INFECTION AND IMMUNITY, Dec. 2006, p. 6590–6598 Vol. 74, No. 120019-9567/06/$08.00�0 doi:10.1128/IAI.00868-06Copyright © 2006, American Society for Microbiology. All Rights Reserved.

Mac-1� Cells Are the Predominant Subset in the Early HepaticLesions of Mice Infected with Francisella tularensis�

John W. Rasmussen,1 Jeronimo Cello,1 Horacio Gil,1† Colin A. Forestal,1 Martha B. Furie,1,2,3

David G. Thanassi,1,3 and Jorge L. Benach1,3*Center for Infectious Diseases,1 Department of Pathology,2 and Department of Molecular Genetics and

Microbiology,3 Stony Brook University, Stony Brook, New York 11794

Received 31 May 2006/Returned for modification 7 July 2006/Accepted 17 September 2006

The cell composition of early hepatic lesions of experimental murine tularemia has not been characterized withspecific markers. The appearance of multiple granulomatous-necrotic lesions in the liver correlates with a markedincrease in the levels of serum alanine transferase and lactate dehydrogenase. Francisella tularensis, detected byspecific antibodies, can be first noted by day 1 and becomes associated with the lesions by 5 days postinoculation.These lesions become necrotic, with some evidence of in situ apoptosis. The lesions do not contain B, T, or NK cells.Rather, the lesions are largely composed of two subpopulations of Mac-1� cells that are associated with the bacteria.Gr-1� Mac-1� immature myeloid cells and major histocompatibility complex class II-positive (MHC-II�) Mac-1�

macrophages were the most abundant cell phenotypes found in the granuloma and are likely major contributors incontrolling the infection in its early stages. Our findings have shown that there is an early development of hepaticlesions where F. tularensis colocalizes with both Gr-1� Mac-1� and MHC-II� Mac-1� cells.

Francisella tularensis is a facultative intracellular bacterialzoonotic pathogen that causes a disease known as tularemia.Human infection can follow ingestion of contaminated food orwater, contact of open skin wounds with infected animal car-casses, bites from various blood-sucking arthropods, or inha-lation of aerosolized bacteria (2, 17, 18). In its natural setting,tularemia is the third most common tick-borne disease; it isalso the second most common laboratory-associated infectionin the United States (25, 39). Although tularemia has declinedsteadily since World War II, interest in Francisella continuesnot only as a model of study for intracellular bacteria but alsodue to its potential use as a biological weapon (16).

Two subspecies of F. tularensis, F. tularensis subsp. tularensis(type A) and F. tularensis subsp. holarctica (type B), are highlyinfectious in humans. Type B strains cause only moderateillness and are usually nonfatal. Meanwhile, type A strainscause potentially lethal infections in humans, particularly fol-lowing exposure to aerosolized organisms. For this reason,type A F. tularensis is considered a potential biological warfareagent (16) and has been classified as a category A agent ofbioterrorism by the Centers for Disease Control and Preven-tion. An attenuated live vaccine strain (LVS) derived from typeB F. tularensis does not cause illness in humans but causes adisease in mice that resembles human tularemia (3, 21). There-fore, the LVS strain has been used extensively for experimentalstudies on the pathogenesis of tularemia. The involvement ofthe liver in both clinical and experimental tularemia regardless

of the portal of entry or host species has been known for a longtime (5, 18, 42, 43).

Single or multiple randomly distributed irregular microab-scesses of mononuclear cells and a few neutrophils in the hepaticparenchyma have been seen as early as 1 day postinoculation(DPI) in murine tularemia (13). These microabscesses grow intowell-circumscribed granulomas composed mostly of macrophagesby 4 to 5 DPI. Hepatocytes can be infected by F. tularensis, andthese cells can harbor large numbers of bacteria (11–13, 15, 33; H.Zheng and M. B. Furie, unpublished observations). With time,the developing granulomas become prominent in the entire liver,and the cytoplasm of many hepatocytes becomes completely filledwith bacteria (15). Liver infection from LVS has also been used tostudy protective immunity and mouse strain susceptibility (12).Livers from LVS-immunized C57BL/6 mice contained small- tomedium-sized areas of focal inflammatory necrosis with both ne-crotic and apoptotic hepatocytes, while the liver pathology ofLVS-immunized BALB/c mice was milder. This mouse strain wasmore resistant to intradermal and aerosol inoculation (12). Thus,in murine tularemia, pathogen virulence, genetic background ofthe host, and route of inoculation all play a role in pathogenesis,specifically in the liver.

While the liver pathology of tularemia is well recognized ina number of experimental models, characterization of the in-filtrating cells of the lesions has not been done with specificmarkers, nor, for that matter, has the process of cell death inliver infection been documented specifically. In this study, weused experimental sublethal tularemia infection of C3H/HeNmice to characterize the liver infiltrates and other signs ofhepatic dysfunction. We report that subpopulations of cellsexpressing Mac-1 associate with F. tularensis during the earlydevelopment of hepatic lesions.

MATERIALS AND METHODS

Bacteria. F. tularensis LVS (ATCC 29684; American Type Culture Collection,Manassas, VA) was cultured in Mueller-Hinton (MH) broth (BD Biosciences,

* Corresponding author. Mailing address: Center for Infectious Dis-eases, 5120 Centers for Molecular Medicine, Stony Brook, NY 11794-5120. Phone: (631) 632-4225. Fax: (631) 632-4294. E-mail: jbenach@notes.cc.sunysb.edu.

† Present address: Laboratorio de Espiroquetas y Patogenos Espe-ciales, Centro Nacional de Microbiologıa, Instituto de Salud Carlos III,Majadahonda 28220, Spain.

� Published ahead of print on 25 September 2006.

