Conditional mutation of Smc5 in mouse embryonic stem cells perturbs condensin localization and mitotic progression

Pluripotent stem cells (PSCs) possess unlimited differentiation potential and are considered to be a source of cells for regenerative medicine. PSCs also share multiple characteristics with cancer cells and are prone to acquisition of chromosomal abnormalities in cell culture. Although the underlying mechanisms that initiate genomic instability are yet to be elucidated, DNA replication stress and defective chromosome condensation, decatenation and segregation are likely contributing factors (Na et al., 2014; Lamm et al., 2015; Damelin et al., 2005).

It is expected that the structural maintenance of chromosomes (SMC) complexes play crucial roles in PSC preservation, proliferation and differentiation. The SMC complex family includes cohesin, condensins and Smc5/6, and each complex contributes to maintaining eukaryotic genome integrity. Malfunction of SMC complexes is often associated with severe developmental disorders. Mutations and misexpression of cohesin lead to developmental disorders collectively known as cohesinopathies (Bose and Gerton, 2010; Skibbens et al., 2013). Condensins are crucial for early embryonic development and cerebral cortex formation in mice (Nishide and Hirano, 2014). Misregulation of condensin II has been implicated in the neurodevelopmental disorder autosomal recessive primary microcephaly (MCPH) (Hirano, 2012; Trimborn et al., 2006). Recently, hypomorphic mutations of a Smc5/6 component, NSMCE2, have been reported to cause primordial dwarfism and microcephaly in human (Payne et al., 2014). Furthermore, SMC mutations are associated with cancer (Leiserson et al., 2015; Jacome et al., 2015). Cohesin mutations have been identified in myeloid leukemias (Leeke et al., 2014). The condensin component Smc2 and the Smc5/6 complex component Smc6 are overexpressed in cancers, such as colorectal, neuroblastoma and breast cancer (Davalos et al., 2012; Murakami-Tonami et al., 2014; Stevens et al., 2011). Furthermore, mutation of Smc5 is linked to the development of brain metastases (Saunus et al., 2015).

Each SMC complex is composed of two SMC components forming a V-shaped heterodimer, which is bridged by non-SMC subunits (Hirano, 2006, 2012). Cohesin comprises the Smc1 and Smc3 heterodimer, bridged by the α-kleisin subunit Rad21 and one of two stromal antigen proteins, Stag1 or Stag2. The canonical function of the cohesin complex is to hold sister chromatids together following DNA replication. Cohesin removal is required to ensure chromosome segregation during cell division (Nasmyth and Haering, 2009). There are two condensin complexes, condensin I and condensin II, both promote compaction and disentanglement of sister chromatids prior to chromosome segregation (Hirano, 2012). Condensin I and II share the core Smc2 and Smc4 heterodimer; however, they are made unique by their complex specific non-SMC subunits. In mammals, the Smc5/6 complex contains a Smc5 and Smc6 heterodimer and four non-SMC elements Nsmce1– Nsmce4 (also known as Nse1–Nse4) (Hirano, 2006). In addition, two Smc5/6 complex localization factors (Slf1 and Slf2) have recently been discovered (Räschle et al., 2015).

Studies using budding and fission yeast mutants have shown that the Smc5/6 complex is required for replication fork stability, facilitating the resolution of joint molecules and preventing the formation of aberrant joint molecules that can lead to mitotic catastrophe (reviewed in Carter and Sjögren et al., 2012; Jeppsson et al., 2014; Langston and Weinert, 2015; Murray and Carr, 2008; Verver et al., 2016; Wu and Yu, 2012). The distinct roles of the Smc5/6 complex in mammalian cells have yet to be defined. However, localization and small interfering RNA (siRNA) knockdown studies in mammalian cells suggest that the complex is required during DNA replication, DNA repair and chromosome segregation (Wu et al., 2012; Gallego-Paez et al., 2014; Gomez et al., 2013).

Faithful chromosome segregation depends on cooperative functioning of the SMC complexes and multiple cell cycle kinases including polo-like kinases (Plks), cyclin-dependent kinases (Cdks) and Aurora kinases. For instance, Plk1-mediated phosphorylation of cohesin stimulates removal of arm cohesin during prometaphase (Giménez-Abián et al., 2004). Condensins are phosphorylated by Cdk1, Plk1 and Aurora B kinases to ensure proficient chromosome condensation (Abe et al., 2011; Lipp et al., 2007; Tada et al., 2011). In addition, condensins are required for appropriate localization of Aurora B and Plk1 kinases during the prophase-to-metaphase transition and ensure accurate chromosome segregation (Abe et al., 2011; Kim et al., 2014; Green et al., 2012; Kitagawa and Lee, 2015). Components of the Smc5/6 complex have been reported to be phosphorylated by Plk1 and Aurora B kinases during mitosis (Hegemann et al., 2011). However, mechanistic links between Smc5/6 complex and cell cycle kinases have yet to be determined.

To assess the requirements for the Smc5/6 complex in stem cell genome maintenance, we aimed to use a knockout mouse approach. Previous studies have reported that Smc5/6 components are essential for early embryonic development in mouse (Ju et al., 2013; Jacome et al., 2015). Therefore, we created a Smc5 conditional knockout mouse, which we used to investigate functions of the Smc5/6 complex in mouse embryonic stem cells (mESCs). Cre-ER^T2-mediated mutation of Smc5 impacted mitotic progression, leading to the formation of chromosomal bridges, appearance of lagging chromosomes during anaphase and, ultimately, to aneuploidy. mESCs accumulated in the G2 phase of the cell cycle and activated apoptotic signaling. Microscopy studies revealed the irregular distribution of condensin, Plk1 and Aurora B in Smc5-depleted mitotic cells, which correlated with distorted chromosome structure and abnormal spindle morphology. In summary, our data demonstrate that the absence of functional Smc5/6 complex in mESCs leads to rapid cell death as a result of disrupted genomic integrity and mitotic failure.

