The Role of Autophagy during Group B Streptococcus Infection of Blood-Brain Barrier Endothelium*

Background: Penetration of brain endothelium by Group B streptococcus (GBS) is the first step in the development of meningitis. Results: Autophagy is activated in response to GBS infection. Conclusion: Autophagy induction occurs through GBS toxin expression, while key autophagic proteins contribute to GBS destruction. Significance: Understanding the role of autophagy in brain endothelium may inform novel strategies to prevent the pathogenesis of bacterial meningitis. Bacterial meningitis occurs when bloodborne pathogens invade and penetrate the blood-brain barrier (BBB), provoking inflammation and disease. Group B Streptococcus (GBS), the leading cause of neonatal meningitis, can enter human brain microvascular endothelial cells (hBMECs), but the host response to intracellular GBS has not been characterized. Here we sought to determine whether antibacterial autophagy, which involves selective recognition of intracellular organisms and their targeting to autophagosomes for degradation, is activated in BBB endothelium during bacterial infection. GBS infection resulted in increased punctate distribution of GFP-microtubule-associated protein 1 light chain 3 (LC3) and increased levels of endogenous LC3-II and p62 turnover, two hallmark indicators of active autophagic flux. Infection with GBS mutants revealed that bacterial invasion and the GBS pore-forming β-hemolysin/cytolysin (β-h/c) trigger autophagic activation. Cell-free bacterial extracts containing β-h/c activity induced LC3-II conversion, identifying this toxin as a principal provocative factor for autophagy activation. These results were confirmed in vivo using a mouse model of GBS meningitis as infection with WT GBS induced autophagy in brain tissue more frequently than a β-h/c-deficient mutant. Elimination of autophagy using Atg5-deficient fibroblasts or siRNA-mediated impairment of autophagy in hBMECs led to increased recovery of intracellular GBS. However, electron microscopy revealed that GBS was rarely found within double membrane autophagic structures even though we observed GBS-LC3 co-localization. These results suggest that although autophagy may act as a BBB cellular defense mechanism in response to invading and toxin-producing bacteria, GBS may actively thwart the autophagic pathway.

Bacterial meningitis occurs when bloodborne pathogens invade and penetrate the blood-brain barrier (BBB), provoking inflammation and disease. Group B Streptococcus (GBS), the leading cause of neonatal meningitis, can enter human brain microvascular endothelial cells (hBMECs), but the host response to intracellular GBS has not been characterized. Here we sought to determine whether antibacterial autophagy, which involves selective recognition of intracellular organisms and their targeting to autophagosomes for degradation, is activated in BBB endothelium during bacterial infection. GBS infection resulted in increased punctate distribution of GFP-microtubule-associated protein 1 light chain 3 (LC3) and increased levels of endogenous LC3-II and p62 turnover, two hallmark indicators of active autophagic flux. Infection with GBS mutants revealed that bacterial invasion and the GBS pore-forming ␤-hemolysin/cytolysin (␤-h/c) trigger autophagic activation. Cell-free bacterial extracts containing ␤-h/c activity induced LC3-II conversion, identifying this toxin as a principal provocative factor for autophagy activation. These results were confirmed in vivo using a mouse model of GBS meningitis as infection with WT GBS induced autophagy in brain tissue more frequently than a ␤-h/c-deficient mutant. Elimination of autophagy using Atg5-deficient fibroblasts or siRNA-mediated impairment of autophagy in hBMECs led to increased recovery of intracellular GBS. However, electron microscopy revealed that GBS was rarely found within double membrane autophagic structures even though we observed GBS-LC3 co-localization. These results suggest that although autophagy may act as a BBB cellular defense mechanism in response to invading and toxinproducing bacteria, GBS may actively thwart the autophagic pathway.
