CHMP7 and LEM2 exhibit complex domain architectures (Figure 1A) and in interphase are localised to the ER and the INM, respectively. We took RNA-interference (Figure 1B) and stable cell line (Figure 1C) approaches to explore the CHMP7/LEM2 axis during M-exit. In early mitosis, GFP-CHMP7 and LEM2-mCh co-localised in a hybrid ER/NE, were then transiently co-enriched at sites of ESCRT-III assembly at the reforming NE, and subsequently adopted distinct ER and NE identities as cells exited mitosis (Figure 1C, Figure 1—video 1). As expected (Gu et al., 2017), LEM2-depletion prevented the transient assembly of CHMP7 at the reforming NE (Figure 1—figure supplement 1, Figure 1—video 2). We generated cells stably expressing LEM2-mCh or LEM2^δLEM-mCh (Figure 1—figure supplement 2A) and found that the LEM-domain was necessary for efficient LEM2 and CHMP7 enrichment at the reforming nuclear envelope, and subsequent retention of LEM2 at the NE (Figure 1—figure supplement 2B and Figure 1—video 3). In cells co-expressing GFP-CHMP7 and LEM2^δLEM-mCh, both proteins instead co-assembled in the peripheral ER after nuclear envelope reformation (Figure 1—figure supplement 2C). Deletion of the CHMP7-interaction domain from LEM2 (LEM2^δ415-485-mCh, von Appen et al., 2020) also resulted in the generation of clusters of LEM2 and nuclear envelope morphology defects in the following interphase (Figure 1—figure supplement 2D). These data suggest that LEM2 must effectively gain chromatin tethers during mitotic exit to localise CHMP7 to the reforming NE, and that persistent LEM2 in the peripheral ER can seed inappropriate CHMP7 clustering during mitotic exit. In S. japonicus, Cmp7 disaggregates Lem2 clusters during interphase and mitotic exit (Pieper et al., 2020; Lee et al., 2020). Here, we found that CHMP7 depletion led to persistent enrichment of LEM2 at the reforming NE (Figure 1D, Figure 1E and Figure 1—video 4), suggesting that CHMP7 acts additionally to disassemble LEM2 during M-exit. Importantly, depletion of IST1, an ESCRT-III subunit with well-characterised roles in nuclear envelope sealing (Vietri et al., 2015), did not lead to persistent LEM2 clusters (Figure 1—figure supplement 3), which is consistent with recent findings from yeast and worms suggesting functional diversity across ESCRT-III components during NE reassembly (Penfield et al., 2020; Pieper et al., 2020). Persistent assembly of LEM2 in CHMP7-depleted cells led to the formation of NE morphology defects during cytokinesis that persisted into the subsequent interphase (Figure 1F and Figure 1G). These LEM2 clusters did not incorporate other INM proteins such as LAP1 or Emerin (Figure 1—figure supplement 4A). VPS4 was neither recruited to the reforming NE in the absence of CHMP7, nor recruited to the LEM2 clusters induced by CHMP7-depletion (Figure 1—figure supplement 4B–C), suggesting that the clusters may form as a consequence of impaired remodelling of ESCRT-III. In cells bearing persistent LEM2 clusters, we observed a breakdown in nucleocytoplasmic compartmentalisation (Figure 1H) and the localisation of DNA-damage markers to the site of these clusters, suggesting that their formation is detrimental to cellular health (Figure 1—figure supplement 5A). In S. japonicus, Lem2 clusters generated in the absence of ESCRT-III sequester heterochromatin (Pieper et al., 2020). We wondered if persistent LEM2 aggregation during mitotic exit may perturb chromatin structure in CHMP7-depleted mammalian cells. We followed endogenous Heterochromatin Protein 1 (HP1) after CHMP7 depletion and instead found diminished HP1 foci in these cells (Figure 1—figure supplement 5B–C). This defect was more pronounced in cells exhibiting LEM2-mCh clusters (Figure 1—figure supplement 5D) and whilst there were a few remaining HP1 clusters in CHMP7-depleted cells, these did not co-localise with LEM2 (Figure 1—figure supplement 5E). Thus, as well as roles in spindle disassembly (Vietri et al., 2015; von Appen et al., 2020) and membrane fusion (Vietri et al., 2015; Olmos et al., 2016; Gu et al., 2017), we suggest that CHMP7 also controls a separate process of LEM2 dissolution during mitotic exit to allow both generation of a normal nuclear envelope and proper organisation of heterochromatin during the following interphase.

Figure 1 with 9 supplements see all

Download asset Open asset

Figure 1—source data 1

CHMP7-dependent dissolution of LEM2 clusters during M-exit.

