Spastin couples microtubule severing to membrane traffic in completion of cytokinesis and secretion

Abstract

Mutations in the gene encoding the microtubule (MT)-severing protein spastin are the most common cause of hereditary spastic paraplegia, a genetic condition in which axons of the corticospinal tracts degenerate. We show that not only does endogenous spastin colocalize with MTs, but that it is also located on the early secretory pathway, can be recruited to endosomes and is present in the cytokinetic midbody. Spastin has two main isoforms, a 68 kD full-length isoform and a 60 kD short form. These two isoforms preferentially localize to different membrane traffic pathways with 68 kD spastin being principally located at the early secretory pathway, where it regulates endoplasmic reticulum-to-Golgi traffic. Sixty kiloDalton spastin is the major form recruited to endosomes and is also present in the midbody, where its localization requires the endosomal sorting complex required for transport-III-interacting MIT domain. Loss of midbody MTs accompanies the abscission stage of cytokinesis. In cells lacking spastin, a MT disruption event that normally accompanies abscission does not occur and abscission fails. We suggest that this event represents spastin-mediated MT severing. Our results support a model in which membrane traffic and MT regulation are coupled through spastin. This model is relevant in the axon, where there also is co-ordinated MT regulation and membrane traffic.

Figure 1. Schematic diagram of spastin's domain structure and constructs used

Numbering refers to amino acid position. The regions of the protein against which the antibodies used in the study (3G11/1 and spastin86–340) were raised are shown, as is the position of lysine 388, mutated to arginine in the K388R constructs.

Figure 2. Endogenous spastin's subcellular location

A–C) Hela cells labeled with spastin (A) and calreticulin (B) show infrequent colocalized puncta (box 1) and tubules (box 2). D–F) Overexpression of VPS4-E235Q (E) results in redistribution of endogenous spastin (D) to VPS4-E235Q-positive endosomes. G–I) Spastin (G) tubulin (H) colocalization was detected in the cytoplasm of MRC5 cells. The zoomed box shows colocalization on a filament. J–L) Spastin (J) and tubulin (K) showed strong colocalization in the midbody of cells undergoing cytokinesis. In (A–C) and (G–I), arrowheads indicate structures showing co-localisation. In these and subsequent micrographs, right hand panels show the merged images; the colour of each marker in the merged image is shown by the colour of its lettering in the non-merged panels. Scale bars in these and subsequent micrographs = 10 μm. Formaldehyde fixation

Figure 3. Spastin's recruitment to membrane compartments is isoform specific

A–C) Coexpression of 60 kD myc-spastin (A) and VPS4(E235Q) (B) in Hela cells shows strong recruitment of the 60 kD myc-spastin isoform to VPS4(E235Q) endosomes. D–F) Sixty-eight kiloDalton myc-spastin (D) and VPS4(E235Q) (E) coexpressed in Hela cells show minimal colocalization (arrowhead). G–I) Hela cells were transfected with 60 KD myc-spastin (G) and VSVG-GFP (H). VSVG-GFP has just been released from the ER in a synchronized pulse and is concentrated in puncta. There is no colocalization between 60 kD myc-spastin and VSVG-GFP. J–L) In contrast, there is strong colocalization between 68 kD myc-spastin (J) and VSVG-GFP (K), just after VSVG-GFP release in a synchronized pulse from the ER of HeLa cells. Formaldehyde fixation.

Figure 4. Sixty-eight kiloDalton spastinK388R delays ER–Golgi traffic of VSVG-GFP

