Differential Ubiquitination as an Effective Strategy Employed by the Blood-Brain Barrier for Prevention of Bacterial Transcytosis

ABSTRACT

The protective mechanisms of blood-brain barrier (BBB) prohibiting entry of pathogens into central nervous system (CNS) are critical for maintenance of brain homeostasis. These include various intracellular defense mechanisms that are vital to block transcytosis of neurotropic pathogens into the CNS. However, mechanistic details of coordination between these defense pathways remain unexplored. In this study, we established that BBB-driven ubiquitination acts as a major intracellular defense mechanism for clearance of Streptococcus pneumoniae, a critical neurotropic pathogen, during transit through BBB. Our findings suggest that the BBB employs differential ubiquitination with either K48- or K63-ubiquitin (Ub) chain topologies as an effective strategy to target S. pneumoniae toward diverse killing pathways. While K63-Ub decoration triggers autophagic killing, K48-Ub directs S. pneumoniae exclusively toward proteasomes. Time-lapse fluorescence imaging involving proteasomal marker LMP2 revealed that in the BBB, the majority of the ubiquitinated S. pneumoniae was cleared by proteasome. Fittingly, inhibition of proteasome and autophagy pathway led to accumulation of K48-Ub- and K63-Ub-marked S. pneumoniae, respectively, and triggered significant increases in intracellular S. pneumoniae burden. Moreover, genetic impairment of either K48- or K63-Ub chain formation demonstrated that although both chain types are key in disposal of intracellular S. pneumoniae, K48-Ub chains and subsequent proteasomal degradation have more pronounced contributions to intracellular S. pneumoniae killing in the BBB. Collectively, these observations, for the first time, illustrated a pivotal role of differential ubiquitination deployed by BBB in orchestrating a symphony of intracellular defense mechanisms for interception and degradation of S. pneumoniae, blocking its entry into the brain, which could be exploited to prevent bacterial CNS infections.

IMPORTANCE The blood-brain barrier (BBB) represents a unique cellular barrier that provides structural integrity and protection to the CNS from pathogen invasion. Recently, ubiquitination, which is key for cellular homeostasis, was shown to be involved in pathogen clearance. In this study, we deciphered that the BBB deploys differential ubiquitination as an effective strategy to prevent S. pneumoniae trafficking into the brain. The different ubiquitin chain topologies formed on S. pneumoniae dictated the selection of downstream degradative pathways, namely, autophagy and proteasomes, among which the contribution of the proteasomal system in S. pneumoniae killing is more pronounced. Overall our study revealed how the BBB deploys differential ubiquitination as a strategy for synchronization of various intracellular defense pathways, which work in tandem to ensure the brain’s identity as an immunologically privileged site.

INTRODUCTION

The human body is composed of multiple cellular barriers that primarily safeguard deeper tissues that are susceptible to infection by pathogens. Among these host cellular barriers, the blood-brain barrier (BBB) stands out, as this specialized structural and functional barrier aids maintenance of central nervous system (CNS) homeostasis by blocking entry of toxic metabolites and pathogens circulating in the bloodstream into the CNS, keeping it immunologically sterile (1, 2). Despite the protective architecture of BBB, some neurotropic pathogens, such as Streptococcus pneumoniae (3), Haemophilus influenzae (4), Escherichia coli K1 (5), etc., invade brain endothelial cells and breach the BBB transcellularly to cause meningitis. Transcellular invasion into the endothelial cells also provides a safe refuge to the pathogens, protecting them from the systemic immunity and facilitating relapse of bacterial infection (6). Due to all these factors, regardless of advances in antibiotic therapy and vaccines, bacterial meningitis still remains a burning global health issue with a high rate of morbidity and mortality (7).

Streptococcus pneumoniae (pneumococcus), a Gram-positive alpha-hemolytic bacteria, is the most common cause of bacterial meningitis worldwide (8). Employing a plethora of surface proteins like PspC/CbpA (9, 10) RrgA (Pilus-1) (11), and phosphorylcholine (12), S. pneumoniae adheres to the brain endothelium and subsequently invades it via various endocytic pathways (13). Following invasion, S. pneumoniae thwarts intracellular killing by a classical endolysosomal pathway by secreting a cholesterol-dependent pore-forming toxin, pneumolysin (Ply). Escape of S. pneumoniae from lysosomal killing by virtue of Ply elicits autophagy, facilitating intracellular degradation of S. pneumoniae and arming the host with an alternate strategy to curb bacterial dissemination. However, in a substantial fraction of intracellular S. pneumoniae, Ply can pierce the autolysosomal membrane, leading to its cytosolic escape (14). Clearance of this cytosolic S. pneumoniae pool that dodged host intracellular killing pathways is critical to avert bacterial dissemination into the CNS.

Recently, we and others have reported that in brain endothelial cells, cytosol-exposed S. pneumoniae gets ubiquitinated, which led to its clearance and limited its entry into the brain (14, 15). Apart from facilitating pathogen clearance, the ubiquitin (Ub) system plays a major role in a wide range of cellular functions, like signal transduction, endosome trafficking, protein sorting, protein degradation, etc. (16). The ubiquitin molecule attached to the substrate by the action of E1-, E2-, and E3-conjugating enzymes can form unique structures on substrates by polyubiquitination, utilizing either the N-terminal methionine residue (M1) or any lysine residue among the eight residues it possesses (K6, K11, K27, K29, K33, K48, and K63) (17). The unique ubiquitin topologies thus formed are selectively recognized by various polyubiquitin binding effector proteins (18), subsequently governing downstream signaling events (16).

