Myeloid Src kinases regulate phagocytosis and oxidative burst in pneumococcal meningitis by activating NADPH oxidase

Abstract

Myeloid cells, including neutrophils and macrophages, play important roles in innate immune defense against acute bacterial infections. Myeloid Src family kinases (SFKs) p59/61^hck (Hck), p58^c-fgr (Fgr), and p53/56^lyn (Lyn) are known to control integrin β₂ signal transduction and FcγR-mediated phagocytosis in leukocytes. In this study, we show that leukocyte recruitment into the cerebrospinal fluid space and bacterial clearance is hampered in mice deficient in all three myeloid SFKs (hck^−/−fgr^−/−lyn^−/−) during pneumococcal meningitis. As a result, the hck^−/−fgr^−/−lyn^−/− mice developed increased intracranial pressure and a worse clinical outcome (increased neurologic deficits and mortality) compared with wild-type mice. Impaired bacterial killing was associated with a lack of phagocytosis and superoxide production in triple knockout neutrophils. Moreover, in hck^−/−fgr^−/−lyn^−/− neutrophils, phosphorylation of p40^phox was absent in response to pneumococcal stimulation, indicating a defect in NAPDH oxidase activation. Mice lacking the complement receptor 3 (CR3; CD11b/CD18), which belongs to the β₂-integrin family, also displayed impaired host defense against pneumococci, along with defective neutrophil superoxide production, but cerebrospinal fluid pleocytosis was normal. Cerebral expression of cytokines and chemokines was not decreased in both mouse strains, indicating that CR3 and myeloid SFKs are dispensable for the production of inflammatory mediators. Thus, our study demonstrates the pivotal role of myeloid SFKs and CR3 in mounting an effective defense against CNS infection with Streptococcus pneumonia by regulating phagocytosis and NADPH oxidase-dependent superoxide production. These data support the role of SFKs as critical mediators of CR3 signal transduction in host defense.

INTRODUCTION

Src family kinases (SFKs) have been identified to be essential in regulating vascular permeability as well as leukocyte function [1]. Myeloid cells such as neutrophils and macrophages play a pivotal role in the innate immune defense against pathogens. After activation, leukocytes adhere to the vasculature and migrate toward the inflammatory stimulus. At the site of infection, pathogens are cleared by phagocytic uptake into leukocytes, which stimulates release of reactive oxygen species (ROS) leading to bacterial killing.

Myeloid leukocytes and microglial cells express three SFK members: p59/61^hck (Hck), p58^c-fgr (Fgr), and p53/56^lyn (Lyn) [2, 3]. In vitro studies have shown that hck^−/−fgr^−/−lyn^−/− macrophages have impaired FcγR-mediated phagocytosis and respiratory burst but normal internalization of complement-opsonized particles [4]. Likewise, hck^−/−fgr^−/−lyn^−/− macrophages have normal production of NO and TNF-α and even an increased secretion of IL-1 and IL-6 in response to LPS [5]. Moreover, migration of macrophages is impaired in hck^−/−fgr^−/−lyn^−/− mice challenged i.p. with thioglycolate [6]. Like macrophages, hck^−/−fgr^−/−lyn^−/− neutrophils failed to undergo respiratory burst after adhesion-mediated activation [7]. However, no defect in migration of hck^−/−fgr^−/−lyn^−/− neutrophils has been demonstrated in vitro, in standard chemotaxis models, or in vivo using a peritonitis model [8]. Deficiency of myeloid SFKs also aborts development of hemorrhagic vasculitis in the Shwartzman model, without affecting the migratory ability of kinase-deficient neutrophils, by blocking signaling through complement receptor 3 {CR3; CD11b/CD18, α_Μβ₂, membrance-activated complex 1 (Mac-1) [9]}. Despite the plethora of in vitro studies, the knowledge about the role of myeloid Src kinases in mediating the inflammatory response to infections in vivo is limited. In one study, it was shown that hck^−/−fgr^−/− mice challenged with the protozoan Leishmania major were able to resolve the infection, and the same double-mutant mice failed to eliminate the bacterial pathogen Listeria monocytogenes, indicating a marked defect in the innate immune response to this organism [10]. However, investigations in clinically relevant disease models are still missing.

Meningitis, caused by Streptococcus pneumoniae, is the most common bacteria-induced CNS inflammation in adults and is associated with high mortality and morbidity [11]. Bacterial meningitis is an acute, infectious disease of the CNS, which is characterized by a predominantly neutrophilic, inflammatory response. Clinical and neuropathologic studies have demonstrated that the cerebral inflammation causes intracranial complications such as brain edema, increased intracranial pressure (ICP), and cerebrovascular insults, which often result in a fatal outcome [11].

