The Arabidopsis homolog of Scc4/MAU2 is essential for embryogenesis

The establishment of sister chromatid cohesion and its controlled release is necessary for the proper segregation of chromosomes at mitosis and meiosis (Moschou and Bozhkov, 2012). Sister chromatids are bound and held together by multi-protein subunit complexes called cohesins (Nasmyth, 2001). Mutations in genes encoding cohesin subunits and their regulatory proteins are found in different types of cancer and in cohesinopathies, developmental disorders such as Cornelia de Lange syndrome, Roberts syndrome and the age-related increased occurrence of trisomy 21 (Barbero, 2013).

The cohesin complex belongs to the class of structural maintenance of chromosome (SMC) complexes (Cobbe and Heck, 2004) and comprises four highly conserved core subunits: an SMC1 subunit, an SMC3 subunit, a stromalin subunit (e.g. Scc3; also known as STAG2) and a kleisin subunit [e.g. Scc1 (also known as Rad21) and Rec8] (Fig. 1A). Components of the cohesin complex are often represented by several homologs, and the function of cohesin complex is at least partially defined by the combination of the different subunit homologs it comprises. Most of the studied organisms have at least two kleisin subunit orthologs, functionally separated into those predominantly involved in mitosis or meiosis (Schleiffer et al., 2003). The Arabidopsis genome encodes four kleisin subunits (Fig. 1A): the meiosis-specific SYN1 (Cai et al., 2003) and SYN3, involved in gene expression of meiotic genes, but also expressed in somatic cells (Yuan et al., 2012; Jiang et al., 2007), and SYN2 and SYN4, which have been suggested to participate in mitotic cell division (da Costa-Nunes et al., 2006).

In non-plant species, e.g. Saccharomyces cerevisiae, Saccharomyces pombe, Homo sapiens, Xenopus, Drosophila melanogaster and Coprinopsis cinerea, the cohesin ring is loaded on chromatin and locked around DNA by the Scc2–Scc4 protein complex (Scc2 is also known as the NIBPL, and Scc4 as MAU2) in an ATP-dependent manner (Ciosk et al., 2000; Dorsett, 2004; Watrin et al., 2006; Arumugam et al., 2003). The cohesin-loading complex defines both the timing and initial location of cohesin rings on chromatin (Ciosk et al., 2000). Scc2 physically interacts with cohesin and its C-terminus is sufficient to entrap DNA in a cohesin ring in vitro (Chao et al., 2015).

The Scc4 subunit of the cohesin-loading complex contains multiple tetratricopeptide repeats (TPRs) forming a barrel-like superhelix essential for interaction with the N-terminus of Scc2 (Chao et al., 2015; Hinshaw et al., 2015). While Scc2 is required for closing the cohesin ring around DNA, Scc4 is necessary for delivering the complex to DNA in vivo (Chao et al., 2015; Hinshaw et al., 2015). A highly conserved region on the surface of Scc4 is required for targeting the Scc2–Scc4 complex to centromeric regions of chromosomes (Hinshaw et al., 2015).

Studies in yeast and animal systems have established cohesin as an essential actor in virtually all aspects of chromosome biology, including chromosome segregation, maintenance of genome stability, regulation of gene expression via defining chromatin structure, 3D genome organization (Watrin et al., 2016) and binding of enhancer RNAs (eRNAs) (Li et al., 2013). Loss-of-function mutants of Arabidopsis thaliana (At)SCC2 show phenotypes similar to mutants of the cohesin complex subunits, e.g. Atsmc1 and Atsmc3 (also known as titan8 and titan7, respectively) (Tzafrir et al., 2002; Liu et al., 2002). All these mutations cause embryonic lethality and formation of giant endosperm nuclei, indicating that these genes are indispensable for plant growth and development (Sebastian et al., 2009; Liu et al., 2002). In addition, transient induction of AtSCC2 silencing causes aberrations in meiosis, culminating in pollen abortion (Sebastian et al., 2009).

Here, we identify and characterize AtSCC4, the Arabidopsis ortholog of Scc4. We show that it physically interacts with the N-terminal domain of AtSCC2 and is immobilized within interphase nuclei in an AtSCC2-independent manner. We demonstrate that each subunit of AtSCC2–AtSCC4 complex is indispensable for plant embryogenesis and for controlling cell fate at the early stages of embryogenesis, but differ in their impact on seed endosperm development. Finally, we establish a novel in vivo assay for tracking cohesin-loading dynamics in Arabidopsis embryos, which allowed us to reveal that mitotic cohesin re-colocalizes to chromatin during late telophase and that AtSCC4 is required for this process.

A BLASTP search for Scc4 homologs in Arabidopsis using the Homo sapiens Scc4 (HsMau2) as a query revealed a single gene, At5g51340 (E-value=7×10⁻¹⁰). The At5g51340 encodes a putative protein of 726 amino acids with 78% and 24% of HsMau2 query coverage and identity, respectively. Protein sequence analysis revealed that the two typical tetratricopeptide repeats (TPR) that are present in Scc4 orthologs are arranged in tandem between amino acids 525 and 600, within a predicted structural motif TPR 12 (pfam13424, Marchler-Bauer et al., 2015) (Fig. 1B).

