the presence of LPS andIFN� (“LI”) together with 100 nM of geldanamycin (G) or 17-AAG(A). After 4 hr, total RNA was prepared and NOS2 mRNA levelsestimated by competitive reverse transcriptase-polymerase chain reac-tion (RT-PCR) using cDNA obtained from 50 ng of starting totalRNA versus 20 fg of a NOS2 competitive standard (lower band). Thecalculated ratios of the upper band (PCR product derived from NOS2mRNA) to lower band (PCR product derived form internal standard)

were 0.7 (control), 5.3 (LI), 1.3 (Geld), 2.1 (LI � Geld), 0.6 (AAG),and 2.4 (LI � AAG). The gel shown is representative of two indepen-dent experiments with similar results. B: Promoter analysis. C6-2.2cells containing a 2.2-�B fragment of the rat NOS2 promoter drivingluciferase expression were incubated in media (control) or LPS plusIFN� in the presence of the indicated concentration of geldanamycin.Reporter gene activity was measured in whole cell lysates after 6 hr.Results are mean S.E.M. of three to six measurements. RLU, relativelight units.

466 Murphy et al.

Since astrocyte inflammatory responses, includingNOS2 expression, contribute to the pathogenesis of EAE,we tested the effects of geldanamycin on the developmentof EAE in C57BL/6 mice (Fig. 8). EAE was induced usingthe encephalitogenic MOG35–55 peptide, and the micereceived a single injection of geldanamycin (300 ng, i.p.)

2 days after the booster injection, a time point at which noclinical signs are evident. Geldanamycin significantly re-duced the incidence of disease (Fig. 8A) from 100% (9/9)to 42% (3/7; P .0001 unpaired t-test) and the averageclinical scores of the entire group (Fig. 8B; P .0001,Mann Whitney ranked sum analysis). However, compar-ison of only those mice (n � 4) which became ill in thegeldanamycin group to the control group revealed nodifferences in their average daily clinical scores (Fig. 8B,closed squares; P � 0.05, Mann Whitney ranked sumanalysis), the day of disease onset (control: 9.1 0.6, n �9 versus geldanamycin: 8.5 0.5, n � 4, P � 0.05 byunpaired t-test), the average maximal clinical scores (con-trol: 2.0 0.2, n � 9 versus Geldanamycin: 1.8 0.5,n � 4, P � .05 unpaired t-test), or maximal clinical scores(3.0 in both in both groups). Thus, a single injection ofgeldanamycin prior to appearance of clinical signs reduceddisease incidence, but had no significant effect on diseasein the mice which became ill.

DISCUSSIONOur results indicate that incubation with ansamycins

reduces nitrite production in C6 and primary astrocytes,and experiments with geldanamycin demonstrated a de-crease in steady-state NOS2 mRNA levels and promoteractivation. Treatment with geldanamycin caused induc-tion of a HSR as revealed by an increase in levels of theHSP70 protein, while incubation with 17-AAG increasedthe mRNA levels of IkB�, itself being recently identifiedas a HSP. The effects of 17-AAG were partially reversedby coincubation with quercetin, a bioflavonoid whichblocks the induction of a HSR, suggesting that inductionof a HSR accounts, at least in part, for the suppressiveactions of ansamycins on NOS2 expression. Together,these findings demonstrate that ansamycins can exert po-tent anti-inflammatory actions on brain glial cells, whichmay offer therapeutic benefit in neuroinflammatory dis-eases as suggested by our results showing a reduction in theincidence of EAE.

Whether the effects of ansamycins on NOS2 expres-sion are mediated through their ability to act as tyrosinekinase inhibitors, and/or by their induction of a HSR, isnot yet clear. In fact, these two functions may be inextri-cably linked since the mechanism believed to account forHSR induction is via binding of ansamycins to the amino-terminal nucleotide-binding site on HSP90, leading torelease of associated HSF1 and activation of a HSR (Ali etal., 1998; Bagatell et al., 2000). We used the bioflavonoidquercetin to begin to dissect the contribution of these twoprocesses, since this molecule has been shown to blockbinding of activated HSF1 to its cognate DNA sequence,reduce the activation of the HSR, and decrease HSPexpression (Hosokawa et al., 1990; Lee et al., 1994; Nagaiet al., 1995; Wu and Yu, 2000). Our results demonstratethat the suppression of nitrite production due to 50 nM17-AAG was partially, and dose-dependently reversed bycoincubation with quercetin up to a concentration of5 �M. Since quercetin alone did not cause any increase innitrite production, these results are consistent with the