6590

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

Sparks, MD) supplemented with 2% IsoVitaleX Enrichment (BD Biosciences),0.1% glucose, 63 mM CaCl2, 53 mM MgCl2, and 34 mM ferric pyrophosphateand incubated at 37°C in 5% carbon dioxide. Mid-log-phase bacteria were frozenin 1-ml aliquots at �80°C (20, 24). Bacteria from frozen aliquots were grown onChocolate II agar (BD Biosciences) at 37°C in 5% carbon dioxide. Single colo-nies were inoculated into prewarmed (37°C) MH broth, grown for 16 to 18 h,serially diluted in MH broth to 105 CFU/ml as determined by an optical densityreading at 600 nm, and verified by growth on Chocolate II agar.

Mice. Female C3H/HeN mice were purchased from Charles River Laborato-ries (Wilmington, MA) and used from 6 to 10 weeks of age. All mice were housedin microisolator cages with free access to food and water. Mice received intra-dermal injections of 105 CFU of F. tularensis LVS. At various time pointspostinoculation, mice were euthanized, which was immediately followed by bloodand organ collection. All animal procedures were approved by an institutionalreview board. The number of viable bacteria in blood was determined by streak-ing samples onto Chocolate II agar plates and counting the numbers of colonies.

White blood cell counts and enzymes. Total white blood cell counts were donemanually by use of Petroff-Hausser chambers. Differentials were determined byenumeration from Giemsa-stained peripheral blood smears. Serum clinicalchemistries for liver and kidney function were done by the Research AnimalDiagnostic Laboratory, Columbia, MO. The tests included determinations foralanine transferase (ALT), alkaline phosphatase, direct and total bilirubin, lac-tate dehydrogenase (LDH), creatinine, and blood urea nitrogen.

Cell isolation. Following euthanization of mice, livers were perfused with largevolumes of Hanks’ balanced salt solution (Invitrogen, Grand Island, NY) untilthe organ was blanched. Once removed, livers were minced and incubated indigestive medium (0.05% collagenase A [Roche, Indianapolis, IN] and 0.002%DNase I [Sigma, St. Louis, MO] in Hanks’ balanced salt solution) at 37°C and at80 rpm for 30 min to provide a single-cell suspension of tissue. Cells werecollected and centrifuged for 10 min at 400 � g followed by suspension on aPercoll gradient (GE Healthcare, Piscataway, NJ) and centrifugation for 30 minat room temperature (RT) at 400 � g in a swing-out rotor. Mononuclear cellswere enumerated by using Petroff-Hausser chambers prior to antibody stainingfor flow cytometry.

Flow cytometry. Mononuclear cells (106 cells) were resuspended in fluores-cence-activated cell sorter buffer (0.2% bovine serum albumin [Sigma] and0.09% NaN3 [Sigma] in phosphate-buffered saline [PBS] [Invitrogen]) and incu-bated with anti-Fc�R antibody (clone 2.4G2) (BD Pharmingen, San Diego, CA)before appropriate amounts of conjugated antibodies or isotype controls wereadded and incubated for 30 min at 4°C (see below). Cells were washed twicewith fluorescence-activated cell sorter buffer and centrifuged for 5 min at400 � g at 4°C before being fixed in 500 �l 1% formalin in PBS. At least10,000 viable cells were acquired on the basis of forward light and side lightscattering and then quantified by using a BD FACSCalibur instrument andanalyzed with WinList software (Verity Software House, Topsham, ME).Two-tailed P values were calculated using an unpaired t test with InStatsoftware (GraphPad, San Diego, CA).

Antibodies for flow cytometry and immunofluorescence. The following anti-bodies were used for flow cytometry and confocal microscopy: fluorescein iso-thiocyanate (FITC) anti-mouse CD45R/B220 (clone RA3-6B2), FITC anti-mouse CD11c (clone HL3), FITC anti-mouse CD49b/Pan natural killer (NK)cells (clone DX5), R-phycoerythrin (PE) anti-mouse CD3 (clone 17A2), PEanti-mouse CD45R/B220 (clone RA3-6B2), PE anti-mouse CD11c (clone HL3),PE anti-mouse I-A/I-E (major histocompatibility complex class II [MHC-II])(clone M5/114.15.2), PE anti-mouse Ly-6G and Ly-6C (Gr-1) (clone RB6-8C5),peridinin chlorophyll a protein (PerCP) anti-mouse CD4 (clone RM4-5), PerCP-Cy5.5 anti-mouse Mac-1 (CD11b) (clone M1/70), allophycocyanin (APC) anti-mouse NK1.1 (clone PK136), and APC anti-mouse CD8 (clone 53-6.7) from BD

Pharmingen; Alexa Fluor 488 anti-mouse CD4 (clone GK1.5), Alexa Fluor 647anti-mouse CD8a (clone 53-6.7), and Alexa Fluor 647 anti-mouse Mac-1(CD11b) (clone M1/70) from Biolegend (San Diego, CA); and Alexa Fluor 488anti-mouse F4/80 (clone CI:A3-1) from Serotec (Raleigh, NC). Isotype-matchedantibodies (all from BD Pharmingen) were used as controls for nonspecificbinding. Polyclonal rabbit anti-F. tularensis LVS was harvested after four injec-tions of heat-killed organisms. FITC anti-rabbit immunoglobulin G (IgG) fromChemicon Int. (Temecula, CA) or Alexa Fluor 555 anti-rabbit IgG from Molec-ular Probes (Eugene, OR) was used as a secondary antibody to F. tularensisantisera.

Hematoxylin and eosin staining and immunohistology on tissue sections.Livers were aseptically removed and immediately fixed in 10% neutral bufferedformalin, embedded in Blue Ribbon paraffin (Surgipath, Richmond, IL), sec-tioned at 5 �m, stained with hematoxylin and eosin, dehydrated in gradedalcohols, cleared with xylene, and mounted with Acrymount (Statlab MedicalProducts, Lewisville, TX). Tissue sections were examined by light microscopy.

