A Septin Double Ring Controls the Spatiotemporal Organization of the ESCRT Machinery in Cytokinetic Abscission

Abstract

Abscission is the terminal step of mitosis that physically separates two daughter cells [1, 2]. Abscission requires the endocytic sorting complex required for transport (ESCRT), a molecular machinery of multiple subcomplexes (ESCRT-I/II/III) that promotes membrane remodeling and scission [3-5]. Recruitment of ESCRT-I/II complexes to the midbody of telophase cells initiates ESCRT-III assembly into two rings, which subsequently expand into helices and spirals that narrow down to the incipient site of abscission [6-8]. ESCRT-III assembly is highly dynamic and spatiotemporally ordered, but the underlying mechanisms are poorly understood. Here, we report that, after cleavage furrow closure, septins form a membrane-bound double ring that controls the organization and function of ESCRT-III. The septin double ring demarcates the sites of ESCRT-III assembly into rings and disassembles before ESCRT-III rings expand into helices and spirals. We show that septin 9 (SEPT9) depletion, which abrogates abscission, impairs recruitment of VPS25 (ESCRT-II) and CHMP6 (ESCRT-III). Strikingly, ESCRT-III subunits (CHMP4B and CHMP2A/B) accumulate to the midbody, but they are highly disorganized, failing to form symmetric rings and to expand laterally into the cone-shaped helices and spirals of abscission. We found that SEPT9 interacts directly with the ubiquitin E2 variant (UEV) domain of ESCRT-I protein TSG101 through two N-terminal PTAP motifs, which are required for the recruitment of VPS25 and CHMP6, and the spatial organization of ESCRT-III (CHMP4B and CHMP2B) into functional rings. These results reveal that septins function in the ESCRT-I-ESCRT-II-CHMP6 pathway of ESCRT-III assembly and provide a framework for the spatiotemporal control of the ESCRT machinery of cytokinetic abscission.

Keywords: ESCRT; ESCRT-III; SEPT9; TSG101; abscission; cell division; cytokinesis; endosomal sorting complex required for transport; intercellular bridge; midbody; septins.

Figure 1.. A septin double ring flanks the midbody between cleavage furrow closure and abscission.

(A-B) Time-lapse images of MDCKs expressing mCherry-SEPT9 (A) and SEPT2-YFP (B) show septins transitioning from an hourglass-like shape (arrows) to a double bar (insets). (B) Time-lapse frames of MDCK-mCherry-SEPT9 cells expressing GFP-α-tubulin. Arrowheads point to a double septin bar, which becomes progressively dimmer. Plots show a fluorescence line scan along nine microns left and right of the ICB center. (C) Confocal images of endogenous α-tubulin and SEPT9, SEPT2, SEPT7, SEPT6. Outlined ICBs and double septin bars (arrowheads) are shown in higher magnification. (D-E) Structured illumination super-resolution images of mCherry-SEPT9 (D) and endogenous SEPT2 (E) were 3D volume rendered and rotated from various angles. A ~45 degree rotation of the ICB around the y-axis (D) reveals that the septin bar (arrowhead) flanking the midbody is a ring. See also Figures S1 and S4, and Videos S1 and S2.

Figure 2.. Assembly of the ESCRT machinery of abscission is spatiotemporally coupled to the septin double ring.

(A) SEPT9 localization with respect to endogenous TSG101, VPS25, CHMP4B and GFP-CHMP6 in the outlined ICB areas (arrows) of MDCK cells. Arrowheads are fixed in position and point to SEPT9-ESCRT overlap. (B-D) Western blot (B) of MDCK-GFP-CHMP4B cell lysate and time-lapse images of GFP-CHMP4B with SEPT9-mCherry (C) and mCherry-SEPT2 with GFP-CHMP2A (D). Plots show the mean pixel intensity of septins (magenta) and GFP-CHMP2A/CHMP4B (green) across the width of the midbody. Asterisks highlight fluorescence overlap and green arrows point to the formation of ESCRT-III cones. (E) Mean fluorescence intensity of GFP-CHMP4B and SEPT9-mCherry rings after normalization to a scale of 0 (min) to 1 (max) prior to abscission (t=0; n = 3 cells). (F-H) Quantification of the fluorescence intensity of SEPT9-mCherry (percent of maximum) in the frame preceding the emergence a GFP-CHMP4B cone (F), and timing of SEPT9-mCherry ring disassembly and GFP-CHMP4B cone formation with respect to the ICB cut (t = 0) in individual cells (G) and as an average ± SD (H; n = 12 cells). See also Figure S2 and Videos S3 and S4.

