role of macrophage oxidative burst in the action of anthrax lethal

Original Articles

Role of Macrophage Oxidative Burst inthe Action of Anthrax Lethal Toxin

Philip C. Hanna,* Benjamin A. Kruskal,tR. Allen B. Ezekowitz,,t Barry R. Bloom,* and R. John Collier**Department of Microbiology and Molecular Genetics and the ShipleyInstitute of Medicine, Harvard Medical School, Boston, Massachusetts,U.S.A. tDivision of Hematology and Infectious Diseases, TheChildren's Hospital, Department of Pediatrics, Harvard MedicalSchool, Boston, Massachusetts, U.S.A.tHoward Hughes Medical Institute and Departments of Microbiologyand Immunology, Cell Biology, Molecular Pharmacology, and MolecularBiology, Albert Einstein College of Medicine, Bronx, New York, U.S.A.

ABSTRACT

Background: Major symptoms and death from sys-temic Bacillus anthracis infections are mediated by theaction of the pathogen's lethal toxin on host macro-phages. High levels of the toxin are cytolytic to mac-rophages, whereas low levels stimulate these cells toproduce cytokines (interleukin- 1(3 and tumor necrosisfactor-a), which induce systemic shock and death.Materials and Methods: Experiments were per-formed to assess the possibility that the oxidative burstmay be involved in one or both of lethal toxin's effectson macrophages. Toximediated cell lysis, superoxideanion and cytokine production were measured. Effectsof antioxidants and macrophage mutations were ex-amined.Results: RAW264.7 murine macrophages treatedwith high levels of toxin released large amounts ofsuperoxide anion, beginning at about 1 hr, which cor-relates with the onset of cytolysis. Cytolysis could beblocked with various exogenous antioxidants or withN-acetyl-L-cysteine and methionine, which promote

production of the endogenous antioxidant, glutathi-one. Mutant murine macrophage lines deficient in pro-duction of reactive oxygen intermediates (ROIs) wererelatively insensitive to the lytic effects of the toxin,whereas a line with increased oxidative burst potentialshowed elevated sensitivity. Also, cultured bloodmonocyte-derived macrophages from a patient withChronic Granulomatous Disease, a disorder in whichthe phagocyte's oxidative burst is disabled, were totallyresistant to toxin, in contrast to control monocytes.Conclusions: These results imply that the cytolyticeffect of the toxin is mediated by ROIs. Additionally,cytokine production and consequent pathologiesshowed partial dependence on macrophage ROIs. An-tioxidants moderately inhibited toxin-induced cyto-kine production in vitro, and BALB/c mice pretreatedwith N-acetyl-L-cysteine or mepacrine showed partialprotection against lethal toxin. Thus ROIs are involvedin both the cytolytic action of anthrax lethal toxin andthe overall pathologic process in vivo.

INTRODUCTIONA key component of the host's ability to survivebacterial challenge is the innate capacity ofphagocytes to ingest and destroy the invading

Address correspondence and reprint requests to: R. JohnCollier, Department of Microbiology and Molecular Genet-ics, Harvard Medical School, 200 Longwood Avenue,D1-502, Boston, MA 02115, U.S.A.

organism. Phagocytes, neutrophils and mono-cytes/macrophages (M4s), maintain an exten-sive portfolio of mechanisms for killing phagocy-tosed organisms, which include: extremes of pH,lysozyme secretion, release of hydrolytic en-zymes, membrane-damaging defensins andother small cationic proteins, and the uniqueability to generate large amounts of reactive oxy-

Copyright C 1994, Molecular Medicine 7

8 Molecular Medicine, Volume 1, Number 1, November 1994

gen and nitrogen intermediates (ROIs and RNIs,respectively). Some of these agents may be toxicto the phagocytes, which must therefore tightlyregulate their production and/or restrict them tosubcellular locales where their effects can be tol-erated. The potential exists, therefore, for patho-gens to divert these killing mechanisms towardphagocytes themselves.

Anthrax has historically been a useful modelfor studying infectious diseases and host-parasiterelationships (1,2). In particular, Bacillus anthra-cis, the etiologic agent, has played an importantrole in elucidating M4--bacterium interactions.Metchnikoff (3), using transparent tissues of liv-ing animals attached to his microscope stage,employed the anthrax bacillus to demonstratethat large blood cells left the circulatory system(diapedesis), migrated toward a site of infection,and ingested the invading pathogens. He namedthese cells macrophages. Modern studies havefurther linked B. anthracis and its virulence fac-tors to the M+, and these interactions are nowknown to be intimately involved in the clinicalmanifestations of the disease.

Anthrax in humans, and most other animals,is initiated by introduction of B. anthracis sporesinto the body by penetration of the epidermis,the gut, or the respiratory epithelia (4). Afterspore germination, the bacteria multiply andmay spread to the regional lymph nodes andfrom there to the blood stream where they canreach >108/ml. Systemic anthrax is nearly al-ways fatal, with nonspecific, shock-like symp-toms. Of particular note is the suddenness of thefatal shock; in animals, the first overt symptom isoften death itself (4).

Virulence of B. anthracis depends upon threeplasmid-encoded factors: an antiphagocytic poly-D-glutamic acid capsule and two protein toxins(5). Edema toxin (EdTx) is believed to be respon-sible for the edema seen in the cutaneous form ofanthrax and is composed of two noncovalentlylinked proteins, protective antigen (PA, 83 kD)and edema factor (EF, 88 kD). Lethal toxin(LeTx), the central effector of shock and deathduring systemic anthrax, is also comprised of twoproteins, PA and lethal factor (LF, 89 kD). PA, theprotein in common, serves to bind either EF or LFto cell surface receptors and mediate their trans-fer into the cytoplasm. There, EF functions as aCa++-calmodulin-dependent adenylate cyclase(6). The biochemical action of LF is unknown,but a recent report indicates that LF shares lim-ited sequence homology to active site regions of afamily of Zn+±-dependent proteases (7).

