Interleukin-6 Is Essential for Primary Resistance to Francisella tularensis Live Vaccine Strain Infection

Abstract

We employed Francisella tularensis live vaccine strain (LVS) to study mechanisms of protective immunity against intracellular pathogens and, specifically, to understand protective correlates. One potential molecular correlate identified previously was interleukin-6 (IL-6), a cytokine with pleotropic roles in immunity, including influences on T and B cell functions. Given its role as an immune modulator and the correlation with successful anti-LVS vaccination, we examined the role IL-6 plays in the host response to LVS. IL-6-deficient (IL-6 knockout [KO]) mice infected with LVS intradermally or intranasally or anti-IL-6-treated mice, showed greatly reduced 50% lethal doses compared to wild-type (WT) mice. Increased susceptibility was not due to altered splenic immune cell populations during infection or decreased serum antibody production, as IL-6 KO mice had similar compositions of each compared to WT mice. Although LVS-infected IL-6 KO mice produced much less serum amyloid A and haptoglobin (two acute-phase proteins) than WT mice, there were no other obvious pathophysiological differences between LVS-infected WT and IL-6 KO mice. IL-6 KO or WT mice that survived primary LVS infection also survived a high-dose LVS secondary challenge. Using an in vitro overlay assay that measured T cell activation, cytokine production, and abilities of primed splenocytes to control intracellular LVS growth, we found that IL-6 KO total splenocytes or purified T cells were slightly defective in controlling intracellular LVS growth but were equivalent in cytokine production. Taken together, IL-6 is an integral part of a successful immune response to primary LVS infection, but its exact role in precipitating adaptive immunity remains elusive.

INTRODUCTION

Francisella tularensis is a facultative, intracellular bacterium that can cause life-threatening illness in humans. F. tularensis subsp. tularensis, which is present only in North America, is also known as type A Francisella, and it is highly virulent in people (1). F. tularensis subsp. holarctica, which is the only biotype found in Europe and Asia, is designated type B Francisella, and infections by this strain are rarely lethal in humans. A live attenuated vaccine for F. tularensis was developed in the 1940s from type B Francisella (2–4). One such vaccine, designated the live vaccine strain (LVS), has been tested in human experimental studies since the 1950s, but it is not currently licensed for use in the United States (3, 5, 6).

LVS infection of mice is a useful model system for understanding primary and adaptive immunity to Francisella, as well as other intracellular pathogens. LVS given to C57BL/6J mice intraperitoneally or intravenously is lethal at low doses, with a 50% lethal dose (LD₅₀) of <10 CFU (7). C57BL/6J mice are more resistant to intranasal (i.n.) LVS infection, with an LD₅₀ of ∼10³ CFU, and have a high resistance to subcutaneous or intradermal (i.d.) primary infection, with an LD₅₀ of ∼10⁶ CFU. Further, mice vaccinated with a sublethal dose of LVS can be used to study adaptive immunity; once mice have been primed by sublethal infection, they survive high secondary LVS challenge doses that are lethal for naïve mice.

Successful immune responses to LVS are dependent on several biological effectors derived from the innate and adaptive arms of the immune system. Among the most important components of the adaptive immune response are T cells, including CD4⁺, CD8⁺, and CD4⁻ CD8⁻ double-negative T cells (7, 8). Other cell types contributing to anti-Francisella immunity include B cells, natural killer (NK) cells, and dendritic cells (DC). Recently, we exploited this model system to develop a panel of immune mediators whose relative gene expression correlated with the degree of protection after vaccination (9). In this study, wild-type mice were immunized with a panel of vaccines of various efficacies. Gene expression analyses were performed using cells recovered from in vitro cocultures of Francisella-immune splenocytes with LVS-infected macrophages. These studies identified several genes, including mediators already known to be important, such as gamma interferon (IFN-γ) and tumor necrosis factor alpha (TNF-α), which were differentially regulated in cells from mice given the most protective vaccines compared to those given less effective vaccines (9–11). Among the highly differentially regulated molecules of unknown importance was IL-6.

IL-6 is a highly pleotropic molecule, with demonstrated roles in diverse biological functions, such as the acute-phase protein response (12) and both innate and adaptive immune responses. For example, IL-6 regulates dendritic cell development, chemokine production, B cell development and antibody secretion, and T cell maturation (13–17). IL-6 has been demonstrated to be important for primary resistance to several pathogens, including Listeria monocytogenes (18), Chlamydia trachomatis (19), Escherichia coli (20), and Yersinia enterocolitica (21). The role of IL-6 in immunity to F. tularensis, however, is not yet well defined. Several reports have shown that IL-6 is induced during primary infection of mice with either the virulent strain of F. tularensis SchuS4 or the attenuated LVS (22–26). In contrast to a potential role for prediction of vaccine-induced protection, IL-6 has also been suggested as a predictor of mortality in BALB/c mice infected intranasally with a dose of LVS that approximated the expected LD₅₀ (27). In this study, IL-6 was highly expressed in mice that became moribund after infection, but there was little IL-6 expression in recovering mice at the same time point. There is also evidence that IL-6 is expressed during successful vaccination with LVS and during recall challenge with virulent SchuS4 (25, 28). Thus, the available data indicate that IL-6 plays a role in resistance to many bacterial pathogens, is induced during several stages of immune responses to LVS, and is differentially expressed during both Francisella infection and vaccination of mice. We therefore hypothesized that IL-6 is important for host resistance to LVS, and we directly investigated its potential role in protection.

MATERIALS AND METHODS

Mice.

Male C57BL/6 and C57BL/6 IL-6^−/− mice (IL-6 knockout [KO] mice [29]) at ages 4 to 8 weeks were purchased from Jackson Laboratories (Bar Harbor, ME). All mice were housed in sterile microisolator cages and were given autoclaved food and water ad libitum. Animal studies using LVS were performed under protocols approved by the Animal Care and Use Committee (ACUC) of CBER, FDA. Experiments, including F. tularensis SchuS4 challenge were performed at the Rocky Mountain Laboratories (RML) under protocols approved by the RML ACUC. Both sets of protocols stressed practices and procedures designed to strictly minimize any suffering. Within each experiment, all animals were age matched.

Bacteria and growth conditions.

F. tularensis LVS (ATCC 29684) and F. tularensis strain SchuS4 (provided by Jeannine Peterson, Centers for Disease Control and Prevention, Fort Collins, CO) were grown in modified Mueller-Hinton (MH) broth (Difco Laboratories, Detroit, MI) to mid-logarithmic phase as previously described (30) and then frozen in 0.5-ml aliquots at −70°C until use. A sample from each batch of bacterial stock used for in vivo studies was subjected to quality control experiments to determine the number of CFU, to determine the proportion of dead bacteria using the Live/Dead BacLight bacterial viability kit (Invitrogen/Life Technologies, Grand Island, NY), to confirm typical colony morphologies, and to determine the intraperitoneal (i.p.) and intradermal (i.d.) LD₅₀s and times to death in adult male BALB/cByJ mice. Only bacterial stocks that exhibited an i.p. LD₅₀ of ≤3 CFU, time to death between 5 and 7 days after an i.p. dose of 10¹ or 10² CFU, and an i.d. LD₅₀ of ≥10⁵ CFU were used.

Bacterial infections.

Groups of C57BL/6J and IL-6 KO mice were infected i.d. with a dose range of LVS from 1 × 10¹ to 5 × 10⁶ CFU, delivered in 0.1 ml of sterile phosphate-buffered saline (PBS; Lonza Walkersville, MD) containing <0.01 ng of endotoxin/ml. Groups of mice were also infected i.n. Before intranasal infection, mice were anesthetized with 0.1 ml of a cocktail of Ketaject ketamine-HCl (1.5 mg/0.1 ml; Phoenix Pharmaceuticals, St. Joseph, MO) and AnaSed (0.3 mg/0.1 ml; Lloyd Laboratories, Shenandoah, IA) diluted in sterile PBS and given i.p. Mice were infected i.n. with a dose range of LVS of 2 × 10¹ to 2 × 10⁴ CFU diluted in sterile PBS, delivered in a 20-μl volume to a single nostril. Mice were also challenged via the i.p. route with 1 × 10⁶ to 5 × 10⁶ CFU LVS. Each LVS infection dose was also plated on MH plates to assess the actual number of CFU delivered, and the colonies were enumerated after 3 days incubation at 37°C, 5% CO₂. In addition, we performed two experiments in which groups of 10 WT and IL-6 KO mice were vaccinated with 10² CFU LVS i.d. and after 7 or 8 weeks (for experiment 1 or experiment 2, respectively) were challenged with ∼25 CFU of the virulent SchuS4 strain of F. tularensis given subcutaneously (s.c.). After infection, mice were monitored and were euthanized when clearly unable to reach food and water, according to the established criteria.

