Autophagy-mediated reentry of Francisella tularensisinto the endocytic compartment aftercytoplasmic replicationClaire Checroun*, Tara D. Wehrly*, Elizabeth R. Fischer, Stanley F. Hayes, and Jean Celli*

*Tularemia Pathogenesis Section, Laboratory of Intracellular Parasites, and Microscopy Core Unit, Rocky Mountain Laboratories, National Institute ofAllergy and Infectious Diseases, National Institutes of Health, Hamilton, MT 59840

Edited by E. Peter Greenberg, University of Washington School of Medicine, Seattle, WA, and approved August 3, 2006 (received for review March 6, 2006)

Intracellular bacterial pathogens evade the bactericidal functionsof mammalian cells by physical escape from their phagosome andreplication into the cytoplasm or through the modulation ofphagosome maturation and biogenesis of a membrane-boundreplicative organelle. Here, we detail in murine primary macro-phages the intracellular life cycle of Francisella tularensis, a highlyinfectious bacterium that survives and replicates within mamma-lian cells. After transient interactions with the endocytic pathway,bacteria escaped from their phagosome by 1 h after infection andunderwent replication in the cytoplasm from 4 to 20 h afterinfection. Unexpectedly, the majority of bacteria were subse-quently found to be enclosed within large, juxtanuclear, LAMP-1-positive vacuoles called Francisella-containing vacuoles (FCVs). FCVformation required intracytoplasmic replication of bacteria. Usingelectron and fluorescence microscopy, we observed that the FCVscontained morphologically intact bacteria, despite fusing withlysosomes. FCVs are multimembranous structures that accumulatemonodansylcadaverine and display the autophagy-specific proteinLC3 on their membrane. Formation of FCVs was significantlyinhibited by 3-methyladenine, confirming a role for the autophagicpathway in the biogenesis of these organelles. Taken together, ourresults demonstrate that, via autophagy, F. tularensis reenters theendocytic pathway after cytoplasmic replication, a process thus farundescribed for intracellular pathogens.

macrophage pathogenesis tularemia trafficking

Intracellular bacterial pathogens have devised various strate-gies for circumventing the microbicidal functions of mamma-lian cells, in order to survive and multiply intracellularly (1, 2).The two canonical strategies used by these pathogens are (i)physical escape from the degradative endocytic compartmentsby means of lysis of the phagosomal membrane and replicationwithin the cytoplasm, often accompanied by actin-based motil-ity, as exemplified by Listeria, Shigella, or Rickettsia spp.; or (ii)maintenance inside a pathogen-tailored, membrane-bound com-partment stalled along, or segregated from, the endocytic com-partment, as is the case for Salmonella, Mycobacterium, Chla-mydia, Brucella, or Legionella spp. (13).Francisella tularensis is a Gram-negative, highly infectious,

facultative intracellular pathogen that causes tularemia, a zoo-notic disease affecting humans and other mammals with signif-icant mortality (4). Given its high infectivity and lethality, F.tularensis has raised concerns as a potential bioterrorism agent;however, little is known about its pathogenesis. F. tularensis iscapable of infecting various mammalian cell types, among whichmacrophages constitute a survival and replication niche essentialto the virulence of this bacterium. Early studies in rodentmacrophages have suggested that intracellular Francisella residesinside a phagosome that does not fuse with lysosomes and whoseacidification is essential for intracellular survival (5, 6). How-ever, recent evidence has challenged these results by demon-strating the phagosomal escape of virulent and attenuatedFrancisella strains into the cytoplasm of various murine or

human primary macrophages or macrophage-like cells, followedby bacterial replication (79). Phagosomal escape occurs 24 hpostinfection (p.i.), suggesting some early interactions with theendocytic pathway. After replication, bacterial egress is thoughtto occur via the induction of programmed cell death (4, 10). Ithas also recently been proposed that Francisella-induced celldeath is an innate immune macrophage response to cytoplasmicbacteria aimed at restricting bacterial multiplication (11). Here,we have investigated Francisella interactions with the endocyticcompartment in a synchronized infection model of murinebone-marrow-derivedmacrophages (BMM).We show that, afterlimited interactions with the endocytic pathway, phagosomaldisruption occurs rapidly. Moreover, we describe a postreplica-tion stage of Francisella intracellular trafficking whereby bacte-ria reenter the endocytic compartment via an autophagy-mediated process to reside in large fusogenic vacuoles. Thisfinding represents a unique trafficking event for an intracellularbacterium and suggests that bacterial pathogens can cyclethrough different host cell compartments during their intracel-lular cycle.