Searches for potential proteins that interact with At5g51340 using the STRING database (Szklarczyk et al., 2015) (Fig. 1C) predicted that it might form complexes with plant cohesin subunits and the cohesin-regulating proteins, thus supporting the hypothesis that At5g51340 (hereafter referred to as AtSCC4) is involved in sister chromatid cohesion.

A phylogenetic analysis of Scc4 orthologs revealed that plant Scc4 orthologs form a separate clade (Fig. 1D; Fig. S5). Scc4 orthologs demonstrated a high evolutionary divergence of their primary structure (Fig. S5) and significant similarity of their secondary structures. Conservation of tertiary structure might explain how highly different Scc4 orthologs play similar roles in loading of cohesin onto chromatin.

Despite having a conserved function in cell division, Scc4 family members show quite high divergence even within the same kingdom (Fig. 1D), suggesting that Scc4 might have also acquired specific functions in each lineage.

Analysis of the GENEVESTIGATOR microarray database revealed that AtSCC4 mRNA levels do not correlate with the mRNA levels of AtSCC2 and those of some of the kleisins (Fig. S1A). To further gain insight into the tissue-specific expression pattern of AtSCC4, we generated transgenic Arabidopsis lines expressing β-glucuronidase (GUS) under control of the AtSCC4 promoter (a 2 kb region upstream of the AtSCC4 start codon). Interestingly, we could detect GUS expression in all tissues examined, both in meristematic and non-meristematic cells (Fig. S1B).

We established transgenic Arabidopsis lines expressing an AtSCC4–GFP fusion protein under the control of the AtSCC4 native promoter to assess the turnover rate of AtSCC4 and further verify the presence of the AtSCC4 in different cell types (Fig. 2A). Expression of the full-length fusion protein in transgenic lines was confirmed by immunodetecting GFP (Fig. S1C). AtSCC4 was consistently observed in the nuclei and cytoplasm of differentiated and meristematic cells (Fig. 2A), indicating that, similarly to other Scc4 orthologs, AtSCC4 protein might act not only during cell division but also in the interphase (Peters et al., 2008; Ball et al., 2014).

In cells at the interphase and preprophase stages, most of the AtSCC4–GFP localized in the nucleus, but not in the nucleolus, and a weak signal was also detected in the cytoplasm (Fig. 2A,B), consistent with observations made in non-plant organisms, e.g. Homo sapiens, Drosophila melanogaster and Caenorhabditis elegans (Seitan et al., 2006; Benard et al., 2004). During prometaphase, metaphase, anaphase and telophase, the GFP signal was cytoplasmic (Fig. 2B) indicating that AtSCC4 does not decorate chromatin during cell division, when it would be expected that cohesin would be removed from chromatids by the wings-apart-like (WAPL) and separase pathways (Buheitel and Stemmann, 2013; Gerlich et al., 2006; Moschou and Bozhkov, 2012). Shortly after completion of cell division, AtSCC4–GFP regained its nuclear localization (Fig. 2B). Split-nuclei inverse fluorescence recovery after photobleaching (iFRAP) (Fig. 2C,D) analyses showed that a significant portion of AtSCC4 is immobilized in the interphase nuclei, most probably by its association with chromatin (Fig. 2D,E). This result suggests the possible involvement of the plant cohesin-loading complex in sculpting chromatin of interphase nuclei (Lopez-Serra et al., 2014).

We obtained two Atscc4 transfer (T)-DNA insertional alleles from the SAIL and SALK mutant collections (Alonso, 2003). The corresponding alleles were designated Atscc4-1 and Atscc4-2; the lines were shown to carry palindrome insertions in the 7th and 8th exons, respectively (Fig. S2A,B). Plants heterozygous for Atscc4-1 and Atscc4-2 showed no obvious growth phenotypes and had a similar AtSCC4 mRNA level to that in wild-type (Fig. S2C).

We were unable to identify homozygous mutants among the progeny of the heterozygous plants. The Arabidopsis genome is diploid and the 1:2:0 (wild-type:heterozygous:homozygous) segregation pattern for the Atscc4 T-DNA insertions suggested that disruption of AtSCC4 could lead to embryo lethality (Fig. S2D). We examined siliques from Atscc4-1/AtSCC4 and Atscc4-2/AtSCC4 plants, and found that ∼25% of the seeds were aborted in both mutant backgrounds (Fig. 3; Fig. S3A). Interestingly, we did not detect abnormalities in the transmission of T-DNA when pollen or ovules of mutants were used for crossing with ovules or pollen of wild-type (Fig. S2D). Lack of the phenotype in Atscc4 gametes might be a consequence of carryover of AtSCC4 protein or mRNA from the parental tissues.

We analyzed seed abortion rate in the complemented Atscc4-1/AtSCC4 and Atscc4-2/AtSCC4 lines expressing AtSCC4–GFP under the control of the AtSCC4 native promoter (Atscc4-1/AtSCC4 pAtSCC4::AtSCC4 and Atscc4-2/AtSCC4 pAtSCC4::AtSCC4, respectively) and determined that in the T1 generation of the complemented lines seed abortion decreased to the expected 6.25% (Fig. 3; Fig. S3A). Furthermore, in the T3 generation we were able to detect four viable homozygous-knockout Atscc4-2 pAtSCC4::AtSCC4 plants in three independent complemented lines (Fig. S3B). Collectively, our results demonstrate that the observed seed abortion phenotype was caused by disruption of AtSCC4.

