PDCD10 (CCM3) regulates brain endothelial barrier integrity in cerebral cavernous malformation type 3: role of CCM3-ERK1/2-cortactin cross-talk

Abstract

Impairment of brain endothelial barrier integrity is critical for cerebral cavernous malformation (CCM) lesion development. The current study investigates changes in tight junction (TJ) complex organization when PDCD10 (CCM3) is mutated/depleted in human brain endothelial cells. Analysis of lesions with CCM3 mutation and brain endothelial cells transfected with CCM3 siRNA (CCM3-knockdown) showed little or no increase in TJ transmembrane and scaffolding proteins mRNA expression, but proteins levels were generally decreased. CCM3-knockdown cells had a redistribution of claudin-5 and occludin from the membrane to the cytosol with no alterations in protein turnover but with diminished protein-protein interactions with ZO-1 and ZO-1 interaction with the actin cytoskeleton. The most profound effect of CCM3 mutation/depletion was on an actin-binding protein, cortactin. CCM3 depletion caused cortactin Ser-phosphorylation, dissociation from ZO-1 and actin, redistribution to the cytosol and degradation. This affected cortical actin ring organization, TJ complex stability and consequently barrier integrity, with constant hyperpermeability to inulin. A potential link between CCM3 depletion and altered cortactin was tonic activation of MAP kinase ERK1/2. ERK1/2 inhibition increased cortactin expression and incorporation into the TJ complex and improved barrier integrity. This study highlights the potential role of CCM3 in regulating TJ complex organization and brain endothelial barrier permeability.

Keywords: Brain endothelial barrier; CCM3; Cortactin; Permeability; Tight junction.

Figure 1

a) mRNA microarray and b) tissue array for junctional proteins present in brain tissue of control (age and gender matched), CCM3 lesion (case with CCM3 c.474+1G>A mutation) and sCCM (cases with sporadic form of CCM). For the mRNA microarray array (a), data represent relative mRNA expression, while the tissue array (b) is total florescence intensity of 30 analyzed spots on CCM lesions or blood vessels in controls (see Material and Method section). Values are means ± SD, n = 3 control brain, n = 2 CCM3 cases, n = 3 sporadic CCM cases (sCCM). *p<0.05, **p<0.01 and ***p<0.001 compared to control samples.

Figure 2

a) Representative immunofluorescent images of junctional protein expression in CCM3 and sCCM brain lesions. Mag. 10x and 20x (H&E control, CCM3 and sCCM). Scale bar 20μm. b) Representative images of Western blot analysis of TJ proteins (claudin-5, occludin, ZO-1 and JAM-A), AdJ proteins (Ve-cadherin and β-catenin) and an actin binding protein, cortactin in all CCM3 and sCCM laser capture microdissection samples. Relative protein expression in control, CCM3 and sCCM samples was determined by semi-quantitative analysis. Values are means ± SD, n = 3 control brain, n = 2 CCM3 cases, n = 3 sporadic CCM cases (sCCM). *p<0.05 and **p<0.01 compared to control samples.

Figure 3

a) Permeability coefficient of brain endothelial monolayers for FITC-inulin (5 kDa), Cascade Blue-dextran (20 kDa) and Texas red-dextran (40 kDa) in control (scrambled siRNA) and CCM3-KD (CCM3 siRNA) transfected cells measured daily from 1 to 7 days after initial plating. Data represent means±SD for n=5 independent, experiments. *p<0.05, **p<0.01 and ***p<0.001 comparing control and CCM3-KD cells. The total cell density in all experimental groups was assessed after each permeability measurement by MTT assay and are presented as the measured OD value for formazan. Note that there are no significant changes in density of cells (1–1.5×10⁵ cells/well) over the time period of 1–7 days or between control and CCM3-KD groups. b) Real time RT-PCR and Western blot analysis of claudin-5, occludin, JAM-A and ZO-1 expression in control and CCM3-KD (siRNA) cells 7 days after initial plating. Data represent means±SD, n=3; **p<0.01 and ***p<0.001 comparing control and CCM3-KD cells. Western blot image is one of three independent experiments. GAPDH is an internal control. c) TJ protein (claudin-5, occludin, JAM-A and ZO-1) turnover rate was analyzed following treatment with cycloheximide in CCM3-KD and control cells. Semi-quantitative densitometry for TJ protein was performed and normalized to β-tubulin as an internal control in three independent experiments. Cycloheximide (CXD) was added for 0–24 h in concentration of 5 μg/ml. Time point 0 is start for CXD treatment and for control and CCM3-KD cells was taken as 100% of certain protein expression. The half-life of the proteins was: claudin-5, control cells = 6.4h (R²=0.902), CCM3-KD cells = 6.7h (R²=0.929). Occludin, control = 5.5h (R²=0.968) and CCM3-KD = 4.0h (R²=0.960). ZO-1, control = 8.8h (R²=0.883) and CCM3-KD = 7.4h (R²=0.845). JAM-A, control = 5.6h (R²=0.923) and CCM3-KD = 4.6h (R²=0.889), n=3 independent experiments. There were no significant differences in TJ protein turnover between control and CCM3-KD cells. d) Cell fractionation and Western blot analysis of claudin-5, occludin, JAM-A and ZO-1 distribution in CCM3-KD and control cells showed an increased accumulation of claudin-5 and occludin in the cytosol fraction (CF, TritonX-100 soluble fraction) in CCM3-KD cells compared to control cells where almost all of those proteins was in the membrane fraction (MF, TritonX-100 insoluble). ZO-1 was distributed in the cytosol fraction (CF, TritonX-100 soluble) and actin cytoskeleton fraction (ACF, TritonX-100 insoluble) in control cells, while in CCM3-KD cells was predominantly present in the CF. Blots represent one of three successful experiments. Cytochrome P450 reductase, calpain and vimentin represent the markers for membrane, cytosol and actin cytoskeleton fractions. e) Co-immunoprecipitation (IP) of ZO-1 with either claudin-5, occludin, JAM-A or β-actin showed close association of claudin-5, occludin, JAM-A and β-actin with ZO-1 in control cells and diminished or completely lost interaction in CCM3-KD cells. Western blotting image is one of three independent experiments. Input represents the 50% of total cell extract used for immunoprecipitation.