Results and DiscussionFrancisella Rapidly Disrupts Its Phagosomal Membrane to Access theMacrophage Cytoplasm. To examine early interactions of Fran-cisella with the endocytic compartment of murine BMMs, wesynchronized bacterial uptake by BMMs as described in Mate-rials and Methods. Five minutes after infection, intracellularbacteria colocalized with the early endosome marker earlyendosome antigen-1 (EEA-1; Fig. 1A), indicating interactionswith early endosomes. Such interactions were transient and wererapidly followed by the acquisition of the late endosomallysosomal marker lysosomal-associated membrane protein 1(LAMP-1; Fig. 1A), indicating a progressive maturation of theFrancisella-containing phagosome along the endocytic pathway.The percentage of LAMP-1-positive phagosomes peaked at 20min p.i. (52 5.7%; all data are given as mean SD) and thendecreased to 7.7 2.9% at 60 min p.i. (Fig. 1A), suggestingsignificant phagosomal escape by 60 min p.i., earlier thanpreviously reported in murine (7) or human (8) phagocytes. Todetermine whether the loss of LAMP-1 colocalization with

Author contributions: C.C. and J.C. designed research; C.C., T.D.W., E.R.F., S.F.H., and J.C.performed research; C.C., E.R.F., S.F.H., and J.C. analyzed data; and C.C. and J.C. wrote thepaper.

The authors declare no conflict of interest.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: BMM, bone marrow-derived macrophage; FCV, Francisella-containing vac-uole; LAMP-1, lysosomal-associated membrane protein 1; LVS, live vaccine strain; MDC,monodansylcadaverine; p.i., postinfection; TEM, transmission electron microscopy; 3-MA,3-methyladenine.

To whom correspondence should be addressed at: Laboratory of Intracellular Parasites,Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, Na-tional Institutes of Health, 903 South Fourth Street, Hamilton, MT 59840. E-mail:jcelli@niaid.nih.gov.

1457814583 PNAS September 26, 2006 vol. 103 no. 39 www.pnas.orgcgidoi10.1073pnas.0601838103

intracellular bacteria was due to phagosomal disruption, wedeveloped a fluorescence microscopy assay of phagosomal in-tegrity, based on the sequential use of digitonin and saponin todifferentially label bacteria that are cytoplasmic or within acompromised phagosome and those enclosed within an intactphagosome (see Supporting Materials and Methods, which ispublished as supporting information on the PNAS web site). Inan infection of BMMs with F. tularensis live vaccine strain (LVS),36 5.7% and 95 3.1% of intracellular bacteria weredetectable at 20 min and 60 min p.i., respectively, by cytoplas-mically delivered antibodies (Fig. 1B; and Fig. 6E, which ispublished as supporting information on the PNAS web site),indicating that these bacteria had significantly escaped fromtheir phagosome or compromised their phagosomal membrane.

Consistently, infected BMMs analyzed by transmission electronmicroscopy (TEM) contained an increasing percentage of bac-teria surrounded by degraded membranes from 30 min to 2 h p.i.(55% at 30 min, 75% at 1h, and 90% at 2 h p.i.; Fig. 1 B and C).Most phagosomes displayed 75% of degraded membranes by1 h p.i. This finding demonstrates that F. tularensis LVS rapidlydisrupts its phagosome after uptake by murine BMMs andreaches the cytoplasm. Thereafter (Fig. 1B and data not shown),95% of the intracellular bacteria were cytoplasmic, and rep-lication occurred from 4 h onward, as reported previously (79).

Intracellular Francisella Localize to Large Vacuoles After Intracyto-plasmic Replication. While examining the fate of replicatingcytoplasmic bacteria, we unexpectedly detected clusters of bac-teria that were not accessible to cytoplasmically delivered anti-bodies at 24 h p.i. (Fig. 2A). Immunostaining of infected BMMsfor the late endosomallysosomal marker LAMP-1 revealed thatthese bacterial clusters were surrounded by a LAMP-1-positivemembrane (Fig. 2B), demonstrating that bacteria were enclosedwithin large endocytic vacuoles. These organelles, named Fran-cisella-containing vacuoles (FCVs), were juxtanuclear, hetero-geneous in size (215 m in diameter) and numbers of enclosedbacteria (data not shown) and were present in cells that alsocontained cytoplasmic bacteria. Formation of FCVs increasedwith time, with the percentage of FCV-containing BMMs reach-ing 46 8.1% at 24 h p.i. and 81 8.4% at 36 h p.i. (Fig. 2C).The kinetics and extent of FCV formation did not dependsignificantly on the apparent moi used, although higher infectionrates favored FCV formation (Fig. 7, which is published assupporting information on the PNAS web site). FCV formationoccurred after the net intracellular growth of Francisella (20 hp.i.; Fig. 2C), suggesting that FCV formation is a postreplicationevent requiring a specific intracellular bacterial load. Impor-tantly, the virulent F. tularensis subsp. holarctica strain FSC200and F. tularensis subsp. tularensis strain Schu S4 also formedFCVs in murine BMMs, with comparable kinetics after cyto-plasmic replication (Fig. 8, which is published as supportinginformation on the PNAS web site), indicating that this phe-nomenon is not due to the attenuation of LVS.We next quantified the proportion of cytoplasmic and vacu-