Although ample evidence suggests the crucial role of ubiquitination in pathogen clearance (14, 19), the mechanistic details regarding how ubiquitination triggers bacterial killing and the role of specific chain topologies formed on bacteria and their subsequent contribution toward choice of bacterial killing pathways remain elusive. This is particularly true for brain endothelium, where ubiquitination could play a major role in its sentinel function by blocking pathogen entry into the CNS. In this study, using Streptococcus pneumoniae as a model pathogen, we, for the first time, portray a comprehensive picture of ubiquitination of a pathogen with various ubiquitin chain topologies during transit through the BBB. Our study demonstrated that the differential ubiquitination governing diverse killing pathways, particularly the K48-ubiquitination-directed proteasomal pathway, is an efficient strategy deployed by the brain endothelium to curb pathogen transit to the CNS.

RESULTS

Ubiquitination triggers both lysosome dependent and independent clearance of S. pneumoniae in brain endothelium.

Following invasion in brain endothelium and during transcytosis, catastrophic damage to the pneumococcus-containing vacuolar (PCV) membrane by the pore-forming toxin pneumolysin triggers ubiquitination of S. pneumoniae. However, mechanisms adopted by the brain endothelial cells to clear such ubiquitinated S. pneumoniae remain elusive. To determine this, we first tested whether ubiquitinated PCVs associate with lysosomes for subsequent degradation. In classical bacterial killing pathways, bacteria containing endosomes gradually get acidified and finally fuse with lysosomes for ultimate destruction (20). We used LysoTracker as an acidotropic probe to evaluate acidification of the vacuoles (Fig. 1A and B). Surprisingly, confocal microscopy analysis revealed that in brain endothelial cells only a minor subset of the ubiquitinated S. pneumoniae associated with LysoTracker (Ub⁺ Lyso⁺) (35.61% ± 4.9%), while the majority (64.38% ± 4.89%) did not get sequestered in acidified vacuoles (Ub⁺ Lyso⁻) (Fig. 1C). Since nonacidification or neutralization of PCVs can occur even after maturation and fusion with lysosomes, due to activity of different bacterial virulence factors, including cholesterol-dependent cytolysins (21), we evaluated accumulation of cathepsin B, a lysosomal cysteine protease in the ubiquitinated PCVs (Fig. 1D and E). Our results suggest that a significant subset (61.3% ± 2%) of the ubiquitinated PCVs did not accumulate cathepsin B (Fig. 1F). Simultaneous presence of S. pneumoniae in acidified lysosome-like vacuoles as well as nonacidified compartments implies that brain endothelial cells employ two parallel pathways for clearance of ubiquitinated S. pneumoniae. While very few S. pneumoniae cells are driven toward the lysosome-dependent killing, the major subset of ubiquitinated S. pneumoniae clearance involves a hitherto speculative nonlysosomal pathway.

FIG 1.

Ubiquitination triggers both lysosome dependent and independent clearance of S. pneumoniae (SPN here) in brain endothelium. (A) Confocal micrograph showing association of ubiquitin (red)-positive S. pneumoniae (green) with LysoTracker (pink) at 9 h p.i. hBMECs were infected with GFP-expressing S. pneumoniae and stained with antiubiquitin Ab and the acidotropic dye LysoTracker deep red. The arrow designates bacteria shown in insets. Scale bar, 5 μm. (B) Fluorescent line scan across the yellow line in the merged inset in panel A. (C) Quantification of ubiquitinated S. pneumoniae’s association with LysoTracker in hBMECs at 9 h p.i. n ≥ 100 bacteria per coverslip. Data are presented as means ± SD from triplicate experiments. Statistical analysis was performed using Student's t test. **, P < 0.005. (D) Confocal micrograph showing association of ubiquitin (red)-positive S. pneumoniae (blue) with lysosomal enzyme cathepsin B (green) at 9 h p.i. hBMECs stably expressing mStrawberry-Ub (red) were infected with S. pneumoniae and stained with anti-cathepsin B Ab. The arrow designates bacteria shown in insets. Scale bar, 5 μm. (E) Fluorescent line scan across the yellow line in the merged inset in panel D, depicting colocalization of S. pneumoniae with cathepsin B. (F) Quantification of colocalization of ubiquitinated S. pneumoniae with cathepsin B in hBMECs at 3 h p.i. Data are presented as means ± SD from triplicate experiments. n ≥ 100 bacteria per coverslip. Statistical analysis was performed using Student's t test. **, P < 0.005.

Ubiquitinated S. pneumoniae is cleared by both autophagy and proteasomal pathways.