Bacterial meningitis is a serious infection of the central nervous system (CNS) that can develop rapidly into a life-threatening infection even in previously healthy children or adults. The Gram-positive bacterium Streptococcus agalactiae, known as group B Streptococcus (GBS), 2 is the leading cause of meningitis in newborn infants (1). Although antibiotic therapy has changed GBS meningitis from a uniformly fatal disease to an often curable one, the overall outcome remains unfavorable as 25-50% of surviving infants suffer permanent neurological sequelae of varying severity, including cerebral palsy, mental retardation, blindness, deafness, and seizures (2). Infection is initiated when bloodborne bacteria cross the blood-brain barrier (BBB) in a complex interplay between endothelial cells and microbial gene products. The human BBB, which is composed of a single layer of specialized human brain microvascular endothelial cells (hBMECs), separates the brain and its surrounding tissues from the circulating blood, tightly regulating the flow of nutrients and molecules promoting the proper biochemical conditions for normal brain function (3,4). Although the BBB serves as a critical barrier to protect the CNS against microbial invasion, disruption of the BBB is a hallmark event in the pathophysiology of bacterial meningitis. This disruption may be due to the combined effect of bacterial entry, direct cellular injury by bacterial cytotoxins, and/or activation of host inflammatory pathways that compromise barrier function. GBS produces a pore-forming ␤-hemolysin/cytolysin (␤-h/c) that has been shown to directly damage brain endothelial cells (5) and activate proinflammatory mediators, promoting the development of GBS meningitis in vivo (6,7). To gain entry into the CNS and the subarachnoid space, GBS must persist in the blood stream and interact with and penetrate brain endothelium; however, the exact mechanism(s) of bacterial transit across the BBB is not known. It is likely that GBS tropism for the BBB is the primary step in the pathogenesis of meningitis. Many GBS surface components have been identified that contribute to the initial interaction with hBMECs, including invasion-associated gene A (iagA), which is required for proper anchoring of lipoteichoic acid to the cell wall (8); Srr-1 (9, 10); FbsA (11); Lmb (12); HvgA (13); alpha C protein (14); and pili components, which consist of the pilus backbone protein PilB and pilus tip adhesin PilA (15,16). GBS is able to enter or "invade" brain endothelium apically and exit the cell on the basolateral side, thereby crossing the BBB transcellularly (5,7). Electron microscopy (EM) studies have demonstrated the presence of GBS in membrane-bound vesicles within hBMECs (5), suggesting the involvement of endocytic pathways. However, little is known about how GBS persists and traffics through the BBB or the host defenses deployed to combat its intracellular presence in brain endothelium.
Macroautophagy, hereafter referred to as autophagy, is an evolutionarily conserved degradation process that utilizes the lysosomal machinery to recycle damaged, aggregated, or aged cytoplasmic constituents. Cargo is initially captured into the autophagosome through the formation of an isolation membrane called the phagophore, which is ultimately destined for lysosomal fusion resulting in cargo degradation (17,18). Autophagy is initiated by the interactions between multiple autophagy (ATG) proteins (19). ATG5, ATG12, and microtubule-associated protein 1 light chain 3 (LC3)/GATE-16/GABA receptor-associated protein are vital for the formation of the initial phagophore and maturation of the autophagosome. LC3 is conjugated to phosphatidylethanolamine, a lipid constituent of plasma membranes, by the ATG5-ATG12-ATG16L1 complex to allow for autophagosome expansion (19). Recently, antimicrobial autophagy, a selective type of autophagy also known as xenophagy, has emerged as a potent host defense mechanism against intracellular bacterial and viral pathogens (17,20). Several pathogenic bacteria such as Salmonella enterica serovar Typhimurium (Salmonella typhimurium), Listeria monocytogenes, Shigella flexneri, and group A Streptococcus (GAS) have been shown to activate the autophagic pathway (21)(22)(23). Multiple mechanisms have been described as to how these and other pathogens are recognized by the cell to induce the autophagic process (24). Further modulation or evasion of these pathways by bacteria may be critical for their intracellular survival and disease manifestation.
In the present study, we examined the hypothesis that selective autophagy may play a role in host defense against meningeal pathogens such as GBS. Our results demonstrate that GBS infection triggers a robust autophagic response in brain endo-thelium and that this response contributes to limiting intracellular bacteria. Experiments with isogenic GBS mutants lacking the ␤-h/c toxin or surface components that promote cellular invasion indicate that these virulence factors impact autophagy induction. Furthermore, our studies demonstrate that the GBSsecreted ␤-h/c toxin is sufficient to activate an acute autophagic response in BBB endothelium but that this response may not be adequate to reduce the majority of intracellular GBS.
Construction of Green Fluorescent Protein-expressing GBS-The pDESTerm plasmid expressing GFP was provided by John Buchanan and Victor Nizet (University of California, San Diego). Competent bacterial cells were created by propagating GBS in THB with 0.6% glycine to early log phase. Cells were then centrifuged at 4000 rpm for 30 min at 4°C. The supernatant was removed, and bacteria were washed with ice-cold 0.625 M sucrose buffer. Bacteria were centrifuged again as described above, and again the supernatant was removed. 1 g/l plasmid was added to the competent GBS in a 0.1-cm electroporation cuvette, and cells were electroporated at 1500 V for 2-4 ms. All steps were performed on ice. Recovery medium (THB with 0.25 M sucrose) was added to the cells, and cells were incubated at 37°C for 1 h. The culture was then plated on THB agar plates containing 5 g/ml erythromycin and incubated at 37°C with 5% CO 2 . Colonies were then assessed for fluorescence by microscopy and fluorescent-activated cell sorting.