olar bacteria recovered from infected BMMs by flow cytometryanalysis. In these experiments, digitonin permeabilization wasused to specifically label cytoplasmic bacteria, while all bacteriawere detectable through GFP expression, as described in Mate-rials and Methods. We first controlled for the specificity oflabeling cytoplasmic bacteria. In the absence of cytoplasmicdelivery of Alexa Fluor 647-conjugated anti-Francisella antibod-ies, no intracellular bacteria were labeled (Fig. 3A), whereas themajority of intracellular bacteria (92%) were labeled after BMMlysis (Fig. 3B). At 16 h p.i., 81% of intracellular GFP-expressingLVS were detected by cytoplasmically delivered Alexa Fluor647-conjugated anti-Francisella antibodies (Fig. 3C) and hencewere cytoplasmic, in agreement with the low frequency of FCVin BMMs at this time point (Fig. 2C). In contrast, 72% of bacteriawere inaccessible to cytoplasmically delivered antibodies at 24 hp.i. (Fig. 3D) and hence were vacuolar, demonstrating that themajority of intracellular Francisella are located within FCVs atthis time point.To assess whether FCV formation requires metabolically

active bacteria, we blocked bacterial protein synthesis by addingchloramphenicol at 14 h p.i. and examined FCV formationthereafter. Chloramphenicol treatment stopped bacterial mul-tiplication without significant killing (Fig. 2D) and significantlyprevented FCV formation (22 6.1% at 28 h p.i.) comparedwith untreated BMMs (64 7.1%, P 0.01; Fig. 2D). Thisphenomenon was reversible, inasmuch as bacterial replicationand FCV formation resumed when chloramphenicol was chased

Fig. 1. Phagosomal escape of F. tularensis LVS occurs rapidly in murineBMMs. (A) Confocal microscopy images and quantitation of endocytic markeracquisition by phagosomes during early trafficking events. BMMs were in-fected with LVS for the indicated times, fixed, and processed for immunoflu-orescence using anti-Francisella and either EEA-1 or LAMP-1 antibodies. Co-localization of bacteria with either EEA-1 or LAMP-1 was scored for 100bacteria per condition. Arrows indicate areas magnified in Insets. (Scale bars:10m; Insets, 2m.) (B) Quantitation of Francisella escape into the cytoplasm.LVS-infected BMMs were processed for the phagosomal integrity assay, TEM,or LAMP-1 and Francisella immunofluorescence staining. Phagosomal escapewas measured as the percentage of intracellular bacteria labeled after digi-tonin permeabilization (filled circles, cytoplasmic bacteria) or as the percent-age of bacteria surrounded by degraded membranes (open squares, repre-sentative TEM analysis), and LAMP-1 colocalization with 100 bacteria per timepoint was scored (open circles). At least 50 bacteria per time point wereanalyzed by TEM in each experiment. (C) Representative TEM micrograph ofan LVS-infected BMM at 1 h p.i. The bacterium is surrounded by degradedmembranes, which indicates phagosomal disruption. (Scale bar: 0.5 m.)

Checroun et al. PNAS September 26, 2006 vol. 103 no. 39 14579

MICRO

BIOLO

GY

at 20 h p.i. (data not shown). Hence, bacterial protein synthesis,or the full extent of replication, is required for FCV formation.Induction of programmed cell death has been associated with

Francisella replication (10). To examine whether FCVs origi-nated from phagocytosed bodies from neighboring Francisella-infected BMMs that had undergone cell death, we measured celldeath in LVS-infected BMMs via the lactate dehydrogenaserelease assay and the TUNEL assay. Under our infectionconditions, cell death was very low at 24 h p.i. ( 15%; Fig. 9,which is published as supporting information on the PNAS website). We also quantitated FCV formation either in C57BL6BMMs treated with the pan-caspase inhibitor Z-VAD (Biomol,Plymouth Meeting, PA) to inhibit programmed cell death or in

BMMs derived from ASC mice, which do not undergoprogrammed cell death when infected with Francisella becauseof a defect in the inflammasome-associated adaptor protein ASC(11). In both conditions, FCV formation was not affectedcompared with untreated C57BL6 BMMs (Fig. 9). Collectively,these results rule out an exogenous, cell-death-related origin forFCVs.Next, FCV ultrastructure was examined by TEM. At 24 h p.i.,

many individual bacteria or groups of bacteria were enclosed inmembrane-bound compartments (Fig. 4 A and B), a feature thatwas not observed at earlier time points (data not shown). In mostcases, FCVs were large membrane-bound vacuoles filled withundegraded bacteria (Fig. 4 B and E). Interestingly, thesecompartments displayed double-membrane structures reminis-cent of autophagic vacuoles (Fig. 4 C and D, arrows).