In vivo blocking of IL-6.

IL-6 was inhibited in vivo by injecting mice with an anti-IL-6 monoclonal antibody (clone 20F3; BioXCell, West Lebanon, NH). Groups of five WT mice were given 500 μg of anti-IL-6 antibody in a 100-μl volume via i.p. injection 24 h prior to infection, and groups of five IL-6 KO mice and five WT mice were given 100 μl PBS as a control. All three groups were given 10⁵ CFU LVS i.d. Two days after infection, anti-IL-6-treated mice were given a second injection of 300 μg of anti-IL-6 via i.p. injection.

Assessment of bacterial organ burdens and tissue pathology.

Bacterial burdens in the organs were determined at the indicated time points after infection. Mice were euthanized, and organs were removed aseptically and transferred to sterile homogenizer bags containing 5 ml of sterile PBS/organ. Organs were disrupted using a stomacher (Seward, England), and the homogenates were serially diluted and plated onto MH plates for colony enumeration. Organ homogenates were also frozen and stored at −70°C to be used for cytokine analysis. Blood from the femoral artery and heart was also collected. Whole blood was directly plated to enumerate CFU, or it was collected and diluted in 0.5 mg/ml EDTA to prevent clotting for subsequent analyses. For organ homogenates, the limit of detection was 50 CFU/organ; for blood samples, the limit of detection was 10¹ CFU/ml. For assessment of antibodies, sera were isolated from whole blood by using Sarstedt serum gel microtubes (Fisher Scientific, Pittsburg, PA) and then frozen until further analysis. In some experiments, portions of each lung, liver, and spleen were removed and preserved in 10% formalin. Samples were then sent to American Histolabs, Inc. (Gaithersburg, MD), where the tissues were embedded in paraffin and sectioned for slides and the slides were stained with hematoxylin and eosin (H&E). Pathology slides were analyzed by board-certified pathologists (O. Foreman and M. R. Anver) in a blinded fashion.

Blood analysis.

Whole blood and serum samples were analyzed by the Department of Laboratory Medicine, Clinical Center, National Institutes of Health, in Bethesda, MD. Whole blood complete blood counts (CBC) were performed on a Cell Dynn 3700 analyzer (Abbott Diagnostics, Abbott Park, IL). Serum chemistry analysis was performed on a Dimension Vista 1500 analyzer (Siemens Healthcare Diagnostics, Tarrytown, NY); analytes tested in the serum are listed below in Table 2.

Table 2.

CBC analyses of whole blood from wild-type and IL-6 KO mice infected i.d. with 10⁴ CFU LVS^a

CBC parameter	Composition of whole blood in mouse strain onb:
	Day 0 (naïve)		Day 3		Day 7
	WT	IL-6 KO	WT	IL-6 KO	WT	IL-6 KO
WBC	5.7 (5.4–6.0)	8.8 (7.6–14.4)	4.7 (4.2–5.2)	4.2 (3.6–4.8)	9 (7.04–12.6)	9.1 (7.6–22.1)
RBC	8.5 (7.3–9.8)	7.8 (6.5–12.4)	10.1 (7.0–13.2)	6.8 (5.4–10.9)	7.3 (4.6–8.2)	5.8 (4.3–9.3)
Hemoglobin	13.4 (11.6–15.2)	12.4 (10.4–15.2)	15.8 (11.6–20)	10.8 (8.8–16.3)	10.8 (7.1–12.4)	9 (6.8–14.4)
Hematocrit (PCV)	37.6 (31.6–43.6)	34.4 (28–50.3)	42.6 (30–55.2)	28.4 (22.8–47.9)	32.5 (21.4–36)	25 (19.5–39.2)
MCV	43.6 (43.5–44.3)	43.4 (40.6–44.2)	42.1 (41.9–42.3)	42.1 (41.9–44.1)	44.1 (42.2–46.4)	43.4 (42.2–45)
Platelets	1162 (1,056–1,268)	972 (732–1,108)	946 (888–1,004)	524 (236–576	668* (419–769)	216* (104–304)
PMNs	21.7 (21.3–22.1)	8.2 (5.9–21.2)	44 (28.1–59.8)	30.4 (27.8–33.2)	18.2 (14.5–28.3)	17 (6.4–32.3)
Lymphocytes	50.1 (48.4–51.8)	80.3 (65.6–82)	16.7 (16.1–17.3)	28.3 (18.7–37.1)	20.4 (14.6–46.1)	17.3 (13.9–54.2)
Monocytes	19.6 (16.1–23)	4.4 (2.6–9.4)	27.8 (15.2–40.3)	36.9 (15.8–40.1)	39.1 (20.1–46.4)	43.5 (31.5–51)
Eosinophils	2.3 (0.5–4.2)	0.8 (0.5–6.5)	0.9 (0.2–1.5)	1.3 (0.8–7.5)	0.7 (0.5–4.8)	0.7 (0.1–3.6)
Basophils	6.3 (5.8–6.8)	5.4 (2.8–8.7)	10.8 (7.4–14.1)	6.4 (6.2–9.5	15.6 (10.6–18.6)	11.7 (7.8–17.4)

^aGroups of six WT and six IL-6 KO mice were infected i.d. with 10⁴ CFU LVS. Three mice from each group, as well as naive mice, were sacrificed at day 3 and day 7 after infection. Equivalent amounts of blood from each mouse were collected and pooled, sera were isolated, and sera samples were analyzed by the Department of Laboratory Medicine, Clinical Center, NIH. Data are presented as the median (range) for each analyte, using all combined results from four independent experiments for the naïve groups and from up to seven independent experiments for the infected groups. *, values from the sera samples from LVS-infected WT and IL-6 KO mice were significantly different from each other, based on ANOVA. WBC, white blood cells; RBC, red blood cells; PCV, packed cell volume; MCV, mean corpuscular volume; PMNs, polymorphonuclear leukocytes.

^bUnits for WBC and platelets values are K/μl. Unit for RBC is millions/μl. Unit for hemoglobin is g/dl. Units for hematocrit, PMNs, lymphocytes, monocytes, eosinophils, and basophils are percentages. Unit for MCV is fl.

Necropsy, histology, and serum analysis performed at SAIC.

Groups of three C57BL/6J and IL-6 KO mice were infected with 10⁴ CFU LVS i.d., a dose ∼25-fold above the LD₅₀ for the IL-6 KO mice. At day 6 after infection, a time point chosen based on experimental data as likely being just before the IL-6 KO mice would succumb but at which they would have fulminant infection, the mice were transferred to SAIC—Frederick, Inc. (SAIC) and euthanized. Blood was taken for CBC analysis, which was performed on a Hemavet 950FS analyzer (Drew Scientific Group, Waterbury, CT), and was also analyzed by manual cell count. A full necropsy was performed, in which 46 organs and tissues were examined for gross findings. In addition, samples of each were fixed in 10% buffered neutral formalin, paraffin embedded, sectioned at 5 μm, and stained with hematoxylin and eosin for pathology evaluation.

Characterization of antibody responses.