Given that clearance of the majority of ubiquitinated S. pneumoniae required a lysosome-independent cellular pathway, we sought to test the host proteasomal system as an alternate degradative pathway employed by BBB for tackling the invaded S. pneumoniae. To accomplish this, we generated a transfected hBMEC line stably expressing mStrawberry-Ub and GFP-LMP2, one of the subunits of 26S immunoproteasome by transfecting cells with pMRX-IRES-Puro-GFP-LMP2 and pMRX-IRES-BLAST-mStw-Ub plasmids. Time-lapse live-cell imaging of ubiquitinated S. pneumoniae labeled with DRAQ5 for an elongated period surprisingly revealed accumulation of GFP-LMP2 to approximately 62.5% ± 2.5% of the ubiquitinated S. pneumoniae (Fig. 2E). We also observed gradual degradation of the ubiquitinated S. pneumoniae while being associated with GFP-LMP2 (Fig. 2A and B; see also Movie S1 in the supplemental material), establishing key roles of the proteasome in S. pneumoniae clearance in brain endothelium. Live-cell imaging was also performed on an hBMEC cell line stably expressing both mStrawberry-Ub and LC3-GFP to decipher the contribution of the autophagy-lysosome pathway in S. pneumoniae clearance. We observed that approximately 32.5% ± 3.2% mStrawberry-Ub-marked, DRAQ5-labeled S. pneumoniae displayed sequestration by LC3-GFP ring-like structures with time and eventual degradation while remaining confined within these LC3-GFP structures (Fig. 2C and D; Movie S2). To further evaluate which process among autophagy and proteasomal pathways is more efficient in tackling intracellular S. pneumoniae burden during its trafficking through the BBB, we treated HBMECs with either 3-methyladenine (3-MA; autophagy inhibitor) or MG132 (26S proteasome inhibitor) and performed an intracellular survival assay. Our results suggest that at noncytotoxic concentrations (Fig. S1A), both 3-MA and MG132 significantly improved survival ability of S. pneumoniae in brain endothelium without affecting internalization, corroborating our time-lapse microscopic data (Fig. S1B to D). Intriguingly, inhibiting proteasomal machinery led to 3.8- and 2.5-fold survival advantage to the intracellular S. pneumoniae compared to 2.6- and 1.9-fold survival benefit provided by blocking cellular autophagy at 6 and 12 h postinfection (p.i.), respectively (Fig. 2F). Collectively, these results suggest that although both autophagy and proteasomal degradation are essential for promoting ubiquitinated S. pneumoniae clearance in brain endothelium, contribution of the proteasomal pathway is more pronounced.

FIG 2.

Ubiquitinated pneumococcus is cleared by both autophagy and proteasomal pathways. (A and C) Representative time-lapse montage for degradation of ubiquitinated S. pneumoniae in a proteasome (A)- and autophagy (C)-dependent manner. hBMECs stably expressing mStrawberry-Ub (red) and GFP-LMP2 (green) or LC3-GFP (green) were infected with DRAQ5-stained WT S. pneumoniae (magenta). Live imaging of infected cells was performed for extended hours beginning at 6 h p.i. under a confocal microscope and time-lapse series with intervals of 30 min are represented. The stills in panels A and C correspond to movies shown as Movies S1 and S2, respectively. Scale bars, 2 μm. (B and D) Temporal quantification of ubiquitin, S. pneumoniae, and LMP2 (B) or LC3 (D) fluorescence intensities, relative to fluorescence in the cytosol (Cyt) (for Ub and LMP2 or LC3) or fluorescence signal of hBMEC nuclei (Nuc) (for S. pneumoniae). Relative mean intensity for LMP2 or LC3 (I_PCV/I_Cyt) in the case of ubiquitinated S. pneumoniae displayed values of >1.0, depicting proteasome (B)- or autophagy (D)-dependent killing of S. pneumoniae, respectively. In both cases relative mean intensity values for bacteria (I_{S. pneumoniae}/I_Nuc) remains >1.0 for substantially long period of time before finally disappearing. (E) Percentage of ubiquitinated S. pneumoniae undergone LC3-dependent or LMP2-mediated killing as examined by time-lapse imaging of hBMECs stably expressing Ubq-mStrawberry and LC3-GFP or GFP-LMP2, respectively. n = 50. (F) Comparison of intracellular survival of S. pneumoniae in hBMECs following pretreatment with either 3-MA (1 mM), MG132 (10 μM), or dimethyl sulfoxide (DMSO) (vehicle). Fold change in intracellular survival at indicated time points postinfection is calculated by multiplicity of percent S. pneumoniae survived (relative to 0 h) in the presence of 3-MA or MG132 compared to DMSO. Data are presented as means ± SD from triplicate experiments. Statistical analysis was performed using two-way analysis of variance (ANOVA) (Bonferroni test). **, P < 0.005; ***, P < 0.001.

Differential ubiquitination of S. pneumoniae with varied chain topologies governs choice of host degradative pathways.

Since specific ubiquitin chain topologies formed on the substrate dictate downstream signaling pathways triggering precise degradation pathways, we aimed to decipher the composition of the ubiquitin coat covering intracellular S. pneumoniae. We hypothesized that a differential ubiquitin coat is the decisive factor in shunting the ubiquitinated S. pneumoniae toward autophagy or proteasomal pathway in brain endothelium. To accomplish this, intracellular S. pneumoniae was stained with different ubiquitin linkage-specific antibodies (Ab). Among the different ubiquitin linkage topologies, K48-Ub and K63-Ub chains are primary signals for degradation (22, 23). Interestingly, we observed that brain endothelium decorates S. pneumoniae with both types of ubiquitin chain topologies (Fig. 3A to D). However, quantification of such events at various time points postinfection revealed significant differences in the kinetics of their contribution toward total ubiquitinated S. pneumoniae, especially at later stages of infection. At 6 h p.i., K63-Ub S. pneumoniae had reduced drastically from 68.15% ± 6.25% to 43.38% ± 4.81%, subsequently to 31.54% ± 3.33% at 9 h, and finally to 17.9% ± 3.69% at 12 h p.i. However, unlike K63-Ub chains, contribution of K48-Ub chains to the total ubiquitinated S. pneumoniae remained fairly constant until extended time of infection (70.87% ± 1.94%, 67.87% ± 5.67%, 64% ± 2% and 73.64% ± 3.18% at 3, 6, 9, and 12 h p.i., respectively) (Fig. 3E). This could be because with increased time, progressively greater damage to the PCV membrane could lead to a higher degree of decoration with K48-Ub chain topology. Alternatively, faster degradation kinetics of K63-Ub chain types could have triggered its drastic reduction with time. S. pneumoniae was also found to be decorated with K48- and K63-Ub chain types in lung epithelial cells (data not shown). This suggests that differential ubiquitination can be employed as a general strategy by both host epithelial and endothelial cells for protection against pneumococcal infection.