Injection of LeTx mimics symptoms of sys-temic anthrax in test animals, and LeTx-deficientstrains of B. anthracis are highly attenuated(8-9). Although most cell types bind and inter-nalize the toxin, only M4s are killed (10). Athigh toxin doses, these cells are lysed within 1-2hr in vitro. We found that mice depleted of M4sare resistant to LeTx, and these mice can beresensitized by injecting cultured M4s but notother types of cells (11). At low concentrations,LeTx is not cytolytic but stimulates cytokine pro-duction in cultured Mos. Furthermore, passiveimmunity to Il-1IO and TNFa protects mice fromtoxin challenge, implying that these cytokinesmediate animal death (11). Thus it is clear thatthe MO plays a central role in anthrax LeTxaction, although the underlying biochemistry ofthe effects seen in these cells is not yet under-stood.

We recently described a cascade of biochem-ical and physiological changes in LeTx-chal-lenged Mos, which culminates in cytolysis (12).That study implicated an early increase in plasmamembrane permeability to Na+ and K+, causingmisregulated cellular colloidosmotic equilibrium,influx of water into the cell, hydrolysis of ATP,and eventual Ca"+ influx (and serious Ca++-induced toxicities). These initial insults triggereddownstream effects, including cessation of virtu-ally all macromolecular synthesis, changes in cellmorphology, and generalized membrane rup-ture. These events resemble, at least superficially,those reported for cells exposed to oxidants. Wetherefore hypothesized that damaging ROIsmight play a role in the responses of Mos to theLeTx. The current study explores this possibility.

MATERIALS AND METHODSAnthrax Toxin PreparationB. anthracis strain Sterne is nonencapsulated andwas maintained as spore stocks. Cultures weregrown in defined RM toxin-production medium(13). Culture supernatants were sterilized bypassage through a 0.22 Am filter (Millipore, Bed-ford, MA, U.S.A.) and concentrated to 1 1 withthe Minitan Ultrafiltration System (Millipore).Ammonium sulfate was added to 75%, and theprotein pellet was collected and suspended in 20mM tris(hydroxymethyl)aminomethane (Tris)-HCI, pH 8.0, and dialyzed extensively against thesame buffer. An efficient one-step purificationwas performed by FPLC MonoQ anion exchange(Pharmacia, Piscataway, NJ, U.S.A.) with a 20

P. C. Hanna et al.: Anthrax Toxin Induces Oxidative Burst

mM Tris-HCl, pH 8.0, buffer and linear 0-400mM NaCl gradient elution over 40 min. PAeluted at 130-140 mM NaCl, EF at 150-175mM, and LF at 250-270 mM. PA, EF, and LFwere each determined to be 90-95% pure bysodium dodecyl sulfate-polyacrylamide gel elec-trophoresis (SDS-PAGE) analysis, and none ofthe toxin components alone affected cells. Forcytotoxicity assays, unless otherwise noted, thefinal concentration of lethal toxin was PA = 0.1,tg/ml; LF = 0.1 ,ug/ml.

Cell CultureMurine macrophage lines RAW 264.7 (ATCCTIB-71), J774 (ATCC TIB-67), and IC-21 (ATCCTIB - 186) were obtained from the American TypeCulture Collection (Rockville, MD, U.S.A.) andwere maintained in minimal essential medium(MEM) supplemented with Earl's salts, 10% fe-tal calf serum (FCS), penecillin (200 ,ug/ml),streptomycin (100 ,ug/ml), and 10 mM N-2-hy-droxyethylpiperazine-N'-2-ethanesulfonic acid(HEPES), pH 7.0. Murine macrophage linesJ774.16, C3C, and D9 are originally derived fromthe J774 parent (14-15) and maintained in Dul-becco's modified Eagle's medium (DMEM) con-taining high glucose (4,500 mg/1) supplementedwith 10% FCS, 1% nonessential amino acids,5% NCTC109, HEPES, and antibiotics as above.Cells were grown in a humidified atmosphere of5% carbon dioxide and 95% air at 37°C in 75cm plastic flasks (Corning, Corning, NY, U.S.A.).For cytotoxicity experiments, cells were har-vested by scraping with a rubber policeman,plated at a density of 5 X 105 cells/ml in 2 cm224-well dishes (Corning), and grown for 36-48hr to approximately 80-90% confluence (1-2 X106 cells/well). For superoxide anion productionassays, cells were gently removed from dishes byincubation in sterile medium containing 1.5%lidocaine at 370C for 5 min. Cells were immedi-ately pelleted and washed twice in lidocaine-freemedium and suspended in medium in siliconized50 ml plastic centrifuge tubes at a concentrationof 1 X 107/ml at 370C with occasional swirling.

Human blood monocytes were purified from50 ml of blood by dilution 1:1 with sterile phos-phate-buffered saline (PBS) and maintained at220C for 1 hr. Diluted blood was then layeredonto an equal volume of HISTOPAQUE- 1 119(Sigma, St. Louis, MO, U.S.A.) in 50 ml centri-fuge tubes and spun 700 X g for 30 min. The toplayer (serum) was removed, heat-inactivated,and saved for medium completion (DMEM, 10%

autologous serum, antibiotics, and HEPES asabove). The mononuclear cell layer was diluted1:2 in sterile PBS and spun for 5 min at 200 x gto remove any red cells. This was repeated for atotal of three washes. Cells were plated in com-pleted medium at a density of 1 X 106 cells/ml in24-well dishes and incubated (as above) for 2days to allow for monocyte differentiation intomacrophages before assay for LeTx cytotoxicity.