To further investigate the cause of the seed abortion phenotype in Atscc4/AtSCC4 mutants, we tracked phenotypes of Atscc4/Atscc4, Atscc4/AtSCC4 and wild-type embryos throughout successive stages of embryogenesis. Zygotes of all backgrounds were normal, and the first cell division produced small apical and elongated basal cells (Fig. 4Aa,i). At the octant stage of embryo development, we observed the first aberrations in mutant embryos when cell division in the embryo proper became unsynchronized resulting in the disruption of the typical bilateral symmetry (Fig. 4Al). From the dermatogen stage onwards, ∼25% of the embryos in siliques from Atscc4-1/AtSCC4 and Atscc4-2/AtSCC4 exhibited cell division defects in the embryo proper (Fig. 4Am–o), and at the heart stage, knockout embryos began to deteriorate (Fig. 4Ap). Following the octant stage, cell division defects were also observed in the suspensor of mutant embryos, wherein supernumerary cells were formed (Fig. 4Am–o), sometimes leading to a raspberry-like phenotype (Yadegari et al., 1994). These phenotypes were very similar to the embryo development aberrations associated with depletion of another subunit of the cohesin complex, AtSCC2 (Sebastian et al., 2009). Interestingly, in contrast to AtSMC1, AtSMC3 or AtSCC2 loss-of-function mutants, AtSCC4 deficiency did not cause severe developmental aberrations in the endosperm (Fig. 4B) (Tzafrir et al., 2002; Liu et al., 2002). This indicates that the AtSCC2–AtSCC4 complex might play different roles in embryo and endosperm development, and that there might be additional cell division-related functions specific for each subunit.

The suspensor is a temporary structure of the plant embryo that serves as a conduit of nutrients and hormones to sustain the embryo proper (Yeung and Meinke, 1993; Friml et al., 2003). Although suspensor cells in Arabidopsis have embryogenic potential, they are destined to die during embryo maturation (Bozhkov et al., 2005). It has been suggested that the embryogenic potential of the suspensor cells is suppressed by the embryo proper and that the plant hormone auxin plays an important role in this process. Indeed, redistribution of auxin in the suspensor cells of Arabidopsis after laser ablation of the embryo proper precedes re-initiation of cell proliferation (Liu et al., 2015; Gooh et al., 2015).

To examine whether the increased embryogenic potential of suspensor cells in cohesin-loading complex mutants is associated with the redistribution of auxin response maxima, we introduced the auxin-response maxima reporter DR5rev::3xVENUS-N7 (Heisler et al., 2005) into Atscc4-1/AtSCC4 and Atscc2-2/AtSCC2 backgrounds. While in the wild-type-like embryos the reporter was mostly observed in the upper part of the suspensor, Atscc4-1 embryos displayed irregular distribution of the reporter (Fig. 5). Before the globular stage, the reporter was detectable in all cells of mutant embryos, with an atypically high intensity in the embryo proper. This abnormal pattern of auxin-response maxima coincided with the loss of cell division synchrony in the embryo proper and subsequently with the loss of bilateral symmetry and retardation of cell proliferation (Figs 4A and 5). From the globular stage onward, when cells of the Atscc4-1 embryo proper showed signs of degradation, the auxin-response reporter was mostly visible in the suspensor cells, with the highest intensity observed in the basal cells (Fig. 5). This inverse distribution of auxin response maxima in the mutant embryos preceded the onset of ectopic cell division in the suspensors (Figs. 4A and 5). Our observations strengthen the notion that the embryo proper of Arabidopsis produces an inhibitory signal blocking the embryogenic potential of the suspensor. Since similar patterns in the spatial distribution of auxin response maxima were observed in the Atscc2-2 embryos (Fig. 5), we assumed that AtSCC2 and AtSCC4 might act in the same pathway during embryonic pattern formation.

In yeast and animals, Scc2 physically interacts with Scc4 (Lopez-Serra et al., 2014). To investigate whether a similar complex is formed in plants, we first employed a yeast two-hybrid assay. We found that AtSCC4 associates with the N-terminus of AtSCC2 (Fig. 6A). To confirm this interaction in plants, we co-immunoprecipitated Myc-tagged AtSCC4 using a CFP-tagged AtSCC2 as a bait. For this experiment, both proteins were transiently co-expressed under the constitutive 35S CaMV (Cauliflower mosaic virus) promoter in Nicotiana benthamiana (Fig. 6B).

