Genome-wide RNAi screen reveals ALK1 mediates LDL uptake and transcytosis in endothelial cells - PubMed (original) (raw)

doi: 10.1038/ncomms13516.

John H Chidlow 1 2, Chitra Rajagopal 1 2, Michael G Sugiyama 3 4, Joseph W Fowler 1 2, Monica Y Lee 1 2, Xinbo Zhang 2 5, Cristina M Ramírez 2 5, Eon Joo Park 1 2, Bo Tao 1 2, Keyang Chen 6, Leena Kuruvilla 1, Bruno Larriveé 7, Ewa Folta-Stogniew 8, Roxana Ola 7, Noemi Rotllan 2 5, Wenping Zhou 1 2, Michael W Nagle 9, Joachim Herz 10, Kevin Jon Williams 6 11, Anne Eichmann 7, Warren L Lee 3 4 12, Carlos Fernández-Hernando 2 5, William C Sessa 1 2

Affiliations

Genome-wide RNAi screen reveals ALK1 mediates LDL uptake and transcytosis in endothelial cells

Jan R Kraehling et al. Nat Commun. 2016.

Abstract

In humans and animals lacking functional LDL receptor (LDLR), LDL from plasma still readily traverses the endothelium. To identify the pathways of LDL uptake, a genome-wide RNAi screen was performed in endothelial cells and cross-referenced with GWAS-data sets. Here we show that the activin-like kinase 1 (ALK1) mediates LDL uptake into endothelial cells. ALK1 binds LDL with lower affinity than LDLR and saturates only at hypercholesterolemic concentrations. ALK1 mediates uptake of LDL into endothelial cells via an unusual endocytic pathway that diverts the ligand from lysosomal degradation and promotes LDL transcytosis. The endothelium-specific genetic ablation of Alk1 in Ldlr-KO animals leads to less LDL uptake into the aortic endothelium, showing its physiological role in endothelial lipoprotein metabolism. In summary, identification of pathways mediating LDLR-independent uptake of LDL may provide unique opportunities to block the initiation of LDL accumulation in the vessel wall or augment hepatic LDLR-dependent clearance of LDL.

PubMed Disclaimer

Conflict of interest statement

M.W.N. is an employee of Pfizer Worldwide Research and Development, but the company had no influence in study design, data collection and analyses. Other authors declare no competing financial interests.

Figures

Figure 1

Figure 1. Screen to identify pathways regulation LDL uptake.

(a) Summary of the results from the RNAi screen. Results from genome-wide RNAi screen are red while follow-up screens are in blue. (b) Robust z-score-scores from the original screen. The inverse sigmoidal robust z-score-score distribution of genes indicates that genes can either increase or decrease DiI-LDL uptake and (c) shows the reproducibility between two individual sets of plates from the screen. (d) Representative image (scale bar, 50 μm.) of a gene knockdown of ALK1 resulting in less DiI-LDL uptake, whereas the loss of ALK1 (e) did not affect the uptake of FITC transferrin (scale bar, 50 μm). (f) Pie chart of 140 hits tested for DiI-LDL uptake inhibition in the follow-up screen, showing the number (and percentage) of active siRNAs for each hit. (g) Pathway clustering of the final 34 hits.

Figure 2

Figure 2. Validation of ALK1 in mediating LDL uptake into endothelium.