Titers of specific anti-LVS serum antibodies were determined by enzyme-linked immunosorbent assay (ELISA) as described previously (31). Briefly, Immulon 1 plates were coated with live LVS, washed, and blocked with 10% calf serum, and serum samples were serially diluted. In each assay, sera from naïve mice were used as a negative control, and sera from LVS-hyperimmune mice, generated by repeated immunization of mice with LVS, were used as a positive control. Horseradish peroxidase-labeled antibodies (anti-IgM or anti-IgG that detected IgG1, IgG2a, IgG2b, and IgG3; Southern Biotech, Birmingham, AL) were added, and 2,2′-azinobis(3-ethylbenthiazolinesulfonic acid) peroxidase substrate (Kirkegaard & Perry Laboratories, Gaithersburg, MD) was used for color development. The endpoint titer was defined as the lowest dilution of serum that gave an optical density at 405 nm greater than the optical density at 405 nm when 3 standard deviations were added to the OD value of the matched dilution of normal prebleed mouse serum and that was also greater than 0.025.

Preparation of splenocytes.

Spleens were aseptically removed and homogenized in 2% fetal calf serum (FBS; HyClone, Logan, UT). Single-cell suspensions were generated, red blood cells were lysed using ACK lysis buffer, and cell viability was determined using trypan blue dye exclusion.

In vitro overlay assay.

Cocultures were performed in 24- or 48-well tissue culture plates as described previously (32–37). Briefly, bone marrow macrophages (BMMØ) were cultured in complete Dulbecco's modified Eagle's medium (DMEM supplemented with 10% heat-inactivated FCS [HyClone, Logan, UT], 10% L-929-conditioned medium, 0.2 mM l-glutamine, 10 mM HEPES buffer, and 0.1 mM nonessential amino acids) in 24-well plates. Confluent adherent macrophage monolayers were infected for 2 h with F. tularensis LVS at a multiplicity of infection (MOI) of 1:20 (bacterium:BMMØ ratio), washed, treated for 60 min with 50 μg/ml gentamicin, and washed extensively with antibiotic-free medium. Single-cell suspensions of splenic lymphocytes derived from vaccinated mice (5 × 10⁶/well, or as indicated) were added to LVS-infected macrophages. At 72 h after infection, supernatants from harvested cells were collected and stored at −70°C until analyzed for nitric oxide and cytokines as described below. Intracellular bacterial burdens in adherent infected macrophages were determined by lysing the macrophages with water and plating the lysate as previously described (32). In some experiments, the indicated wells with lymphocytes cocultured with LVS-infected macrophages were left either untreated or were treated with 25 μg/ml anti-mouse IL-6 antibody (rat IgG1 MAB406; R&D Systems, Minneapolis, MN) or 25 μg/ml of IgG1 isotype control (BD Pharmingen, San Diego, CA), as indicated.

In vitro T cell purification.

Thy 1.2⁺ T cells were purified from the total immune splenocyte preparation by using the Dynal Mouse Flow Comp Pan T (CD90.2) isolation kit (Invitrogen, Oslo, Norway) according to the manufacturer's instructions. The purity of the T cell fraction was established by multiparameter flow cytometry to be approximately 90 to 95% CD45⁺ TCRβ⁺ T cells (data not shown).

Flow cytometry.

Single-cell suspensions prepared from spleens were stained for a panel of murine cell surface markers and analyzed using a Becton, Dickinson LSR II flow cytometer (San Jose, CA) and FlowJo software (Tree Star, Inc.) as previously described with minor modifications (9, 33, 34). Briefly, cells were washed and resuspended in flow cytometry buffer (PBS–2% serum). Nonspecific binding of antibodies was inhibited by blocking Fc receptors with anti-CD16 (Fc block; BD Pharmingen) for 10 min on ice. To discriminate live from dead cells, a staining step was performed using a commercially available kit and following the manufacturer's instruction (Live/Dead staining kit; Invitrogen). The cells were then washed and stained for cell surface markers. Antibody concentrations were optimized separately for use in nine-color staining protocols, using appropriate fluorochrome-labeled isotype-matched control antibodies. The following antibodies were used: anti-B220 (clone RA3-6B2), anti-CD19 (clone 1D3), anti-T cell receptor β (anti-TCRβ; clone H57-597), anti-CD4 (clone RM4-5), anti-CD8β (H35-17.2), anti-NK1.1 (clone PK136), anti-CD11b (clone M1/70), anti-Gr-1 (clone RB6-8C5), and anti-CD11c (cloneHL3), each labeled with a variety of fluorochromes as needed (the above antibodies were purchased from BD Pharmingen).

Cytokine, acute-phase protein, and nitrite measurements.

Organ homogenates from spleens, lungs, and livers, sera, and supernatants recovered from in vitro cocultures were assayed using standard sandwich ELISAs, according to the manufacturer's instructions (BD Pharmingen; Cayman Biochemical, Ann Arbor, MI; Life Diagnostics, West Chester, PA). The absorbance was read at 405 nm on a VersaMax tunable microplate reader with a reference wavelength of 630 nm (Molecular Devices, Sunnyvale, CA). Cytokine or acute-phase protein concentrations were determined by comparing unknown values to a standard curve made with recombinant protein at known concentrations (BD Pharmingen; Life Diagnostics; Cayman Chemical), using four-parameter fit regression within the SOFTmax Pro ELISA analysis software (Molecular Devices). For cytokine assays, antibody pairs and standards were purchased from BD Pharmingen. Nitric oxide was estimated in culture supernatants using the Griess reaction. Samples of supernatants were incubated with an equal volume of commercial Griess reagent (Life Technologies, Grand Island, NY), and absorbance was measured at 548 nm. Nitrite (NO₂) was measured by comparison to serially diluted NaNO₂ as a standard, using four-parameter fit regression as described above.

Protein quantitation in organ homogenates using a Quantibody cytokine array.

A panel of murine cytokines and chemokines were assessed using Quantibody cytokine arrays (Ray Biotech, Norcross, GA), according to the manufacturer's protocols. Each organ homogenate sample was incubated in an individual well. Wells were washed and incubated with capture antibodies, followed by incubation with biotinylated detection antibodies. After a final incubation with streptavidin-conjugated Alexa Fluor 55 streptavidin, the chips were analyzed on a GenePix 4000A laser fluorescence scanner (Molecular Devices, Sunnyvale, CA) on the Cy3 (532-nm) channel, and data were extracted using GenePix software. Protein quantitation was accomplished using a standard curve of serial dilutions of a known quantity of each protein run in tandem on each chip. Quantibody mouse cytokine arrays 4, 5, and 6 were used, each providing the detection capability of 40 unique proteins, for a total of 120 cytokines and chemokines analyzed. A list of the proteins available on each chip is listed on the manufacturer's website (Ray Biotech). The organ homogenates from lungs, livers, and spleens from two mice per group from two independent experiments were analyzed on each chip set.

Statistical analyses.

The statistical significance of differences within parameters was assessed using Student's t test or analysis of variance (ANOVA; SigmaPlot; Systat Software, Inc., San Jose, CA). The LD₅₀s were calculated using the method described by Reed and Muench (38).

RESULTS

Characterization of the production and role of IL-6 during primary LVS infection.

To assess the role of IL-6 in immunity to F. tularensis LVS, we examined whether IL-6 protein was induced and found systemically during LVS infection in vivo. Male C57BL/6 mice were infected i.d. with 10⁴ CFU of LVS. At days 1, 3, and 6 after infection, mice were euthanized, and LVS CFU were enumerated in lungs, livers, and spleens. In parallel, the homogenates were assessed for IL-6 protein. The LVS bacterial burdens in lungs, livers, and spleens of the infected animals rose and peaked about day 3 after infection, with greater than 10³ bacteria in each organ, where CFU levels remained at day 6 (Fig. 1A). IL-6 was detected in homogenates of uninfected organs but increased significantly in response to LVS infection within 1 day in livers, peaked at day 3, and declined thereafter (Fig. 1B). There was a significant increase by day 3 in the spleens as well (P ≤ 0.05 for day 3 compared to naïve animals); however, in contrast to previous results using intranasal infection (23), levels of IL-6 did not increase in response to infection in the lungs. Thus, IL-6 can readily be detected in several infected organs in response to active LVS infection.

Fig 1.

