Synaptojanin 1-mediated PI(4,5)P2 hydrolysis is modulated by membrane curvature and facilitates membrane fission

Abstract

Phosphatidylinositol-4,5-bisphosphate [PI(4,5)P₂] plays a fundamental role in clathrin-mediated endocytosis. However, precisely how PI(4,5)P₂ metabolism is spatially and temporally regulated during membrane internalization and the functional consequences of endocytosis-coupled PI(4,5)P₂ dephosphorylation remain to be explored. Using cell-free assays with liposomes of varying diameters, we show that the major synaptic phosphoinositide phosphatase, synaptojanin 1 (Synj1), acts with membrane curvature generators/sensors, such as the BAR protein endophilin, to preferentially remove PI(4,5)P₂ from curved membranes as opposed to relatively flat ones. Moreover, in vivo recruitment of Synj1's inositol 5-phosphatase domain to endophilin-induced membrane tubules results in fragmentation and condensation of these structures largely in a dynamin-dependent fashion. Our study raises the possibility that geometry-based mechanisms may contribute to spatially restricting PI(4,5)P₂ elimination during membrane internalization and suggests that the PI(4,5)P₂-to-PI4P conversion achieved by Synj1 at sites of high curvature may cooperate with dynamin to achieve membrane fission.

Figure 1. PIP₂ metabolism is affected by the size of substrate liposomes

A) Relationship between the liposome diameter and relative curvature. The diameters illustrated are those of the membrane pores through which extrusion of liposomes were performed and thus reflect the theoretical diameters of liposomes used. For actual measured diameters, see Extended Experimental Procedures. B) Autoradiography of a TLC showing levels of incorporated radiolabeling on phospholipids when liposomes were incubated with brain cytosol in the presence of [γ-³²P]-ATP. C) Quantification of phospholipid radiolabeling by phosphorimaging. The counts of PIP2, PIP and PA were normalized to total counts. D) The ratio of PIP₂/PIP for the different sized liposomes. n = 6 for 800 nm and 50 nm, n = 4 for 400 nm and 100 nm. Data are represented as mean ± SEM. See also Fig. S1.

Figure 2. The curvature effect in PIP₂ metabolism is mediated by the proteins Synj1 and endophilin

A) Western blot of brain cytosol from Synj1^+/+ (WT) and Synj1^−/− (KO) newborn mice. B) Analysis of the hydrolysis of soluble PI(4,5)P₂ when incubated with brain cytosol from Synj1^+/+ and Synj1^−/− newborn mice. C) The ratio of PIP₂/PIP when 800 nm and 50 nm liposomes are incubated with brain cytosol from Synj1^+/+ (WT) and Synj1^−/− (KO) newborn mice in the phospholipid radiolabeling assay. D) Western blot on the rat brain cytosols depleted of either Synj1 (using recombinant SH3 domain of endophilin fused to GST, which also results in the removal of dynamin) or endophilin (using recombinant PRD domains of both Synj1 and dynamin 1 fused to GST). E) The ratio of PIP₂/PIP when 800 nm and 50 nm liposomes are incubated with the biochemically-depleted cytosols. Data are represented as mean ± SEM. ***, p < 0.001; n=6 in A) and n=6–10 in B). See also Fig. S1.

Figure 3. Synj1 preferentially hydrolyzes PI(4,5)P₂ on high curvature membranes

A) Western blot analysis and B) quantification by infrared detection of Synj1, endophilin and dynamin recruitment to membranes when brain cytosol is incubated with 800 nm and 50 nm liposomes. C) Normalization of Synj1 and dynamin to endophilin levels. D) Time course of PI(4,5)P₂ hydrolysis by Synj1 using measurements of free phosphate when 800 nm or 50 nm liposomes are incubated with recombinant Synj1 (12 nM) in the presence or absence of recombinant endophilin (120 nM). E) Measurement of BODIPY- PI(4,5)P₂ hydrolysis in conditions where His-Synj1 recruitment is equalized using NiNTA liposomes. Data are represented as mean ± SEM. ***, p < 0.001; ** p < 0.01; n=12 in B–C), n=4 in D), and n=15 in E). See also Fig. S2 and Table S1.

