Chlamydia pneumoniae has been identified and associated with multiple sclerosis (MS) and Alzheimer's disease (AD) pathogenesis, although the relationship of this organism in these diseases remains controversial. We have hypothesized that one potential avenue of infection is through the junctional complexes between the blood–brain barrier (BBB) endothelia. C. pneumoniae is characteristically a respiratory pathogen, but has been implicated in atherosclerosis, coronary artery disease, and neuroinflammatory conditions. C. pneumoniae infection may lead to endothelial damage, junctional alterations, and BBB breakdown. Therefore, in this study, C. pneumoniae infection of human brain microvascular endothelial cells (HBMECs) resulted in increased expression of the zonula adherens proteins β-catenin, N-cadherin, and VE-cadherin, and decreased expression of the tight junctional protein occludin, as determined by immunocytochemistry and Western blot analyses. These events may underlie a mechanism for the regulation of paracellular permeability while maintaining barrier integrity during C. pneumoniae infection associated with neuropathologies such as MS and AD.
Infection with bacterial and viral organisms has been associated with changes in endothelia that result in the upregulation of surface adhesion molecule expression, the release of cytokines/chemokines, and most importantly, the increased permeability of junctional complexes in endothelial monolayers [3,4]. Recent evidence suggests that signaling elements promoting cytoskeletal modifications link changes at the cell surface, involving surface adhesion molecule activation, to junctional alterations in endothelial monolayers. The migration of immune cells into the brain implicates surface adhesion molecules, in particular, ICAM-1 on endothelium and LFA-1 on leukocytes. This interaction may be enhanced in neuroinflammatory conditions such as MS and AD. Signal transduction involving the Rho GTPases is linked to β-catenin, influencing the function of the zonula adherens. In this regard, some organisms can trigger signaling cascades resulting in pro-inflammatory cytokine release, and actin reorganization. In endothelial cells, such activation is consistent with ruffle formation, the phagocytic phenotype, spindle-shaped morphology, and a potential association with the bacterial inclusion [9,10]. All are typically observed in endothelia following C. pneumoniae infection.
Our laboratory previously has demonstrated that C. pneumoniae infection upregulated the expression of surface adhesion molecules in brain endothelia and monocyte cells, and also promoted the migration of monocytes across brain endothelial cell monolayers in vitro (submitted, Brain). In this study, surface-to-junction cross-talk was examined to determine the effects of C. pneumoniae infection on the zonula adherens, particularly the cadherins and catenins, as well as occludin in the zonula occludens. VE-cadherin is the only cadherin specific to endothelia, and localized exclusively to the intercellular junctions. VE-cadherin, through β-catenin, connects to the actin cytoskeleton, and polymerization events cause changes in cell morphology, and junctional complex assembly. N-cadherin, initially isolated from neural tissue, is present on the entire cell surface of brain endothelial cells, and is important in junctional assembly. The adhesive function of cadherins requires their association with the actin cytoskeleton via catenins. β-Catenin is responsive to changes at the cell surface and relays them to the junction, thereby affecting barrier permeability through cytoskeletal rearrangements. These proteins have been shown to control peri-junctional actin, and may be integral to a unifying mechanism for regulating barrier integrity. In the BBB, the endothelial tight junctions create a rate-limiting barrier. The tight junctional barrier typically is formed by rows of occludin, extending in a zipper-like formation into the paracellular space. Barrier permeability and occludin expression are often studied following infection with a variety of communicable agents [12–14]. Occludin is a prime target in immune cell migration and pathogen infiltration. Organisms that may affect occludin functioning at the BBB include HIV-1, Neisseria meningitidis, and Herpes species to name a few. Thus, the regulation of occludin expression during C. pneumoniae infection may be critical to the tight junctional permeability properties of human brain microvascular endothelial cells (HBMECs). Therefore, in this study, cell surface/cell junction linkage properties were examined following C. pneumoniae infection of representative BBB microvascular endothelial cells.
2Materials and methods
HBMECs (BB-19s were a gift from Dr. Jeymohan Joseph of the NIH/NIA, Bethesda, MD, USA) and were cultured in human endothelial SFM basal growth media with l-glutamine (Invitrogen, Life Technologies, Carlsbad, CA, USA) supplemented with 10% human serum (Sigma, St. Louis, MO, USA), 2 mM glutamine, 1% Pen/Strep, 25 μg ml−1 endothelial cell growth supplement (Collaborative Biomedical Products, Bedford, MA, USA), 40 μg ml−1 heparin (Sigma), and G418 (Invitrogen Corp., Gibco-BRL, Carlsbad, CA, USA) at 200 μg ml−1. Cells were cultured in T-75 flasks coated with gelatin (Invitrogen, Gibco-BRL). Cells were harvested between passages 4–9.
