The TRIM69-MST2 signaling axis regulates centrosome dynamics and chromosome segregation

Abstract Stringent control of centrosome duplication and separation is important for preventing chromosome instability. Structural and numerical alterations in centrosomes are hallmarks of neoplastic cells and contribute to tumorigenesis. We show that a Centrosome Amplification 20 (CA20) gene signature is associated with high expression of the Tripartite Motif (TRIM) family member E3 ubiquitin ligase, TRIM69. TRIM69-ablation in cancer cells leads to centrosome scattering and chromosome segregation defects. We identify Serine/threonine-protein kinase 3 (MST2) as a new direct binding partner of TRIM69. TRIM69 redistributes MST2 to the perinuclear cytoskeleton, promotes its association with Polo-like kinase 1 (PLK1) and stimulates MST2 phosphorylation at S15 (a known PLK1 phosphorylation site that is critical for centrosome disjunction). TRIM69 also promotes microtubule bundling and centrosome segregation that requires PRC1 and DYNEIN. Taken together, we identify TRIM69 as a new proximal regulator of distinct signaling pathways that regulate centrosome dynamics and promote bipolar mitosis.


INTRODUCTION
Accura te DNA replica tion and faithful / equal separa tion of replicated chromosomes into daughter cells is essential for maintaining genome stability and pre v enting cancer.The machinery that segregates the replicated chromosomes between daughter cells is termed the mitotic spindle ( 1 ).The spindle is comprised of two centrosomes (which serve as Microtubule-Organizing Centers or MTOCs), microtubules (MTs), and kinetochores.Beginning in prophase, each centrosome nucleates microtubules which either interdigitate with MTs from the other centrosome, or which attach to kinetochores ( 2 ).Some MTs emanating from the centrosomes also interact with factors at the cell cortex.The resulting MT network and associated factors collectively generate the forces that determine chromosome dynamics in mitotic cells.
Similar to DNA replication, centrosome duplication is tightly coupled / coordinated with the cell cycle and involves many of the protein kinases and ubiquitin ligases that regulate DNA synthesis.Polo-like kinase 4 (PLK4) is a master regulator of the centrosome cycle whose activity initia tes centriole duplica tion ( 3 , 4 ).PLK4 ov ere xpression alone leads to accumulation of multiple daughter centrioles adjacent to the mother centriole ( 3 , 4 ).PLK4 stability is tightly regulated by the Skp1-Cul1-F-box (SCF) complex SCF-bTrCP which restricts initiation of centriole duplication to a short time window at the G1 / S transition ( 5 , 6 ).G2 cells normally contain two centrosomes, each comprised of two centrioles that are linked by rootletin, C-NAP1 and other proteins ( 5 ).Prior to mitosis the linker connecting centrioles is dissolved in a process termed disjunction.Disjunction is mediated by a protein kinase cascade initiated by PLK1 which phosphorylates and activates MST2 / MST2.Activated MST2 phosphorylates and activates NIMA related kinase 2 (NEK2A), which in turn phosphorylates centr osomal pr otein 250 (C-NAP1), initiating untethering of the duplicated centrosomes ( 7 ).After disjunction, the centrosomes are separated and moved to opposite poles by the activity of the MT-dependent motor protein Eg5 ( 8 ).The force generated by Eg5-mediated sliding of antiparallel microtubules is sufficient to separate the centrosomes e v en when disjunction is impaired ( 7 ), demonstrating that the MST2-NEK2A and Eg5 pathways are redundant.Perturbation of disjunction and separation can se v erely compromise subsequent mitotic e v ents.For e xample, incomplete spindle pole separation leads to higher rates of kinetochore mis-attachments, spindle multipolarity, chromosome missegregation and genomic instability ( 9 ).
Structural and numerical centrosome abnormalities are hallmarks of many cancers and are likely to contribute to genetic instability and tumorigenic phenotypes.Based on centrosomal abnormalities observed in early-stage tumors it has been suggested that altered centrosome biology may facilitate tumor initiation ( 10 , 11 ).Numerical alterations in centrosomes, most commonly in the form of centrosome amplification are frequently observed in cancer.Centrosome amplification has been observed in many solid and hematological cancer types and in many cancer cell lines ( 12 ).Notably, in a study of the NCI-60 panel, up to 62% of populations of lung cancer cell lines contained > 4 centrioles.Studies in both flies and mammals show that centrosome amplification can be causally linked to aneuploidy and tumorigenesis ( 13 , 14 ).Centrosome amplification can arise via cytokinesis failure, mitotic slippage, cell-cell fusion, ov er duplication of centrioles and e xcessi v e de novo centriole assembly ( 10 , 11 ).
A 'Centrosome Amplification CA20' gene signature has been de v eloped which comprises centrosome structural genes and genes that promote centrosome amplification such as polo-like kinase 4 (PLK4) ( 15 ).In a pan-cancer analysis of 9721 tumors in the TCGA, CA20 was associated with genomic instability, alteration of specific chromosomal arms, and poor pro gnosis ( 16 ).Notabl y, CA20 was highl y associated with distinct clinical and molecular features of breast cancer.
The presence of two centrosomes at mitosis is required for a bipolar spindle, whilst e xcessi v e numbers of centrosomes lead to spindle multipolarity.Since many cancer cells harbor supernumerary centrosomes, they m ust ada pt to withstand the presence of multipolar spindles.Four main processes are no w kno wn to avert lethal spindle multipolarity due to excess centr osomes: centr osome clustering (the best characterized mechanism in cancer cells), centrosome inactiva tion, centrosome degrada tion, and centrosome loss by extrusion ( 12 ).
The first molecule described to have a role in centrosome clustering was the minus-end-directed motor dynein ( 17 ).Subsequently, a screen for proteins involved in preventing spindle multipolarity in human cancer cells identified the chromosomal passenger complex, Ndc80 microtubulekinetochore attachment complex, sister chromatid cohesion, and microtubule formation via the augmin complex as r equir ements for centrosomal clustering ( 18 ).Taken together, such studies indicate that factors controlling tension of the mitotic spindle apparatus are important for clustering of supernumerary centrosomes to form pseudo-bipolar spindles that support cytokinesis and viability ( 11 ).Howe v er, cells with pseudo-bipolar spindles fail to position all centr osomes appr opriately on the bipolar axis and are prone to chromosome segregation defects ( 19 ).Thus, centrosome clustering in cancer cells may be a survival mechanism that also fuels further genome instability and dri v es tumor evolution.
Because centrosome clustering is necessary for viability of tumors harboring supernumerary centrosomes, disruption of centrosome clustering has been proposed as a potential therapeutic strategy in cancer ( 10 , 20 ).Lethal spindle multipolarity is also the mechanism of action of the antimitotic chemotherapeutic drug paclitaxel and clustering mechanisms confer resistance to this agent (21)(22)(23).Accordingly, there has been considerable interest in defining the centrosome clustering mechanisms deployed by neoplastic cells since these processes r epr esent potential therapeutic targets.
In a preliminary analysis of TCGA data, we identified an association between the E3 ubiquitin ligase TRIM69 and centrosome amplification in cancer.Protein phosphorylation cascades and ubiquitin signaling e v ents are jointly required for coordinating cell cycle events including centrosome duplication and movements ( 5 , 24-26 ).Ther efor e, we tested a hypothetical role for TRIM69 in regulating centrosome dynamics.Our results re v eal a novel signaling pathwa y in v olving the protein kinase MST2 as a do wnstream effector of TRIM69 in a pa thway tha t regula tes centrosome disjunction.Additionally, we show that the TRIM69 promotes centrosome clustering through PRC1 and DYNEIN in cancer cells.

Generation of stable cell lines
To generate the doxy cy cline (Dox)-inducib le GFP-Plk4 U2OS cell lines stab ly e xpressing RFP-H2B, the cDNA fragment encoding GFP-Plk4 was PCR amplified from pEGFP-C3-PLK4-3xFLAG (Addgene , Cambridge , MA, USA) and subcloned into the pinducer20 plasmid, which placed it under transcriptional control of a doxy cy clineregulated promoter.High-titer lentivirus was produced in HEK293T cells, U2OS cells were infected with lentiviruscontaining medium containing 8 mg / ml polybrene (Sigma-Aldrich) in individual wells of a six-well plate.Medium was changed after 24 hours and stably-transduced cells were selected by growth in medium containing 1000 mg / ml G418 (ThermoFisher).Then the cells were infected with lentivirus expressing RFP-H2B and selected by growth in medium containing hygromycin B (ThermoFisher).To avoid clonal selection of idiosyncratic cells, pools of stably-infected cells were used for all experiments.
To r econstitute CAS9-r esistant TRIM69A in MDA-MB-231 TRIM69A −/ − cells, the PAM sequences of wildtype and E3 ligase-m utant TRIM69A cDN As (corresponding to sites targeted by TRIM69A sgRNAs in the knockout cell line) were mutated to synonymous codons.TRIM69A cDNAs harbouring silent mutations were subcloned into pINDUCER20 and the resulting lentiviral vector was packaged to generate high-titer virus.MDA-MB-231 TRIM69A −/ -cells were infected with pINDUCER-20 TRIM69A lenti viruses.Stab ly-transduced cells were selected in growth medium containing 1000 mg / ml G418 (ThermoFisher).Doxy cy cline-inducib le reconstitution of TRIM69A expression was validated by SDS-PAGE and immunoblotting.

