IL-10 administration reduces PGE-2 levels and promotes CR3-mediated clearance of Escherichia coli K1 by phagocytes in meningitis

To investigate the role of IL-10, 3-d-old WT and IL-10^−/− mice were infected intranasally with 10³ CFU of E. coli K1 and various parameters of the resulting infection and disease were compared. Analysis of serum samples at various time points revealed that IL-10 production peaked around 12 h after infection, whereas peak production of TNF, IFN-γ, IL-1β, IL-12, and IL-6 was observed at 72 h after infection in WT mice (Fig. 1, A and B, and Fig. S1, A–D). In contrast, these proinflammatory cytokines were observed in significantly higher quantities by 24 h after infection in IL-10^−/− mice (P < 0.01 compared with WT animals; Fig. 1 B and Fig. S1, A–D). A high level of bacteremia (10⁶ CFU of bacteria) was observed in WT animals starting from 48 h, whereas IL-10^−/− mice showed similar bacteremia levels by 12 h after infection (Fig. 1 C). The WT animals developed meningitis as demonstrated by positive CSF cultures, between 72 and 96 h after infection, whereas IL-10^−/− mice showed the pathogens in CSF by 16 h. IL-10^−/− animals succumbed to infection within 48 h even at an inoculum size of 10² CFU of E. coli (unpublished data). Histopathological examination of the brains at 72 h (WT) and 24 h (IL-10^−/−) after infection indicated that the pathological conditions in this model are very similar to that of humans. Spotty neuronal necrosis and scattered apoptotic bodies were seen both in the cortex and hippocampal dentate gyrus in the brain of E. coli K1–infected mice compared with control uninfected animals (Fig. 1 D). Acute inflammation in the brain and spreading along penetrating vessels into the cerebral cortex and white matter was prominent. A high number of neutrophils was observed at the base of the brain involving leptomeninges, cisternal areas, and choroidal fissures. Increase in cellularity was conspicuous in the molecular layer of cerebral cortex. Acute hemorrhage in the choroid plexus was also observed. These pathological changes were much more pronounced in the brains of IL-10^−/− mice compared with WT animals. On par with these results, the tissue damage markers lactate dehydrogenase and malondialdehyde (MDA) were increased dramatically in brains of IL-10^−/−–infected animals in comparison to WT-infected mice, whereas glutathione levels decreased, indicating oxidative stress (Fig. S1, E–G). Quantification of the influx of various cells into brain tissue showed a moderate influx of microglia, granulocytes, B cells, and macrophages in WT mice, which was even more prominent in IL-10^−/− mice (Fig. 1, E–H). In contrast, the influx of CD4⁺ and CD8⁺ T cells were similar up to 24 h after infection in both these animals (Fig. S2, A and B). IL-10 production in brain also robustly increased by 12 h after infection and declined by 72 h, which correlated with IL-10 mRNA, as RT-PCR showed high amounts of IL-10 transcripts in infected animals compared with control uninfected mice (Fig. S2, C–E). These results suggest that IL-10 plays a vital role in controlling the inflammatory response during the acute phase; however, the production of IL-10 was down-regulated during the chronic phase of E. coli K1 infection.