Cells of the mononuclear phagocyte system and polymorphonuclear neutrophils (PMNs) mediate the innate immune response in pneumococcal infections [12]. Phagocytosis of the bacterium is mainly mediated by complement, in particular, the interaction of C1q and C3 with CR3 [13]. Neutrophil activation during phagocytosis of pneumococci leads to respiratory burst and production of superoxides (ROS) by the NADPH oxidase of phagocytic cells [14]. After activation, neutrophils, monocytes, and macrophages secret chemokines (e.g., CXCL-2) and cytokines (e.g., IL-1β, IL-6, and IL-12), which further recruit leukocytes into the brain and boost the inflammatory reaction [15].

It has been shown that p60^c-Src (c-Src) does not play a pivotal role in the pathophysiology of pneumocoocal meningitis [16]; however, the relevance of the myeloid Src kinases remains unknown. Using mice deficient in all three myeloid SFKs (hck^−/−fgr^−/−lyn^−/−), we show in this study for the first time that these animals have a defective host defense against S. pneumoniae as a result of reduced leukocyte recruitment, impaired phagocytosis, and blocked activation of NADPH oxidase during acute bacterial meningitis. Moreover, we demonstrate that deficiency in CD11b, which is upstream of myeloid SFKs, also causes impairment of phagocytosis in pneumococcal meningitis, and leukocyte recruitment, in contrast to hck^−/−fgr^−/−lyn^−/− mice, is unaffected in CD11b^−/− mice.

MATERIALS AND METHODS

Mouse model of pneumococcal meningitis

All studies were approved by the government of Upper Bavaria. A well-characterized mouse model of pneumococcal meningitis was used, as described earlier [17]. In brief, meningitis was induced by transcutaneous injection of 15 μl 10⁷ cfu/ml S. pneumoniae type 3 into the cisterna magna. Twenty-four hours after infection, mice were clinically evaluated and anesthetized with ketamine/xylazine. Physiological (temperature, weight, tremor, seizure, vigilance, pilorection) and motor (beam-balancing, postural reflex, paper-crunching) parameters were assessed by a clinical score, reaching from 0 (no clinical deficits) to 20 points (for detailed description, see Malipiero et al. [18]). A catheter was inserted into the cisterna magna to measure ICP and to determine cerebrospinal fluid white blood cell (WBC) counts. Thereafter, mice were deeply anesthetized and perfused with ice-cold PBS. The cerebrum was removed and rapidly frozen.

To determine cerebellar bacterial titers, the cerebellum was removed immediately after the mice were killed; it was homogenized in sterile saline, and serial dilutions were plated on blood agar plates. Likewise, blood samples were drawn at the end of the experiments and plated on blood agar plates. Only typical pneumococcal cultures were observed on the plates.

Mouse cytokine antibody array

For determination of cytokine expression, a custom-made antibody array containing 15 different inflammatory cytokines/chemokines (IL-1, IL-4, IL-6, IL-12p40, CCL-2, CCL-12, CXCL-1, CXCL-4, CXCL-9, CXCL-16, Axl, L-selectin, M-CSF, G-CSF, stem cell factor) was used (Raybiotech, tebu-bio, Offenburg, Germany) according to the manufacturer’s instructions. Briefly, frozen brain sections were homogenized in sample buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 1.5 mM MgCl₂, and protease inhibitors). The homogenates were centrifuged, and 50 μl of the supernatant was used for each analysis. Additionally, the protein concentration of the supernatant was measured using the Nanoquant assay (Carl Roth GmbH, Karlsruhe, Germany). Protein samples of all experiments within one group were pooled. Membranes were blocked and incubated with 3000 μg brain tissue homogenate and incubated with a biotin-conjugated antibody mix. After consecutive washes, streptavidin-conjugated peroxidase was added. Subsequently, they were washed, incubated with peroxidase substrate, exposed to an X-ray film, and analyzed using imaging software. Relative expression levels of cytokines were determined by comparing signal intensities normalized to positive control spots. If the OD of a protein was below the detection limit (OD <5% of positive control) in unchallenged controls but not in infected/traumatized animals, the protein was considered to be “induced.” If the OD of a protein was above the detection limit in unchallenged controls, and its ratio of infected versus uninfected and sham-operated versus traumatized was >2, the protein was considered to be “up-regulated”; if the ratio was <0.5, the protein was considered to be “down-regulated.”

Immunoassay for murine IL-1β, CXCL-2, and L-selectin

Immunoreactive IL-1β, CXCL-2 (MIP-2), and L-selectin were determined using commercially available ELISA kits (Quantikine Assay kits, R&D Systems GmbH, Wiesbaden-Nordenstadt, Germany). Briefly, frozen brain sections were homogenized in sample buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 1.5 mM MgCl₂, and protease inhibitors). The homogenates were centrifuged, and 50 μl of the supernatant was used for each determination. Additionally, the protein concentration of the supernatant was measured using the Nanoquant assay (Carl Roth GmbH). Protein concentrations were expressed as pg/mg total protein.