Yeast two-hybrid assay
The yeast two-hybrid assay was carried out as described previously ( 34 ).The full length coding sequences of TRIM69A, TRIM69B, MST2 and MST1 were fused to either Nterminal GAL4 DNA activation domain (AD) in the pDEST-GADT7 vector or N-terminal GAL4 DNA binding domain (DBD) in pDEST-GBKT7.The Saccharomyces cerevisiae y east str ain Y187 tr ansformed with GAL4-DBD fusion protein was mated with the yeast strain AH109 transformed with GAL4-AD fusion protein.The pDEST-GADT7-GUS construct was used as a negati v e control.The fresh diploids on the double dropout (DDO) medium were placed on selecti v e triple dropout medium (TDO, without Leu, Trp and His) plus 1 mM 3-aminotriazole (3-AT) and quadruple dropout (QDO) medium (without Leu, Trp, His and Ade), plates were incubated at 30 • C for 3 days before been collected.

Adenovirus construction and infection
TRIM69A, TRIM69B and Control adenoviruses were constructed and purified as described previously ( 35 ).In brief, cDNAs encoding TRIM69A, TRIM69B were subcloned into the pACCMV shuttle vector.The resulting shuttle vectors were co-transfected with the pJM17 adenovirus plasmid into HEK293T cells.Recombinant adenovirus clones were isolated by plaque purification and verified by restriction analysis and Southern blotting.The empty vector Ad-Control (used to control for adenovirus infections) was deri v ed similarly but by co-transfection of the parental pACCMV shuttle vector with pJM17.Adenovirus particles were purified from 293T cell lysates by polyethylene glycol precipitation, CsCl gradient centrifugation, and gel filtration column chromato gra phy.Adenovirus preparations were quantified by A260 measurements.Cells were typically infected with 0.1 −1.0 × 10 10 pfu / ml by direct addition of purified virus to the culture medium.

RNA interference
For H1299, MDA-MB-231and U2OS cell lines, siRNAs wer e r e v erse-transfected using Lipofectamine 2000.In brief, siRNAs were incubated with Lipofectamine 2000 and serum-free OptiMEM for 15 min at room temperature in the dark.Cells were then trypsinized and resuspended in 1 ml of OptiMEM and added directly into the siRNA / OptiMEM / Lipofectamine solution to gi v e a plating density of 50%, and then they were incubated for 48 h.The siRNA sequences were listed in Supplementary Table S1.
For image acquisition, we used an Andor Dragonfly Spinning Disk Confocal Microscope (OXFORD In-struments America, Concord, MA) mounted on a Leica DMi8 microscope stand, equipped with an HC PL APO 100 ×/ 1.40 OIL CS2 Leica objecti v e.The pinhole size was set to 40 m.The camera was a Zyla Plus 4.2MP sCMOS with 2048 × 2048 pixels, with an effecti v e pixel size of 0.063 um.A piezoelectric Z stage was used to acquire Z stacks at 0.147 um intervals.Z stack size ranged between x and y um.Excitation lasers were 405 nm (for DAPI), 488 nm (for AlexaFluor 488) and 561 nm (for AlexaFluor 594).Emission filters were 445 / 46 (for DAPI), 521 / 38 (for AlexaFluor 488) and 594 / 43 (for AlexaFluor 594).Images were deconvolved in Autoquant.
H1299 cell lines stab ly e xpressing RFP-H2B or Doxinducible GFP-PLK4 U2OS cell lines with stable RFP-H2B expr ession wer e seeded on Chamber ed Coverglass from Lab-Tek II (ThermoFisher, 155382).H1299 RFP-H2B cell lines were transfected with siRNA for 48 h before exposed to paclitaxel for 24 h while U2OS cell lines induced PLK4 expression 24 h post transfection.Time-lapse microscopy was performed on a Keyence BZ-X810 using a 40 × objecti v e. Images were taken at 2 min interval for 24 h.Best focus projections of the time series were exported into AVI format.Image sequences were generated using ImageJ and manually quantified.

Image analysis
Quantification of the spindle tubulin intensity was done on a sum-intensity projection of all z-planes in which the spindle was positioned.We used the Polygon selections tool in Fiji / ImageJ (National Institutes of Health, Bethesda, MD, USA) to encompass the area of the spindle and measure the mean spindle intensity.Mean spindle intensity was background corrected by subtracting the mean intensity of the cytoplasm and normalized by dividing it by the number of z-planes in which the spindle was positioned.Spindle PRC1 intensity was quantified in the same manner.
By using the Line tool in Fiji / ImageJ, the tubulin signal intensity of a cross-section of an interphase bundle was measured by drawing a 5-pixel-thick line perpendicular to the tubulin signal.The tubulin intensity profile was corrected by subtracting the mean background from the cytoplasm.The signal intensity of the interphase bundle was calculated as the area under the peak using SciDavis (Free Software Foundation Inc, Boston, MA, USA).The same tubulin intensity profiles were used to calculate the thickness of interphase bundles, i.e. by measuring the width at the base of the tubulin signal intensity peak.Fi v e bundles per inspected cell were analyzed.Quantification of the colocalization was done using ImageJ plug-in JAcoP.

RN A e xtr action, r everse tr anscription and r eal-time PCR
RNA samples were extracted with RNeasy Mini Kit (QI-AGEN, Valencia, CA, USA).Re v erse transcription assay was performed by using the iScript cDNA Synthesis Kit (BIO-RAD , Hercules , CA, USA) according to the manufacturer's instructions.Real-time PCR was performed by using iTaq Uni v ersal SYBR Green Supermix (BIO-RAD).For quantification of gene expression, the 2 -Ct method was used.GAPDH expression was used for normalization.The sequence information for each primer used for gene expression analysis was listed in the Supplementary Table S1.

Ubiquitination Assay
HA-MST2, My c-TRIM69A and My c-TRIM69A E3 mut were transfected into 293T cells together with or without Flag-Ub.The cells were then treated with MG132 (20 M) for 8 h and lysed by RIPA buffer with Protease Inhibitor Cocktail and Phostop (Roche).Magnetic beads containing covalently conjugated antibodies against the HA tag, Myc tag or Flag tag (MBL interna tional Corpora tion) were added to the extracts, and incubations were performed for 3 h at 4 • C using rotating racks.Then the samples were examined via western blotting.In some experiments cell lysis and recovery of ubiquitinated proteins was performed under denaturing conditions.For those experiments, Flag-TRIM69A (WT or E3 mut), HA-MST2 (WT or ubiquitination site mutant) and His-Myc-ubiquitin plasmids were co-transfected into cells.After 48 h, the transfected cells wer e tr eated with MG132 (20 M) for 8 h.Cells were r ecover ed by scraping and aliquots of the harvested cells ( ∼10%) were reserved for quantification of protein expression.The remaining cells were lysed with denaturing buffer A (6 M guanidine HCl, 0.1 M Na 2 HPO 4 / NaH 2 PO 4 and 5 mM imidazole) and sonicated briefly.For each experimental condition, 2 mg of cell lysate was incubated with 50 l of TALON Metal Affinity Resin (Takara, 635501) and rotated at 4 • C overnight.The TALON beads were then sequentially washed once with buffer A, twice with buffer B (1.5 M guanidine HCl, 25 mM Na 2 HPO 4 / NaH 2 PO 4 , 20 mM Tris-Cl pH 6.8 and 10 mM imidazole) and three times with buffer T1 (25 mM Tris-Cl pH 6.8 and 15 mM imidazole).After the washes, beads were boiled in 100 l of 2 × Laemmli loading buffer containing 200 mM imidazole to release the ubiquitinated proteins.Beads were removed by centrifugation and released proteins were analyzed by SDS-PAGE and immunoblotting.

Luciferase assay
Luciferase activities were measured by using the dual luciferase reporter assay (Promega) according to the manufactur er's protocol.pRL / TK-luciferase r eporter plasmid was used as a second reporter.The data were obtained by analyzing triplicated samples.In general, 100 ng expression plasmid, 60 ng 8xGTIIC-luciferase (Addgene), and 3 ng pRL-TK (internal control) were co-transfected into H1299 cells plated in 24-well plates.48 h later, cells were harvested and luciferase activities wer e measur ed by using the dual luciferase reporter assay (Promega, Madison, WI, USA) according to the manufactur er's protocol.All r eporter assays were completed at least in triplicate, and the results were shown as average values ± standard deviations (SD) from one r epr esentati v e e xperiment.

Clonogenic survival assays
For experiments in H1299 RFP-H2B, MDA-MB-231 TRIM69A KO and U2OS, cells were seeded at a density of 2000 cells / well in triplicate in six-well plates.Cells were transfected with siRNA for 24 h before seeding to the plates.Growth medium was replenished e v ery 3 days.Colonies were stained with 0.05% crystal violet in 1 × PBS containing 1% methanol and 1% formaldehyde.The ImageJ plugin ColonyArea was used to automatically quantify stained colonies.