IL-6 is produced in response to intradermal Francisella tularensis LVS infection. Male C57BL/6 mice were infected i.d. with 10⁴ CFU of F. tularensis LVS. At the indicated time points after infection, groups of three mice were euthanized, and the lungs, livers, and spleens were harvested and homogenized. A group of three naïve mice was included as controls. (A) CFU counts in organ homogenates were determined and are expressed as the mean total CFU per organ ± the standard deviation. *, P ≤ 0.05, comparing CFU in infected mice on day 1 versus day 3 for each organ, using Student's t test. There were no significant differences in CFU between day 3 and day 6 for any organ (P > 0.05). (B) Organ homogenates were assayed for the presence of IL-6 by ELISA; results are expressed as the mean pg/ml ± the standard deviation. Results shown are from one representative experiment of three of a similar design. *, P ≤ 0.05, comparing IL-6 (in pg/ml) in livers or spleens from naive mice versus levels in organs obtained on day 1 after infection and comparing levels on day 1 to day 3 after infection, as determined with Student's t test.

To directly examine the importance of IL-6 in immunity to LVS, WT mice and mice deficient in IL-6 production (IL-6 KO) were compared in terms of responses to primary LVS infection. Groups of IL-6 KO and C57BL/6 (WT) mice were infected i.d. with graded doses of LVS ranging from 10² to 5 × 10⁶ CFU and monitored for survival over time. As reported previously (7), very few WT mice succumbed to infection except at the highest dose, 5 × 10⁶ CFU, and all mice in this group died (Fig. 2A; 1 × 10⁶ and 5 × 10⁶ doses not shown); the i.d. LD₅₀ for the WT mice was 2 × 10⁶ CFU. In stark contrast, IL-6 KO mice were much more susceptible, with 60% of the mice succumbing to the lowest dose, 10² CFU (Fig. 2A). The calculated LD₅₀ for the IL-6 KO mice was 4 × 10², nearly 4 logs less than that of WT mice. To determine whether the increased susceptibility of the IL-6 KO mice was related to the route of infection, i.n. LVS infections were also compared. WT and IL-6 KO mice were infected with graded doses of LVS ranging from 2 × 10¹ to 2 × 10⁴ CFU. WT mice exhibited an intranasal LD₅₀ of ∼6 × 10³ CFU LVS, but IL-6 KO mice exhibited increased susceptibility to i.n. LVS infection, with an LD₅₀ of ∼5 × 10² CFU (Fig. 2B).

Fig 2.

IL-6 KO mice exhibit increased susceptibility to both intradermal and intranasal Francisella tularensis LVS infection compared to C57BL/6J mice. Groups of three to five mice were infected via the i.d. route (A) with the indicated amount of LVS or by the i.n. route (B) following anesthesia. Survival was monitored for at least 30 days, and the LD₅₀ was calculated. Results shown are from one representative experiment of two of a similar design. Overlapping lines are offset for clarity. The results for the 1 × 10⁶ CFU dose (for WT, 5 of 5 survived; for IL-6 KO, 0 of 5 survived) and 5 × 10⁶ CFU dose (no survivors for WT or IL-6 KO) after intradermal infection are not shown.

We further examined the role of IL-6 during infection by directly blocking IL-6 in vivo in WT mice with an anti-IL-6 monoclonal antibody prior to infection with 10⁵ CFU LVS i.d. Whereas untreated WT mice survived, all but one mouse treated with anti-IL-6 succumbed within 10 days. The extent and times to death of anti-IL-6-treated mice were similar to IL-6 KO mice given the same dose (Fig. 3). Thus, IL-6 is critical for survival of LVS infection.

Fig 3.

C57BL/6J mice treated with anti-IL-6 antibody exhibit increased susceptibility to i.d. Francisella tularensis LVS infection. A group of five C57BL/6 mice were given 500 μg of anti-IL-6 antibody intraperitoneally. After 24 h, the five antibody-treated mice, five IL-6 KO mice, and five C57BL/6 mice (WT) were infected i.d. with 10⁵ CFU LVS. Two days after infection, the antibody-treated group were given a second dose of 300 μg anti-IL-6 antibody. Survival was monitored for at least 20 days after infection. This experiment was performed twice with similar results, and pooled results for 10 mice per group are shown.

Given the increased susceptibility of IL-6 KO mice during primary infection, we further characterized the role of IL-6 during the early response period in more detail. Wild-type and IL-6 KO mice were infected with 10⁴ i.d., a dose that approached the LD₁₀₀ for the IL-6 KO mice. At days 3, 6, and 8 after infection, organs from each infected mouse strain were examined for bacterial burdens as well as cytokine content. By day 6 after inoculation, the IL-6 KO mice had approximately 10-fold more bacteria in lungs, livers, and spleens, a significant increase compared to those from the WT mice (Fig. 4). However, there were no other consistent differences in bacterial loads in these organs at other time points (including lungs on day 3, although higher levels in WT mice were observed in this particular example). Organ homogenates were also assayed for TNF-α (Fig. 4), IL-12p40, IL-12p70, monocyte chemoattractant protein 1 (MCP-1), IFN-γ, IL-18, and IL-17A (data not shown). Despite the differences in bacterial burdens on day 6, there were no significant differences in the concentrations of TNF-α (Fig. 4) or any of the tested cytokines in any organ over time (data not shown). This suggests that the increased susceptibility of the IL-6 KO mice is not due to overt differences in production of any of these immune mediators.

Fig 4.

IL-6 KO mice exhibit increased bacterial burdens in organs following i.d. infection with 10⁴ CFU LVS, but there were no difference in tissue levels of proinflammatory cytokines. C57BL/6 and IL-6 KO mice were infected i.d. with 10⁴ CFU LVS, a dose that is lethal for the IL-6 KO but not C57BL/6J mice. Three mice per group were sacrificed at day 3, 6, or 8 after infection, and organs were harvested, homogenized, and plated on MH plates to enumerate CFU. Homogenates were also assayed for cytokines by sandwich ELISA. Data shown are for TNF-α in each organ, but other cytokines measured included IL-12p40, IL-12p70, MCP-1, IFN-γ, IL-18, and IL-17A. There were no statistical differences in the levels of cytokines between the wild-type or IL-6 KO mice in any organ at any time point tested. Each data point represents the average ± standard deviation of three samples. *, P ≤ 0.05 by Student's t test. Results shown are from one representative experiment of seven of a similar design.

To explore other possible differences, we also measured the levels of 120 cytokines and chemokines in lung, liver, and spleen homogenates obtained from WT and IL-6 KO mice infected with 10⁴ CFU LVS i.d. on day 3 after infection, the peak time of IL-6 production. The protein concentrations from two mice from each group, from two independent experiments, were averaged and compared between samples from WT and IL-6 KO mice. Only one protein was upregulated more than 2-fold across lungs, livers, and spleens of the IL-6 KO mice compared to WT mice, namely, granulocyte colony-stimulating factor (G-CSF) (data not shown). There were no proteins downregulated in all three organs in the IL-6 KO mice. IL-23 and amphiregulin (AR) were each slightly upregulated in livers of IL-6 KO mice but downregulated greater than 3-fold in the spleens and lungs (data not shown).

In addition to examining LVS burden in tissues, the level of bacteremia was assessed in whole blood taken from WT and IL-6 KO mice infected with 10⁴ CFU LVS i.d. at days 3 and 7 after infection. It was previously reported that in BALB/c mice infected with LVS, persistent bacteremia at day 7 after infection was a marker of morbidity (27). Indeed, IL-6 KO mice given 10⁴ CFU LVS i.d. exhibited detectable increases in bacteremia at days 3 and 7 after infection, while the wild-type mice exhibited few bacteria in blood (Fig. 5).

Fig 5.

IL-6 KO mice exhibit an increase in bacteremia following i.d. infection with 10⁴ CFU LVS. C57BL/6J and IL-6 KO mice were infected i.d. with 10⁴ CFU LVS. Three mice per group were sacrificed at day 3 or 7 after infection, and blood was harvested and plated on MH plates to enumerate CFU. Each data point represents the average ± standard deviation of three samples. *, P ≤ 0.05 by the Student t test. Results shown are from one representative experiment of three of a similar design.

