Sorting nexin-2 is associated with tubular elements of the early endosome, but is not essential for retromer-mediated endosome-to-TGN transport

Abstract

Sorting nexins are a large family of phox-homology-domain-containing proteins that have been implicated in the control of endosomal sorting. Sorting nexin-1 is a component of the mammalian retromer complex that regulates retrieval of the cation-independent mannose 6-phosphate receptor from endosomes to the trans-Golgi network. In yeast, retromer is composed of Vps5p (the orthologue of sorting nexin-1), Vps17p (a related sorting nexin) and a cargo selective subcomplex composed of Vps26p, Vps29p and Vps35p. With the exception of Vps17p, mammalian orthologues of all yeast retromer components have been identified. For Vps17p, one potential mammalian orthologue is sorting nexin-2. Here we show that, like sorting nexin-1, sorting nexin-2 binds phosphatidylinositol 3-monophosphate and phosphatidylinositol 3,5-bisphosphate, and possesses a Bin/Amphiphysin/Rvs domain that can sense membrane curvature. However, in contrast to sorting nexin-1, sorting nexin-2 could not induce membrane tubulation in vitro or in vivo. Functionally, we show that endogenous sorting nexin-1 and sorting nexin-2 co-localise on high curvature tubular elements of the 3-phosphoinositide-enriched early endosome, and that suppression of sorting nexin-2 does not perturb the degradative sorting of receptors for epidermal growth factor or transferrin, nor the steady-state distribution of the cation-independent mannose 6-phosphate receptor. However, suppression of sorting nexin-2 results in a subtle alteration in the kinetics of cation-independent mannose 6-phosphate receptor retrieval. These data suggest that although sorting nexin-2 may be a component of the retromer complex, its presence is not essential for the regulation of endosome-to-trans Golgi network retrieval of the cation-independent mannose 6-phosphate receptor.

Fig. 1

SNX2 binds PtdIns(3)P and PtdIns(3,5)P₂. (A) Bacterially expressed SNX2 was incubated with sucrose-loaded liposomes formed from a lipid mixture of PtdCho, PtdSer and PtdEth supplemented with 20% of the relevant phosphoinositide species. SNX2 association with pelleted liposomes was assayed by western blotting with SNX2-specific antisera. Results are representative of more than three independent experiments. (B) Relative amounts of phosphoinositides were titrated down while keeping total lipid mass constant. SNX2 association with the lipid pellet was determined by volume integration of western blots. Results were averaged for at least three separate experiments. (C) The PX domain mutant SNX2(K211A) is unable to associate with phosphoinositides. Results are representative of more than three independent experiments.

Fig. 2

SNX2 is capable of sensing membrane curvature but does not induce membrane tubulation in vitro or in vivo. (A) Variously curved liposomes were formed by extrusion, and binding of proteins was assayed by sedimentation. SNX2 bound preferentially to highly curved membranes, as did SNX1. The epsin1 ENTH domain (which strongly tubulates membranes), and Dab2 (which does not change liposome shape) were insensitive to curvature. Amount of bound protein was normalised to the value for 0.8 μm liposomes. Average values for three experiments ± s.d. are shown. (B) Liposomes derived from brain were incubated with 20 μM BSA (i) or 20 μM of full length recombinant SNX1 (ii,iii) or SNX2 (iv,v). Whereas budding tubular profiles and occasionally tubular networks were observed with SNX1, no tubules were observed when using SNX2. For SNX1 (ii,iii) and SNX2 (iv,v) the data represents two independent experiments. (C). HeLa cells were transiently transfected with a construct encoding for GFP-SNX2. After 48 hours of incubation, cells expressing low and high levels of GFP-SNX2 were fixed and imaged by confocal microscopy. At low levels, SNX2 was associated with punctate cytoplasmic vesicles, which contrasted with the swollen endosomal vacuoles observed under conditions of high level expression. Bars, 10 μm.

Fig. 3

At low-level expression, GFP-SNX2 is associated with elements of the early endosome. HeLa cells were transiently transfected with GFP-SNX2 and, after 24 hours, cells were fixed and stained for the early endosomal markers EEA1 (A) and SNX1 (B), internalised receptors for transferrin (C) and EGF (D), and the late endosomal marker LAMP1 (E). Images are representative of more than 10 imaged cells in each case. Bars, 10μm.

Fig. 4

In HeLa cells endogenous SNX2 and SNX1 co-localise on an early endosomal compartment. (A) HeLa cells were treated with control, SNX2-specific, SNX1-specific or both SNX2 and SNX1 siRNA duplexes for 72 hours. At this time, the levels of SNX2 and SNX1 were determined by western blotting with SNX2 or SNX1 antisera. In all cases loading was controlled by probing samples with an anti-tubulin antibody. Quantification of the data revealed that, for individual siRNA treatment using SNX2-specific or SNX1-specific duplexes, the level of SNX2 and SNX1 expression was suppressed by 82% and 93%, respectively. Under conditions of simultaneous duplex treatment, SNX2 and SNX1 were suppressed by 81% and 92%, respectively. (B) HeLa cells were treated with control, SNX2-specific, SNX1-specific or both SNX2 and SNX1 siRNA duplexes for 72 hours prior to fixation. Cells were stained for endogenous SNX1 and SNX2. Bar, 10 μm. (C) HeLa cells were fixed and co-stained with SNX1- and SNX2-specific antibodies to reveal the endogenous distribution of these proteins. Bar, 10 μm.