Detection of caspase-3 was achieved by dewaxing and rehydration of paraffinsections with xylene and graded alcohols, followed by quenching of endogenousperoxidase with methanol and hydrogen peroxide and blocking with Tween-bovine serum albumin. Rabbit anti-cleaved caspase-3 (Asp175) from Cell Sig-naling Technology (Danvers, MA) was diluted in blocking solution and added tosections for overnight incubation at RT. Sections were then washed and treatedwith polyclonal biotinylated anti-goat IgG (Vector Laboratories, Burlingame,CA) for 1 h at RT. Sections were washed, and avidin-biotinylated enzymecomplex reagent (Vector Laboratories) was added for 45 min at RT, followed byfive washes and incubation with diaminobenzidine (Sigma-Aldrich Corporation,St. Louis, MO) for 10 min. Sections were rinsed in water, counterstained withhematoxylin, dehydrated in graded alcohols, and cleared with xylene.

Terminal deoxynucleotidyl transferase-mediated dUTP-X nick end labeling

FIG. 1. Mean serum levels � standard deviations of (A) ALT and(B) LDH in mice infected with sublethal doses of LVS. The dashedline represents the upper limit of the normal range for each enzyme.Each diamond represents the mean value � standard deviation forthree mice.

TABLE 1. Total and differential peripheral blood leukocyte countsfrom mice inoculated with F. tularensis LVS

DPI Mean total cell counts (106 cells/ml) � SD(lymphocyte:neutrophil:monocyte ratio)a

0 ................................................................. 4.0 � 0.5 (89:6:5)1 ................................................................. 8.6 � 2.8 (49:48:3)2 ................................................................. 5.25 � 2.2 (55:41:4)3 ................................................................. 7.8 � 4.3 (67:30:3)4 ................................................................. 6.3 � 1.1 (82:9:9)5 ................................................................. 7.5 � 2.0 (85:12:3)

a Three mice were used at each time point.

VOL. 74, 2006 Mac-1� CELLS IN LIVER TULAREMIA 6591

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

(TUNEL) assays were performed according to the manufacturer’s protocol usingan in situ cell death detection kit with tetramethylrhodamine red (Roche AppliedScience, Indianapolis, IN).

For detection of bacteria in the liver, paraffin sections were treated for 30 minwith rabbit anti-F. tularensis LVS IgG after dewaxing and rehydration of thesections. Secondary alkaline phosphatase-labeled goat anti-rabbit IgG fromZymed (San Francisco, CA) was added for 30 min at RT, and Vulcan Fast redchromogen (Biocarta, San Diego, CA) was then used to visualize the bacteria.

Immunofluorescent staining of frozen tissue sections. Tissues removed frommice were immediately placed into freshly made 1% formalin in PBS fromInvitrogen and gently shaken for 1 h at 4°C. The tissues were removed, blotteddry, placed into freshly made 30% sucrose in PBS at 4°C, and left overnight. Thetissues were removed, blotted dry, placed into Neg �50 freezing compound(Richard-Allan Scientific, Kalamazoo, MI), rapidly frozen in isopentane that hadbeen cooled with liquid nitrogen, and stored at �80°C. For some experiments,organs were immersed in OCT compound (Sakura Finetek, Torrance, CA) andthen frozen and stored as described above.

Frozen tissue sections were cut at 5 �m in the cryostat at �25°C, air dried, andfixed in acetone for 30 s. Twenty microliters of the various antibodies (see above)diluted in 0.01 M PBS (pH 7.4) was applied to sections and incubated in the darkfor 25 min. Slides were washed three times in PBS, and when appropriate,secondary antibodies were added for 25 min in the dark. Mouse spleens, treatedin the same manner, were used as positive controls for the antibodies used in thisstudy. After washing, slides were mounted in Opti-Mount (Richard-Allan Sci-entific, Kalamazoo, MI). The slides were examined by phase-contrast and epi-

fluorescence microscopy using a Nikon Eclipse E600 microscope, and imageswere captured using a Spot camera (Diagnostic Instruments, Inc.). Slides forconfocal microscopy were analyzed using a Leica DM IRE2 confocal micro-scope. Images of the red, green, and blue emission signals were captured sepa-rately with the Leica LCS software package. Images were processed using AdobePhotoshop.

RESULTS AND DISCUSSION

The pathology of liver involvement in experimental tulare-mia has been studied with standard histopathological proce-dures, and the existence of granulomatous necrotic lesions hasbeen noted previously in several studies (13, 15). To date,however, the infiltrating cells of the hepatic lesions have notbeen characterized with respect to specific markers, and theextent of apoptosis has not been examined with markers spe-cific to this type of cell death.

Intradermal inoculation of C3H/HeN mice with F. tularensisLVS led to bacteremia for the first 5 DPI. In the periphery, thisbacteremia was accompanied by leukocytosis with an initialreversal in the ratio of lymphocytes to neutrophils in the dif-

FIG. 2. Hematoxylin- and eosin-stained sections of livers of mice infected with sublethal doses of LVS. (A) Normal liver. (B) Low-power viewof hepatic parenchyma showing multiple lesions (arrows) of inflammatory cell infiltrates in the liver of a mouse inoculated with LVS 5 days earlier.(C) Mononuclear cell inflammatory cell infiltrate of a perivascular focus of inflammation in the liver of a mouse infected with LVS 5 days earlier.Some neutrophils are evident, as indicated by arrows. (D) A 5-day-old lesion within the hepatic parenchyma showing necrotic hepatocytes, pyknoticnuclei, and residual mononuclear cell infiltrate. Bars, 150 �m (A and B) and 75 �m (C and D).