RT-PCR analysis of C1r and C3

Total RNA was extracted from frozen brain sections with TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) and reverse-transcribed using Superscript II (Invitrogen Life Technologies). The cDNA was amplified by PCR with gene-specific primers (MWG Biotech, Ireland) of the following sequences: C3 sense, 5′-CACCGCCAAGAATCGCTAC-3′, C3 antisense, 5′-GATCAGGTGTTTCAGCCGC-3′; C1r sense, 5′-CTTCCGCTACATCACCAC-3′, C1r antisense, 5′-GCTAACTTATCTTCTGTTA-3′. Mouse β-actin was coamplified as an internal control using the following primer sequences: sense, 5′-GGACTCCTATGTGGGTGACGAGG-3′, antisense, 5′-GGGAGAGCATAGCCCTCGTAGAT-3′. Linearity of DNA amplification was determined for each primer set in experiments establishing the PCR procedures in terms of cDNA amounts and number of cycles applied. PCR products were separated on a 1.5% agarose gel, stained with ethidium bromide, visualized by UV illumination, and photographed. Densitometry was performed on the negative image, and the relative absorbances of C3 and C1r were normalized by relation to absorbance of β-actin RT-PCR products.

Oxidative burst and p40^phox phosphorylation

Induction of neutrophil oxidative burst by serum-coated S. pneumoniae was performed as described [19]. Bone marrow neutrophils from hck^−/−fgr^−/−lyn^−/−, CD11b^−/−, and CD18^−/− mice were purified as described [8] and incubated in microtiter plates (5×10⁵ cells/well) with serum-opsonized S. pneumoniae at a multiplicity of infection (MOI) of 40. Mouse serum was obtained from recombination activating gene 1 (RAG1)-deficient mice and hence, was devoid of antibodies, thus allowing us to avoid engagment of FcRs by opsonized bacteria. Oxidative burst was assessed by reduction of cytochrome C [19] over the indicated time course.

Phosphorylation of p40^phox was determined by Western blotting using a polyclonal antibody that recognizes phospho-threonine (Thr154; Cell Signaling Technology, Beverly, MA, USA). Bone marrow-purified neutrophils were prestimulated with TNF-α (50 ng/ml) or IFN-γ (1 ng/ml) for 30 min at 37°C and then incubated with serum-opsonized S. pneumoniae at a MOI of 40 for 45 min at 37°C. Cell lysates were prepared for Western blotting as described [8]. As a positive control, some cell samples were stimulated with PMA at 50 nM for 45 min, and lysates were prepared.

Experimental groups

C57BL/6 mice were injected i.c. with PBS (n=4); C57BL/6 mice were injected i.c. with S. pneumoniae (n=12); and hck^−/−fgr^−/−lyn^−/− mice were injected i.c. with S. pneumoniae (n=12). Gene-targeted mice were generated by Meng and Lowell [5], as described previously, and then backcrossed for 11 generations into a C57BL/6 background. C57BL/6 mice were injected i.c. with S. pneumoniae (n=9), and CD11b^−/− mice were injected i.c. with S. pneumoniae (n=8). J. Leusen (University Medical Center Utrecht, the Netherlands) kindly provided CD11b^−/− [20] mice. The government of Upper Bavaria (Germany) approved all of the experiments.

Statistical analysis

All values are expressed as mean ± sd. Data sets were compared using the unpaired Student’s t-test. Differences were considered significant at P < 0.05.

RESULTS

Deficiency in Hck, Fgr, and Lyn modulates pathophysiologic alterations during pneumococcal meningitis

In wild-type (wt) mice, meningitis caused an increase in cerebrospinal fluid WBC counts (10,016±7953/μL) and ICP (13.7±4.5 mm Hg), which were significantly higher compared with PBS-injected controls (29±14/μL and 2.8±0.8 mm Hg, respectively; Fig. 1A). Meningitis was associated with a poor clinical outcome, as determined by the increase in clinical score (7.8±2.3 vs. 0.0±0.0 in controls; P<0.05; Fig. 1C). None of the wt mice died within the first 24 h after infection.

Fig. 1.

Effect of Hck, Fgr, and Lyn deficiency on pathophysiologic alterations during pneumococcal meningitis. wt or hck^−/−fgr^−/−lyn^−/− mice were injected i.c. with PBS (wt control, n=4) or S. pneumoniae (labeled, infected, n=12 each group), using 1.5 × 10⁵ cfu, and then monitored for 24 h. After that time, animals were clinically evaluated (C) and anesthetized to measure ICP (A) and to determine cerebrospinal fluid (CSF) WBC counts (B). *, P < 0.05, compared with uninfected controls; ^#, P < 0.05, compared with infected wt mice.