Mass spectrometry
Proteomics preparation after affinity purification of TRIM69 isoforms and MST2: Immunoprecipitated protein samples were subjected to on-bead trypsin digestion as previously described ( 37 ).Briefly, after the last wash buffer step during affinity purification, beads were resuspended in 50 l of 50mM ammonium bicarbonate (pH 8).On-bead digestion was performed by adding 1 g trypsin and incubating overnight at 37 • C while shaking.The next day, 0.5 g trypsin was added and incubated at 37 • C for an additional 3h.Beads were recovered by centrifugation and supernatants transferred to fresh tubes.The beads were washed twice with 100 l LC-MS grade water, and washes were added to the original supernatants.Samples were acidified by adding formic acid to final concentration of 2%.Peptides were desalted using peptide desalting spin columns (Thermo Fisher), lyophilized and stored at -80 • C until further analysis.
For phosphoproteomics sample preparation, cell lysates (400 g; n = 3) were lysed in 8M ur ea, r educed with 5mM DTT for 45 min at 37 • C and alkylated with 15mM iodoacetamide for 30 min in the dark at room temperature.Samples were digested with LysC (Wako, 1:50 w / w) for 2 h at 37 • C, then diluted to 1M urea and digested with trypsin (Promega, 1:50 w / w) overnight at 37 • C. The resulting peptide samples were acidified to 0.5% trifluoracetic acid, desalted using desalting spin columns (Thermo Fisher), and the eluates were dried via vacuum centrifugation.Peptide concentration was determined using Quantitati v e Colorimetric Peptide Assay (Thermo Fisher).
Samples were labeled with TMTpro (Thermo Fisher), for a total of two TMTpro 16plex sets.125 g of each sample was reconstituted with 50 mM HEPES pH 8.5, then individually labeled with 250 g of TMTpro reagent for 1 h at room temperature.Prior to quenching, the labeling efficiency was ev aluated b y L C-MS / MS analysis of a pooled sample consisting of 1 ul of each sample.After confirming > 98% efficiency, samples were quenched with 50% hydroxylamine to a final concentration of 0.4%.Labeled peptide samples were combined 1:1, desalted using Thermo desalting spin column, and dried via vacuum centrifugation.The dried TMT-labeled samples (six TMT sets total) were fractionated using high pH re v ersed phase HPLC ( 38 ).Briefly, the samples were offline fractionated over a 90 min run, into 96 fractions by high pH re v erse-phase HPLC (Agilent 1260) using an Agilent Zorbax 300 Extend-C18 column (3.5-m, 4.6 × 250 mm) with mobile phase A containing 4.5 mM ammonium formate (pH 10) in 2% (v ol / v ol) LC-MS grade acetonitrile, and mobile phase B containing 4.5 mM ammonium formate (pH 10) in 90% (v ol / v ol) LC-MS grade acetonitrile.The 96 resulting fractions were then conca tena ted in a non-continuous manner into 24 fractions and 5% of each were aliquoted, dried down via vacuum centrifugation and stored at -80 • C until further analysis.The remaining 95% of each fraction was further conca tena ted into 3 fractions and dried down via vacuum centrifugation.For each fraction, phosphopeptides were enriched with the High Select Fe-NTA kit (Thermo Fisher) per manufacturer's protocol.The Fe-NTA eluates were dried down via vacuum centrifugation and stored at -80 • C until further analysis.

LC / MS / MS and data analysis
For the affinity purification samples: The peptide samples were analyzed in duplicate by LC / MS / MS using an Easy nLC 1200 coupled to a QExacti v e HF mass spectrometer (Thermo Fisher).Samples were injected onto an Easy Spray PepMap C18 column (75 m id × 25 cm, 2 m particle size) (Thermo Fisher) and separated over a 2 h method.The gradient for separation consisted of 5-45% mobile phase B at a 250 nl / min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in ACN.The QExacti v e HF was operated in data-dependent mode where the 15 most intense precursors were selected for subsequent fragmentation.Resolution for the precursor scan ( m / z 300-1600) was set to 120 000, while MS / MS scans resolution was set to 15 000.The normalized collision energy was set to 27% for HCD.Peptide match was set to pr eferr ed, and pr ecursors with unknown charge or a charge state of 1 and ≥7 were excluded.
For the global phosphoproteomics samples: Two sets of 24 fractions for the proteome analysis and two sets of 3 FeNTA-enriched fractions for the phosphoproteome analysis were analyzed by LC / MS / MS using an Easy nLC 1200 coupled to an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher).Samples were injected onto an Easy Spray PepMap C18 column (75 m id × 25 cm, 2 m particle size) (Thermo Fisher) and separated over a 120 min method.The gradient for separation consisted of 5-42% mobile phase B at a 250 nl / min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN.
For the proteome fractions, the Lumos was operated in SPS-MS3 ( 39 ), with a 3s cycle time.Resolution for the precursor scan (m / z 400-1500) was set to 120000 with a AGC target set to standard and a maximum injection time of 50 ms.MS2 scans consisted of CID normalized collision energy (NCE) 32; AGC target set to standard; maximum injection time of 50 ms; isolation window of 0.7 Da.Following MS2 acquisition, MS3 spectra were collected in SPS mode (10 scans per outcome); HCD set to 55; resolution set to 50 000; scan range set to 100-500; AGC target set to 200% with a 100 ms maximum inject time.
For the phosphoproteome fractions, the Lumos was operated in MS2 ( 40 ) with a 3s cycle time.Resolution for the precursor scan ( m / z 400-1500) was set to 60 000 with a AGC target set to standard and a maximum injection time of 50 ms.For MS2 scans, HCD was set to 35; AGC target set to 200%; maximum injection time of 120 ms; isolation window of 0.7 Da; resolution set to 50 000.
Raw data files were processed using Proteome Discoverer version 2.4 (Thermo Fisher) and searched against the re vie wed human database (containing 20 203 entries), appended with a common contaminants database, using Sequest.Enzyme specificity was set to trypsin and up to two missed cleavage sites w ere allow ed.For the affinity purification samples, methionine oxidation, N-terminus acetyla tion, serine / threonine / tyrosine phosphoryla tion and lysine diglycine were set as variable modifications.For TMT proteome and phosphoproteome samples, cysteine carbamidomethylation and TMTpro were set as a fixed modification on peptide N-terminus and lysine; methionine oxidation was set as a variable modification.For phosphoproteome samples, serine / threonine / tyrosine phosphorylation were set as variable modifications.The Percolator node was used to calculated false discovery rates (FDR).A peptide FDR of 1% was used to filter all data.For PTMs, the ptmRS node was used to localize modification sites.For the affinity purification samples, the Minora node was used to extract peak areas and the 'Precursor Ions Quantifier' node was used for relati v e quantitation of peptides / proteins across samples.For the proteome and phosphoproteome samples, the 'Reporter Ions Quantifier' node was used to extract reporter ion abundances (intensities).Normalization and statistical analysis to calculate p-values and log2 fold changes were all performed in Proteome Discoverer.

Statistics and reproducibility
Statistical analysis was performed using Microsoft Excel and GraphPad Prism 6. Student's t -test was used to determine P values for all data involving comparisons between two groups.Results are expressed as the mean ± standard error of the mean (SEM) of two independent experiments.

TRIM69 expression is associated with Centrosome Amplification 20 (CA20) gene expr ession signatur e in basal breast cancers
We recently de v eloped a predicti v e classifier of breast cancer subtypes based on mRNA expression profiling of DNA repair genes ( 48 ).In our analyses of TCGA data, we unexpectedly noticed that expression of the TRIM69 mRNA in basal breast cancers was very strongly associated with gene signatures for centrosome amplification (CA20, Figure 1 A,  B), aneuploidy (Figure 1 C), and Homologous Recombination Deficiency (HRD, Figure 1 D).TRIM69 encodes an E3 ubiquitin ligase of unknown function with possible roles in tolerance of mitotic stress ( 49 ).Interestingly, expression of the Cancer Testes Antigen (CTA) MAGE-A4, which encodes a pathological cancer-specific activating binding partner of TRIM69 ( 50 ) was also strongly associated with the basal subtype ( 48 ).Ther efor e, we sought to identify roles of TRIM69 and mechanistically define its putati v e role(s) in genome maintenance and mitotic processes.