6592 RASMUSSEN ET AL. INFECT. IMMUN.

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

ferential as well as a modest increase in the percentage ofcirculating monocytes at 4 DPI (Table 1). A similar pattern ofleukocytosis and reversal of the differential has been demon-strated for experimental infections of other strains of mice(13). There were marked increases in serum levels of ALT(Fig. 1A) and LDH (Fig. 1B). This pattern is consistent withearly inflammation in the liver without reducing the ability ofthe liver to conjugate and secrete bilirubin, as evidenced by thenormal values obtained for direct and indirect bilirubin andalkaline phosphatase (data not shown). Kidney function waswithin normal limits.

Here, we confirm the previous findings of widespread, earlyfoci characterized by the infiltration of a large number ofmononuclear cells that are morphologically consistent withmacrophages and a few neutrophils in the liver (Fig. 2B) (12,13). These lesions had a focus of mononuclear infiltration. Asthese lesions matured, necrotic hepatocytes with pyknotic nu-clei were common within the inflammatory foci (Fig. 2C andD). The evolution of the granulomatous response is typical,where neutrophils with a short half-life appear early and

mononuclear cells persist. Neutrophils are known to be im-portant for defense in primary tularemia infection (14, 40).The perivascular location of many of the granulomas (Fig.2C) suggests that the infiltrate derives from circulating cellsfrom the blood as opposed to an expansion of the residentcells.

Bacteria invade the liver parenchyma early in randomly dis-tributed locations (Fig. 3B). Some bacteria appear to be asso-ciated with Kupffer cells based on the location of these cells onthe sinusoids, but others are within hepatocytes (Fig. 3B andC). Figure 3C shows a hepatocyte swollen with bacteria, similarto what has been observed previously by others (15). Theseheavily infected hepatocytes could become focal points for thedevelopment of the granulomas. In the granuloma at 5 DPI, itis difficult to determine whether the bacteria are extracellularor associated with hepatocytes, macrophages, or both (Fig.3D). Nonetheless, there is severe damage to hepatocytes, andnumerous bacterial colonies are present in the lesions. Ourresults indicated that hepatic dysfunction in tularemia is likelyto be a contributor to the morbidity and mortality of this

FIG. 3. Detection of bacteria in livers of mice inoculated with sublethal doses of LVS 5 days earlier. Bacteria (red) were detected with rabbitanti-F. tularensis LVS followed by alkaline phosphatase-conjugated anti-rabbit serum. Vulcan Fast red was used to detect alkaline phosphatase.(A) Normal liver. (B) Low-power view of hepatic parenchyma showing multiple sites of bacterial infection. (C) Hepatocytes infected with F.tularensis. Some of the hepatocytes are heavily infected and swollen with bacteria. (D) Bacteria concentrated within the granuloma. Bars, 150 �m(A and B) and 50 �m (C and D).

VOL. 74, 2006 Mac-1� CELLS IN LIVER TULAREMIA 6593

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

infection, although in some instances, liver disease can bereversible.

To characterize the cells in the lesions observed in the liversof infected mice, immunofluorescence microscopy of frozenliver sections was performed using specific cell surface markersfor macrophages, lymphocytes, and F. tularensis antibodies.Infection of the liver parenchyma was already present on thefirst day after inoculation (Fig. 4). In later stages, the vastmajority of the mononuclear cells within the granulomas wereF4/80� Mac-1� (Fig. 5). A few F4/80� Mac-1� cells werefound in the borders of the lesion; these cells may represent apopulation of Kupffer cells (28). The F4/80� Mac-1� cells mayrepresent monocytes/macrophages recruited from the blood tothe liver. Cells with this phenotype have been shown to trafficfrom the peripheral blood to the inflamed retina in a murinemodel of autoimmune uveoretinitis (45). Blood monocytes canexpress both F4/80 and Mac-1 markers (23), so it is possiblethat the phenotype of the infiltrating mononuclear cells may bederived from blood, with a subsequent downregulation of theF4/80 marker. Another possibility is that F4/80� Mac-1� cellsare a subset of resident macrophages similar to those in thespleen, optic nerve, and the connective tissue of the lung inwhich the expression of the F4/80 antigen is downregulated byinflammatory stimuli (6, 8, 19, 44). Regardless of the possibleorigin of Mac-1� F4/80� cells, F. tularensis was found to beassociated with these cells, correlating with a Mac-1� pheno-type that has been shown to be involved in the hepatic killingof other bacteria (7).

Mac-1 can be expressed on a variety of cells, including gran-ulocytes, T cells, B cells, NK cells, dendritic cells (DCs), andmonocytes. To further characterize the Mac-1� infiltrating

FIG. 4. Immunofluorescent detection of F. tularensis and inflam-matory cells in the livers of mice inoculated with sublethal doses 1 dayearlier viewed by confocal microscopy. (A) Merged-image, low-powerview of Mac-1� cells (Alexa Fluor 647, blue), F. tularensis (secondaryanti-rabbit Ig, Alexa Fluor 555, red), and F4/80 (Alexa Fluor 488,green). (B) Hepatocytes infected with F. tularensis (red). (C) Mergedimage of a Mac-1� cell (neutrophil-like, blue) infected with F. tula-rensis (red).

FIG. 5. Immunofluorescent detection of F. tularensis and inflammatory cells in the livers of mice inoculated with sublethal doses 5 days earlierviewed by confocal microscopy. (A) F4/80� cells (Alexa Fluor 488, green). (B) F. tularensis (secondary anti-rabbit Ig, Alexa Fluor 555, red) in fociof inflammatory cells. (C) Mac-1� cells (Alexa Fluor 647, blue). (D) Merged image of panels A to C. Note the colocalization of Mac-1� cells (blue)and F. tularensis (red). Also note the peripheral location of the F4/80� cells around the foci of infection. Some cells are both F4/80� and Mac-1�

and can be seen in merged images of panels E and F.