FCV Formation Requires Autophagy. Because the ultrastructure ofFCVs suggested an autophagic nature, we sought to determinewhether these organelles displayed additional autophagic fea-tures. We first examined whether FCVs accumulated the auto-phagic probe monodansylcadaverine (MDC). At 24 h p.i., 74 8.9% of the LAMP-1-positive FCVs strongly labeled with MDC(Fig. 5 A and C), indicating an autophagic origin for FCVs. Toconfirm this result, we expressed in BMMs aGFP fusion with theautophagosomal membrane-associated protein LC3 (12) thatspecifically labels autophagosomes and examined GFP-LC3recruitment on FCVs. GFP-LC3 was enriched on the majority of

Fig. 2. Intracellular Francisella become enclosed in large vacuoles afterintracytoplasmic replication. (A) Confocal micrographs of an LVS-infectedBMM at 24 h p.i., subjected to the phagosomal integrity assay. Cytoplasmicbacteria (red and green, appearing yellow in the overlay) are labeled afterdigitonin permeabilization, whereas clustered bacteria are detected onlyafter saponin permeabilization (red). Calnexin staining (blue) allows detec-tion of digitonin-permeabilized cells. (B) Confocal micrographs of an LVS-infected BMM at 24 h p.i. BMMs were infected with LVS, fixed, and processedfor immunofluorescence with Francisella LPS and LAMP-1 antibodies. Bacte-rial clusters (green) are enclosed in LAMP-1-positive, membrane-bound com-partments (red) termed Francisella-containing vacuoles (FCVs). Arrows indi-cate FCVs. (Scale bars: 10 m.) (C) Kinetics of intracellular replication and FCVformation. BMMs were infected with LVS for the indicated times. Intracellularbacteria were enumerated from cfus, and FCV formation was measured as thepercentage of infected cells harboring LAMP-1-positive FCVs. (D) Effect ofinhibition of bacterial protein synthesis on FCV formation and replication.BMMs were infected with LVS and left untreated or treated at 14 h p.i. with10 gml chloramphenicol (indicated by arrow), and FCV formation (filledshapes) or intracellular growth (open shapes; cfu) was measured. The asteriskindicates statistically significant differences between control and chloram-phenicol-treated BMMs at 28 h p.i. (P 0.05, two-tailed unpaired Students ttest).

Fig. 3. Flow-cytometry-based quantitation of cytoplasmic and vacuolarFrancisella at 16 and 24 h p.i. BMMs were infected with GFP-expressing LVS,and cytoplasmic bacteria were labeled using AlexaFluor 647-conjugated anti-Francisella antibodies after digitonin permeabilization. (A) Negative controlof AlexaFluor 647 labeling of cytoplasmic, GFP-expressing bacteria whenBMMs infected for 24 h were processed without digitonin permeabilization.(B) Positive control of AlexaFluor 647 labeling of GFP-expressing bacteriarecovered from BMMs infected for 24 h. The majority (92%) were labeled aftertotal lysis. (C) Analysis of GFP-expressing bacteria recovered from BMMsinfected for 16 h, showing that the majority (81%) were labeled with AlexaFluor 647 and hence are cytoplasmic. (D) Analysis of GFP-expressing bacteriarecovered from BMMs infected for 24 h, showing that the majority (72%) werenot labeled with AlexaFluor 647 and hence are vacuolar. Percentages shownin red refer to the proportions of vacuolar bacteria. Data are from oneexperiment representative of three.

14580 www.pnas.orgcgidoi10.1073pnas.0601838103 Checroun et al.

FCVs at 24 h p.i. (62 9.6% of LAMP-1-positive FCVs; Fig. 5B and C). GFP-LC3 enrichment was dependent on LC3 associ-ation with membranes because the GFP-LC3C22, G120A mutantform, which cannot be processed and conjugated to autophago-somal membranes (12), was not significantly recruited to FCVs(8.8 1.9%; Fig. 5 B and C). Taken together, these resultsdemonstrate the autophagic origin of FCVs. To extend thesefindings, we examined the effect of autophagy inhibition on FCVformation. Infected BMMs were treated at 14 h p.i. with 5 mM3-methyladenine (3-MA), and FCV formation was analyzedat 24 h p.i. In uninfected cells, such treatment inhibitedautophagosome formation by 76% upon amino acid starvation(Fig. 5D). 3-MA significantly reduced FCV formation such thatonly 18 4.5% of BMMs contained FCVs, compared with43.3 11.1% of control BMMs (P 0.04; Fig. 4D), yet treatmentdid not significantly affect the intracellular yield of bacteria (Fig.10, which is published as supporting information on the PNAS

web site). Collectively, these results demonstrate that FCVsdisplay autophagic features and require autophagy for theirformation but are not involved in either intracellular prolifera-tion or killing.