Infection Assays-hBMECs and ATG5 knock-out (KO) and WT MEFs were grown to confluence (ϳ10 5 cells/well) and washed three times prior to GBS infection. GBS was grown in THB to midlog phase (ϳ10 8 cfu/ml; A 600 ϭ 0.4), washed in PBS, resuspended in RPMI 1640 medium plus 10% FBS, and used to infect monolayers at a multiplicity of infection (m.o.i.) of 1 or 10 for various time points. Plates were then centrifuged at 1000 rpm for 3 min to synchronize infection and incubated at 37°C in 5% CO 2 . After infection, cells were treated with penicillin (5 g/ml) and gentamycin (100 g/ml) to kill extracellular GBS for various time points. Cells were treated with 0.1 ml of 0.25% trypsin, EDTA solution and lysed with addition of 0.4 ml of 0.025% Triton X-100 by vigorous pipetting. The lysates were then serially diluted and plated on THB agar to enumerate bacterial cfu. Cell lysate was collected and stored at Ϫ80°C until further use. In specified experiments, hBMECs were pretreated with rapamycin (5 M; Calbiochem, catalog number 553211) and bafilomycin (100 nM; LC Laboratories, catalog number B-1080) for 1 h prior to GBS infection.
Transmission Electron Microscopy-hBMECs were incubated with GBS at 37°C with 5% CO 2 . Cells were washed three times with PBS. Cells were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer for 90 min and rinsed 3 times in 0.1 M cacodylate buffer for 10 min for each rinse. Samples were postfixed in 1% osmium tetroxide for 90 min and then dehydrated at increasing concentrations of ethanol and acetone for 10 min each. Samples were embedded in Epon acetone and baked at 60°C overnight. Thin sections were cut using a diamond knife on a Leica microtome, stained with uranyl acetate and lead citrate, and viewed using an FEI Tecnai 12 transmission electron microscope.
Immunofluorescence Staining-hBMECs were fixed with 4% paraformaldehyde prior to mounting with VectaShield with DAPI (Vector Laboratories, catalog number H-1200). For GFP-LC3 studies, hBMECs were transduced with GFP-LC3 adenovirus overnight in RPMI 1640 medium containing 2% FBS before infection with GBS. After infection, cells were fixed in 4% paraformaldehyde prior to solubilization in 0.1% Triton X-100 and subsequent staining with a GBS-specific antibody (Acris, catalog number BM5557P) or anti-von Willebrand factor antibody (Sigma, catalog number F3520). For tissue visualization of GFP-LC3 mice, brains were harvested and fixed in optimal cutting temperature compound (O.C.T.) (VWR, catalog number 25608-930) and then frozen at Ϫ80°C. Brain tissue was then sectioned using a Leica cryostat. Samples were visualized using a Zeiss Axiovert 200 inverted fluorescence microscope (Carl Zeiss) or a Zeiss LSM 710 confocal microscope (Carl Zeiss).
Preparation of Heat-killed and Paraformaldehyde-fixed GBS-Bacteria were grown to midlog phase in THB and subsequently pelleted. Pelleted bacteria were reconstituted in PBS and then boiled at 95°C for 5 min. After boiling, bacteria were diluted accordingly for an m.o.i. of 10 and administered to hBMECs. For paraformaldehyde fixation, bacteria were grown to midlog phase in THB, pelleted, resuspended in 1% paraformaldehyde, and incubated for 10 min at room temperature. The fixed pellet was then washed three times with PBS and added to hBMECs at an m.o.i. of 10.
Preparation of GBS Hemolytic Extract-GBS hemolytic extract was prepared as described previously (6,37). Briefly, WT GBS NCTC 10/84 and isogenic ⌬cylE strains were grown to an A 600 of 0.4 -0.6 in THB. Bacteria were then pelleted and resuspended in PBS containing 1% glucose and 1% starch. After 1 h at room temperature, bacteria were sterile filtered using a 0.2-m syringe filter. After filtration, bacteria were pelleted and washed once in PBS. The washed pellet was then incubated for 1 h at 37°C. The bacteria were then pelleted and resuspended in PBS plus 1% starch and 1% glucose, and the supernatant was sterile filtered using a 0.2-m syringe filter. And equal volume of ice-cold methanol was added to the filtered supernatant and incubated on ice for 1 h. Methanol/ supernatant was then pelleted and resuspended in 1 ml of PBS. The hemolytic titer of the isolated extract was quantified as described previously (38).