Meningitis in hck^−/−fgr^−/−lyn^−/− mice caused an increase in cerebrospinal fluid WBC counts, which was significantly lower (4909±4083/μL; P<0.05) compared with infected wt mice (Fig. 1B). In contrast, ICP was significantly higher in hck^−/−fgr^−/−lyn^−/− mice (18.4±4.0 mm Hg; P<0.05) compared with infected controls. (Fig. 1A). Lack of myeloid Src kinases was associated with a worse clinical score (16.3±2.0; P>0.05) and higher mortality rate (two of 12) compared with infected wt mice after 24 h (Fig. 1C).

Bacterial clearance is impaired in neutrophils lacking Hck, Fgr, and Lyn

In infected wt mice, cerebellar S. pneumoniae titers were log 5.96 ± 0.77 cfu/organ, and blood titers were log 2.94 ± 0.60 cfu/ml 24 h after infection (Fig. 2, A and B). No bacteria could be detected in the brain or blood of PBS-injected mice (not shown). Deficiency in Hck, Fgr, and Lyn significantly impaired bacterial clearance in the brain and peripheral blood (Fig. 2, A and B). Cerebellar and blood S. pneumoniae titers were increased significantly compared with infected wt mice (log 7.90±0.44 cfu/organ and log 5.01±0.47 cfu/ml, respectively; P<0.05; Fig. 2, A and B).

Fig. 2.

Effect of Hck, Fgr, and Lyn deficiency on bacterial clearance. S. pneumoniae bacterial titers in cerebellar tissue (A) and peripheral blood (B) were determined as described in Materials and Methods from infected mice shown in Figure 1. *P < 0.05, compared with infected wt mice. Samples of cerebrospinal fluid were also gram-stained to observe bacterial loads directly in wt (C) and hck^−/−fgr^−/−lyn^−/− mice (D). Wright-Giemsa stains of isolated/washed neutrophils from wt (E) and hck^−/−fgr^−/−lyn^−/− mice (F) are shown. White arrows show intracellular S. pneumoniae, present in wt but not hck^−/−fgr^−/−lyn^−/− leukocytes; black arrows show cell surface-bound S. pneumoniae, present in both neutrophil types. Images are representative of 10 different fields from five different mice per cohort.

To begin to determine the mechanism of the impaired host defense present in the hck^−/−fgr^−/−lyn^−/− mice, we stained cerebrospinal fluid smears from the infected animals. Although many segmented neutrophils and only a few pneumococci could be seen in the cerebrospinal fluid of infected wt mice (Fig. 2C), many bacteria among a few leukocytes were detected in the cerebrospinal fluid of hck^−/−fgr^−/−lyn^−/− animals (Fig. 2D). Wright-Giemsa-stained cerebrospinal fluid neutrophils from wt mice contained significant numbers of obvious S. pneumoniae organisms attached to the cell surface and intracellularly (Fig. 2E). Surprisingly, the hck^−/−fgr^−/−lyn^−/− neutrophils had predominately only cell surface-bound S. pneumoniae, and few organisms were taken up into the mutant neutrophils (Fig. 2F). This suggests that phagocytosis of the serum-opsonized S. pneumoniae was defective in Hck/Fgr/Lyn-deficient neutrophils.

Cerebral up-regulation of G-CSF, CXCL-1, and CXCL-2 in hck^−/−fgr^−/−lyn^−/− mice with bacterial meningitis

Using an antibody array, we compared the expression of 15 different cytokines and chemokines in the brains of infected wt and hck^−/−fgr^−/−lyn^−/− mice. In accordance with previous studies [21], we detected induction/up-regulation of IL-6, IL-12, CCL-2, CCL-12, CXCL-1, CXCL-16, L-selectin, and G-CSF in the brains of infected wt mice, and the remaining seven proteins were unchanged in this assay (data not shown). In brains of infected, triple knockout (KO) mice, cytokine/chemokine expression remained strong, with a further increase in expression of G-CSF and CXCL-1, and L-selectin expression was reduced (Fig. 3A). Using immunoassays, we confirmed the expression of CXCL-2, IL-1β, and L-selectin. CXCL-2 was up-regulated in hck^−/−fgr^−/−lyn^−/− mice (70±12 pg/mg compared with 33±24 pg/mg in infected wt mice; P<0.05), and IL-1β expression was unchanged in KO mice compared with wt mice (32±12 pg/mg compared with 29±19 pg/mg; P=not significant; Fig. 3, B and C). L-selectin was reduced by 44% in triple KO mice (0.86±0.43 pg/mg compared with 1.54±0.81 pg/mg in infected wt mice; P=0.0525), which reflects the reduced cerebrospinal fluid leukocyte count in the hck^−/−fgr^−/−lyn^−/− animals (Fig. 3D). These data indicate that despite the low cerebrospinal fluid pleocytosis in hck^−/−fgr^−/−lyn^−/− mutants, these mice are able to mount a normal or exaggerated production of cytokines/chemokines following intracranial S. pneumoniae infection.