FCVs Interact with Late EndocyticLysosomal Compartments. Giventhe presence of LAMP-1 on FCVmembranes, we tested whetherFCVs are mature, fusogenic autolysosomes. With live cell im-aging, 74 2.9% of FCVs formed by GFP-expressing LVS at24 h p.i. accumulated Alexa Fluor 568-dextran, a fluorescent

Fig. 4. FCVs are double membrane-bound vacuoles containing intact bac-teria. BMMs were infected with LVS and processed for TEM at 24 h p.i. (A andB) TEM micrographs showing individual bacteria or groups of bacteria en-closed by double membranes (indicated by arrows). (C and D) Magnificationsof the boxed areas in A and B, respectively, showing double membranes(arrows) surrounding bacteria. (E) Ultrastructure of a typical FCV showingclustered, intact bacteria enclosed by a membrane. (Scale bars:A, B, and E, 0.5m; C and D, 0.2 m.)

Fig. 5. FCVs display autophagic features, and their biogenesis requiresautophagy. (A) LVS-infected BMMs were labeled with MDC (blue) beforefixation at 24 h p.i. and immunofluorescence staining of Francisella (green)and LAMP-1 (red). (B) BMMs were transduced to express GFP-LC3 or GFP-LC3C22, G120A (GFP-LC3C22) and infected with LVS for 24 h p.i. before fixationand immunostaining of Francisella (blue), GFP (green), and LAMP-1 (red).Insets are single-channel fluorescence images of FCVs. Arrows indicate FCVs.(C) Quantitation of MDC accumulation and recruitment of GFP-LC3 or GFP-LC3C22, G120A on LAMP-1-positive FCVs at 24 h p.i. For each condition, 100 FCVswere scored per experiment. (D) Effect of autophagy inhibition on FCVformation. (Left) LVS-infected BMMs were treated with 5 mM 3-MA at 14 hp.i., and FCV formation was scored at 24 h p.i. As a positive control forautophagy inhibition, uninfected, GFP-LC3-expressing BMMs were left un-treated or pretreated with 3-MA for 1 h, then starved for 4 h to induceautophagosome formation. (Right) Autophagy was then scored as the per-centage of cells containing GFP-LC3-positive vesicles. Asterisks indicate datasignificantly different from untreated controls (P 0.05, two-tailed unpairedStudents t test).

Checroun et al. PNAS September 26, 2006 vol. 103 no. 39 14581

MICRO

BIOLO

GY

f luid-phase marker preloaded into lysosomes (Fig. 11 A and C,which is published as supporting information on the PNAS website). Consistently, 76 11% of FCVs contained the luminallysosomal hydrolase cathepsin D (Fig. 11 B and C), confirmingthat FCVs fuse with lysosomes, and 73 5.6% of FCVs alsoaccumulated the acidotropic probe LysoTracker Red DND-99(Fig. 11 C and D), indicating that they are acidic organelles.Hence, FCVs are fusogenic, matured autolysosomes.Autophagy in mammalian cells has been associated with

innate defense mechanisms against intracellular pathogens, re-trieving cytoplasmic bacteria for degradation (13, 14) or resum-ing the maturation of phagosomes stalled along the endocyticpathway (15). Predictably, some cytoplasmic pathogens, such asShigella, have evolved strategies to avoid autophagy (16),whereas membrane-bound pathogens such as Legionella pneu-mophila, Porphyromonas gingivalis,Coxiella burnetii, and Brucellaabortus may take advantage of this mammalian process (1720).Here, we show that, after cytoplasmic replication, F. tularensisreenters the endocytic pathway via the autophagic machinery ofprimary murine macrophages. In addition to uncovering adistinctive stage of the Francisella intracellular cycle, we dem-onstrate that an intracellular pathogen with a cytoplasmic rep-lication phase can subsequently resume interactions with theendocytic compartment.FCV formation could be a host- or bacterial-driven process.