To further explore the mechanisms underlying the susceptibility of IL-6 KO mice, blood from WT and IL-6 KO mice infected with 10⁴ CFU LVS i.d. was also analyzed for CBC, and serum samples were isolated for serum chemistry analyses. Both tests were used to compare blood and sera from naïve WT and IL-6 KO mice. Even though inbred mouse strains were studied, there was considerable variability among individual mice and groups from experiment to experiment for several of the analytes. The ranges of values for both CBC and serum chemistry are presented in Tables 1 and 2, respectively. The levels of several serum analytes changed over the course of infection, however. There were few clear or informative differences between the WT and IL-6 KO mice, partially due to the large range for some of the values (Table 1). There was a significant increase in alanine aminotransferase (ALT) and aspartate transaminase (AST) in the IL-6 KO mice at day 7, the time of increased bacterial burden (P ≤ 0.04 by ANOVA), but no other serum analyte levels were significantly different between sera from LVS-infected WT and IL-6 KO mice at day 7. Both groups also had similar patterns of blood leukocyte populations, as determined by CBC (Table 2) and manual count (data not shown). The only notable and consistent difference was reduced numbers of platelets in IL-6 KO mice at day 3 and day 7 (P ≤ 0.009 by ANOVA; Table 2).

Table 1.

Serum chemistry analysis of wild-type and IL-6 KO mice infected ID with 10⁴ CFU LVS^a

Analyte	Concn of analyte found in mouse strain on:
	Day 0 (naïve)		Day 3		Day 7
	WT	IL-6 KO	WT	IL-6 KO	WT	IL-6 KO
Glucose	328 (321–475)	472 (204–504)	148* (68–217)	364* (228–460)	272 (180–404)	179 (124–288)
Cholesterol	72 (56–88)	66 (64–75)	112 (91–146)	72 (52–88)	76 (72–100)	60 (54–74)
Triglycerides	136 (101–148)	208 (169–272)	84 (52–131)	128 (94–176)	153 (132–178)	86 (54–110)
Sodium	146 (140–162)	160 (148–161)	156 (148–162)	156 (143–162)	146 (137–159)	155 (149–161)
Chloride	111 (107–134)	133 (112–134)	126 (111–138)	125 (106–138)	113 (108–132)	126 (116–133)
Magnesium	1.0 (0.35–1.77)	0.48 (0.43–1.42)	0.8 (0.48–1.96)	0.97 (0.48–1.46)	1.86 (0.54–2.05)	1.5 (0.56–2.07)
Alkaline phosphatase	89 (78–150)	86.5 (80–104)	33* (20–56)	67* (59–196)	31 (24–37)	86 (68–102)
ALT	83 (50–207)	56 (48–177)	87 (56–290)	133 (76–496)	279* (181–336)	816* (414–1,005)
AST	287 (198–1,194)	248 (204–921)	449 (180–1,230)	401 (232–1,352)	644* (473–904)	1,703* (626–1,962)
Amylase	1,134 (961–1,316)	1,408 (1,324–2,248)	671* (516–816)	1,043* (554–5,834)	1,220 (956–1,692)	1,276 (900–1,956)
Creatinine kinase	2,609 (1,219–3,948)	3,628**	2,021 (2,008–3,188)	3572 (878–5,219)	1,004 (239–1,770)	4,458 (3,112–5,804)
LDH	864 (684–949)	816 (588–828)	1,144 (624–2,331)	864 (828–2,335)	1,450 (1,440–1,646)	3,867 (3,423–4,311)
Uric acid	5 (4.5–8.7)	9 (5.9–9.6)	6 (4–7.2)	7.9 (6–9.2)	6.8 (6.6–12.4)	6.5 (6–9.9)

^aGroups of six WT and six IL-6 KO mice were infected i.d. with 10⁴ CFU LVS. Three mice from each group, as well as naive mice, were sacrificed at day 3 and day 7 after infection. Equivalent amounts of blood from each mouse were collected and pooled, sera were isolated, and sera samples were analyzed by the Department of Laboratory Medicine, Clinical Center, NIH. Data are presented as the median (range) for each analyte, in pg/ml, using all combined results from four independent experiments for the naïve groups and up to 7 for the infected groups. *, values from the samples from LVS-infected WT and IL-6 KO mice were significantly different from each other based on ANOVA; **, only 1 sample was available. Units for sodium, chloride, and magnesium values are mmol/liter. Units for glucose, cholesterol, triglycerides, and uric acid are mg/dl. Units for alkaline phosphatase, ALT, AST, amylase, creatinine kinase, and LDH are units/liter.

Because IL-6 induces acute-phase proteins and IL-6 KO mice have a demonstrated deficiency in this process, we measured the levels of several key acute-phase proteins during LVS infection (29, 39, 40). WT and IL-6 KO mice were given 10⁵ CFU LVS i.d., and sera and liver tissues were harvested at 24 h and 48 h after infection. We measured the levels of serum amyloid A (SAA), haptoglobin, fibrinogen, and prostaglandin E2 (PGE₂) by ELISA. There was a 5- to 10-fold induction of SAA and haptoglobin between 24 h and 48 h after infection in WT serum and livers (Fig. 6). Both liver homogenates and serum samples from LVS-infected IL-6 KO mice had greatly decreased levels of SAA and haptoglobin compared to those from WT mice (P ≤ 0.05 by Student's t test). However, there were no demonstrable differences in the levels of fibrinogen or PGE₂ between the WT and IL-6 KO mice (data not shown). Further, there were no obvious differences in other mediators considered to be acute-phase proteins, such as TNF-α (Fig. 4) or KC (data not shown).

Fig 6.

IL-6 KO mice exhibit a decrease in certain acute-phase proteins following i.d. infection with 10⁵ CFU LVS. C57BL/6J and IL-6 KO mice were infected i.d. with 10⁵ CFU LVS. Three mice per group were sacrificed at day 1 or 2 after infection, and serum and livers were harvested. SAA and haptoglobin levels were determined by sandwich ELISA. Each data point represents the average ± standard deviation of three samples. *, P ≤ 0.05 by the Student t test. Results shown are from one representative experiment of two of a similar design.

IL-6 is known to be important for immune cell maturation and trafficking (14). We therefore examined the composition of immune cells in tissues during infection. Wild-type and IL-6 KO mice were infected with 10⁴ CFU LVS i.d., and at days 1 and 6 after infection, splenocytes were analyzed by multiparameter flow cytometry. The proportions of B cells, T cells, and myeloid cells were similar between spleens of LVS-infected WT and IL-6 KO mice at both time points examined (Table 3). We also examined the cell compositions in peritoneal cavities and bone marrow of LVS-infected WT and IL-6 KO mice on days 3 and 7, but we found no obvious differences in these compartments (data not shown).

Table 3.

Splenocyte subpopulations in mice infected with 10⁴ CFU LVS i.d.^a

Cell type and marker(s)	% of cells with indicated marker(s) in mouse strain on:
	Day 1		Day 6
	WT	IL-6 KO	WT	IL-6 KO
B cells
CD19+ B220+	33.3 ± 4.4	33.7 ± 2.5	48.5 ± 1.3	54.5 ± 4.7
T cells
Total α/β TCR+	52.3 ± 3.2	50.3 ± 6	31 ± 2.5	31.4 ± 3.3
CD4+	31.6 ± 3	28.1 ± 4	17.3 ± 4.3	14.7 ± 2.5
CD8+	23.6 ± 10.1	20.3 ± 1.9	13.5 ± 3.6	15.0 ± 3
CD4− CD8−	2 ± 0.8	2.4 ± 2	2.2 ± 0.3	1.5 ± 0.4
Macrophages
CD11b+ CD11c− Gr-1−	2.9 ± 0.8	4 ± 0.6	3.4 ± 0.5	2.6 ± 1.1
Dendritic cells
CD11c+	0.7 ± 0.1	0.9 ± 1.9	2.7 ± 1.5	3.4 ± 2
Neutrophils
Gr-1hi CD11bhi	6.0 ± 4.4	3.9 ± 2.4	4.6 ± 2.3	2.5 ± 1.5
NK cells
NK1.1+	3.6 ± 1.8	3.7 ± 1.4	3.4 ± 0.5	2.3 ± 0.5

^aWild-type C57BL/6 and IL-6KO mice were infected i.d. with 10⁴ CFU LVS. Groups of three mice from each group were sacrificed on days 1 and 6 after infection, spleens were harvested and pooled, and splenocytes were prepared for analyses by multiparameter flow cytometry. Cells were analyzed on an LSR II cytometer using appropriate single-color compensation and isotype controls; only live nonaggregated cells were included in the analyses. Data represent means ± standard deviations of data combined from three independent experiments.