Fig. 6. Ansamycins increase IkB� expression. Primary astrocytes wereincubated with LPS and IFN� (LI) together with 10 nM of 17-AAG,or with 10 nM 17-AAG alone. After 4 hr, total RNA was prepared andlevels of NOS2, glyceraldehyde 3-phosphate dehydrogenase (GDH),and IkB� determined by RT-PCR. The analysis for NOS2 was donein the presence of 20 fg of a NOS2 competitive standard. The gelshown is representative of two separate experiments, and similar resultswere also obtained after 24 hr incubation.

Fig. 7. Quercetin partially reverses inhibition by 17-AAG. Primaryastrocytes were incubated with LPS and IFN�, in the presence of either0 or 50 nM 17-AAG, and with the indicated concentrations of thebioflavonoid quercetin. Nitrite production was measured 24 hr laterusing Griess reagent. The data are mean S.E.M., and is percent nitriteaccumulation compared to incubation in the absence of any drugs.*P 0.05 versus control, n � 3).

Heat Shock Response Prevents EAE 467

interpretation that, at least at low doses, quercetin preventsinduction of the HSR by ansamycins. At higher concen-trations of quercetin, no further reversal of the effects of17-AAG was observed, most likely due to the fact thatquercetin at those concentrations begins to exert inhibi-tory effects on NOS2 expression.

Inhibitory effects of geldanamycin, as well as severalother protein tyrosine kinase inhibitors, on NOS2 expres-sion have been reported previously. In vascular smoothmuscle cells (VSMC), geldanamycin blocked both LPSand IL-1�-induced cGMP and nitrite production, with50% inhibition at roughly 100 nM (Marczin et al., 1993).In the same study, a second tyrosine kinase inhibitor,genistein, also blocked nitrite production, albeit at muchhigher concentrations (10 to 50 �M). The authors con-cluded that geldanamycin blocked an unspecified proteintyrosine kinase (PTK) which is needed for signaling lead-ing to NOS2 expression. Similarly, Joly et al. (1997) foundthat geldanamycin reduced LPS and cytokine inducednitrite production in VSMC as well as in brain endothelialcells, although a structurally distinct PTK inhibitor tyr-phostin had no effect, and genistein was only effective inthe endothelial cells when used at high concentrations.These authors concluded that NOS2 expression involvesnovel PTKs that can be inhibited by geldanamycin but notby two other PTK inhibitors. The fact that neither tyr-phostin nor genistein have been shown to induce a HSR,suggests the alternate possibility that the inhibitory actionsof ansamycins are mediated, at least in part, by theirinduction of a HSR. The possibility exists that the “novel”

PTKs presumed by others to mediate the inhibitory effectsof geldanamycin may in fact reside within the HSP90complex itself, to which the ansamycins (geldanamycin,17-AAG, as well as Herbimycin-A) but not genistein nortyrphostin have access.

It was previously shown that the HSR reducesNF�B activation due to stabilization of the inhibitoryprotein IkB� (Feinstein et al., 1996, 1997; DeMeester etal., 1997), and IkB� was determined to be a HSP by virtueof the presence of a near consensus HS element within itspromoter region (Wong et al., 1999). We also demon-strated that induction of a HSR by whole-body hyper-thermia of rats induced IkB� expression in brain (Henekaet al., 2000). Our current data indicate that incubationwith 17-AAG increases the steady-state levels of the IkB�mRNA, thus demonstrating that ansamycins can induceglial IkB� expression, and suggesting that as in the case forthermally induced HSR, ansamycins reduce NOS2 ex-pression due to increased IkB� levels and therefore de-creased NF�B activation.