https://cdn.elifesciences.org/articles/59999/elife-59999-fig1-data1-v2.xlsx

Download elife-59999-fig1-data1-v2.xlsx

We reasoned that spatiotemporal regulation of the LEM2/CHMP7 interaction would be essential both during mitotic exit and in the subsequent interphase. CHMP7 contains predicted Nuclear Export Sequences (NESs) in Helices 5 and 6 that are thought to limit its exposure to INM proteins, including LEM2, during interphase (Vietri et al., 2020b; Thaller et al., 2019). We confirmed the presence of active NESs in helices 5 and 6 of CHMP7 (Figure 2A and Figure 2—figure supplement 1A to 1C) and demonstrated that transient perturbation of exportin function resulted in nuclear accumulation of GFP-CHMP7 (Figure 2—figure supplement 1D and E). Transient transfection of NES-compromised CHMP7 constructs led to sequestration of LEM2-mCh in nuclear and cytosolic clusters during interphase, with an additive effect observed upon disrupting both NESs (Figure 2B and Figure 2—figure supplement 1F). We were unable to generate cells constitutively expressing GFP-CHMP7^NES1-&NES2-, however, by transient transduction of cells with retroviruses driving weak GFP-CHMP7^NES1-&NES2- expression, we could demonstrate the inappropriate capture of LEM2 during interphase in GFP-CHMP7^NES1-&NES2--positive nuclear envelope and cytoplasmic clusters (Figure 2—figure supplement 1G). These data highlight the essential nature of the trans-nuclear envelope segregation of LEM2 and CHMP7, suggesting the existence of a surveillance system poised to monitor the integrity of this barrier (Vietri et al., 2020b; Thaller et al., 2019) and showing that if this segregation is compromised then LEM2 and CHMP7 can inappropriately cluster.

Figure 2 with 4 supplements see all

Download asset Open asset

Figure 2—source data 1

Nucleocytoplasmic distribution of GFP-CHMP7 or GFP-CHMP7 delta-Helix6.

https://cdn.elifesciences.org/articles/59999/elife-59999-fig2-data1-v2.xlsx

Download elife-59999-fig2-data1-v2.xlsx

The second NES in Helix6 overlaps with a predicted type-1 MIM (Schöneberg et al., 2017, Figure 2A), suggesting that this region may regulate both nucleocytoplasmic compartmentalisation and the ability of CHMP7 to engage the AAA-ATPase, VPS4. While deletion of Helix6 mimicked abrogation of Helix6’s NES by elevating CHMP7’s nucleoplasmic localisation, it also induced the appearance of CHMP7 clusters at the NE and in the peripheral ER (Figure 2C and D). We wondered if this clustering was due to impaired CHMP7 disassembly due to compromised interaction with VPS4. However, we found that CHMP7’s C-terminus was unable to bind the isolated VPS4 MIT domain, and careful analysis revealed that charge substitution at two critical acidic residues has inactivated the predicted type-1 MIM in CHMP7’s Helix6 (Figure 2A and Figure 2—figure supplement 2).

We next followed cells expressing GFP-CHMP7^δHelix6 through mitotic exit. Here, we were surprised to find that this protein failed to assemble at the NE, but instead assembled in a timely, but spatially inappropriate manner in the peripheral ER (Figure 2E, Figure 2—video 1), sequestering both downstream ESCRT-III subunits, such as IST1 (Figure 2—figure supplement 3A and B), and LEM2-mCh (Figure 2F). These assemblies persisted into the next interphase, retaining incorporation of ESCRT-III components and LEM2-mCh and preventing LEM2 from adopting its normal interphase INM localisation (Figure 2G). Given the absence of regulated nucleocytoplasmic transport during this phase of mitosis, and the absence of a functional MIM in CHMP7 (Figure 2—figure supplement 2), these data argue against canonical VPS4-binding or the CHMP7-NESs as contributing to this spatially inappropriate assembly. Removal of the C-terminal regulatory region of ESCRT-III subunits is thought to convert them into an ‘open’ conformation, facilitating their polymerisation (Shim et al., 2007). We found that GFP-CHMP7^δHelix6 could more efficiently precipitate partner ESCRT-III subunits (Figure 2—figure supplement 3C), suggesting that like the described S. cerevisiae Chm7^open (Webster et al., 2016), its autoinhibition has been relieved.

When analysing our time-lapse data, we noticed that interphase assemblies of GFP-CHMP7^δHelix6 were efficiently disassembled upon entry into the next mitosis, but reformed again during mitotic exit (Figure 3A, Figure 3B and Figure 3—video 1). Indeed, GFP-CHMP7 and GFP-CHMP7^δHelix6 were largely indistinguishable during early anaphase (Figure 2F). We wondered if there existed a ‘reset’ mechanism on the assembly status of CHMP7 during mitotic entry, to prepare it for spatiotemporally controlled polymerisation during mitotic exit. Interrogation of the Phosphosite database (Hornbeck et al., 2014) reveals that CHMP7 and LEM2 possess a number of annotated sites of post-translational modification. By capturing GFP-CHMP7 from interphase or mitotic cells, we discovered that CHMP7 was phosphorylated upon mitotic onset (Figure 3C). We mapped sites of mitotic phosphorylation to Ser3 in CHMP7’s N-terminus and to Ser441 in CHMP7’s C-terminus (Figure 3D, Figure 3—figure supplement 1). Both sites conform to consensus sequences ([K/H]-[pS]-[P] or [pS]-[P]-[X]-[R/K]) for the major mitotic kinase CDK1 and we could detect modification of these sites in immunoprecipitated samples from M-phase extracts using phospho-specific antibodies directed against these CDK1 substrate phosphorylation sites (Figure 3C).