Mouse Model of Hematogenous Meningitis-Animal experiments were approved by Institutional Animal Care and Use Committee at San Diego State University Protocol APF 13-07-011D and performed using accepted veterinary standards. We utilized a mouse model of hematogenous GBS meningitis as described previously (6, 8, 15). 8-week-old male CD-1 mice (n ϭ 10) were injected intravenously with 7-8 ϫ 10 7 cfu of NCTC 10/84 GBS or isogenic ⌬cylE mutant. At the time of morbidity or the experimental end point (24 h), mice were euthanized, and blood and brain were collected. One half of the brain was homogenized and processed for Western blot analysis in radioimmune precipitation assay buffer, and the other half was homogenized and plated on THB agar plates for enumeration of cfu. GFP-LC3 transgenic mice (39) (aged 10 -12 weeks) were similarly infected with WT GBS or the ⌬cylE mutant (n ϭ 3) or injected with PBS control, and at the experimental end point (24 h), brain tissue was isolated and cryopreserved in optimal cutting temperature (O.C.T.) compound prior to immunofluorescence microscopy.
Statistical Analyses-GraphPad Prism version 5.0f was used for statistical analysis. Unpaired t tests or one-way analysis of variance was used for analysis. Statistical significance was accepted at p Ͻ 0.05.

Bacterial Infection Induces Autophagy in Brain Endothelial
Cells-To investigate autophagy activation in hBMECs, we analyzed the processing and lipidation of LC3. Upon initiation of autophagy, the cytosolic LC3-I form is converted to LC3-II, which is covalently linked to phosphatidylethanolamine and associated with autophagosomal membranes (40). Ectopically expressed LC3, which is N-terminally tagged with GFP (GFP-LC3), is diffusely distributed in the cytosol, but upon proteolysis of the C terminus and lipidation, it is recruited into autophagosomes, which are evident as fluorescent puncta. Initially, we infected hBMECs with known meningeal pathogens that are capable of invading brain endothelial cells, including GBS (5), B. anthracis (31), and S. aureus (32). Following transduction of hBMECs with an adenovirus expressing GFP-labeled LC3 (Ad-GFP-LC3) (41) and subsequent bacterial infection, we observed a significant number of LC3 puncta compared with the uninfected control (Fig. 1A). We further examined two well studied GBS clinical isolates shown to cause experimental meningitis, COH1 (serotype III) and NCTC 10/84 (serotype V) (6,10). Increased formation of LC3 puncta can be visualized after COH1 WT GBS infection in comparison with an uninfected control (Fig. 1B). Furthermore, Western blot analysis of endogenous LC3 levels in hBMECs during infection revealed a significant increase in LC3-II levels compared with LC3-I at early time points (Fig. 1, C and D). We also investigated the turnover of the autophagic adaptor protein p62/sequestome-1 (42), which interacts with LC3 and is an indicator of active autophagy (43)(44)(45)(46)(47). We observed an early decrease in p62 compared with the cytosolic marker GAPDH following GBS infection (Fig. 1, C and D), which is consistent with autophagic flux. Furthermore, analysis of our previously obtained data on the hBMEC transcriptional response following GBS infection (15) revealed a number of modulated genes within autophagy gene networks, including Atg10 and Xbp1 (data not shown). Collectively, these data indicate that autophagy in BBB endothelium is activated early during pathogen infection.
Bacterial Invasion and Toxin Production Activate Autophagy-To test whether GBS must be actively replicating to induce autophagy, we incubated hBMECs with heat-killed or paraformaldehyde-fixed GBS. As shown in Fig. 2, A-D, neither heat-killed nor paraformaldehyde-fixed bacteria were able to induce substantial LC3-II conversion or p62 turnover compared with infection with live GBS. These data demonstrate that live bacterial infection is required to induce autophagy in brain endothelium, which may involve active bacterial transcription and protein synthesis and/or bacterial uptake into host cells. We next examined a subset of GBS virulence factors known to play a role in the pathogenesis of meningitis, specifically those involved in hBMEC attachment and invasion and in toxin production. Using GBS mutant strains lacking cylE, which codes for ␤-h/c activity, and pili components pilA and pilB, which promote GBS interaction with hBMECs, we assessed their ability to induce autophagy in BBB endothelium. Infection with the ⌬pilA and ⌬pilB mutants induced LC3-II conversion and p62 turnover similar to those of WT GBS, whereas strains lacking the ␤-h/c toxin resulted in less LC3-II (Fig. 2, E and F). Similar results were observed for a ⌬cylE mutant in a different GBS WT parental background (Fig. 2, G  and H). Additionally, the GBS ⌬iagA mutant, which exhibits reduced bacterial invasion (8), resulted in less LC3-II conversion. We sought to further investigate the role of the ␤-h/c toxin in autophagy activation in brain endothelium. Compared with the WT strain, the GBS mutant lacking the ␤-h/c toxin induced significantly less LC3 puncta (Fig. 3, A and B) and reduced endogenous levels of LC3-II as well as increased p62 (Fig. 3, C, D, and E). These data suggest that GBS toxin production is required for autophagy activation in brain endothelium.