Fig. 3.

Cerebral up-regulation of CXCL-1, G-CSF, and CXCL-2 in hck^−/−fgr^−/−lyn^−/− (HFL^−/−) mice with bacterial meningitis. Cerebrospinal fluid from infected wt and hck^−/−fgr^−/−lyn^−/− mice from Figure 1 were assayed for a number of chemokines or cytokines by array (A) or ELISA (B–D), as described in Materials and Methods. Representative array results demonstrating increased expression of CXCL-1 and G-CSF, but lower levels of soluble L-selectin, from mutant mice are shown (POS, positive control). ELISA results are averaged from all mice in each cohort (see Fig. 1). *, P < 0.05, compared with uninfected controls; ^#, P < 0.05, compared with infected wt mice.

Deficiency in CD11b also results in impaired host defense to intracranial S. pneumoniae infection

As loss of myeloid SFKs has been demonstrated to cause decreased intracellular signaling from CR3 (Mac-1 or CD11b) [9], we asked whether mice lacking this leukocyte receptor also manifested a defect in resistance to S. pneumoniae meningitis. In this experimental cohort, meningitis in wt control mice caused an increase in cerebrospinal fluid WBC counts (10,740±8409/μL) and ICP (12.6±6.3 mm Hg), which was significantly higher compared with PBS-injected controls (29±14/μL and 2.8±0.8 mm Hg, respectively), along with a correspondingly poor clinical outcome (8.6±2.5 vs. 0.0±0.0 in controls; Fig. 4, A–C). None of the wt mice died within the first 24 h after infection. Like the hck^−/−fgr^−/−lyn^−/− mice (Fig. 1C), the CD11b^−/− animals displayed a significantly worse clinical outcome (13.5±5.0) and higher mortality rate (one of eight) compared with infected wt mice after 24 h (Fig. 4C). However, no significant differences in ICP or cerebrospinal fluid pleocytosis between CD11b^−/− (10.6±6.4 mm Hg and 6787±5452/μL, respectively) and wt mice were present, although the cerebrospinal fluid count trended toward the lower level in the mutant mice (Fig. 4, A and B).

Fig. 4.

Effect of CD11b deficiency on pathophysiologic alterations during pneumococcal meningitis. wt or CD11b^−/− mice were injected i.c. with PBS (wt control, n=4) or S. pneumoniae (labeled, infected: wt, n=9; CD11b^−/−, n=8), using 1.5 × 10⁵ cfu and then monitored for 24 h. After that time, animals were clinically evaluated (C) and anesthetized to measure ICP (A) and to determine cerebrospinal fluid WBC counts (B). *, P < 0.05, compared with uninfected controls; ^#, P < 0.05, compared with infected wt mice.

Consistent with their poor clinical outcome, the CD11b^−/− mice manifested significantly higher S. pneumoniae titers in the cerebellum and in peripheral blood compared with wt mice 24 h after infection (CD11b^−/−=log 7.40±0.48 cfu/organ and log 4.53±0.73 cfu/ml; wt=5.96±1.30 cfu/organ and log 3.57±1.1 cfu/ml, respectively; Fig. 6, A and B), indicating defective phagocytosis in CD11b^−/− cerebrospinal fluid neutrophils. Consistent with this, the CD11b^−/− mice showed no significant differences in cerebrospinal fluid CXCL-2 or L-selectin levels compared with wt mice (Fig. 6, C and D).

Fig. 6.

Impaired bacterial clearance but normal cytokine expression in CD11b^−/− mice. S. pneumoniae bacterial titers in cerebellar tissue (A) and peripheral blood (B) were determined, as described in Materials and Methods, from infected mice shown in Figure 4. *, P < 0.05, compared with infected wt mice. Cerebrospinal fluid CXCL-2 (C) and L-selectin (D) were determined by ELISA, as described in Materials and Methods, from mice shown in Figure 4. *, P < 0.05, compared with uninfected controls.

Overall, these data indicate that like the hck^−/−fgr^−/−lyn^−/− mice, loss of CR3 present in CD11b^−/− mice results in impaired host defense to S. pneumoniae meningitis.