We furthered our investigation into the cause of the mortality of the IL-6 KO mice by examining tissue pathology and performing formal necropsies. WT and IL-6 KO mice were infected with 10⁴ CFU LVS i.d. Lungs, livers, and spleens were obtained at days 1, 3, or 7 after infection, processed, and analyzed for pathology. There were no obvious differences between the WT and IL-6 KO mice in tissue pathology in any of the organs at any time point examined (for example, see Fig. S1 in the supplemental material, specifically, the data for days 3 and 6 for spleens). Similarly, WT and IL-6 KO mice that were infected with 10⁴ CFU LVS i.d. were subjected to complete necropsy on day 6 after infection, a time point at which we observed the greatest difference in bacterial burden in the organs and nearing the time to death of IL-6 KO mice infected with this dose. As expected, inflammatory lesions were seen in the livers of all infected mice, whereas the IL-6 KO mice also had splenic and lymph node granulomatous inflammation. Of note, despite evidence for reduced levels of platelets, there were no histologic lesions indicative of disseminated intravascular coagulopathy, and sections of pituitary and adrenal glands appeared to be normal in both WT and KO LVS-infected mice. Other lesions present in LVS-infected IL-6 KO mice, but not WT mice, were hyperplasia of crypts in the small intestine, apoptosis in crypts of the large intestine (cecum, colon, and rectum), and apoptosis of pancreatic acinar cells.

Characterization of the role of IL-6 during LVS vaccination and secondary challenge.

Given the requirement for IL-6 in resistance to primary LVS infection, we examined the potential roles of IL-6 during the adaptive immune response to LVS. Previously, our laboratory identified IL-6 as an immune mediator whose relative expression correlated with the degree of vaccine-induced protection against LVS lethal challenge (9). WT and IL-6 KO mice were therefore vaccinated with a sublethal dose of LVS, allowed to clear the primary infection, and at 8 or 12 weeks after initial priming were administered a secondary challenge with large lethal doses of LVS. In the first experiment, WT and IL-6 KO mice were primed with 10⁴ CFU LVS i.d. Although few IL-6 KO mice survived this dose, those that did were challenged 8 weeks later with 5 × 10⁶ CFU LVS i.p. (a dose that is more than 100,000× the LD₅₀ for naïve mice). All naïve mice succumbed to infection with 5 × 10⁶ CFU i.p. within 4 days. Four of five LVS-primed IL-6 KO mice survived the secondary i.p. challenge, similar to the LVS-primed WT mice (Table 4). In a second set of experiments, WT and IL-6 KO mice were primed with either 10² CFU LVS i.d. or 10² CFU LVS i.n. Twelve weeks later, mice that survived the primary infection were challenged with 10⁶ CFU LVS i.p. or 8 × 10⁷ CFU LVS i.n. Naïve control mice succumbed to infection within 4 days following i.p. challenge and 6 days following i.n. challenge. LVS-immune IL-6 KO mice, primed either via the intradermal or intranasal route, survived secondary challenge, as did LVS-immune WT mice, for both routes of challenge (Table 4).

Table 4.

Mice that survive primary LVS infection survive high-level secondary challenge with F. tularensis LVS^a

Expt no.	Strain	Primary infection (CFU)	Secondary challenge (CFU)	No. of survivors/total no. infected
1	C57BL/6	PBS	5 × 106 i.p.	0/5
	C57BL/6	104 i.d.	5 × 106 i.p.	4/5
	IL-6 KO	104 i.d.	5 × 106 i.p.	4/5
2	C57BL/6	PBS	106 i.p.	0/3
	C57BL/6	PBS	8 × 107 i.n.	0/3
	C57BL/6	102 i.d.	106 i.p.	4/5
	IL-6 KO	102 i.d.	106 i.p.	4/4
	C57BL/6	102 i.d.	8 × 107 i.n.	5/5
	IL-6 KO	102 i.d.	8 × 107 i.n.	3/4
	C57BL/6	102 i.n.	106 i.p.	5/5
	IL-6 KO	102 i.n.	106 i.p.	4/4

^aWild-type C57BL/6 and IL-6 KO mice were vaccinated by primary i.d. infection with F. tularensis LVS (10⁴ or 10² CFU, as indicated), by intranasal infection (10² CFU), or with PBS as a control. Eight (experiment 1) or 12 (experiment 2) weeks later, surviving mice were challenged i.p. with 1 × 10⁶ to 5 × 10⁶ LVS i.p. or with 8 × 10⁷ LVS i.n., as indicated.

These results indicated that IL-6 was not strictly required for protection when vaccinated mice were lethally challenged with the attenuated LVS strain. To assess the role of IL-6 during secondary challenge with fully virulent type A F. tularensis, groups of 10 WT mice and 10 IL-6 KO mice each were primed with 10² CFU LVS i.d. and then challenged at 7 or 8 weeks with ∼25 CFU of the virulent SchuS4 strain. However, there was no obvious difference in the time to death for the WT and IL-6 KO mice or in the total number of mice in each group that succumbed to infection (data not shown).

Nonetheless, compensatory alterations in the relative contributions of B and T cell responses in LVS-vaccinated IL-6 KO mice compared to WT mice may result in the observed lack of difference in protection. Because IL-6 is known to be important for B cell differentiation and antibody production (14), antibody levels were examined in mice after LVS infection. Groups of WT and IL-6 KO mice were infected with 10² CFU LVS i.d., and serum samples obtained on day 28 were evaluated by ELISA. Sera from IL-6 KO mice had slightly elevated levels of IgM at day 28 (endpoint titers of 2,560 for sera from WT mice and 5,120 for sera from IL-6 KO mice) and slightly reduced levels of IgG compared to the WT mice (endpoint titers of 20,480 for sera from WT mice and 10,240 for sera from IL-6 KO mice). Nonetheless, anti-LVS antibody production was not substantively impaired in the IL-6 KO mice (Fig. 7).

Fig 7.

Sera from LVS-infected IL-6 KO mice have slightly elevated IgM levels compared to wildtype mice, but equivalent levels of IgG, at 28 days after infection. C57BL/6J and IL-6 KO mice were prebled (day 0) and then infected i.d. with 10² CFU LVS. Serum samples were collected 28 days after infection, and endpoint titers of specific anti-LVS antibodies were determined by ELISA. Negative-control serum from uninfected normal mice (NMS) and positive-control serum from LVS-vaccinated hyperimmune mice (IMS) were included as comparators. Reciprocal endpoint titers were as follows: for IgM, WT = 2,560, IL-6 KO = 20,480; for IgG, WT = 20,480, IL-6 KO = 10,240. Results shown are from one representative experiment of three of a similar design.

We previously demonstrated the value of an in vitro coculture system. This approach for evaluating adaptive T cell immunity quantifies the ability of LVS-primed splenocytes to control intramacrophage bacterial growth (32, 37). Here, immune lymphocytes were harvested from the spleens of LVS-vaccinated WT and IL-6 KO mice and compared for their ability to control intramacrophage growth of LVS. In addition, bone marrow macrophages derived from both WT and IL-6 KO mice were compared to assess the potential contribution of IL-6 from LVS-infected macrophages. The initial uptake of LVS by macrophages derived from WT or IL-6 KO mice was equivalent (Fig. 8A, day 0). Bacteria also replicated in macrophages from either WT or IL-6 KO mice to equivalent levels by day 3 after infection (“no splenocytes”). Splenocytes from neither naïve WT nor naïve IL-6 KO mice controlled infection, as expected. Splenocytes from WT mice controlled intracellular LVS replication in a dose-dependent manner (Fig. 8A). On a per cell basis, splenocytes from LVS-immune WT mice controlled intracellular growth significantly better than those from the immune IL-6 KO mice. This outcome was observed when WT immune cells were cocultured with LVS-infected IL-6 KO macrophages, or when IL-6 KO immune cells were cocultured with LVS-infected WT macrophages (data not shown). Thus, there is a subtle but statistically significant defect in the ability of LVS-immune splenocytes from vaccinated IL-6 KO mice to control intramacrophage LVS infection in vitro, compared to splenocytes from LVS-immune wild-type mice.