(a) Quantitative PCR analysis for the knockdown efficiency of the ALK1 siRNA in human endothelial cells (HUVEC). Data represent the mean±s.e.m. and are representative of three experiments in duplicate. *P<0.05, Student's _t_-test. (b) Western blot analysis showing the knockdown efficiency of the ACVRL1 siRNA in HUVEC based on the BMP9 (10 ng ml−1) induced phosphorylation of canonical SMAD 1/5 phosphorylation. A non-cropped western blot for this experiment can be found in Supplementary Fig. 9a. (c) DiI-LDL uptake was reduced in various human (EA.hy926 and HUVEC) and mouse (MLEC) endothelial cells treated with ACVRL1/Acvrl1 siRNA. Scale bar, 50 μm. (d) 125I-LDL uptake into WT and _Acvrl1_fl/fl/Ldlr −/− MLEC. _Acvrl1_fl/fl/Ldlr −/− MLEC were infected with AdGFP (control) or AdCre/GFP, then the uptake (includes bound LDL) of 125I-LDL was examined. Cells were kept in LPDS overnight before uptake studies. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test. (e) Cells were treated with control siRNA, LDLR siRNA or ACVRL1 siRNA and placed into regular media, washed and exposed to increasing concentrations of DiI-LDL. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test. (f) Uptake analysis of DiI-HDL, -LDL and -VLDL (2.5 μg ml−1) into endothelial cells treated with control siRNA and ACVRL1 siRNA. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test. (g) Uptake of oxidized LDL (2.5 μg ml−1) into EA.hy926 cells treated with control siRNA and ACVRL1 siRNA cultured overnight in LPDS media supplemented with 25 μg ml−1 LDL. Data represent the mean±s.e.m. and are representative of three experiments. (h) Comparison of DiI-LDL uptake into _Ldlr_-KO MEFs transfected with either GFP (negative control), ALK1-GFP, ALK2-GFP or LDLR-GFP (positive control). Data represent the mean±s.e.m. and are representative of three experiments in duplicates. *P<0.05, Student's _t_-test. (i) Uptake of DiI-LDL in EA.hy926 cells in the presence of increasing concentrations of ALK1-Fc was measured. Addition of 10−9 M ALK1-Fc is equimolar to DiI-LDL (500 ng ml−1). Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test.

Figure 3

Figure 3. ALK1 deficiency does not affect sterol sensing in the endothelium.

(a) Quantitative PCR analysis of SREBP2-dependent genes after knockdown of DNM2 or ACVRL1. The loss of DNM2 increases SREBP2-dependent gene expression, whereas the loss of ALK1 does not. Data represent the mean±s.e.m. and are representative of three experiments in duplicates. *P<0.05, Student's _t_-test. (b) Western blot analysis of the BMP9 (10 ng ml−1) induced phosphorylation of SMAD 1/5. HUVEC were incubated in LPDS and exposed to BMP9 for 60 min. In lane 2, cells were pretreated with LDL (25 μg ml−1) to downregulate LDLR. In lanes 3 and 4, ALK1 or LDLR was silenced with siRNA, respectively. A non-cropped western blot for this experiment can be found in Supplementary Fig. 9b. (c) The loss of ALK1 does not influence LDLR on the cell surface. Flow cytometric analysis of cell surface LDLR levels in endothelial cells treated with control, ACVRL1 or LDLR siRNAs. The Ab C7 was used for LDLR and IgG is an isotype control and data quantified in right panel. Data represent the mean±s.e.m. and are representative of three experiments in triplicates. *P<0.05, Student's _t_-test. (d) 125I-LDL internalization and degradation in cells treated with control, ACVRL1 or LDLR siRNAs. EA.hy926 cells were pre-incubated with LDL (25 μg ml−1) overnight and the internalization and degradation of 125I-LDL was after 4 h of incubation. ACVRL1 siRNA reduced internalization and had no effect on LDL degradation, whereas the LDLR siRNA (as a positive control) reduced LDL internalization and led to less degradation of LDL. Data represent the mean±s.e.m. and are representative of three experiments in duplicates. *P<0.05, Student's _t_-test. (e) Loss of ALK1 does not increase cellular free cholesterol. Filipin-III staining was examined in endothelial cells treated with siRNAs for ACVRL1, LDLR, DNM2 and NPC2 siRNA or treated with U18666 to enhance free cholesterol. Scale bar, 10 μm. Data are representative of at least four experiments. ns, not significant.

Figure 4

Figure 4. ALK1 rescues LDL uptake and promotes LDL uptake independent of its kinase activity.