We recently demonstrated that induction of a HSRin mice immunized to develop EAE, the animal model forMS, potently blocks the incidence and severity of clinicaland histological symptoms (Heneka et al., 2001). In thatmodel, the HSR was induced by raising the body tem-perature of the mice to 41 0.5°C for 20 min, done at9 days after the initial immunization (2 days after thebooster immunization). In the current study, mice weretreated once with geldanamycin at this same time point. Incontrast to the hyperthermia, which both reduced disease

Fig. 8. Geldanamycin reduces clinical signs of Experimental autoim-mune encephalomyelitis (EAE). C57BL/6 mice were immunized withmyelin-oligodendrocyte glycoprotein (MOG)35–55 peptide, and in-jected once with geldanamycin (300 ng, i.p.; filled circles) or vehicleonly (empty circles) at 2 days after the booster injection. A: Incidenceof disease. In the control group, incidence reached 100% (9/9), whilein the geldanamycin group it reached 42% (3/7). B: Average clinicalscores. The scores for the entire geldanamycin group (filled circles)

include those mice which did show any clinical signs (P .0001,vehicle (n � 9) versus geldanamycin (n � 7), Mann Whitney rankedsums analysis). The scores of those mice (n � 4) in the geldanamycingroup which showed clinical signs are plotted separately (filled squares).For these mice, the average day of disease onset was not different fromthe control group (6.3 0.9 n � 9, control vs. 6.5 0.5, n � 4,geldanamycin, unpaired t-test). Maximal clinical scores were 3.0 inboth groups.

468 Murphy et al.

incidence as well as severity in the animals which becameill, geldanamycin only reduced disease incidence, and themice which became ill displayed clinical symptoms similarto that of control mice. The reasons for these differencesare not clear, but differences in both the quantitative (i.e.,duration or magnitude of the elicited response) and qual-itative (i.e., cellular specificity) effects of hyperthermiaversus ansamycins on induction of the HSR contribute tothe differential effects observed. Nevertheless, our currentfindings suggest that pharmacological treatments can beused to induce a HSR under normothermic conditions,thus avoiding potentially deleterious side effects, whileblocking potentially damaging expression of inflammatorycytokines including NOS2, during trauma and disease.

REFERENCESAkiyama H, Arai T, Kondo H, Tanno E, Haga C, Ikeda K. 2000. Cell

mediators of inflammation in the Alzheimer disease brain. Alzheimer DisAssoc Disord 14 Suppl 1:S47–S53.

Ali A, Bharadwaj S, O’Carroll R, Ovsenek N. 1998. HSP90 interacts withand regulates the activity of heat shock factor 1 in Xenopus oocytes. MolCell Biol 18:4949–4960.

Ayad O, Stark J, Fiedler MM, Menendez IY, Ryan MA, Wong HR. 1998.The HSR inhibits RANTES gene expression in cultured lung epithelium.J Immunol 161:2594–2599.

Bagatell R, Paine-Murrieta GD, Taylor CW, Pulcini EJ, Akinaga S, Ben-jamin IJ, Whitesell L. 2000. Induction of a heat shock factor 1-dependentstress response alters the cytotoxic activity of hsp90-binding agents. ClinCancer Res 6:3312–3318.

Baldwin AS, Jr. 2001. Series introduction: the transcription factor NF-kappaB and human disease. J Clin Invest 107:3–6.

Benveniste EN, Nguyen VT, O’Keefe GM. 2001. Immunological aspectsof microglia: relevance to Alzheimer’s disease. Neurochem Int 39:381–391.

Cechetto DF. 2001. Role of nuclear factor kappa B in neuropathologicalmechanisms. Prog Brain Res 132:391–404.

Chen YC, Shen SC, Lee WR, Hou WC, Yang LL, Lee TJ. 2001.Inhibition of nitric oxide synthase inhibitors and lipopolysaccharide in-duced inducible NOS and cyclooxygenase-2 gene expressions by rutin,quercetin, and quercetin pentaacetate in RAW 264.7 macrophages. J CellBiochem 82:537–548.

DeBoer C, Meulman PA, Wnuk RJ, Peterson DH. 1970. Geldanamycin,a new antibiotic. J Antibiot Tokyo 23:442–447.

DeMeester SL, Buchman TG, Qiu Y, Jacob AK, Dunnigan K, HotchkissRS, Karl I, Cobb JP. 1997. Heat shock induces IkappaB-alpha andprevents stress-induced endothelial cell apoptosis. Arch Surg 132:1283–1287.

de Vera ME, Kim YM, Wong HR, Wang Q, Billiar TR, Geller DA. 1996.Heat shock response inhibits cytokine-inducible nitric oxide synthaseexpression in rat hepatocytes. Hepatology 24:1238–1245.

Feinstein DL, Galea E, Aquino DA, Li GC, Xu H, Reis DJ. 1996. Heatshock protein 70 suppresses astroglial inducible nitric oxide synthaseexpression by blocking NF�B activation. J Biol Chem 271:17724–17732.