Fig 8.

Splenocytes from LVS-immune IL-6 KO mice are less able to control intracellular growth of LVS than splenocytes from primed WT mice. (A) WT and IL-6 KO mice were infected i.d. with 10⁴ CFU LVS. Two mice from each group were sacrificed, and splenoctyes were isolated and pooled; splenocytes were also obtained from naive mice. Bone marrow-derived Mϕ from WT and IL-6 KO mice were infected with LVS, and then the indicated numbers of homologous lymphocytes were added to triplicate wells of infected macrophages; 5 × 10⁶ naive splenocytes were added per well. Intracellular bacterial burdens were assessed immediately after infection at 0 h and at 72 h after the initiation of cocultures. (B) Supernatants were harvested at 72 h and assayed for RNI levels by using the Griess reagent. (C) Bone marrow-derived macrophages from IL-6 KO mice were infected with LVS, and then the indicated numbers of naive total lymphocytes or LVS-immune enriched Thy1.2⁺ T cell subpopulations were added to triplicate wells of infected macrophages; 5 × 10⁵ naive T cells were added per well. *, P ≤ 0.05 by the Student t test in a pairwise comparison of each cell concentration. Results shown are from one representative experiment of seven of a similar design.

The ability of LVS-immune splenocytes to control intracellular LVS growth relies on the production of IFN-γ, TNF-α, and the resulting production of reactive nitrogen intermediates, including nitric oxide (RNI) (32–34, 36). Supernatants were therefore collected from cocultures and assessed for levels of RNI. There were no significant differences in the levels of RNI between matched cocultures with naive or immune splenocytes from IL-6 KO mice compared to those from WT mice (Fig. 8B). Similarly, there were also no significant differences in the levels of other proinflammatory cytokines tested in the supernatants from cocultures with IL-6 KO cells compared to those with WT cells, including IFN-γ, TNF-α, IL-12p40, IL-12p70, MCP-1, IL-18, IL-10, and IL-17A (data not shown).

T cells are the primary cellular mediators in this coculture assay, and consequently purified T cells are much more potent at controlling intracellular LVS replication than total splenocyte populations (32, 33). Therefore, the activities of T cells from immune WT and IL-6 KO animals were directly compared. T cells were enriched from splenocytes derived from LVS-immune IL-6 KO and WT mice and added to IL-6 KO LVS-infected macrophage monolayers. Cultures containing T cells enriched from IL-6 KO mice had significantly more bacteria than WT controls (Fig. 8C). The cell fraction remaining after T cell enrichment contained B cells, NK cells, and other myeloid cells. As expected (32), this cell fraction did not control LVS replication (data not shown). As for total splenocytes, we measured IFN-γ and RNI in overlay supernatants, and there were no significant differences in levels of either endpoint in WT and IL-6 KO mice (data not shown).

Given the diminished control seen with the IL-6 KO splenocytes and T cells, we sought to determine whether IL-6 was directly necessary for LVS control in the in vitro coculture assay, or if it was important for in vivo priming of cells. We studied the interaction of LVS-infected macrophages and LVS-primed splenocytes from WT mice in the presence or absence of anti-IL-6 blocking antibodies. We observed similar intracellular LVS growth inhibition when splenocytes were incubated with anti-IL-6 antibody, isotype control antibody, or without antibody (Fig. 9A). Indeed, IL-6 increased with the addition of immune splenocytes, with or without the isotype control, compared to naïve wells or those with no splenocytes, but it was successfully blocked in cocultures treated with anti-IL-6 antibody (Fig. 9B). Thus, IL-6 was not required to control intracellular growth in the overlay assay when cells from LVS-primed WT mice were studied.

Fig 9.

Blocking IL-6 in vitro does not dampen LVS growth inhibition by splenocytes. C57BL/6J mice were infected i.d. with 10⁴ CFU LVS, and splenocytes were isolated. Immune splenocytes from WT mice were added to LVS-infected WT macrophages. In addition, one set of wells contained anti-IL-6 blocking antibodies, and another contained IgG1 isotype control antibodies. (A) Intracellular bacterial burdens were assessed immediately after infection at 0 h and at 72 h after the initiation of cocultures. Results are shown for one representative experiment of two that incorporated IL-6 blockade. (B) Supernatants were harvested at 72 h and assayed for cytokines by sandwich ELISA for IL-6. *, P ≤ 0.05 by the Student t test in a pairwise comparison of each cell concentration. Results shown are from one representative experiment of four of a similar design.

DISCUSSION

In this study, we found that IL-6 is critical for resistance to primary LVS infection during either intradermal or intranasal LVS infection. IL-6 KO mice, and also wild-type mice treated with anti-IL-6 blocking antibodies, infected with LVS exhibited increased bacterial organ burdens and substantial bacteremia; ultimately, LVS infection results in death of IL-6 KO mice at much lower doses than for LVS-infected WT mice. An extensive search for a definitive mechanism by which IL-6 participates has not led to clear conclusions, however. The dramatic differences in survival were not reflected by obvious differences in blood chemistry, hematology, production of cytokines in infected tissues, or changes in proportions of splenic or blood leukocyte populations. While we did see an increase in AST/ALT levels, which may suggest liver distress, there were no significant differences in other markers of liver injury, such as creatinine kinase and lactate dehydrogenase (LDH), and there were no obvious differences in liver pathology. None of the lesions observed during pathological studies and necropsies were considered severe or extensive enough to result in peracute mortality. Thus, the specific cause of death of LVS-infected IL-6 KO mice was not established by either assessment of immunological or physiological parameters, nor by pathological studies. In fact, there is little understanding of the cause of death in WT mice lethally infected with LVS or virulent Francisella (27). The significance of the observations that G-CSF was increased in all organs of LVS-infected IL-6 KO mice compared to WT mice and that IL-23 and amphiregulin were differentially expressed remains to be determined, but future studies of these molecules may be helpful. Subtle differences in immunological responses, such as slightly increased levels of anti-LVS IgM serum antibodies and decreases in LVS-specific T cell functions, were detected. These differences were modest, however, and did not seem sufficient to explain the increased susceptibility of LVS-infected IL-6 KO mice. Nor did these difference impact the survival of LVS-vaccinated IL-6 KO mice upon secondary lethal challenge administered either i.p. or i.n.

LVS-infected IL-6 KO mice had decreased levels of serum amyloid A and haptoglobin compared to LVS-infected WT mice (Fig. 6). These proteins are part of the acute-phase response, a set of early physiological reactions to injury, trauma, or infection (29, 39, 40). Upon activation, several acute-phase proteins are rapidly produced, which in turn can modulate other physiological responses. While decreased levels of SAA and haptoglobin were observed in LVS-infected IL-6 KO mice, there were no differences in levels of TNF-α (Fig. 4), fibrinogen, PGE₂, or KC/CXCL1 (data not shown). Thus, the acute-phase response is not globally disrupted. Further studies will be necessary to examine potential roles of SAA or haptoglobin. SAA is an apolipoprotein that can bind high-density lipoprotein (HDL), a relationship that has linked SAA to atherosclerosis (41). SAA has also been shown to directly bind to the outer membrane protein A of some, but not all, Gram-negative bacteria (42), but whether this molecule binds to that of Francisella is unknown. Further, the exact function of SAA and the role that it may play in protecting a host against a bacterial pathogen is not clear. SAA is a chemoattractant for neutrophils, and by itself it can also induce G-CSF to promote neutrophilia (43). Here, we did not see evidence of dysregulated neutrophil responses in blood or in splenic homogenates in LVS-infected IL-6 KO mice (Tables 2 and 3 and data not shown). Haptoglobin is a glycoprotein that can bind free hemoglobin (Hb), preventing excess Hb from causing renal damage (44). More directly, haptoglobin may be bacteriostatic by binding free hemoglobin and sequestering it, thus limiting iron sources from the bacteria (45). However, whether or not haptoglobin is bacteriostatic for LVS awaits future study.