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 3—source data 1

GFP-CHMP7 delta-Helix6 cluster disassembly during M-phase.

https://cdn.elifesciences.org/articles/59999/elife-59999-fig3-data1-v2.xlsx

Download elife-59999-fig3-data1-v2.xlsx

To understand how M-phase phosphorylation affected CHMP7’s behaviour, we turned to a previously described sedimentation assay (von Appen et al., 2020). We found that recombinant CHMP7 could be directly phosphorylated by CDK1/CCNB1 in vitro and that CDK1-phosphorylation reduced the capacity of CHMP7 to sediment (Figure 4A and B). Importantly, incorporation of phosphomimetic residues at Ser3 and Ser441 (CHMP7 S3D/S441D and CHMP7 S3E/S441E) reduced this sedimentation (Figure 4C and D). These data suggest that the M-phase phosphorylations placed by CDK1 restrict the ability of CHMP7 to interact with itself and form sedimentable clusters.

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 4—source data 1

Influence of CDK1-phosphorylation on the CHMP7/LEM2 interaction.

https://cdn.elifesciences.org/articles/59999/elife-59999-fig4-data1-v2.xlsx

Download elife-59999-fig4-data1-v2.xlsx

We found that LEM2’s C-terminus (LEM2^CT) could be co-sedimented with CHMP7 (Figure 4—figure supplement 1A and B), but its sedimentation with CHMP7 S3D/S441D and CHMP7 S3E/S441E was reduced (Figure 4—figure supplement 1C and D), suggesting that the CHMP7 assembly formed in these assays could interact with LEM2^CT. Given that CHMP7 alone may form non-specific aggregates which could pellet by sedimentation, we next used a sedimentation-independent interaction assay to test the effect of CHMP7 phosphorylation on its ability to bind LEM2^CT. Here, we discovered that whilst CHMP7 could interact well with LEM2^CT, either CDK1-phosphorylated CHMP7 or CHMP7 S3D/S441D were less able to interact with LEM2^CT (Figure 4E–H). These data suggest that CDK1 phosphorylation of CHMP7 reduces both its ability to self-associate and its ability to bind LEM2. As the LEM2 interaction is thought to activate polymerisation of CHMP7, we next examined the nature of the LEM2/CHMP7 assemblies produced by negative-stain electron microscopy. We observed that LEM2^CT induced polymerisation of CHMP7 in vitro (Figure 4I, Figure 4—figure supplement 2), producing curved filaments with periodicity matching the published dimensions and repeat units of the CHMP7 polymer observed previously (von Appen et al., 2020). Consistent with the reduced interaction between CHMP7 S3D/S441D and LEM2^CT (Figure 4G and H), these filaments were not produced when LEM2^CT was incubated with CHMP7 S3D/S441D (Figure 4I). In addition to curved filaments of CHMP7, we also observed curved lattices of CHMP7, suggesting that both linear and lateral polymeric interactions were induced (Figure 4I, Figure 4—figure supplement 2). These in vitro data support the hypothesis that CDK1-phosphorylation of CHMP7 acts to restrict its polymerisation by suppressing self-interaction and interaction with LEM2.

We next used monopolar spindle assays (Hu et al., 2008) to examine the synchronised behaviour of CHMP7 during mitotic exit. Here, we synchronised cells at prometaphase using the Eg5 inhibitor STLC (Skoufias et al., 2006) and then employed RO-3306, a highly specific CDK1 inhibitor (Vassilev et al., 2006), to force synchronous mitotic exit. We used previously described CAL-51 cells in which the CHMP7 locus had been homozygously edited to encode mNG-CHMP7 (Olmos et al., 2016). Following STLC arrest, we found that mNG-CHMP7 initiated assembly 16.32 ± 0.88 min after CDK1 inhibition and polymerised in a wave around the chromatin mass from the spindle-distal to the spindle-engaged face (Figure 5A and Figure 5—video 1). Assembly of endogenous CHMP7 persisted for 2.67 ± 0.22 min (Figure 5C), mimicking the assembly dynamics around the telophase nuclear envelope in asynchronous cells both in duration and in that CHMP7 in the peripheral ER was resistant to the assembly signal. In CHMP7-depleted HeLa cells stably expressing siRNA-resistant GFP-CHMP7 (GFP-CHMP7^R), CHMP7 assembled with similar kinetics, initiating 16.93 ± 1.21 min after CDK1 inhibition and persisting for 3.32 ± 0.13 min (Figure 5—figure supplement 1A, Figure 5C and Figure 5—video 2). Importantly, LEM2-mCh displayed similar kinetics, becoming enriched at the monopolar nuclear envelope 16.21 ± 2.23 min after CDK1 inhibition, and persisting for 4.11 ± 0.44 min. In contrast to CHMP7, LEM2-mCh localised more strongly at the spindle-engaged face of the NE (Figure 5B–C and Figure 5—video 3) before being depolymerised and maintaining NE localisation (Figure 5B). Enrichment at the spindle-engaged face is perhaps indicative of its phase-separated interaction with microtubules (von Appen et al., 2020). We next asked whether microtubules were necessary for CHMP7 assembly by using nocodazole instead of STLC to arrest cells at pro-metaphase. In this case, endogenous mNG-CHMP7 assembled and disassembled at the nuclear envelope with normal kinetics (Figure 5C, Figure 5—figure supplement 1B – 1D, Figure 5—video 4), suggesting that microtubules are dispensable for ESCRT-III assembly at this site. Interestingly, endogenous mNG-CHMP7 usually assembled synchronously around the perimeter of the chromatin mass (Figure 5A). However, in cells that were able to form a pseudo-furrow (Hu et al., 2008), CHMP7 assembled at the furrow-proximal face, before spreading to the rest of the chromatin, suggesting that there may be a furrow-directed spatial component to ESCRT-III dynamics at the reforming NE (Figure 5—figure supplement 1B, Figure 5—video 5). Consistent with experiments from cycling cells (Figure 1), in CHMP7-depleted cells stably expressing LEM2-mCh, LEM2-mCh was poorly disassembled after nuclear envelope enrichment (Figure 5—figure supplement 1E and Figure 5—video 6). These persistent LEM2-mCh assemblies did not colocalise with microtubules, suggesting that they do not form as a consequence of impaired spindle disassembly (Figure 5—figure supplement 1F and G). Biochemical analysis of whole cell lysates from these monopolar spindle assays revealed phosphorylation of total CHMP7 at both Ser3 and Ser441 persisted for 20 min following CDK1 inhibition, with dephosphorylation completing 30 min after CDK1 inhibition (Figure 5D and Figure 5E). Importantly, these data suggest that ESCRT-III-dependent nuclear envelope reformation occurs in a time period when the bulk of CHMP7 in the peripheral ER is phosphorylated, and that complete CHMP7 dephosphorylation only occurs after nuclear envelope reformation. We wondered if there were spatiotemporal differences in the dephosphorylation kinetics of CHMP7. Using biochemical fractionation, we found that only a minor pool of CHMP7 could be detected in a chromatin-associated fraction during M-exit (Figure 5—figure supplement 2A) and we discovered that CHMP7 in this pool was dephosphorylated in advance of the larger extra-chromatin ER and cytosolic pool (Figure 5—figure supplement 2B – 2C). We next raised an antibody that was able to detect pSer3 CHMP7 in mitotic cells and discovered that the pool of endogenous CHMP7 assembling at the reforming NE was not illuminated by this antibody (Figure 5—figure supplement 2D – 2I). These data suggest that phosphorylation of the bulk of CHMP7 in the peripheral ER persists during M-exit, whereas the pool of CHMP7 in contact with the reforming NE is dephosphorylated in advance.