Figure 4

a) and b) Real time RT-PCR and Western blot analysis of cortactin (CTTN) expression in control and CCM3-KD cells showed higher cortactin mRNA expression in CCM3-KD cells but significantly decreased protein levels. mRNA and protein levels are presented as expression relative to GAPDH, an internal control. Data represent means±SD, n = 3; *p<0.05, ***p<0.001 between control and CCM3-KD cells. Western blot image is one of three independent experiments. c) Cortactin expression was examined by double label immunofluorescence with ZO-1 in control and CCM3-KD cells. Cortactin staining was present in control cells in cell cytosol and as continuous staining at cell borders associated with ZO-1. In CCM3-KD cells, cortactin staining completely disappeared from the cell border. Scale bar 20μm. d) The initial cell fractionation and Western blot analysis of cortactin distribution showed cortactin present in the membrane (MF) and actin cytoskeleton fraction (ACF) in control cells, while CCM3-KD cells had a redistribution of cortactin to the cytosol (CF) and actin cytoskeleton fraction (ACF). Additional analysis of the actin cytoskeleton fraction by fractionation into high density Triton100X insoluble (HD-Triton insol) and low density Triton100X insoluble (LD-Triton insol) fractions. Cortactin presence in the membrane actin cytoskeleton (LD-Triton insol) and cytoplasmic actin (HD-Triton insol) fractions was diminished in CCM3-KD cells compared to control cells. Western blot image is one of three independent experiments. e) Co-immunoprecipitation of either ZO-1 or β-actin with cortactin showed close association of cortactin with ZO-1 and β-actin in control cells and diminished or completely lost interaction in CCM3-KD cells. Western blot image is one of three independent experiments. Input represents the 50% of total cell extract used for immunoprecipitation. f) Rescuing of CCM3-KD cells with CCM3 cDNA restored both CCM3 and cortactin protein expression. Western blot image is one of three independent experiments. Semi-quantitative densitometry for cortactin and CCM3 protein expression was performed and normalized to β-tubulin as an internal control in five independent experiments. Graph represents means±SD, n = 5; ** p<0.01, ***p<0.001 comparing control and CCM3-KD cells and CCM3-KD and CCM3-KD+CCM3 cells respectively. G) It also improved brain endothelial barrier integrity in CCM3-KD cells with decreased permeability to FITC-Inulin. Graph represents means±SD, n = 3; ***p<0.001.

Figure 5

a) Cortactin (CTTN) protein turnover rate was analyzed following treatment with cycloheximide (CXD) in CCM3-KD and control cells. Semi-quantitative densitometry for cortactin was performed using β-tubulin as internal control. Cycloheximide (5 μg/ml) was added for 0–24 hrs. Time point 0 is start for CXD treatment and for control and CCM3-KD cells was taken as 100% cortactin expression. The half-life of cortactin was in 5.0h in controls (R²=0.916) and 2.3h in CCM3-KD cells (R²=0.972). Cortactin in CCM3-KD cells had a decreased half-life compared with control cells. b) Cortactin degradation in control and CCM3-KD cells was also examined in the presence of leupeptin (25μM), an inhibitor of lysosmal enzymes, or MG-132 (10 μM), an inhibitor of proteasome activity, during treatment with CXD. Representative Western blot images and semi-quantitative densitometry (graph) showed that inhibition of proteasome activity by MG132 prevented cortactin degradation in both control and CCM3-KD cells and abolished the difference in degradation between the two types of cells. In contrast, leupeptin did not affect cortactin degradation in either cell type. Graphs show means ± SD, n = 3 independent experiments. c) The ubiquitination of cortactin (CTTN) in control and CCM3-KD cells was detected by Western blot using anti-ubiquitin antibody. d) Cortactin protein expression in CCM3-KD cells was rescued by inhibiting proteasome activity with MG132 or by treatment with the ubiquitin E3 ligase inhibitor SMER3 (50 μM). Western blot image is one of three independent experiments. e) Permeability coefficient (PC) for FITC-inulin in CCM3-KD and control cells after treatment with the proteasome inhibitor MG132. Data represent means ± SD for n=3 independent experiments, ***p<0.001.