6594 RASMUSSEN ET AL. INFECT. IMMUN.

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

mononuclear cells, immunofluorescence microscopy was per-formed using specific markers for cell types known to expressMac-1. Neutrophils were ruled out as a major cell type con-tributing to the infiltrate of the granulomas by morphology(Fig. 2C and D). CD3�, CD4�, and CD8� T cells and B220�

B cells were not detected in the granulomatous areas of hepa-tocyte necrosis at 1 and 5 DPI (data not shown). However, theMac-1� cells did colocalize with markers specific for myeloidcell populations that were most consistent with macrophagesand DCs. One population expressed both macrophage (Mac-1�) and granulocyte (Gr-1�) markers and was the predomi-nant phenotype in the granuloma (Fig. 6A). The lesions alsocontained a significant population that had an MHC-II�

Mac-1� phenotype (Fig. 6B) and a CD11c� Mac-1� DC phe-notype (Fig. 6C). NK cells (NK1.1� and CD49/DX5�) wereseen in the liver tissue at 5 DPI but were not associated withthe hepatic lesions (Fig. 6D).

Coexpression of both Gr-1 and Mac-1 is indicative of animmature myeloid cell that can differentiate into either a ma-ture granulocyte, macrophage, or DC (for a review, see refer-ence 38). Immature myeloid cells (Gr-1� Mac-1�) do notexpress MHC-II molecules and exhibit dull F4/80 expression

(29, 34), correlating with the predominant phenotype (F4/80�

Mac-1�) seen at 5 DPI (Fig. 5). Also known as myeloid sup-pressor cells, these cells accumulate and inhibit the T-cell im-mune response in tumor-bearing mice (22, 36, 41). In addition,they have been found to have immunosuppressive effects inmice infected with various pathogens (1, 26, 35). However, ithas been noted that depending on the cytokine milieu that ispresent, Gr-1� Mac-1� cells can either activate or inactivatethe T-lymphocyte immune response. Bronte et al. (9) previ-ously showed that when cultured in vitro with proinflammatorycytokines (gamma interferon and tumor necrosis factor alpha),Gr-1� Mac-1� cells differentiated into functional antigen-pre-senting cells. However, when these cells were cultured with ananti-inflammatory cytokine (interleukin-4), the cells greatly in-creased T-cell suppression. Therefore, the function of imma-ture myeloid cells is dependent upon the host inflammatoryresponse initiated by a pathological process.

To further quantify the abundance of cellular populations,flow cytometry analysis was performed on liver tissue fromuninfected mice and mice infected with F. tularensis LVS at 5DPI. Markers were used for T cells (CD3), B cells (B220), DCs(CD11c and Mac-1), NK cells (DX5 and NK1.1), macrophages(F4/80, MHC-II, and Mac-1), and immature myeloid cells(Gr-1 and Mac-1) (Table 2). Results of the quantification ofmononuclear cells in the liver by flow cytometry were consis-tent with the imaging results. Markers for B and T cells weresimilarly expressed in both uninfected mice and mice inocu-lated with F. tularensis LVS 5 days earlier. A twofold increaseof NK cell marker expression by 5 DPI was noted. This mayindicate that NK cells are upregulated to aid the innate re-sponse and cytokine secretion, even though NK cells were notseen in the granulomas. Total CD11c expression, which isindicative of DCs, increased 2.5-fold by 5 DPI. Furthermore,the numbers of myeloid DCs (CD11c� Mac-1�), which werefound in the lesions, also increased but did not reach statisticalsignificance. The most significant increase of all cellular phe-notypes was Mac-1� cells (2.3% uninfected to 21.3% at 5 DPI),which correlates to the majority of cells seen in the hepaticlesions. The bulk of Mac-1� cells were CD11c�, indicating that

TABLE 2. Flow cytometry analysis of cell marker expression oftotal cell counts from livers of mice inoculated with F. tularensis

LVS 5 days earlier compared to that from uninfected livers

Cell marker

Mean % � SDa

Uninfected totalcells

Total cells at5 DPI

CD3� 7.6 � 2.9 6.1 � 1.4B220� 3.4 � 1.3 3.1 � 1.7CD11c� 1.1 � 0.6 2.8 � 1.8CD11c� Mac-1� 0.3 � 0.3 2.1 � 1.6DX5� NK1.1� 2.3 � 0.6 4.8 � 0.9b

Mac-1� 2.3 � 0.3 21.3 � 3.1c

F4/80� Mac-1� 1.8 � 1.3 2.8 � 0.8MHC-II� Mac-1� CD11c� 1.2 � 0.2 10.4 � 5.2d

Gr-1� Mac-1� 1.5 � 0.3 13.1 � 5.3e

a Results are the means � standard deviations for three mice.b P � 0.015.c P � 0.0005.d P � 0.039.e P � 0.020.

FIG. 6. Immunofluorescent detection of F. tularensis and inflam-matory cells in the livers of mice inoculated with sublethal doses 5 daysearlier viewed by confocal microscopy. (A) Merged image of Mac-1�

cells (Alexa Fluor 647, blue), Gr-1� cells (PE, red), and F. tularensis(secondary anti-rabbit Ig, FITC, green). Gr-1� cells colocalize withMac-1� cells (pink) and make up the granuloma. (B) Merged image ofMac-1� cells (Alexa Fluor 647, blue), MHC-II� cells (PE, red) and F.tularensis (secondary anti-rabbit Ig, FITC, green). The granuloma alsoconsists of a cell population, MHC II� Mac-1�, that also associateswith the bacteria. (C) Merged image of Mac-1� cells (Alexa Fluor 647,blue), CD11c� cells (PE, red), and F. tularensis (secondary anti-rabbitIg, FITC, green). DCs (CD11c� Mac-1�) were found in associationwith the granuloma. (D) Merged image of NK1.1� cells (APC, blue),F. tularensis (secondary anti-rabbit Ig, Alexa Fluor 555, red), andDX5� (FITC, green). NK cells were seen in the tissue but not inassociation with the bacteria or the hepatic lesions.