Next, we examined the genetic interaction between AtSCC2 and AtSCC4 by crossing heterozygous plants carrying an Atscc4-1 or Atscc4-2 mutant allele with heterozygous plants carrying an Atscc2-2 or Atscc2-3 mutant allele (Sebastian et al., 2009). We estimated the seed abortion rate in 200 siliques from 20 individual plants of the F1 generation. Seed abortion in siliques from Atscc2/AtSCC2;Atscc4/AtSCC4 followed the predicted 9:7 segregation (number of viable seeds:number of aborted seeds). (The segregation rate was predicted based on the following assumptions: only one copy of the complementation insert was present in the diploid genome of the T1 generation of the complemented lines; the complementation insert was inherited independently from the corresponding Atscc4 T-DNA insertion; only seeds containing embryos homozygous for the Atscc4 T-DNA insertion would be aborted; and the presence of the complementation insert would restore viability of the embryos homozygous for the Atscc4 T-DNA insertion) (Fig. S2E). We next assessed embryo development in the double Atscc2/AtSCC2; Atscc4/AtSCC4 mutants by examining ∼600 embryos. We did not observe an additive effect of the double mutation on the phenotype of early embryos when compared to the single Atscc2/AtSCC2 or Atscc4/AtSCC4 mutants. Collectively, these results confirm that AtSCC2 and AtSCC4 act in the same pathway during plant embryogenesis.

Yeast Scc2 interacts with the cohesin ring and facilitates its loading on DNA in vitro in the absence of Scc4 (Chao et al., 2015), while Scc4 is required for loading cohesin rings at specific sites on chromosomes (Hinshaw et al., 2015). Immobilization of AtSCC4 in the nuclei is most likely a result of its association with chromatin and is indicative that AtSCC4 might serve as an adapter between the AtSCC2–cohesin complex and the sites of cohesin loading on chromatin. We investigated whether the immobilization of AtSCC4 was dependent on AtSCC2 by performing the split nuclei iFRAP assay with AtSCC4–GFP expressed under AtSCC4 promoter in Atscc2-2 embryos (Fig. 7). The results revealed no statistically significant difference in the proportion of AtSCC4 immobilized on chromatin in wild-type or AtSCC2-deficient backgrounds, indicating that chromatin-mediated immobilization of AtSCC4 in the nuclei is independent of AtSCC2.

In non-plant model organisms, e.g. Saccharomyces cerevisiae, Saccharomyces pombe, Homo sapiens, Xenopus, Drosophila melanogaster and Coprinopsis cinerea, the cohesin ring is loaded on chromatin by the Scc2–Scc4 complex (Fernius et al., 2013; Dorsett, 2004). Arabidopsis cohesin loading complex subunit AtSCC2 has been reported to control localization of cohesin during meiotic cell division, which was demonstrated using immunostaining of fixed plant material (Sebastian et al., 2009).

To examine the role of AtSCC4 in loading cohesin in vivo, we established a dedicated live-cell imaging assay. Constitutive expression of cohesin subunits using 35S promoter was shown to perturb Arabidopsis development (Yuan et al., 2014). To avoid this effect, we used the ABA INSENSITIVE 3 (ABI3) promoter, which is relatively weak and, in seeds, mostly embryo specific (Devic et al., 1996; Ng et al., 2004), to drive expression of TagRFP fusions of the four known Arabidopsis kleisin subunits, RAD21.1/SYN2, RAD21.1.2/SYN3, RAD21.3/SYN4 and Rec8/SYN1 (da Costa-Nunes et al., 2006; Dong et al., 2001). The obtained constructs were introduced into wild-type and Atscc4-1/AtSCC4 backgrounds. Notably, the T1 generation of all but the SYN4-expressing plants were highly susceptible to even mild environmental changes, and displayed decreased fertility phenotypes, e.g. development of short siliques with few seeds (Fig. S4). Interestingly, the aberrant phenotypes seemed to be dependent on the condition, probably following induction of ABI3 promoter in response to short-term drought caused by strong ventilation in one of our growth chambers. SYN4–TagRFP-expressing plants did not display any noticeable phenotype in T1 and T2 generations and were used for further experiments. Arabidopsis plants of corresponding backgrounds expressing free TagRFP under the control of the ABI3 promoter were used as a control.

TagRFP–SYN4 localized predominantly in the nuclei of embryonic cells, while free TagRFP was detected in both nuclei and cytoplasm (Fig. 8A). In accordance to the predicted activity of the ABI3 promoter (Ng et al., 2004; supported also by the eFP browser and GENEVESTIGATOR), we observed a progressive increase of TagRFP-SYN4 or free TagRFP expression during embryogenesis, starting from the octant stage (Fig. 8B). In dividing cells, TagRFP–SYN4 was excluded from chromatin during metaphase and anaphase and colocalized with it again at late telophase (Fig. 8C). Surprisingly, we could detect a pool of TagRFP–SYN4 residing between separating chromosomes during anaphase (Fig. 8C). Under our experimental conditions, we failed to find embryonic cells in prometaphase to verify whether localization of SYN4 also correlates with the distribution of AtSCC4 at this stage. Collectively, our results suggest that, in plants, cohesin is loaded on chromatin during the G1 phase and is already removed from chromatin by metaphase.

Previous experiments performed on normal rat kidney (NRK) cells indicated that cohesin diffuses throughout the nucleus and cytosol and binds to unreplicated DNA with a residence time in the range of minutes (Gerlich et al., 2006). This ‘dynamic binding mode’ is thought to be the result of continuous loading of cohesin onto chromatin and subsequent release by WAPL (Buheitel and Stemmann, 2013).