Measurement of SuperoxideAnion Production

02- was measured by the SOD-inhibitable reduc-tion of ferricytochrome c (15,16). Briefly, mac-rophages (1 X 107/ml) were kept suspended in50 ml centrifuge tubes at 37°C in the presence of160 ,uM cytochrome c (type III; Sigma) in thepresence or absence of LeTx (PA = 0.1 ,ug/ml, LF= 0.1 ,ug/ml). Reference samples also contained75 ,g/ml of SOD (Sigma). At the indicated times(0-8 hr), 1 ml samples were gently removed,placed into cuvettes, and analyzed in a BeckamnDU-640 spectrophotometer. Reduction of cyto-chrome c was measured at 550 nm. The amountof SOD-inhibitable 02- production was calcu-lated on the basis of an extinction coefficient ofcytochrome c of 21 mM-1 cm-' and is expressedas a function of time in nanomoles/ 107 cells.

Cytotoxicity AssaysFor cell-rupture studies measuring the efflux of51Cr from cells, nearly confluent monolayers ofcells were prepared as described above and incu-bated overnight with the cytoplasmic markerNa25' CrO4 (1 gCi/ml, 1 Ci = 37 GBq; Dupont/NEN, Boston, MA, U.S.A.) (1 1, 12). Cultures werethen washed twice and challenged at 370C withLeTx or controls in the presence or absence ofthe indicated antioxidants. After 4 hr of chal-lenge, culture supernatant was gently removedand counted. The amount of radioactivity re-leased from the cells was calculated withthe formula: percent maximal release = 100 x(toxin-induced release - spontaneous release)/(maximal release - spontaneous release).

Spontaneous release represents radiolabelreleased into the medium without toxin chal-lenge. The maximal release of cytoplasmic labelrepresents the total available cytoplasmic radio-activity at the end of radiolabeling (i.e., beforechallenge) and was determined after cell mem-brane rupture by the addition of 1 ml of 1%Triton X- 100 to each well. In these studies, max-

9


imal release of Cr5' was approximately 12,000cpm/well, and spontaneous release was approx-imately 900 cpm/well.

Measurement of Cytokine ProductionMacrophage production of TNF-a and IL- I3, as afunction of LeTx-challenge (6 hr incubation inthe presence of the indicated antioxidants), wereperformed by ELISA (Genzyme, Boston, MA,U.S.A.) as specified by the manufacturer, andpresented values represent the sum of extracel-lular plus intracellular cytokine pools (11).

In Vivo Lethality AssaysSix-month-old BALB/c mice (approximately 1:1,male:female) were lightly anesthetized and chal-lenged with the indicated amounts of anthraxLeTx via tail vein injections as described previ-ously (11). Nearly simultaneous injection of theindicated antioxidant was performed at a seper-ate tail vein site.

Other Materials and MethodsUnless otherwise indicated, all chemicals werepurchased from Sigma Chemicals. Protein deter-minations were performed with Coomassie pro-tein assay reagent (Pierce, Rockford, IL, U.S.A.)with bovine serum albumin as a standard.

RESULTSProduction of microbicidal ROIs during oxidativeburst is initiated and regulated by the NADPHoxidase complex. The primary species of ROI,superoxide anion (02-), is generated by the sin-gle electron reduction of molecular oxygen. 02-can then be reduced further to form other reac-tive species, such as singlet oxygen (021), hydro-gen peroxide (H202), and hydroxyl radical (-OH).Additionally, H202 can be halogenated to formthe toxic agent hypochlorate (-OC1). These pow-erful oxidants are known to disrupt membraneintegrity, intermediate metabolism, and proteinfunctions. The fact that similar disruptions areseen in macrophages treated with anthrax LeTxled us to ask if ROIs play a role in LeTx action.We have applied three experimental approachesto this question: direct measurement of 02 pro-duction by toxin-challenged M4s; tests of exo-genous and endogenous antioxidants as protec-tive agents against toxin; and determination of

LeTx-sensitivity of a variety of murine and hu-man macrophage cell lines that are altered inoxidative burst potential.

A cultured murine RAW264.7 macrophageline was tested for 02 expression as a functionof time after addition of cytolytic concentrationsof toxin (Fig. 1). Production of 02- as measuredby cytochrome c reduction was first observedapproximately 1 hr after toxin addition, whichcoincided with the onset of cytolysis. Both PAand LF were required for these effects. Maximalstimulation of the oxidative burst by LeTx wasapproximately two to three times that observedwith PMA, a potent stimulator of the respiratoryburst (220 nmol 02/107 cells in 2 hr for LeTxversus 90 nmol 02 107 cells for PMA). Presum-ably the binding, processing, internalization, andputative enzymatic steps account for the lag pe-riod seen with the toxin.

To assess the possibility that the high levelsof ROIs induced by LeTx contribute to lysis, sev-eral types of potential protective agents wereadded to the challenge medium. Data in Fig. 2indicate that exogenous cell-permeable antioxi-dants protect M4s from LeTx in a dose-depen-dent manner. These include reducing agents(BME, DTT), solvents that can trap free radicals(ethanol, DMSO), and vitamins (ascorbate anda-tocopherol). Oxidized DTT conferred no pro-tective effect and served as a control. Mepacrine,an anticancer drug that inhibits phospholipasesand also acts as an oxygen scavenger, was alsoprotective. It is probably the latter propertythat gives mepacrine its protective capabilities,since we found that several other classes ofphospholipase inhibitors were not protective(e.g., 7,7 - dimethyl - (5Z,8Z)- eicosadienoic acid;dimethyl- DL - 2,3 - distearoly - oxypropyl - 2' -hydroxyethyl - ammonium acetate; 0-tricyclo[5.2.1.0(2.6)]DEC -9- YL - dithiocarbonate; anti-inflammatory peptide 2; hexadecylphosphocho-line; data not shown).