Figure 5 with 10 supplements see all

Download asset Open asset

Figure 5—source data 1

Dynamics of CHMP7 assembly at the reforming nuclear envelope.

https://cdn.elifesciences.org/articles/59999/elife-59999-fig5-data1-v2.xlsx

Download elife-59999-fig5-data1-v2.xlsx

We next used CHMP7-depletion in cells stably expressing RNAi-resistant versions of CHMP7 (CHMP7^R) to examine the role of this phosphorylation during M-exit. We found that CHMP7 bearing non-phosphorylatable residues at positions 3 and 441 (GFP-CHMP7^R S3A/S441A) assembled at the NE with similar kinetics and intensity to GFP-CHMP7^R, but displayed limited precocious clustering on the peripheral ER during early M-phase (Figure 5F–H, Figure 5—figure supplement 3A and B). During telophase, these precocious clusters of GFP-CHMP7^R S3A/S441A grew larger (Figure 5H, Figure 5—video 7) and similarly to the clusters induced by GFP-CHMP7^δHelix6 or LEM2^δLEM, incorporated ESCRT-III components and persisted into the next interphase (Figure 5—figure supplement 3C). These data suggest that M-phase phosphorylation of CHMP7 acts to suppresses its inappropriate assembly on the peripheral ER during both M-phase and M-exit. We next examined the behaviour of CHMP7 bearing phosphomimetic residues at position 3 and 441 (GFP-CHMP7^R S3D/S441D) and were surprised to find that these mutations reduced, although did not eliminate, its assembly at the reforming NE (Figure 5F–H). Using the degree of cleavage furrow ingression at the point of maximal recruitment as a proxy for progression through M-phase, we found that the residual GFP-CHMP7^R S3D/S441D was recruited later than WT CHMP7 (Figure 5G). These data suggest that incorporation of phosphomimetic residues at Ser3 and Ser441 limits its assembly at the reforming NE.

We next looked to rescue the LEM2 clustering and nucleocytoplasmic compartmentalisation phenotypes in cells stably expressing both LEM2-mCh and GFP-NLS using siRNA-resistant HA-tagged versions of CHMP7. Consistent with a limited assembly of GFP-CHMP7^R S3D/S441D at the reforming NE, we found that like HA-CHMP7^R, HA-CHMP7^R S3D/S441D, and HA-CHMP7^R S3E/S441E could rescue both the clustering of LEM2 and the nuclear compartmentalisation breakdown induced by CHMP7 depletion (Figure 6A and B). Importantly, in CHMP7-depleted cells expressing HA-CHMP7^R S3A/S441A, cytoplasmic and NE clusters of LEM2 persisted, and the nuclear compartmentalisation defect induced by CHMP7-depletion was only partially rescued (Figure 6A–B and Figure 6—figure supplement 1A). In CHMP7-depleted cells stably expressing both LEM2-mCh and GFP-CHMP7^R S3A/S441A, LEM2-mCh was incorporated into precocious GFP-CHMP7^R S3A/S441A-positive clusters that formed during M-exit (Figure 6C, Figure 6—figure supplement 1B and Figure 6—video 1). These clusters displayed fusogenic behaviour (Figure 6—figure supplement 1C and Figure 6—video 2) and long-term culture of cells bearing LEM2-mCh and GFP-CHMP7^R S3A/S441A resulted in the loss of LEM2 from the INM with its retention in singular extra-nuclear clusters (Figure 6—figure supplement 1D and Figure 6—video 2). These data highlight CDK1-phosphorylation of CHMP7 as a mechanism to restrict polymerisation and precocious capture of LEM2 in the peripheral ER during mitotic exit. Further, they suggest that advanced dephosphorylation of CHMP7 in a chromatin associated pool licenses its polymerisation, ensuring that the nucleus reforms with, and retains, its proper complement of INM proteins for the following interphase.