To analyze the mobility of plant cohesin, we performed iFRAP in interphase nuclei. The method was verified using wild-type plants expressing free TagRFP or TagRFP–SYN4 under the control of the ABI3 promoter (Fig. 8D). We demonstrated that, in the wild-type background, TagRFP showed high motility while TagRFP–SYN4 was immobilized in the nuclei (Fig. 8D), thus confirming that iFRAP is an efficient method to study SYN4 motility.

Next, we assessed the efficacy of cohesin immobilization in nuclei of AtSCC4 knockouts. We performed iFRAP analyses at the octant to globular stages of embryogenesis, when Atscc4/Atscc4 embryos are still viable. In the wild-type background only a small fraction (∼10%) of TagRFP–SYN4 signal was lost in the region of interest (ROI), suggesting strong immobilization of TagRFP–SYN4 (Fig. 8E,F). By contrast, in Atscc4/Atscc4 embryos, TagRFP–SYN4 was mostly mobile (Fig. 8E,F). These results strongly indicate the requirement of AtSCC4 protein for immobilizing plant cohesin in nuclei, most probably by anchoring it onto chromatin.

We demonstrate that, as with some other Scc4 orthologs (Seitan et al., 2006), most of the AtSCC4 localizes in the nucleus and is excluded from the nucleolus, while a small amount of the protein can be detected in the cytoplasm (Fig. 2A,B). Furthermore, a large fraction of the AtSCC4 in interphase nuclei is immobile, most probably due to its association with chromatin.

Our genetic experiments confirm that AtSCC2 and AtSCC4 act in the same pathway, suggesting that plant Scc2–Scc4 is a functional complex. Similar to with loss-of-function mutations of some other cohesin-regulating proteins [e.g. AtSCC2 (Sebastian et al., 2009), separase AtESP (Liu and Makaroff, 2006) and either cohesin subunits, AtSMC1 and AtSMC3 (Tzafrir et al., 2002; Liu et al., 2002)], knockout of AtSCC4 is embryonic lethal (Fig. 3; Figs S2D,E and S3A), thus confirming the crucial role of AtSCC4 in regulating cohesin. Loss-of-function mutants depleted of either cohesin or of cohesin-regulating proteins exhibit distinct phenotypes. Embryos of AtSMC1 or AtSMC3 mutants (ttn mutants) arrest their development at very early stages, before the number of cells in the embryo exceeds five and cells start to enlarge (Liu et al., 2002). By contrast, embryos deficient in AtSCC2, AtSCC4 or AtESP, can undergo multiple cell divisions prior to arrest (Liu and Makaroff, 2006; Yang et al., 2009). Efficient regulation of plant DNA packaging might require many more cohesin molecules, as compared to cohesin-regulating proteins. In such case, cohesin-regulating proteins carried over from heterozygous diploid precursors of gametes might be sufficient to sustain viability of a corresponding knockout embryo through multiple rounds of cell division. Alternatively, the stronger phenotype of SMC depletion might be due to additional important functions executed by SMC proteins, e.g. establishment of the microtubule arrays in the spindle and phragmoplast (Tzafrir et al., 2002). Interestingly, in contrast to AtSCC2, AtSMC1 and AtSMC3 loss-of-function mutants, seeds of an AtSCC4-deficient background did not show severe abnormalities of endosperm development (Fig. 4B). This indicates that the two subunits of the AtSCC2/4 complex might play different roles in endosperm and development of the embryo proper. Alternatively, milder phenotypes of the AtSCC4 loss-of-function mutants might be explained by the presence of AtSCC4 protein carried over from the parental tissues through gametes.

The embryo phenotype observed in the AtSCC4-depleted embryos is similar to the previously reported phenotype of AtSCC2 knockouts (Sebastian et al., 2009). With a low frequency, embryos of both mutants survived to develop into raspberry-like embryos. Raspberry-like embryos first display loss of symmetry and synchrony of cell divisions in the embryo proper, followed by developmental arrest (Yadegari et al., 1994). Instead of proliferating, cells of the embryo proper begin to expand, vacuolate and eventually die, mimicking the normal fate of suspensor cells. While suspensor cells acquire an embryo-proper-like fate and begin to divide. The mechanism underlying this shift of cell fate is still unknown. Laser ablation experiments have shown that removal of the embryo proper results in the induction of an embryo-proper-like pathway in the terminally differentiated suspensor cells, suggesting that the embryo proper produces a signal suppressing the embryogenic potential of suspensor cells (Liu et al., 2015; Gooh et al., 2015). This developmental transition is associated with the redistribution of auxin. Here, we provide genetic evidence that auxin redistribution precludes ectopic cell proliferation in the suspensor and the degradation of the embryo proper (Fig. 5).

In yeast and animal model organisms, Scc4 forms a stable complex with Scc2 (Ciosk et al., 2000; Watrin et al., 2006). The interacting interface comprises Scc4 TPR repeats, tandemly arranged short helices forming a superhelix that engulfs the unstructured N-terminus of Scc2 (Chao et al., 2015; Hinshaw et al., 2015). We demonstrate that AtSCC4 interacts with the N-terminus of AtSCC2 in yeast and in plants (Fig. 6), but this interaction is not necessary for immobilization of AtSCC4 in the nuclei (Fig. 7). This result indicates that, similar to for yeast Scc4 (Hinshaw et al., 2015), AtSCC4 might be determining sites for cohesin loading by recognizing receptor proteins on chromatin. Further work will show whether AtSCC4 forms an interacting scaffold for CHROMOSOME TRANSMISSION FIDELITY 7 (AtCtf) protein and how the Scc2–Scc4 complex is distributed along chromatin.