To examine whether the administration of IL-10 would protect these animals, a single dose of various concentrations of recombinant human IL-10 (rh-IL-10) or PBS was injected intraperitoneally into mice at 48 h after infection. Doses of 0.1 µg and 1.0 µg per animal did not show a significant difference in the bacterial load compared with infected WT animals. However, administration of 2.5 or 5 µg of rh-IL-10 after 48 h after infection, by which time the bacteremia levels were already high, rescued the animals from infection by 75 and 100%, respectively (unpublished data). The bacterial count and inflammatory response in rh-IL-10–treated (5 µg) animals was reduced to basal levels in blood, and the animals appeared normal by 6 d after infection (Fig. 2, A and B). This rescue by rh-IL-10 also prevented, and probably reversed, the macrophage and PMN infiltration into the brain, and thus resolved the inflammation and significantly reduced the immunopathology (Fig. 2 C). Similarly, IL-10^−/− animals were also rescued from E. coli infection by rh-IL-10 administration at 12 h after infection (unpublished data). Despite the survival of E. coli K1–infected infants as a result of antibiotic treatment, the neurological sequelae impose long-lasting effects on the quality of their life (Shah et al., 2005; Karageorgopoulos et al., 2009). In addition, the number of E. coli K1 strains that are resistant to third generation antibiotics is on the rise, and therefore the morbidity and mortality rates might increase considerably (Ramdani-Bouguessa et al., 2006; Boyer-Mariotte et al., 2008; Dubois et al., 2009). Therefore, to examine whether rh-IL-10 administration also prevents antibiotic-resistant E. coli infection, newborn mice were infected with cefotaxime-resistant or sensitive strains of E. coli K1 (E. coli CTX^R or E. coli CTX^S) and treated with cefotaxime starting at 48 h after infection (10 mg/kg body weight). The antibiotic therapy failed to rescue the E. coli CTX^R–infected animals treated with cefotaxime, and all the mice succumbed to infection by 96 h after infection, whereas animals given E. coli CTX^S survived. In addition, the animals infected with E. coli CTX^R and treated with cefotaxime still showed high levels of bacteremia and CSF cultures were positive for meningitis, whereas animals with E. coli CTX^S demonstrated declining bacteria levels over the time (Fig. 2, D and E). In contrast, the animals infected with E. coli CTX^R and treated with rh-IL-10 (5 µg/animal) survived well and showed no signs of bacteremia and the occurrence of meningitis found in mice with E. coli CTX^S infection (Fig. 2, D and E). To test whether cefotaxime treatment of newborn mice reveals normal brain histology, pathological examination of brain tissues from animals infected with E. coli CTX^R and CTX^S was performed. A significant loss of neuronal cells was observed, supporting the notion that antibiotics cannot restore normalcy in the brain (Fig. 3 A). This indicates that the neurological sequelae in infants, despite treatment with antibiotics, could be caused by the residual damage to the brain. In contrast, the histopathology of the brains obtained from mice infected with E. coli CTX^R and treated with rh-IL-10 showed no signs of damage (Fig. S2 F). Notably, injection of 1 µg of rh-IL-10 along with cefotaxime restored the morphology of brain and other organs to normal condition (unpublished data). These results demonstrate that administration of exogenous IL-10 protects newborn mice from brain damage and the occurrence of meningitis by controlling proinflammatory cytokines and by enhancing the clearance of E. coli K1 during later stages of infection.

Although the histology of infected brains of rodents have been compared in some studies (Glode et al., 1977; Ballok et al., 2004), which showed similar neuronal damage to that of human brains, the differences in the morphology of the whole brains have not established. Therefore, to examine whether the gross morphological changes of infected mouse brain are similar to that of human brain in E. coli infection, virtual histology (micro-CT) of the brains of infected mice was performed. Magnetic resonance images (MRIs) of a neonate with E. coli K1 meningitis admitted to CHLA were also examined and compared (Fig. 3, B and C, and Videos 1 and 2). In both mice and the neonate, we observed hydrocephalus, gray matter, and white matter volume loss and dilatation of the lateral ventricles. These results indicate that in newborn mice and humans, E. coli K1 meningitis caused significant brain damage. Brains of infected and antibiotic-treated animals still showed significant edema (Video 3). However, no such brain damage was visible in rh-IL-10–treated animals (Video 4). Furthermore, volumetric analysis of infected brains demonstrated a significant loss of a portion of the forebrain in mice upon infection, which was prevented by rh-IL-10 treatment (Fig. 3 C and Videos 5–7).