Feinstein DL, Galea E, Reis DJ. 1997. Suppression of glial nitric oxidesynthase induction by heat shock: effects on proteolytic degradation of Ikappa B-alpha. Nitric Oxide 1:167–176.

Galea E, Feinstein DL, Reis DJ. 1992. Induction of calcium-independentnitric oxide synthase activity in primary rat glial cultures. Proc Natl AcadSci USA 89:10945–10949.

Galea E, Reis DJ, Feinstein DL. 1994. Cloning and expression of induciblenitric oxide synthase from rat astrocytes. J Neurosci Res 37:406–414.

Green LC, Wagner DA, Glogowski J, Skipper PL, Wishnok JS, Tannen-

baum SR. 1982. Analysis of nitrate, nitrite, and [15N] nitrate in biologicalfluids. Anal Biochem 126:131–138.

Hauser GJ, Dayao EK, Wasserloos K, Pitt BR, Wong HR. 1996. HSPinduction inhibits iNOS mRNA expression and attenuates hypotensionin endotoxin challenged rats. Am J Physiol 271:H2529–2535.

Heneka MT, Sharp A, Klockgether T, Gavrilyuk V, Feinstein DL. 2000.The heat shock response inhibits NF-�B activation, nitric oxide synthasetype 2 expression, and macrophage/microglial activation in brain. J CerebBlood Flow Metab 20:800–811.

Heneka MT, Sharp A, Murphy P, Lyons JA, Dumitrescu L, Feinstein DL.2001. The heat shock response reduces myelin oligodendrocyteglycoprotein-induced experimental autoimmune encephalomyelitis inmice. J Neurochem 77:568–579.

Hosokawa N, Hirayoshi K, Nakai A, Hosokawa Y, Marui N, Yoshida M,Sakai T, Nishino H, Aoike A, Kawai K. 1990. Flavonoids inhibit theexpression of heat shock proteins. Cell Struct Funct 15:393–401.

Housby JN, Cahill CM, Chu B, Prevelige R, Bickford K, Stevenson MA,Calderwood SK. 1999. Non-steroidal anti-inflammatory drugs inhibit theexpression of cytokines and induce HSP70 in human monocytes. Cyto-kine 11:347–358.

Johns TG, Kerlero dR, Menon KK, Abo S, Gonzales MF, Bernard CC.1995. Myelin oligodendrocyte glycoprotein induces a demyelinating en-cephalomyelitis resembling multiple sclerosis. J Immunol 154:5536–5541.

Joly GA, Ayres M, Kilbourn RG. 1997. Potent inhibition of induciblenitric oxide synthase by geldanamycin, a tyrosine kinase inhibitor, inendothelial, smooth muscle cells, and in rat aorta. FEBS Lett 403:40–44.

Kim HK, Cheon BS, Kim YH, Kim SY, Kim HP. 1999. Effects of naturallyoccurring flavonoids on nitric oxide production in the macrophage cellline RAW 264.7 and their structure-activity relationships. Biochem Phar-macol 58:759–765.

Lee YJ, Erdos G, Hou ZZ, Kim SH, Kim JH, Cho JM, Corry PM.1994. Mechanism of quercetin-induced suppression and delay of heatshock gene expression and thermotolerance development in HT-29 cells.Mol Cell Biochem 137:141–154.

Linington C, Lassmann H. 1987. Antibody responses in chronic relapsingexperimental allergic encephalomyelitis: correlation of serum demyelinat-ing activity with antibody titre to the myelin/oligodendrocyte glycopro-tein MOG. J Neuroimmunol 17:61–69.

Manjeet KR, Ghosh B. 1999. Quercetin inhibits LPS-induced nitric oxideand tumor necrosis factor-alpha production in murine macrophages. IntJ Immunopharmacol 21:435–443.

Marczin N, Papapetropoulos A, Catravas JD. 1993. Tyrosine kinase inhib-itors suppress endot. Am J Physiol 265:H1014–H1018.

Mattson MP, Camandola S. 2001. NF-kappaB in neuronal plasticity andneurodegenerative disorders. J Clin Invest 107:247–254.

Meng X, Shames BD, Ao L, Meldrum DR, Chan D, Banerjee A, HarkenAH. 1999. Heat shock protein 72 suppresses macrophage TNF-alphaproduction by inhibition of NF kappa B intranuclear translocation. Shock11:138.