To investigate the role of the Smc5/6 complex in PSCs, we established two sets of mESC lines with experimental (Smc5^flox/del, Cre-ER^T2; called Smc5-13^exp and Smc5-1^exp) and control (Smc5^wt/flox, Cre-ER^T2; called Smc5-3^cont and Smc5-6^cont) genotypes. Cell lines with both genotypes similarly displayed characteristic mESC colony morphology and expressed the pluripotency-associated markers Oct4 (also known as Pou5f1), Sox2, SSEA-1 (also known as Fut4) and a high level of alkaline phosphatase activity (Fig. 1A; Fig. S1A). After injection into immuno-deficient mice, mESCs formed teratomas representing derivatives of all three embryonic germ layers (Fig. 1B; Fig. S1B). Thus, using in vitro and in vivo assays, we confirmed pluripotency of established mESC lines. As an additional control, we established a wild-type cell line with the same C57BL/6J genetic background (Fig. S1A).

To initiate Cre-ER^T2 recombinase activity and excise the floxed Smc5 exon 4, two experimental and two control mESC lines were treated with 0.2 µM 4-hydroxytamoxifen (4-OH TAM) (Fig. 1C; Fig. S2A). This dose was sufficient to excise the targeted sequence within 3 days of treatment (Fig. 1D; Fig. S2B). The deletion of Smc5 exon 4 in experimental cell lines was accompanied by the loss of Smc5 protein expression and a dramatic decrease in the Smc6 level compared to control cell lines (Fig. 1E; Fig. S2C).

Published studies and our observations have revealed that some cells can escape Cre recombinase activity and overtake cell cultures (Yoshida et al., 2010). Therefore, after 3 days of 0.2 µM 4-OH TAM treatment, we kept a concentration of 0.05 µM 4-OH TAM in culture. Cells were evaluated for changes in gene and protein expression, and cell cycle perturbations during the next three passages after initiating 4-OH TAM treatment (Fig. S2A). For simplicity, we refer to the wild-type Smc5 as ‘Smc5⁺’, floxed Smc5 as ‘Smc5^flox’ and deleted Smc5 as ‘Smc5⁻’.

Smc5 depletion in mESCs resulted in extensive cell death after 5–8 days of 4-OH TAM treatment (Fig. 2A; Fig. S2D). However, this was not the case for immortalized mouse embryonic fibroblasts (MEFs). Instead, the proliferation rate of fibroblasts diminished after about 10 days of drug addition, but cell death was negligible (Fig. S2E). mESCs proliferate very rapidly compared to MEFs. To exclude possible influence of slow cell expansion on the appearance of phenotypic changes, we assessed MEF proliferation during 17 days. After 14 and 17 days of 4-OH TAM treatment, MEFs increased in cell number by less than two-fold, whereas untreated MEFs increased by six-fold (Fig. S2E). PCR analysis revealed that excision of Smc5 exon 4 was much less efficient in MEFs compared to mESCs, requiring up to 10 days of 4-OH TAM treatment (Fig. S2F). We speculated that a high expression level of Smc5/6 components is essential for pluripotent mESCs, and lower levels of Smc5/6 are required for cells at more advanced developmental stages. To support this hypothesis, we evaluated Smc5 and Smc6 expression during mESC differentiation in embryoid bodies. Notably, our data showed a decrease in protein amount with prolonged embryoid body culture (Fig. 2B). This result suggests that higher levels of the Smc5/6 complex are required for the highly proliferative mESCs compared to differentiated cells.

In contrast to somatic cells, PSCs are very sensitive to stress factors and readily undergo differentiation or apoptosis under suboptimal conditions (Weissbein et al., 2014). We explored whether depletion of Smc5 would lead to changes in mESC gene expression profile (Fig. 2C; Fig. S2G). We did not observe a decrease in pluripotency markers Oct4 and Nanog. The mesendoderm marker brachyury (T) was expressed in all cell samples at a low level. No expression of the endoderm marker α-fetoprotein was detected. However, after 5 and 8 days (P2 and P3) of 4-OH TAM treatment to experimental mESCs, we observed an upregulation of NeuroD1, which is an early marker of ectoderm. A slight increase in the expression of later ectoderm marker Pax6 was observed for the mESC Smc5-13^exp (experimental; Smc5-3^font/del, Cre-ER^T12) cell line (Fig. S2G). We also observed upregulation of Myc in experimental mESCs after 5 and 8 days of 4-OH TAM treatment (Fig. S2G). Thus, the depletion of Smc5 protein in mESCs does not lead to downregulation of pluripotency markers, but causes a shift in the pluripotent state by inducing the expression of differentiation markers.

The rapid cell death we observed could be mediated by the stress response factor p53, which also impedes somatic cell reprogramming and promotes ESC differentiation (Weissbein et al., 2014; Lin et al., 2012). The basal p53 level in ESCs is generally higher than in somatic cells, and regulation of p53 signaling is different (Lin et al., 2012). ESCs activate p53 signaling in response to DNA damage; however, they do not arrest at the G1 phase of the cell cycle as is the case for somatic cells (Weissbein et al., 2014). Previous studies on mouse and human ESCs have shown that acetylation of the C-terminus of p53 plays a crucial role in its activation in response to genotoxic stress (Feng et al., 2005; Chung et al., 2014). Depletion of Smc5 in our experimental mESCs did not change total p53 level. However, after 5 and 8 days (P2 and P3) of 4-OH TAM treatment we observed p53 acetylation at Lys379 (Fig. 2D), indicating stimulation of p53 transcriptional activity.