We established a method for in vivo detection of cohesin immobilization dynamics during embryogenesis (Fig. 8). Overexpression of cohesin proteins affects meiotic cell division in Arabidopsis (Yuan et al., 2012). Therefore, we established lines expressing kleisin subunits under the control of the ABI3 promoter, which has a relatively low and localized activity (Ng et al., 2004). Plants expressing TagRFP–SYN4 under control of the ABI3 promoter did not show any aberrant phenotype and the expression level of SYN4 kleisin turned out to be optimal for microscopy. As expected, cohesin decorated plant chromatin during interphase, was removed by metaphase and colocalized with chromatin again at late telophase (Fig. 8C). The iFRAP assay confirmed immobilization of SYN4 in the nuclei, indicating its anchoring on chromatin in wild-type background and demonstrated requirement of AtSCC4 for cohesin loading in vivo (Fig. 8D–F). Importantly, analysis of dynamics of AtSCC4 and SYN4 colocalization with chromatin suggests that cohesin localization on chromatin upon cell division coincides with localization of AtSCC4 in the nuclei. This observation reinforces the mechanistic role of AtSCC4 in cohesin loading on chromatin.

In this study, we identified and characterized the plant homolog of Scc4 and confirmed its evolutionary conserved features, such as a direct interaction with the N-terminal domain of Scc2. Our in vivo cohesin loading assay revealed the strict requirement of AtSCC4 for cohesin immobilization in the nuclei that most probably occurs due to loading of cohesin on chromatin. We also demonstrated the importance of the AtSCC2–AtSCC4 complex for cell fate determination. Further studies into the role of the cytoplasmic fraction of AtSCC4, as well as the interactome and possible post-translational modifications of AtSCC4, will provide clues to understanding its role in the spatiotemporal transcriptional regulation of gene expression and chromatin maintenance.

Scc4 homologs were identified using the default parameters of Protein Basic Local Alignment (PBlast) and Delta-BLAST tools. Obtained sequences were aligned using the Clustal Omega tool (EMBL-EBI). Accession numbers are available in the Table S1 and alignment is available in the Fig. S5. The phylogenetic tree was built using the neighbor-joining method and software Geneious v6.1.8 (Saitou and Nei, 1987), with 2000 bootstrap test iterations (Felsenstein, 1985) and visualized with support threshold set to 70%.

The prediction of putative interactors of AtSCC4 was performed using STRING database (http://string-db.org; Szklarczyk et al., 2015).

Expression of AtSCC2, AtSCC4, SYN1, SYN2, SYN3 and SYN4 in Arabidopsis tissues was estimated by using the Anatomy tool of the GENEVESTIGATOR database (https://www.genevestigator.com; Zimmermann et al., 2004) and the spatiotemporal expression tool of the eFP browser (http://bar.utoronto.ca/~dev/eplant/; Fucile et al., 2011).

Sequences of all oligonucleotide primers used in this study can be found in Table S2.

The promoter of AtSCC4 was amplified from the Arabidopsis genomic DNA using primers MAU2promFWGW/MAU2promRVGW and cloned into the pDONR/Zeo vector (Invitrogen, ThermoFisher Scientific, Waltham, USA). The obtained entry clone was recombined with pGWB3 vector (Nakagawa et al., 2007).

Full length AtSCC4 (At5g51340) and its promoter was amplified from Arabidopsis genomic DNA using Fuc-Fw/MAU2RVGW primers and cloned into the pDONR/Zeo. The obtained entry clone was recombined with pGWB4 vector (Nakagawa et al., 2007).

Proteins were extracted from rosette leaves with Laemmli sample buffer (Laemmli, 1970), separated on 12% polyacrylamide gels and transferred onto a PVDF membrane. Anti-GFP (clone JL-8, Cat 632381, Lot A5033481, Clontech, Takara Bio Europe, Saint-Germain-en-Laye, France) was used at dilution 1:2000; HRP-conjugated anti-mouse-IgG (Cat NA931, Amersham, GE Healthcare, Uppsala, Sweden) was used at dilution 1:5000. The reaction was developed using an ECL Prime kit (Amersham, GE Healthcare, Uppsala, Sweden) and detected using ChemiDoc XRS+ and ImageLab software (BioRad, Solna, Sweden).

The promoter of the ABI3 (AT3G24650) gene was amplified from Arabidopsis genomic DNA using ABI3prFw/ABI3prRe primers and inserted into pGWB560 and 561 vectors using HindIII/XbaI sites. The coding sequences (CDS) of SYN1 (At5g05490), SYN2 (At5g40840.2), SYN3 (At3g59550) and SYN4 (At5g16270) were amplified using cDNA derived from 2-week-old leaves of Arabidopsis Col-0 plants and primers Fw_SYN1/RV_NS_SYN1, Fw_SYN2/RV_NS_SYN2, Fw_SYN3/RV_NS_SYN3 and Fw_SYN4/RV_NS_SYN4. PCR products were cloned into pDONR/Zeo. Obtained entry clones were recombined with pGWB560 ABI3 and pGWB561 ABI3 vectors.