Given the role of autophagy in innate defense against intracel-lular pathogens, FCVs could result from a macrophage responseto the intracellular bacterial load, aimed at restricting bacterialmultiplication. This hypothesis is, nonetheless, questioned by thefact that autophagy is not induced earlier when Francisellainitially reach the cytoplasm, inasmuch as autophagy occursrapidly in response to other cytoplasmic bacteria, includingGroup A Streptococcus (13), Salmonella (14), or Listeria ren-dered metabolically inactive with chloramphenicol (21). Incontrast, chloramphenicol treatment of Francisella-infectedmacrophages during (Fig. 2D) or before replication (4 h p.i., datanot shown) did not induce the autophagic uptake of bacteria.Instead, Francisella remained cytoplasmic and intact for up to16 h posttreatment (data not shown). Because FCV biogenesisrequires active bacterial protein synthesis andor substantialreplication of bacteria, FCV formation could be a temporallyregulated, bacterially driven process. Francisella appears toprevent a macrophage autophagic response before and duringbacterial replication. Once replication is completed, the macro-phage autophagic response develops or is actively triggered bythe bacteria, inducing their autophagic uptake and allowing FCVformation. The biogenesis of such organelles, which is obviouslya significant event of the Francisella intracellular cycle, involvesthe majority of bacteria and occurs with both attenuated andhighly virulent strains. Additionally, the report of bacteriaenclosed by double membranes within peritoneal cells of miceinfected with LVS (6) potentiates the significance of FCVs bysuggesting that they form in vivo.The reentry of Francisella into the endocytic degradative

pathway after cytoplasmic replication is a unique phenomenonin the trafficking of intracellular pathogens. Because FCVspossess features of autolysosomes, intravacuolar Francisellalikely encounter a lysosomal environment. Yet no sign ofbacterial degradation was observed in most FCVs, suggestingthat Francisella can resist phagolysosomal bactericidal conditionsonce it reenters the endocytic pathway. The role of FCVs in theintracellular cycle and pathogenesis of Francisella remains to beelucidated. FCV formation does not affect intracellular survivaland replication, suggesting that the vacuoles are not involved inbacterial proliferation. As a postreplication stage, FCVs maysubject the bacteria to environmental cues required to induce asubset of genes involved in egress andor reinfection. Given theirinteractions with the endocytic compartment, FCVs might also

provide intracellular Francisella with access to the macrophagemembrane trafficking functions, promoting bacterial egressthrough exocytosis. Studies of the biogenesis and dynamics ofFCVs could address the role of these intracellular organelles inFrancisella pathogenesis. Given the increasing prominence ofautophagy in bacterial pathogenesis, future studies of FCVs mayalso provide much-needed information about this importantmammalian response.

Materials and MethodsBacterial Strains and Culture Conditions. F. tularensis subsp. holarc-tica strains LVS (ATCC 29684) and FSC200 (22) were obtainedfrom Francis Nano (University of Victoria, Victoria, BC, Can-ada) and Anders Sjostedt (Umea University, Umea, Sweden),respectively. F. tularensis subsp. tularensis strain Schu S4 (23) wasobtained from Rick Lyons (University of New Mexico, Albu-querque, NM). To generate GFP-expressing bacteria, the plas-mid pFNLTP6groE-gfp (24) was introduced by electroporationinto the LVS strain, as described in ref. 24. Bacteria were grownon cysteine heart agar (Becton Dickinson, Sparks, MD) supple-mented with 9% heated sheep blood (CHAB), and kanamycin(10 gml) when required, for 3 days at 37C in 7% CO2. Forinfections with LVS, two to three fresh colonies were inoculatedinto tryptic soy broth (Becton Dickinson) supplemented with0.1% L-cysteine (TSB-C) and grown overnight at 37C withshaking to an OD600 1.5. Virulent holarctica or tularensisstrains were scraped off freshly streaked CHAB plates andresuspended in TSB-C before infection. To enumerate viableintracellular bacteria, infected macrophages were lysed in steriledistilled water, and serial dilutions were plated on CHAB plates.

Macrophage Culture and Infection. To generate BMMs, bonemarrow cells were collected from dissected femurs of 6- to12-week-old C57BL6 female mice (Harlan, Indianapolis, IN),and macrophages were derived in 150-mm non-tissue-culture-treated dishes, as described in ref. 25. After 5 days, looselyadherent BMMs were washed with PBS, harvested by incubationin chilled cation-free PBS on ice for 10 min, resuspended incomplete medium, and seeded onto 12-mm glass coverslips in24-well plates (immunofluorescence, 1 105 per well) orWillCo-dish glass-bottomed 35-mm dishes (live cell imaging, 1105 per dish; WillCo Wells BV, Amsterdam, The Netherlands).BMMs were further cultured for 2 days before infection. BMMsderived from ASC knockout mice were kindly provided byDavid Weiss and Denise Monack (Stanford University, Stan-ford, CA).BMM infections were performed at an moi of 50 by centri-