Two control experiments suggest that theseagents exert their protective effects at a step sub-sequent to toxin internalization. (1) Antioxidantconcentrations that gave 75-90% protectionfrom LeTx had virtually no effect on EdTx action(Table 1). Since LeTx and EdTx are believed toenter cells by the same pathway (18), this impliesthat the entry process is unaffected. (2) The pro-tective effects of BME, DTT, and EtOH were rap-idly reversed (within 30 min) after removing theinhibitor from the medium (data not shown).Removal of dimethylsulfoxide (DMSO), mepa-crine, and vitamins C and E did not restore cy-


v

.

2 .uco

U

V

0

e-

Agent (mM)

FIG. 1. Time course of lethal toxin-induced0°- production and lysisM4 production of 02- (a) and cytotoxicity (b) werecompared after exposure to LeTx or controls. RAW264.7 cells were grown and prepared as described.Superoxide anion production was determined bymonitoring changes in Absorbance,50 due to cyto-chrome c reduction. Cells to be monitored for cyto-toxicity were loaded with Na25' CrO4 for 16 hr, andwashed twice before challenge. 51Cr release was nor-malized for spontaneous leakage (<1,000 cpm/106cells) and plotted as the percentage of total radioac-tivity in the cells (approximately 12,000 cpm/ 106cells). Challenge concentrations were: PA = 0.1 ,tglml, LF = 0.1 ,g/ml, PMA = 10 ng/ml. Experimentsmeasuring 02 production and cytotoxicity wererun in parallel (triplicate samples); with three exper-imental series run for both 51Cr and 02 production.SEM values were <13% of the indicated value (02measurement) and <5% of the indicated value (5"Crassay) (n = 3). Higher doses of PMA (highest dose =

0.1 j,g/ml) did not significantly increase 02 pro-duction and did not induce cytotoxicity (datanot shown).

u)c) C;

->,.0

6c

4'

Antioxidant (mM)

FIG. 2. Exogenous antioxidants protect M4sfrom lethal toxinRAW 264.7 MO cultures were prepared as describedand challenged with lethal toxin (PA = 0.1 ,ug/ml,LF = 0.1 ,ug/ml) in the presence of the indicatedconcentrations of antioxidants: reducing agents (a),organic solvents (b), or dietary vitamins (c). Protec-tion from the cytolytic effects of the LeTx was scoredusing the 51Cr release assay described in Materialsand Methods. Specific release of approximately 65%of the available 51Cr from cells represents 100% cy-totoxicity due to LeTx. Antioxidants alone, at thegiven concentrations, were not cytotoxic to MO cul-tures. Experiments were run in duplicate with twoexperimental series each. SEM values for each datapoint was <9% of the indicated value.

00

a

~4

0

Time (hours)

11

8C


TABLE 1. Effects of antioxidants on edematoxin and lethal toxin actions in RAW264.7 M4cultures

Percent ofInitial Toxin

Activity

Antioxidant Concentration EdTxa LeTxb

None - 100% 100%DTT 10 mM 91 3

BME 10 mM 94 6

EtOH 10%(V/V) 99 12DMSO 10% (v/v) 108 7

ascorbate 100 mM 95 21

a-tocopherol 100 mM 90 18mepacrine 100 mM 88 10NACc 50 mM 99 5

met" 100 mM 103 9

"EdTx-induced increase of cellular cAMP was determinedby radioimmunoassay, using a cAMP scintillation proximityassay. 100% activity is defined as the amount of cAMP incells (950 fmol/,g protein) 45 min after addition of toxin(0.1 ujig PA + 0.1I,ug EF/ml), after normalization for cAMPpresent in untreated cells.bLeTx-induced cytotoxicity was measured by release of 5"Crfrom prelabeled cells (see Fig. 1 legend).NAC and met were preincubated with cells for 16 hr be-fore toxin challenge (see Fig. 3).

totoxicity, presumably because they are retainedwithin cells.

All cells, including M4s, contain mM con-

centrations of glutathione (GSH, y-glu-cys-gly),which serves (among other purposes) to coun-

teract oxidants (19). The thiol group on the Cysresidue of GSH acts as an endogenous scavenger,

and in the process, GSH is oxidized to a GS-SGdimer (or a GS-S-protein adduct). GS-SG is thenreduced to regenerate GSH or exported from thecell via an ATP-driven pump. When oxidantsoverload GSH's capacity for reductive detoxifica-tion, cell damage occurs. It is possible to increaseGSH concentrations several-fold by supplement-ing cells with certain metabolic precursors, N-acetyl-L-cysteine, or methionine. We thereforetested Mts for LeTx sensitivity after preincuba-tion with these precursors or with the other nor-

mal amino acids as controls. As shown in Fig. 3,both NAC and Met provided strong protectiveeffects against LeTx (Fig. 3a) whereas glycine didnot. Among the other (20) amino acids tested,

e)CZ

O Eo x

5%e

100.001 .01 .1 1 10Glutathione precursor (mM)

100

80t-

Ln 60C3

.= _

22r 40o ._

e 20

0 2 10 50 100 200

Buthionine sulfoximine (mM)

FIG. 3. Glutathione precursors protect M4+sfrom lethal toxin(A) Cellular levels of glutathione, an endogenousantioxidant, were increased by preincubating RAW264.7 cultures with the indicated concentrations ofprecursor (N-acetyl-L-cysteine or methionine) for 16hr prior to LeTx challenge (PA = 0.1 ,ug/ml, LF =

0.1 jig/ml). Cytotoxicity was scored using the 51Cr-release assay described in Materials and Methods.Control amino acids did not protect M4s from LeTxchallenge. Experiments were run in triplicate withtwo experimental series. SEM values for each datapoint were <11% of the indicated value. (B) Pre-venting precursor incorporation into glutathione bybuthionine sulfoximine (BSO), a competitive inhibi-tor of glutathione synthetase, blocks the protectiveability of N-acetyl-L-cysteine. MO cultures were pre-incubated in the presence or absence of theN-acetyl-L-cysteine (50 mM), in the presence of theindicated amounts of BSO (0-200 mM) for 16 hrprior to LeTx challenge (PA = 0.1 ,ug/ml, LF = 0.1jig/ml). LeTx toxicity was assayed by 51Cr-release aspreviously described. Experiments were run in du-plicate with two experimental series. SEM values foreach point were <12% of the indicated value.