Another common feature of apoptotic signaling involves activation of caspases and subsequent cleavage of target proteins including Parp1 (Chaitanya et al., 2010; Luo and Kraus, 2012). Smc5 depletion in mESCs led to cleavage of Parp1 at P2 and P3 (Fig. 2D). This correlated with substantial cell death at these time points (Fig. 2A).

Given the important roles reported for the Smc5/6 complex in DNA replication, repair and chromosome segregation, we hypothesized that its deficiency would lead to changes in cell cycle dynamics and distribution. Indeed, the decrease in cell growth within two passages was associated with significant changes in cell cycle distribution (Fig. 2E). PSCs have a relatively short G1 phase, and prevalence of a cell population in S phase (Li et al., 2012). After 5 and 8 days (P2 and P3) of 4-OH TAM treatment, we observed a significant decrease in the percentage of experimental cells in S phase (minus 10.19% and 10.3% for P2 and P3, respectively, compared to plus 1% for P1) and accumulation of cells in G2 phase (plus 10.87% and 10.7% for P2 and P3, respectively, compared to plus 1.29% for P1) (Fig. 2E). These changes in cell cycle distribution could be a result of DNA damage during interphase and the inability of cells to segregate chromosomes (Mankouri et al., 2013). We also detected an increase in the polyploid cell population (∼7%, Fig. 2F). Giemsa staining of Smc5-depleted cells confirmed the presence of cells with an abnormal karyotype (Fig. 2G). We did not observe changes in cell cycle distribution or polyploid cell population in either the untreated (Smc5^−/flox and Smc5^+/flox) or control mESCs treated with 4-OH TAM (Smc5^+/−) (Fig. 2E,F).

To investigate the reasons for disrupted mitotic progression and death of Smc5-depleted mESCs, we used whole-cell immunofluorescence microscopy. First, we characterized the localization of Smc5/6 components during cell cycle progression in mESCs with functional Smc5. Smc6 and Nse1 proteins were enriched within the nucleus during interphase, and accumulated at the pericentromeric regions of chromosomes during prometaphase, metaphase and anaphase (Fig. 3A,B; Fig. S3A,B). At late stages of interphase, we also observed perinuclear granules of Nse1. These also persisted in the cytoplasm during anaphase (Fig. 3B). Similarly, antibodies against Smc5 detected protein enrichment near centromeres during prometaphase, metaphase and anaphase (Fig. 3C). In telophase and interphase, Smc5 immunocytochemistry showed nuclear and cytoplasmic staining (Fig. 3C). To support our data further, we performed immunostaining of chromosome spreads, which confirmed pericentromeric enrichment of Smc5/6 components on mitotic chromosomes and localization along the condensed chromosome arms (Fig. 3D).

We also observed Smc6 accumulation in distinct nuclear foci during interphase and prominent spindle pole localization through mitosis and up to late anaphase (Fig. S3C,E). Smc6 localization correlated with centrosome positioning during mitosis (Fig. S3C). Interestingly, Mad2 showed a similar localization pattern, including co-localization with Smc6 in interphase nuclei foci (Fig. S3D,F). Nse1 was also detected at spindle poles in mitotic cells (Fig. S3G). Because of abundant cytoplasmic staining with anti-Smc5 antibodies, we did not obtain conclusive images with Smc5 spindle pole localization. However, we propose that this is also the case for Smc5, based on the decline in Smc6 levels upon Smc5 mutation, and the similar localization observed for Smc6 and Nse1.

Smc5-depleted mESCs demonstrated a dramatic increase in abnormal mitotic cells harboring lagging chromosomes, chromosome bridges or both (Fig. 4A,B). We also observed a prominent increase in mitotic cells with abnormal spindles demonstrating aberrant chromosome segregation (Fig. 4B). Approximately 80% of Smc5-depleted mESCs could not complete accurate chromosome segregation. In contrast, 4-OH-TAM-treated control cells did not demonstrate an increase in mitotic abnormalities (Fig. 4A). To evaluate the appearance of abnormal mitotic cells in more detail, we also collected control and Smc5 mutant cells after nocodazole-induced mitotic block and release at time points corresponding to day 3, 4 and 5 of 4-OH TAM treatment (Fig. S4A). Numerous abnormal mitotic cells were clearly detected after 4 and 5 days of 4-OH TAM treatment, but not after 3 days (Fig. S4A).

Our study of Smc5/6 complex localization during normal mESC mitosis showed that the complex accumulates at pericentromeric regions (Fig. 3). Depletion of Smc5 protein led to disappearance of Smc5 from mitotic chromosomes (Fig. 4C). We also observed the absence of Smc6 and Nse1 components from pericentromeric regions of chromosomes in cells undergoing abnormal mitosis (Fig. 4D). These data underline the importance of Smc5/6 complex in chromosome segregation.