fuging bacteria suspended in complete medium onto prechilledmacrophages at 400 g for 10 min at 4C. BMMs were thenrapidly warmed to 37C for 2 min in a water bath to triggerphagocytosis and further incubated for a total of 20 min at 37Cin 7% CO2. BMMs were extensively washed with DMEM toremove extracellular bacteria and incubated for 40 min incomplete medium and then for 60 min in 100 gml gentamicin-containing medium to kill the remaining extracellular bacteria.Thereafter, infected BMMs were incubated in gentamicin-freemedium. Such conditions allowed for a synchronized entry ofbacteria, leading to 29 3.8% of BMMs infected with one to twobacteria (Fig. 7). At each time point, BMMs were washed threetimes with PBS before processing.When required, chloramphen-icol (Sigma, St. Louis, MO) was added (10 gml). Autophagywas induced by incubating cells in Hanks balanced salt solution(Mediatech, Herndon, VA) supplemented with 1 gliter D-glucose for 4 h to mimic amino acid starvation conditions.Autophagy was inhibited by treating BMMs with 5 mM 3-MA(FlukaSigmaAldrich, St. Louis, MO), and the percentage ofGFP-LC3-expressing BMMs containing LC3-positive vesicleswas scored.

14582 www.pnas.orgcgidoi10.1073pnas.0601838103 Checroun et al.

Fluorescence Microscopy. BMMs were fixed with 3% paraformal-dehyde in PBS, pH 7.4, for 10 min at 37C and processed forimmunofluorescence staining as described in ref. 20. The pri-mary antibodies used were mouse monoclonal anti-F. tularensisLPS (US Biologicals, Swampscott, MA), rat monoclonal anti-mouse LAMP-1 (1D4B; developed by J. T. August; obtainedfrom the Developmental Studies Hybridoma Bank; developedunder the auspices of the National Institute of Child Health andHuman Development; and maintained by the Department ofBiological Sciences, University of Iowa, Iowa City, IA), rabbitpolyclonal anticalnexin and mouse monoclonal antiprotein di-sulfide isomerase (PDI; Stressgen Biotechnologies, Victoria, BC,Canada), rabbit polyclonal anti-human cathepsin D (provided byStuart Kornfeld, Washington University, St. Louis, MO), goatpolyclonal anti-EEA1 (N-19; Santa Cruz Biotechnologies, SantaCruz, CA), and rabbit polyclonal anti-GFP (Molecular Probes,Eugene, OR). The secondary antibodies used were Alexa Fluor488-conjugated and Alexa Fluor 568-conjugated (MolecularProbes) and cyanin-5-conjugated (Jackson ImmunoResearch,West Grove, PA) donkey anti-mouse, anti-rat, anti-rabbit, andanti-goat antibodies. To label autophagosomes, BMMs wereincubated with 50 MMDC for 1 h, followed by a 30-min chasebefore fixation. Samples were observed on either a Nikon(Melville, NY) Eclipse E800 epifluorescence microscope forquantitative analysis or a Carl Zeiss (Thornwood, NY) LSM 510confocal laser scanning microscope for quantitative analysis andimage acquisition. MDC fluorescence was imaged by epifluo-rescence using a DAPI filter and a Carl Zeiss Axiocam digitalcamera mounted on the LSM 510 confocal microscope, con-comitant with the confocal acquisitions of the other fluorescentemissions. Confocal images of 1,024 1,024 pixels were acquiredand assembled using Adobe Photoshop CS software (AdobeSystems, San Jose, CA).

Flow Cytometry Quantitation of Vacuolar Francisella.To evaluate theproportions of cytoplasmic and vacuolar Francisella, BMMs insix-well plates (5 105 per well) were infected with GFP-expressing LVS and washed with KHM buffer (110 mM potas-sium acetate20 mM Hepes2 mM MgCl2, pH 7.3), and theirplasma membranes were selectively permeabilized with 50gml digitonin in KHM buffer for 1 min at room temperature.After washing with KHM buffer, mouse anti-Francisella LPS

antibodies conjugated to Alexa Fluor 647 (Molecular Probes)were specifically delivered to the macrophage cytoplasm (seeSupporting Materials and Methods) for 15 min at 37C to labelcytoplasmic bacteria. After washing in KHMbuffer, BMMs werelysed in water and lysates were centrifuged at 200 g for 5 minto remove cellular debris. Supernatants containing all intracel-lular bacteria were analyzed by using a FACSCalibur flowcytometer (BD Biosciences, San Jose, CA) for GFP and AlexaFluor 647 fluorescence. Data were analyzed with FlowJo soft-ware, version 6.3.2 (Tree Star, Ashland, OR). Under theseconditions, cytoplasmic bacteria were detected as both GFP- andAlexa Fluor 647-positive, whereas vacuolar bacteria were onlyGFP-positive. At least 90% of intracellular bacteria were GFP-positive after a 24-h infection (data not shown), confirming thatthe analysis was performed on the majority of the bacterialpopulation.