(a) Analysis of DiI-LDL uptake in EA.hy926 treated with ACVRL1 siRNA (+) in the presence of increasing amounts of expressed ALK1 (using various MOI of AdALK1-GFP). DiI-LDL uptake was normalized by Hoechst dye stained nuclei. Data represent the mean±s.e.m. and are representative of three experiments in triplicates. *P<0.05, Student's _t_-test. (b) ALK1 dose dependently increases DiI-LDL uptake in _Ldlr_-KO MEFs. DiI-LDL data are normalized for ALK1-GFP expression in _Ldlr_-KO MEFs by using various MOI of AdALK1-GFP. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test. (c) HeLa cells were transfected with either GFP, WT, constitutively active (CA, Q201D) and inactive variant (IA, R374Q) ALK1 constructs and p-SMAD1/5 levels were examined in the absence or presence of BMP9 (10 ng ml−1) (Data are mean±s.e.m., experiment was performed three times). A non-cropped western blot for this experiment can be found in Supplementary Fig. 10a. (d) DiI-LDL uptake analysis of cells expressing GFP and the different variants of ALK1. Data represent the mean±s.e.m. and are representative of three experiments in triplicates. *P<0.05, Student's _t_-test. (e) Western bot analysis of p-SMAD1/5 after starvation using BMP9 (10 ng ml−1) or pharmacological inhibitors (ALK1ecto,400 ng ml−1: tenfold molar excess or LDN193189, 50 nM). A non-cropped western blot for this experiment can be found in Supplementary Fig. 10b. (f) BMP9 stimulation or inhibition does not affect DiI-LDL uptake into endothelial cells. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test. ns, not significant.

Figure 5

Figure 5. Cellular and direct binding of LDL to ALK1.

(a) _Ldlr_-KO MEFs were infected with AdGFP or AdALK1-GFP and the concentration-dependent binding of LDL at 4 °C determined. Cells were kept in LPDS overnight. Insert shows western blot for the expression of ALK1-GFP with HSP90 as a loading control. A non-cropped western blot for this experiment can be found in Supplementary Fig. 11a. Table shows _K_d and _B_max. Data represent the mean±s.e.m. and are representative of three experiments in triplicates. *P<0.05, Student's _t_-test. (b) SPR analysis of binding of ALK1 to LDL. The analytes Fc (0.1, 0.2, 0.6, 1.8, 5.4 μM), LDLRecto (5, 14, 43, 129, 388 nM) and ALK1ecto: (0.19, 0.57, 1.7, 5.1, 15.3 μM) were assayed for binding to immobilized LDL. Sensorgram depicts binding events from a representative experiment and table shows _K_d calculated from three independent experiments. (c) SPR analysis of LDLRecto binding on naive or ALK1ecto saturated LDL. ALK1ecto (15.3 μM) was prebound to LDL and exposed to LDLRecto (5, 14, 43, 129, 388 nM). Results show no competition between ALK1 and LDLR in binding to LDL, indicating distinct binding domains on LDL for ALK1 and LDLR. Table shows _K_d calculated from three independent experiments. (d) SPR analysis of ALK1ecto binding on naive or LDLRecto saturated LDL. LDLRecto (388 nM) was prebound to LDL and exposed to ALK1ecto (0.19, 0.57, 1.7, 5.1, 15.3 μM). This inverse experiment of Fig. 6c shows no competition between ALK1 and LDLR in binding to LDL and the inset shows _K_d calculated from three independent experiments. (e) SPR analysis of binding of Fc, BMP9, ALK1ecto and ALK1ecto/BMP9 complex to LDL (all proteins at 2 μM). This result indicates that LDL and BMP9 bind separate domains on ALK1 as ALK1 binding to LDL is not inhibited by the presence of BMP9. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test.

Figure 6

Figure 6. Time-dependent internalization and co-localization of DiI-LDL and ALK1.