Figure 6 with 3 supplements see all

Download asset Open asset

Figure 6—source data 1

CHMP7 phosphorylation prevents inappropriate LEM2 clusters forming in the peripheral ER during M-exit.

https://cdn.elifesciences.org/articles/59999/elife-59999-fig6-data1-v2.xlsx

Download elife-59999-fig6-data1-v2.xlsx

STR-profiled, mycoplasma-free vials of HeLa, 293T, GP2-293 and CAL-51, CAL-51 mNG-CHMP7^+/+ or HeLa cells stably expressing GFP-CHMP7^R have been described previously (Olmos et al., 2016) and were banked and obtained from the Crick Cell Services Science Technology Platform. Cells were cultured in DMEM containing 10% FBS, Penicillin (100 U/mL) and Streptomycin (0.1 mg/mL). Stable cells lines were generated by transduction using MLV-based retroviruses or HIV-1-based lentiviruses as described previously (Olmos et al., 2015), and selected using Puromycin (1 μg/mL) or G418 (750 μg/mL) as necessary. Where necessary, cells were sorted to monoclonality by limiting dilution or FACS.

Retroviral expression vectors containing the human CHMP7 coding sequence (both WT, siRNA-resistant and N-terminal scanning mutagenesis) have been described previously (Olmos et al., 2016). A codon optimised human LEM2 sequence was obtained by genesynthesis (GeneWIZ) and cloned EcoRI-NotI into a modified version of pMSCVneo bearing an EcoRI-NotI-XhoI polylinker leading into mCherry. Derivatives and mutations of CHMP7 and LEM2 were obtained by standard molecular biology protocols and cloned into pCMS28-GFP-CHMP7, pLHCX-HA-CHMP7^R, or pMSCVneo LEM2-mCh as required. A cDNA for VPS4A was cloned with a C-terminal LAP tag (Cheeseman and Desai, 2005) into pCMS28. A lentiviral expression vector for GFP-NLS was prepared by adding the c-MYC NLS to the C-terminus of GFP in pLVXP-GFP (a kind gift from Dr Michael Way, Crick). The VPS4 MIT domain (residues 1–122) was amplified by PCR and cloned EcoRI-NotI sites of pCAGGS-GST EcoRI-NotI-XhoI. To replace the putative MIM in CHMP7 Helix6 with a known MIM from CHMP2A, CHMP7 δHelix6 was prepared by exchanging residues 420–428 with a BamHI restriction site. Oligos encoding the CHMP2A MIM (reverse translation of Helix6: SALADADADLEERLKNLRRD) were cloned into the BamHI site. For expression of recombinant proteins, the open reading frames of full-length CHMP7 and LEM2 C-terminal domain 395–503 (LEM2^CT), or derivatives, were cloned with an N-terminal TEV-recognition site that cleaves at the 1^st Methionine as EcoRI-NotI inserts into a version of pET28a comprising N-terminal tandem His₆ affinity tag, an N-utilisation sequence A (NusA) solubilisation tag (Davis et al., 1999) and EcoRI-NotI-XhoI polylinker.

For retroviral transduction, above constructs in retroviral packaging vectors were transfected with pVSVG into GP2-293 cells (Clontech), constructs in lentiviral packaging vectors were transfected with pCMV8.91 and pVSVG into 293 T cells. Supernatants were harvested, clarified by centrifugation (200 x g, 5 min), filtered (0.45 μm) and used to infect target cells in the presence of 8 μg/mL polybrene (Millipore) at MOI < 1. Antibiotic selection was applied after 48 hr.

An antibody against GAPDH (MAB374) was from Millipore; Calnexin (ab22595) was from Abcam; Tubulin (DM1A) was from Sigma; CHMP2A (10477–1-AP) was from Proteintech; IST1 (51002–1-AP) was from Proteintech; CHMP7 (16424–1-AP) was from Proteintech; GFP (7.1/13.1) was from Roche; mCherry (ab167453) was from Abcam; HA.11 (16B12) was from (Biolegend); γH2AX (05–636) was from Sigma; 53BP1 (NB100-305) was from (Novus Biologicals). Anti-CDK1 substrate antibodies were from Cell Signaling Technology (9477S and 2325S). Anti-Histone H3 pS10 was from Cell Signaling Technology (9701S). Anti-LEM2 was from Sigma (HPA017340); anti-HSP90 (F8) was from Santa Cruz; anti-Emerin (10351–1-AP) was from Proteintech. Anti-LAP1 (21459–1-AP) was from Proteintech. Alexa conjugated secondary antibodies were from Invitrogen and HRP-conjugated secondary antibodies were from Millipore. IRDye 800 CW (925–32210) and IRDye 680 RD (925–68071) were from LI-COR Biosciences. Anti-peptide antibodies against CHMP7 pSer3 (3891 and 3892) were generated by immunisation of rabbits with KLH-conjugated peptides (MWpSPEREAEAPAGGC) by GenScript Biotech (Netherlands) B.V.

