Defective autophagy is a key feature of cerebral cavernous malformations

Abstract

Cerebral cavernous malformation (CCM) is a major cerebrovascular disease affecting approximately 0.3-0.5% of the population and is characterized by enlarged and leaky capillaries that predispose to seizures, focal neurological deficits, and fatal intracerebral hemorrhages. Cerebral cavernous malformation is a genetic disease that may arise sporadically or be inherited as an autosomal dominant condition with incomplete penetrance and variable expressivity. Causative loss-of-function mutations have been identified in three genes, KRIT1 (CCM1), CCM2 (MGC4607), and PDCD10 (CCM3), which occur in both sporadic and familial forms. Autophagy is a bulk degradation process that maintains intracellular homeostasis and that plays essential quality control functions within the cell. Indeed, several studies have identified the association between dysregulated autophagy and different human diseases. Here, we show that the ablation of the KRIT1 gene strongly suppresses autophagy, leading to the aberrant accumulation of the autophagy adaptor p62/SQSTM1, defective quality control systems, and increased intracellular stress. KRIT1 loss-of-function activates the mTOR-ULK1 pathway, which is a master regulator of autophagy, and treatment with mTOR inhibitors rescues some of the mole-cular and cellular phenotypes associated with CCM. Insufficient autophagy is also evident in CCM2-silenced human endothelial cells and in both cells and tissues from an endothelial-specific CCM3-knockout mouse model, as well as in human CCM lesions. Furthermore, defective autophagy is highly correlated to endothelial-to-mesenchymal transition, a crucial event that contributes to CCM progression. Taken together, our data point to a key role for defective autophagy in CCM disease pathogenesis, thus providing a novel framework for the development of new pharmacological strategies to prevent or reverse adverse clinical outcomes of CCM lesions.

Keywords: CCM; ROS; autophagy; endothelial‐to‐mesenchymal transition (EndMt); mTOR.

© 2015 The Authors. Published under the terms of the CC BY 4.0 license.

Figures

Figure 1. KRIT1-ablated cells display autophagy suppression

1. Immunoblot analysis of p62 and LC3 I/II in KRIT1 wt and KRIT1-KO endothelial cells. Actin was used as a loading control. Quantification of total LC3 on actin is reported (*P = 0.02712). The results are representative of three independent experiments.

2. Representative images of p62 dots in KRIT1 wt and KRIT1-KO endothelial cells. Scale bar, 20 μm. Magnifications in insets. Right, quantitative analysis of p62 distribution of dots is reported (four independent experiments; n = 35 cells per group). *P = 0.00542 (dotted); *P = 0.00014 (nuclear).

3. Immunoblot analysis of p62 and LC3 I/II in KRIT1-KO and KRIT1-KO re-expressing KRIT1 (KO+KRIT1) MEFs. Left, immunoblot showing KRIT1 levels in KRIT1-KO and KO+KRIT1 cells. Right, immunoblots for p62 and LC3 I/II. Actin was used as a loading marker. Quantification of total LC3 on actin is reported (*P = 0.01248). The results are representative of three independent experiments.

4. Representative images of p62 dots in KO+KRIT1 (top) and KRIT1-KO cells (bottom). Scale bar, 50 μm. Magnifications in insets. Right, quantitative analysis of the number of p62 dots per cell is shown (four independent experiments; n = 50 cells per group). *P = 7.18e−14.

5. Immunoblot analysis of hBMECs transiently transfected with control siRNA or KRIT1 siRNA. Left, evaluation of siRNA efficiency with antibody directed against KRIT1. Right, immunoblots for p62 and LC3 I/II. Actin was used as a loading marker. Quantification of total LC3 on actin is reported (*P = 0.03071). The results are representative of three independent experiments.

6. Immunoblot analysis of EA.hy926 cells transiently transfected with control siRNA or KRIT1 siRNA. Left, evaluation of siRNA efficiency with antibody directed against KRIT1. Right, immunoblot for p62 and LC3 I/II. Actin was used as a loading marker. Quantification of total LC3 on actin is reported (*P = 0.02527). The results are representative of three independent experiments.

7. Immunofluorescence analysis of p62 (green) and ProteoStat Aggresome staining detection reagent (red) in KRIT1 wt and KRIT1-KO lung endothelial cells. The yellow signal in the merged images represents an overlapping spatial relationship between green and red fluorescence. Magnification in insets. Scale bar, 50 μm. The images are representative of four independent experiments.