The production of high amounts of proinflammatory cytokines, especially TNF in mice infected with E. coli K1, might be responsible for the exacerbated pathophysiology and organ dysfunction observed in these animals. Because the administration of rh-IL-10 reduced proinflammatory cytokines at one end of the treatment regimen and cleared the pathogen on the other end, it is reasonable to speculate that suppressing the overproduction of TNF may also produce protective effects similar to that of rh-IL-10. Neutrophils and macrophages provide protection at the site of infection and are the primary source of inflammatory cytokines. Therefore, we examined the production of TNF production in neutrophils and macrophages infected with E. coli K1 in vitro at various time points and also the efficacy of anti-TNF antibody to neutralize the produced cytokine. The untreated or isotype-matched control antibody-treated neutrophils or macrophages infected with E. coli produced TNF in detectable levels at 2 h after infection and increased 8-fold at 6 h after infection (Fig. 4 A and Fig. S3 A). Preincubation of the phagocytes with anti-TNF antibody completely neutralized the TNF production. The survival of E. coli K1 in anti-TNF antibody pretreated neutrophils or macrophages was not affected although the bacteria were completely taken up by the cells by 6 h after infection (Fig. 4, B and C, and Fig. S3, B and C). In addition, anti-TNF antibody failed to rescue the neutrophils or macrophages from E. coli K1–mediated apoptosis (Fig. 4 D and Fig. S3 D). To examine the effect of anti-TNF antibody in vivo, infected animals were treated with anti-TNF antibody at 48 h after infection. Mice infected with E. coli with or without administration of control antibody showed a gradual increase in TNF production in both blood and brain, whereas administration of anti-TNF antibody neutralized the production of the cytokine by 72 h after infection, and hence was not detectable by ELISA (Fig. 5, A and B). However, the bacterial load and the occurrence of meningitis in anti-TNF antibody–treated animals were similar to that of untreated or control antibody-treated mice (log₁₀ 8.0 ± 0.5 CFU of E. coli in the blood and with 90% positive CSF cultures at 72 h after infection; Fig. 5, C and D). Histological examination of the brain sections from anti-TNF antibody–treated animals exhibited similar damage to the brain compared with control antibody or untreated animals (Fig. 5 E). These findings suggest that neutralizing the production of TNF does not confer protection against E. coli K1 meningitis unlike the administration of rh-IL-10.

The ability of rh-IL-10 treatment to clear the pathogen from the circulation of infected mice suggests increased phagocytosis of the bacteria by neutrophils and/or macrophages probably mediated by IL-10R. Therefore, we examined whether IL-10R1 is expressed in uninfected mouse neutrophils by flow cytometry. The data revealed that IL-10R1 expressed in good quantity on the surface of neutrophils (Fig. 6 A). Furthermore, the effect of rh-IL-10 treatment of neutrophils before bacterial infection on the phagocytic activity of neutrophils isolated from WT animals was examined ex vivo. E. coli K1 enters and survives in neutrophils isolated from WT mice; however, neutrophils pretreated with rh-IL-10 showed enhanced uptake of E. coli K1 and killing of bacteria within 6 h after infection (Fig. 6, B and C). Neutrophils from WT mice infected with E. coli K1 underwent rapid apoptosis compared with cells from rh-IL-10–treated animals (Fig. 6 D). Neutrophils from IL-10^−/− mice showed no capacity to kill the bacteria and underwent apoptosis very rapidly ex vivo. They could, however, be rescued from E. coli–induced apoptosis with rh-IL-10 pretreatment. Similar results were observed with neutrophils isolated from human peripheral blood (Fig. S4, A–C). Collectively, these data indicate that bacteria might prevent bactericidal events in neutrophils either by suppressing the apoptotic mechanisms once inside neutrophils or by modulating their entry into the cells, such that the normal phagocytic arsenal is circumvented. To test the latter hypothesis, we examined the expression of surface pathogen associated recognition receptors on neutrophils isolated from mice infected with E. coli K1. Flow cytometry analysis revealed increased expression of TLR2 and FcγRI (CD64), but reduced expression of TLR4 and CR3 (Fig. 6 E). Notably, neutrophils from infected and rh-IL-10–treated mice showed significantly increased expression of CR3 and TLR4 and down-regulation of TLR2 and FcγRI. A similar receptor profile was observed on human neutrophils (Fig. S4 D). Introducing siRNA to CR3 into neutrophils in vitro further substantiated the role of CR3 in the phagocytosis of E. coli K1 by neutrophils after rh-IL-10 treatment. E. coli entered and survived in neutrophils transfected with CR3-siRNA and pretreatment with rh-IL-10 had no effect on uptake and killing of the bacterium (Fig. 6, F and G). The protective effect of rh-IL-10 on the apoptosis of infected neutrophils was also abrogated by suppressing the expression of CR3 (Fig. 6 H). These data suggest that rh-IL-10 administration up-regulates the expression of CR3 in neutrophils leading to enhanced killing of E. coli K1. In addition to neutrophils, macrophages also play an important role in the clearance of invading pathogens. Therefore, we also examined the effect of rh-IL10 on the phagocytosis of E. coli and surface expression of various receptors upon infection in macrophages. Peritoneal macrophages from both WT and IL-10^−/− exhibited similar uptake and intracellular survival of E. coli, whereas rh-IL-10 treatment reduced the survival of bacteria in these cells significantly (Fig. S5, A and B). The viability and the expression of surface receptors were also similar to that of the expression observed in neutrophils (Fig. S5, C and D). In addition, no difference in uptake, survival, apoptosis, and receptor profile of CR3 knockout macrophages even after treatment with rh-IL-10 compared with control siRNA-treated macrophages was observed (Fig. S5, E–H). This indicates that rh-IL-10 enhances CR3 expression so that the bacteria enter through CR3 for efficient destruction by phagocytes.