Fig. 5

Endogenous SNX2 and SNX1 co-localise on tubular elements of the early endosome. Ultrathin cryosections of HepG2 cells showing the co-localization of SNX2 with SNX1 and CI-MPR on endosomal tubules. SNX2 (15 nm gold particles) and SNX1 (10 nm gold particles) co-localize on tubules emanating from endosomal (E) vacuoles (arrow in A) and on numerous vesicular-tubular profiles in close vicinity of endosomal vacuoles (arrows in B). SNX2 (15 nm gold particles) associates with CI-MPR (10 nm gold particles)-positive endosomal buds (arrows in C) and similar endosome-associated vesicular-tubular profiles, as shown in B (arrow in D). Bars, 200 nm

Fig. 6

Suppression of SNX2 or the joint suppression of SNX2 and SNX1 has no gross affect on EGF or transferrin receptor sorting. (A) HeLa cells were treated with either control, SNX2-specific siRNA or jointly with SNX1- and SNX2-specific siRNA duplexes for 72 hours. Cells were serum-starved for 3 hours and labelled with 1 kBq per well ¹²⁵I-EGF for 1 hour at 4°C, allowed to internalise surface bound ¹²⁵I-EGF for 5 minutes at 37°C, and then returned to 4°C. Cells were chased into 100 ng/ml cold EGF-containing media for various times at 37°C. Recycled, degraded and internalised fractions were subjected to gamma counting. Data is the average±s.d. from three independent experiments. (B) HeLa cells were treated with control, SNX2-specific siRNA or jointly with SNX1- and SNX2-specific siRNA duplexes for 72 hours. Cells were serum-starved for 3 hours and labelled with 1 kBq per well ¹²⁵I-transferrin for 60 minutes at 37°C. Cells were chased into 50 μg/ml cold transferrin-containing media for various times at 37°C. Recycled, degraded and internalised fractions were subjected to gamma counting. Data is the average±s.d. from three independent experiments. (C) Western analysis of a typical SNX1 and SNX2 suppression achieved during one of the three sets of receptor trafficking assays.

Fig. 7

Endogenous SNX2 shows a high degree of co-localisation with the retromer component mVps26. (A) Co-localisation of endogenous SNX2 and endogenous mVps26. Cells were plated on coverslips, fixed and stained for SNX2 and mVps26. (B) Suppression of SNX1, SNX2, or joint suppression of both SNX1 and SNX2 does not affect mVps26 localisation. HeLa cells were treated with control, SNX1-specific siRNA, SNX2-specific siRNA or jointly with SNX1- and SNX2-specific siRNA duplexes for 72 hours. Cells were then fixed and stained against endogenous SNX1, endogenous SNX2 and endogenous mVps26. Individual or joint suppression of SNX1 and SNX2 does not affect punctate distribution of mVps26. Bars, 20 μm (A,B).

Fig. 8

Suppression of SNX2 does not result in a detectable defect in the steady-state distribution of the CI-MPR. (A) HeLa cells were treated with control, SNX1-specific siRNA, SNX2-specific siRNA or jointly with SNX1- and SNX2-specific siRNA duplexes for 72 hours. Cells were fixed and stained against endogenous SNX1, endogenous SNX2 and endogenous CI-MPR. In cells subjected to RNAi against SNX2, the CI-MPR remains at a perinuclear structure that has been previously characterised as the TGN. In cells in which SNX1 and SNX2 have been suppressed, the CI-MPR undergoes a limited redistribution at steady state to peripheral structures. Bars, 20 μm. (B) HeLa cells were treated twice with control, SNX1-specific siRNA, SNX2-specific siRNA or jointly with SNX1- and SNX2-specific siRNA duplexes for two consecutive periods of 72 hours. At this time, cells were washed into cycloheximide (40 μg/ml)-containing media for 0, 4, 8 or 12 hours. Cells were lysed and the levels of endogenous CI-MPR present were resolved by western blotting with CI-MPR-specific antisera, followed by volume integration. Results are presented graphically from the averages of four independent experiments.

Fig. 9

In HeLaM cells stably expressing a CD8-CI-MPR chimera, suppression of SNX2 induces a subtle change in the rate of sorting to the TGN. (A) Images depicting the kinetics of delivery of a cell-surface-labelled CD8-CI-MPR chimera to the TGN in HeLaM cells. Bars, 20 μm. (B) HeLaM cells stably expressing a CD8-CI-MPR chimera were treated with control, SNX1-specific, SNX2-specific, or SNX1- and SNX2-specific siRNA and subjected to an anti-CD8 antibody uptake experiment as described in Materials and Methods. Briefly, CD8-CI-MPR at the cell surface were labelled with anti-CD8, and allowed to internalise for 2, 8, 16 or 24 minutes. Cells were fixed and the amount of anti-CD8 that had reached the TGN46-labelled TGN was quantified by immunofluoresence using LCSLite (Leica). Results are expressed ± the s.e.m. from three independent experiments (except the 2 minute time point, which is from two independent experiments). The key indicates the siRNA treatment. (C) Western blot indicating representative levels of SNX1 and SNX2 suppression.