To further enhance our observations, we synchronized control and Smc5-depleted mESCs in the G2 phase of the cell cycle and analyzed cell progression by flow cytometry (Fig. 4E). In support of our previous data obtained using asynchronous cells, we observed a mitotic delay and accumulation of Smc5-deficient mESCs in G2 phase (Figs 2E and 4E). A previous study employing human somatic RPE-1 cells has demonstrated the requirement of Smc5/6 complex for timely progression of DNA synthesis (Gallego-Paez et al., 2014). Given that DNA replication stress can further cause defects in stem cell chromosome condensation and abnormal segregation (Lamm et al., 2015), we synchronized mESCs in the G1 phase of cell cycle and observed the progression through the S phase (Fig. 4E). Owing to accumulation of Smc5-depleted mESCs in G2 phase, we could not obtain a G1-enriched mESC population similar to controls. However, the majority of cells entered S phase and did not demonstrate a delay in DNA replication. There was a minor cell population that remained in G1 phase. Given that mESCs do not demonstrate G1 cell cycle arrest, accumulation of cells in G1 phase could be explained by acquisition of chromosomal damage during mitosis (Fig. 4E).

Based on previous siRNA-mediated knockdown studies of the Smc5/6 complex in mammalian cell lines, we hypothesized that the chromosome bridges and lagging chromosomes we observed after Smc5 depletion could be attributed to defects in chromosome cohesion, condensation and/or topoisomerase IIα (Topo IIα) function (Wu et al., 2012; Gallego-Paez et al., 2014; Gomez et al., 2013).

Timely removal of pericentromeric cohesin is a prerequisite for accurate sister chromatid segregation during the metaphase-to-anaphase transition (Giménez-Abián et al., 2004). Thus, we tested the localization of Rad21, a cohesin subunit, in mitotic mESCs (Fig. 5A–C). In cells with functional Smc5/6, cohesin was enriched at pericentromeric regions in prometaphase and dissociated from chromosomes after anaphase onset (Fig. 5A,C). We observed similar localization pattern for cohesin in Smc5-depleted cells, suggesting that aberrant cohesin removal was not the reason for lagging chromosomes and DNA bridges (Fig. 5B,C).

Incomplete decatenation of centromeric DNA and the presence of unresolved joint molecules can lead to chromatin bridge formation in mitotic cells (Mankouri et al., 2013). The separation of sister centromeres requires the activity of Topo IIα (Liu et al., 2014). Previous studies on human Smc5/6-depleted RPE-1 cells have reported mislocalization of Topo IIα from pericentromeric regions to arms and distal ends of chromosomes (Gallego-Paez et al., 2014). However, we did not observe a defect in Topo IIα localization. In Smc5-depleted mESCs, that were undergoing chromosome missegregation, Topo IIα remained enriched at the pericentromeric regions (compare Fig. 5E with Fig. 5F). Analysis of mitotic chromosome spreads did not reveal mislocalization of Topo IIα (Fig. 5D).

Condensins are required for structural organization of mitotic chromosomes ensuring accurate segregation during anaphase (Hirano, 2012; Thadani et al., 2012). Previous studies have shown aberrant distribution of condensin in Smc5/6-depleted human RPE-1 cells (Gallego-Paez et al., 2014). Thus, we investigated whether the localization of condensin subunit Smc4 was affected. In mESCs with a functional Smc5/6 complex, we observed enrichment of condensin at pericentromeric regions during prometaphase (Fig. 6A). In contrast, most Smc5-depleted prometaphase cells demonstrated decreased condensin accumulation at pericentromeric regions, and the condensin signal was increased along chromosome arms (Fig. 6B). Quantification of the condensin signal along chromosome arms and pericentromeric regions supported our observations (Fig. 6C–E). We also evaluated condensin distribution on mitotic chromosome spreads from control and Smc5 mutant mESCs, which confirmed the data obtained with whole-cell immunocytochemistry (Fig. 6F,G). In summary, our results demonstrate that the Smc5/6 complex is essential for normal chromosome condensation in mESCs.

Dynamic binding to chromatin and DNA supercoiling activity of condensin is regulated by mitotic kinases, including Aurora B and Polo-like kinase 1 (Plk1) (Hirano, 2012; Thadani et al., 2012). In mESCs, Plk1 was enriched at pericentromeric regions of chromosomes from prometaphase through mitosis and also redistributed to spindle poles and microtubules during metaphase and anaphase (Fig. 7A). Smc5-depleted mitotic cells displayed diminished Plk1 enrichment at pericentromeric regions (Fig. 7B).

Aurora B is required for kinetochore localization of spindle assembly checkpoint (SAC) proteins, including Mad2 (also known as MAD2L1), which are collectively required to ensure amphitelic kinetochore attachment to microtubules prior to the metaphase-to-anaphase transition (Maresca, 2011; Schuyler et al., 2012). Although we observed persistence of Mad2 protein at pericentromeric regions in prometaphase mESCs and on chromosome spreads, the Mad2 signal was diminished after Smc5 depletion in mitotic whole cells (Fig. 7C–F). Mad2 was generally absent in anaphase cells, consistent with its role in SAC (Fig. 7C,D) (Schuyler et al., 2012). The decrease in Mad2 level in Smc5-depleted cells suggests that SAC was either satisfied prior to chromosome segregation or SAC function was disrupted (Schuyler et al., 2012).

In addition to its canonical role in regulating the SAC, Aurora B kinase is required for appropriate chromosome condensation (Kitagawa and Lee, 2015). In mESCs, we detected Aurora B along chromosome arms and near centromeres during prophase, near centromeres during prometaphase and metaphase, and on spindle microtubules at the cleavage furrow during anaphase (Fig. 8A). Smc5-depleted cells displayed aberrant distribution of Aurora B in mitotic cells. Aurora B was enriched at centromeric regions, but in contrast to control cells, persisted along chromosome arms (Fig. 8B–E).