TRIM69A is dynamically associated with the centrosome during mitosis
To define mechanisms by which TRIM69 regulates centrosomes and mitosis we first determined the subcellular localization of the TRIM69 protein.The TRIM69 transcript encodes two variant proteins, TRIM69A and a smaller truncated species termed TRIM69B which lacks the N-terminal RING domain ( 49 , 51 ) ( Supplementary Figure S1A).We ectopically expressed HA-tagged versions of TRIM69A and TRIM69B in MDA-MB-231 breast cancer and in H1299 lung adenocarcinoma cells using adenoviral vectors, then determined subcellular distribution of the two TRIM69 variants using biochemical fractionation and immunoblotting.As shown in Figure 2 A, TRIM69A primarily localized to the detergent-insoluble perinuclear cytoskeleton (CSK) fraction while TRIM69B was lar gely deter gent-soluble in both MDA-MB-231 and H1299 cells.
To investigate connections between TRIM69 and centrosomes we determined more precisely the subcellular distribution of TRIM69 in cells transitioning from S-phase to mitosis.Interestingly, in S-phase, TRIM69A localized to brightly-stained fiber arrays which encircle the nucleus (Figure 2 B upper panel).On entering mitosis TRIM69A became highly concentrated at the centrosomes and was also detectable in the astral and kinetochore microtubules.In anaphase and telophase, TRIM69A co-localization with the centrosomes was progressi v ely reduced, yet TRIM69A remained associated with the microtubule networks.During telophase, TRIM69A also formed a ring around the periphery of the daughter nuclei, likely corresponding to the nascent microtubule fibers encompassing new nucleus.Unlike TRIM69A, TRIM69B was distributed broadly throughout the cell and did not localize specifically with the centrosome (Figure 2 B, lower panel).The dynamic association of TRIM69A with the centrosome is fully consistent with a fundamental role for this E3 ligase in regulating its dynamics.

MST2 and MST1 ar e no vel inter action partners of TRIM69A
To identify potential effectors of TRIM69 in regulating centrosome dynamics we performed mass-spectrometry analysis of immunopurified TRIM69 complexes and defined the TRIM69 protein-interaction network.As shown in Figure 2 C, we identified the Serine / Threonine protein Kinases MST2 and MST1 as the most abundant components of the TRIM69 complex.Other TRIM69-associated proteins we identified included Nuclear Pore proteins NUP205 and NUP133, as well as Dynein Cytoplasmic 1 Heavy Chain 1 (DYNC1H1).The MST2 / 1 kinases mediate the Hippo pathway which controls cell proliferation, mitosis and cell polarity ( 52 ).Moreover, the Hippo pathway is triggered in part by extra centrosomes ( 53 ).Ther efor e, we initially focused on MST2 / 1 as potential components of TRIM69 signaling in a pathway regulating centrosome dynamics.
We performed independent co-immunoprecipitation experiments to validate the mass spectrometry experiments and demonstrate that MST2 / 1 are members of the TRIM69 complex (Figure 2 D).Using immunofluorescence microscopy we observed co-localization of MST2 and TRIM69 (Figure 2 E).Similar to TRIM69A, HA-MST2 colocalized with centrosomes during mitosis (Figure 2 F), consistent with a potential role for TRIM69 and MST2 in regulating centrosomes.To determine the mechanism of interaction between TRIM69 and MST2 / 1, we performed yeast two-hybrid (Y2H) assays.As shown in Figure 2 G, our Y2H assays detected strong reciprocal interactions between TRIM69A and MST2 / 1 (Figure 2 G, upper and lower panels), indicating that TRIM69A interacts directly with MST2 / 1.By comparison, TRIM69B interactions with MST2 / 1 were weak and barely detectable.We conclude that MST2 and MST1 are novel TRIM69A-binding partners.
Next, we asked whether TRIM69A promotes MST2 / 1 ubiquitination.We analyzed le v els of MST2 ubiquitination in cells ov er-e xpressing wild-type TRIM69A or a catal yticall y-inacti v e TRIM69A mutant harboring C > A substitutions in amino acids 61 and 64 of the RING domain ( 54 ).As shown in Figures 3 A-B, TRIM69A was auto-ubiquitinated and also stimulated MST2 ubiquitination.Howe v er, auto-ubiquitination acti vity was reduced by ∼60% for the catal yticall y-inacti v e TRIM69A mutant when compared with WT TRIM69A.MST2-induced ubiquitination activity was also reduced (by ∼70%) in cells expressing catal yticall y-inacti v e TRIM69A when compared with cells expr essing WT TRIM69A (Figur e 3 A, B).We conclude that MST2 is likely to be a TRIM69 substrate.Howe v er, we hav e not observed any effect of TRIM69 on MST2 le v els or protein stability (Figure 3 C ). Ther efor e, TRIM69-dependent MST2 ubiquitination is most likely unrelated to proteasomal degradation.
To determine the type of TRIM69-induced polyubiquitin chain linkage assembled on MST2, we co-expressed MST2 with a panel of ubiquitin m utants, for w hich each has only one of the se v en possib le lysines availab le for polymer chain assembly ( 55 ).As shown in Supplementary Figure S1B, TRIM69A promoted both K6-and K29-, but not K48-linked ubiquitination of MST2.This result suggests tha t TRIM69A-media ted ubiquitina tion does not promote proteasome-media ted degrada tion of MST2.
Our proteomic analysis of the MST2 complex from cells ectopically expr essing TRIM69A r evealed a di-Gly ubiquitin remnant at K279 of MST2.Ther efor e, we tested a K279 > A MST2 mutant for TRIM69A-dependent ubiquitination.Our results showed that this bona-fide ubiquitination site at K279 is not absolutely r equir ed for TRIM69Adependent ubiquitination of MST2.The caveat of this experiment is that while many E3 ligases have a pr eferr ed ubiquitination site on their substrate proteins, mutating that pr eferr ed site will often result in ubiquitin conjuga tion a t alternati v e lysine residues.Often, it is necessary to remove many, or e v en all lysine residues on a target protein in or der to abrogate ubiquitination.Ther efor e, based on our results with the K279A mutant we can not exclude the possibility that K279 is a TRIM69A target site.There are 6 other lysine residues in the vicinity ( ∼40 AAs) of K279 and it is possible that those residues are targeted for ubiquitination when K279 is unavailable.
We noticed that ectopically-expressed TRIM69A promoted redistribution of HA-MST2 to a chromatin-and cytoskeleton-enriched detergent-insoluble CSK fraction (see 'Input' panels of Figure 3 D).We also detected robust association of TRIM69A and MST2 in the detergentinsoluble CSK fraction.Interestingly, the RING finger TRIM69A mutant (TRIM69A E3 mut) failed to redistribute MST2 to the CSK-insoluble fraction.The truncated TRIM69B variant (which lacks the RING domain) was also unable to redistribute MST2 to the CSK compartment (Figure 3 D).In immunofluorescence experiments, HA-MST2 was distributed broadly in the cell in the absence of TRIM69A (Figure 3 E).Howe v er, HA-MST2 was redistributed to filamentous structures encircling the  nucleus when co-expressed with WT TRIM69A (but not the TRIM69A RING finger mutant, see Figure 3 E).Consistent with a role for TRIM69 in regulating MST2 subcellular localization, depleting endogenous TRIM69 using siRNA decreased the amount of CSK-associated MST2 (Figure 3 F).Taken together, the results of Figure 2 identify MST2 as a novel TRIM69A binding partner and substrate.Moreover, the RING domain of TRIM69A is critical for both associating with and regulating the subcellular localization of MST2.

TRIM69 / STK interactions do not affect the Hippo pathway
MST2 and MST1 ar e cor e protein kinases of the Hippo pathway, a conserved signal transduction cascade that contr ols transcriptional pr ograms involved in diverse processes including cell proliferation, survival ( 56 ), and the centrosome pathway ( 57 ).Ther efor e, we tested a role for TRIM69-STK signaling in regulating the Hippo signaling cascade and its transcriptional endpoints.As shown in Supplementary Figure S2 (panels A and B), neither depletion nor overexpression of TRIM69A affected levels of phosphoproteins (such as p-MOB1, P-YAP) that critically regulate the Hippo pa thway.Knockdown ef ficiency of TRIM69A in these experiments is shown in Supplementary Figure S2C.Simi-larly, TRIM69 ablation or over expr ession did not affect Hippo pathway transcriptional endpoints.Staurosporineinduced r epr ession of Hippo-r esponsi v e genes such as CYR61, CTGF, and ANKRD1 (Supplementary Figure S2D) was unaffected by TRIM69 status (Supplementary Figure S2E).As e xpected, ectopically-e xpressed MST2 did r epr ess luciferase r eporter acti vity dri v en by a Hippo pathway-responsi v e 8xGTIIC promoter (Supplementary Figure S2F).Howe v er, Hippo pathway reporter gene activity was unaffected by TRIM69 (Supplementary Figure S2G).Ther efor e, TRIM69-MST2 / 1 interaction does not regulate centrosome behaviour and mitotic progression via the Hippo pathway.