A piece of rosette leaf was ground in 30 μl of 0.5 M NaOH and resuspended in 370 μl of 0.1 M Tris-HCl pH 8.0; 3 μl of DNA solution were used per 15 μl of PCR. Primers used for genotyping are listed in Table S1; genotyping of AtSCC4 T-DNA insertion lines is presented in the Figs S2A,B and S3A. Genotyping of AtSCC2 T-DNA insertion lines was previously described in Sebastian et al. (2009). For genotyping of the complementation lines, the AtSCC4 wild-type allele was detected using primers AtSCC4 3UTR Re/Atscc4-F, the Atscc4-2 T-DNA insertion was detected using primers Atscc4-R/LB1 SALK, and the complementation insert pAtSCC4::AtSCC4-GFP was detected with primers Atscc4-F/attB2.

For the co-immunoprecipitation (co-IP) experiment, the AtSCC4 and AtSCC2 CDS were recombined into pGWB418 and pGWB641 (Nakagawa et al., 2007) vectors, respectively, and introduced into Agrobacterium strains GV3101. N. benthamiana leaves were co-infiltrated with cultures carrying Myc–SCC4 and CFP–SCC2, or Myc–SCC4 and free CFP. Samples were collected 2 days after infiltration, frozen and used for co-IP. Briefly, tissues were processed in nuclei isolation buffer [1 M hexyleneglycol, 20 mM PIPES-KOH pH 7.6, 10 mM MgCl2, 0.1 mM EGTA pH8, 20 mM NaCl, 60 mM KCl, 1× protease inhibitor (Cat P9599, Sigma, Stockholm, Sweden), 1% Triton X-100 and 5 mM β-mercaptoethanol] and incubated for 15 min at 4°C. Extracts were filtered through Miracloth and Partec CellTric filters (Cat 25004-0042-2316, Sysmex, Kungsbacka, Sweden) and centrifuged at 1500 g for 10 min at 4°C. The pelleted nuclei were resuspended and incubated for 1 h at 4°C in nuclei lysis buffer (50 mM Tris-HCl, pH 8, NaCl 400 mM and 0.1% NP-40) followed by sonication and centrifugation at 6000 g for 5 min at 4°C. Finally, nine volumes of nuclei buffer (50 mM Tris-HCl, pH 8, 1× protease inhibitor and 0.1% NP-40) were added to the supernatant. Co-IP was performed according to the manufacturer's instructions (μMACS GFP Isolation Kit; Cat 130-091-125, Miltenyi Biotec, Lund, Sweden). 1:1000 anti-Myc (Cat 11667149001, Roche, Stockholm, Sweden),1:2000 anti-GFP (clone JL-8, Cat 632381, Lot A5033481, Clontech, Takara Bio Europe, Saint-Germain-en-Laye, France), 1:5000 HRP-conjugated anti-mouse-IgG (Cat NA931, Amersham, GE Healthcare, Uppsala, Sweden) and the ECL Prime kit (Amersham, GE Healthcare, Uppsala, Sweden) were used for the immunoblotting.

RNA was extracted from 3-week-old seedlings of Col-0 wild-type, Atscc4-1/AtSCC4 and Atscc4-2/AtSCC4, from three biological replicates per genotype. 400 ng RNA were used per reverse transcription reaction with a Maxima kit (Cat. K1671, ThermoFisher Scientific, Waltham, MA). qPCR was performed using CFX thermal cycler (Bio-Rad), DyNAmo Flash qPCR kit (Cat F415S, ThermoFisher Scientific, Waltham, MA) and primers MAU2_SAIL_qPCR_F/R, MAU2_SALK_qPCR_F/R, PP2AqPCRFw/Re and HELqPCRFw/Re. Quantification of AtSCC4 gene expression was assessed by ΔΔCt method (Livak and Schmittgen, 2001) and normalized to PP2A (At1g13320.1) and RNA helicase (At1g58050.1) expression (Czechowski et al., 2005).

Arabidopsis seeds were surface sterilized using a 10% commercial bleach, rinsed with milli-Q water, and vernalized at 4°C for 24–48 h. Seeds were sown on half-strength Murashige and Skoog (MS) plates containing 0.8% (w/v) agar and 1% (w/v) sucrose. Plates were incubated under long-day conditions (16 h at 120 µE m⁻² s⁻¹ light intensity at 22°C, with an 8 h night at 20°C). For root microscopy assays, plates were kept vertically.

For growth in soil, vernalized seeds were sown directly on soil. Plants were grown under standard long-day conditions (16 h at 120 µE m⁻² s⁻¹ light intensity at 22°C, with an 8 h night at 20°C).

Arabidopsis Col-0 plants were transformed as described previously (Clough and Bent, 1998) using Agrobacterium tumefaciens strain GV3101.

GUS detection was performed essentially as described previously (Fincato et al., 2012).

