SETDB1 regulates microtubule dynamics

Abstract Objectives SETDB1 is a methyltransferase responsible for the methylation of histone H3‐lysine‐9, which is mainly related to heterochromatin formation. SETDB1 is overexpressed in various cancer types and is associated with an aggressive phenotype. In agreement with its activity, it mainly exhibits a nuclear localization; however, in several cell types a cytoplasmic localization was reported. Here we looked for cytoplasmic functions of SETDB1. Methods SETDB1 association with microtubules was detected by immunofluorescence and co‐sedimentation. Microtubule dynamics were analysed during recovery from nocodazole treatment and by tracking microtubule plus‐ends in live cells. Live cell imaging was used to study mitotic kinetics and protein–protein interaction was identified by co‐immunoprecipitation. Results SETDB1 co‐sedimented with microtubules and partially colocalized with microtubules. SETDB1 partial silencing led to faster polymerization and reduced rate of catastrophe events of microtubules in parallel to reduced proliferation rate and slower mitotic kinetics. Interestingly, over‐expression of either wild‐type or catalytic dead SETDB1 altered microtubule polymerization rate to the same extent, suggesting that SETDB1 may affect microtubule dynamics by a methylation‐independent mechanism. Moreover, SETDB1 co‐immunoprecipitated with HDAC6 and tubulin acetylation levels were increased upon silencing of SETDB1. Conclusions Taken together, our study suggests a model in which SETDB1 affects microtubule dynamics by interacting with both microtubules and HDAC6 to enhance tubulin deacetylation. Overall, our results suggest a novel cytoplasmic role for SETDB1 in the regulation of microtubule dynamics.

me2/3 are usually associated with gene repression and heterochromatin formation, 1 and indeed SETDB1 is involved in silencing the X chromosome, 5,6 repetitive elements 7-10 and specific genes. [11][12][13] SETDB1 is important for various developmental processes including embryogenesis, 14,15 neurogenesis, 16 immune cell development 17 and germ line development. 8 SETDB1 was also shown to methylate nonhistone proteins such as inhibitor of growth protein 2 (ING2), 18 p53 19 and upstream binding factor (UBF). 20 SETDB1 is considered a nuclear protein; and indeed, its substrates either histone H3 or non-histone proteins, are mainly nuclear proteins. However, SETDB1 contains both nuclear localization signal (NLS) motifs and nuclear export signal (NES) motifs. 21,22 In addition, a cytoplasmic pool of SETDB1 has been found in several types of cells including HeLa cells, 23 HEK293 cells, 24 mouse embryonic fibroblasts, 22 differentiated myoblasts 21 and human melanoma biopsies. 25 Cytoplasmic localization of SETDB1 is thought to facilitate methylation of newly synthesized histones before their incorporation into nucleosomes 26 or to restrict the enzyme of methylating nucleosomal H3K9. 21,22 In recent years, SETDB1 has been considered an oncogene. Its genomic location is commonly amplified in melanoma 27,28 and its expression levels are increased in various types of cancer to support cancer cell proliferation, migration and invasion. These types of cancer include melanoma, 27,28 colorectal cancer, 29,30 liver cancer 19,31 and lung cancer. 32,33 More recently, SETDB1 was also linked to adaptive resistance of tumour cells to various drugs 2 and repression of the innate immune response. 7,34 Currently, SETDB1 is thought to promote cancer mainly by its nuclear activity of methylating H3K9 or transcription factors such as p53. 19 However, since a substantial amount of SETDB1 can be found in the cytoplasm, we looked for cytoplasmic activity of SETDB1 that may also promote cancer formation and progression.
Here we show that SETDB1 co-sedimented with microtubules (MTs) and that cytoplasmic SETDB1 partially co-localized with the MT network. Reduced SETDB1 levels increased MT stability as measured by EB1-tracking and MT recovery from nocodazole treatment and interfered with mitotic progression. Over-expression of either wild type (WT) or catalytic dead (CD) SETDB1 increased MT polymerization rate to the same extent, suggesting that SETDB1 can affect MT dynamics by an additional mechanism to substrate methylation. Finding interaction between SETDB1 and the tubulin deacetylase histone deacetylase 6 (HDAC6), along with increased tubulin acetylation levels after reduction in SETDB1 levels, suggest that SETDB1 may affect MT dynamics by supporting HDAC6 activity.