Figure 6

a) Phosphorylation status of cortactin in control and CCM3-KD cells. Western blot analysis of the cytosol (triton-soluble fraction) and both actin cytoskeleton fractions, low density TritonX-100 insoluble (membrane) and high density Triton-100 insoluble actin cytoskeleton (cytosol) fractions, showed increased serine phosphorylation (p-Ser) of cortactin in the cytosol fraction in CCM3-KD cells while tyrosine phosphorylation (p-Tyr) was depleted in both actin cytoskeleton fractions. Western blot image is one of three independent experiments for 1–5 days after CCM3 siRNA transfection. b) Representative images of immunofluorescence and Western blot analysis of cortactin in CCM3-KD and control cells showed accumulation of p-Serine405 (pS405) phosphorylated cortactin in CCM3-KD cells in cytosol fraction (calpain + fraction). Rescuing CCM3 in CCM3-KD cells decreased the p-Serine405 (pS405) cortactin. Western blot image is one of three independent experiments. c) Cortactin overexpression in CCM3-KD cells (CTTN^over) rescued the expression of cortactin. Western blot analysis indicated the cortactin presence in CCM3-KD cells at day 1 after transfection. The overexpression of cortactin also lead to establish the cortactin-ZO-1 and cortactin-β-actin interaction in CCM3-KD cells indicating that rescuing cortactin may improve the cortactin availability for “physiological” function. Western blot image is one of three independent experiments. d) Overexpression of cortactin in CCM3-KD cells also improved brain endothelial barrier integrity with decreased permeability to FITC-inulin. Graph represents means ± SD, n = 3; ***p<0.001 comparing to control and **p<0.01 comparing CCM3-KD and CCM3-KD CTTN^over cells.

Figure 7

a) CCM3-KD cells showed a steady increased activity of ERK1/2 compared to control cells. ERK1/2 activity was measured by Western blot analysis of phospho-ERK1/2 (Western blot) as well as ERK1/2 activation assay (graph). The positive control for ERK1/2 activation was epidermal growth factor (EGF; 10ng/ml). Data represent an average ± SD for n=3 independent experiments, *p<0.05, ***p<0.001 vs. control cells. b) Inhibition of ERK1/2 activity either via PD098059 or transient transfection with ERK1 siRNA showed depleted pS405 cortactin in cell cytosol and increased total cortactin expression in CCM3-KD cells. Western blot image is one of three independent experiments. c) ERK1/2 inhibition either with PD098059 or ERK1/2 siRNA was also associated with stabilization of brain endothelial barrier and decreased permeability to FITC-Inulin. Data represent an average±SD for n=4 independent experiments. *p<0.05 vs. no PD098059 treatment, ***p<0.001 vs. no ERK1/2 siRNA transfection.

Figure 8

a) Stable transfection with cortactin shRNA (CTTN-KD) in hBMEC depleted cortactin (Western blot) and increased endothelial barrier permeability (a similar effect as CCM3-KD). The permeability of FITC-inulin (5 kDa) and dextran (20 kDa) was measured 1–10 days after initial plating in CTTN-KD and control transfected brain endothelial monolayers and a permeability coefficient (PC) determined. In control cells, the permeability of FITC-inulin and dextran progressively declined with time. In contrast, with CTTN-KD cells, the reduction in dextran permeability was delayed and incomplete over 10 days, while for inulin the permeability did not decrease with time. Data are means SD; n=5 independent experiments. b) CTTN-KD cells had altered TJ protein immunofluorescence manifested as fragmented staining for claudin-5, occludin and ZO-1 and robust short stress fiber formation visualized by phalloidin staining. c) Coimmunoprecipitation and Western blot analysis showed loss of interaction between ZO-1 with claudin-5 or occludin and β-actin with ZO-1. Western blot image is one of three independent. Input represents the 50% of total cell extract used for immunoprecipitation.

Figure 9

Possible scenario of the signaling events induced by CCM3 mutation/depletion in brain endothelial cells that result in tight junction complex instability and brain endothelial barrier hyperpermeability.