To determine whether the ␤-h/c toxin could independently induce autophagy in hBMECs, cell-free extracts from GBS were prepared in PBS plus 2% starch to extract stabilized ␤-h/c activity from the bacterial surface as described previously (6,37). Hemolytic titers were determined and were similar to that observed previously (data not shown). hBMEC monolayers were incubated with cell-free extracts from either WT or ␤-h/c mutant strains. As shown in Fig. 4, A-C, extracts containing ␤-h/c induced LC3-II conversion and p62 turnover in a dosedependent fashion, whereas extracts from a ␤-h/c mutant did not result in autophagy activation, indicating that other secreted GBS products had negligible stimulatory effects. Under these experimental conditions, the ␤-h/c extract did not result in substantial cell death (Fig. 4D). These data indicate that brain endothelium responds directly to the GBS ␤-h/c to activate autophagy pathways independent of live bacterial challenge.
The GBS ␤-h/c Promotes Autophagy Activation in Vivo-Our results suggested a prominent role for the GBS ␤-h/c toxin in promoting an autophagic response in brain endothelium. To test this hypothesis in vivo, we used our murine model of GBS hematogenous meningitis as described previously (6). Groups of mice (n ϭ 9) were infected intravenously with WT GBS (NCTC 10/84) or the isogenic ⌬cylE mutant. As we have demonstrated previously (6,37), the majority of WT-infected mice died and exhibited high bloodstream and brain bacterial loads compared with ⌬cylE-infected mice (Fig. 5, A and B). At the time of death or sacrifice, brains were harvested and processed to obtain protein lysates for subsequent Western blot analysis of LC3 and p62 proteins. We observed increased LC3-II and a concomitant decrease in p62 in the majority of WT GBS-in-fected mice compared with mice infected with the ⌬cylE mutant (Fig. 5, C and D). We also similarly infected GFP-LC3 transgenic mice (48) with WT or ⌬cylE mutant strains. Fluorescence microscopy of representative brains revealed increased GFP-LC3 puncta in brain tissue following GBS WT infection compared with mutant infection or PBS injection (Fig. 6A). Additionally, we observed that GBS co-localized with GFP-LC3 within endothelial structures (Fig. 6B) and that GFP-LC3 co-localized with von Willebrand factor, a marker of endothelial cells (Fig. 6C). These results confirm that autophagy is activated in response to GBS infection and the ␤-h/c toxin in vivo.
Autophagy Contributes to Bacterial Clearance-To examine the importance of autophagic activation in combating intracellular GBS, we first used an MEF cell line that is deficient in ATG5 (35). This protein is required for autophagy initiation, and cell lines lacking ATG5 are not able to form the ATG5-ATG12-ATG16L1 initiation complex required to elongate the developing autophagosome (35). Following GBS infection and antibiotic treatment to remove extracellular bacteria (see "Experimental Procedures"), intracellular GBS was enumerated in WT and ATG5 KO MEFs. In the absence of ATG5, significantly more intracellular GBS was recovered (Fig. 7A). To establish the protective role of autophagy in brain endothelium, we first pretreated hBMECs with rapamycin, which is known to induce autophagy through its ability to inhibit the protein kinase mTOR complex 1, a pivotal negative regulator of autophagy (49 -51). The intracellular bacterial load after pretreatment with rapamycin was significantly lower than in untreated controls (Fig. 7B). This suggests that stimulation of autophagy in hBMECs limits GBS intracellular survival. We also utilized bafilomycin A1, an antibiotic known to inhibit autophagosome-lysosome fusion (52), and observed that bafilomycin pretreatment led to a significant increase in the num- Image analysis was performed using ImageJ software to determine LC3-II/LC3-I and p62/GAPDH ratios. All experiments were repeated at least three times in triplicate; data represent the mean Ϯ S.D. from a representative experiment. *, p Ͻ 0.05; **, p Ͻ 0.005; ****, p Ͻ 0.0001. Error bars represent S.D. DIC, differential interference contrast. DECEMBER 26, 2014 • VOLUME 289 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 35715
ber of intracellular GBS cfu recovered (Fig. 7B). To corroborate these data, we utilized siRNAs directed toward ATG5 and ATG12 to inhibit autophagy in hBMECs during GBS infection. During treatment with siATG5, we observed reduced levels of ATG5 and LC3-II conversion during GBS infection and a slight, although significant, increase in recovered intracellular GBS compared with treatment with the siRNA scrambled control (Fig. 6, C and F). Similar results were obtained when ATG12 was silenced in hBMECs (Fig. 6, D and G). To further examine elimination of genes critical for autophagy activation but not LC3 functionality, we knocked down FIP200, a ULK1-interacting protein essential for autophagy induction (53). We observed that knockdown of FIP200 did not similarly result in increased recovery of intracellular GBS (Fig. 3, E and H). These results suggest that LC3 recruitment and ensuing activation are key contributors to limiting intracellular GBS survival.