Cell lysates and fractions were denatured by boiling in denaturing LDS-sample buffer (Life Technologies) and resolved using SDS-PAGE using precast Novex gels (Life Technologies). Resolved proteins were transferred onto nitrocellulose by western blotting (wet transfer, 100V, 1 hr) and were probed with the indicated antisera in 5% milk or BSA. HRP-conjugated secondary antibodies were used to probe membranes, following by incubation with ECL Prime enhanced chemiluminescent substrate (GE Healthcare) and visualised by exposure to autoradiography film; alternatively, IRdye fluorophore conjugated secondary antibodies were used to probe membranes, following by visualisation by infrared imaging (LICOR Odyssey Fc). Where necessary, 10% polyacrylamide gels were prepared manually containing 200 μM MnCl₂ and either 25 μM PhosTag-acrylamide (whole cell lysates) or 50 μM PhosTag-acrylamide (immunoprecipitants). PhosTag-acrylamide was from Fujifilm Wako Chemicals. EDTA-free Laemmli buffer was used to prepare lysates for PhosTag-gels, prestained molecular weight markers (NEB) were supplemented with twice the volume of 10 mM MnCl₂ and 1 vol of 4x EDTA-free Laemmli buffer. After electrophoresis, PhosTag gels were soaked for 10 min in Tris-Gly transfer buffer containing 10 mM EDTA, 10 min in Tris-Gly transfer buffer containing 1 mM EDTA and 10 min in Tris-Gly transfer buffer prior to electro-transfer.

HeLa and CAL-51 cells were transfected using Lipofectamine-3000 (Life Technologies) according to the manufacturer’s instructions. 293GP2 cells were transfected using linear 25 kDa polyethylenimine (PEI, Polysciences, Inc), as described previously (Carlton and Martin-Serrano, 2007).

HeLa cells were seeded at a density of 1E5 cells/mL and were transfected with siRNA at 20 nM, 2 hr after plating using RNAi-MAX (Invitrogen), for 72 hr. The following targeting sequences that have previously been demonstrated to achieve potent and specific suppression of the targeted CHMP were employed: Control, Dharmacon Non-targeting control D-001810; CHMP7-1 (GGGAGAAGATTGTGAAGTTdTdT), CHMP7-2 (GGAGGUGUAUCGUCUGUAUdTdT); LEM2 and IST1, siGenome Dharmacon SmartPools.

Both CHMP7 and LEM2^CT were purified as previously described (Gu et al., 2017; von Appen et al., 2020) with some modifications. Escherichia coli BL21(DE3) cells expressing the 2xHis₆-NusA-TEV-tagged proteins were grown at 37°C to OD 0.6 in LB medium, transferred to 18°C and induced with 0.25 mM IPTG overnight. Cells expressing full length CHMP7, CHMP7 S3D/S441D, or CHMP7 S3E/D441E were harvested by centrifugation and lysed with TNG1000 lysis buffer (50 mM Tris-HCl pH 7.5, 1 M NaCl, 10% w/v Glycerol) supplemented with 1 tablet of cOmplete Protease Inhibitors (Roche), 20 mM imidazole, 5 mM β-Mercaptoethanol, 0.2% Triton X-100, 1 mM PMSF, 10 μg/mL DNaseI, and 1 mg/mL lysozyme. Cells were then sonicated and clarified by centrifugation at 40,000 × g for 60 min. Supernatants were incubated with Ni-NTA beads at 4°C for 2 hr. Beads were extensively washed with complete TNG1000 lysis buffer, transferred into a TNG1000 buffer with 0.01% Triton X-100, and eluted in four steps of this buffer containing stepwise increasing imidazole concentrations (50, 150, 250, and 400 mM). The eluted protein was dialysed overnight against TNT buffer (50 mM Tris-HCl pH 7.6, 150 mM NaCl, 0.5 mM TCEP) with 5% glycerol in the presence of recombinant TEV protease. The sample was further dialysed for two more times in fresh TNT buffer, before incubating with new Ni-NTA beads for re-capturing the 2xHis₆-NusA tag. The supernatant was then injected into a Superdex 200 Increase 10/300 GL column pre-equilibrated with TNT buffer, monomeric CHMP7 fractions were pulled, aliquoted and snap-frozen in liquid nitrogen. 4 L of bacterial culture yielded 1 mg of CHMP7 wildtype, 1.8 mg of CHMP7 S3D/S441D and 1.6 mg of CHMP7 S3E/S441E. BL21(DE3) cells expressing LEM2^CT or HA-LEM2^CT were processed similarly, with some modifications: lysis and washes were carried out in TNG500 lysis buffer (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 10% w/v Glycerol); no stepwise imidazole elution was performed; TEV cleavage was achieved on the Ni-bound protein overnight. Fractions containing LEM2^CT were pulled, concentrated, aliquoted and snap-frozen in liquid nitrogen. Yield was 4 mg of protein per 3 L of bacterial culture. Before subsequent experiments, proteins were thawed quickly and cleared by centrifugation at 22,000 × g for 20 min.