VOL. 74, 2006 Mac-1� CELLS IN LIVER TULAREMIA 6595

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

most Mac-1� cells were not DCs. In addition, levels of F4/80�

Mac-1� cells did not increase significantly, correlating with thelow levels of this subpopulation shown in Fig. 5.

The largest increases in Mac-1� subpopulations observed byflow cytometry analysis were the Gr-1� Mac-1� immature my-eloid cells and the MHC-II� Mac-1� CD11c� macrophages,confirming the results from immunofluorescence staining ofliver tissue. Representative plots from flow cytometry areshown in Fig. 7A and B. Correlating to their abundance in thegranulomas, Gr-1� Mac-1� and MHC-II� Mac-1� CD11c�

cells are likely major contributors in controlling early F. tula-rensis LVS infection. The Gr-1� Mac-1� cells could function asimmunosuppressive cells to inhibit the immune response andallow for bacterial survival or as a means to wall off the infec-tion until an inflammatory response develops. It is tempting tospeculate that under a proinflammatory response, Gr-1�

Mac-1� cells differentiate into functional antigen-presentingcells largely as MHC-II� Mac-1� CD11c� macrophages and toa lesser extent as myeloid DCs (CD11c� Mac-1�). These cellphenotypes would correlate with those seen within the hepaticlesions (Fig. 6).

We have observed and confirmed a necrotic process (12, 13)that is clearly evident within the liver abscesses at 5 DPI (Fig.2). Apoptosis of hepatocytes is the hallmark of murine listeri-osis, which is caused by another intracellular organism with apredilection for liver involvement (37). In addition, a numberof in vitro studies with murine macrophages have shown that F.tularensis infection is able to trigger the apoptotic cascade (27,30–32). Based on those studies, we examined the extent ofapoptosis in the livers of infected mice. Although apoptoticcells were detected in the livers of mice infected with F. tula-

FIG. 7. Flow cytometry analysis of predominant cell populations inthe livers of F. tularensis LVS-infected mice. Accumulation of Gr-1�

Mac-1� immature myeloid cells (A) and MHC-II� Mac-1� cell ex-pression (B) in a mouse 5 DPI compared to uninfected control areshown. A representative experiment is shown for both A and B.

FIG. 8. Detection of apoptotic cell markers by immunohistochemistry in the livers of mice inoculated with sublethal doses of F. tularensis 5 daysearlier. (A) Low-power view of anti-cleaved caspase-3 reactivity (brown) within the foci of inflammation and infection. (B and C) High-power viewof anti-cleaved caspase-3 reactivity (brown) within the foci of inflammation and infection. (D) Low-power phase-contrast view of liver parenchymacontaining foci of inflammation and infection. (E) TUNEL-positive cells (red) (same field as panel D). (F) Merged photograph of panels D andE showing cells with fragmented DNA. Bars, 150 �m (A) and 75 �m (B to G).

6596 RASMUSSEN ET AL. INFECT. IMMUN.

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

rensis using both caspase-3 and TUNEL markers (Fig. 8), theselevels of apoptosis were qualitatively less than those observedin a sublethal Listeria infection (37, 46). Both the necrotic andapoptotic pathways of cell death appear to be important intularemia, as is also true for some other bacterial infections ofthe liver. For example, Listeria induces the apoptosis of hepa-tocytes (37), but it is also known that murine macrophagessuccumb to Listeria infection in vitro by necrosis (4). Further-more, apoptotic death of CD8� T cells has been attributed toa function of Gr-1� Mac-1� cells (10). Infection of the liver bymicroorganisms results in the death of hepatocytes, and my-eloid cells could be major contributors to this process.

F. tularensis causes liver damage, as evidenced by elevatedserum ALT and LDH levels and by the granulomatous-necrotic-apoptotic lesions that appear by 5 DPI. These lesions arecomposed mostly of Mac-1� cells from two myeloid popula-tions (Gr-1� Mac-1� and MHC-II� Mac-1�) that are associ-ated with the bacteria, thus suggesting that these cells areimportant for controlling the infection. These cells appear tobe recruited cells and are accumulated specifically to regulatethe infection in its early stage and may therefore prove usefulin the development of vaccines against F. tularensis infectionsto stimulate the activation of these particular cell subsets.

ACKNOWLEDGMENTS

This work was supported by a grant from the National Institutes ofHealth, AI 055621.

We appreciate the assistance of Gloria Monsalve and Patricio Mena.

REFERENCES

1. al-Ramadi, B. K., M. A. Brodkin, D. M. Mosser, and T. K. Eisenstein. 1991.Immunosuppression induced by attenuated Salmonella. Evidence for medi-ation by macrophage precursors. J. Immunol. 146:2737–2746.

2. Anda, P., J. Segura del Pozo, J. M. Diaz Garcıa, R. Escudero, F. J. GarcıPena, M. C. Lopez Velasco, R. E. Sellek, M. R. Jimenez Chillaron, L. P.Sanchez Serrano, and J. F. Martınez Navarro. 2001. Waterborne outbreakof tularemia associated with crayfish fishing. Emerg. Infect. Dis. 7:575–582.

3. Anthony, L. S., and P. A. Kongshavn. 1987. Experimental murine tularemiacaused by Francisella tularensis, live vaccine strain: a model of acquiredcellular resistance. Microb. Pathog. 2:3–14.

4. Barsig, J., and S. H. Kaufmann. 1997. The mechanism of cell death inListeria monocytogenes-infected murine macrophages is distinct from apop-tosis. Infect. Immun. 65:4075–4081.

5. Baskerville, A., and P. Hambleton. 1976. Pathogenesis and pathology ofrespiratory tularaemia in the rabbit. Br. J. Exp. Pathol. 57:339–347.