| Plasmids and molecular cloning
Plasmids expressing GFP-fused WT and H1224K SETDB1 fused to GFP were generated by PCR using pREV-SETDB1 as a template (a kind gift from Slimane Ait-Si-Ali, Centre National de la Recherche

| MT co-sedimentation assay
MT co-sedimentation assay was performed as reported previously. 37 Briefly, cells were lysed in PIPES buffer: 80 mM PIPES pH 6.8, 1 mM MgCl 2 , 1 mM EGTA, 100 mM NaCl, 1% Triton X-100, and 1Â protease inhibitor cocktail (539134, Millipore, Burlington, MA, USA) for 30 min on ice. Cell lysates were cleared by two repetitive centrifugations at 20,000g for 20 min at 4 C. The supernatants supplemented with 1 mM GTP and 40 μM Paclitaxel were incubated at 4 C or 37 C for 30 min for tubulin polymerization before centrifugation at 20,000g for 30 min at 4 C or 37 C, respectively.
The resulting pellets and supernatants were subject to Western blot analysis.

| MT recovery assay
Cells plated on fibronectin-coated coverslips were treated with 7 μg/ml of nocodazole for 3 h. Following three washings with cold DMEM to remove the nocodazole, the cells were incubated at 37 C in pre-warmed complete growth medium for the indicated periods of time, fixed and immunostained as described above. Quantitative data analysis was performed with ImageJ/Fiji software (NHI) by manual delineation of the area covered by MTOC-linked MTs. At least 30 cells were measured for each condition in each repetition and the average size of treated cells was normalized to the one of control cells. Average scores of three repetitions were calculated and statistical significance was determined by the Student's t-test.

| Time-lapse imaging
For live-cell imaging, cells were plated in 35-mm glass-bottom dishes.
Time-lapse images were collected with a coolSNAP HQ2 CCD camera (Photometrics, Tuscon, AZ, USA) mounted on an Olympus 1X81 fluorescent microscope at 37 C and 7% CO 2 . Frames were captured every 3 s for 1.5 min (movies to track growing MT ends) or every 3 min for 10 h (movies to monitor mitotic progression). Acquired images were analysed by ImageJ/Fiji software. To analyse growing MT ends, EB1-GFP comets were tracked manually using the MTrackJ ImageJ plugin. 38 Comets were analysed in each frame considering the distal site of the comet as the comet point. Data analysis was done according to duration and length tracked. To analyse mitosis progression, mitotic events were followed in terms of time and success rate.   Figure 1A). Immunostaining for SETDB1 verified this observation and revealed a partial co-localization of SETDB1 with MTs in B16-F1 cells ( Figure 1B,C). In WM266.4 cells, the cytoplasmic fraction of SETDB1 was smaller than in B16-F1 cells, though some colocalization of cytoplasmic SETDB1 with MTs could still be detected ( Figure 1D,E). To validate this observation, we used an additional antibody against SETDB1 that revealed a similar pattern of partial colocalization ( Figure 1D,E). Co-localization of SETDB1 with MTs was also found in HeLa cells ( Figure 1F). Moreover, this pattern was kept during mitosis in B16-F1 cells and in HeLa cells ( Figure 1G). Analysis of the localization of over-expressed GFP-fused SETDB1 revealed a similar pattern of partial co-localization with MTs ( Figure 1H,I).