TABLE 2. Correlation of M46 sensitivity to LeTx with ability to mount an oxidative burst

PMA-Induced 02-Productiona LeTx-Cytotoxicityb

(nmol/107 (Relative (% of Maximal (Relativecells/hr) Activity) Release) Activity)

A. Murine LinesJ774 21 100% 60 100%J774.16 34 162 69 1153C3 14 67 52 87D-9 11 52 45 75RAW264.7 75 357 79 134IC-21 <1 <5 0 0

B. 1° Human MonocytesLN (CGD Patient) 0 0 <1 <2PH (Control) 23 109 44 73

aSuperoxide anion production was determined by cytochrome c reduction assay after addition of 10 ng/ml PMA, as described inFig. 1.bCytotoxicity was assayed by 51Cr-release, as described in Fig. 1. Relative activities are compared using values derived from J774MO line as the 100% standard (see text).

only glutamic acid showed any protective effectand this was weak (data not shown). While bothGly and Glu are precursors of GSH they are nor-mally not rate limiting in its synthesis. The hy-pothesis that NAC and Met protected cells bystimulating GSH synthesis is supported by thefinding that buthionine sulfoximine (BSO) (20),a competitive inhibitor of y-glu-cys synthetase,relieved the protective effects of these precursorsagainst LeTx action (Fig. 3b). Data in Fig. 3b alsoindicate that shrinking the normal GSH pools byBSO-pretreatment increases MO susceptibility toLeTx. Neither NAC nor Met significantly reducedEdTx activity (Table 1), implying that the protec-tive effects of these amino acids occurred onlyafter toxin entry into cells.

To determine if hyperstimulation of theNADPH oxidase system is ultimately responsiblefor LeTx-induced lysis, we tested various MOlines differing in capability to generate oxidants.The transformed murine peritoneal MO line,J774, was chosen as the standard (sensitivity =100%). Although the J774 line is somewhat at-tenuated in ROI production relative to freshlyprepared murine monocyte cultures, this cell lineis well characterized and there are ROI mutantsderived from it that have increased or decreased

ROI potentials. For example, the J774. 16 mutantis more active and the 3C3 and D9 mutants areless active in ROI production (Table 2). LinesIC-21 and RAW264.7, which are not derivedfrom J774, are also well characterized in thisregard, IC-2 1 being totally unable andRAW264.7 fully able to mount a ROI response.Results presented in Table 2 indicate a strongcorrelation between toxin sensitivity and theability to generate ROIs. The remarkable insen-sitivity of IC-21 (21) cells to LeTx is apparentlydue to the well-documented inability of thesecells to generate superoxide anion (22).

The link between ROIs and LeTx cytotoxicitywas further investigated in cultured primary hu-man M4's. Blood monocyte cultures from anadult patient with a rare autosomal recessiveform of chronic granulomatous disease wastested (23). The patient has a missense mutationand a heterozygous deletion at 16p24 that en-compasses the p22 phox gene encoding the 22 kDlight chain subunit of cytochrome b558. The fail-ure to assemble the b cytochrome of the NADPHoxidase complex renders this patient's phago-cytes unable to generate reactive oxygen inter-mediates. Monocyte cultures from this CGD pa-tient proved to be entirely resistant to LeTx,

13


'0o

none DTr BME EtOH DMSO vit C vit E NAC

Antioxidant

FIG. 4. Antioxidants decrease toxin-inducedexpression of TNFa and IL-1pjRAW 264.7 cultures were prepared as described inMaterials and Methods and challenged with sublyticamount of LeTx known to induce a large cytokineresponse (PA = 0.1 ,ug/ml, LF = 10-6 ,ug/ml) in thepresence of various antioxidants: DTT and BME (10mM), ethanol and DMSO (10%, v/v), vitamins Cand E and mepacrine (100 mM). N-acetyl-L-cysteine(50 mM) was preincubated with MO cells for 16 hrprior to LeTx challenge to allow for incorporationinto glutathione. After 6 hr in toxin-containing me-dia, cells were lysed with 0.1% Triton X- 100 andtotal (cellular plus extracellular) TNFa (a) and IL-1,B(b) concentrations were determined by ELISA. Ex-periments were run in duplicate with two experi-mental series. SEM values for each point were<18% of the indicated value.

while control human monocytes were suscepti-ble (Table 2).

Evidence that MO lysis by LeTx is dependenton ROIs raises the question of the possible in-volvement of the other MO-derived free radicalgas, nitric oxide (NO), and other RNIs. Thesecompounds, like the ROIs, are inducible, micro-bicidal and, at high concentrations, toxic to hostcells. Several lines of evidence suggest that RNIsare unlikely to be involved, however. First, lysis

of M4s by LeTx occurs within 1-2 hr, whilehigh-level NO production is reported to occuronly 10-12 hr after induction. Second, we foundthat neither NMMA nor aminoguanidine, bothof which are inhibitors of NO synthetases, pro-tected M4s from the toxin (data not shown).Lastly, the cell line IC-21, a well characterizedMO model for inducible NO-mediated killing ofpathogens (22), is entirely insensitive to the ef-fects of LeTx. Although these findings make itunlikely that NO is directly responsible for cellkilling, possible regulatory roles for low-levelRNI expression have not been ruled out.