To confirm that CR3 expression is critical in vivo for rh-IL-10 mediated protection against E. coli infection and brain damage, CR3 expression in mice was suppressed using Invivofectamine reagent. Groups of animals were transfected with premixed control siRNA or CR3-siRNA and Invivofectamine reagent by intraperitoneal injection. CR3 expression in neutrophils, peritoneal macrophages, liver, and spleen was then examined by flow cytometry. The expression of CR3 was completely absent in neutrophils, macrophages, and other tissues of transfected mice (Fig. 7 A and not depicted). Determination of bacterial load in the blood of CR3-siRNA–transfected and –infected animals showed that E. coli entered the blood within 2 h and multiplied to a log₁₀ CFU of >5 by 24 h similar to that of control siRNA transfected mice (Fig. 7 B). Administration of rh-IL-10 in control siRNA transfected animals reduced the bacterial load, whereas it had no effect on CR3 knocked down animals. The animals showed positive CSF cultures and histopathological analysis revealed brain damage and infiltration of neutrophils similar to that of WT animals infected with E. coli (Fig. 7, C and D). The protective effect of rh-IL-10 administration during high-level bacteremia on cytokine production in CR3 knocked down mice was completely abrogated (Fig. 7 E, only TNF data are depicted). Neutrophils from CR3-siRNA–transfected and –infected animals revealed greater expression of TLR2 and FcγRI and did not change in animals with rh-IL-10 treatment (Fig. 7 F). Similar results were also observed with peritoneal macrophages isolated from CR3 knocked down mice (unpublished data). These results collectively demonstrate that CR3 expression was significantly suppressed during the progression of the disease by E. coli K1 and the protective effects of rh-IL-10 are caused by restoration of CR3 expression in neutrophils and macrophages.

Two important mediators of inflammatory response during bacterial meningitis are nitric oxide and PGE-2 (Kim, 2003). Although there is limited information on the role of prostanoids, such as PGE-2 in meningitis, one clinical study showed that 72 of 80 infants and children with meningitis exhibited elevated levels of CSF PGE-2 (Mustafa et al., 1990). Notably, PGE-2 has also been involved in down-regulation of CR3 expression in monocytes and macrophages (Zeidler et al., 2000). Therefore, to examine whether PGE-2 is responsible for rh-IL-10–induced protection against meningitis caused by E. coli, the production of PGE-2 in neutrophils and macrophages was examined. Neutrophils infected with the bacteria showed significantly greater production of PGE-2 over the time in comparison to uninfected cells, which levels were reduced by >80% by rh-IL-10 treatment (Fig. 8 A). In contrast, E. coli–infected neutrophils in which PGE-2 expression is suppressed by introducing siRNA to PGE synthase 2 (PGES-2) demonstrated no production of PGE-2 in comparison to control siRNA-transfected cells (Fig. 8 B). The PGES-2 siRNA–transfected neutrophils, despite engulfing the bacteria in a manner similar to that of control siRNA/neutrophils, killed very efficiently and no detectable bacteria were observed by 6 h after infection (Fig. 8, C and D). In contrast, intracellular E. coli in control siRNA/neutrophils survived and multiplied over the same time period. Similarly, suppression of PGES-2 expression in neutrophils renders the cells resistant to apoptosis despite infection with bacteria (Fig. 8 E). In addition, E. coli–infected PGES-2 siRNA/neutrophils showed increased expression of CR3 and TLR4 and decreased expression of TLR2 and FcγRI, similar to expression levels observed in rh-IL-10–pretreated and infected neutrophils (Fig. 8 F). We therefore examined whether mice infected with E. coli induce the production of PGE-2. On par with the in vitro data, the production of PGE-2 in blood increased twofold by 72 h in comparison to the levels produced at 24 h after infection and administration of rh-IL-10 at 48 h after infection significantly reduced circulating PGE-2 by 72 h (Fig. 8 G). The levels of PGE-2 in the brains of infected animals exhibited a similar increase. These levels decreased to baseline after rh-IL-10 treatment (Fig. 8 H). Similar data on PGE-2 production was also obtained with peritoneal macrophages (Fig. S6, A–F), substantiating the role of PGE-2 in rh-IL-10–induced protective effects during the development of E. coli K1 meningitis.

Currently, the primary mode of treatment of E. coli meningitis is limited to antibiotics, which reduced the mortality and frequency of complications, but the incidence of serious neurological disability remains alarmingly high (Kim, 2003). Trivedi et al. (2007) showed that focal anisotropy values were significantly greater in patients with E. coli infection after antibiotic treatment than in the controls, indicating that the inflammation is still active. Moreover, the emergence of antibiotic-resistant E. coli further complicates treatment options (Dubois et al., 2009), warranting the necessity of new treatments that control both bacterial load and inflammatory response. Our findings provide novel insights into the role of IL-10 in the control of pathogen load at times of high-grade bacteremia in a newborn mouse model of E. coli meningitis. Although we expected that exogenous administration of IL-10 would decrease disease severity because of suppression of the proinflammatory response, its effect in the control of pathogen load was most surprising. In the majority of situations, IL-10 has a suppressive effect on both macrophages and dendritic cells (Moore et al., 2001; Demangel et al., 2002) and most of the animal models have shown that IL-10 inhibits antimicrobial response while protecting from immunopathology (Hunter et al., 1997; Wilson et al., 2005). In contrast to this paradigm, IL-10 plays the opposite role in E. coli infection; it helps control the phagocytic activity of immune cells to clear the pathogens while reducing inflammation. Some studies have reported that persistent bacteremia is the result of loss of pathogen clearance caused by apoptosis of lymphocytes (Hotchkiss et al., 2001; Alileche et al., 2005; Wesche-Soldato et al., 2007). Also, there is evidence that patients dying from sepsis have markedly increased lymphocyte apoptosis in the spleen (Toti et al., 2004; Felmet et al., 2005). Recently, it has been shown that IL-10 generates pathogen-clearing phagocytes (monocytes and macrophages), which are resistant to complement lysis, and thereby enabled to survive longer in a hostile inflammatory environment (Koch et al., 2009). Our findings suggest that IL-10 treatment prevents neutrophil and macrophage apoptosis induced by E. coli, and thereby increases the phagocytosis of the bacteria. In animal models using adult mice, it has been shown that both Gram-negative bacteria and mycobacteria induced the recruitment of neutrophils, which secrete abundant amounts of IL-10 (Zhang et al., 2009). In acute mycobacterial infection, neutrophil-derived IL-10 controlled the inflammatory response of dendritic cells, monocytes, and macrophages in the lung (Zhang et al., 2009). However, diminished production of IL-10 mRNA and IL-10 protein in term and preterm neonates compared with adults was observed after stimulating with LPS (Schultz et al., 2004). We also observed decreased IL-10 production and increased levels of proinflammatory cytokines in blood and brain of E. coli–infected neonatal mice 24 h after infection. Therefore, IL-10 produced by neutrophils may not be enough, and excessive IL-10 quantities may be required to subdue the inflammatory response and improve phagocytosis of pathogens by neutrophils and macrophages in neonates.