FIG 3.

S. pneumoniae is ubiquitinated with different ubiquitin chain topologies in brain endothelium. (A and C) Confocal micrographs showing association of ubiquitinated (red) S. pneumoniae (magenta) with specific ubiquitin chain types (K48 in panel A and K63 in panel B) at 6 h p.i. hBMECs stably expressing mStrawberry-Ub was infected with S. pneumoniae and stained with anti-K48-Ub (A) or anti-K63-Ub (C) linkage-specific Abs. The arrow designates event shown in insets. Scale bar, 5 μm. (B and D) Fluorescent line scan across the yellow line in the merged inset in panels A and C are shown in panels B and D, respectively. (E) Contribution of K48-Ub- and K63-Ub-decorated S. pneumoniae toward total ubiquitinated S. pneumoniae at 3, 6, 9, and 12 h p.i. n ≥ 100 ubiquitinated bacteria per coverslip. Data are represented as means ± SD from triplicate coverslips. Statistical analysis was performed using one-way ANOVA (Tukey's multiple-comparison test). ns, nonsignificant; *, P < 0.05; ***, P < 0.001.

We then aimed to determine the role of K63-Ub and K48-Ub chains in driving intracellular S. pneumoniae toward autophagy or a proteasomal system, the two established pathways involved in clearance of ubiquitinated intracellular pathogens. LC3-GFP and GFP-LMP2 were used as markers for autophagy and proteasomal system, respectively. We observed clear association of LC3 with K63-Ub-marked S. pneumoniae (Fig. 4C and D; Fig. S2C and D), while K48-ubiquitinated S. pneumoniae exhibited exclusive accumulation of GFP-LMP2 signal (Fig. 4A and B; Fig. S2A and B). Quantification of such events revealed that majority of K48-Ub-marked S. pneumoniae (57.33% ± 2.51%) was found to be associated with GFP-LMP2, while a minor population (9.32% ± 2.3% and 0%) exhibited colocalization with LC3-GFP (Fig. 4E). Further, MG132 treatment of the hBMECs to impair proteasomal activity caused accumulation of K48-Ub S. pneumoniae without affecting degradation of K63-Ub-positive S. pneumoniae (Fig. 4G), implying that in brain endothelium, K48-Ub drives S. pneumoniae toward proteasomal killing. On the contrary, 45.76% ± 3.28% of the K63-Ub S. pneumoniae population was found to be colocalized with LC3-GFP, while only 8.33% ± 1.52% showed association with GFP-LMP2 (Fig. 4F). Additionally, treatment of hBMECs with 3-methyladenine led to 2-fold significant increase in the K63-Ub-marked S. pneumoniae and showed no effect on the K48-ubiquitinated S. pneumoniae (Fig. 4G). These observations collectively demonstrate that brain endothelial cells employ K63 ubiquitination to preferentially target intracellular S. pneumoniae toward autophagic killing by sequestering it inside LC3-marked autophagosome, while K48 ubiquitination promotes S. pneumoniae killing via proteasomal machinery. We also evaluated the association of encapsulated S. pneumoniae strain TIGR4 and its isogenic capsule mutant (TIGR4:Δcps) with K48/LMP2 and K63/LC3. No difference was observed in S. pneumoniae–K48-LMP2 (Fig. S3A) and S. pneumoniae–K63-LC3 (Fig. S3B) association levels between wild-type (WT) TIGR4 and isogenic capsule mutant derivative. Similarly, no difference was observed in degree of association of K63/LC3 with piliated (TIGR4) or nonpiliated (R6) S. pneumoniae strains (Fig. 4F, Fig. S3B). However, association with proteasomes was higher in cases of nonpiliated S. pneumoniae strain R6 (57.33%) (Fig. 4E) than for piliated strain TIGR4 (29.56%) (Fig. S3A). This could be attributed to higher expression of the pore-forming toxin pneumolysin by S. pneumoniae strain D39 and its nonencapsulated derivative R6 compared to TIGR4 (14). Higher expression of pneumolysin would trigger catastrophic damage to the endosomal membrane, facilitating S. pneumoniae’s escape to hBMEC cytosol, where it would get marked with K48-Ub, followed by proteasomes. Collectively, these showcased that neither capsule nor pilus components, such as pilus adhesion RrgA, play a role in ubiquitination-dependent selection of degradative pathways, such as proteasome or autophagy, in brain endothelium.

FIG 4.

Differential ubiquitination of S. pneumoniae with varied chain topologies governs choice of host degradative pathways. (A and C) Confocal micrographs showing association of K48-ubiquitinated (red) S. pneumoniae (blue) with LMP2 (green) (A) or K63-ubiquitinated (magenta) S. pneumoniae (blue) with LC3 (green) (C) in hBMEC at 9 h p.i. hBMECs stably expressing GFP-LMP2 (A) or LC3-GFP (C) were infected with Hoechst-stained S. pneumoniae and stained with anti-K63-Ub and anti K48-Ub linkage-specific Abs. The arrow designates event shown in insets. Scale bar, 5 μm. (B and D) Fluorescent line scans across the yellow line in the merged inset in panels A and C are shown in panels B and D, respectively. (E and F) Quantitation of association of LC3 and LMP2 with K48-Ub (E)- or K63-Ub (F)-positive S. pneumoniae. n ≥ 50 K48/K63-ubiquitinated bacteria per coverslip. Data are represented as means ± SD from triplicate hBMEC cultures. Statistical analysis was performed using unpaired Student's t test. ***, P < 0.001. (G) Quantification of S. pneumoniae ubiquitinated with K48-Ub or K63-Ub chain topologies at 9 h p.i. following pretreatment of hBMECs with 3-MA (1 mM) or MG132 (10 μM). n ≥ 50 K48/K63-ubiquitinated bacteria per coverslip. Data are represented as means ± SD from triplicate coverslips. P values are from one-way ANOVA with Dunnett’s multiple-comparison test. ns, nonsignificant; ***, P < 0.001.