Figure 3.. Septin 9 depletion abrogates the recruitment of VPS25 and CHMP6, and impairs proper ESCRT-III assembly.

(A-B) Image (A) of SEPT9 in MDCK cells transfected with shRNAs/mCherry (insets; asterisks). Quantification (B) of mean SEPT9 fluorescence intensity per area (± SEM) in cells treated with shRNAs (n = 30-33). (C) Percentage of total cells that completed abscission under 2 h or within 2-8 h after cleavage furrow closure, and percent of cells that failed to undergo abscission or died within 10 h (n = 19). Plot shows the mean duration (± SD) of completed abscissions in control (n = 19) and SEPT9-depleted cells (n = 13). (D) TSG101 and α-tubulin in the ICBs (arrows) of shRNA-treated MDCKs. Quantification of the mean area and fluorescence intensity (± SEM) of TSG101 in the center of the ICB (n = 32-34). (E-F) Images show the localization of a-tubulin, VPS25 (E) and CHMP6-GFP (F) in the ICBs of MDCKs treated with shRNAs. Graph shows the percentage of ICBs with or with no VPS25 (n = 50) and CHMP6-GFP (n = 25-26). (G) Localization of α-tubulin and CHMP4B in ICBs (arrow) of cells transfected with shRNAs. A gallery of defective CHMP4B phenotypes is shown and arrowheads point to distended accumulations of CHMP4B. Quantification shows percent ICBs without CHMP4B (absent) and with abnormally localized CHMP4B or two rings and 1-2 cones of CHMP4B in control (n = 64) or SEPT9-depleted (n = 86) cells. See also Figure S3.

Figure 4.. Assembly of ESCRT-III requires a direct interaction between the UEV domain of TSG101 and PTAP motifs of SEPT9.

(A) Schematic of SEPT9 (isoform 1) domains and location of PTAP motifs. (B) Western blots (anti-His/-GST) of protein binding assays with GST or GST-UEV (TSG101 aa1-145) and His-SEPT9 or His-SEPT9-Δ(PTAP)_2x. Quantification shows the normalized intensity of His-SEPT9-Δ(PTAP)_2x relative to His-SEPT9 (set to 1) in four independent experiments (expt). (C-D) Western blots (anti-SEPT9) of immunoprecipitations (IPs) with control and anti-TSG101 IgGs (C). SEPT9-mCherry or SEPT9-Δ(PTAP)_2x-mCherry were immunoprecipitated with RFP_Trap_A and blotted for mCherry and TSG101 (D). Graph shows normalized intensity of TSG101 in IP of SEPT9-Δ(PTAP)_2x-mCherry relative to SEPT9-mCherry (set to 1). (E) Images show PLA signal in the outlined ICBs of cells labeled with antibodies against SEPT9, TSG101, VPS25 and CHMP4B. PLA signal (mean ± SD) quantification for TSG101 alone (n = 14), SEPT9 alone (n = 10), SEPT9 plus TSG101 (n = 45), SEPT9 plus VPS25 (n = 10), and SEPT9 plus CHMP4B (n = 32). (F) Distribution of the percentage of ICBs with and without PLA signal (SEPT9-TSG101) based on maximum diameter of their MT stalk (n = 45). (G) BiFC signal between SEPT9-VN and TSG101-VC and mean fluorescence intensity (± SEM) for MDCKs expressing VN plus VC (n = 29), SEPT9-VN plus TSG101-VC (n = 39), SEPT9-VN plus TSG101-VN (n = 29), and SEPT9-VN plus VC (n = 26). (H) Quantification of cells that completed abscission <2 h or within 2-8 h and failed to undergo abscission or died within 10 h in SEPT9-depletion rescue experiments with SEPT9-mCherry (n = 16) or SEPT9-Δ(PTAP)_2x-mCherry (n = 12). (I-M) MDCK cells were transfected with plasmids co-expressing SEPT9 shRNA and shRNA-resistant SEPT9-mCherry or SEPT9-Δ(PTAP)_2x-mCherry. Cells were stained for endogenous TSG101 (I), CHMP4B (J) and VPS25 (L), or imaged for CHMP6-GFP expression (M). Images show the localization of TSG101 and CHMP4B in the outlined ICBs (arrows). Arrowheads point to SEPT9-ESCRT overlap. Quantifications show the mean surface area and fluorescence intensity (± SEM) of TSG101 (I; n = 38 - 41), and percent of ICBs without or with ESCRTs of normal and abnormal organization in cells stained for CHMP4B (n = 30-37), VPS25 (n = 40-46) and CHMP6-GFP (n = 55-78). See also Figure S4.