In an earlier study (1 1) we showed: (1) thatmacrophages are required for the lethal action ofLeTx in mice; (2) that sublytic doses of LeTx(orders of magnitude lower than those requiredfor cell lysis) induce production of large amountsof TNFa and IL-1 3 in vitro; and (3) that passiveimmunity to these cytokines enabled mice tosurvive LeTx challenge. We concluded that theM4-cytokine response is central to the lethalshock symptoms of anthrax. It is therefore im-portant to discover if ROIs play a role in regula-tion of M4 cytokine expression and to determinewhether or not scavengers of free radical oxygenprotect animals from LeTx. The effect of variousoxygen scavengers on toxin-induced cytokineproduction by cultured M4s is shown in Fig. 4.The results indicate that DTT, BME, EtOH,DMSO, and NAC significantly reduce theamount of TNFa and IL-1(, generated in re-sponse to low-dose LeTx treatment while vita-mins C and E appear to have little, if any, effect.

Those antioxidants that are relatively non-toxic to animals were also tested in vivo. Six-month-old BALB/c mice were injected withN-acetyl-L-cysteine, mepacrine or vitamin C si-multaneously to LeTx challenge. Although noneof these compounds fully protected mice fromhigh toxin doses, a significant reduction in mor-tality, both in terms of increased survival andincreased time to death, was observed in micereceiving N-acetyl-L-cysteine or mepacrine at in-termediate and low toxin doses (Table 3). Noprotection was observed with ascorbate. Thesedata (Fig. 4, Table 3) suggest that certain antioxi-dants may neutralize the regulatory effects ofROIs, thus lowering cytokine production by hostMO's and reducing the fatal pathologies associ-ated with anthrax lethal toxin in the victim. Whyall antioxidants were not equally protective isunclear but may involve differences in metabolicbreakdown rates.

'A

81.0(D

t

I


TABLE 3. Effect of antioxidants on lethal toxin-induced mortality of mice

Mice Surviving (n = 8)

Day Day Day Day Day Day Day Day DayToxin Dose Antioxidantd 0 1 2 3 4 5 6 7 14e

High Dose" None 8 3 1 0 0

NAC 8 7 5 3 3 2 1 0 0Ascorbate 8 6 4 2 1 0 0Mepacrine 8 6 5 4 2 1 0 - 0

Medium Doseb None 8 6 3 1 1 1 1 1 1

NAC 8 8 7 6 4 3 3 3 3Ascorbate 8 5 4 2 2 1 0 0Mepacrine 8 7 6 4 3 3 3 3 3

Low dose' None 8 7 6 4 3 3 2 2 2

NAC 8 8 7 6 6 6 6 6 6Ascorbate 8 6 5 5 4 4 3 3 3Mepacrine 8 6 6 5 5 5 5 5 5

Anthrax lethal toxin was administered to anesthetized animals in 300 ,ul sterile PBS via the tail vein. Antioxidants or mock wereadministered to animals in 200 Al sterile PBS at the same time as toxin at a second tail vein site."LeTx = 100 ,Ag PA + 20 ,tg LF per animal."LeTx = 50 ,ug PA + 10 ,ug LF per animal.'LeTx = 50 j,g PA + 5 jig LF per animal.dAntioxidant dose = 800 mg per animal.eNo deaths after day 14.

DISCUSSIONWhile macrophages and other professionalphagocytes generate reactive oxygen intermedi-ates for microbicidal purposes, this potent de-fense mechanism constitutes a risk to the hostitself. High concentrations of oxidants are knownto modify vital residues in central homeostaticregulatory proteins (e.g., especially those in-volved in Ca++ homeostasis) (24). Of the nu-cleophilic centers in proteins that are potentialtargets for oxidants, sulfhydryl (SH) groups areby far the most susceptible, and there is goodevidence that modification of protein thiolgroups can result in cell death. ROIs can alsoinitiate a destructive peroxidative cascade thatconsumes a large percentage of plasma mem-brane lipids (24). Both classes of effects are di-sastrous to the cell.

Recently, circumstantial evidence from threedifferent sources hinted that lysis of the macro-phage by anthrax lethal toxin may involve theoxidative burst. First is the finding by Friedlander(10) that, although LeTx is fully capable of en-

tering most cell types, cytotoxicity is apparentlylimited to Mos, a major producer of ROIs. Nophenotype associated with intoxication has beenpresented for other cell types. Second, LeTx-challenged M4s were found to undergo a cas-cade of cellular events that culminate in necroticlysis (12). Those pathologies resemble, at leastsuperficially, the ones seen with high concentra-tions of reactive oxidants. Third, anthrax toxinsequence data reveal no cysteines in the entiretoxin complex, which includes three proteins to-taling 2,278 amino acids (5). We therefore hy-pothesized an ROI link between LeTx actions andMO lysis.

In support of this hypothesis, data presentedhere indicate that M4s generate ROIs in responseto LeTx challenge. The concentrations of super-oxide anion produced in response to the toxinare two to three times higher than that inducedby the potent stimulant PMA (Fig. la) or otherknown activators of the NADPH oxidase com-plex, such as LPS, f-met-leu-phe, or whole killedbacteria (data not shown). Kinetics of LeTx-in-

15


duced lysis closely paralleled release of ROIsfrom cells (Fig. lb), both beginning around60-75 min. We also found that adding exoge-nous antioxidants to M4s or boosting the endo-genous MO glutathione antioxidant system pro-tected cells from the lethal effects of the toxin(Figs. 2 and 3), adding further support for therole of oxidants in cytolysis. Interestingly, sincePMA alone did not induce cell death (Fig. lb), itis possible that the greater ROI burden imposedby LeTx overloads the macrophage's innate ca-pacity to protect itself from its own metabolites.It is also possible that LeTx misdirects ROI releaseto a cellular compartment that lacks endogenousantioxidants. Thus, if these oxidizing agents canbe counteracted by experimental or therapeuticmeans before irreversible damage occurs, the cellsurvives. Finally, the finding that M4s com-pletely deficient in the ability to generate ROIs(both the murine line IC-2 1 and primary humanmonocytes from a patient with CGD; Table 2) arecompletely resistant to LeTx strongly implicates acentral role for the NADPH oxidase complex andoxygen-related toxicities. It is somewhat surpris-ing that the other major ROI-producing cell, theneutrophil, was found in earlier studies not to belysed by the LeTx (25-26). It is possible that thiscell's toxin resistance may be due to a block at anearlier step in the intoxication process. Alterna-tively, PMNs readily degranulate in response tomany signals and it may be that LeTx directssecretion of ROIs in PMNs.