HBMECs were plated at 90% confluence in T-25 flasks (2–2.5×106 cells), allowed to attach for 24 h, and infected with 5×105 IFU ml−1 of the AR-39 strain of C. pneumoniae from HEp-2 cell lysate (American Type Culture Collection, Manassas, VA, USA), followed by incubation at 37°C for 4, 8, 12, 24, and 48 h. All cultures tested were negative for mycoplasma using the ATCC mycoplasma detection kit Version 2.0 (ATCC 90-1001K). Cells were rinsed in Hank's buffered salt solution (HBSS), trypsinized for 2 min, and collected in 15-ml tubes, and centrifuged at 1000 rpm at 4°C for 5 min. HBMECs were resuspended in HBSS, counted, and split for immunofluorescence, immunocytochemistry and electrophoresis. Control/uninfected flasks were prepared exactly as above and harvested at matched time points. Control experiments have determined that uninfected HEp-2 cell lysates show no effect on junctional protein expression levels in our cells.
Uninfected and infected cells were cytospun onto Fisher Superfrost/plus microscope slides (Fisher Scientific, Pittsburgh, PA, USA) and fixed in 2% paraformaldehyde for 20 min at 4°C. Cytospun slides were incubated with 25 μl of a chlamydial specific antibody (Imagen, Dako Corp., Carpenteria, CA, USA). Slides were viewed on a Nikon Eclipse E800 microscope, and images were captured on an Optronics DEI-750 CA camera. Infection was determined by number of inclusions per high-powered field per slide.
HBMECs were resuspended in HBSS at 1×105 200 μl−1 and cytospun onto Superfrost/plus slides (Fisher Scientific) at 200 μl per slide. Slides were fixed in cold 2% paraformaldehyde for 30 min, rinsed in phosphate-buffered saline (PBS), and quenched for endogenous peroxidase with 3% H2O2. Slides were blocked in 2% fetal bovine serum and incubated undiluted for 2 h at 25°C with a β-catenin antibody (Dr. Karen Knudson of Lankenau Medical Research Center, Philadelphia, PA, USA). Following reblocking, samples were incubated with a secondary horseradish peroxidase (HRP)-conjugated IgG1 antibody (Amersham, Inc., Piscataway, NJ, USA) (1:200) for 1 h at 25°C, rinsed in PBS and incubated in SigmaFast brand 3,3′-diaminobenzidine (Sigma) for 5 min. Slides were counterstained for 1 min in hematoxylin (Fisher Scientific), coverslipped with Gelvatol, and viewed on a Nikon Eclipse E800 microscope with an Optronics DEI-750 CA camera.
2.5Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE)/Western blot analysis
HBMECs were resuspended in 6 M urea at 1.2×106 cells ml−1 and were sonicated for 20 s per sample. Cells were assayed for protein concentration using the Pierce BCA protein assay kit (Rockford, IL, USA). Protein (10 μg per lane) was loaded onto NOVEX 4–20% gradient Tris–glycine gels (Invitrogen Corp.) and electrophoresed. Kaleidoscope molecular mass markers (Bio-Rad, Hercules, CA, USA) were used as standards. Samples were run in combination with an uninfected 48 h control followed by each infection time point, or in uninfected/infected pairs per time point. Gels were transferred onto nitrocellulose (Schleicher and Schuell, Keene, NH, USA) for immunoblotting. Blots were rinsed in PBS for 30 min, blocked in 5% non-fat milk and 0.1% Tween 20 (Blotto), and incubated with primary antibody diluted in Blotto for 2 h at 25°C. Blots were reblocked in Blotto, incubated with a HRP-conjugated secondary antibody for 1 h at 25°C, rinsed in PBS, and protein bands were detected using the enhanced chemiluminescence (ECL) detection system on Hyperfilm™ ECL™ (Amersham, Inc.). Primary antibodies were anti-β-catenin, and anti-N-cadherin (neat) (gifted, Dr. Karen Knudson), anti-VE-cadherin (1:200), and anti-occludin (1:100) (Transduction Laboratories, Inc., San Jose, CA, USA). Densitometry was performed for all immunoblots using Gel Pro Analyzer software. All experiments were repeated in triplicate.