TRIM69A stimulates phosphorylation of MST2 by PLK1 and promotes centrosome disjunction
To elucidate pathways regulated by TRIM69-MST2 signaling, we defined the MST2 interactome in the presence and absence of co-expressed TRIM69A.We identified se v eral high-confidence TRIM69A-inducib le MST2 interactors in the detergent-insoluble CSK fraction including: MST1 (a known MST2 heterodimerization partner); zinc ribbon domain containing 2 (a centromeric protein); USP10 (a de-ubiquitinating enzyme); NDE1 (a centroso-mal protein which regula tes d ynein function and microtubule organization); and TNPO2 (which mediates docking of the importin / substrate complex to the nuclear pore complex); and PLK1 (a proximal kinase involved in centrosome disjunction).Importantly, MST2-binding of these proteins was induced specifically by WT TRIM69A and not by the catal yticall y-inacti v e TRIM69A mutant (Figure 4 A).
In addition to its canonical role in the Hippo pathway, MST2 mediates PLK1-induced centrosome separation by promoting recruitment of NEK2A to the linked centrioles.NEK2A subsequently phosphorylates C-NAP1 and rootletin to stimulate centrosome disjunction ( 7 ).Ther efor e, we tested a role for TRIM69 in regulating PLK1-MST2 signalling and the centrosome cycle.
As shown in Figure 4 C, WT TRIM69A (but not the E3 ligase mutant) promoted complex formation between MST2 and PLK1 in the detergent-insoluble CSK compartment.Consistent with our co-IP results, we also detected TRIM69A-inducible co-localization of HA-MST2 with PLK1 (Figure 4 B).PLK1-mediated phosphorylation of MST2 at S15, S18 and S316 is an important e v ent in centrosome disjunction ( 7 ).Analysis of MST2 phosphopeptides from the LC-MS / MS e xperiment re v ealed increased phosphorylation of S15, S316, T336, T384, S385 and se v eral other residues in the detergent-insolub le fraction of TRIM69A-ov ere xpressing cells (Figure 4 D).We further validated TRIM69-dependent phosphorylation of MST2 using Phos-tag phosphate affinity gel electrophoresis ( 58 ).As shown in Figure 4 E, Phos-tag gel electrophoresis re v ealed a TRIM69A-inducib le phosphorylated MST2 species (labelled 'P-MST2' in Figure 4 E).TRIM69Ainduced phosphorylation of MST2 was also detectable using a phospho-specific antibody against pS316 (Supplementary Figure S3C).To determine whether there were preferred TRIM69-inducible MST2 phosphorylation sites, we constructed a series of MST2 mutants containing individual or combinatorial S15A / S316A / T336A / T384A / S385A substitutions.TRIM69A-induced MST2 phosphorylation was not significantly affected f or an y of the MST2 mutants harbouring individual alanine substitutions in S15, S316, T336, T384 or S385.Howe v er, mutating all fiv e residues completely abolished TRIM69A-induced MST2 phosphorylation (Supplementary Figure S3A,B).We conclude that TRIM69A promotes multi-site phosphorylation of MST2.During the centrosome cycle, MST2 phosphorylates the downstream protein kinase NEK2A to promote centrosome disjunction ( 7 ).Our Phos-tag gel electrophoresis experiments also re v ealed that TRIM69A promotes NEK2A phosphorylation by MST2 (Figure 4 F), further consistent with a role for TRIM69 in activating the MST2-NEK2A signaling cascade.
Next we performed co-immunoprecipitations to determine how defecti v e MST2 phosphorylation impacts association with its binding partners.Remar kab ly, the TRIM69Ainduced association of HA-MST2 with PLK1 was abrogated in the MST2 S15 > A mutant, but not the other phosphorylation-resistant MST2 variants (Figure 4 H).We used IF microscopy to quantify co-localization of PLK with WT and phosphorylation site mutant forms of MST2. Figure 4 G shows tha t co-localiza tion of MST2 S15 > A with PLK1 was specifically reduced when compared with MST2 WT.These results are suggesti v e of a ternary complex involving TRIM69, PLK, and MST2 that is both positi v ely and negati v ely regulated by different MST2 phosphorylation e v ents.
To directly test the role of TRIM69A in regulating centrosome separation we determined the effect of TRIM69A ov ere xpression or ablation on inter-centriolar distance.As r eported pr e viously ( 7 ), MST2 or NEK2A ov ere xpression reduced the centrosomal staining of C-NAP1 (Figure 5 A-D) and reduced the number of cells with low inter-centriolar distance (Figure 5 C).Interestingly, TRIM69A ov ere xpression fully phenocopied the effects of ov ere xpressed MST2 or NEK2A (our positi v e controls for stimulation of centrosome disjunction) (Figure 5 A-D) Conversely, TRIM69 ablation led to an increase in the number of unseparated centrioles and increased C-NAP1 intensity relati v e to control cells (Figure 5 E-G).W ha t's more, MST2 S15A mutant which abolished the interaction with PLK1 partially inactivates MST2 in promoting centrosome disjunction (Figure 5 A-D).Taken together, the results of Figure 4 and Figure 5 suggest that TRIM69A promotes PLK1-mediated phosphorylation of MST2 to facilitate the disjunction phase of the centrosome cycle.

TRIM69A-deficiency induces centrosome scattering and proliferation defects
Gi v en the high expression of TRIM69 in CA20-high cancers (Figure 1 ), we considered the possibility that TRIM69 and MST2 may also have roles in promoting centrosome clustering and averting mitotic defects that can arise from multipolar spindles ( 10 , 20 , 59 ).For these experiments we initially chose to work with the MDA-MB-231 breast cancer cell line which harbors amplified centrosomes ( 60 ).We used gene editing to generate TRIM69 −/ − clonal derivati v es of MDA-MB-231 cells (Supplementary Figure S4).When compared with the parental TRIM69 + / + MDA-MB-231 cells, TRIM69 −/ − cells showed decreased clonogenic survival (Figure 6 A).We quantified centrosomes in TRIM69 + / + and TRIM69 −/ − cells that were treated with 5 nM taxol, a first-line chemotherapeutic agent which stabilizes microtubules and causes spindle-multipolarity ( 22 ).As shown in Figure 6 B, two independent clones of TRIM69 −/ − cells showed increases in spindle multipolarity and centrosome scattering when compared with parental MDA-MB-231 cells.In a complementary approach to determine the effect of TRIM69-deficiency on centrosome dynamics, we ablated HAUS1, a factor r equir ed for centrosome integrity ( 61 ) in TRIM69 + / + and TRIM69 −/ − cells and quantified scattered centrosomes.As expected, HAUS1-depletion led to an ∼2-fold increase in centrosome scattering in two independent TRIM69 −/ − clones when compared with parental TRIM69 wild-type MDA-MB-231 cells (Figure 6 C).We used Doxy cy cline-regulated promotors to conditionally r econstitute TRIM69A expr ession in TRIM69 −/ − cells (Figure 6 D).Expression of WT TRIM69A (but not of catal yticall y-inacti v e TRIM69A E3 mut) rescued the centrosome scattering in TRIM69 −/ − cells (Figure 6 E).Ectopically-expressed WT TRIM69A also promoted taxolresistant cell proliferation (Figure 6 F).The results of Figure 6 suggest that TRIM69 allows cancer cells to tolerate

TRIM69 regulates centrosome dynamics and mitotic progression
TRIM69 ablation in H1299 lung cancer cells (which harbor supernumerary centrosomes) also led to reduced clonogenic survival (Supplementary Figure S5A).Similar to MDA-MB-231 cells, treatment of H1299 cells with HAUS1 siRNA led to spindle m ultipolarity, w hich was partially rescued by ectopic ov er-e xpression of MYC-TRIM69A (Supplementary Figure S5B).Human cells with e xcessi v e centrosomes and multipolar spindles experience prolonged mitosis ( 62 ).To determine whether TRIM69 / MST2 facilitate mitosis, we used time-lapse microscopy to study cell cycle progression of control (siCon), TRIM69-depleted, or MST2 / MST1depleted H1299 cells in the presence of paclitaxel.As shown in Supplementary Figure S5C-E, siRNA-mediated ablation of MST2 / MST1 or TRIM69A led to prolonged mitosis.The mean times between nuclear envelope breakdown (NEB) and anaphase in control, TRIM69A-, and MST2 / 1ablated cells were 53, 86 and 75 min, respecti v ely (Supplementary Figure S5D, E).TRIM69A or MST2 / 1-depletion also led to increases in paclitaxel-dependent micronucleation when compared to the control cultures (Supplementary Figure S5F).Finally, depleting TRIM69 or MST2 and MST1 led to enhanced centrosome scattering and multipolar spindles in the presence of paclitaxel, fully recapitula ting the ef fects of TRIM69A abla tion (Supplementary Figure S5G).Knockdown efficiency of TRIM69A and MST2 / 1 is shown in Supplementary Figure S5H.
In a complementary approach to test the role of TRIM69-MST2 signaling in tolerating e xcessi v e centrosomes and spindle multipolarity, we determined the effects of TRIM69 and MST2 / 1 depletion on mitotic progression of PLK4-ov ere xpressing cells.PLK4 regulates centriole replication and causes centrosome amplification when ov ere xpressed ( 3 , 63 ).Moreov er, PLK4 is one of the 20 centrosome amplification (CA) signature genes whose expres-sion is positi v ely related with TRIM69A (Figure 1 ).We generated human U2OS cells which induce expression of PLK4 from a doxy cy cline (Dox)-responsi v e promoter.As expected, Dox-inducible PLK4 expression led to centrosome amplification (evident from pericentrin staining) and a decrease in clonogenic survival (Figure 7 A, B).Interestingly, TRIM69-depletion caused increased lethality in PLK4ov ere xpressing cells (+ Dox) when compared with control (-Dox) cultures (Figure 7 A, B).We also performed li v e-cell imaging to study the effects of TRIM69A and MST2 / 1 depletion on mitotic progression in the presence and absence of ov ere xpressed PLK4.In uninduced U2OS cells (which rarely have amplified centrosomes), siRNA-mediated ablation of TRIM69A or MST2 / 1 had no effect on mitotic timing, spindle multi-polarity, micro-nuclei, multinucleation, or lagging chromosomes (Figure 7 C-G).Interestingly howe v er, in PLK4-ov ere xpressing cells (+ Dox), depletion of TRIM69 or MST2 + MST1 led to increased numbers of multi-polar spindles and other mitotic defects including lagging chromosomes, anaphase bridges and mi-cronuclei (Figure 7 C-G).Conversel y, ectopicall y-expressed TRIM69A significantly corrected the spindle multipolarity and other mitotic defects caused by PLK4 induction.TRIM69A ov ere xpression also promoted cell proliferation in cells experiencing PLK4-induced mitotic stress (Supplementary Figure S6).Taken together the results of Figure 7 and Supplementary Figures S5 and S6 show that MST2 / 1ablation phenocopies the mitotic defects resulting from TRIM69A-deficiency.We conclude that the TRIM69-MST2 / 1 signaling axis resolves multipolar spindles and pre v ents lethal mitoses.