Excessive Aurora B activity during mitosis induces defective chromosome congression and causes chromosome missegregation (Dobrynin et al., 2011). Thus, we tested Aurora B activity by assessing histone H3 phosphorylation at the Ser10 residue (H3S10). In mESCs phosphorylation of H3S10 is detectable in the S phase of cell cycle and reaches the maximum during mitosis (Mallm and Rippe, 2015). Our analysis did not reveal aberrancies in the H3S10 phosphorylation pattern (Fig. S4B). Interestingly, H3S10 phosphorylation signal revealed chromatin fibers that appeared to connect chromosomes during mitosis (Fig. S4B).

Finally, intrigued by the increased amount of abnormal mitotic cells after Smc5 depletion, we used H3S10 phosphorylation to calculate the percentage of cells at different stages of mitosis, as well as cells in prophase. We recorded an increase in the population of cells in late prophase (Fig. S4C). This finding suggests that defects observed in Smc5-depleted mESCs during mitosis are due to aberrancies that occur during the prophase-to-prometaphase transition.

PSCs are unique cells capable of differentiation into numerous tissue-specific cell types and share some characteristics with cancer cells (Evans and Kaufman, 1981; Martin, 1981; Thomson et al., 1998). Studying PSCs might help elucidate processes supporting genetic stability of stem cells, reveal factors leading to malignant transformation and possibly find therapeutic targets to cure cancers. Functions of the Smc5/6 complex have been investigated in different model systems and cell types from budding yeast to human cell lines (reviewed in Carter and Sjögren et al., 2012; Jeppsson et al., 2014; Langston and Weinert, 2015; Murray and Carr, 2008; Verver et al., 2016; Wu and Yu, 2012). These reports have provided information about the crucial roles of the Smc5/6 complex in preserving genomic integrity. However prior to our research, mammalian PSCs had not been assessed. Furthermore, previous reports studying the function of Smc5/6 using human cells have been hindered by siRNA off-target effects (Wu et al., 2012). Our conditional knockout strategy using Cre-ER^T2 recombinase-induced mutation of Smc5 avoided off-target effects, and resulted in efficient depletion of Smc5 as well as Smc6. This suggests that Smc5 and Smc6 are stabilized within the context of the Smc5/6 complex.

We revealed varying effects of Smc5 depletion in embryonic versus somatic cells. mESCs begin to undergo apoptosis shortly after Smc5 depletion, whereas MEFs display diminished proliferative capacity. Observed outcomes highlight different needs for the Smc5/6 complex or distinct regulation of cellular processes in these cell types. Similar observations have been reported when condensin I and II are both depleted in mESCs compared to MEFs (Fazzio and Panning, 2010). Species- and cell-type-specific differences have been also reported by others. For example, after depletion of the core condensin subunit Smc2, mouse neural stem cells (NSCs) progressed through mitosis, underwent defective chromosome segregation and DNA damage-induced apoptosis. However, the depletion of Smc2 in human RPE-1 cells causes senescent-like phenotype and cell cycle arrest in G1 phase (Nishide and Hirano, 2014). Additionally, Smc5 knockdown in chicken DT40 lymphoma cells and human RPE-1 cells does not affect cell viability, although cells showed lower proliferation rates (Stephan et al., 2011; Gallego-Paez et al., 2014).

Contrary to somatic cells, ESCs do not have effective G1/S phase cell cycle checkpoints. Instead, ESCs with acquired DNA damage proceed to S phase and accumulate in G2 phase (Weissbein et al., 2014; van der Laan et al., 2013; Desmarais et al., 2012). We observed an increase in the G2 population of Smc5-depleted mESCs. We also observed a minor fraction of polyploid cells, which suggests that some cells did not complete mitosis and underwent another round of DNA replication. Our results echo the outcomes of condensin knockdown studies in mESCs (Fazzio and Panning, 2010).

Our observations also complement a recent report using budding yeast, which demonstrated that the absence of Smc5/6 during S-phase does not impact upon DNA replication efficiency and cell proliferation, whereas the absence of Smc5/6 during the G2/M transition causes cell cycle arrest and lethality (Menolfi et al., 2015). That study specified that Smc5/6 is required to resolve recombination structures formed from endogenous replication stress and late replication pausing sites. Human PSCs are prone to genomic instability due to replicative stress and impaired chromosome condensation (Lamm et al., 2015). It is likely that Smc5/6 plays a crucial role in avoiding these sources of genomic instability in PSCs. We speculate that perturbation of Smc5/6 function could result in PSC transformation and tumor formation from other cell types with impaired intra-S or decatenation checkpoints.