3.1C. pneumoniae infects HBMECs
Uninfected HBMECs did not label for C. pneumoniae (Fig. 1A), while extensive immunofluorescent punctate labeling was revealed in 48 h C. pneumoniae-infected HBMECs (Fig. 1B), demonstrating infection with the organism.
3.2C. pneumoniae infection upregulates β-catenin, N-cadherin, and VE-cadherin of HBMECs
HBMECs were analyzed by immunocytochemistry and Western immunoblotting for changes in the cytoplasmic junctional molecule β-catenin, and by Western immunoblotting for N-cadherin, and VE-cadherin. As compared to control/uninfected cultures (Fig. 2A), increased immunoreactivity of β-catenin was detected 8 h post-C. pneumoniae infection (Fig. 2B), and was more pronounced at 24 and 48 h post-infection (Figs. 2C, D). Western immunoblotting of HBMECs harvested at 48 h represented uninfected control cultures (Fig. 3A, lane 1). Western immunoblotting performed on 24–48 h infected samples demonstrated increased immunoreactivity for β-catenin beginning at 24 h post-infection (Fig. 3A, lane 2), with even more pronounced immunoreactivity at 36 and 48 h (Fig. 3A, lanes 3, 4, respectively). VE-cadherin displayed a similar pattern to that of β-catenin. Uninfected HBMECs were analyzed at the 48 h time point, again representing uninfected cultures (Fig. 3B, lane 1). Increased immunoreactivity was observed initially at 24 h post-infection for VE-cadherin (Fig. 3B, lane 2), and appeared more pronounced at 36 and 48 h post-infection (Fig. 3B, lanes 3, 4, respectively). N-cadherin was analyzed for changes in protein expression after C. pneumoniae infection. Again, control uninfected HBMECs were harvested at the 48 h time point (Fig. 3C, lane 1). With N-cadherin, less pronounced increases in immunoreactivity were observed up to 36 h post-infection, with no change at 24 h (Fig. 3C, lane 2), a slight increase in protein expression at 36 h, but a defined increase at 48 h post-infection (Fig. 3C, lanes 3, 4, respectively). Semi-quantitative densitometry demonstrated increased protein expression from infected vs. uninfected culture lysates. Between 24 and 48 h post-infection: β-catenin protein expression increased 1.3–9.4-fold, N-cadherin increased 1.0–2.94-fold, and VE-cadherin increased 2.4–4.0-fold as compared to controls.
3.3C. pneumoniae infection decreases occludin immunoreactivity transiently in the zonula occludens of HBMECs
HBMECs were analyzed by Western immunoblotting for changes in occludin protein expression. Occludin expression was monitored with matched uninfected samples (Fig. 4, lanes 1, 3, 5, and 7) and infected samples (Fig. 4, lanes 2, 4, 6, and 8) for each time point. Although a double banding pattern that may represent post-translational modification was apparent, differences in uninfected versus C. pneumoniae-infected HBMECs at the 24 h time point were minimal (Fig. 4, lanes 1, 2, respectively). A decrease in immunoreactivity for C. pneumoniae-infected HBMECs, as compared to uninfected HBMECs, was exhibited at the 36 h time point (Fig. 4, lane 4). Loss of immunoreactivity in C. pneumoniae-infected HBMECs remained consistent at the 48 h time point, and comparable to that of the 36 h time point (Fig. 4, lane 6). Semi-quantitative densitometry demonstrated at 36–48 h post-infection that there was a 57–64% decrease in occludin protein expression. However, increased immunoreactivity was observed at 72 h post-infection, to suggest a transient loss at 36–48 h, rather than a permanent loss of the tight junctional protein during C. pneumoniae infection (Fig. 4, lane 8). Here, densitometry revealed that occludin protein expression returned to control values within <1%. These data are consistent with occludin expression as a determinant of tight junction permeability, whereby infection with C. pneumoniae transiently decreased the protein expression of occludin in HBMECs.