Visualization of Intracellular GBS-Thus far our results suggest that although the autophagic pathway is activated in brain endothelium during GBS infection, it may play a limited role in reducing the intracellular pool of GBS. Thus we sought to determine whether GBS was found in double membrane structures, which are characteristic of autophagosomes. Using transmission electron microscopy (TEM) analysis, we observed that GBS resides primarily in single membrane-bound compartments within hBMECs and was never free in the cytoplasm at 4 h (Fig. 8A) and up to 24 h (data not shown) postinfection. Single membrane structures were often damaged, and a small population had double or multiple membranes (Fig. 8, A and B). Using confocal microscopy, we further analyzed GBS co-localization with LC3 in hBMECs over time as described under "Experimental Procedures." By 4 h, we observed that ϳ40% of intracellular GBS co-localized with LC3 (Fig. 8, C and D), sug- gesting that less than half of the intracellular pool may be shuttled into the autophagic pathway.

DISCUSSION
To penetrate the CNS, bacterial pathogens may directly invade BBB endothelium and traverse the barrier in a process called transcytosis. We and others have demonstrated that meningeal pathogens, including GBS, are capable of transcellular passage, but the exact mechanisms of intracellular survival and trafficking are not well understood. It is likely that intracellular host defenses may be activated to combat invasive bacteria, but it is unknown whether autophagy in brain endothelium represents an important BBB defense mechanism or whether meningeal pathogens ultimately thwart or utilize this pathway for survival and BBB traversal. Our results provide new evidence that autophagy/xenophagy is activated in brain endothelium during GBS infection and contributes to limiting intracellular organisms. We demonstrate that inactivation of ATG5  . GBS ␤-h/c extract independently activates autophagy. A, Western blot analysis was performed on cell lysates collected from hBMECs incubated for 2 h with the indicated dilutions of cell-free ␤-h/c extracts recovered from NCTC 10/84 WT or the isogenic ⌬cylE mutant. B and C, image analysis was performed using ImageJ software to determine LC3-II/LC3-I and p62/GAPDH ratios. D, hBMECs were incubated with the indicated dilutions of cell-free ␤-h/c extracts recovered from NCTC 10/84 WT for 2 h and then stained with trypan blue to measure cell viability. All experiments were repeated at least three times in triplicate; data represent the mean Ϯ S.D. from a representative experiment. *, p Ͻ 0.05. Error bars represent S.D. DECEMBER 26, 2014 • VOLUME 289 • NUMBER 52 FIGURE 5. Autophagy is induced in vivo following GBS infection. CD1 male mice were injected intravenously with WT (n ϭ 9) or ⌬cylE GBS strain (n ϭ 9) or injected with PBS (n ϭ 2). A, bacterial counts (cfu) in mouse brain and in blood at the time of death. B, Kaplan-Meier survival plot. Significance was assessed using a log rank (Mantel-Cox) test. C, Western blot analysis was performed on protein lysates from brain harvested from mice infected with NCTC 10/84 and isogenic ⌬cylE mutant for LC3-I, LC3-II, and p62. D, image analysis was performed using ImageJ software to determine LC3-II/LC3-I ratios. **, p Ͻ 0.005; ****, p Ͻ 0.0001. Error bars represent S.D. FIGURE 6. Visualization of autophagy activation in brain endothelium. A, representative brain samples from GFP-LC3 transgenic mice infected with NCTC 10/84 GBS and isogenic ⌬cylE mutant. GFP-LC3 puncta were observed in brain tissue in WT-infected mice compared with mice infected with the ⌬cylE mutant or injected with PBS. Scale bar, 20 m. B, immunofluorescence for GBS in WT-infected GFP-LC3 transgenic mice. GBS co-localizes with GFP-LC3 within endothelial portions of the brain. Scale bar, 10 m. C, immunofluorescence for von Willebrand factor (VWF) in GFP-LC3 transgenic WT-infected mice demonstrates that endothelial cells are producing active GFP-LC3. Scale bar, 10 m. and ATG12, two key autophagy proteins involved in LC3 processing and autophagosome formation, resulted in increased GBS survival. Conversely, we observed that activation of autophagy using rapamycin prior to GBS infection restricts the recovery of intracellular bacteria. Analysis of various GBS mutants deficient in factors previously determined to play a role in disease pathogenesis led to the discovery of the ␤-h/c toxin as a key virulence factor associated with autophagy activation. Our data suggest that ␤-h/c secretion is sufficient to promote an autophagic response in brain endothelium, a response that is aimed to potentially eliminate intracellular GBS.