6. Breel, M., M. Van der Ende, T. Sminia, and G. Kraal. 1988. Subpopulationsof lymphoid and non-lymphoid cells in bronchus-associated lymphoid tissue(BALT) of the mouse. Immunology 63:657–662.

7. Brengman, M. L., D. Wang, K. B. Wilkins, N. Sakamoto, T. Arai, E. P.Ceppa, A. S. Klein, and G. B. Bulkley. 2003. Hepatic killing but not clearanceof systemically circulating bacteria is dependent upon peripheral leukocytesvia Mac-1 (CD11b/CD18). Shock 19:263–267.

8. Brissette-Storkus, C. S., S. M. Reynolds, A. J. Lepisto, and R. L. Hendricks.2002. Identification of a novel macrophage population in the normal mousecorneal stroma. Investig. Ophthalmol. Vis. Sci. 43:2264–2271.

9. Bronte, V., E. Apolloni, A. Cabrelle, R. Ronca, P. Serafini, P. Zamboni, N. P.Restifo, and P. Zanovello. 2000. Identification of a CD11b(�)/Gr-1(�)/CD31(�) myeloid progenitor capable of activating or suppressing CD8(�) Tcells. Blood 96:3838–3846.

10. Bronte, V., M. Wang, W. W. Overwijk, D. R. Surman, F. Pericle, S. A.Rosenberg, and N. P. Restifo. 1998. Apoptotic death of CD8� T lympho-cytes after immunization: induction of a suppressive population of Mac-1�/Gr-1� cells. J. Immunol. 161:5313–5320.

11. Chen, W., R. KuoLee, H. Shen, and J. W. Conlan. 2004. Susceptibility ofimmunodeficient mice to aerosol and systemic infection with virulent strainsof Francisella tularensis. Microb. Pathog. 36:311–318.

12. Chen, W., H. Shen, A. Webb, R. KuoLee, and J. W. Conlan. 2003. Tularemiain BALB/c and C57BL/6 mice vaccinated with Francisella tularensis LVS andchallenged intradermally, or by aerosol with virulent isolates of the patho-gen: protection varies depending on pathogen virulence, route of exposure,and host genetic background. Vaccine 21:3690–3700.

13. Conlan, J. W., W. Chen, H. Shen, A. Webb, and R. KuoLee. 2003. Experi-mental tularemia in mice challenged by aerosol or intradermally with viru-lent strains of Francisella tularensis: bacteriologic and histopathologic stud-ies. Microb. Pathog. 34:239–248.

14. Conlan, J. W., R. KuoLee, H. Shen, and A. Webb. 2002. Different hostdefences are required to protect mice from primary systemic vs pulmonaryinfection with the facultative intracellular bacterial pathogen, Francisellatularensis LVS. Microb. Pathog. 32:127–134.

15. Conlan, J. W., and R. J. North. 1992. Early pathogenesis of infection in theliver with the facultative intracellular bacteria Listeria monocytogenes, Fran-cisella tularensis, and Salmonella typhimurium involves lysis of infected hepa-tocytes by leukocytes. Infect. Immun. 60:5164–5171.

16. Dennis, D. T., T. V. Inglesby, D. A. Henderson, J. G. Bartlett, M. S. Ascher,E. Eitzen, A. D. Fine, A. M. Friedlander, J. Hauer, M. Layton, S. R. Lil-libridge, J. E. McDade, M. T. Osterholm, T. O’Toole, G. Parker, T. M. Perl,P. K. Russell, and K. Tonat. 2001. Tularemia as a biological weapon: medicaland public health management. JAMA 285:2763–2773.

17. Ellis, J., P. C. Oyston, M. Green, and R. W. Titball. 2002. Tularemia. Clin.Microbiol. Rev. 15:631–646.

18. Evans, M. E., D. W. Gregory, W. Schaffner, and Z. A. McGee. 1985. Tula-remia: a 30-year experience with 88 cases. Medicine (Baltimore) 64:251–269.

19. Ezekowitz, R. A., J. Austyn, P. D. Stahl, and S. Gordon. 1981. Surfaceproperties of bacillus Calmette-Guerin-activated mouse macrophages. Re-duced expression of mannose-specific endocytosis, Fc receptors, and antigenF4/80 accompanies induction of Ia. J. Exp. Med. 154:60–76.

20. Forestal, C. A., J. L. Benach, C. Carbonara, J. K. Italo, T. J. Lisinski, andM. B. Furie. 2003. Francisella tularensis selectively induces proinflammatorychanges in endothelial cells. J. Immunol. 171:2563–2570.

21. Fortier, A. H., M. V. Slayter, R. Ziemba, M. S. Meltzer, and C. A. Nacy. 1991.Live vaccine strain of Francisella tularensis: infection and immunity in mice.Infect. Immun. 59:2922–2928.

22. Gabrilovich, D. I., M. P. Velders, E. M. Sotomayor, and W. M. Kast. 2001.Mechanism of immune dysfunction in cancer mediated by immature Gr-1�myeloid cells. J. Immunol. 166:5398–5406.

23. Geissmann, F., S. Jung, and D. R. Littman. 2003. Blood monocytes consistof two principal subsets with distinct migratory properties. Immunity 19:71–82.

24. Gil, H., J. L. Benach, and D. G. Thanassi. 2004. Presence of pili on thesurface of Francisella tularensis. Infect. Immun. 72:3042–3047.

25. Goethert, H. K., I. Shani, and S. R. Telford III. 2004. Genotypic diversity ofFrancisella tularensis infecting Dermacentor variabilis ticks on Martha’s Vine-yard, Massachusetts. J. Clin. Microbiol. 42:4968–4973.