8. Immunofluorescence analysis of p62 (green) and ProteoStat Aggresome staining detection reagent (red) in KRIT1-KO re-expressing KRIT1 (KO+KRIT1) and KRIT1-KO MEFs. The yellow signal in the merged images represents an overlapping spatial relationship between green and red fluorescence. Magnification in insets. Scale bar, 50 μm. The images are representative of four independent experiments.

Source data are available online for this figure.

Figure 2. KRIT1 loss-of-function activates the mTOR-ULK1 pathway

1. Immunoblot analysis with antibodies directed against phosphorylated mTOR (Ser 2448), total mTOR, phosphorylated p70 S6 Kinase (Ser 371), total p70 S6 Kinase, phosphorylated 4E-BP1 (Thr 37/46), and total 4E-BP1; actin was used as a loading marker. Where indicated, KRIT1 wt and KRIT1-KO endothelial cells were treated with 100 nM Torin1 for 4 h. The results are representative of three independent experiments.

2. Immunoblot analysis of total ULK1 and actin in KRIT1 wt and KRIT1-KO endothelial cells. Where indicated, cells were treated with 100 nM Torin1 for 4 h. The results are representative of three independent experiments.

3. Immunoblot analysis of p62, LC3 I/II, and actin in KRIT1 wt and KRIT1-KO endothelial cells treated with 100 nM Torin1 or 500 nM rapamycin for 4 h. The results are representative of three independent experiments.

4. Immunoblot analysis with antibodies directed against phosphorylated mTOR (Ser 2448), total mTOR, phosphorylated p70 S6 Kinase (Ser 371), total p70 S6 Kinase, phosphorylated 4E-BP1 (Thr 37/46), and total 4E-BP1; actin was used as a loading marker. Where indicated, KRIT1-KO re-expressing KRIT1 (KO+KRIT1) and KRIT1-KO MEFs were treated with 100 nM Torin1 for 4 h. The results are representative of three independent experiments.

5. Immunoblot analysis of phosphorylated ULK1 (Ser 757), total ULK1, and actin in KRIT1 KO+KRIT1, and KRIT1 KO MEFs. Where indicated, cells were treated with 100 nM Torin1 for 4 h. The results are representative of three independent experiments.

6. Immunoblot analysis of p62, actin, LC3 I/II in KO+KRIT1 and KRIT1-KO cells. Where indicated, cells were treated with 100 nM Torin1 for 4 h or 500 nM rapamycin for 4 h. The results are representative of three independent experiments.

7. KRIT1 wt and KRIT1-KO endothelial cells were transiently transfected with mRFP-GFP-LC3. Where indicated, the cells were treated with 100 nM Torin1 for 4 h or 2 μM xestospongin B for 4 h. The differences in the autophagic flux were evaluated by counting the yellow LC3 I/II dots/cell (RFP+GFP+) and red LC3 dots/cell (RFP+GFP−) for each condition. Yellow dots: autophagosomes; red dots: autophagolysosomes. *P = 5.74e−5 (red dots, WT ctrl vs. WT Tor1); *P = 9.62e−5 (red dots, WT ctrl vs. WT xesto); *P = 0.00727 (red dots, WT ctrl vs. KO ctrl); #P = 0.00046 (red dots, KO ctrl vs. KO Tor1). The data are expressed as the mean ± s.e.m.

8. KO+KRIT1 and KRIT1-KO MEFs were transiently transfected with the mRFP-GFP-LC3 tandem construct. Where indicated, the cells were treated with 100 nM Torin1 for 4 h or 2 μM xestospongin B for 4 h. The differences in the autophagic flux were evaluated by counting the yellow LC3 I/II dots/cell (RFP+GFP+) and red LC3 dots/cell (RFP+GFP−) for each condition. Yellow dots: autophagosomes; red dots: autophagolysosomes. *P = 0.00023 (red dots, KO+KRIT1 ctrl vs. KO+KRIT1 Tor1); *P = 0.00045 (red dots, KO+KRIT1 ctrl vs. KO+KRIT1 xesto); #P = 3.08e−6 (red dots, KO ctrl vs. KO Tor1); ##P = 6.73e−5 (yellow dots, KO+KRIT1 ctrl vs. KO ctrl). The data are expressed as the mean ± s.e.m. of four independent experiments.