K48-Ub chains contribute primarily to controlling intracellular S. pneumoniae burden in brain endothelium.

After deciphering the composition of the ubiquitin coat of PCVs and their individual role, we next investigated the comparative contribution of different ubiquitin chain topologies in promoting clearance of intracellular S. pneumoniae during its passage through the brain endothelium. To accomplish this, we inhibited formation of K48 and K63 ubiquitin chains individually in hBMECs by expressing Ub^K48R and Ub^K63R mutant ubiquitins from pLKO-Puro-tet-Ub^K48R-GFP and pLKO-Puro-tet-mStrawberry-Ub^K63R plasmids, respectively. Given that ubiquitination is indispensable for maintenance of cellular homeostasis and both stable expression of mutant ubiquitin or deletion of WT ubiquitin is lethal for host cell, we expressed mutant ubiquitins under a tetracycline-inducible promoter using a Tet-ON/OFF system in brain endothelial cells. This ensured that only in the presence of tetracycline, cells expressed mutant Ub (Ub^KxR) along with Ub^WT, which is expressed from the genome. We hypothesized that simultaneous expression of Ub^WT and Ub^KxR in hBMECs would prematurely thwart polyubiquitin chain formation due to incorporation of Ub^KxR in growing polyubiquitin chains by dominant-negative effect. The aborted ubiquitin chain topology formed cannot dictate downstream signaling, leading to inefficient degradation of the cargo and its subsequent accumulation (Fig. 5A). Visualization of hBMECs transfected with fluorescently labeled mutant ubiquitins following treatment of noncytotoxic concentrations of tetracycline (Fig. S4A) under a confocal microscope showed similar expression of Ub^K48R-GFP or mStrawberry-Ub^K63R or mStrawberry as a control (Fig. 5B). As expected, hBMECs expressing Ub^K48R-GFP and mStrawberry-Ub^K63R blocked formation of K48-Ub and K63-Ub chain topologies, respectively, and did not perturb the formation of opposite ubiquitin chain types (Fig. 5C). Critically, hBMECs expressing mutant ubiquitin variants (Ub^K48R-GFP and mStrawberry-Ub^K63R) did not show any defect in pneumococcal invasion (Fig. S4B), and they were unable to restrict intracellular S. pneumoniae as efficiently as hBMECs with only Ub^WT (Fig. 5D), confirming a vital role of ubiquitination in clearance of intracellular S. pneumoniae in brain endothelium. Further, our findings suggest that although impairment of K63-Ub chain formation promoted S. pneumoniae survival by a modest margin (1.85-fold), prevention of decoration with K48-Ub chains significantly improved intracellular survival ability of S. pneumoniae by 3.91-fold compared to Ub^WT expressing hBMEC (Fig. 5D). These findings collectively corroborate our other observations and unambiguously demonstrate that K48 ubiquitination and subsequent targeting toward proteasomal killing is an effective strategy adopted by brain endothelium to control theintracellular S. pneumoniae pool.

FIG 5.

K48-Ub chains contribute primarily to controlling intracellular S. pneumoniae burden. (A) Schematic representation of inhibition of S. pneumoniae decoration with K48-Ub or K63-Ub chain topologies by expressing Ub^K48R and Ub^K63R mutant ubiquitins under a tetracycline-inducible promoter. Incorporation of Ub^K48R and Ub^K63R variants in a growing polyubiquitin chain would thwart further elongation and subsequent signaling, affecting subsequent degradation of substrates by chain-specific pathways. (B) Confocal micrographs depicting hBMECs expressing Ub^K48R-GFP, mStrawberry-Ub^K63R, or mStrawberry following transfection with pLKO-Puro-tet-Ub^K48R-GFP, pLKO-Puro-tet-mStrawberry-Ub^K63R, and pLKO-Puro-tet-mStrawberry plasmids, respectively, in the presence or absence of tetracycline (40 μg/ml). Scale bars, 5 μm. (C) Evaluation of inhibition of K48-Ub or K63-Ub chain topology formation in hBMECs transfected with either Ub^K48R-GFP or mStrawberry-Ub^K63R mutant variants in the presence of tetracycline following immunoblotting with anti-K48-Ub or anti-K63-Ub linkage-specific Abs. GAPDH served as a loading control. (D) Intracellular survival efficiency of S. pneumoniae in hBMECs expressing Ub^K48R-GFP or mStrawberry-Ub^K63R mutant variants at 6 h were calculated as percentage survival relative to 0 h. Data are presented as means ± SD from triplicate experiments relative to 0 h. Statistical analysis was performed using two-way ANOVA (Bonferroni test). ns, nonsignificant; ***, P < 0.001.