A more challenging question is whether tox-in-induced ROI production plays an extendedrole in anthrax pathogenesis beyond bursting ofthe macrophage. It is now well established inother systems that sublytic concentrations of ox-idants (e.g., H202) can modify gene expressionpatterns of immune cells (27). This gives rise tothe hypothesis that ROIs act as second messen-gers in signal transduction pathways responsiblefor expression of host defense-oriented genes.For instance, NF-KB is an inducible mediator ofimmune cell gene control (27,28). In the restingimmune cell, NF-KB remains inactive in the cy-toplasm, bound to an inhibitory protein calledIKB, for inhibitor of KB. However, upon immunechallenges that initiate the production of ROIs,NF-KB dissociates from IKB and translocates tothe nucleus, where it serves as a transcriptionalactivator of genes such as IgK, cytokines, MHCclass I, HLA, serum amyloid A, j32-microglobulin,HIV, and other viral genes (28). Antioxidantshave been shown to inhibit this activation and toinhibit expression of genes under NF-KB control

(27). Although the precise mechanism of oxi-dant-mediated NF-KB dissociation from lKB isunclear, it seems to be through action upon IKB.It is theorized that ROIs could directly degrade ormodify IKB by modifying susceptible Cys resi-dues. Alternatively, a frequently observed effectof oxidant stress is the induction of proteolysis,especially by activation of metalloproteases,which could then degrade IKB. There are alsoreports of changes in phosphorylation states ofIKB correlating with separation from NF-KB invitro. There have been even more indirect mech-anisms proposed for NF-KB activation.

The current study suggests that ROIs mayhave a regulatory role in MO responses to LeTx,perhaps in a secondary messenger paradigm notunlike that seen with NF-KB. Data presented inFig. 4 and Table 3 indicate that several classes ofantioxidants inhibit LeTx-induced cytokine ex-pression in vitro and protect, at least to somedegree, against the cytokine-mediated shock pa-thologies seen in the toxin-challenged mouse.Why antioxidants did not entirely abolish LeTx-induced cytokine production is not yet known,but it may be that even if small amounts ofoxidants escape neutralization a significant am-plified response can occur. Additionally, antioxi-dants may have differing affinities to the variousROI species, some neutralizing the more damag-ing ones. Why all antioxidants did not protectanimals equally may also involve pharmacolog-ical differences, such as delivery, subcellular lo-calization, or breakdown rates of these agents.

It is interesting to speculate that the molec-ular target (as yet unknown) for LeTx action isinvolved, directly or indirectly, with metabolicregulation of the NADPH oxidase complex, andthat stimulation of the oxidative burst can resultin two tiers of phenomena: changes in MO geneexpression (low dose) and cytolysis (high dose).A dual role for ROIs in anthrax is consistent witha "dual-role" model for LeTx pathogenicity. Atthe onset of anthrax, small amounts of LeTx arereleased from the, as yet, few bacilli. These sub-lytic toxin doses would initiate a small or mod-erate ROI response in the susceptible MO, induc-ing them to express defense-related geneproducts (e.g., cytokines), but not compromisingcellular integrity. We have shown previouslythat some of these products (e.g., IL-113) accu-mulate within the MO cytoplasm during LeTxchallenge (11). As the concentrations of toxinrise in the host, from bacterial counts approach-ing 109/ml blood, lytic thresholds of LeTx (andtherefore of ROIs) are reached, which may cause

P. C. Hanna et al.: Anthrax Toxin Induces Oxidative Burst 17

M4s lysis. Newly formed inflammatory media-tors may then be released suddenly into the cir-culation. This model could account for the dra-matic "sudden-death" noted during anthraxinfection.

In summary, the subversion of the phago-cytic respiratory burst by LeTx represents a newclass of virulence strategies for bacterial patho-gens. For the bacillus, it seems a reasonable tacticto produce a toxin that incapacitates the majorobstacle to proliferation. Additionally, as a patho-gen in which no live animal-to-live animaltransmission has been reported, killing of thehost might help ensure evolutionary success. Forthe anthrax victim, the macrophage represents aparadox. First, at the cellular level, M4s play akey role in both clearance of the invading bacil-lus as well as for generation of lethal, shock-inducing cytokines. Second, at the metaboliclevel, oxidants account for the killing of mostbacteria and for coordinating an aspect of hostdefense by acting as second messengers. But ex-aggerated production of unscavenged ROIs dam-ages the integrity of the MO with resultant re-lease of inflammatory mediators into the tissuesand blood stream. Although antioxidants, due totheir ability to neutralize many LeTx actions,seem to hold potential as anthrax therapies, theymay in fact compromise the phagocyte's overallabilities to kill bacteria via ROIs and thereforemay be counter-productive in ameliorating theinfection. Further knowledge is required to de-termine the best course of antibiotic, antitoxin,anticytokine, and antioxidant therapies neededto care for those people with anthrax.