| SETDB1 affects MT growth rate
To evaluate whether SETDB1 can affect MT functions, we first tested the impact of SETDB1 levels on the rate of MT growth during recovery from nocodazole treatment. B16-F1 cells were transfected with control short interfering RNA (siRNA) or SETDB1 siRNA, which reduced SETDB1 levels by about 50%, probably due to its long halflife ( Figure S1). Then cells were treated with nocodazole for 3 h to depolymerize MTs. Following nocodazole washout, the cells were Movies S1 and S2). Notably, MT growth duration, growth length and growth rate were increased in SETDB1 siRNA-transfected cells by 25%, 38% and 11%, respectively, in comparison to control cells ( Figure 5A-C). On the other hand, MT catastrophe rate in SETDB1 siRNA-treated cells was reduced by 15% in comparison to control cells ( Figure 5D). These results suggest that SETDB1 could be a negative regulator of MT growth.

| SETDB1 is important for proper cell division
The finding that SETDB1 regulates MT dynamics led us to test if SETDB1 is important for cell division, a process that is heavily dependent on MT organization. Indeed, reduced SETDB1 levels attenuated the proliferation of HeLa cells by 20% ( Figure 6A), while increasing unsuccessful mitotic events from 4.1% to 22.1% ( Figure 6B). Measuring the lengths of the different cell division stages in cells that were able to finish mitosis successfully revealed a significant increase in the duration of both early and late mitosis in SETDB1 siRNA transfected cells in comparison to Ctl siRNA-transfected cells. In SETDB1 siRNAtransfected cells, the durations from nuclear envelope breakdown (NEB) to anaphase and from anaphase to the appearance of a cleavage furrow were increased by 17% and 40%, respectively, in comparison to control cells ( Figure 6C-H).

| Catalytic-dead SETDB1 can affect MT growth rate
SETDB1 is a well-established methyltransferase that was found to methylate both histones 4 and non-histone proteins. 19,24,40 Moreover, recently α-tubulin was found to be methylated at lysine 40 by SET domain containing 2 (SETD2). 41 To evaluate if the effect of SETDB1 on MT growth rate is dependent on its methyltransferase activity, we repeated the MT recovery assay while over-expressing either WT The H1224K mutation was shown before to completely impair the methyltransferase activity of SETDB1. 4 As shown in Figure 7, overexpression of the CD SETDB1 attenuated MT recovery from nocodazole treatment to the same extent as over-expression of WT SETDB1.
Thus, SETDB1 could affect MT dynamics by a mechanism that is independent of its enzymatic activity.

| SETDB1 affects tubulin acetylation
The molecular mechanism by which SETDB1 affects MT dynamics seemed to be at least partially independent of its methyltransferase activity. To identify a molecular mechanism by which SETDB1 could affect MT dynamics, we took into account two features: the first is that nuclear SETDB1 is known to participate in a complex with HDAC1 and HDAC2 to repress transcription. 42 The second is that acetylation of Lys40 in α-tubulin (acetylated tubulin) is a key tubulin post-translational modification, which is associated with stable MTs in various cellular contexts. 43,44 We hypothesized that SETDB1 may interact with a tubulin HDAC and affect its activity. Since the major tubulin deacetylase is HDAC6, we tested if SETDB1 can interact with it and affect tubulin acetylation levels. As shown in Figure 8A, HDAC6 co-immunoprecipitated with SETDB1, suggesting an in vivo interaction between these two factors. In addition, reduction in SETDB1 levels led to an increase of 110% in the levels of tubulin acetylation ( Figure 8B,C), thus supporting the hypothesis that SETDB1 may serve as a co-factor of HDAC6 in the cytoplasm to support tubulin deacetylation to regulate MT dynamics.