Observing cells with abnormal mitotic spindles and missegregating chromosomes prompted us to investigate mitotic regulators involved in chromosome congression and segregation. We showed that Smc5 depletion in mitotic mESCs resulted in a decrease of condensin signal at the pericentromeric regions and persistence of condensin along chromosome arms. Abnormal condensin distribution on mitotic chromosomes correlated with diminished Plk1 enrichment at pericentromeric regions. A previous study on human cells showed that stable localization of Plk1 to the kinetochores requires an interaction with the condensin II subunit Ncapg2 (Kim et al., 2014). Plk1 is required for stabilizing kinetochore–microtubule attachments by counteracting the spindle assembly checkpoint function of Aurora B (Suijkerbuijk et al., 2012). As we observed reduced Plk1 localization, but no effect on the enrichment of Aurora B at centromeric regions, it is likely that Aurora B is overactive during mitosis. Studies using mammalian cell lines have shown that Plk1 and Aurora B are also required to coordinate chromosome condensation. Plk1 phosphorylates the condensin II subunit Cap-D3 (also known as Ncapd3), which leads to correct chromosome assembly prior to segregation (Abe et al., 2011). Association of condensin I to the chromatin is regulated by Aurora-B-mediated phosphorylation (Lipp et al., 2007; Tada et al., 2011). As we observe increased enrichment of condensin and Aurora B along chromosome arms, Smc5/6 might be essential for coordinating the release of Aurora B from chromosome arms after condensin I deposition. A systematic phosphorylation screen of human mitotic protein complexes demonstrated that Smc5, Smc6 and NSMCE4 are phosphorylated, and these phosphorylation events are sensitive to chemical inhibition of either Aurora B or Plk1 (Hegemann et al., 2011). Therefore, determining the importance of these modifications during mitosis might provide insight into the mechanistic link between the Smc5/6 complex and these cell cycle kinases.

Depletion of the protein degradation chaperone complex Cdc48–Ufd1–Npl4 (Cdc48 is also known as p97 and VCP; Npl4 is also known as NPLOC4), in mitotic HeLa cells has been reported to cause abnormal distribution of Aurora B along chromosome arms (Dobrynin et al., 2011). Furthermore, the persistence of Aurora B resulted in defects in chromosome congression and segregation, similar to what we observe in Smc5 mutant mESCs. The Ufd1 subunit of this chaperone complex has been shown to bind ubiquitin and small ubiquitin-like modifications (SUMO) (Nie et al., 2012). SUMO modification of Aurora B stimulates its removal from chromosome arms during prometaphase (Fernandez-Miranda et al., 2010). The Smc5/6 complex components Nse1 and Nse2 are E3 ubiquitin and SUMO ligases, respectively (reviewed in Verver et al., 2016). Our study allows us to speculate that the Smc5/6 complex might be required for the ubiquitin or SUMO modification of Aurora B, which ensures timely removal of Aurora B from chromosome arms. Interestingly, in budding yeast, it has been shown that Nse2 is required for normal SUMOylation levels of Bir1 (also known as survivin and Birc5), which is a direct interaction partner of Aurora B (Yong-Gonzales et al., 2012). However, direct SUMO modification by Nse2 might not be essential, as mice expressing an Nse2 mutant allele deficient for SUMOylation activity did not show any phenotypic aberration (Jacome et al., 2015).

During DNA replication, sister chromatids become topologically intertwined and these catenane structures must be resolved by Topo IIα to permit chromosome segregation (Liu et al., 2014). Mutation of Smc5 in mESCs resulted in chromosome segregation defects reminiscent of a defect in Topo IIα function. Similar to our observations, Smc5/6 depletion in human RPE-1 cells results in the presence of DNA bridges and lagging chromosomes during mitosis (Gallego-Paez et al., 2014). This phenotype was attributed to impaired localization of Topo IIα and condensin to pericentromeric regions. Although we did not observe mislocalization of Topo IIα in our Smc5 mutant mESCs, we did so for condensin, and this might be the direct cause of chromosome missegregation. In budding yeast, it has been shown that condensin is required to facilitate Topo-IIα-mediated resolution of catenanes prior to the metaphase-to-anaphase transition (Charbin et al., 2014). Our results suggest that the Smc5/6 complex is required for stable localization of condensin to the pericentromeric regions, and in its absence it is conceivable that Topo IIα cannot efficiently stimulate sister chromatid decatenation. Alternatively, from work using DT40 cells and budding yeast, it has been reported that condensin and Topo IIα demonstrate opposing activities, where condensin promotes DNA compaction and Topo IIα promotes DNA relaxation (Samejima et al., 2012; Leonard et al., 2015). Smc5/6 might be required to mediate the balance between condensin and Topo IIα during mitosis, which is essential for the efficient resolution of intertwined sister chromatids.

Smc5/6 components primarily localize to the pericentromeric heterochromatin, but also along chromosome arms in mESCs. This is consistent with what was previously published using MEFs (Gomez et al., 2013). These observations are in contrast to what has been reported using RPE-1 cells, where Smc5 and Smc6 are enriched within the nucleus during interphase, but then greatly diminished from the chromatin during mitosis (Gallego-Paez et al., 2014). However, some Smc5 and Smc6 is still present in the chromatin fraction during mitosis in RPE-1 cells, and the immunocytochemistry with anti-GFP antibodies demonstrates strong chromatin signal in mitotic HeLa cells expressing EGFP–Smc5 used in the same study. Therefore, the differences might be easily reconciled with further detailed assessments.

Budding yeast Smc5 binds directly to microtubules and mutation of the microtubule-binding region of Smc5 results in compromised microtubule stability (Laflamme et al., 2014). We report that Smc5/6 components localize to the spindle poles, and depletion of Smc5 results in the presence of abnormal mitotic spindles. It is possible that the Smc5/6 complex is involved in the regulation of centrosome function, and required for stabilizing microtubules that promote normal mitotic spindle formation.

Our study has demonstrated that the Smc5/6 complex is crucial for mESC maintenance and cell cycle progression. Destabilization of the Smc5/6 complex in mESCs affects multiple aspects of the cell cycle and leads to mitotic catastrophe. We have shown that Smc5/6 is required for regular distribution and functioning of condensin and mitotic kinases Plk1 and Aurora B. We evaluated condensin localization by visualizing Smc4, a common subunit of condensin I and II, which prompts future investigation of each condensin complex. Our study provides the platform for further detailed investigation of Smc5/6 functions in stem cell maintenance, DNA replication and mitosis. Collectively, this will lead to a better understanding of stem cell genomic instability and the developmental abnormalities caused by Smc5/6 perturbation.