DISCUSSION

Bacterial meningitis occurs because of uncontrolled bacterial replication in the CNS following invasion of the blood-brain barrier (BBB) (24). However, the brain endothelium functions as a sentinel to thwart easy passage of pathogens across the BBB, protecting the CNS from bacterial invasion. One of the major mechanisms employed by the BBB to block bacterial trafficking is endolysosomal killing. Although certain neurotropic pathogens have devised strategies to efficiently evade the endolysosomal pathway, the brain endothelium adopts alternate means present in its arsenal to clear these escaped pathogens, inhibiting BBB breach. Streptococcus pneumoniae, the most prominent meningitis causative microorganism, has been shown to rupture the endosomal membrane during transit through the BBB due to expression of the pore-forming toxin pneumolysin. This blocks endosome maturation and subsequent lysosomal fusion. Our previous findings demonstrate that brain endothelium ubiquitinates a large subset of such intracellularly persistent S. pneumoniae cells for eventual killing (14). In this study, we extended our findings to reveal the details of different molecular mechanisms adopted by brain endothelial cells to dispose such ubiquitinated S. pneumoniae by decorating them with various ubiquitin chain topologies (Fig. 6). This lowers pneumococcal burden in BBB following bacterial invasion and prevents their access to the central nervous system for establishment of infection.

FIG 6.

Model depicting differential ubiquitination dependent intracellular clearance of S. pneumoniae during transit through blood-brain barrier. Following internalization in brain endothelium, secretion of Ply triggers extensive damage to PCV membrane. This leads to ubiquitination of the PCVs with various chain topologies. K63 ubiquitination results in sequestration and eventual degradation of S. pneumoniae in autophagosomes following fusion with lysosomes. On the other hand, K48 ubiquitination selectively triggers association and subsequent killing while being associated with proteasomes. A coordinated action of these distinct intracellular defense mechanisms ensures fruitful blockage of S. pneumoniae transcytosis across brain endothelium, offering protection to the CNS.

In accordance with the intracellular fate of other pathogens (25), we also observed acidification of a subset of ubiquitinated PCVs, which ultimately fused with lysosomes for degradation in brain endothelium. This was expected, as several host autophagy receptors have ubiquitin binding domains to sequester the cargo, including pathogens, in autophagosomes for eventual fusion and degradation in lysosomes. Activation of antipneumococcal autophagy was earlier reported in lung epithelial and brain endothelial cells (14, 26, 27). However, we could demonstrate that the specificity for such a process is provided by K63-Ub chain types formed on PCVs. Intriguingly, the major population of ubiquitinated PCVs in the BBB was neither acidified nor displayed characteristic lysosomal markers. Intracellular bacteria like Streptococcus pyogenes, Streptococcus pneumoniae, Listeria monocytogenes, etc., which produce cholesterol-dependent cytolysins such as streptolysin O (SLO), pneumolysin (Ply), and listeriolysin O, respectively, have been reported to rupture not only endosomal but also autophagosomal membrane by virtue of their pore-forming ability (14, 28, 29). Progressively larger pores formed by these toxins trigger the escape of ubiquitinated pathogens from autophagosomes. However, such ubiquitinated autophagosome-escaped pathogens must be cleared by other mechanisms before they transcytose through the BBB. Coexistence of a large fraction of such nonlysosomal, ubiquitinated S. pneumoniae populations propelled us to speculate that apart from autophagy, brain endothelium must deploy a parallel degradative pathway for clearance of ubiquitinated intracellular S. pneumoniae. Our findings reveal that, indeed, the host proteasomal pathway contributes significantly to intracellular ubiquitinated S. pneumoniae clearance. Although proteasomal involvement in bacterial killing was proposed previously for Mycobacterium tuberculosis (30), we, for the first time, using time-lapse fluorescence imaging, showed degradation of ubiquitinated bacteria in close association with proteasomal machinery (Fig. 2). Indeed, association of proteasome with S. pneumoniae is reported in mouse brain tissue (15). This further substantiates our claim for the involvement of proteasomes as a major mechanism deployed by brain endothelium to restrict pneumococcal passage across the BBB, which could be extrapolated to other neurotropic pathogens.

To unravel the factors governing the specificity of the bacterial degradation pathways, we then explored for the contribution of the ubiquitin chain topologies formed on the PCVs. Ubiquitin chain types formed on the substrate, which includes cellular proteins, organelles, and even bacteria, are reported to govern the choice of further clearance pathways (31). The selectivity of the ubiquitin binding adaptor proteins associated with either autophagy or proteasome dictate the choice of pathways. High local concentrations of K63-linked polyubiquitin chains acquiring an extended shape preferentially bind to the autophagic adaptors p62, NBR1, and NDP52, driving the cargo toward autophagy (32). On the contrary, K48-linked ubiquitin polymers adopt a compact conformation, favoring a sandwich-like binding by the ubiquitin binding adaptor proteins associated with proteasome (33). Our findings, for the first time, demonstrated a comprehensive composition of the ubiquitin coat of intracellular S. pneumoniae involving K48- and K63-Ub chain topologies (Fig. 3). We envisage that this composition of ubiquitin coat on pneumococcus could be of similar order in any host cell type. The difference in the formation kinetics of K48-Ub and K63-Ub suggests that although at early stages of infection PCV-bound S. pneumoniae gets ubiquitinated with different types of ubiquitin topology, at prolonged infection intervals a large subset of S. pneumoniae is present in cytosol without the shield of any membrane remnants. This is supported by a progressive steep decrease in the association of S. pneumoniae with K63-Ub chains. Since the majority of K63-ubiquitin signal was earlier observed to be associated with PCV membrane (27), it could be projected that K63-Ub signal is left behind along with damaged endosome remnant when S. pneumoniae escaped to host cytosol.