LVS-infected IL-6 KO mice had mildly increased LDH, a low hematocrit, and significantly decreased platelets (Tables 1 and 2), which may be indicators of hemolysis. LVS-infected IL-6 KO mice also had increased bacteremia (Fig. 5). Whether or not increased bacteremia and the mild hemolysis observed are symptoms of morbidity, or are a contributing cause of the susceptibility of the IL-6 KO mice, remains to be determined.

Of note, the difference in LD₅₀s for intranasal infection between WT and IL-6 KO mice was not as pronounced as it is for intradermal infection (Fig. 2). Whether this simply reflects a smaller possible range of differences due to the lower LD₅₀ for intranasal LVS infection, or is also related to mechanistic differences in the role of IL-6 during intranasal versus intradermal LVS infection is unclear. There are ample precedents for mechanistic differences between systemic and mucosal immune responses. An example has already been described for Francisella, where neutrophil or IFN-γ depletion reduced survival of i.d. F. tularensis LVS infection but not aerosol LVS infection (46). Similarly, qualitatively and quantitatively different immune responses have been reported for animals immunized by intranasal versus intradermal or enteric routes against respiratory syncytial virus (RSV) (47).

Although IL-6 has been linked to resistance to several other pathogens, in the other examples the mechanisms also remain incompletely understood. For Chlamydia trachomatis, the importance of IL-6 is infection route dependent. In a murine model of pneumonitis, IL-6 was important for inhibiting C. trachomatis growth in the lungs and improving survival in infected mice (19). In contrast, a model of C. trachomatis infection in the murine female genital tract showed IL-6 to be dispensable for immunity (48). The role of IL-6 in protection against Mycobacterium tuberculosis is also complex and may be route dependent. IL-6-deficient mice had increased mortality and bacterial organ burdens when given M. tuberculosis intravenously, a phenotype accompanied by a dramatic shift in proinflammatory cytokine production (49). However, blocking IL-6 with an anti-IL-6 receptor antibody did not yield the same increased susceptibility to intravenous M. tuberculosis (50). In a low-dose aerosol infection model of M. tuberculosis, IL-6 KO mice had a moderate increase in CFU in lungs, but not in spleens or livers, and this was accompanied by a decrease in IFN-γ production (51). Here, IL-6 KO mice had increased susceptibility to both intradermal and intranasal LVS infection and no detectable changes in systemic levels of IFN-γ.

IL-6 KO mice infected with Listeria monocytogenes exhibited a 100- to 1,000-fold increase of CFU in organs in infected IL-6 KO mice and a higher mortality rate than for WT mice (18, 29). Listeria-infected IL-6 KO mice had decreased levels of neutrophils in the blood, although there was no inherent defect in number of neutrophils in the blood of naïve IL-6 KO mice. Treatment with recombinant IL-6 (rIL-6) did not rescue neutrophil-depleted, L. monocytogenes-infected IL-6 KO mice, suggesting that the role of IL-6 depends on neutrophils (18). Similar results were seen for Candida albicans, where there was also a defect in blood neutrophil counts during infection in IL-6 KO mice (52). As with Listeria, adding rIL-6 did not rescue neutrophil-depleted IL-6 KO mice. In this study, we measured the levels of neutrophils and other immune cell populations in the spleens and blood of LVS-infected WT and IL-6 KO mice by using both multiparameter flow cytometry and hematological analysis methods (Table 2 and data not shown). However, we did not find any consistent significant differences in any of the cell populations analyzed, including neutrophils.

Increased susceptibility of IL-6 KO mice to infection has also been linked to dysregulation of other cytokines. In the case of C. albicans, excess IL-10 in IL-6 KO mice appears to be directly related to the increased sensitivity (52). Similarly, one of the functions of IL-6 in protection against Yersinia enterocolitica is to modulate inflammatory cytokines, including TNF-α and IL-10 (21). However, in our infection model, we found no substantive differences in the levels of these or many other inflammatory cytokines or chemokines in LVS-infected IL-6 KO mice compared to WT mice, in either infected tissues or in sera (Fig. 4 and data not shown).

In addition to cytokine dysregulation, infection of IL-6 KO mice with Y. enterocolitica, an enteric pathogen, resulted in increased inflammation in tissues, gastrointestinal (GI) bleeding, and intestinal adhesions (21). Necropsy of IL-6 KO mice infected with LVS also revealed intriguing histopathological findings in the GI tract and pancreas, although LVS is not typically considered an enteric pathogen. Future studies will investigate whether these changes in pathology in IL-6 KO mice, which were not deemed substantial enough to directly contribute to morbidity, are nonetheless linked to the susceptibility of IL-6 KO mice.

IL-6 is perhaps best known for its important roles in inducing B cell differentiation and antibody production (14). However, following immunization with LVS, we found somewhat increased levels of anti-LVS IgM and similar levels of IgG and the IgG subclasses (Fig. 5 and data not shown) in WT and IL-6 KO mice. We also found similar levels of B cells in the spleens of infected WT and IL-6 KO mice (Table 3). B cells are an important part of immunity to F. tularensis (32, 53, 54); although only serum antibodies were measured here, an intact or even increased B cell response in some compartments may compensate for a slightly impaired T cell response, thus contributing to the overall resistance of LVS-primed IL-6 KO mice to secondary challenge.

IL-6 has been shown to play a role in priming T cell growth and differentiation (reviewed by Kishimoto (14). Indeed, we found that enriched T cells from LVS-immune IL-6 KO mice were less able to control intramacrophage LVS infection than were those derived from LVS-primed WT mice (Fig. 6 and 7), suggesting that IL-6 is important for optimal priming, expansion, or maintenance. Notably, the decreased functions of LVS-primed IL-6 KO mice were not readily attributable to differences in production of IFN-γ, nitric oxide, or any of the other cytokines tested.

Nonetheless, substantial frequencies of LVS-specific splenocytes, and T cells in particular, developed in the absence of IL-6 during LVS infection, and this deficiency was not sufficient to impact protection against maximal secondary challenge (Table 4). Similarly, IL-6 was dispensable for the priming of IFN-γ secretion in splenocytes of C. albicans-primed mice (18). In the latter study, exogenous rIL-6 treatment protected C. albicans-infected Rag-2 KO mice, indicating that the role of IL-6 depends on myeloid cells but is not related to its influence on lymphocytes.

IL-6 remains a molecule of interest because the relative expression of the IL-6 gene in splenocytes from Francisella-vaccinated mice was found to correlate with the degree of protection against LVS lethal challenge (9). These results implied that IL-6 may be important during adaptive immunity and contribute directly to the mechanism of protection. Ideally, the final selection of predictive correlates would include those with direct mechanistic relevance, as these are most likely to be robust across different vaccination circumstances. However, our data indicate that IL-6 is dispensable for the acquisition of a protective secondary response, as LVS-vaccinated IL-6 KO mice survived maximal lethal secondary challenges with LVS (Table 4), or with the virulent type A F. tularensis SchuS4, just as well as WT mice. There are several possible interpretations of the apparent discrepancy between the identification of IL-6 as a predictive correlate after vaccination and its apparent lack of a role during secondary immune responses in this study. First, although we are unaware of gross anatomical or physiological differences between IL-6 KO mice and WT mice, it is possible that such differences, which are not representative of normal physiology, obscure differences directly related to the absence of IL-6 itself. Second, IL-6 may not be an important effector for adaptive immunity to LVS, and it has been shown to be differentially regulated in studies using differential vaccination to search for protective correlates because it is linked to, or coregulated with, another molecule that is directly important for adaptive immunity to LVS. However, IL-6 was clearly important during primary challenge, which obviously argues for an important if unexpected mechanistic role for IL-6 during early responses to Francisella. Third, and we believe likely, there may be other immune moieties with overlapping roles, and such molecules may compensate for the absence of IL-6 during the adaptive response. Collectively, the present study both highlights a critical contribution of IL-6 during early responses to LVS that is not obviously linked to its known functions and emphasizes the need for a mechanistic understanding of the specific role of potential correlates in order to better consider how best to apply them to predict successful vaccination.