Mimnaugh EG, Chavany C, Neckers L. 1996. Polyubiquitination andproteasomal degradation of the p185c-erbB-2 receptor protein-tyrosinekinase induced by geldanamycin. J Biol Chem 271:22796–22801.

Nagai N, Nakai A, Nagata K. 1995. Quercetin suppresses heat shockresponse by down regulation of HSF1. Biochem Biophys Res Commun208:1099–1105.

Neckers L, Schulte TW, Mimnaugh E. 1999. Geldanamycin as a potentialanti-cancer agent: its molecular target and biochemical activity. InvestNew Drugs 17:361–373.

Omura S, Iwai Y, Takahashi Y, Sadakane N, Nakagawa A, Oiwa H,Hasegawa Y, Ikai T. 1979. Herbimycin, a new antibiotic produced by astrain of Streptomyces. J Antibiot Tokyo 32:255–261.

Raso GM, Meli R, Di Carlo G, Pacilio M, Di Carlo R. 2001. Inhibitionof inducible nitric oxide synthase and cyclooxygenase-2 expression byflavonoids in macrophage J774A.1. Life Sci 68:921–931.

Heat Shock Response Prevents EAE 469

Rothwell NJ, Luheshi GN. 2000. Interleukin 1 in the brain: biology,pathology and therapeutic target. Trends Neurosci 23:618–625.

Scarim AL, Heitmeier MR, Corbett JA. 1998. Heat shock inhibitscytokine-induced nitric oxide synthase expression by rat and human islets.Endocrinology 139:5050–5057.

Schmitz ML, Bacher S, Kracht M. 2001. I kappa B-independent control ofNF-kappa B activity by modulatory phosphorylations. Trends BiochemSci 26:186–190.

Schulte TW, Neckers LM. 1998. The benzoquinone ansamycin 17-allylamino-17-demethoxygeldanamycin binds to HSP90 and shares im-portant biologic activities with geldanamycin. Cancer Chemother Phar-macol 42:273–279.

Simon MM, Reikerstorfer A, Schwarz A, Krone C, Luger TA, Jaattela M,Schwarz T. 1995. Heat shock protein 70 overexpression affects theresponse to ultraviolet light in murine fibroblasts. evidence for increasedcell viability and suppression of cytokine release. J Clin Invest 95:926–933.

Snyder YM, Guthrie L, Evans GF, Zuckerman SH. 1992. Transcriptionalinhibition of endotoxin induced monokine synthesis following heat shockin murine peritoneal macrophages. J Leukocyte Biol 51:181–187.

Stasiolek M, Gavrilyuk V, Sharp A, Selmaj K, Feinstein DL. 2000. Lacta-cystin inhibits astroglial Nitric oxide synthase type 2: effects on IkB-betaexpression. J Biol Chem 275:24847–24856.

Uehara Y, Hori M, Takeuchi T, Umezawa H. 1986. Phenotypic change

from transformed to normal induced by benzoquinonoid ansamycinsaccompanies inactivation of p60src in rat kidney cells infected with Roussarcoma virus. Mol Cell Biol 6:2198–2206.

Uehara Y, Fukazawa H, Murakami Y, Mizuno S. 1989. Irreversible inhi-bition of v-src tyrosine kinase activity by herbimycin A and its abrogationby sulfhydryl compounds. Biochem Biophys Res Commun 163:803–809.

Welch WJ. 1992. Mammalian stress response: cell physiology, structure/function of stress proteins, and implications for medicine and disease.Physiol Rev 72:1063–1081.

Whitesell L, Mimnaugh EG, De Costa B, Myers CE, Neckers LM. 1994.Inhibition of heat shock protein HSP90-pp60v-src heteroprotein com-plex formation by benzoquinone ansamycins: essential role for stressproteins in oncogenic transformation. Proc Natl Acad Sci USA 91:8324–8328.

Wong HR, Finder J, Wasserloos K, Pitt BR. 1995. Expression of iNOS incultured rat pulmonary artery smooth muscle cells is inhibited by HSR.Am J Physiol 269:L843–848.

Wong HR, Ryan MA, Menendez IY, Wispe JR. 1999. Heat shockactivates the I-kappaBalpha promoter and increases I-kappaBalphamRNA expression. Cell Stress Chaperones 4:1–7.

Wu BY, Yu AC. 2000. Quercetin inhibits c-fos, heat shock protein, andglial fibrillary acidic protein expression in injured astrocytes. J NeurosciRes 62:730–736.

470 Murphy et al.