TEM. Infected BMMs on 12-mm Aclar coverslips were processedas described in ref. 26, except that BMMs were fixed for 24 h andpostfixed in 1%OsO4. Samples were viewed in a Hitachi (Tokyo,Japan) H7500 TEM at 80 kV, fitted with a Hamamatsu (Bridge-water, NJ) CCD camera C474295 and Advantage HRHR-Bdigital image software (AMT, Danvers, MA) for imaging. Toassess phagosomal membrane integrity, 50100 phagosomeswere analyzed per condition in two independent experiments.FCV ultrastructural analysis was performed on 50 vacuoles inthree independent experiments.

Statistical Analyses.All data are given as mean SD from at leastthree independent experiments, unless otherwise stated. Statis-tical analyses were performed by using either a one-wayANOVAwith Tukey posttest or an unpaired, two-tailed Student t test. AP value 0.05 was considered significant.

We thank Leigh Knodler, Rey Carabeo, and Samantha Gruenheid forcritical reading of the manuscript; Francis Nano, Anders Sjostedt, RickLyons, Tamotsu Yoshimori, Stuart Kornfeld, David Weiss, and DeniseMonack for providing strains, plasmids, antibodies and macrophages;Holger Lorenz for technical advice on digitonin permeabilization; andRon Messer for advice and help with flow cytometry. This work wassupported by the Intramural Research Program of the National Institutesof Health, National Institute of Allergy and Infectious Diseases.

1. Knodler LA, Celli J, Finlay BB (2001) Nat Rev Mol Cell Biol 2:578588.2. Meresse S, Steele-Mortimer O, Moreno E, Desjardins M, Finlay B, Gorvel JP

(1999) Nat Cell Biol 1:E183E188.3. Amer AO, Swanson MS (2002) Curr Opin Microbiol 5:5661.4. Oyston PC, Sjostedt A, Titball RW (2004) Nat Rev Microbiol 2:967978.5. Anthony LD, Burke RD, Nano FE (1991) Infect Immun 59:32913296.6. Fortier AH, Leiby DA, Narayanan RB, Asafoadjei E, Crawford RM, Nacy CA,

Meltzer MS (1995) Infect Immun 63:14781483.7. Golovliov I, Baranov V, Krocova Z, Kovarova H, Sjostedt A (2003) Infect

Immun 71:59405950.8. Clemens DL, Lee BY, Horwitz MA (2004) Infect Immun 72:32043217.9. Santic M, Molmeret M, Abu Kwaik Y (2005) Cell Microbiol 7:957967.10. Lai XH, Golovliov I, Sjostedt A (2001) Infect Immun 69:46914694.11. Mariathasan S, Weiss DS, Dixit VM, Monack DM (2005) J Exp Med

202:10431049.12. Kabeya Y, Mizushima N, Ueno T, Yamamoto A, Kirisako T, Noda T,

Kominami E, Ohsumi Y, Yoshimori T (2000) EMBO J 19:57205728.13. Nakagawa I, Amano A,Mizushima N, Yamamoto A, Yamaguchi H, Kamimoto

T, Nara A, Funao J, Nakata M, Tsuda K, et al. (2004) Science 306:10371040.14. Birmingham CL, Smith AC, Bakowski MA, Yoshimori T, Brumell JH (2006)

J Biol Chem 281:1137411383.

15. Gutierrez MG, Master SS, Singh SB, Taylor GA, Colombo MI, Deretic V(2004) Cell 119:753766.

16. Ogawa M, Yoshimori T, Suzuki T, Sagara H, Mizushima N, Sasakawa C (2005)Science 307:727731.

17. Amer AO, Swanson MS (2005) Cell Microbiol 7:765778.18. Dorn BR, Dunn WA, Jr, Progulske-Fox A (2001) Infect Immun

69:56985708.19. Gutierrez MG, Vazquez CL, Munafo DB, Zoppino FC, Beron W, Rabinovitch

M, Colombo MI (2005) Cell Microbiol 7:981993.20. Pizarro-Cerda J, Meresse S, Parton RG, van der Goot G, Sola-Landa A,

Lopez-Goni I, Moreno E, Gorvel JP (1998) Infect Immun 66:57115724.21. Rich KA, Burkett C, Webster P (2003) Cell Microbiol 5:455468.22. Johansson A, Berglund L, Eriksson U, Goransson I, Wollin R, Forsman M,

Tarnvik A, Sjostedt A (2000) J Clin Microbiol 38:2226.23. Eigelsbach HT, Braun W, Herring RD (1951) J Bacteriol 61:557569.24. Maier TM, Havig A, Casey M, Nano FE, Frank DW, Zahrt TC (2004) Appl

Environ Microbiol 70:75117519.25. de Chastellier C, Frehel C, Offredo C, Skamene E (1993) Infect Immun

61:37753784.26. Rockey DD, Fischer ER, Hackstadt T (1996) Infect Immun 64:42694278.

Checroun et al. PNAS September 26, 2006 vol. 103 no. 39 14583

MICRO

BIOLO

GY