Seeds were harvested and fixed overnight at 4°C in EtOH with acetic acid (9:1, v/v), treated with 1% SDS (w/v) and 0.2 M NaOH for 1 h at room temperature (RT), rinsed with water and incubated in 25% bleach solution for 1 min at RT. After rinsing with water, seeds were incubated in 1% (v/v) periodic acid at RT for 40 min, washed with water again and submerged in Schiff's reagent supplemented with propidium iodide (100 mM sodium metabisulfite and 0.15 N HCl; propidium iodide to a final concentration of 1 mg/ml was freshly added) until visibly stained. Excess staining was washed away with water, and seeds were mounted in chloral hydrate solution (4 g chloral hydrate, 1 ml glycerol and 2 ml water), left for 1 h and then observed using a Zeiss CLSM780 microscope (Carl Zeiss GmbH, Jena, Germany) with excitation at 488 nm, emission at 568–735 nm and a 1 AU pinhole.

Seeds were taken out of the siliques, incubated on a sample glass in chloral hydrate solution (4 g chloral hydrate, 1 ml glycerol and 2 ml water) for 2 h at RT and observed using DIC optics with a AxioScope A1 microscope (Carl Zeiss GmbH, Jena, Germany) and ZEN lite software (Carl Zeiss GmbH, Jena, Germany).

Atscc4 and AtScc2 T-DNA lines were crossed with plants expressing DR5rev::3xVenus-N7 (Heisler et al., 2005). The F1 generation was genotyped to confirm the presence of the corresponding T-DNA. Siliques from selected plants were opened with a needle and fixed for 1 h in 3% (w/v) paraformaldehyde, 50 mM PIPES pH 6.8. Seeds were taken out, placed on a sample glass in Milli-Q water and gently pressed with the coverslip to release embryos. Only one silique at a time was analyzed to ensure that all embryos are at approximately the same developmental stage. Venus signal was detected using excitation at 514 nm, emission at 500–530 nm and the Zeiss CLSM 780 microscope. The experiment was repeated three times using plants grown either in the greenhouse or in the growth chamber; 28 to 42 embryos were imaged in each experiment.

Seeds or complete seedlings were fixed for 15–30 min in 3% (w/v) paraformaldehyde, 50 mM PIPES pH 6.8, rinsed with PBS three times and incubated for 15 min at RT in 4 µg/ml DAPI in PBS. Roots of seedlings were mounted and imaged using the Zeiss CLSM 780 microscope. Prior to imaging, embryos were taken out of seeds as described above.

Yeast two-hybrid analysis was performed using the Matchmaker Gold system (Clontech, Takara Bio Europe, Saint-Germain-en-Laye, France) using Y2HGold (MATa/trp1-901/leu2-3/112/ura3-52/his3-200/gal4Δ/gal80Δ/LYS2::GAL1UAS-Gal1TATA-His3/GAL2UAS-Gal2TATA-Ade2/URA3::MEL1UAS-Mel1TATA-AUR1-C/MEL1) and Y187 (MATα/ura3-52/his3-200/ade2-101/trp1-901/leu2-3, 112/gal4Δ/gal80Δ/URA3::GAL1UAS-GAL1TATA-lacZ) strains for bait and prey constructs, respectively (Maier et al., 2008). The AtSCC4 (At5g51340) CDS was amplified from cDNA derived from Arabidopsis Col-0 plants with MAU2FWGW/MAU2RVGW primers, cloned into the pDONR/Zeo and recombined with pGBKT7. The N-terminal part of AtSCC2 (At5g15540) was amplified using the same cDNA and FwGWscc2/RVtruncation1GWscc2 primers, cloned into pDONR/Zeo and recombined into the prey vector pGADT7. Yeast transformation was performed by the LiAc/PEG method as described in Yeast Protocols Handbook (Clontech). Growth on quadruple dropout medium (QDO; –Ura,–His,–Trp,–Leu) plates indicated a positive interaction.

Zeiss CLSM 780 and ZEN Black software ZEN black (Carl Zeiss GmbH, Jena, Germany) were used for iFRAP expereiments. Selected half of a nucleus was bleached using corresponding laser lines at 100% transmittance and 100 iterations. Typically, optical section size was optimized to fit the average diameter of nuclei in samples (6-9 µm). Three pre-bleach and at least 50 post-bleach images were acquired. Fluorescence intensity was measured in the not subjected to photobleaching regions of interest and in the photobleached regions. Average intensities of the three scans right before or after photobleaching were used for quantification. Loss of intensity or the retained intensity were calculated as the percentage of the corresponding pre-bleached value. A measurement made outside the bleached region was used as the background. Loss of intensity in a nucleus not subjected to photobleaching was used to assess bleaching caused by scanning. Changes in the intensity in a fully bleached nucleus were used to assess the rate of possible import of fluorescent protein synthesized de novo into nuclei.

ZEN black (Carl Zeiss GmbH, Jena, Germany) or ImageJ v 1.41 software (http://rsb.info.nih.gov/ij/index.html) were used for image analysis. Statistical analysis was performed using JMP software v11.

We would like to thank Dr Eugene I. Savenkov for sharing material for some of the experiments and critical review of the results.