Our studies also demonstrated that the IL-10–induced bactericidal activity of neutrophils and macrophages is caused by increased expression of CR3 and TLR4, which was suppressed during the infection without treatment. The role of CR3-mediated clearance of E. coli from the blood was further substantiated in mice by suppressing CR3 expression using siRNA and Invivofectamine. Because CR3 plays a role in transmigration of neutrophils from the blood to the site of inflammation, it is possible that lack of occurrence of meningitis in CR3 knockdown mice could be caused by inefficient migration of neutrophils, which may be acting as reservoirs for E. coli multiplication during the early stages of infection. However, mice deficient in CR3 have been developed and found to have normal migration of neutrophils in the peritoneal cavity after chemical and TNF stimulus (Lu et al., 1997). LFA-1 has been shown to overshadow the contribution of CR3 for neutrophil migration and adhesion to endothelial cells, indicating that CR3 is not absolutely necessary (Ding et al., 1999). In agreement, histopathological sections of brains of CR3 knocked down and infected mice also exhibited neutrophil infiltration similar to that of WT-infected mice, indicating neutrophil migration is not defective in CR3-suppressed mice. Furthermore, increased formation of PGE-2 was observed in the blood of E. coli–infected mice and also by neutrophils and macrophages ex vivo upon infection. Biogenesis of PGE-2 has been shown to occur through TLR2 signaling during apoptotic neutrophil uptake by macrophages in acute BCG infection (D’Avila et al., 2008). The PGE-2 production observed in E. coli K1 meningitis corroborates this mechanistic concept, as we also observed significant increase in TLR2 production in neutrophils and macrophages upon infection with E. coli K1. IL-10 administration suppressed the formation of PGE-2 by neutrophils and macrophages, in which case the expression of TLR2 in these cells suppressed and induced the expression of CR3. It has been demonstrated that IL-10 inhibits the production of PGE-2 induced by LPS in monocytes (Niiro et al., 1994); however, PGE-2 generated during efferocytosis enhanced the production of IL-10 (Medeiros et al., 2009). Thus, the relationship between IL-10 and PGE-2 may be dependent on the biological phenomena that are taking place or the presence of other pro- and antiinflammatory cytokines in the milieu. Collectively, these results suggest that E. coli entry via receptors other than CR3 target PGE-2 to enhance bacterial survival and replication in host cells.

Studies in experimental animals and humans with bacterial meningitis have shown that two forms of neuronal injury are predominant: necrotic cortical injury and apoptotic hippocampal injury, as seen in E. coli–infected mice in this study (Nau et al., 1999; Kim, 2003). Neuronal injury associated with bacterial meningitis probably involves many microbial and host factors. Adjunct therapy with dexamethasone has been shown to reduce hearing impairment and other neurological sequelae in children with H. influenzae type b meningitis, but dexamethasone was shown to aggravate hippocampal injury in experimental models of pneumococcal meningitis (Wald et al., 1995; Leib et al., 2003). In contrast to these observations, IL-10 treatment completely prevented and/or restored the damaged brain as seen in micro-CT images and histopathology. Prevention of meningitis after reaching a threshold level of bacteremia is a very important therapeutic strategy, as the diagnosis of meningitis is difficult until after the bacteria reaches the central nervous system. The finding of prevention of meningitis caused by antibiotic-resistant strains by exogenous IL-10 is significant to treat these patients. This function indicates that an adjunct therapy containing a regimen of antibiotics and small quantities of IL-10 would be possible to treat neonates. Since IL-10 therapy has already been used in adults for some autoimmune diseases like psoriasis, it is possible that administration of small quantities of exogenous IL-10 would not have adverse effects on neonates (Friedrich et al., 2002). Although, the concept of treating neonates with E. coli infections by exogenous IL-10 remains to be tested, these findings provide a strong rationale for exploration of IL-10 therapies beyond antibiotic treatment alone. This IL-10 therapy may also form the basis of future therapeutic strategies to treat meningitis.

Breeding pairs of C57BL/6 IL-10^−/− mice and C57BL/6 WT mice were obtained from The Jackson Laboratory.