TRIM69A promotes MT bundling and regulates centrosome clustering through PRC1 and Dynein
Our proteomics analyses identified Dynein as a component of the TRIM69 complex (Figure 2 C).Ther efor e, we considered MTs, MT-based motors and MT-bundling proteins such as Dynein and PRC1 (which are implicated in clustering of supernumerary centrosomes) ( 17 , 18 , 64 ) as candida te media tors of TRIM69A-dependent centrosome clustering.As shown in Figure 8 A, B, TRIM69 WT (but not catal yticall y-dead TRIM69A) induced the formation of MT bundles which contained increased le v els of acetylated tubulin, a marker of MT stability ( 65 , 66 ).Conversely, TRIM69A knockdown led to decreased tubulin spindle intensity in metaphase cells (Figure 8 C).
Interestingly, TRIM69A co-localized with PRC1 (a MTbinding and bundling protein r equir ed for mitotic progression ( 67)) in both interphase and mitotic cells (Figure 8 D).Moreover, TRIM69A promoted localization of PRC1 to bundled MTs (Figure 8 E) suggesting that TRIM69A plays a pr oximal r ole in PRC1-mediated MT bundling and centrosome clustering.Ov er-e xpressed PRC1 fully recapitulated the TRIM69A-induced MT bundling phenotype (Figure 8 F), further consistent with participation of TRIM69 and PRC1 in a common pa thway tha t regula tes MT reorganiza tion.TRIM69A also promoted localization of the cytoskeletal motor Dynein to bundled MTs (Figure 8 G).To test whether bundled MTs are necessary for TRIM69A-dependent centrosome movements, we measured centrosome clustering in cells depleted of PRC1 and Dynein respecti v ely.As shown in Figure 8 H, depleting PRC1 or Dynein completely abrogated TRIM69induced centrosome clustering.Quantification of tubulinstaining re v ealed that depleting PRC1 or Dynein led to decreased MT bundle width and cross section intensity in interphase (Figure 8 I-K) and reduced total tubulin spindle intensity in metaphase (Figure 8 L, M).Knockdown efficiencies for PRC1 and DYNEIN are shown in Figure 8 N.Although TRIM69 re-localized MST2 to bundled microtubules (Supplementary Figure S7D), MST2 was dispensable for TRIM69A-induced centrosome clustering (Figure 8 I-M).
Our mass spectrometry analyses identified nuclear pore proteins NUP205 and NUP133 as components of the TRIM69 complex (Figure 2 ).Centrosome separation and movement along the nuclear envelope is critically dependent on molecular linkages between nuclear pores and the MT network.Ther efor e, we determined the effect of TRIM69 on the distribution of tubulin, NUPs, and other potential mediators of centrosome clustering.As shown in Supplementary Figure S7, TRIM69A WT induced colocalization of MST2 with DYNLL1 (Supplementary Figure S7A), CENPF (Supplementary Figure S7B), nuclear pore complex proteins, as evidenced by co-staining with Mab414 (Supplementary Figure S7C) and bundled microtubules (Supplementary Figure S7D).In co-IP experiments, we also detected TRIM69-induced interaction of MST2 with NUPs and MT motors (Supplementary Figure S7E).We conclude that TRIM69A functions upstream of PRC1 in a pathway that leads to MT bundling, reorganizes MTmotor-nucleoporin networks and allows dynein-dependent centrosome clustering.