Superoxide production is absent in neutrophils lacking myeloid Src kinases, CD11b, and CD18

As SFKs are known to be required for signal transduction initiated by engagement of CD11b, CD18, and other leukocyte integrins following stimulation by extracellular matrix proteins [7], we determined whether ligation of these surface receptors with serum-opsonized S. pneumoniae would also induce neutrophil activation. For this experiment, we compared superoxide (ROS) production in wt, hck^−/−fgr^−/−lyn^−/−, CD11b^−/−, and CD18^−/− neutrophils treated with serum-opsonized S. pneumoniae using a cytochrome C reduction assay. As shown in Figure 5A, loss of SFKs in neutrophils completely blocked ROS production following bacterial challenge, similar to the profound defect seen following adhesion to fibrinogen or fibronectin [22]. CD18^−/− mutant neutrophils (which obviously lack multiple integrin adhesion receptors) were also profoundly impaired (Fig. 5A), but consistent with their milder impairment of host defense in vivo, the CD11b^−/− neutrophils showed a significant, but not complete, block in ROS production following bacterial challenge (Fig. 5A).

Fig. 5.

Absent ROS production in hck^−/−fgr^−/−lyn^−/− neutrophils is caused by impaired NADPH oxidase activation. (A). Superoxide release of wt, CD11b^−/−, CD18^−/−, and hck^−/−fgr^−/−lyn^−/− neutrophils stimulated with serum-opsonized S. pneumonia (MOI=40) was performed as described in Materials and Methods; error bars, sd of triplicate measurements. Western blot analysis of p40phox phosphorylation on Thr154 in response to (Lane 1) no stimulation; (Lane 2) incubation for 30 min at 37°C with 50 ng/mL TNF-α (upper gel) or 1 ng/mL IFN-γ (lower gel), followed by 45 min at 37°C; (Lane 3) incubation for 30 min at 37°C with 50 ng/mL TNF-α (upper gel) or 1 ng/mL IFN-γ (lower gel), followed by incubation with S. pneumonia (MOI=40) for 45 min 37°C; or (Lane 4) 50 nM PMA for 45 min at 37°C.

Activation of NADPH oxidase is impaired in hck^−/−fgr^−/−lyn^−/− neutrophils

To test whether reduced ROS production is caused by impaired NADPH oxidase activation, we investigated phosphorylation of p40^phox in response to S. pneumoniae challenge. Neutrophils were primed with TNF-α or IFN-γ, followed by incubation with serum-opsonized bacteria. In wt leukocytes, phosphorylation on Thr154, indicating NADPH oxidase activation, was detectable 45 min after stimulation with serum-opsonized bacteria in the presence of TNF-α (Fig. 5B, WT PMN, Lane 3). In contrast, p40^phox phosphorylation was absent after stimulation with bacteria in neutrophils lacking Hck, Fgr, and Lyn (HFL^−/− PMN, Lane 3). Similar results were seen in neutrophils primed by incubation with IFN-γ (Fig. 5B, lower). In contrast, wt and mutant neutrophils showed robust phosphorylation of p40^phox when stimulated by phorbol ester (upper and lower, Lanes 4). These results indicate the loss of myeloid SFKs results in a block in p40^phox phosphorylation, thus impairing NADPH oxidase activation and ROS production following challenge by serum-opsonized S. pneumoniae. Given the impaired ROS production in CR3 (CD11b)-deficient mice, it is likely that SFKs are acting downstream of this receptor following binding of serum-opsonized pneumoccoci.

Expression of cerebral complement is unaltered in hck^−/−fgr^−/−lyn^−/− mice

As shown in a recent study, expression of complement factors C1r and C3 is normally absent in brains of uninfected mice but is induced during pneumococcal meningitis [23]. Likewise, we found increased cerebral expression of C1r and C3 mRNA in infected control mice (data not shown). No difference in complement expression was found in infected hck^−/−fgr^−/−lyn^−/− mice compared with infected wt mice, indicating that impaired bacterial killing was not a result of complement deficiency (Fig. 7, A and B).

Fig. 7.

Complement expression is unchanged in hck^−/−fgr^−/−lyn^−/− mice. mRNA levels for the complement factors C1r (A) and C3 (B) were determined in the brains of infected wt versus hck^−/−fgr^−/−lyn^−/− mice, obtained from Figure 1, as described in Materials and Methods.

DISCUSSION

We have recently reported that c-Src does not influence brain edema formation and leukocyte function during bacterial meningitis [16]. In this study, we showed that the myeloid Src kinases Hck, Fgr, and Lyn, in fact, play a pivotal role in the effective defense against CNS infection with S. pneumonia.