E. coli K1 is a rifampicin-resistant mutant of E. coli K1 strain RS 218 (serotype O18:K1:H7) and is sensitive to cefotaxime (CTX^S; Prasadarao et al., 1996). Cefotaxime-resistant strain of E. coli (CTX^R) was isolated from CSF of a newborn as previously described (Dubois et al., 2009). Bacteria were grown in brain heart infusion broth (Difco Laboratories) with appropriate antibiotics.

The animal studies were approved by the institutional animal care and use committee of the Saban Research Institute at CHLA and followed National Institutes of Health guidelines for the performance of animal experiments. 3-d-old mice were randomly divided into various groups and infected intranasally with 10³ CFU/10 µl of bacteria. Control mice received pyrogen-free saline through the same route. Blood was collected from the tail or facial vein at different times after infection and plated on LB agar containing rifampicin. CSF samples were collected aseptically under anesthesia by cisternal puncture and directly inoculated into broth containing antibiotics in which the growth of E. coli was considered positive for meningitis. Mice were perfused intracardially with 0.9% saline to remove blood and contaminating intravascular leukocytes. Various tissues were aseptically harvested and homogenized in sterile PBS, and the bacterial counts were determined by plating 10-fold serial dilutions on rifampicin LB agar plates.

To assess the effect of rh-IL-10 on meningitis, groups of 3-d-old mice were infected with E. coli CTX^S or E. coli CTX^R and treated intraperitoneally at 48 h after infection with various doses of rh-IL-10 (0.1–5 µg/animal; (PeproTech), with cefotaxime (10 mg/kg body weight) or with cefotaxime and rh-IL-10 (1 µg/animal) together. For anti-TNF antibody treatment, groups of 3-d-old mice were infected with E. coli CTX^S and treated intraperitoneally at 48 h after infection with one dose of 5 µg/animal of murine anti-TNF antibody or isotype-matched control antibody (R&D Systems).

Half of the tissue samples were fixed in 10% buffered formalin, routinely processed, and embedded in paraffin. 4–5-µm-thick sections were cut using a Leica microtome and stained with hematoxylin and eosin (H&E; Mittal et al., 2009b).

The levels of TNF, IL-1β, IL-6, IL-10, IL-12p70, and IFN-γ, were measured in blood and brain homogenates of infant mice infected with E. coli using ELISA kits from Biosource according to the manufacturer’s instructions. PGE-2 levels were determined by EIA kit (Cayman Chemicals).

Characterization of leukocytes derived from brains of newborn mice was done using flow cytometry as described previously (Mittal and Prasadarao, 2008) using various antibodies obtained from eBioscience. Murine microglia and macrophages were identified by staining with anti-CD45 (LAC)-biotin and anti-F4/80-FITC, respectively followed by avidin-PE/Cy5. Inflammatory leukocytes recruited to the brain are CD45^high F4/80⁻, macrophages are CD45^high F4/80⁺, and microglia are CD45^low F4/80⁺. CD4⁺ and CD8⁺ T lymphocytes were stained with rat anti–mouse-CD4 followed by goat anti–rat-phycoerythrin and anti-CD8-FITC. B lymphocytes were detected by staining with anti-CD45R (B220)-FITC and anti-CD45 (LCA)-biotin followed by avidin-phycoerythrin/Cy5. Granulocytes were stained with anti-Ly6-G (GR-1), followed by goat anti–rat-PE and F4/80-FITC. Granulocytes were defined as Ly6-G⁺ F4/80⁻. Control staining included incubation of brain-derived leukocytes with unlabeled or fluorochrome-labeled isotype-matched control antibodies. Flow cytometry was performed on a FACScan instrument (BD) and the data were analyzed with CellQuest Software.

For detection of IL-10 by RT-PCR, RNA was isolated from the brains of uninfected and infected mice with TRIZOL-LS-reagent (Invitrogen). RNA was quantified using a NanoDrop machine. RT-PCR was performed using the following primer sequences: sense, 5′-ACCTGGTAGAAGTGATGCCCCAGGCA-3′; antisense, 5′-CTATGCAGTTGATGAAGATGTACCA-3′. Mouse GAPDH was used as an internal control, using the following primer sequences: sense, 5′-CATCACCATCTTCCAGGAGCG-3′; antisense, 5′-GAGGGGCCATCCACAGTCTTC-3′. Negative controls without RT were performed in parallel for every reaction, to exclude amplification of contaminating DNA. The amplified products were separated on a 1% agarose gel and were stained with ethidium bromide. The gels were photographed, and optical densities were determined by using a computer imaging analysis system (Visitron Systems GmbH). RT-PCR was performed three times and the mean density of PCR products was determined to calculate the ratio of IL-10 to GAPDH mRNA.