Autophagy Activation in Brain Endothelium by GBS
Our results demonstrate that hBMECs respond to GBS infection with a robust autophagic response as we observed activation of LC3 and autophagic clearance of the key autophagy adaptor protein p62. Activation was dependent on live bacteria as heat-killed or formalin-fixed GBS failed to induce conversion to LC3-II (Fig. 2, A-D). This led us to investigate which bacte-rial virulence determinants may be responsible for autophagy activation in BBB endothelium. We observed that factors associated with GBS invasion into hBMECs such as properly anchored lipoteichoic acid may contribute to autophagy activation. We also found that production of the GBS ␤-h/c toxin was an important contributor to autophagy activation in hBMECs. This is a well characterized GBS virulence factor shown to promote GBS invasion and intracellular survival in a variety of cell types as well as immune activation and disease progression (6,37,54,55). Compared with WT GBS infection, we found that infection with a ␤-h/c-negative mutant resulted in significantly less autophagy induction in hBMECs in vitro and in brain tissue of infected mice. Activation appears to not require bacterial invasion or even the bacterial cell itself as extracts containing ␤-h/c activity also directly stimulated autophagy activation. Thus ␤-h/c appears to be a key mediator in provoking an acute autophagic response in the brain endothelium and may be an important contributor to disease progression. Whether this  DECEMBER 26, 2014 • VOLUME 289 • NUMBER 52 action is elicited by direct interaction of the toxin with endothelial signal transduction systems or activation is a secondary result of cellular injury that is mediated by the toxin remains to be elucidated. Recent studies investigating other bacteriumhost interactions have shown that autophagy can be stimulated by toxins from Vibrio cholerae (56) and B. anthracis (57) and pore-forming toxins from GAS (58) and S. aureus (59). Interestingly, it has been recently suggested that hemolytic and cytolytic activity of GBS is due to the ornithine rhamnolipid pigment and not due to a pore-forming protein toxin (60). Furthermore, this associated carotenoid pigment has been shown to promote GBS intracellular survival in phagocytic cells (37). Thus it will be of interest to determine the exact mechanism of autophagy activation by the GBS ␤-h/c.

Autophagy Activation in Brain Endothelium by GBS
Research devoted toward the understanding of how antibacterial autophagy may defend against intracellular microbes has recently become of increasing interest. Classically cytosolic intracellular bacterial pathogens such as Listeria, Shigella, and GAS that disrupt phagosomal membranes and escape from these vesicles may be targeted for sequestration by autophagy, leading to their degradation (20,61,62). GAS escapes the endocytic pathway and enters the cytoplasm using a pore-forming cytolysin, streptolysin O (22,63). Cytosolic GAS is then isolated into autophagosomes and rapidly undergoes lysosomal degradation. Autophagy may also target vesicular bacteria as is the case for Mycobacterium and Salmonella (20,62). During infection of epithelial cells, S. typhimurium damages and escapes endosomal membranes, becomes ubiquitinated, and is recognized by autophagic adaptor proteins p62, nuclear dot protein 52, and optineurin for eventual binding to LC3 (64 -66). Using TEM, we have visualized GBS mainly within single membranebound vesicles, which is consistent with early observations and . Intracellular bacteria were quantified according to the intracellular structure in which they resided (n ϭ 25). Bacteria are present in membrane-bound vesicles, damaged membranes, multiple membranes, and putative autophagic structures as indicated by arrowheads. Scale bar, 500 nm. C, transfection of an mCherry-LC3 plasmid into hBMECs was performed as described under "Experimental Procedures." hBMECs were infected with GFP COH1 WT for 4 h prior to treatment with extracellular antibiotics 1, 2, 4, and 6 h postinfection. Quantification of the amount of GFP COH1 WT co-localizing with mCherry-LC3 was gathered by counting at least 100 cells with intracellular GBS, and data are means Ϯ S.E. from a representative experiment performed in triplicate. D, representative confocal microscopic images from 4 h postinfection. Scale bar, 10 m. Error bars represent S.D. DIC, differential interference contrast.