Contrary to a previous report that showed reduction in the abundance of K63-Ub S. pneumoniae in the absence of autophagy (27), our findings involving recruitment of LC3-positive autophagosome to the K63-Ub-positive S. pneumoniae coupled with accumulation of K63-Ub-positive S. pneumoniae only in autophagy inhibitor-treated cells unambiguously proves that K63-Ub directs S. pneumoniae toward autophagic killing. This is in accordance with other intracellular pathogens, like M. tuberculosis, that has also been reported to be targeted toward autophagy following ubiquitination with K63-Ub chains by the E3 ligase Parkin (25). Critically, exclusive association of LMP2 with K48-Ub-positive S. pneumoniae and further accumulation of K48-Ub S. pneumoniae when proteasomal machinery of the brain endothelium was impaired established that K48-Ub is responsible for the proteasomal killing of S. pneumoniae. Although previously it was observed that K48-Ub S. pneumoniae accumulates in Atg5 knockout murine embryonic fibroblasts (MEFs) (27), we did not observe any effect of autophagy inhibition on the abundance of K48-Ub-positive S. pneumoniae in hBMECs. We argue that this association of K48-Ub-positive S. pneumoniae with autophagy could be a specific phenotype in MEFs, as in the majority of the host cells K48 ubiquitination maintains protein homeostasis by directing them to the proteasome (34).

Numerous pathogens like M. tuberculosis, Salmonella, etc., have been shown to be decorated with K63-, K48-, and M1-ubiquitin chain topologies (25, 27, 30, 35). However, relative contributions of individual chain types to the fate of intracellular pathogens have remained elusive. Along with linkage type, the length of the chain plays a crucial role dictating further ubiquitin-mediated signals (36). Insufficient length of the ubiquitin polymer impairs association of the ubiquitin binding adaptor proteins, arresting further degradation of the cargo (37). We targeted this attribute of ubiquitin chain length to compare the impact of each ubiquitin chain type on S. pneumoniae degradation. Adopting a novel approach to impair formation of one specific type of ubiquitin chains (either K48 or K63), we demonstrated that although both K48- and K63-ubiquitin chains play important roles in clearance of intracellular S. pneumoniae, K48-Ub chains and subsequent proteasomal degradation have more telling contributions to ubiquitinated S. pneumoniae killing in brain endothelial cells. Interestingly, unlike S. pneumoniae, an increase in intracellular M. tuberculosis load is similar in the absence of K48 and K63 ubiquitin chains (30), implying that the contribution of each ubiquitin chain type to clearance of bacteria varies for different intracellular pathogens.

Collectively, these observations, for the first time, illustrated a symphony of ubiquitin-mediated intracellular defense mechanisms deployed by the blood-brain barrier to prevent pathogen replication. Synergy between such defense mechanisms allows the brain endothelium to function as a sentinel for protection of CNS from bacterial entry and maintenance of homeostasis.

MATERIALS AND METHODS

Cell culture and transfections.

Human brain microvascular endothelial cells (hBMECs; provided by K. S. Kim, Johns Hopkins University) (38) were used as a model for the BBB. hBMECs have been extensively used for studying interaction of host with meningeal pathogens and to characterize host response (39–43). Short tandem repeat (STR) profiling was used to authenticate the hBMEC line using the FTA sample collection kit for human cell authentication service, ATCC 135-XV. Additionally, hBMECs were found to express VE-cadherin and caudin-5, the markers of brain endothelium (see Fig. S5 in the supplemental material). hBMECs are routinely cultured in RPMI 1640 medium (Gibco) supplemented with 10% fetal bovine serum (Gibco), 10% Nu-serum (Corning), and 1% nonessential amino acids (Gibco) at 37°C in 5% CO₂. Stably transfected hBMECs overexpressing LC3-GFP, GFP-LMP2, and mStrawberry-Ub fusion proteins were constructed by transfecting hBMECs with pBABEpuro LC3-GFP vector (22405; Addgene), pMRX-IRES-GFP-LMP2 (GFP-LMP2 was subcloned from pCND3-GFP-LMP2, a kind gift from J. Neefies, The Netherlands Cancer Institute, The Netherlands) (44), and pMRX-IRES-mStrawberry-ubiquitin (45) (kind gift from T. Yoshimori, Osaka University, Japan). Ub^K48R-GFP (46) (kind gift from D. Gray, Ottawa Regional Cancer Centre, Canada), mStrawberry-Ub^K63R (47) (kind gift from L. Penengo, University of Zuric-Vetsuisse, Switzerland), and mStrawberry were stably expressed in hBMECs under tetracycline-inducible promoter in pLKO-Puro-tet plasmid (21915; Addgene). All transfections were performed using Lipofectamine 3000 reagent (Invitrogen), and selections were done in the presence of 2 μg/ml puromycin (Sigma-Aldrich) and 2 μg/ml blasticidin hydrochloride (HiMedia) where necessary. For inhibition studies, hBMECs were pretreated with 3-methyladenine (1 mM, 24 h) or MG132 (10 μM, 1 h) before infection with bacteria. The concentrations used for 3-MA and MG132 are noncytotoxic, as found out by 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. For induction to express mutant ubiquitin variants (Ub^K48R-GFP and mStrawberry-Ub^K63R), cells were treated with freshly prepared tetracycline (40 μg/ml, 15 h) prior to the assay, and formation of ubiquitin variants was verified using immunoblotting.

Bacterial strains and constructs.