ACKNOWLEDGMENTSThe authors would like to thank Marla Stein-beck, Kate Beauregard, and Jill Milne for criticaldiscussion of this study and J.M. also for deter-mination of cAMP values. This work was sup-ported by NIH grants AI-22021, AI-22848 (RJC),AI-08649 (PCH), and AI-07118 (BRB). R.A.B.E.is an Established Investigator of the AmericanHeart Association. B.A.K. is the recipient of aCharles A. Janeway, Child Health ResearchScholarship.

REFERENCES1. Koch R. (1881) Mitt. Kaiserliche Gesundheit-

samte 1: 174-206.2. Pasteur L. (1881) De l'attenuation des virus

et de leur retour a la virulence. C. R. Acad. Sci.Agric. Bulg. 92: 429-435.

3. Metchnikoff E. (1905) Immunity in InfectiveDiseases. Cambridge University Press, Lon-don.

4. Stephen J. (1986) Anthrax toxin. In: DornerF, Drews, J (eds). Pharmacology of BacterialToxins. Pergamon Press, Oxford, pp.381-395.

5. Leppla SH. (1991) The anthrax toxin com-plex. In: Alouf J (ed). Sourcebook of BacterialProtein Toxins. Academic Press, New York, pp.277-302.

6. Leppla SH. (1984) Bacillus anthracis calm-odulin-dependent adenylate cyclase: Chem-ical and enzymatic properties and interac-tions with eucaryotic cells. Adv. Cyclic.Nucleotide Prot. Phosphoryl. Res. 17: 189-198.

7. Klimpel KR, Arora N, Leppla SH. (1993) An-thrax toxin lethal factor has homology to thethermolysin-like proteases and displays pro-tease activity. Abstracts, Amer. Soc. Microbiol.Annual Meeting 45 (abstract).

8. Cataldi A, Labruuyere E, Mock M. (1990)Construction and characterization of a pro-tective antigen-deficient Bacillus anthracisstrain. Mol. Microbiol. 4: 1111-1117.

9. Pezard C, Berche P, Mock M. (1991) Contri-bution of individual toxin components tovirulence of Bacillus anthracis. Infect. Immun.59: 3472-3477.

10. Friedlander AM. (1986) Macrophages aresensitive to anthrax lethal toxin through anacid-dependent process. J. Biol. Chem. 261:7123-7126.

11. Hanna PC, Acosta D, Collier RJ. (1993) Onthe role of macrophages in anthrax. Proc.Natl. Acad. Sci. U.S.A. 90: 10198-10201.

12. Hanna PC, Kochi S, Collier RJ. (1992) Bio-chemical and physiological changes inducedby anthrax lethal toxin in J774 macrophage-like cells. Mol. Biol. Cell 3: 1269-1277.

13. Leppla SH. (1988) Production and purifica-tion of anthrax toxin. Meth. Enzymol. 165:103-116.

14. Tanaka Y, Kiyotaki C, Tanowitz H, andBloom BR. (1982) Reconstitution of a vari-ent macrophage cell line defective in oxygenmetabolism with a H202-generating system.Proc. Natl. Acad. Sci. U.S.A. 79: 2584-2558.

15. Damiani G, Kiyotaki C, Soeller W, Sasada M,Peisach J, Bloom BR. (1980) Macrophagevarients in oxygen metabolism. J. Exp. Med.152: 803-822.


16. Johnston RBJ. (1984) Measurement of 02secreted by monocytes and macrophages.Meth. Enzymol. 105: 365-369.

17. Markert M, Andrews PC, Babior BM. (1994)Measurement of 02- production by humanneutrophils. The preparation and assay ofNADPH oxidase-containing particles fromhuman neutrophils. Meth. Enzymol. 105:358-350.

18. Friedlander AM. (1990) The anthrax toxins.In: Saelinger CB (ed). Trafficking of BacterialToxins. CRC Press, Boca Raton, pp. 121-138.

19. (1989) In: Coenzymes and Cofactors, Vol III:Glutathione, Part A. Wiley Interscience,New York.

20. Griffith OW. (1981) Depletion of glutathioneby inhibition of biosynthesis. Meth. Enzymol.77: 59-63.

21. Singh Y, Leppla SH, Bhatnagar R, and Fried-lander AM. (1989) Internalization and pro-cessing of Bacillus anthracis lethal toxin bytoxcin-sensitive and -resistant cells. J. Biol.Chem. 264: 11099-11102.

22. Scott P, James S, Sher A. (1985) The respi-ratory burst is not required for killing ofintracellular and extracellular parasites by a

lymphokine-activated macrophage cell line.Eur. J. Immunol. 15: 553-558.

23. Ezekowitz RAB. (1992) Cronic granuloma-tous disease: An update and a paradigm forthe use of interferon-gamma as adjunct im-munotherapy in infectious diseases. CurrentTop. Microbiol. Immunol. 181: 283-292.

24. Fawthrop DJ, Boobis AR, Davies DS. (1991)Mechanisms of cell death. Arch. Toxicol. 65:437-444.

25. Wright GG, Mandell GL. (1986). Anthraxtoxin blocks priming of neutrophils by lipo-polysaccharide and by muramyl dipeptide. J.Exp. Med. 164: 1700-1709.

26. O'Brien J, Friedlander A, Dreier T, Ezzell J,Leppla S. (1985) Effect of anthrax toxincomponents on human neutrophils. Infect.Immunol. 46: 306-310.

27. Schreck R, Rieber P, Baeuerle PA. (1991)Reactive oxygen intermediates as apparentlywidely used messengers in the activation ofthe NF-KB transcription factor and HIV-1.EMBO J. 10: 2247-2258.

28. Lenardo MJ, Baltimore D. (1989) NF-KB: Apleiotropic mediator of inducable and tissue-specific gene control. Cell 58: 227-229.

role of macrophage oxidative burst in the action of anthrax lethal

Documents