We found that the cerebrospinal fluid leukocyte count was reduced significantly in infected hck^−/−fgr^−/−lyn^−/− mice compared with uninfected wt mice (Fig. 1B). This effect is attributed, at least in part, to impaired β₂ integrin-mediated signaling in triple KO myeloid leukocytes. Integrin receptor β₂ (CD18) has been shown to be a major determinant of leukocyte adhesion in bacterial meningitis. Tuomanen et al. [24] showed that application of an anti-CD18 antibody effectively blocked the development of leukocytosis in the cerebrospinal fluid in a rabbit model of bacterial meningitis, which was accompanied by a decreased blood-brain barrier (BBB) injury. Similar results were obtained using an antibody against the endothelial counterpart of integrin β₂ complexes, ICAM-1 (CD54), in a rat model of pneumococcal meningitis [25]. Src kinases and β₂ integrins, together with the urokinase plasminogen activator receptor (uPAR; CD87) establish a membrane-bound complex in leukoyctes [26]. In addition, the Src kinase Hck is activated by binding uPAR to its ligand, uPA [27]. We reported recently that similar to this study, uPAR deficiency significantly reduced cerebrospinal fluid pleocytosis by ∼50% during bacterial meningitis [17]. These results indicate that myeloid SFKs together with uPAR may in fact regulate leukocyte adhesion during inflammation in vivo. Of the two major β2 integrins on neutrophils, CD11a/CD18 (LFA-1) has been implicated more in neutrophil recruitment, and CD11b/CD18 (CR3 or Mac-1) seems to play a dominant role in leukocyte activation [9, 28]. This would be consistent with the lack of major effects on neutrophil recruitment that we observed in the CD11b^−/− mice (Fig. 4B). It was also shown in vitro that myeloid SFKs are necessary to achieve full adhesion to P- and E-selectin [7, 29] and to aggregated platelets [30]. The importance of endothelial selectins in meningitis was demonstrated in P- and/or E-selectin-deficient mice using a cytokine-induced acute meningitis model: Cerebrospinal fluid leukocytosis and BBB disruption were reduced dramatically in these mice [31]. All of these results indicate that myeloid SFKs are a converging point in the regulation of leukocyte adherence mediated by β₂ integrins, uPAR, and selectins.

Cerebral cytokine expression was not reduced in infected hck^−/−fgr^−/−lyn^−/− and CD11b^−/− mice (Figs. 3and 6). This indicates that the reduced cerebrospinal fluid leukocyte infiltration in hck^−/−fgr^−/−lyn^−/− mice and the impaired bacterial clearance in both strains are not a result of a decreased production of inflammatory mediators. This is in accordance with previous studies showing that nitrite production and cytokine secretion are unaffected in hck^−/−fgr^−/−lyn^−/− macrophages in response to LPS [5]. In addition, activation of NF-κB, which controls gene expression during inflammation, was unaltered in these cells [5, 6]. In a mouse model of pneumococcal pneumonia, production of proinflammatory cytokines in the lung of infected CD11b^−/− mice was even higher compared with infected wt mice [32]. Moreover, release of TNF-α in alveolar macrophages after stimulation with pneumococci was unaffected by CD11b deficiency [32]. All of these observations indicate that SFKs and CD11b are not required for the expression of inflammatory cytokines in myeloid cells. These data would also suggest that the primary cell source of inflammatory cytokines in pneumococcal meningitis is from perivascular or leptomeningeal macrophages, rather than the invading leukocytes themselves. If the invading leukocytes were the major contributors to cytokine production, the low cerebrospinal fluid leukocytosis in hck^−/−fgr^−/−lyn^−/− or CD11b^−/− mice would have predicted a poor cytokine response, which was not observed.

In contrast to this study, bacterial load in the brains of uPAR-deficient mice during meningitis was not affected by the reduced cerebrospinal fluid pleocytosis [33], indicating that the number of leukocytes does not necessarily correlate with the ability to clear a pathogen. Therefore, impaired leukocyte recruitment alone cannot explain the impaired bacterial clearance in hck^−/−fgr^−/−lyn^−/− mice. We found that engulfment of pneumococci was reduced in hck^−/−fgr^−/−lyn^−/− granulocytes (Fig. 2, E and F), which suggests that myeloid SFKs are necessary for bacterial phagocytosis. In fact, the role of Hck, Fgr, and Lyn as key regulators of phagocytic cell activation is well-documented for macrophage FcγR-mediated phagocytosis but has not been observed for phagocytosis of C3-opsonized particles such as zymosan [4]. Engagement of the macrophage receptor dectin-1 by zymosan particles may explain the ability of SFK macrophages to engulf these particles [34].

Our in vitro and in vivo studies now indicate that phagocytosis of pneumococci by neutrophils is also dependent on the activity of Src kinases. The innate immune response is clearly the most important factor in controlling pneumococcal infection in the nonimmune host [12]. Outside the CNS, innate resistance to S. pneumoniae is at least partially mediated by naturally occurring antiphosphocholine antibodies [35] and C-reactive protein [36]. Both require an intact complement system but not FcγRs [37]. Similar to this study, mice deficient in complement C1q and C3 have an impaired bacterial killing during pneumococcal meningitis [23], indicating that complement-mediated phagocytosis is dependent on the activity of myeloid SFKs (cerebral expression of C1r and C3 was not altered in hck^−/−fgr^−/−lyn^−/− mice). Additionally, our in vitro studies show that pneumococci opsonized with serum devoid of specific antipneumococcal antibodies (as mouse serum from RAG1-deficient mice was used) are not engulfed by hck^−/−fgr^−/−lyn^−/− granulocytes. Both results indicate that in contrast to macrophages, complement-dependent phagocytosis in granulocytes is regulated by SFKs. However, the role of the FcγRs in bacterial meningitis is unknown, and a possible involvement cannot be ruled out completely.