Mice were bred by the investigators at The Jackson Laboratory (JAX, Bar Harbor, ME) and Johns Hopkins University (JHU, Baltimore, MD) in accordance with criteria of the NIH and USDA. All animal procedures were conducted with approval from the IACUC of JAX and JHU.

mESC clone EPD0395_1_F05 (C57BL/6N-A/a genetic background) bearing a ‘knock-out first’ allele of Smc5 (Smc5^{tm1a(KOMP)Wtsi}) were acquired from the Knockout Mouse Project (http://www.mousephenotype.org/data/alleles/MGI:2385088/tm1a%28KOMP%29Wtsi).

Chimeras were obtained by microinjection of EPD0395_1_F05 mESCs into C57BL/6J blastocyst-stage mouse embryos and were assessed for germline transmission. Heterozygous progeny were bred with a C57BL/6J Flp recombinase deleter strain (B6.129S4-Gt(ROSA)26Sor^tm1(FLP1)Dym/RainJ, JAX) to remove the SA-LacZ and Neo selection cassette and produce the floxed exon 4 (designated Smc5^flox).

To produce offspring heterozygous for the deleted exon 4 (designated Smc5^del), heterozygous Smc5^flox males were mated to Sox2-Cre C57BL/6J (B6.Cg-Tg(Sox2-cre)1Amc/J, JAX) mice. Then, heterozygous Smc5^del mice were bred to mice harboring the conditional Cre-ER^T2 (B6.129-Gt(ROSA)26Sor^{tm1(cre/ERT2)Tyj}/J, JAX), which resulted in progeny heterozygous for the Smc5^del allele and hemizygous for the Cre-ER^T2 genotype. These mice were bred to homozygous Smc5^flox mice to derive experimental (Smc5^flox/del, Cre-ER^T2) and control (Smc5^wt/flox, Cre-ER^T2) genotypes.

For mESC derivation and culture we used 2i medium described by Ying et al. (2008) and Hanna et al. (2010). Briefly, mESC culture medium included 1:1 mixture of DMEM/F12 and neurobasal medium with 1% N2 and 2% B27 supplements, 1 mM L-Glutamine, 1% MEM NEAA (Invitrogen), 50 µM β-mercaptoethanol (Sigma), 50 µg/ml BSA (Sigma), 10 ng/ml human leukemia inhibitory factor (Cell Signaling), 1 µM MEK inhibitor PD 0325901 and 3 µM GSK-3 inhibitor CHIR 99021 (Tocris). mESC lines were established from whole embryos (Behringer et al., 2014). After 5–6 passages on primary MEF feeder, mESC lines were switched to a feeder-free culture on 0.2% gelatin and passaged every 3 days with a seeding density 5×10³/cm². mESCs were analyzed within 20 passages of culture.

Primary MEFs were derived from F1 FVBxC57BL/6J embryos at 13.5 days post coitum (dpc) (Behringer et al., 2014). MEF culture medium consisted of 89% DMEM (Invitrogen), 10% fetal bovien serum (FBS; Hyclone), 1% penicillin and streptomycin (100 U/100 µg). For feeder preparation, γ-irradiated MEFs (20 Gy) were plated into six-well plates precoated with 0.1% gelatin at density 2×10⁴/cm². Feeder was used for mESC culture on the next day after cell plating. The standard NIH 3T3 protocol was used for establishing immortal MEF cell lines.

For embryoid body formation, mESC aggregates were grown in a mixture of differentiation medium [15% FBS (HyClone) and 85% DMEM/F12 (Invitrogen)] and mESC culture medium at a ratio of 2:1. After 3 days, medium was changed to differentiation medium only. Medium was changed every 3 days. Embryoid bodies were collected for western blot analysis on day 0, 6, 12 and 18 of differentiation.

mESCs were mechanically collected from tissue culture plastic, washed once in PBS and resuspended in 80% of DMEM/F12 and 20% Geltrex (Invitrogen). mESCs (2×10⁶–4×10⁶) were injected subcutaneously into right hind limb of the NOD/SCID mouse (The Jackson Laboratory). Teratomas were collected after 4 weeks with a size of 1–2 cm³. Isolated teratomas were fixed in 10% formalin, embedded in paraffin, sectioned onto glass microscope slides with a 5-µm thickness and stained with hematoxylin and eosin.

Cre recombinase activity was induced by 4-hydroxytamoxifen (4-OH TAM) (Sigma) at a dose of 0.2 µM, which was added into cell culture medium for 3 days (mESCs) or 10 days (MEFs). After 3 or 10 days of treatment the dose of 4-OH TAM was decreased to 0.05 µM and kept at this level until the end of experiments. Fresh 4-OH TAM was added daily into mESC culture and every 2 days for MEFs. Excision of intron 4 was confirmed by PCR analysis on days 1–3 of treatment in mESCs, and on day 5 and day 10 of treatment for MEFs. For further information see the Results section and Fig. S2A.

Cells were expanded in large quantity and simultaneously collected for DNA, RNA, western blotting and cell cycle analyses. Cells were washed in PBS, counted and snap-frozen in aliquots for further evaluation or were processed immediately. Cell viability was determined based on Trypan Blue die incorporation using a Bio-Rad TC20 cell counter.