DISCUSSION
Here, we identify the protein kinases MST2 and MST1 as new binding partners and effectors of the E3 ubiquitin ligase TRIM69.We demonstrate that TRIM69 stimula tes forma tion of an MST2-PLK1 complex and promotes phosphorylation of MST2 at S15, a known PLK1 site ( 7 ).PLK1-mediated MST2 phosphorylation at S15 is necessary for subsequent phosphorylation of NEK2A to dissociate c-NAP1 from daughter centrioles ( 7 ).Thus, we provide a new molecular mechanism by which TRIM69 promotes MST2-and PLK1-mediated centrosome disjunction.Importantly, the TRIM69-mediated linker dissolution mechanism defined here is distinct from the growth factor receptor / GRK2-media ted pa thway of centrosome disjunction which does not involve PLK1 ( 68 ).
A limitation of this work is that w e w ere unable to interrogate the endogenous TRIM69A protein.In numerous experiments with appropriate positi v e and negati v e controls, none of the available commercial antibodies (or antibodies that we generated in-house) detected endogenous TRIM69A or TRIM69B.Our results suggest that TRIM69A is present in cells at very low levels and / or that the protein is not very imm uno genic.Ther efor e, out of necessity we have studied tagged forms of TRIM69 in our experiments.Historically, studies with ectopically-expressed proteins have been critical for generating mechanisms of action and paradigms for E3 ligase signaling and genome maintenance.Some landmark studies in the centrosome field have relied primarily on ectopically-expressed proteins to define biochemical interactions and signaling mechanisms that mediate centrosome disjunction ( 7 , 69 ).Ne v ertheless, in the future when better reagents are available, it will be necessary to validate roles of endogenous TRIM69 and other factors in regulating centrosome biology.
In addition to pr omoting centr osome disjunction, we show that TRIM69 stimulates centrosome clustering, both after taxol treatment (which promotes multi-polar mitoses) and following PLK4 ov ere xpression (which promotes centrosome replication and amplification).There is no evidence that the canonical PLK1 / MST2-mediated centrosome disjunction pathway can explain centrosome clustering activity of TRIM69-MST2.Ther efor e, we must consider possible mechanisms whereby TRIM69-MST2 might promote centrosome clustering.
An elegant genetic screen for mediators of centrosome clustering in Drosophila cells yielded three categories of genes: (1) participants in the Spindle Assembly Checkpoint (SAC); (11) genes encoding MT-associated proteins and motors with roles in spindle focusing and (111) genes involved in cell adhesion-based centrosome movement ( 64 , 70 ).While we have not examined connections between TRIM69 and the SAC or cell adhesion, TRIM69A certainly has hallmarks of centrosome-clustering genes in ca tegory #2, i.e.MT-associa ted proteins and motors.For example, we show that TRIM69 induces robust MT bundling, and also forms complexes with MT motors (DYNEIN), MT-associated proteins (PRC1) and nucleoporins that tether centrosomes to the nucleus and regulate centrosome movements.
Clustering of supernumerary centrosomes is often associated with MT stabilization and involves the same media tors tha t bundle MTs into bipolar spindle arrays in normal cells ( 71 ).Forces responsible for centrosome clustering can be generated by MT-bundling complexes that reside where anti-parallel MTs overlap ( 59 ).For example, the MT-bundling protein PRC1, crosslinks microtubules into antipar allel arr ays and cooper ates with motor proteins to control the dynamics and size of bundled regions ( 72 ).Thus PRC1 is necessary for central spindle formation ( 73 ) and kinetochore tension in metaphase ( 74 , 75 ).Moreover, PRC1 facilitates clustering of supernumerary centrosomes in metaphase, which pre v ents spindle multipolarity ( 18 ).
In addition to MT bundling and stabilization, MT motors and their connections with the nuclear envelope are important for centrosome clustering and separation.In normal cells, dynein stabilizes interphase microtubule arrays and determines centrosome position ( 76 ).Dynein media tes a ttachment and migra tion of centrosomes along the nuclear envelope during interphase / prophase and facilita tes a ttachment of centrosomes to spindle poles in mitosis ( 77 ).In cells with supernumerary centrosomes, dynein helps coalesce the excess centrosomes into pseudo-bipolar spindles.Interference with d ynein localiza tion promotes centrosome scattering and multipolarity ( 17 ).During G2 and prophase, MTs emanating from centrosomes connect to the nuclear envelope via two independent and coopera ting d ynein-media ted mechanisms: (i) nuclear pore proteins RanBP2 and NUP358 at the cytoplasmic side of the nuclear membrane recruit BICD2 which tethers d ynein / d ynactin to NPCs.(ii) The nucleoporin NUP133 interacts with CENPF in G2 / M. In turn, CENPF recruits NudE / NuDEL which interact with dynein ( 78 ).These d ynein-media ted connections between the nuclear pores and the MT networks help drive centrosome separation and critically maintain centrosome association with the nuclear envelope ( 78 ).Depletion of nucleoporins leads to supernumerary centrosomes and multipolar spindles ( 79 ) further consistent with a role for nuclear pore complexes in centrosome clustering.Taken together, our results suggest that PRC1 and DYNEIN, together with TRIM69 play important roles in facilitating formation of stable bundled MT networks and regulating their associated proteins, including NUPs, to promote centrosome clustering and centrosome separation.
Here, we propose that TRIM69A promotes association of PRC1 with antiparallel MT bundles, including those that connect the extra centrosome (s) with the bipolar spindle.Stability of these overlap regions supports extra centrosome coalescence with the centrosomes from the bipolar spindle by zippering up these overlaps with the spindle.Likewise, TRIM69A promotes Dynein association with microtubules and this likely accounts for the minus end directed pulling of an extra centrosome towards a centrosome of the bipolar spindle.Our finding that PRC1 and Dynein depletions have similar effects on spindle multipolarity suggest that these two microtubule-associated proteins act cooperati v ely in this process.
In our study, a catal yticall y-dead TRIM69A m utant was also inacti v e for MST2-relocalization, centrosome disjunction, and centrosome clustering activities.Further studies are necessary to identify the putati v e TRIM69A substrate(s) whose ubiquitination is necessary for regulating and interacting MST2 and centrosome dynamics.To address this issue, we have performed an unbiased screen for TRIM69A-induced ubiquitination e v ents.As shown in Supplementary Figure S5F, we identified se v eral proteins (including nucleoporins) that were ubiquitinated in a TRIM69-inducible manner and which r epr esent potential mediators of centrosome disjunction or clustering.We are also considering the alternati v e hypothesis that TRIM69A media tes MST2 localiza tion and regula tion of centrosome function in a manner that is separable from its E3 ubiquitin ligase activity.The E3 ligase-inacti v e TRIM69A mutant used in this study harbors substitutions in the RING domain.In addition to abrogating E3 ligase activity, RING domain mutations could also pre v ent pr otein-pr otein interactions.Some RING finger-containing E3 ubiquitin ligases have important biological activities that are independent of their catalytic activities.For example the E3 ubiquitin ligase RAD18 mono-ubiquitinates PCNA to promote Trans-Lesion Synthesis at stalled DNA replication forks.Howe v er, RAD18 also binds directly to DN A pol ymerase Eta (Pol ), and acts as a molecular chaperone that deposits Pol a t DNA replica tion forks independent of its PCNA ubiquitination activity ( 80 , 81 ).RAD18 also acts as a molecular chaperone for the HR factor RAD51B ( 82 ) independently of its catalytic activity.Based on this precedent, TRIM69A functions in centrosome disjunction and clustering could be mediated solely through chaperoning and re-localization of MST2.
TRIM69 belongs to a family of SUMO-Targeted Ubiquitin Ligases (STUBLs) which associate with SUMOylated target proteins ( 83 , 84 ).Ther efor e, it will also be important to identify the putati v e SUMOylated proteins that bind and recruit TRIM69A.Appropriate SUMOylation / de-SUMOylation is important for centromere-MT attachment and hundreds of key mitotic factors are SUMOylated ( 84 ) and r epr esent potential r eceptors for r ecruiting TRIM69A.Moreover, our proteomics experiments identified nucleoporins as major binding partners of TRIM69 and MST2.The nucleoporin NUP358 is an E3 SUMO ligase complex ( 85 ) and is localized to the centrosomes ( 86 ).Ther efor e, NUP358 or its SUMOylated substrates might be important for recruiting TRIM69A to the centrosomes.
Many of the tripartite motif (TRIM) proteins have roles in cell cycle phase tr ansitions, particular ly mitotic progression, and are implicated in human diseases including cancer ( 87 ).Interestingly, the E3 ligase TRIM37 pre v ents the formation of aberrant centr osomal pr otein assemblies that function as extra MTOCs and cause segregation defects ( 88 , 89 ).Our work identifies new cancer-relevant mitotic mechanisms for TRIM69, thereby expanding our understanding of biological roles of the TRIM family members.Song et al recently reported that TRIM69 binds and stabilizes MTs in interferon-stimulated myeloid cells to limit viral spread ( 90 ).Gi v en the di v erse cellular functions of P values are indicated in the Figure legends.Microscopy images shown are representati v e of at least 10 fields from two independent experiments.Western blot images are representati v e of two independent experiments.All biological and biochemical experiments were performed with appropriate internal negati v e and / or positi v e controls as indicated.

Figure 1 .
Figure 1.TRIM69A expression is associated with Centrosome Amplification 20 (CA20) gene expression signature, Homologous Recombination-Deficiency (HRD) and Aneuploidy in basal breast cancers.( A ) Heatmap showing relati v e e xpr ession of CA20 signatur e genes in r elation to TRIM69A mRNA expression and various other classifiers.Annotations for PAM50 subtype, BRCA1 germline muta tion sta tus, TP53 RNA subtype ( 43 ), self-reported race, and DNA Repair group are provided along with continuous CA20, DDR, and HRD signatur e scor es.Heatmap clustering was performed by both samples (columns) and genes (rows) using centroid linkage.( B ) Violin plot depicting distribution of CA20 scores by TRIM69A expression (low = bottom three quartiles, high = top quartile).P-value represents comparison of CA20 between TRIM69A high and low groups by Wilcoxon rank-sign test.( C ) Violin plot of aneuploidy scores separated b y TRIM69A expression group.P-v alue r epr esents comparison of aneuploidy scores ( 44 ) in TRIM69A high versus low samples by Wilco x on rank-sign test t .( D ) Violin plot of HRD scores according to TRIM69A expr ession group.P -value r epr esents comparison of HRD scores ( 44 ) in TRIM69A high versus low samples by Wilco x on rank-sign test.This is a violin plot of samples separated as TRIM69A High or Low.HRD score is plotted on the y-axis.Wilcoxon P -value was performed.

Figure 2 .
Figure 2. TRIM69A interacts with MST2 and MST1 mainly in a detergent-insoluble subcellular compartment.( A ) Immunoblots showing relative distribution of TRIM69A and TRIM69B between detergent-soluble and detergent-insoluble cell fractions in H1299 and MDA-MB-231 cells.( B ) Representati v e confocal microscopy images of HA-TRIM69A / B-expressing cells showing subcellular distribution of HA-TRIM69 (green) in relation to pericentrin (red) for each cell cycle stage (identified based on nuclear morphology).Scale = 5 m. ( C ) H1299 cells transduced with viruses encoding HA-TRIM69A, HA-TRIM69B, or an 'empty' adenovirus vector for control.Cells were biochemically-fractionated to generate detergent-soluble and detergent-insoluble e xtracts.The detergent-insolub le e xtracts were dissocia ted using DNAse and sonica tion.HA-TRIM69A and HA-TRIM69B complex es wer e immunopurified from the extracts and analyzed by mass-spectrometry.The table lists the most abundant proteins that specifically co-purified from with HA-tagged TRIM69 variants.Max Ratios were selected with a stringent filter to avoid a ratio < 2 f or an y sample.The bar chart shows fold-enrichment of various TRIM69-associated proteins in anti-HA-immunoprecipitates from HA-TRIM69A / B-expressing cells when compared with control (empty vector) cultures.( D ) Imm unoblot showing co-imm unoprecipitation of MYC-TRIM69A with HA-MST2.( E ) Confocal microscopy images of a r epr esentati v e H1299 cell showing co-localization of MYC-TRIM69A (red) with HA-MST2 (green).( F ) Confocal microscopy images of representati v e H1299 cells at different cell cycle stages showing subcellular distribution of HA-MST2 (green) in relation to pericentrin (red).Scale = 5 m. ( G ) Yeast 2-hybrid assays showing extent to which TRIM69A or TRIM69B interact with MST2 and MST1.

Figure 3 .
Figure 3. TRIM69A regulates MST2 ubiquitination and subcellular distribution.( A ) TRIM69 (WT), but not catal yticall y-inacti v e TRIM69A (E3 mut) promotes conjugation of Ub to MST2. ( B ) TRIM69 WT but not catal yticall y-inacti v e TRIM69A (E3 mut) undergoes autoubiquitination.( C ) MYC-TRIM69A does not affect MST2 stability in cy clohe ximide (CHX, 10 g / ml)-treated 239T cells.The half-life of MST2 was determined based on results from two different independent e xperiments.Relati v e le v els of MST2 as detected by immunoblotting were quantified using ImageJ software and normalized to GAPDH le v els.( D ) Wild-type MYC-TRIM69A (but not MYC-TRIM69A E3 mut or MYC-TRIM69B) redistributes MST2 to the detergent insoluble compartment in H1299 cells.( E ) Confocal microscopy images of r epr esentati v e H1299 cells showing TRIM69A-induced subcellular redistribution of HA-MST2.Scale bar r epr esents 5 m. ( F ) Imm unoblot showing that m ultiple independent TRIM69-directed siRNAs reduce the amount of endogenous MST2 associated with the detergent-insoluble compartment in H1299 and MDA-MB-231 cells.