Immunocytochemistry and SDS–PAGE/Western blot analysis demonstrated that C. pneumoniae infection in HBMECs upregulated the cytoplasmic protein β-catenin as well as the transmembrane proteins VE-cadherin and N-cadherin of the zonula adherens. In contradistinction, C. pneumoniae infection of HBMECs resulted in a downregulation of the tight junctional protein occludin, albeit in a transient manner. As previously stated, an upregulation of integrins on endothelial cells can stimulate the activation of the Rho signaling cascade. The results of this activation may alter the expression of β-catenin and the cadherins, and contribute to permeability fluxes at the tight junctional complex. Our results demonstrate an upregulation in β-catenin, VE-cadherin, and N-cadherin, and a subsequent or simultaneous transient loss of occludin after C. pneumoniae infection in HBMECs. These results suggest that a modification of occludin may result in a compensatory response at the cadherin/catenin level to maintain the integrity of the endothelial monolayer. This compensatory response may constitute a unifying mechanism that regulates paracellular permeability while maintaining integrity of the brain endothelial monolayer following C. pneumoniae infection.
Brain endothelial cells are targets for signaling pathways, immune reactions, and pathogen invasion. Ligand–receptor binding at the cell surface has been linked to paracellular events. These paracellular events, including receptor activation, alter membrane polarity and protein expression. Modifications in cytoskeletal elements that affect membrane morphology, such as the formation of filopodia, or membrane ruffling in phagocytosis, can occur following junctional signaling through the cadherins. Junctional modifications, as in the cases of immune cell infiltration, and/or pathogen interaction are integral to these events. These interactions can cause activation of several signal transduction cascades, including those involved in phosphorylation and dephosphorylation which ultimately affect the cytoplasmic domain of many membrane-bound proteins, thereby influencing functionality [17,18].
Our data suggest that increased expression of β-catenin is a more prominent feature than post-translational modification following CP infection of HBMECs. Immunocytochemical analysis of HBMECs following CP infection demonstrates β-catenin immunoreactivity at the cell periphery and no nuclear immunoreactivity. This suggests that phosphorylation of β-catenin was unlikely, as phosphorylated β-catenin typically translocates to the nucleus, or is degraded by the ubiquitin-proteasome pathway [19,20]. Likewise, biochemical analysis demonstrated no change in band migration, but an increase in band density over time with infection, suggesting the increase of β-catenin was due to increased expression rather than post-translational modification.
C. pneumoniae entering the CNS may involve direct infection of BBB endothelia, as well as the trafficking of infected monocytes through the BBB. Leukocyte transmigration across HBMEC monolayers has been shown to stimulate transient changes in the expression of numerous junctional proteins, as well as a remodeling of the cytoskeletal elements of HBMECs [3,21]. Our results indicate that the effects of C. pneumoniae infection on occludin expression could allow for transient fluctuations in the permeability of the BBB. Occludin is a stable protein with low turnover. Typically, differences in occludin protein are reflected at the message level, suggesting transcriptional regulation of expression. Our data reflect a decrease in protein expression that possibly is due to increased protein degradation and/or alterations in transcriptional activity over the course of 24–48 h post-infection. However, this appears to be transient because occludin protein levels are restored within 72 h post-infection. Downregulation of occludin and upregulation in the junctional proteins of the zonula adherens suggest a cellular compensatory mechanism to maintain barrier fidelity while greatly altering the permeability profile of the BBB. Even though protein expression is upregulated at the adherens level, this overexpression would not be sufficient to retain the BBB phenotype, as the tight junction proteins principally are responsible for the limited permeability characteristics of the BBB.
In conclusion, C. pneumoniae infection in HBMECs resulted in increased protein expression of β-catenin, VE-cadherin and N-cadherin of the zonula adherens. C. pneumoniae infection of HBMECs resulted in a downregulation of the tight junctional protein occludin in a transient manner. Changes in junctional complex integrity caused by C. pneumoniae infection in HBMECs may account for a complex mechanism governing permeability characteristics of the BBB. Permeability changes are associated with the activation of the Rho signaling cascade, and this activation may alter the expression of β-catenin and the cadherins, as well as alter the permeability of the tight junctional complex. These signaling events directly affect the assembly and sealing of the junctional complexes. As a consequence of C. pneumoniae infection, alterations of the junctional complexes may lead to altered BBB transport mechanisms, increased immune cell infiltration, as well as facilitate pathogen entry through a compromised barrier, potentially having implications for neurodegenerative diseases such as MS and AD.
The authors would like to thank Dr. Karen Knudson of Lankenau Medical Research Center for her generous gift of anti-β-catenin and anti-N-cadherin. Research was supported by NIH/NIAID AI44055, and The Foundation for Research Into Diseases of Aging (FRIDA).