10 μM of recombinant CHMP7 was incubated in TNT buffer (50 mM Tris-HCl pH 7.6, 150 mM NaCl, 0.5 mM TCEP) with 10 mM MgCl₂, 1 μg of recombinant CDK1:CyclinB1 active complex (Millipore) in the presence or absence of 3 mM ATP. After 60 min at 30°C, samples were spun in a benchtop centrifuge at 22,000 × g for 40 min at 4°C. Supernatant (S) and Pelleted (P) fractions were resuspended in equal volumes of 1x Laemmli buffer, run on a PhosTag SDS-PAGE gel. Samples and gels were prepared as described above. Western-blot band intensities were quantified by densitometry using ImageJ and ratios of insoluble pelleted proteins were calculated.

10 μM of recombinant CHMP7, CHMP7 S3D/S441D, or CHMP7 S3E/S441E was incubated in TNT buffer (50 mM Tris-HCl pH 7.6, 150 mM NaCl, 0.5 mM TCEP) for 60 min at 30°C. Samples were spun in a benchtop centrifuge at 22,000 × g for 40 min at 4°C. Supernatant (S) and Pelleted (P) fractions were resuspended in equal volumes of 1x Laemmli buffer, run on an SDS-PAGE gel and analysed by western blotting with antisera raised against CHMP7. Band intensities were quantified by densitometry using ImageJ and ratios of insoluble pelleted proteins were calculated.

5 μM CHMP7, CHMP7 S3D/S441D or CHMP7 S3E/S441E was combined with 10 μM HA-LEM2^CT and diluted in a final volume of 50 μL TKD buffer (50 mM Tris HCl pH 8, 150 mM KCl, 1 mM DTT) supplemented with 1 x cOmplete EDTA free protease inhibitors and incubated at room temperature for 2 hr. Samples were spun in a benchtop centrifuge at 22,000 × g for 40 min at 4°C. Supernatant (S) and Pelleted (P) fractions were resuspended in equal volumes of 1x Laemmli buffer, run on a SDS-PAGE gel and detected by western blot using antisera raised against CHMP7 or HA-tag. Band intensities were quantified by densitometry using ImageJ and ratios of insoluble pelleted proteins were calculated.

Anti-CHMP7 (Proteintech, 16424–1-AP) was covalently coupled to Dynabeads (Life Technologies) at a final concentration of 10 mg/mL, according to the manufacturer’s instructions. Either CHMP7 or CHMP7 S3D/S441D was combined with HA-LEM2^CT at a ratio of 1:2 (μM: μM), diluted in a final volume of 100 μL TKD buffer (50 mM Tris HCl pH 8, 150 mM KCl, 1 mM DTT) supplemented with 1 x cOmplete EDTA-free protease inhibitors and incubated overnight with rotation at 4°C. 5 μg anti-CHMP7 coupled Dynabeads per reaction were washed in TKD, diluted in 100 μl TKD containing 0.1% Tween-20 and incubated with the interacting proteins for 10 min with rotation at 4°C. Immune-complexes were isolated by magnetisation and washing thrice with TKD containing 0.05% Tween-20. Inputs were collected before Dynabead-addition, denatured input and captured fractions were analysed by SDS-PAGE.

10 μM of recombinant CHMP7 was incubated in TNT buffer (50 mM Tris-HCl pH 7.6, 150 mM NaCl, 0.5 mM TCEP) with 10 mM MgCl₂, 1 μg of recombinant CDK1:CyclinB1 active complex (Millipore) in the presence or absence of 3 mM ATP. After 60 min at 30°C, the reaction was combined with 10 μM HA-LEM2^CT to a final CHMP7:LEM2^CT ratio of 1:2 (5 μM:10 μM), diluted in a final volume of 100 μL HiSalt buffer (50 mM Tris-HCl pH 8, 250 mM KCl, 1 mM DTT) and dialysed overnight on Dial-A-Lyzer Mini 3,500 MWCO cassettes (Thermo Scientific) at 4°C into LoSalt buffer (30 mM HEPES pH 8, 25 mM KCl, 1 mM DTT)¹⁵. Immune-complexes were isolated by magnetisation with anti-CHMP7 coupled Dynabeads as described above.

CHMP7 or CHMP7 S3D/S441D was combined with HA-LEM2^CT at a ratio of 1:2 (3 μM:6 μM), diluted in a final volume of 50 μL HiSalt buffer (50 mM Tris-HCl pH 8, 250 mM KCl, 1 mM DTT) and dialysed overnight on Dial-A-Lyzer Mini 3,500 MWCO cassettes (Thermo Scientific) at 4°C into LoSalt buffer (30 mM HEPES pH 8, 25 mM KCl, 1 mM DTT)¹⁵. A thin carbon film was deposited on a freshly cleaved mica sheet in a carbon evaporator (Emitech K950x, 100 mA, four pulses), then transferred onto TEM grids (400 mesh copper, TAAB). The grids were then glow discharged (Emitech K100X) at 25 mA for 30 s in air. A 4 μL sample solution was pipetted onto the glow-discharged carbon-filmed TEM grids. The solution was then blotted from the side of the grid with filter paper after 2 min, followed by staining with 1% Sodium Silicotungstate (SST) aqueous solution for 30 s. Data were collected with a CCD camera in a Tecnai Spirit TEM (FEI) operated at 120 kV.