Source data are available online for this figure.

Figure 3. Defective autophagy underlies major phenotypic signatures of CCM disease

1. Cd44, PAI1 (also known as Serpine1), and Id1 mRNA expression levels in KRIT1 wt and KRIT1-KO endothelial cells were assessed by quantitative real-time PCR. Where indicated, KRIT1 wt and KRIT1-KO endothelial cells were treated with 100 nM Torin1 or 500 nM rapamycin for 16h. The data are expressed as the mean ± s.e.m. Cd44: *P = 0.02848 (KO ctrl vs. KO Rapa); *P = 0.02605 (KO ctrl vs. KO Tor1). PAI1: *P = 0.04446 (KO ctrl vs. KO Rapa); *P = 0.03996 (KO ctrl vs. KO Tor1). Id1: *P = 0.00266 (KO ctrl vs. KO Rapa); *P = 0.01554 (KO ctrl vs. KO Tor1). n = 3 independent experiments.

2. Immunoblot analysis of CD31/Pecam-1, vascular endothelial cadherin (VE-cadherin), and actin in KRIT1-KO endothelial cells that were treated with 100 nM Torin1 or 500 nM rapamycin for 24 h. The results are representative of three independent experiments.

3. Immunoblot analysis of CD31/Pecam-1, vascular endothelial cadherin (VE-cadherin), N-cadherin, alpha-smooth muscle actin (alpha-SMA), and actin in HUVECs transfected with control siRNA or ATG7 siRNA.

4. Formation of capillary-like structures measured by tube formation assays. HUVECs were transfected with control siRNA or ATG7 siRNA for 72 h. Representative phase-contrast (Scale bar, 100 μm) and calcein-fluorescent (Scale bar, 50 μm) images were reported. All data are presented as percentage ± s.e.m from three different experiments performed in duplicate. *P = 1.29e−11.

5. Immunoblot analysis of p62, LC3 I/II, and actin in CCM3 wt and CCM3-KO endothelial cells treated with 100 nM Torin1 or 500 nM rapamycin for 4 h. The results are representative of three independent experiments.

6. Representative immunostaining of retina sections from wt and a model of inducible and endothelial-specific CCM3-KO (CCM3-ECKO) at postnatal day 14. Endothelium was stained with isolectin B4 (ISOB4) (blue). A, artery; V, vein. p62 aggregates can be observed in endothelial cells forming retinal lesions in CCM3-ECKO animals (scale bar: 200 μm). Scale bar of magnifications: 100 μm.

7. Representative immunostaining of brain sections from wt and a model of inducible and endothelial-specific CCM3-knockout mice (CCM3-ECKO) at postnatal day 9. p62 aggregates can be observed in the proximity of CCM lesions (arrows). Cell nuclei (DAPI) are in blue. Scale bar, 30 μm. Smaller panel shows the magnifications of blood vessels (green). Scale bar, 10 μm.

Source data are available online for this figure.

Figure 4. Accumulation of p62 in endothelial cells lining human CCM lesions

p62 immunohistochemical (IHC) staining in human brain tissue.

1. A, B Normal vascular endothelium of autoptic brain parenchyma samples is lacking the typical autophagic p62 granules as shown by the negative staining for p62. Scale bars: (A) 200 μm; (B) 100 μm.

2. C–F Two different representative samples of CCM lesions with a thin, single layer brain endothelium displaying either moderate (C, D) or marked (E, F) positive perinuclear “pearl necklace-like” immunostaining for p62 granules. (C, D), case n° 4 (p62++), and (E, F), case n° 8 (p62+++) are representative of CCM cases listed in Appendix Table S1. Scale bars: (C, E) 200 μm; (D, F) 100 μm. Arrows indicate endothelial p62 positive staining.

3. G–I Hematoxylin and eosin (H&E) staining (G) and p62 immunohistochemical analysis (H, I) of a CCM surgical sample (case n° 6 in Appendix Table S1) containing normal vessels in the peri-lesional area, which served as an internal control. Notice marked p62-positive staining in endothelial cells lining a CCM lesion (H, arrows) and p62-negative staining in endothelial cells lining a normal peri-lesional vessel (I, arrows). Scale bars: (G) 300 μm; (H, I) 100 μm. Background staining in brain parenchyma surrounding CCM lesions may be attributed to either cell debris or p62 immunoreactivity in neuronal and glial cells.