Figure 1.
SNX31 is expressed in a urothelium-enriched and differentiation-dependent manner.
(A) RT-PCR: cDNA's from various mouse tissues were used to amplify sorting nexin 31 (SNX31), SNX17 and glyceraldehyde phosphate dehydrogenase (GAPDH; loading control). Lanes: (1) bladder, (2) kidney medulla, (3) kidney cortex, (4) cornea, (5) trachea, (6) lung, (7) esophagus, (8) spleen, (9) intestine, (10) skeletal muscle, (11) heart, (12) fat (skin), (13) fat (abdomen), (14) vagina, (15) uterus, (16) testis, (17) seminal vesicle, (18) ovary, and (19) water (negative control). M: size markers. (B) Differentiation-dependent expression of SNX31 and UPIIIa (assayed by semi-quantitative RT-PCR) in cultured bovine urothelial cells. Lanes: (1) mouse 3T3 cells (negative control), (2) cultured bovine bladder epithelial cells at 50% confluence, (3) 100% confluence, (4) 3 days post confluence, (5) 6 days post-confluence, (6) 9 days post-confluence, and (7) in vivo bovine bladder epithelium. (C) Monospecific antibodies to SNX31. A rabbit antiserum to SNX31 was raised against a synthetic peptide (positions 423–439 in the C-terminal region of mouse SNX31), affinity-purified and used to immunoblot total cellular lysates of: (Lane 1) mouse (M) bladder urothelium, (2) 293T cells transfected with a mouse SNX31 cDNA, (3) bovine (B) bladder urothelium, (4) 293T cells transfected with a bovine SNX31 cDNA, and (5) 293T cells transfected with an empty pcDNA3 plasmid (negative control). Note that the antibody recognized a single band of 51-kDa mouse SNX31 and a 53-kDa bovine SNX31. (D-E) Localization of SNX31 in the mouse urothelium. A vertical section of mouse bladder was double-stained using a monospecific, rabbit antiserum against SNX31 (green) and a mouse monoclonal antibody against uroplakin IIIa (red). (E) A higher magnification view showing the SNX31-positive vesicles. Note in D and E, that SNX31 was expressed mainly in the terminally differentiated umbrella cells, and that SNX31 vesicles were located slightly below the uroplakin-enriched zone. (F-H) Localization of SNX31 in a whole-mount, horizontal section of mouse urothelium by double-staining. Horizontal sections beneath the apical surface of mouse umbrella cells were double-stained for keratin K20 (red) and SNX31 (green). (F) Note that K20 forms a subapical, chicken wire-like network with no SNX31 vesicles (except the upper right corner showing the deeper cytoplasm). (G) A deeper section (below the K20 positive zone) showing the SNX31-positive vesicles. (H) Enlargement of the boxed area in panel G showing intense staining of the periphery of the SNX31-positive vesicles. B (basal cells), I (intermediate cells), U (umbrella cells), dashed line (basement membrane) and dotted line (urothelial apical surface). Note in (D), and in (F) vs. (G), that the SNX31-positive vesicles are located below the subapical Keratin 20 network. Scale bars = 10 µm (panels D, F and G) or 1 µm (E and H).
Figure 2.
SNX31 is distinguishable from other endocytic marker in mouse urothelial umbrella cells.
Vertical sections of mouse urothelium were double-stained for SNX31 (green; middle row) and several known endosomal markers (red; top row): (A) Clathrin, (B) Caveolin-2, (C) EEA1, and (D) Lamp1. Merged images with nuclear staining (DAPI) are presented (bottom row). (E) A horizontal section from a whole-mounted mouse bladder urothelium double-stained for Lamp1 (E1, red) and SNX31 (E2, green); merged image (E3). Note in (E) the clear separation of SNX31 and Lamp1 staining. Symbols are the same as in Fig. 1. Scale bar = 10 µm.
Figure 3.
SNX31 co-localizes with uroplakins in the multivesicular bodies of mouse umbrella cells.
(A) TEM of normal mouse umbrella cells showing apical plaques (P), fusiform vesicles (FV), multivesicular vesicles (MVB) and lysosomes (Lys). (B–E) EM localization of uroplakin IIIa using a monoclonal antibody AU1 (B and C: Lowicryl procedure; D and E: cryoEM). Note in (B) the uroplakin staining of apical plaques, fusiform vesicles, and multivesicular bodies (MVBs), in (C) and D) the presence of uroplakin not only at the limiting membranes (thick arrows), but also the intra-luminal vesicles (thin arrows), and in (E) the uroplakin association with some MVB invaginations. (F) EM localization of SNX31. Note the detection of SNX31 in both the limiting membranes (thick arrows) and the intra-luminal vesicles (thin arrows), but not in the neighboring fusiform vesicles. (G) Double-labeling of UPIIIa (10-nm immunogold particles; open arrows) and SNX31 (5-nm; filled arrows). Note their co-localization at the limiting membranes of MVB. Scale bar = 1 µm (in panel A), or 0.2 µm (all others).
Figure 4.
Formation of the intraluminal vesicles via the invagination of the MVB membranes.
Normal mouse bladder was fixed by high pressure freezing and freeze-substitution ([29]), cut into 60-nm sections and its urothelial umbrella cells examined by transmission EM. (A) An umbrella cell containing many MVBs lined with uroplakin plaques. (B) The limiting membranes of MVB consist of uroplakin plaques with thickened, rigid-looking leaflets (arrows) interrupted by a hinge area (asterisk), and the intraluminal vesicles (ILVs) are connected with thin, irregular filaments (filled arrowheads), and amorphous materials (open arrowhead). (C and D) Formation of ILVs via the invagination of the limiting membrane of MVBs (open arrows). Scale bar = 1 µm (A), 0.1 µm (B), or 0.2 µm (C and D).
Figure 5.
The endocytosed apical urothelial proteins are targeted to the SNX31-positive MVBs.
(A–C) The luminal surface of the mouse bladder was biotinylated for 15 minutes, followed by a chase period as indicated (30, 90 and 180 min). Samples were co-stained for biotin (red; top row) and SNX31 (green, middle), with merged images shown at the bottom. Note that the internalized, biotinylated proteins (mainly uroplakins; [45]) were initially associated with SNX31-free vesicles (open arrows), and later became co-localized with SNX31 (arrows). (D–F) EM immunolocalization showing the progressive association of the endocytosed (biotinylated) surface-proteins (mainly uroplakins) with the multivesicular vesicles (MVBs). Symbols are the same as in Fig. 1. Scale bars = 10 µm (A to C), or 0.2 µm (D to F).
Figure 6.
SNX31:uroplakin interactions. (A–B) Co-immunoprecipitation of SNX31 and uroplakins.
(A): Affinity-purified, monospecific SNX31 antibodies were crosslinked to an activated aminolink-plus resin (see Methods). 300 µg of total mouse urothelial proteins were incubated with the beads for 4h at 4°C, the immunoprecipitated proteins were resolved by SDS-PAGE and detected using antibodies as denoted (all monospecific recognizing a single band that is shown). Beads whose activated crosslinking side chains were quenched (Q), or control beads (C), were used as negative controls. Note that the anti-SNX31 pulled down not only SNX31, but also UPIIIa and UPIb (of the UPIIIa/Ib pair). UPII was pulled down, but the negative controls indicated that this was nonspecific. (B) A separate co-immunoprecipitation assay with anti-SNX31 crosslinked to the beads under two different crosslinking conditions (aSNX31a and aSNX31b); anti-Gst beads (Gst) were used as an additional negative control. (C) Co-localization of UPIIIa (red) and UPIb (green) in normal mouse urothelium. Note that UPIb and IIIa (UPIb/IIIa pair) co-localized precisely, as expected. (D) Co-localization of UPIIIa (red) and SNX31 (green) in normal mouse urothelium. Note the distinct vesicular pattern of SNX31 in umbrella cells. (E) Co-localization of UPIIIa (red) and UPIb (green) in the urothelium of the Vps33a mutant (Buff) mouse. (F) Co-localization of UPIIIa (red) and SNX31 (green) in the Buff mouse urothelium. Note the tremendously increased amount of SNX31-positive vesicles in (E), comparing with (D). (G–J) In situ Proximity Ligation Assay (PLA) in mouse umbrella cells. Paraffin sections of the mouse bladder were blocked for 1 h in 1% Fish Gelatin. Antibodies for UPIb/UPIIIa or UPIIIa/SNX31 were used to detect these two proteins. The primary antibodies were also used alone, or omitted, as negative controls (data not shown). PLA signal is shown in green. (G) PLA assay of UPIb and UPIIIa interaction in the wild-type mouse urothelium, as a positive control. (H) PLA signal of UPIIIa and SNX31 interaction in normal mouse urothelium. Note that although the interaction was less prominent as compared with UPIb/UPIIIa, punctate signals were detected corresponding most likely to MVBs. (I) PLA signal of UPIb and UPIIIa interaction in the Buff (Vps33a mutant) mouse urothelium. (J) PLA signal of UPIIIa/SNX31 interaction in the Buff mouse urothelium. Note the tremendously increased PLA signals between UPIIIa and SNX31. (K) Quantification of the PLA signals using the Duolink Image Tool software: The average numbers of signals per umbrella cells are presented. Arrows mark the apical surface; other symbols are the same as in Fig. 1. Scale bars = 10 µm (panels C to J)
Figure 7.
Association of SNX31 with uroplakin IIIa during sedimentation equilibrium centrifugation.
Total bladder urothelial proteins from wild type and buff mice were subjected to a flotation analysis on a sucrose density gradient. The total homogenate was loaded at the bottom of a continuous sucrose gradient (0.4 to 2.0 M sucrose), and membranes were separated (fraction 1 represents the lightest and 20 the heaviest). LZ (loading zone) and P (pellet). Following centrifugation, proteins from various fractions were immunoblotted for SNX31, UPIIIa and Lamp1. Note that in normal mouse urothelium (A), SNX31 largely co-distributed with the uroplakin peak with the remaining forming a smear, and that, in Buff mouse (B), almost all the SNX31 co-floated with UPIIIa.
Figure 8.
Association of SNX31 with the early endosomes in MDCK cells.
MDCK cells were transfected with (A) SNX31 alone, or co-transfected SNX31 with (B) uroplakins UPIa/II, (C) UPIb/IIIa, or (D) UPIb/IIIaΔC, fixed 48 hrs post-transfection, and double-stained using antibodies to SNX31 (green) and (A) EEA1, (B) UPIa of the UPIa/II pair, or (C and D) UPIb of the UPIb/IIIa pair (red). Merged images are shown to the right. Note in (A) the co-localization of SNX31 with the EEA1-positive early endosomes, in (B and C) the co-localization of SNX31 with the endocytosed uroplakins, and in (D) that the deletion of the cytoplasmic tail of UPIIIa had no effects on the distribution of SNX31. Scale bar = 10 µm (all panels).
Figure 9.
Uroplakin knockout diminishes the membrane association of SNX31.
(A and B) Immunofluorescent double-staining of uroplakin IIIa (top) and SNX31 (middle) in wild type (A; Wt) and UPII knockout (B; 2KO) mouse bladder urothelium. Note in (B) the diminished UPIIIa staining and the loss of the characteristic MVB-associated SNX31 staining pattern. (C) Effects of UPII-knockout on the mRNA (C1, RT-PCR) and protein levels of SNX31 (C2, Western blot or WB) in wild-type (odd-numbered lanes) and UPII-deficient urothelia (even-numbered). (C1) RT-PCR analyses of (lanes 1–2) UPII, (3–4) SNX31, and (5–6) GAPDH (loading control). (C2) Western blot (WB) analyses of (lanes 1–2) UPII, (3–4) SNX31, and (5–6) actin (loading control). (C3) Quantification of UPII (left) and SNX31 (right) protein amounts normalized against actin, in total mouse protein extracts of the wild-type (dark bars) and UPII KO (grey) mouse bladder urothelia. Values are means±S.D. Note the reduced SNX31 protein levels (∼35% decrease) in the UPII KO mouse (t-test; p<0.01). (D) Effects of UPII-ablation on the membrane association of SNX31. (D1) Equivalent amounts of proteins from various fractions of normal (lanes 1–4) and UPII-KO mouse bladder epithelial cells (lanes 5–8) were immunoblotted for SNX31. Lanes (1 and 5; T) total, (2 and 6; N) nuclear, (3 and 7; MA) membrane-associated, and (4 and 8; Sol) soluble proteins. (D2) Quantification of SNX31 protein, as a fraction of the total protein, in the wild-type (left) and UPII KO (right) mouse urothelial extracts. U (umbrella cell); S (superficial cell of the uroplakin-deficient urothelium); other symbols are the same as in Fig. 1. Note the 3–4-fold decrease in the amount of membrane-associated SNX31 in the UPII KO comparing with the wild-type (t-test; p<0.02). Bars = 10 µm.
Figure 10.
SNX31 preferentially binds to PtdINs(3)P.
(A) Immunoaffinity-purified PIP2:GST and SNX31:HA fusion proteins were used to overlay PIP strips (Echelon) spotted with 100-pmol of various phosphoinositides. The bound proteins were detected with antibodies to GST and HA epitopes. (B) MDCK cells were transfected with SNX31 and stained using affinity-purified anti-SNX31 antibodies (top panel) and EEA1 antibody (middle), with merged images at the bottom. (C) Cells were treated with wortmannin (100-nM, 30 min), a PtdIns(3)P-kinase inhibitor. Abbreviations are: N (nucleus), LPA (lysophosphatidic acid), LPC (lysophosphacholine), PA (phosphatic acid), PC (phosphatidylcholine), PE (phosphatidyl-ethanolamine), PS (phosphatidylserine) and S1P (sphingosine-1-phosphate). Note that PtdIns(3)P-K inhibition abolished the association of both SNX31 and EEA1 with the early endosomal membranes, leading to a diffuse, cytoplasmic staining pattern. Bars = 10 µm.
Figure 11.
A model for the possible roles of SNX31 in the endocytic degradation of uroplakins.
The model illustrates that: (a) The apical surface of mouse bladder urothelium is lined by urothelial plaques consisting of hexagonally packed, 2D crystals of 16–nm uroplakin particles (lollipods). (b) Apical uroplakins are endocytosed by a poorly understood (hence question-marked), clathrin- and caveolin-independent process leading to the possible formation of small, early endosomes (EE) or endocytic ‘fusiform-like vesicles’ (‘FV’ to distinguish them from the regular, exocytic fusiform vesicles, FV; see Figure 8 in [45]). Although the binding of SNX31 to some EE cannot be excluded, such structures have not yet been detected. (c) The multivesicular bodies (MVBs) of bladder urothelial umbrella cells are highly specialized as they are lined by SNX31-associated uroplakin plaques. (d) SNX31 and uroplakins co-enter into the intraluminal vesicles (ILVs). It is hypothesized that the binding of SNX31 to the cytoplasmic tail of UPIIIa can cause the uroplakin particles to dissociate from the plaque and to collapse (changing the shape of the luminal 16-nm particles from circle to triangle in the diagram) thus facilitating the invagination of the uroplakin-containing membrane to form ILVs. (e) MVBs fuse with lysosomes for uroplakin degradation [35], [36]; this process is blocked in Buff mouse carrying a Vps33a mutation. This model depicts the possible roles of SNX31 in facilitating the invagination of the MVB-associated uroplakins enabling them to enter into the intraluminal vesicle compartment. See Discussion for details. EE (early endosome), FV' (endocytic ‘fusiform vesicles-like vesicles’), ILV (intraluminal vesicle), L (lumen), MVB (multivesicular vesicle), SNX31-LBD (lipid binding PX domain), SNX31-PBD (protein-binding domain) and UP (uroplakin).