Complement-dependent phagocytosis is mediated by the CR3 (CD11b/CD18, α_Mβ₂), which belongs to the β₂-integrin family, and it was shown that CD11b is critical for the recognition of C3bi [38], the activation product of C3. Additionally, CR3 can recognize microbes, e.g., type B streptococci, by interacting directly with molecules on their surface [39]. As SFKs are downstream of the CR3 pathway [40], we investigated the effect of CR3 (CD11b^−/−) deficiency in pneumoccocal meningitis to explore the functional differences between the two strains. We found that similar to hck^−/−fgr^−/−lyn^−/− mice, bacterial clearance in the brain and in the blood is in fact impaired during pneumococcal meningitis in CD11b^−/− mice (Fig. 6, A and B). This can be attributed to the impaired ROS production in CD11b^−/− and hck^−/−fgr^−/−lyn^−/− neutrophils, and the milder defect in ROS production in CD11b^−/− correlates with the milder impairment in bacterial clearance (Fig. 5A). However, leukocyte recruitment to the cerebrospinal fluid is not affected in CD11b^−/− mice. This is in accordance with previous studies that showed that CD11b is important for an effective host defense against S. pneumoniae in pneumonia but not for recruitment of neutrophils into the lung [32]. These observations together with our previous study [33] strongly suggest that uPAR in concert with CD18 mediates neutrophil trafficking independent of CD11b; CD11b is essential for phagocytosis and killing of pneumococci; and both events are regulated by myeloid Src kinases.

Previous studies showed that hck^−/−fgr^−/− neutrophils have defective adhesion-dependent ROS production [19] and that Lyn plays an important role in TNF-α-stimulated production of ROS by human neutrophils [41]. This is in accordance with our observations that superoxide production was absent in hck^−/−fgr^−/−lyn^−/− granulocytes stimulated with pneumococci (Fig. 5A). In addition, we found that ROS production was impaired as a result of the lack of NADPH oxidase activation (Fig. 5B). The NADPH oxidase complex comprises two membrane-bound proteins, gp91^phox and p22^phox, and four cytosolic components, p67^phox, p47^phox, p40^phox, and Rac2. During NADPH oxidase activation and assembly, p47^phox and p40^phox function as adaptor proteins of p67^phox translocation [42]. In this study, we show that impaired ROS production in hck^−/−fgr^−/−lyn^−/− granulocytes in response to pneumococcal challenge is associated with a lack of p40^phox phosphorylation on Thr154, indicating that SFKs are required for full NADPH oxidase function. It was shown recently that neutrophils from p40^phox−/− mice have severe defects in oxidant-dependent bacterial killing, pointing out the essential role of p40^phox in NADPH oxidase regulation [43]. Upon activation, p40^phox is phosphorylated at Thr154 and Ser315 [44] by protein kinase Cs (PKCs), in particular, PKCδ (959), which indicates that activation of NADPH oxidase is regulated indirectly rather than directly by Src kinases. In fact, PKCδ^−/− neutrophils have a nearly complete absence of TNF-α-induced ROS production [45]. Therefore, possible activation pathways include tyrosine phosphorylation of PKCδ by SFKs [46] or Src-mediated tyrosine phosphorylation of other NADPH oxidase subunits, such as p47^phox [47], which is necessary for membrane translocation of p40^phox.

Taken together, our study shows that myeloid SFKs Hck, Fgr, and Lyn, but not CD11b, significantly contribute to the leukocyte recruitment into the cerebrospinal fluid space; myeloid SFKs and CD11b play a pivotal role in triggering ROS production and phagocytosis of S. pneumoniae by neutrophils; Hck, Fgr, and Lyn mediate superoxide production by activating neutrophilic NADPH-oxidase; and myeloid SFKs and CD11b are dispensable in inducing cytokine and chemokine expression in response to bacterial challenge. Although reduced ROS production and leukocyte recruitment could be of some benefit by limiting tissue injury during pneumococcal meningitis [48], the downside of SFK inhibition would be an uncontrolled growth and spread of bacteria, resulting in a worsening of the disease, as demonstrated in this study. Therefore, it is questionable whether myeloid-specific tyrosine kinases are good therapeutic targets for acute infectious diseases.