Bro1 stimulates Vps4 to promote intralumenal vesicle formation during multivesicular body biogenesis

Abstract

Endosomal sorting complexes required for transport (ESCRT-0, -I, -II, -III) execute cargo sorting and intralumenal vesicle (ILV) formation during conversion of endosomes to multivesicular bodies (MVBs). The AAA-ATPase Vps4 regulates the ESCRT-III polymer to facilitate membrane remodeling and ILV scission during MVB biogenesis. Here, we show that the conserved V domain of ESCRT-associated protein Bro1 (the yeast homologue of mammalian proteins ALIX and HD-PTP) directly stimulates Vps4. This activity is required for MVB cargo sorting. Furthermore, the Bro1 V domain alone supports Vps4/ESCRT-driven ILV formation in vivo without efficient MVB cargo sorting. These results reveal a novel activity of the V domains of Bro1 homologues in licensing ESCRT-III-dependent ILV formation and suggest a role in coordinating cargo sorting with membrane remodeling during MVB sorting. Moreover, ubiquitin binding enhances V domain stimulation of Vps4 to promote ILV formation via the Bro1-Vps4-ESCRT-III axis, uncovering a novel role for ubiquitin during MVB biogenesis in addition to facilitating cargo recognition.

Figure 1.

Bro1^ΔBOD supports ILV formation. (A) Three-dimensional models reconstructed from 200-nm thick-section electron tomograms of bro1Δ (GOY65) and bro1^ΔBOD (bro1Δ::TEF1p-bro1^ΔBOD; CTY2) cells. The bro1Δ cells have class E compartments, which are flattened stacks of endosomal membranes that generally lack internal vesicles; these stacks are shown in different colors to differentiate individual membranes. For bro1^ΔBOD, the limiting membrane of MVBs are labeled yellow, the ILVs are highlighted in red, and the vacuole limiting membrane is labeled as red mesh. Scale bar = 100 nm. (B) WT (SEY6210), bro1Δ (GOY65), and bro1^ΔBOD (CTY2) were analyzed by ET and quantified for number of ILV per MVB. Asterisk indicates statistically significant differences compared with WT and bro1Δ (P < 0.0001). (C) WT (SEY6210) and bro1^ΔBOD (CTY2) were analyzed by ET and quantified to assess ILV size (diameter), individual BP size (surface area), and the frequency of incomplete ILV budding events (BPs per MVB); 12 MVBs from WT cells (SEY6210) containing 337 ILVs and 11 budding intermediates (BPs), 64 MVBs from bro1^ΔBOD cells (CTY2) containing 831 ILVs and 17 budding intermediates, and 32 class E endosomal compartments from bro1Δ cells (GOY65) were quantified. Data are represented as mean ± SEM. Asterisks indicate statistically significant differences compared with WT (BP per MVB, P = 0.0016; BP size, P = 0.0048).

Figure S1.

Representative images of cells stained with NBD-PC and FM4-64. This figure complements Fig. 3. (A) Three-dimensional models reconstructed from 200-nm thick-section electron tomograms of bro1Δ (GOY65) cells. This image depicts rare MVBs that contains an ILV. Normal-like endosomes are highlighted by yellow limiting membrane, while other colors depict flattened or tubular endosomes devoid of vesicles. The ILVs are highlighted in red. Scale bar = 100 nm. (B) Selection criteria for scoring positive NBD-PC cells. A usable cell is defined by having (1) readily identifiable FM4-64 labeling of vacuole membrane, (2) readily identifiable NBD-PC signal, and (3) defined vacuoles. An unusable cell is defined by having (1) a fragmented vacuole and (2) out of focus. A positive cell is defined by a diffuse NBD-PC signal within the lumen while lacking a distinct ring on the limiting membrane of the vacuole (defined by FM4-64), while a negative cell is defined by the colocalization of NBD-PC and FM4-64 on the limiting membrane of the vacuole. Scale bars = 5 µm. (C) Representative micrographs of bro1^ΔBOD (bro1Δ::TEF1p-bro1^ΔBOD; CTY2) and bro1^ΔBOD vps4Δ (CTY5) cells stained with NBD-PC to assess lipid MVB sorting and FM4-64 to label vacuoles. Scale bars = 5 µm.

Figure 2.

Bro1^ΔBOD does not support efficient MVB cargo sorting. (A) WT (SEY6210), bro1Δ (GOY65), or bro1^ΔBOD (bro1Δ::TEF1p-bro1^ΔBOD; CTY2) cells were transformed with the indicated GFP-tagged cargo plasmid to assess MVB sorting using live-cell fluorescence microscopy. Percentage of cells with WT sorting signal was quantified and calculated from at least 183 cells from four independent experiments performed on four different days from three different transformations. White dashed lines indicate cell boundaries. Scale bars = 5 µm. (B and C) Representative immunoblots showing processing of Sna3-GFP (B) or Mup1-GFP (C), including liberated GFP and ubiquitylated species in WT (SEY6210), vps4Δ (MBY3), bro1Δ (GOY65), bro1^ΔBOD (bro1Δ::TEF1p- bro1^ΔBOD ; CTY2), bro1^ΔBOD snf7Δ (CTY12), or bro1^ΔBOD vps4Δ (CTY5) cells. Pgk1 served as loading control. Data were quantified from four independent experiments performed on four separate days from two transformations. Data are represented as mean ± SD. Asterisks indicate a statistically significant difference (P < 0.05; B: BRO1∆ vs. VPS4∆ P = 0.04, BRO1∆ vs. bro1^∆BOD P = 0.008; C: BRO1∆ vs. VPS4∆ P = 0.0006, BRO1∆ vs. bro1^∆BOD P = 0.0255), and number signs indicate a statistically significant difference compared with WT (P < 0.0001).

Figure S2.

Bro1^ΔBOD does not support efficient MVB cargo sorting of Mup1, Mup1-Ub, and Cos5. This figure complements Fig. 2. (A) Representative immunoblots showing processing of Mup1-GFP or Mup1-GFP-Ub, including liberated GFP in WT (SEY6210), vps4Δ (MBY3), bro1Δ (GOY65), bro1^ΔBOD (bro1Δ::TEF1p- bro1^ΔBOD ; CTY2), bro1^ΔBOD snf7Δ (CTY12), bro1^ΔBOD vps4Δ (CTY5), or bro1^ΔBOD doa4Δ (CTY13) cells. Pgk1 serves as loading control. (B) Representative immunoblots processing of Cos5-GFP, including liberated GFP in WT (SEY6210), vps4Δ (MBY3), bro1Δ (GOY65), bro1^ΔBOD (bro1Δ::TEF1p- bro1^ΔBOD ; CTY2), bro1^ΔBOD snf7Δ (CTY12), or bro1^ΔBOD vps4Δ (CTY5) cells. Pgk1 serves as loading control. Data were quantified from three independent experiments performed on three separate days from two transformations. Data are represented as mean ± SD, Asterisks indicate a statistically significant difference (P < 0.05), and number signs indicate statistically significant difference compared with WT (P < 0.0001). (C) Lysates generated from bro1Δ (GOY65), WT (SEY6210), and TEF1p-bro1^ΔBOD (CTY2) cells expressing GFP-Dap2 were analyzed by immunoblotting using antibodies against GFP and Pgk1 (loading control).

Figure 3.

Bro1^ΔBOD sorting of NBD-PC into the vacuolar lumen is dependent on the Vps4–ESCRT machinery. (A) Domain cartoon of Bro1 (aa 1–844) with interacting factors annotated. (B) Sample micrographs of WT (SEY6210), bro1Δ (GOY65), and bro1^ΔBOD (bro1Δ::TEF1p- bro1^ΔBOD; CTY2) cells costained with NBD-PC and FM4-64, revealing endpoints of the observed phenotypes. Scale bars = 5 µm. (C) NBD-PC– and FM4-64–stained WT (SEY6210), TEF1p-BRO1 (CTY1), vps4Δ (MBY3), bro1Δ (GOY65), TEF1p- bro1^ΔBOD (CTY2), TEF1p-bro1^V (CTY4), bro1^ΔBOD vps4Δ (CTY5), bro1^ΔBOD vps27Δ (CTY29), bro1^ΔBOD vps37Δ (CTY21), bro1^ΔBOD vps22Δ (CTY24), bro1^ΔBOD snf7Δ (CTY12), bro1^ΔBOD vps4Δ(CTY5), and bro1^ΔBOD doa4Δ(CTY13) cells were analyzed by live-cell fluorescence microscopy and quantified for the frequency of cells able to support NBD-PC trafficking to the vacuolar lumen. Data are represented as mean ± SEM. Asterisks indicate statistically significant differences (P < 0.006), and number signs indicate statistically significant difference compared with bro1Δ (P < 0.0001). Percentage of cells with WT sorting signal was quantified and calculated from at least 94 cells from four independent experiments performed on four different days. (D and E) bro1^V (CTY4) were analyzed by ET. 3D reconstructions of the tomogram are shown in D, and ILVs per MVB are quantified in E. 20 MVBs from WT (SEY6210), 32 MVBs from bro1Δ (GOY65), 64 MVBs from bro1^ΔBOD (CTY2) and 72 MVBs from bro1^V (CTY4) were quantified. Data are represented as mean ± SEM. The limiting membrane of normal-like MVBs are labeled yellow, while the limiting membrane of tubular/aberrant MVBs are shown in different colors. ILVs are highlighted in red. A minimum of 13 MVBs from at least 10 cells were quantified. Scale bar = 100 nm. Asterisks indicate statistically significant differences comapred to bro1 (P < 0.0001). (F) WT (SEY6210), bro1Δ (GOY65), or bro1^V (bro1Δ::TEF1p-bro1^V; CTY4) cells were transformed with the GFP-CPS plasmid to assess MVB sorting using live-cell fluorescence microscopy. The percentage of cells with WT sorting signal was quantified and calculated from at least 156 cells from three independent experiments performed on three different days from two different transformations. White dashed lines indicate cell boundaries. Scale bars = 5 µm.

Figure S3.

bro1^ΔBOD-supported ILV formation requires the Vps4–ESCRT machinery. This figure complements Fig. 3. (A) Lysates generated from WT (SEY6210), vps4Δ (MBY3), bro1Δ (GOY65), TEF1p-BRO1 (CTY1), bro1^ΔBOD (370–844; bro1Δ::TEF1p-bro1^ΔBOD; CTY2), bro1^ΔBODΔUBD (388–844; bro1Δ::TEF1p-bro1^ΔBODΔUBD; CTY3), TEF1p-bro1^V (CTY4), and bro1^ΔBOD vps4Δ (CTY5) cells were analyzed by immunoblotting using antibodies against Bro1, Vps4, and Pgk1. Numbers below the Bro1 blot indicate expression levels normalized to BRO1 expression. (B) WT (SEY6210), bro1^ΔBOD (370–844; CTY2), bro1^ΔBODΔUBD (388–844; CTY3), bro1^ΔBOD hse1Δ (CTY30), bro1^ΔBOD vps27Δ (CTY29), bro1^ΔBOD vps37Δ (CTY21), bro1^ΔBOD mvb12Δ (CTY22), bro1^ΔBOD vps22Δ (CTY24), bro1^ΔBOD snf7Δ (CTY12), bro1^ΔBOD vps24Δ (CTY18), bro1^ΔBOD vps2Δ (CTY26), bro1^ΔBOD vta1Δ (CTY27), bro1^ΔBOD vps4Δ (CTY5), and bro1^ΔBOD doa4Δ (CTY13) cells were analyzed by live-cell fluorescence microscopy and quantified for the frequency of cells able to support NBD-PC trafficking to the vacuolar lumen. Data are represented as mean ± SEM. (C) WT (SEY6210), GOY65, and GOY65 cells transformed with BRO1 and BRO1p-bro1^ΔBOD were analyzed by live-cell fluorescence microscopy and quantified for the frequency of cells able to support NBD-PC trafficking to the vacuolar lumen. Error bars indicate SEM. Asterisk indicates statistically significant differences compared with WT (P = 0.0002). (D) Lysates generated from bro1Δ (GOY65) transformed with empty vector, BRO1, and BRO1p-bro1^ΔBOD were analyzed by immunoblotting using antibodies against Bro1 and Pgk1. Numbers below the Bro1 blot indicate expression levels normalized to BRO1 expression. (E) WT (SEY6210), bro1^ΔBOD (CTY2), and bro1^V (CTY4) were analyzed by ET and quantified to assess ILV diameter. 337 ILVs from WT (SEY6210), 831 ILVs from bro1^ΔBOD (CTY2), and 225 ILVs from bro1^V (CTY4) were quantified. Data are represented as mean ± SEM. Data are represented as mean ± SEM. (F) WT (SEY6210), bro1Δ (GOY65), or bro1^V (bro1Δ::TEF1p-bro1V; CTY4) cells were transformed with the indicated GFP-tagged cargo plasmid to assess MVB sorting using live-cell fluorescence microscopy. White dashed lines indicate cell boundaries. Scale bars = 5 µm.

Figure 4.

V domain stimulates Vps4 ATPase activity in vitro. (A) GST-Bro1 fragments and GST-bound beads were incubated with His₆-Vps4. Bound material was visualized by both EZBiolab Instant-Band protein stain and immunoblotting with penta-His antibody. (B) His₆-Vps4, His₆-MIT (WT, I18D or L64D), and Ni-NTA beads were incubated with Bro1V. Bound material was visualized by both EZBiolab Instant-Band protein stain and immunoblotting with anti-Bro1 antiserum. (C) Vps4-specific activity with titration of Bro1V (10 nM to 8 µM). Vps4 (0.5 µM) ATPase assays were conducted using the indicated conditions and resolved by thin-layer chromatography for quantitation and calculation of hydrolysis rates. Dashed lines indicate Vps4-specific activity for 0.5 µM or 1.5 µM Vps4 alone, as indicated. Bro1V alone did not exhibit measurable ATP hydrolysis. (D) Vps4 titrations were performed with or without 4 µM Bro1V. Vps4-specific activity is presented. The vertical dotted line indicates the Vps4 apparent K_m ±Bro1V. (E) V domains of S. cerevisiae Bro1, S. castellii Bro1, and H. sapiens HD-PTP (0.5–4 µM) were titrated against 0.5 µM S. cerevisiae Vps4. Specific activity of Vps4 is expressed as ADP generated per Vps4 molecule per minute. Data are represented as mean ± SEM.

Figure 5.

Bro1 V domain mutations disrupt Vps4 stimulation in vitro without disrupting binding. (A) Model of the V domain based on the S. castellii Bro1 V domain crystal structure (Protein Data Bank accession no. 4JIO, chain A) with amino acid substitution mutations impacting V domain stimulation of Vps4 indicated in red. Conserved residues that when mutated did not impact Vps4 stimulation in vitro are indicated in blue. See Table S4 for individual mutations. (B) Stimulation of Vps4 ATPase activity (0.5 µM) by 4 µM Bro1V(370–709) and Bro1V mutants represented as normalized Vps4 ATPase activity of at least three experiments done in duplicate. Error bars indicate SD, and asterisks indicate a statistically significant difference compared with WT (P < 0.05). (C) Immobilized His₆-Vps4 or Ni-NTA beads alone were incubated with Bro1V, Bro1V^M4, Bro1V^M8, Bro1V^M9, and Bro1V^M10. Bound material was visualized by immunoblotting with anti-Bro1 antiserum. (D) Vps4 (0.5 µM) ATPase activities with titration of Bro1V, Bro1V^M4, Bro1V^M8, Bro1V^M9, and Bro1V^M10 (0.25–5 µM). Vps4-specific activity is expressed as ADP generated per Vps4 molecule per minute. Data are represented as mean ± SEM.

Figure S4.

Bro1V mutants bind Vps4. This figure complements Figs. 5 and 6. (A) Immobilized His₆-Vps4 or Ni-NTA beads alone were incubated with Bro1V WT or Bro1V mutants, and bound material was analyzed by immunoblotting with anti-Bro1 antiserum. M, mutant. (B) Mutant protein expression levels normalized to WT. Bro1 expression levels were normalized to Pgk1 and subsequently normalized to WT/Pgk1 ratios. Quantified from three independent experiments performed on three different days from two sets of transformations. Data are represented as mean ± SEM. Immunoblots were probed against Bro1 and PGK1 using lysates of GOY65 transformed with empty plasmid (pRS414) or BRO1, bro1^M4, bro1^M8, bro1^M9, or bro1^M10 plasmids. Statistical analyses did not reveal differences between WT and mutant forms of Bro1. (C) bro1Δ (GOY65) cells were transformed with GFP-Cps1 and vector, BRO1, BRO1^M1, or BRO1^M7 plasmids to assess the impact of these mutations on GFP-Cps1 sorting. White dashed lines indicate cell boundaries. Scale bar = 5 µm.

Figure 6.

Bro1 V domain mutations disrupting Vps4 stimulation in vitro disrupt MVB sorting in vivo. (A)bro1Δ (GOY65) cells were transformed with empty plasmid (pRS414) or plasmids with the BRO1 promoter and BRO1, bro1^M4, bro1^M8, bro1^M9, or bro1^M10. Localizations of model MVB cargo GFP-Cps1, Ub-GFP-Cps1, or Sna3-GFP were determined using live-cell fluorescence microscopy to assess MVB sorting in these mutant contexts. White dashed lines indicate cell boundaries. Scale bars = 5 µm. (B) The percentage of cells with WT sorting signal was quantified and calculated from at least 100 cells from three independent experiments performed on three different days from three different transformations. (C) Representative immunoblots showing mutant protein expression levels, probing against Bro1 and Pgk1 as a loading control, using lysates of GOY65 transformed with empty plasmid (pRS414) or plasmids with the BRO1 promoter and BRO1, bro1^M4, bro1^M8, bro1^M9, or bro1^M10.

Figure 7.

bro1^M8 perturbs ILV formation in vivo. (A) Three-dimensional models reconstructed from 200-nm thick-section electron tomograms of bro1Δ (GOY65) with BRO1 or bro1^M8 plasmids. The limiting membrane of the MVB is labeled yellow, the ILVs are highlighted in red, and the budding intermediates are colored in green. Scale bars = 100 nm. (B) Quantification of electron tomograms of GOY65 with either BRO1 or bro1^M8 plotting ILV size, BP size, and BPs per MVB; 8 MVBs from bro1Δ (GOY65) with BRO1 plasmid containing 6 BPs and 141 ILVs and 10 MVBs from bro1Δ with bro1^M8 plasmids containing 23 BPs and 204 ILVs were quantified. Data are represented as mean ± SEM. Asterisks indicate statistically significant differences compared with WT (P < 0.006).

Figure 8.

bro1^M4 and bro1^M8 display WT steady-state membrane-associated ESCRT-III in vivo. Subcellular fractionation was performed in bro1Δ (GOY65) cells transformed with an empty vector or BRO1, bro1^M4, or bro1^M8 plasmids. vps4Δ (MBY3) cells were used as a control highlighting the distribution upon complete loss of Vps4 function. Representative immunoblots indicating fractionation of Snf7, Bro1, and Vps4 are shown. Pep12 and Pgk1 were used as membrane and soluble markers, respectively. Quantification represents six experiments; data are represented as mean ± SEM. Asterisks indicate a statistically significant difference compared with bro1Δ (P < 0.05).

Figure S5.

Bro1V mutants bind Ub. This figure complements Fig. 9. (A) Sequence alignment of V domain aa 370–388 from S. castellii and S. cerevisiae. Conserved amino acids are indicated by black circles, and isoleucine 377 and leucine 386, critical for Ub binding, are highlighted in red. (B) Bro1 V domain mutations M4, M8, and M10 (red) that disrupt V domain stimulation of Vps4 ATPase activity are spatially separated from its Ub-binding site using S. castellii Bro1V crystal structure (Protein Data Bank accession no. 4JIO, chain A). (C) Immobilized GST-fused Bro1V, Bro1V^M4, Bro1V^M8, Bro1V^M9, Bro1V^M10, Bro1V^ΔUBD (I377R) and GST alone were incubated with V5 epitope-tagged linear penta-Ub. Bound material was visualized by both Ponceau S protein stain and immunoblotting for the V5 epitope. (D) Vps4 titrations were performed with 4 µM Bro1V WT or Bro1V^ΔUBD (L386R). Vps4-specific activity is presented as mean ± SEM. (E) Lysates generated from bro1Δ (GOY65) transformed with empty vector, BRO1, TEF1p-bro1^ΔBOD, TEF1p-bro1^ΔBOD,ΔUBD (ΔUBD:I377R,L386R) and TEF1p-bro1^ΔBOD,M8 plasmids were analyzed by immunoblotting using antibodies against Bro1 and Pgk1. Numbers below the Bro1 blot indicate expression levels normalized to BRO1 expression.

Figure 9.

Ub potentiates V domain stimulation of Vps4 ATPase activity. (A) Vps4 (0.5 µM) ATPase specific activity in the presence of 1 µM S. cerevisiae Bro1V, S. cerevisiae Bro1V^ΔUBD (I377R), S. castellii Bro1V (370–708), and S. castellii Bro1V^ΔUBD (I377R) ± 50 µM mono-Ub. Data are represented as mean ± SEM. Asterisks indicate a statistically significant difference compared with Vps4 alone (P < 0.05), and number signs indicate a statistically significant difference compared with Vps4 + Bro1V − Ub (P < 0.05). (B) Bro1V titration performed in the presence of 0.5 µM Vps4 with or without 50 µM Ub. The vertical dotted line indicates the Bro1V concentration generating half-maximal stimulation within each context. (C) Vps4 titration (0.05–1.0 µM) in the presence of 4 µM Bro1V WT or Ub-binding mutant (L386R) and 50 µM Ub. Vps4-specific activity (ADP generated per Vps4 molecule per minute) is presented. Data are represented as mean ± SEM. Vertical dotted lines indicate the Vps4 apparent K_m in each context. (D) The Vps4 apparent K_m of Vps4 alone, Vps4 + 4 µM Bro1V, Vps4 + 4 µM Bro1V + 50 µM Ub, Vps4 + 4 µM Bro1V^ΔUBD (L386R), and Vps4 + 4 µM Bro1V^ΔUBD + 50 µM Ub was determined from Vps4 titration experiments in Figs. 4 D, 9 C, and S4 C. Asterisks indicate a statistically significant difference (P < 0.02). (E) Vps4 ATPase activity (0.5 µM) in the presence of 4 µM S. cerevisiae Bro1V, Bro1V^M4, Bro1V^M8, or Bro1V^M10 without or with 50 µM Ub. Vps4-specific activity with Ub addition is normalized to activity without Ub. Data are represented as mean ± SEM. Asterisk indicates a statistically significant difference with or without Ub addition (P = 0.0012). (F) NBD-PC– and FM4-64–stained WT (SEY6210), bro1Δ (GOY65), and GOY65 reexpressing Bro1 or overexpressing Bro1^ΔBOD, Bro1^ΔBOD,ΔUBD (I377R, L386R), and Bro1^ΔBOD,M8 were analyzed by live-cell fluorescence microscopy and quantified for the frequency of cells with NBD-PC in the vacuolar lumen. Data are represented as mean ± SEM. Asterisks indicate a statistically significant difference (P < 0.01), and number signs indicate a statistically significant difference compared with bro1Δ (P < 0.005).

Figure 10.

Model of Bro1 “licensing” ESCRT-III membrane remodeling.(A) Bro1 interacts with ESCRTs via interactions between PRR and early ESCRTs, as well as Bro1 domain (BOD) and ESCRT-III. Depicted are two activities of the Bro1 V domain, binding to the Vps4 MIT domain in a manner distinct from ESCRT-III MIM1 and MIM2 elements (curved black arrow), and stimulation of Vps4 ATPase activity (curved green arrow); Ub binding to the V domain potentiates Vps4 stimulation and promotes ILV formation (curved green arrow). These activities contribute to MVB biogenesis and efficient cargo sorting (straight black arrow). (B) When the BOD is deleted, Bro1^ΔBOD (via the V domain) improperly activates Vps4 and facilitates ILV formation without efficient cargo sorting. (C) Bro1 without the ability to stimulate the Vps4 (Bro1*, e.g., mutants M4, M8, M9, M10) disrupts MVB cargo sorting and perturbs ILV formation. (D) The absence of BRO1 (bro1Δ) abrogates V domain stimulation of Vps4 as well as Bro1 interactions with early ESCRTs and ESCRT-III; thus, bro1Δ exhibits severe defects in both cargo sorting and ILV formation.