26. Goni, O., P. Alcaide, and M. Fresno. 2002. Immunosuppression during acuteTrypanosoma cruzi infection: involvement of Ly6G (Gr1�)CD11b� imma-ture myeloid suppressor cells. Int. Immunol. 14:1125–1134.

27. Hrstka, R., J. Stulık, and B. Vojtesek. 2005. The role of MAPK signalpathways during Francisella tularensis LVS infection-induced apoptosis inmurine macrophages. Microbes Infect. 7:619–625.

28. Hume, D. A., I. L. Ross, S. R. Himes, R. T. Sasmono, C. A. Wells, and T.Ravasi. 2002. The mononuclear phagocyte system revisited. J. Leukoc. Biol.72:621–627.

29. Inaba, K., M. Inaba, M. Deguchi, K. Hagi, R. Yasumizu, S. Ikehara, S.Muramatsu, and R. M. Steinman. 1993. Granulocytes, macrophages, anddendritic cells arise from a common major histocompatibility complex classII-negative progenitor in mouse bone marrow. Proc. Natl. Acad. Sci. USA90:3038–3042.

30. Lai, X. H., I. Golovliov, and A. Sjostedt. 2004. Expression of IglC is necessaryfor intracellular growth and induction of apoptosis in murine macrophagesby Francisella tularensis. Microb. Pathog. 37:225–230.

31. Lai, X. H., I. Golovliov, and A. Sjostedt. 2001. Francisella tularensis inducescytopathogenicity and apoptosis in murine macrophages via a mechanismthat requires intracellular bacterial multiplication. Infect. Immun. 69:4691–4694.

32. Lai, X. H., and A. Sjostedt. 2003. Delineation of the molecular mechanismsof Francisella tularensis-induced apoptosis in murine macrophages. Infect.Immun. 71:4642–4646.

33. Lamps, L. W., J. M. Havens, A. Sjostedt, D. L. Page, and M. A. Scott. 2004.Histologic and molecular diagnosis of tularemia: a potential bioterrorismagent endemic to North America. Mod. Pathol. 17:489–495.

34. Marshall, M. A., D. Jankovic, V. E. Maher, A. Sher, and J. A. Berzofsky.2001. Mice infected with Schistosoma mansoni develop a novel non-T-lym-phocyte suppressor population which inhibits virus-specific CTL inductionvia a soluble factor. Microbes Infect. 3:1051–1061.

35. Mencacci, A., C. Montagnoli, A. Bacci, E. Cenci, L. Pitzurra, A. Spreca, M.Kopf, A. H. Sharpe, and L. Romani. 2002. CD80�Gr-1� myeloid cellsinhibit development of antifungal Th1 immunity in mice with candidiasis.J. Immunol. 169:3180–3190.

36. Otsuji, M., Y. Kimura, T. Aoe, Y. Okamoto, and T. Saito. 1996. Oxidativestress by tumor-derived macrophages suppresses the expression of CD3 zetachain of T-cell receptor complex and antigen-specific T-cell responses. Proc.Natl. Acad. Sci. USA 93:13119–13124.

37. Rogers, H. W., M. P. Callery, B. Deck, and E. R. Unanue. 1996. Listeria

VOL. 74, 2006 Mac-1� CELLS IN LIVER TULAREMIA 6597

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

monocytogenes induces apoptosis of infected hepatocytes. J. Immunol. 156:679–684.

38. Serafini, P., C. De Santo, I. Marigo, S. Cingarlini, L. Dolcetti, G. Gallina, P.Zanovello, and V. Bronte. 2004. Derangement of immune responses bymyeloid suppressor cells. Cancer Immunol. Immunother. 53:64–72.

39. Shapiro, D. S., and D. R. Schwartz. 2002. Exposure of laboratory workers toFrancisella tularensis despite a bioterrorism procedure. J. Clin. Microbiol.40:2278–2281.

40. Sjostedt, A., J. W. Conlan, and R. J. North. 1994. Neutrophils are critical forhost defense against primary infection with the facultative intracellular bac-terium Francisella tularensis in mice and participate in defense against rein-fection. Infect. Immun. 62:2779–2783.

41. Song, X., Y. Krelin, T. Dvorkin, O. Bjorkdahl, S. Segal, C. A. Dinarello, E.Voronov, and R. N. Apte. 2005. CD11b�/Gr-1� immature myeloid cellsmediate suppression of T cells in mice bearing tumors of IL-1beta-secretingcells. J. Immunol. 175:8200–8208.

42. Sutinen, S., and H. Syrjala. 1986. Histopathology of human lymph node

tularemia caused by Francisella tularensis var palaearctica. Arch. Pathol. Lab.Med. 110:42–46.

43. White, J. D., J. R. Rooney, P. A. Prickett, E. B. Derrenbacher, C. W. Beard,and W. R. Griffith. 1964. Pathogenesis of experimental respiratory tularemiain monkeys. J. Infect. Dis. 114:277–283.

44. Witmer, M. D., and R. M. Steinman. 1984. The anatomy of peripherallymphoid organs with emphasis on accessory cells: light-microscopic immu-nocytochemical studies of mouse spleen, lymph node, and Peyer’s patch.Am. J. Anat. 170:465–481.

45. Xu, H., A. Manivannan, R. Dawson, I. J. Crane, M. Mack, P. Sharp, and J.Liversidge. 2005. Differentiation to the CCR2� inflammatory phenotype invivo is a constitutive, time-limited property of blood monocytes and is inde-pendent of local inflammatory mediators. J. Immunol. 175:6915–6923.

46. Zheng, S. J., P. Wang, G. Tsabary, and Y. H. Chen. 2004. Critical roles ofTRAIL in hepatic cell death and hepatic inflammation. J. Clin. Investig.113:58–64.

Editor: J. T. Barbieri

6598 RASMUSSEN ET AL. INFECT. IMMUN.

on March 8, 2021 by guest

http://iai.asm.org/

Dow

nloaded from