images of intracellular GBS in hBMECs (5). We have not observed GBS free in the cytosol even at later time points, although in some cases vesicle membranes appear to be disrupted (Fig. 8A). It has been observed that penicillin may gain access to the cytoplasm of eukaryotic cells (67), which may potentially kill GBS released into the cytoplasm. Although we cannot exclude this possibility, we should note that TEM analysis was performed in the absence of any antibiotics. Furthermore, we did observe a 30% increase in recovered GBS when using only gentamycin (data not shown), but this result is complicated by the fact that gentamycin alone was not as effective at killing extracellular GBS (data not shown). Our findings also suggest that intracellular GBS may traffic through other endosomal pathways. We have observed that GBS can traffic into Rab5-and Rab7-positive endosomes (data not shown), which is consistent with our TEM results suggesting the involvement of the endocytic pathway. The exact mechanisms and key players for recognition of intracellular vesicular GBS are incompletely understood as is the question of whether the ␤-h/c toxin is responsible for the vacuolar membrane damage we observed by TEM.
Pathogenic bacteria that survive within host cells utilize different strategies to avoid being killed in an autophagolysosome (68). These defensive mechanisms include resistance to autophagic engulfment, disruption of trafficking to the lysosome, and resistance to lysosomal killing. S. typhimurium has the capacity to survive and replicate intracellularly due to modulation of amino acid starvation-triggered mTOR inhibition that activates autophagy (69,70). Recently, it has been reported that a clinically relevant serotype of GAS is able to degrade adaptor proteins, including p62, through secretion of a surfaceassociated protease, SpeB, thereby allowing GAS to persist in the cytoplasm (67). Other pathogens such as S. flexneri and L. monocytogenes avoid autophagic recognition by producing virulence factors that bind key autophagy-related proteins such as ATG5 and the ARP2/3 complex, thereby allowing for intracellular persistence and dissemination (46,71,72). Although our results demonstrate that autophagy is activated in BBB endothelium and host factors such as ATG5 and ATG12 contribute to bacterial clearance, GBS is not completely eliminated.
TEM analysis suggests that GBS is not readily sequestered in autophagosomes; however, up to ϳ40% of intracellular GBS co-localized with LC3 at 4 h postinfection (Fig. 8C). It has been demonstrated that LC3 can also be recruited to single membrane phagosomes or vesicles to assist in lysosomal fusion in a process denoted as LC3-associated phagocytosis (LAP) (73)(74)(75)(76). Pathogens such as Burkholderia pseudomallei and Mycobacterium marinum have elicited LAP features in RAW264.7-GFP-LC3 macrophages (77)(78)(79). There is no clear indicator of LAP; however, there has been a universal consensus that the ATG5-ATG12-ATG16L1 complex is required for LAP induction (75,80,81). Defining whether macroautophagy or LAP occurs in the case of pathogen invasion has yet to be differentiated using GFP-conjugated LC3, but modulation of a subset of ATG proteins such as ATG5 and electron microscopic analysis of membrane-bound bacteria have been able to shed light on these subtle differences. Interestingly, inactivation of the ULK1-interacting protein FIP200, which is essential for the for-mation of the isolation membrane during autophagy but not recruitment of LC3 to membranes, did not impact recovery of intracellular GBS. This suggests that LAP may be critical for directing GBS to degradative compartments. Future studies are aimed to further elucidate the contribution of LAP to GBS uptake and intracellular trafficking.
In summary, we have demonstrated for the first time that the BBB endothelium activates autophagy in response to the meningeal pathogen GBS. We present evidence that this pathway may contribute to host cellular defense by controlling the intracellular pool of GBS. In addition, we have identified the GBS pore-forming toxin as the molecular trigger for autophagy activation. However, it will be important to investigate the autophagy adaptor proteins involved in the recognition of intracellular GBS, the impact of infection on autophagic flux, and whether other bacterial factors promote autophagy evasion and GBS intracellular survival. Ongoing studies on the modulation of host autophagy by meningeal pathogens are critical for understanding the host defense of the BBB and developing preventative therapies for CNS infection.