HeLa cells were fixed in MeOH or 4% PFA and subject to processing for immunofluorescence as described previously (Carlton and Martin-Serrano, 2007). Cells were imaged using a Nikon Eclipse Ti2 microscope teamed with an Andor Dragonfly 200 Spinning Disc Confocal imaging system and paired with Andor Zyla sCMOS and Andor iXon EM-CCD cameras. Images were processed in FIJI and exported to Photoshop for assembly into figures. For rescue of LEM2 clustering (Figure 5), cells stably expressing LEM2-mCh and GFP-NLS were transfected in 8-well Ibidi chamber slides with control or CHMP7-targetting siRNA as described above. Media was changed after 6 hr. After 16 hr, cells were transfected with 100 ng pLHCX-CHMP7^R plasmids using Lipofectamine 3000. Media was changed after 6 hr. Cells were fixed 72 hr after the siRNA transfection and stained as described above. Images from three independent experiments were acquired using identical acquisition settings. To minimise effects of overexpression, only dim HA-CHMP7^R-transfected cells (area-normalised HA-fluorescence intensities below 250 AU; background of approximately 100 AU) were selected for quantification. For analysis of pSer3 CHMP7 staining, cells growing on glass coverslips were washed into transport buffer (20 mM HEPES pH 7.3, 110 mM potassium acetate, 5 mM sodium acetate, 2 mM magnesium acetate, 1 mM EGTA, 2 mM DTT and supplemented with protease and phosphatase inhibitors Adam et al., 1990) and containing 50 μg/mL digitonin (CAL-51) or 100 μg/mL digitonin (HeLa) as necessary. Cells were incubated on ice for 5 min, then were fixed in 1% PFA in transport buffer and moved to room temperature. Cells were permeabilised in transport buffer containing 0.5% Saponin and immunofluorescence proceeded as above.

Cells stably expressing the indicated proteins, or edited to express fluorescent proteins, were plated in 4- or 8-chamberslides (Ibidi). Cells were transfected with the indicated siRNA where necessary. Cells were transferred to an inverted Nikon Eclipse Ti2 microscope teamed with an Andor Dragonfly 200 Spinning Disc Confocal unit and paired with Andor Zyla sCMOS and Andor iXon EM-CCD cameras, with attached environmental chamber (Oxolabs) and imaged live using 20x dry or 60x oil-immersion objectives, typically acquiring frames every 30 s. In all cases, 405, 488, or 561 laser lines were used for illumination. To enable multifield overnight imaging of GFP-CHMP7^δHelix6, Videos were acquired using a x20 dry objective, preventing resolution of localisation to the ER or assembly in fine puncta (Figure 3A). For monopolar spindle assays, cells were incubated for 16 hr with S-trityl L-cystine (5 μM) or Nocodazole (50 ng/mL) to arrest cells in M-phase. Cells were released from M-phase through addition of the CDK1 inhibitor RO-3306 to a final concentration of 9 μM. Cells were either imaged live or were collected for biochemical experiments as described below.

For extraction of GFP-CHMP7 for PhosTag SDS-PAGE, cells were lysed on ice in RIPA buffer (150 mM NaCl, 50 mM Tris-Cl pH 7.4, 1% NP-40, 0.2% SDS, 0.5% Sodium deoxycholate) supplemented with protease (HALT, Sigma) and phosphatase (PhosSTOP, Roche) inhibitor cocktails. Clarified lysates were incubated with GFP-trap magnetic agarose beads (Chromotek) for 1 hr, washed thrice using Pulldown Wash buffer (150 mM NaCl, 50 mM Tris-Cl pH 7.4, 0.1% NP-40) after magnetic capture, and released from the resin through boiling in 2 x Laemmli buffer. For biochemical fractionation, synchronised cells at the indicated timepoints were collected on ice, transferred into fractionation buffer (20 mM HEPES pH 7.4, 10 mM KCl, 2 mM MgCl₂, 1 mM EDTA, 1 mM EGTA, 1 mM DTT) supplemented with protease and phosphatase inhibitors and left to swell on ice for 15 min. Cells were lysed by 15 passages through a 27-gauge needle and left on ice for 20 min. Nuclei and extra-nuclear fractions were collected by centrifugation (4000 x g, 10 min), lysed in EDTA-free Laemmli buffer, sonicated and examined by PhosTag SDS-PAGE.

293T cells expressing GST-tagged constructs were lysed on ice in Pulldown Lysis buffer (150 mM NaCl, 50 mM Tris-Cl pH 7.4, 1% NP-40) supplemented with protease inhibitor (HALT, Sigma) cocktail. Clarified lysates were incubated by glutathione sepharose 4β beads (GE healthcare) for 2 hr, washed thrice using Pulldown Wash buffer after centrifugal capture, and released from the resin through boiling in 2 x Laemmli buffer.

Two-tailed Student’s T-tests, or ordinary one-way ANOVA with the indicated post-hoc tests were used to assess significance between test samples and controls and were performed using GraphPad Prism. N-numbers given as the number of independent experiments, n-numbers given as the number of cells analysed.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

JCG is a Wellcome Trust Senior Research Fellow, CLS received a BBSRC LIDO PhD studentship. This work was supported by the Francis Crick Institute which receives its core funding from Cancer Research UK (FC001002, FC001143), the UK Medical Research Council (FC001002, FC001143), and the Wellcome Trust (FC001002, FC001143). We thank the Crick Structural Biology and Advanced Light Microscopy Science Technology Platforms for access to instruments. This research was funded in whole, or in part, by the Wellcome Trust (206346/Z/17/Z, FC001002, FC001143). For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.

© 2021, Gatta et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.