Hela cells were cotransfected with VSVG-GFP and 68 kD myc-spastin (A–H), 68 kD myc-spastin K388R (I–P) or 68 kD myc-spastinΔMTB-K388R (Q–T). VSVG-GFP was then released in a pulse from the ER. A–H) In cells expressing 68 kD myc-spastin (A and E), VSVG-GFP fluorescence (B and F) had left the ER and was strongly associated with the Golgi marker GM130 (C and G) at 10 min (A–D, arrow) and 20 min (E–H) post release. I–L) In contrast, at 10 min post release in cells expressing 68 KD myc-spastin K388R (I), VSVG-GFP (J) remained predominantly in the ER on myc-spastinK388R-positive MT bundles (arrowheads) and only a few VSVG-GFP-positive vesicles had left the ER. In most cells, VSVG-GFP showed minimal or no colocalization with the Golgi (K). M–P) At 20 min, although some VSVG-GFP (N) was retained on the myc-spastinK388R-positive MT bundles (arrowheads), most had emerged. However, much of the VSVG had not reached the Golgi (O), as demonstrated by the presence of green vesicles in the merged image (P). The Golgi often appeared fragmented in cells expressing 68 kD myc-spastinK388R (K and O). Q–T) Ten minutes after VSVG-GFP release in cells expressing 68 kD myc-spastinΔMTB-K388R (Q), VSVG-GFP (R) showed strong colocalization with the Golgi (S). U) Quantification of these results showed that, at each time, VSVG-GFP had reached the Golgi in a smaller percentage of cells transfected with 68 kD myc-spastinK388R (n= 3 experiments), compared with other spastin constructs tested (p < 0.002, two-way anova; n= 3 for each spastin construct). The total number of cells counted at each time is indicated above the relevant bar. Error bars = SEM. Formaldehyde fixation.

Figure 5. Spastin localizes to double-ring structures in the midbody of methanol-fixed cells

A–F) In HeLa cells fixed with methanol, endogenous spastin (A and D) colocalizes with the midbody markers Aurora B (B) and PRC1 (E). Note the double-ring appearance of spastin on either side of the stembody, most obvious in (A). G–I) Spastin (G) also colocalizes with GFP-VPS4A(E235Q) (H) in double-ring structures in the midbody, in methanol-fixed HeLa cells.

Figure 6. The MIT domain but not the MTB domain is required for recruitment of spastin to the midbody

A–C) Myc-spastinΔN194 (A) was not recruited to VPS4(E235Q) structures (B) in the cell body or midbody (box). D–F) 60 kD myc-spastinΔMTB (D) is expressed in a double-ring structure on either side of the stembody, but unlike wild-type 60 kD myc-spastin (Figure S3), does not colocalize with alpha-tubulin (E) throughout the intercellular bridge. Formaldehyde fixation.

Figure 7. Spastin depletion causes the appearance of MT-filled intercellular bridges

A–C) Hela (A and B) and MRC5 (C) cells were labeled with alpha-tubulin following spastin depletion with pooled siRNA oligonucleotides 1–4. Note long intercellular bridges (arrowheads) that were sometimes very convoluted (B). Alpha-tubulin-labeled puncta were often seen in association with these bridges (arrow in B). D and E) The intercellular bridges (arrowheads) were also seen following spastin depletion using two individual spastin siRNA oligonucleotides. Successful spastin depletion in these experiments is verified in (F). G–I) The intercellular bridges (arrowheads) typically joined two cells, as shown in DIC image (G) of YFP–tubulin (H)-expressing HeLa cells depleted of spastin. J–L) Some of the intercellular bridges had the appearance of very elongated midbodies, which labeled with midbody markers [e.g. aurora B; (K)] as well as with MT markers (J). M–O) Hela cells transfected with 60 kD myc-spastinK388R (M) and labeled for alpha-tubulin (N) also displayed similar intercellular bridges (arrowhead). Formaldehyde was used in fixed preparations except (J–L) where methanol was used.

Figure 8. Spastin is required for completion of cytokinesis

A) Time -lapse series of control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) siRNA-treated cells, showing the disruption of MTs in the intercellular bridge, on either side of the stembody (arrowheads) at approximately 2 h after anaphase. Times after anaphase shown in min. B) Time-lapse series of YFP–tubulin cells depleted of spastin. The MT disruption event does not occur. Cells remain attached by a microtubule-rich intercellular bridge (arrow). C) Cumulative timing of MT disruption event shown in (A). In control cells, the time from anaphase to MT disruption was measured. In spastin-depleted cells, where this event is delayed or does not occur, it was often not possible to measure (e.g. when it was after the end of filming). We plotted minimum timings for these cells, underestimating the effect of spastin depletion. Control cells, n= 57; spastin-depleted cells, n= 64, obtained from four independent experiments. p < 0.0001 (Student's t-test). D) Cumulative timing of anaphase onset in YFP–tubulin cells. Time from nuclear envelope breakdown (NEBD) to separation of duplicated chromosomes at anaphase was measured. Control cells, n= 21 (obtained from four experiments); spastin-depleted cells, n= 41 (obtained from three experiments). Error bars = SDs.