| DISCUSSION
SETDB1 is a well-established H3K9 methyltransferase involved in embryogenesis and development 8,[14][15][16][17] as well as in the aetiology of cancer. [45][46][47] Current perception is that SETDB1 affects these processes due to its nuclear localization, while any cytoplasmic localization of SETDB1 serves to either sequester it or methylate newly generated histone H3. 21 22,23,39 and also in our hands. Thus, the effect of OE SETDB1 on MT dynamics suggests that the impact of SETDB1 on MT growth rate is due to the activity of its cytoplasmic pool rather than the nuclear SETDB1.
Recently, the methyltransferases SETD2 and KMT5A were shown to methylate α-tubulin on K40 and K311, respectively, while SETD2 activity was limited to the M phase. 41,48 To assess if SETDB1 may affect MT dynamics by its methyltransferase activity, we evaluated the effects of CD SETDB1 on MT dynamics and found that both WT and CD SETDB1 reduced MT recovery rate from nocodazole treatment.
These results suggest that the effects of SETDB1 on MTs did not require its methyltransferase activity. This observation is in agreement with studies on SETDB1 oncogenic activity in the zebrafish model: over-expression of SETDB1 was found to accelerate melanoma onset in a zebrafish model for melanoma formation and progression. Notably, over-expression of the enzymatically inactive H1224K SETDB1 (CD SETDB1) still significantly accelerated melanoma onset, although to a lower extent than WT SETDB1. 27 More recently, the Caenorhabditis elegans homologue of SETDB1, MET-2, was shown to be able to affect transcription even when its methyltransferase activity was ablated. 49 This suggests that the oncogenic function of SETDB1 is partially contributed by methyltransferase independent activities of the protein.
One of these activities may be the modulation of MT dynamics.
Based on the findings that SETDB1 interacted with HDAC6 and that reduced levels of SETDB1 led to increased tubulin acetylation, we hypothesize that SETDB1 can modulate MT dynamics by affecting the activity of the tubulin deacetylase HDAC6. Tubulin acetylation is associated with more stable MTs ( 43,44 ), increased mechanical stabilization of MTs and reduced MT mechanical breakage. 50 Thus, SETDB1 may support HDAC6 activity to reduce MT stability, leading to F I G U R E 8 SETDB1-HDAC6 interaction. (A) Co-IP of overexpressed SETDB1-GFP and HDAC6-FLAG in HEK293 cells. (B) Western blot analysis of the Ac-tubulin levels in B16-F1 cells transfected with either Ctl siRNA or siRNA against SETDB1. (C) The ratio of Ac-tubulin to Tubulin levels in three repetitions normalized to the same ratio in Ctl siRNA-transfected cells ±SE. Statistical significance was calculated with Student's t-test, *p < 0.05 increased MT catastrophe rate and slower recovery from nocodazole treatment. Notably, altered levels of both tubulin acetylation and HDAC6 were detected in several types of cancer. In some cases, such as in breast cancer and glioblastoma, HDAC6 was shown to support tumour progression. 51 Our results, taken together with previous reports, suggest that SETDB1 may affect various aspects of tumour progression such as cell division and cell migration, at multiple levels by different mechanisms.
At the chromatin level, it may affect transcriptional control of key factors to support cell migration and proliferation 28,30,52,53 and it may promote global chromatin condensation to support cell migration 52,54 ; while at the MT level, it may modulate MT dynamics possibly by affecting the activity and/or the binding of microtubule-associated proteins (MAPs) to MTs such as HDAC6.
Cross-talk between interphase chromatin and MTs was found before: MT-driven mechanical forces were shown to alter chromatin organization, 55,56 the motor protein kinesin KIF4 was found to be involved in heterochromatin formation, transcription and DNA repair. 57 Dynein light chain 1 (DLC1) was found to be involved in transcriptional control. 56 Tau nuclear subcellular localization is thought to associate with DNA damage protection. 58 LIS1, a regulator of cytoplasmic dynein, was shown to interact with histone H1 and MeCP1 and to affect the chromatin binding of the latter. 59 Thus, SETDB1 seems to join this growing group of factors with activities in both the chromatin and the MT worlds.