Figure 4. Endophilin binds to highly phosphorylated phosphoinositides and, upon overexpression, induces tubules originating from the plasma membrane that are enriched in PI(4,5)P₂

A) Binding of endophilin domains (full-length, N-BAR and BAR) to phosphoinositides were evaluated using liposome sedimentation with a panel of liposomes composed of different phospholipids. B) Binding of full-length endophilin to to PI(4,5)P₂ liposomes with or without PS. C) Co-expression of endophilin-GFP and the PLCδ1-PH-mRFP in COS-7 cells. D) Co-expression of endophilin-GFP with the surface potential marker R-pre-mRFP in COS-7 cells. E) Treatment of endophilin-GFP transfected cells with the lipophilic probe FM 4–64 at t= 0 and 15 minutes. Data are represented as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001; n=3 in A) and n=6 in B). Scale bar = 5 μm. See also Fig. S3.

Figure 5. Acute recruitment of the inositol 5-phosphatase domain of Synj1 to endophilin-induced tubules results in membrane fragmentation and condensation

A) Diagram of the heterodimerization constructs: rat endophilin1 without the SH3 domain (endoΔSH3) is fused to the NH₂-terminal side of the FRB domain followed by EGFP; the inositol 5-phosphatase domain of Synj1, either wild type or catalytically dead mutant D730A (Synj1-Ptase, Synj1-PtaseD730 respectively), is fused to the NH₂-terminal side of two FKBP domains followed by mRFP. B) In transfected COS-7 cells, Synj1-Ptase shows a diffuse signal before rapalog treatment (pre-rapa). Rapalog treatment results in Synj1-Ptase recruitment to endophilin-induced membrane tubules and fragmentation and condensation of the membrane tubules take place (post-rapa). C) A time-lapse view of the fragmentation and condensation events upon rapalog treatment from a magnified field of the cell seen in B) as indicated by the dotted square. Examples of fragmentation sites are indicated by the arrows. D) In transfected COS-7 cells, Synj1-PtaseD730A shows a diffuse signal before rapalog treatment (pre-rapa). Rapalog addition results in Synj1-PtaseD730A recruitment to endophilin-induced membrane tubules but no fragmentation/condensation events are observed (post-rapa). E) Reduction of tubule size as a measure of membrane fragmentation and condensation efficiency. To quantify the extent of membrane fission, tubule length before and after rapamycin/rapalog-treatment were measured. Data are represented as mean ± SEM. ***, p < 0.001; *, p < 0.05; n=24,9,5,6,7,8,9,22 (left to right).. Scale bars represent B) 10 μm, C) 5 μm, and D) 5 μm. See also Fig. S3 and Movies S1–4.

Figure 6. Characterization of membrane dynamics during Synj1-induced membrane fission

A) The PI(4,5)P₂ marker PLC1δ-PH localizes to endophilin-induced tubules prior to rapamycin-treatment. This marker translocates into the cytoplasm upon rapamycin-treatment demonstrating the hydrolysis of PI(4,5)P₂ by the Synj1-Ptase. B) Using FM 4–64 as a membrane marker, fragmentation and beading of the membrane are observed. C) FBP17-F-BAR-induced membrane tubules do not undergo membrane re-organization upon Synj1-Ptase recruitment. Scale bars represent A) 5 μm, B) 10 μm (left) and 5 μm (right), and C) 10 μm. See also Movies S5–7.

Figure 7. Synj1-induced membrane fission depends upon dynamin

A) Staining of COS-7 cells expressing the endoΔSH3-FRB-EGFP with anti-dynamin antibody shows that the endophilin-induced tubules are coated with endogenous dynamin 2. B) Pre-treatment of transfected cells with the dynamin inhibitor dynasore prevents the fragmentation and condensation events induced by recruitment of Synj1-Ptase to the endophilin-induced tubules by rapamycin treatment. C) Effect of the single (K44A or K142A) and double (K44A/K142A) dynamin mutants on Synj1-Ptase-induced membrane fragmentation and condensation. Scale bars represent A) 25 μm, B) 5μm, and C) 5 μm (left), 5 μm (middle), and 10 μm (right). See also Movies S8–9.