S. pneumoniae strain R6 (serotype 2, unencapsulated) was obtained from T. J. Mitchell (University of Birmingham, UK). S. pneumoniae was routinely grown in Todd-Hewitt broth (THB) supplemented with 1.5% yeast extract (THY medium) at 37°C in 5% CO₂. To construct tetracycline-resistant S. pneumoniae, the tetR gene was amplified from pYPQ132D (134348; Addgene) and cloned between XbaI and XhoI sites of the shuttle vector pIB166 (48) (kind gift from Indranil Biswas, Kansas University Medical Center) and transformed into wild-type S. pneumoniae strain using competence stimulating peptide 1 (100 ng/ml). Tet^r S. pneumoniae was further used to infect mutant ubiquitin-expressing cell lines as mentioned earlier.

Plasmids, antibodies, and reagents.

Fluorescent S. pneumoniae was developed by transformation of WT S. pneumoniae with an hlpA-GFP fusion cassette (kindly provided by J. W. Veening, University of Groningen, Netherlands). The following primary antibodies were used in this study: cathepsin B (ab58802; Abcam), antiubiquitin K48-specific (05-1307; Merck Millipore), antiubiquitin K63-specific (BML-PW0600-0025; Enzo Life Science), antiubiquitin (Enzo Life Sciences, BML-PW-8810), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (MAB374; Millipore). Fine chemicals used in this study include 3-methyladenine (M9281) and MG132 (474791), from Sigma-Aldrich, LysoTracker deep red (L12492; Invitrogen), and DRAQ5 deep red (564903; BD Pharmingen).

Penicillin gentamicin protection assay.

Infection assays were performed as described earlier (49). Briefly, a fully confluent hBMEC monolayer in 1% collagen coated 24-well plate was infected with S. pneumoniae (optical density at 600 nm of ∼0.4) at a multiplicity of infection (MOI) of 10 for 1 h. Following infection, extracellular S. pneumoniae was killed by incubating infected cells for 2 h in culture medium containing penicillin (10 μg/ml) and gentamicin (400 μg/ml). After several washes, cells were trypsinized (0.025% trypsin EDTA) and lysed (0.025% Triton X-100), and the lysates were spread plated on brain heart infusion (BHI) agar plates for enumeration of bacterial colonies. Bacterial invasion efficiency was calculated as (recovered CFU/initial inoculum CFU) × 100%. To assess intracellular survival of bacteria, cell lysates were prepared similarly and spread plated at indicated time intervals following 1 h of infection and 2 h of antibiotic treatment. Surviving bacteria at different time points were enumerated and represented as percent survival at indicated time points relative to 0 h.

Fluorescence imaging.

For immunofluorescence detection, hBMECs seeded on 1% collagen-coated glass coverslips were infected with S. pneumoniae (MOI, ∼25) as described above. At desired time points postinfection, cells were washed with RPMI to remove extracellular S. pneumoniae and fixed with ice-chilled methanol for 10 min at −20°C. After blocking in 3% bovine serum albumin (BSA) for 2 h, cells were treated with appropriate primary antibody in 1% BSA overnight at 4°C. Cells were then incubated with suitable secondary antibody in 1% BSA for 1 h at room temperature (RT) and finally mounted with VectaShield, with or without 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories), for visualization using a laser scanning confocal microscope (Zeiss LSM780) under 40× or 63× oil objective. For live-cell imaging, transfected hBMECs were grown on glass-bottom petri dishes and infected with S. pneumoniae as described earlier. The time-lapse imaging was performed at multiposition under 63× oil immersion objective of laser scanning confocal microscope (LSM780). The fixed and live images were acquired after optical sectioning and then processed using Zen Lite software (version 5.0).

Western blot analysis.

Following tetracycline treatment for 15 h, hBMECs were washed 2 to 3 times with chilled phosphate-buffered saline and lysed in ice-cold radioimmunoprecipitation assay buffer (50 mM Tris-Cl [pH 7.8], 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 1% SDS) containing protease inhibitor cocktail (Sigma-Aldrich), sodium fluoride (10 mM), phenylmethylsulfonyl fluoride (100 μM), and EDTA (5 mM). Total cellular proteins were separated from the cell debris by centrifugation (15,000 × g for 30 min at 4°C) of cell lysates. Proteins (10 μg or 20 μg) were separated on 12% SDS-PAGE gels and were then transferred to activated polyvinylidene difluoride membrane (Bio-Rad). Following 2 h of blocking in 5% skimmed milk, the blot was incubated with appropriate primary and horseradish peroxidase-conjugated secondary antibodies and finally developed using ECL substrate (Bio-Rad).

MTT cell viability assay.

Effects of inhibitor and antibiotic treatments on the viability of hBMECs were quantified using EZcount MTT cell assay kit per the manufacturer’s protocol (HiMedia). Briefly, treated hBMECs were washed with Hanks’s balanced salt solution containing calcium and magnesium (HBSS-Ca⁺²Mg⁺²); 500 μl of 10% MTT reagent in HBSS-Ca⁺²Mg⁺² from a stock of 5 mg/ml was added to cells and incubated at 37°C for 2 h. Further, cells were treated with 500 μl of solubilizing agent (10% Triton X-100, 90% isopropanol, and 0.02% of 37% HCl) followed by repeated pipetting and incubation at RT for 30 min under shaking conditions to dissolve formazan crystals. Absorbance of the dissolved purple formazan crystals in the cell lysates was determined at 560 nm using a microplate spectrophotometer. Inhibitor-nontreated infected cells were used as a negative control, while Triton X-100 (0.025%) was used as a positive control.

Statistics.

All statistical analysis was performed using GraphPad Prism software (version 5), and significance was set at a P value of <0.05. Statistical tests employed for individual experiments are mentioned in the respective figure legends. At least three independent experiments were performed. The results are presented as the mean ± standard deviations (SD) from one representative experiment.