Figure 4 .
Figure 4. TRIM69A promotes MST2-PLK1 interaction in the detergent-insoluble compartment.( A ) Effect of TRIM69 on the protein-interaction network of MST2 in detergent-soluble and detergent-insoluble compartments.1% FDR was used to filter all peptides / proteins; only proteins with > 1 peptides are reported.( B ) TRIM69A promotes co-localization of MST2 (red) and PLK1 (green) in H1299 cells.Scale bar r epr esents 10 m. ( C ) TRIM69A promotes interaction between MST2 and PLK1 in the detergent-insoluble compartment.( D ) Phosphoproteome profiling analysis showing TRIM69-inducible MST2 phosphorylation sites in the detergent-insoluble compartment.1% FDR was used to filter all peptides / proteins; only proteins with > 1 peptides are reported.( E ) Immunoblots showing phosphorylation of MST2 by TRIM69A in CSK insoluble fraction using SDS-PAGE and Phos-tag phospha te-af finity gel electrophoresis.( F ) Immunoblots showing TRIM69A promoted phosphorylation of NEK2A by MST2 in CSK insoluble fraction using SDS-PAGE and Phos-tag phospha te-af finity gel electrophoresis.( G ) Ef fect of MST2 phosphoryla tion site muta tions on co-localiza tion of MST2 and PLK1.Co-localiza tion was quantified using the ImageJ plug-in JAcoP.Each column r epr esents the mean ± standard error of the mean (SEM) from two independent experiments, n = 20 cells for each condition, * * P < 0.01.Scale bar r epr esents 5 m. ( H ) Effect of MST2 phosphorylation site mutations on MST2 interactions with TRIM69A and PLK1.

Figure 5 .
Figure 5. TRIM69 and MST2 promote centrosome separation.( A -D ) U2OS cells were transfected with the indicated plasmids for 24 h before a single thymidine block / release and treatment with 5 mM Eg5 inhibitor (STLC) to trap cells in prometaphase.(A-B) Representati v e confocal microscopy images showing sim ultaneous imm unostaining for pericentrin (green) and C-Nap1 (red) in U2OS cells harboring ectopically-expressed NEK2A, MST2, or TRIM69A.Scale = 5 m.(C) Effect of ectopically-expressed NEK2A, MST2, or TRIM69A on inter-centrosome distance.Results are from two independent experiments.n > 20 cells were analyzed for each condition.Data are mean ± standard error of the mean (SEM) ( * * P < 0.0001).(D) Effect of ectopically-expressed NEK2A, MST2 or TRIM69A on intensity of C-NAP1 signals at centrosomes.Results are from two independent experiments.n > 20 cells were analyzed for each condition.Data are mean ± standard error of the mean (SEM) ( * * P < 0.0001).( E, F ) Effect of siTRIM69 and siMST2 / 1 on inter-centrosome distance.Results are from two independent experiments; n > 20 cells were analyzed for each condition.Data are mean ± standard error of the mean (SEM) ( * * P < 0.0001).( G ) U2OS cells were transfected with indicated siRNAs for 48 h before a single thymidine block / release and treatment with 5 uM Eg5 inhibitor (STLC) to trap cells in prometaphase.The bar chart shows the effect of siTRIM69 and siMST2 / 1 on intensity of C-NAP1 signals at centrosomes.Results are from two independent experiments.n > 20 cells were analyzed for each condition.Data are mean ± standard error of the mean (SEM).* * P < 0.0001.

Figure 6 .
Figure 6.TRIM69A promotes triple-negati v e breast cancer (TNBC) cell proliferation and allows breast cancer cells to tolerate mitotic stresses due to centrosome amplification.( A ) Clonogenic survival assays using parental MDA-MB-231 ( TRIM69 + / + ) cells or a TRIM69 −/ − deri vati v e cell line.Quantitati v e analysis of colon y f ormation is presented in a bar chart (lower panel).Clonogenic survival of TRIM69 −/ − cells was normalized to colony survival of WT cells.Data points r epr esent the mean of triplicate determinations ± SEM. ** P ≤ 0.001.( B ) Immunofluorescence images of pericentrin (red) and ␤-tubulin (green) staining in taxol-treated MDA-MB-231 cells showing r epr esentati v e normal and clustered centrosomes.Scale = 5 m.The bar chart shows quantification of normal vs. scattered centrosomes in TRIM69 + / + and TRIM69 −/ − deri vati v e cells.Each column r epr esents the mean ± range from two independent experiments, n = 100 cells for each condition.( C ) Immunofluorescence images of pericentrin (red) and ␤-tubulin (green) staining in HAUS1 siRNA-treated MDA-MB-231 cells showing r epr esentati v e normal and clustered centrosomes.Scale = 5 m.The bar chart shows quantification of cells with normal or scattered centrosomes.Bars represent the mean ± range from two independent experiments, n = 100 cells for each condition.( D ) Immunoblots sho wing do xycycline-inducible expression of TRIM69A WT or TRIM69A E3 mut.( E ) The bar chart shows quantification of normal versus scattered centrosomes in TRIM69 + / + and TRIM69 −/ − cells after reconstituting TRIM69A WT or TRIM69A E3 mut.Each column r epr esents the mean ± range from two independent experiments, n = 50 cells for each condition.* P < 0.05.( F ) The bar chart shows quantification of colony formation in TRIM69 + / + and TRIM69 −/ − cells after inducing TRIM69A WT or TRIM69A E3 mut expression with or without taxol trea tment.Da ta points r epr esent the mean of triplicate determinations ± SEM. * P < 0.05.

Figure 8 .
Figure 8. TRIM69A promotes reorganization of microtubules, their associated motors and nucleoporins.( A ) Representati v e images showing effect of TRIM69A and TRIM69A E3 mut expression on immunostaining pattern of ␤-tubulin (green) in H1299 cells.Scale bar r epr esents 5 m. ( B ) Effect of TRIM69A expression on immunostaining pa ttern of acetyla ted ␣-Tubulin (red) in H1299 cells (left panel).Scale bar r epr esents 5 m.Immunoblots showing TRIM69A promoted acetylation of ␣-Tubulin in CSK insoluble fraction (right panel).( C ) The bar charts illustrate the quantification of tubulin spindle intensity after TRIM69A knockdown, number under bars are number of cells quantified.Data are mean ± standard error of the mean (SEM).* * P < 0.0001; n.s., not significant difference.( D ) Confocal microscopy images of a representati v e H1299 cell showing co-localization of MYC-TRIM69A (green) with mCherry-PRC1 (red) (Scale bar represents 5 m).( E ) Effect of TRIM69A expression on immunostaining of ␤-tubulin (green) and distribution of endogenous PRC1 (red) in H1299 cells.Scale bar r epr esents 5 m. ( F ) Confocal microscop y images showing the effect of mCherry-PRC1 (red) ov ere xpression on MT bundling (green) in H1299 cells.( G ) Effect of TRIM69 expression on immunostaining pattern of DYNLL1 (green) and ␤-tubulin (red) in H1299 cells.Scale bar r epr esents 5 m. ( H ) The bar charts illustrate that TRIM69-induced centrosome clustering in paclitax el-tr eated cells is inhibited by siRNAs against PRC1 and Dynein.Bars r epr esent mean ± range from two independent e xperiments.( I ) Representati v e images showing effect of PRC1 and DYNEIN knockdown on interphase MT-bundles after TRIM69A ov ere xpression.Immunostaining pattern of ␤-tubulin (green) is shown.All cells were imaged with the same imaging parameters.Scale bar r epr esents 5 m. ( J ) The bar charts illustrate the quantification of interphase MT-bundle width at the base of the peak r epr esented in Figur e I. n is the number of bundles-fiv e bundles per cell.Data are mean ± standard error of the mean (SEM), * * P < 0.0001; * 0.001 > P > 0.0001; n.s., not significant.( K ) The bar charts illustrate the quantification of interphase MT-bundle cross section intensity r epr esented in figure (I).n is the number of bundles-fiv e bundles per cell.Data are mean ± standard error of the mean (SEM), * * p < 0.0001; * 0.001 > p > 0.0001; n.s., not significant.( L ) Representati v e images showing effect of PR C1, D YNEIN and MST2 / MST1 knockdown on intensity of individual spindles after TRIM69A ov ere xpression in metaphase H1299 cells.Immunostaining pattern of ␤-tubulin (green) and Pericentrin (red) is shown.All cells were imaged with the same imaging parameters.Scale bar r epr esents 5 m. ( M ) The bar charts illustrate the quantification of tubulin spindle intensity r epr esented in (L), numbers under the bars indicate the number of cells quantified.Data are mean ± standard error of the mean (SEM).* * P < 0.0001; n.s., not significant differences.( N )The immunoblots validate effecti v e downregulation of PRC1 and DYNEIN protein in siRNA experiments.