DNA was extracted using the GeneJet Genomic DNA purification kit (Thermo Scientific) and 20 ng was used for each PCR. PCRs were set with HotMaster Taq DNA polymerase (5 Prime). The floxed allele (563 bp) and the wild-type allele (410 bp) were amplified with primers 1 (forward 5′-ACTCAGTCTCACACGGCAAG-3′) and 2 (reverse 5′-ATCCTTCCCACCTTGGAAAC-3′), the floxed allele (644 bp) was amplified with primers 3 (forward 5′-AGAAAGACATCAAACTAACCGCTGGC-3′) and 4 (reverse 5′-GAGATGGCGCAACGCAATTAAT-3′). The deletion allele was amplified as a 763-bp product with primers 1 and 4. PCR reaction conditions were: initial denaturation at 94°C for 2 min, (denaturation at 94°C for 20 s, annealing at 58°C for 30 s, amplification at 72°C for 1 min)×30 cycles, and final extension at 72°C for 10 min.

mESCs (untreated and treated with 4-OH TAM for 5 days) were mitotically arrested by incubating with 0.1 µg/ml KaryoMax colcemid solution (Invitrogen) for 2 h. Cells were collected, placed into hypotonic solution (0.075 M KCl) at 37°C for 20 min and fixed in methanol:acetic acid (3:1). Fixed cells were dropped onto wet, ice-cold glass micro slides (Cardinal Health) and stained with KaryoMax Giemsa stain solution following the manufacturer’s instructions (Invitrogen).

Cell lysates were prepared in RIPA buffer (Santa Cruz Biotechnology) supplemented with protease inhibitor cocktail (Roche). Equal amounts of proteins were fractionated by SDS-PAGE and transferred to PVDF membrane (Bio-Rad). Primary antibodies are provided in Table S1. We used horseradish peroxidase (HRP)-conjugated goat anti-mouse-IgG and anti-rabbit-IgG secondary antibodies (Invitrogen). Signal was detected using Clarity Western ECL Substrate (Bio-Rad) and imaged using Syngene XR5 system.

To synchronize mESCs (untreated and 4-OH TAM-treated for 5 days) in the G2/M phase of cell cycle, nocodazole at a concentration of 50 ng/ml was added to cell culture medium for 6 h. For cell synchronization in G1 phase, following nocodazole treatment, thymidine was used at a concentration 2 mM for the next 4 h. Cells were collected for cell cycle analysis at indicated time points.

Approximately 1×10⁶–2×10⁶ mESCs were collected, washed in PBS and resuspended in 1 ml of PBS. Cells were fixed by adding 3 ml cold (−20°C) absolute ethanol to cell suspension while vortexing and were kept overnight at 4°C. For cell cycle analysis, cells washed in PBS were resuspended in 1 ml of staining solution (0.1% Triton X-100 in PBS, 0.2 mg/ml RNase A and 20 μg/ml propidium iodide) and kept overnight at 4°C in the dark. DNA content was determined by flow cytometry using BD FACS Calibur and Cell Quest software. Total populations were gated to remove cell debris and doublets. Data were analyzed using FlowJo software.

Total RNA was isolated using PureLink RNA kit (Ambion) followed by DNase I (RNase-free, Ambion) treatment. First strand cDNA synthesis was carried out using the Enhanced Avian HS RT-PCR kit (Sigma). PCR amplification of cDNA samples was set with HotMaster Taq DNA polymerase (5 Prime). Primers are listed in Table S2. PCR reaction conditions were: initial denaturation at 94°C for 2 min, (denaturation at 94°C for 20 s, annealing at 55°C for 10 s, amplification at 70°C for 30 s)×28 cycles, and final extension at 70°C for 10 min.

For pluripotency marker analysis, mESCs were plated on glass coverslips (Electron Microscopy Sciences) precoated with 0.1% gelatin and containing MEF feeder. mESCs were fixed with 10% formalin for 20 min, washed in PBS and used for immunostaining. For single-cell analysis, mESCs were fixed in suspension and ∼50,000–100,000 cells were spun onto glass micro slides (Cardinal Health) in Shandon Cytospin 4 centrifuge (200 g, 3 min). For chromosome spread preparation, mESCs were treated with 0.1 µg/ml KaryoMax colcemid solution (Invitrogen) for 1.5 h. Cells were collected, placed into hypotonic solution (PBS:water, 4:6 mix) for 5 min at room temperature and dropped onto ice-cold glass slides. Chromosome spreads were fixed and permeabilized in 10% formalin with 0.5% Triton X-100 for 15 min. Immunocytochemistry was performed as described previously (Pryzhkova et al., 2014). Primary antibodies used are listed in Table S1. Secondary Alexa-Fluor-conjugated (488, 568 or 633) antibodies were goat anti-human-IgG, anti-mouse-IgG and anti-rabbit-IgG antibodies (Invitrogen). To detect cell DNA, we used VectaShield mounting medium with DAPI (Vector Laboratories). The detection of alkaline phosphatase activity was carried out using Vector Red Alkaline Phosphatase Substrate (Vector Laboratories).

Samples were analyzed using either a Zeiss AxioImager A2 or Cell Observer Z1 fluorescent microscopes and captured using AxioCam ERc 5 s (Zeiss) or a ORCA-Flash 4.0 CMOS camera (Hamamatsu), respectively. Images were analyzed and processed using ZEN 2012 blue edition imaging software (Zeiss). Photoshop (Adobe) was used to prepare figure images.

We thank Mike Matunis for critical comments on the manuscript. We thank Laura Morsberger and Raluca Yonescu from the JHU Cytogenetics Core Facility for guidance with karyotype analysis.