Formalin-fixed specimens were sent to Numira Biosciences for micro-CT imaging. Specimens were stained with a proprietary contrast agent. A high-resolution, volumetric micro-CT scanner (μCT40; ScanCo Medical) was used to scan the tissue with the following parameters: 10 µm isometric voxel resolution at 200 ms exposure time, 2,000 views, and 10 frames per view. The micro-CT generated DICOM files were converted into a file format compatible with the image processing software applications. Using Teem (Version 1.10.0; http://teem.sourceforge.net/), SCIRun (Scientific and Computing Imaging Institute, University of Utah, Salt Lake City, Utah), and Numira’s proprietary software tools, images of the samples were generated.

Neutrophils were isolated from the peripheral blood of the human volunteers and mice as described previously (Criss and Seifert, 2008; Luo and Dorf, 2001). IRB of CHLA approved the blood withdrawal protocol from human volunteers. Macrophages were isolated from the peritoneal cavity of mice as previously described (Mittal et al., 2009a). In some experiments, neutrophils and macrophages were transfected with CR3 siRNA (Invitrogen; Oligo ID MSS205544), PGES-2 siRNA (Ambion; Oligo ID 97139), or respective control siRNA using Lipofectamine 2000+ (Invitrogen) transfecting reagent. Phagocytosis of bacteria by neutrophils and macrophages was determined by gentamicin protection assays (Mittal and Prasadarao, 2008). The expression of TLR2, TLR4, CR3, and FcγRI on the surface of neutrophils and macrophages isolated from infected mice was determined by flow cytometry after staining with respective antibodies coupled to various fluorochromes, and at least 10,000 events were collected for analysis. Neutrophils and macrophages were gated using GR-1 and F4/80 staining, respectively, as well as forward and side scatter parameters. Results are expressed as mean fluorescence intensity subtracted from isotype-matched and respective uninfected controls.

To suppress CR3 expression, newborn mice at day 1 were injected with CR3 siRNA or control siRNA in Invivofectamine reagent (premixed in 1:1 ratio, 10 nmol in 10 µl; Invitrogen). Invivofectamine is a proprietary lipid-based nontoxic transfection reagent for in vivo RNAi applications. This reagent stabilizes siRNA in vivo and knocks down the expression of gene of interest. Mice were given four doses of CR3 siRNA intraperitoneally; two doses before infection, one dose immediately after administration of bacteria, and final dose 24 h after infection.

For statistical analysis of data, analysis of variance, Fischer test, Wilcoxon signed rank test, χ² test, and Student’s t test were applied. P values <0.05 were considered statistically significant.

Fig. S1 shows cytokine production and tissue damage markers in E. coli–infected WT and IL-10^−/− mice. Fig. S2 shows CD4⁺ and CD8⁺ T cell responses, IL-10 production, and brain pathology upon E. coli infection in WT and IL-10^−/− mice. Fig. S3 depicts the effect of TNF neutralization on the phagocytic ability and viability of E. coli–infected macrophages in vitro. Fig. S4 shows the phagocytosis, receptor profile, and viability of human neutrophils infected with E. coli in vitro. Fig. S5 shows the phagocytic ability, apoptosis, and receptor profile of peritoneal macrophages pretreated with IL-10 before infection with E. coli. Fig. S6 shows the effect of silencing PGES-2 on the ability of peritoneal macrophages to engulf E. coli, prevent apoptosis, and increase the expression of surface receptors compared with control siRNA-treated cells.

We thank Elaine Tuomanen and Grace Aldrovandi for their suggestions and encouragement during the progress of this work. We also thank Scott Filler, Barbara Driscoll, and Mary Kearns-Jonker for critical reading of the manuscript.

This work was supported by RO1 AI40567 to N.V. Prasadarao.

The authors have no conflicting financial interests.