(a) _Ldlr_-KO MEFs were infected with ALK1-GFP, incubated in LPDS overnight and the time-dependent internalization of DiI-LDL examined at 37 °C. Cells were imaged for ALK1-GFP (green), DiI-LDL (red) and white reflecting co-localization of ALK1-GFP/DiI-LDL (Menders correlation). The internalization of DiI-LDL and its co-localization with DiI-LDL shows at 10–20 min and accumulates in a perinuclear compartment after 60 min. Scale bar, 10 μm. The data is representative for three independent experiments. (b) Analysis of ALK1-GFP localization in control (untreated), LDL (25 μg ml−1) or BMP9 (10 ng ml−1) treated EA.hy926 cells. Cells were infected for 48 h and transferred to LPDS for the remaining 24 h. Cells were treated as described for 1 h at 37 °C, fixed and imaged. Upper panels show original confocal laser scanning microscopy images (green, ALK1-GFP; red, EEA1; blue, nuclei). Lower panels show Menders correlation (green, ALK1-GFP alone; red, EEA1 alone; white, ALK1-GFP/EEA1 co-localized). Scale bar, 20 μm. The data is representative for three independent experiments. (c) Bar graph shows Pearson correlation of these three conditions. The result indicates a co-localization of ALK1 with the early endosome marker EEA1 upon stimulation with either LDL or BMP9 within 1 h. Data represent the mean±s.e.m. and are representative of three experiments. *P<0.05, Student's _t_-test.

Figure 7

Figure 7. ALK1 mediates transcytosis and in vivo uptake of LDL.

(a) HCAECs were treated with PCSK9 to remove LDLR and transfected with either control siRNA or ACVRL1 siRNA. TIRF-based transcytosis assay was performed to measure the effect of ALK1 on LDL transcytosis. Each data point represents over five individual cells measured in six independent experiments. *P<0.05, Student's _t_-test. (b) PCSK9 treated HCAECs were transfected with GFP, ALK1 or ALK2 to measure transcytosis using TIRF imaging. Each data point represents over 15 individual cells measured in three independent experiments. *P<0.05, Student's _t_-test. (c) Transwell assay for LDL transcytosis in HCAECs incubated with 125I-labelled LDL. Data represent the mean±s.e.m. and are representative of three independent experiments with two batches of 125I-LDL. *P<0.05, Student's _t_-test (d) Cross-section imaging of EA.hy926 cells transfected with either control siRNA or ALK1 siRNA or infected with adenovirus encoding GFP or ALK1-GFP (blue, nucleus; green, top panel: lectin/bottom panel: GFP; red, DiI-LDL). The data is representative for three independent experiments. (e) Representative en face images of DiI-LDL uptake into the inner curvature of the aortic arch of _Ldlr_-KO animals with or without endothelial expression of Acvrl1 (blue, nucleus; red, DiI-LDL). Scale bar, 50 μm. (f) Quantification of DiI-LDL uptake into the inner curvature endothelium of the aortic arch _Ldlr_-KO animals with or without endothelial expression of ALK1. Data represent the mean±s.e.m. and are from 4 and 3 mice, respectively. *P<0.05, Student's _t_-test. ns, not significant.

Comment in

References

    1. Skalen K. et al. Subendothelial retention of atherogenic lipoproteins in early atherosclerosis. Nature 417, 750–754 (2002). - PubMed
    1. Tabas I., Williams K. J. & Boren J. Subendothelial lipoprotein retention as the initiating process in atherosclerosis: update and therapeutic implications. Circulation 116, 1832–1844 (2007). - PubMed
    1. Williams K. J. & Tabas I. Lipoprotein retention--and clues for atheroma regression. Arterioscler. Thromb. Vasc. Biol. 25, 1536–1540 (2005). - PubMed
    1. Williams K. J. & Tabas I. The response-to-retention hypothesis of early atherogenesis. Arterioscler. Thromb. Vasc. Biol. 15, 551–561 (1995). - PMC - PubMed
    1. Vasile E., Simionescu M. & Simionescu N. Visualization of the binding, endocytosis, and transcytosis of low-density lipoprotein in the arterial endothelium in situ. J. Cell Biol. 96, 1677–1689 (1983). - PMC - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources