MIDGET connects COP1-dependent development with endoreduplication in Arabidopsis thaliana

Authors


For correspondence (e-mails joachim.uhrig@uni-koeln.de; andrea.schrader@uni-koeln.de).

Summary

In Arabidopsis thaliana, loss of CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) function leads to constitutive photomorphogenesis in the dark associated with inhibition of endoreduplication in the hypocotyl, and a post-germination growth arrest. MIDGET (MID), a component of the TOPOISOMERASE VI (TOPOVI) complex, is essential for endoreduplication and genome integrity in A. thaliana. Here we show that MID and COP1 interact in vitro and in vivo through the amino terminus of COP1. We further demonstrate that MID supports sub-nuclear accumulation of COP1. The MID protein is not degraded in a COP1-dependent fashion in darkness, and the phenotypes of single and double mutants prove that MID is not a target of COP1 but rather a necessary factor for proper COP1 activity with respect to both, control of COP1-dependent morphogenesis and regulation of endoreduplication. Our data provide evidence for a functional connection between COP1 and the TOPOVI in plants linking COP1-dependent development with the regulation of endoreduplication.

Introduction

Sunlight is not only the primary source of energy for plants, but also a fundamental environmental signal that controls development throughout the plant life cycle. Plants perceive light through photoreceptors that initiate signaling cascades resulting in changes in gene expression and photomorphogenesis. These changes are particularly obvious at the seedling stage: light-grown seedlings are characterized by short hypocotyls, unfolded and expanded cotyledons, elongated roots, differentiation of chloroplasts and accumulation of pigments, such as chlorophyll and anthocyanins. Dark-grown seedlings, by contrast, are pale and have elongated hypocotyls, closed cotyledons and short roots, a process called skotomorphogenesis (Whitelam and Halliday, 2007; Franklin and Quail, 2010).

Arabidopsis thaliana mutants undergoing constitutive photomorphogenesis in darkness have been instrumental in identifying three protein complexes as central repressors of photomorphogenesis inhibiting light signaling in darkness: the COP9 signalosome (CSN), the COP1/SPA complex and the COP10–DDB1–DET1 (CDD) complex. The COP1/SPA complex is part of a cullin 4-based E3 ubiquitin ligase, which is thought to be regulated by the CSN and the CDD complex (Laubinger et al., 2004; Yi and Deng, 2005; Zhu et al., 2008; Chen et al., 2010). The CSN is a protein complex conserved in yeast, animals and plants that regulates the activity of cullin RING-type E3 ubiquitin ligases (CRLs) by de-neddylation of the cullin subunits (Wei et al., 2008; Schwechheimer and Isono, 2010).

In darkness, COP1 activity represses light-dependent gene expression by ubiquitylating positive regulators of photomorphogenesis, such as the transcription factors ELONGATED HYPOCOTYL 5 (HY5), LONG HYPOCOTYL IN FAR-RED1 (HFR1) and LONG AFTER FAR-RED LIGHT1 (LAF1), thereby targeting them for degradation by the proteasome (Yi and Deng, 2005). Similarly, COP1 is involved in flowering time regulation by ubiquitylating CONSTANS (CO) and EARLY FLOWERING 3 (ELF3) (Jang et al., 2008; Liu et al., 2008; Yu et al., 2008). In the light, photoreceptors inhibit COP1 function such that activators of the light response are stable (Franklin and Quail, 2010; Yu et al., 2010). Recent results show that blue light inhibits COP1/SPA function through light-dependent interaction of the blue light receptors cry1 and cry2 with SPA1 (Liu et al., 2011; Zuo et al., 2011).

COP1 activity is dependent on the integrity of three conserved protein domains: the N-terminal RING domain mediates COP1 E3-ligase activity (Seo et al., 2003, 2004; Liu et al., 2008), a central coiled-coil domain facilitates COP1 homo-dimerization and interaction with SPA1 (Torii et al., 1998; Hoecker and Quail, 2001; Saijo et al., 2003), and a C-terminal WD40 domain acts as the primary site of target protein binding as well as DDB1 interaction (Yi and Deng, 2005; Chen et al., 2010). In addition, a nuclear localization sequence (NLS), a cytoplasmic localization sequence (CLS) and a sub-nuclear localization signal (SNLS) have been identified (Stacey and von Arnim, 1999; Stacey et al., 1999). The activity of COP1 correlates well with its nucleo-cytoplasmic shuttling regulated by its NLS, CLS and by the CSN (von Arnim and Deng, 1994; Stacey and von Arnim, 1999; Stacey et al., 1999; Schwechheimer and Deng, 2001). The SNLS overlaps with both the CLS and the coiled-coil domain and is necessary and sufficient to localize COP1 to COP1-containing nuclear bodies that are suggested sites of signaling, protein degradation or splicing (Stacey and von Arnim, 1999; Seo et al., 2003, 2004; Shaw and Brown, 2004; Dieterle et al., 2005; Chen, 2008; Liu et al., 2008; Yu et al., 2009).

Regulation of hypocotyl length is one major aspect of COP1-controlled morphogenesis. Hypocotyl length is mainly modulated by cell expansion, not by cell division (Gendreau et al., 1997). Endoreduplication occurs in these cells, during which the genome is repeatedly replicated without subsequent cytokinesis. Cell size positively correlates in most cases with the DNA content and thereby with the number of endocycles (Melaragno et al., 1993; Kondorosi et al., 2000).

Endoreduplication in A. thaliana is critically dependent on the function of the TOPOISOMERASE VI (TOPOVI) complex, composed of ROOT HAIRLESS 2 (RHL2), A. thaliana TOPOISOMERASE 6 SUBUNIT B (AtTOP6B), MIDGET (MID) and ROOT HAIRLESS 1 (RHL1) (Corbett and Berger, 2003; Sugimoto-Shirasu et al., 2005; Breuer et al., 2007; Kirik et al., 2007). The TOPOVI core components are unique to Viridiplantae within the Eukarya. On the one hand, RHL2 and TOPOVI clearly resemble the TOPOVI A- and B-subunit, respectively, from Archaea (Hartung and Puchta, 2001; Hartung et al., 2002; Forterre and Gadelle, 2009). MID, on the other hand, exhibits low sequence similarities with mammalian topoisomerases IIα and is a DNA binding protein with a functional bipartite NLS and an AT-hook (Breuer et al., 2007; Kirik et al., 2007). In addition to endoreduplication above 8C, MID is needed for chromatin condensation and transcriptional silencing (Breuer et al., 2007; Kirik et al., 2007). MID as well as RHL1 are discussed as essential activators of the TOPOVI core components (Sugimoto-Shirasu et al., 2005; Breuer et al., 2007; Kirik et al., 2007).

The TOPOVI complex is hypothesized to indirectly promote endoreduplication and to maintain genome integrity by preventing the accumulation of DNA double-strand breaks (DSBs) (Hartung et al., 2002; Sugimoto-Shirasu et al., 2005; Breuer et al., 2007; Kirik et al., 2007). In mid mutants, the occurrence of DSBs correlates with the ATAXIATELANGIECTASIA-MUTATED- AND RAD3-RELATED (ATM/ATR) dependent activation of DNA damage response presumably leading to the halt of endoreduplication caused by a checkpoint-driven cell cycle arrest (Breuer et al., 2007; Kirik et al., 2007). Interestingly, mutants with defects in TOPOVI core components (RHL2, AtTOP6B) show constitutive photomorphogenesis in darkness (Yin et al., 2002). Similarly, accumulation of DSBs was observed in cop1 and csn mutants. Here genome integrity is compromised such that DSBs occur frequently and the ATM/ATR-dependent DNA damage response is activated, providing an explanation for the observed cell cycle arrest (Dohmann et al., 2008).

In this study, we show that COP1 interacts physically and genetically with the MID protein, a component of the TOPOVI complex, providing a functional link to the TOPOVI complex that connects photomorphogenesis with endoreduplication.

Results

COP1 and MID interact in vivo and in vitro

In order to further dissect COP1-controlled plant development, cDNA libraries were screened using the yeast two-hybrid system with COP1 as bait. More than 100 positive colonies were analysed. We found several interactors of COP1 described previously, including SALT TOLERANCE (STO), SALT TOLERANCE HOMOLOGUE (STH) and HFR1, thereby demonstrating the reliability of the screening procedure (Yi and Deng, 2005). Additionally, we obtained a number of COP1-interacting proteins not described before (Table S1). Six candidates encoded the TOPOVI component MIDGET (MID/BIN4) (Breuer et al., 2007; Kirik et al., 2007).

COP1-interacting proteins can be classified as either targets for COP1-dependent ubiquitylation (e.g. HY5 and HFR1) or regulators/co-factors (e.g. SPA1, cry1 or cry2) (Yi and Deng, 2005). The WD40 domain of COP1 is responsible for selective target binding, and a particular lysine residue (K550) has been shown to form a salt bridge essential for the interaction with a subset of the target proteins (Holm et al., 2001). The COP1–MID interaction, however, is not disrupted by a mutation of lysine 550 to aspartate, a mutation completely abolishing interaction with for example the COP1 target HY5 (Figure 1a). Furthermore, a minimal fragment comprising just the N-terminal 67 amino acids of COP1 is sufficient to interact with the MID protein in yeast. The COP1–MID interaction domain includes the N-terminal half of the RING-finger domain but lacks both the conserved coiled-coil domain and the C-terminal target-binding WD40 domain (Figure 1a), which suggests that MID is probably not a target of COP1. Cop1eid6 is a weak COP1 mutant allele exhibiting hypersensitivity to light (Dieterle et al., 2003). The single amino acid exchange in the mutant COP1 protein is directly adjacent to the COP1–MID interaction domain identified in this study. We investigated whether this particular mutation affects the interaction with MID, and found that the EID6 mutation does not disrupt the MID–COP1 interaction in the yeast two-hybrid assay (Figure S1).

Figure 1.

MID and COP1 interact.(a) Yeast cells were co-transformed with: (1) MID/COP1, (2) MID/GFP, (3) GFP/COP1, (4) MID/COP1K550E, (5) HY5/COP1K550E, (6) COP11–67/MID, (7) COP11–67/GFP and (8) GFP/MID. Cells were selected on double dropout medium (–LW) and tested for interaction on triple dropout media (–LWH). Inset: Schematic representation of the COP1 protein. (b) Co-immunoprecipitation (Co-IP) of YFP–MID with RFP–HA–COP1. Nicotiana benthamiana leaves were co-infiltrated with Agrobacteria harboring the depicted constructs. Immunoprecipitation (IP) was performed with anti-GFP beads. Input and precipitate were analysed with anti-GFP (IP) and anti-HA (Co-IP) antibodies. Coomassie staining served as loading control, combinations with RFP–HA and YFP served as negative controls.

Co-immunoprecipitation (Co-IP) experiments showed that MID and COP1 can associate physically in plant tissue. Constructs encoding RFP–HA–COP1 and YFP–MID were expressed transiently in Nicotiana benthamiana leaves under the control of the CaMV 35S promoter. Both fusion constructs can rescue the mutant phenotypes of the cop1-4 and the mid-2 mutant, respectively, and are therefore considered functional (Figures S2 and S3). Transient co-expression was confirmed by red fluorescent protein (RFP) and yellow fluorescent protein (YFP) fluorescence analysis. Immunoprecipitation using anti-GFP (green fluorescent protein) antibodies co-precipitated the RFP–HA–COP1 fusion protein (Figure 1b, lane 4) while no specific proteins were co-precipitated in the control experiments (Figure 1b, lanes 5, 6). This finding demonstrates a physical association of MID and COP1 in plant extracts.

A characteristic of COP1 target proteins is their degradation in darkness. To further investigate whether or not MID is a target of COP1, we analysed MID protein stability in wild-type and cop1-4 mutant background in dependence of different light regimes. Plants constitutively expressing YFP–MID or HA–MID, respectively, were characterized with respect to their YFP fluorescence and protein accumulation. Seedlings were grown in continuous light for 4 days and then shifted to darkness for 23 h. No significant differences in fluorescence intensity were detected in either wild-type or cop1-4 mutant (Figures S6a). Western blot detection of YFP–MID protein from 7-day-old long day (LD)-grown seedlings showed approximately equal amounts of YFP–MID before and after shift from light to dark conditions under LD regime (Figure S6b) indicating no dark-induced degradation of the MID protein. We even observed an increase of HA–MID in adult plants after 16 h darkness (Figure S6b). These results provide evidence that MID is not a target of COP1.

MID supports COP1 nuclear body formation

Specific sub-nuclear localization is critical for COP1 function (Yi and Deng, 2005). We therefore investigated the MID–COP1 interaction in plant cells with respect to their intracellular localization by bimolecular fluorescence complementation (BiFC) and co-localization experiments. MID and COP1 fused to the N- and C-terminal fragments of YFP, respectively, were transiently expressed in epidermal cells of Allium porrum. YFP fluorescence indicates interaction of the MID protein with COP1 in vivo (Figure 2). BiFC signals were found predominantly in the nucleus in defined dots. Similar sub-nuclear localization has been reported for COP1 interactions with other proteins such as STO or CO (Indorf et al., 2007; Jang et al., 2008). Additionally, BiFC signals were detected in cytoplasmic aggregates that are observed frequently in COP1-expressing cells (Ang et al., 1998; Indorf et al., 2007). BiFC also proved that the N-terminal domain of COP1 (aa 1–67) can interact with MID in the nucleus of plant cells (Figure 2). Interestingly, in contrast to the interaction with full-length COP1, the fluorescence signal is evenly distributed throughout the nucleus (Figure 2). This finding indicates that the fluorescing foci observed for full-length COP1 represent a specific sub-nuclear localization of the intact COP1 protein which is not achieved by the truncated COP11–67 irrespective of its ability to interact with MID in vivo. The specific site of COP1–MID interaction is thus determined by COP1.

Figure 2.

Sub-nuclear localization of COP1–MID interaction. BiFC fusion constructs of MID and COP1 (a–c) or COP11–67(d–f) were transiently expressed in A. porrum and analysed by CLSM. CFP–TALIN served as transformation control (a, d). YFP fluorescence indicates interaction between MID and COP1 or COP11–67, respectively (b, e). Sequentially scanned z-stacks were merged. Right (c, f): Overlay of CFP and YFP channel. Scale bar: 20 μm.

To further study the interrelationship of MID and COP1 with respect to their sub-nuclear localization, we transiently expressed labeled full-length MID and COP1 in the epidermal leaf cells of A. porrum, N. benthamiana and in the A. thaliana mid-2 mutant. In all three backgrounds, YFP–MID is found in the nucleus, and RFP–HA-tagged (N. benthamiana, A. thaliana) or CFP-tagged (A. porrum) COP1 accumulate in nuclear bodies (Figure 3 and S7).

Figure 3.

MID regulates the sub-nuclear localization of COP1 in planta. Co-bombarded epidermal cells of Allium porrum (a–i), mid-2 rosette leaves (j–o) and co-infiltrated leaf epidermal cells of Nicotiana benthamiana (p–y) were analysed by fluorescence microscopy (a–i) and CLSM (j–y). (c, f, i) are enlargements of the nuclei in (a, e, h). Arrows point out cytoplasmic COP1 aggregates. Scale bar: 50 μm (a, b, d,e, g, h); 25 μm (c, f, i); 5 μm (j–o) and 20 μm (p–y).

When transiently co-expressing YFP–MID and tagged COP1, however, the number of visible COP1-containing sub-nuclear foci increased about three-fold in all three systems (Table 1, Figure 3 and, for wild-type background, Figure S7). The presence of MID apparently enhances the capacity of COP1 to accumulate in sub-nuclear foci, suggesting that the MID protein can modulate the localization of COP1 within the nucleus (Table 1 and Figure 3). Preliminary data suggest that MID has a similar effect on the accumulation of the COP1EID6 mutant protein in nuclear bodies (Table S2).

Table 1. MID supports COP1 nuclear body formation Visible sub-nuclear foci, no. ± STDEV (no. of analysed cells)
  Allium porrum Arabidopsis thaliana (mid-2) Nicotiana benthamiana
  1. In all experiments, YFP-MID was used.

  2. MID/COP1: COP1 was either RFP–HA tagged (N. benthamiana, A. thaliana) or CFP tagged (A. porrum). COP1: as a control YFP was co-transformed with RFP–HA–COP1 or CFP–COP1, respectively.

  3. *Number of assays in superscript. aP = 1.4E–07; bP = 1.4E–05; cP = 7.98E–07. STDEV, standard deviation.

MID/COP124 ± 19.5 (56)3,a,*3.7 ± 2.7 (27)3,b9 ± 4.2 (20)2,c
COP17.7 ± 7 (55)3,a1.4 ± 1.5 (32)4,b2.4 ± 1.2 (22)2,c

Taken together, we have shown that MID and COP1 can interact physically in planta and that MID regulates COP1 accumulation in nuclear bodies in planta.

Mid and topoVI mutants have photomorphogenesis defects reminiscent of the cop1 mutant

Most targets of COP1 in A. thaliana are activators of photomorphogenesis, and the corresponding mutants, therefore, display pronounced aspects of skotomorphogenesis in the light. The opposite holds true for regulators/co-factors of COP1, their mutants exhibit constitutive photomorphogenetic phenotypes. We therefore examined weak and strong loss-of-function alleles of midget (mid-1 and mid-2) as well as mutants of the MID-associated TOPOVI (rhl2, hyp6) with respect to light-dependent development.

Similar to the cop1-4 mutant, mid and topoVI mutants consistently showed several morphological aspects of a de-repressed photo- or repressed skotomorphogenesis in the dark: Hypocotyls were short, the apical hook unfolds, and the cotyledons opened up (Figure 4a). The increase in leaf lamina area that has previously been reported for cop1 mutants (Deng and Quail, 1992) was not observed in the mid and topoVI mutants (Figure 4a). Hypocotyl cell files of A. thaliana Col-0 contain 21–25 cells, and increased hypocotyl length in dark-grown seedlings results from enhanced cell elongation and does not involve cell division (Gendreau et al., 1997). We analysed hypocotyl cell files of mid mutants and the cop1-4 mutant and found that both consist of approximately the same number of cells (mid mutant: 22.3 ± 1.5 cells, cop1-4 mutant 22.8 ± 2.2 cells). The reduced hypocotyl length, therefore, in both cases results from reduced cell elongation rather than reduced cell division. These results are in agreement with a previous analysis of the mid/bin4 mutant (Breuer et al., 2007).

Figure 4.

TOPOVI mutants show aspects of photomorphogenesis in the dark. (a) Dark-grown 7-day-old topoVI-mutant seedlings (mid-1, mid-2, rhl2, hyp6), Col-0 and cop1-4. Scale bar: 5 mm. (b) Anthocyanin content of dark-grown Col-0, cop1-4, mid-2 and rhl2 seedlings (MS, 2% sucrose). Error bars are standard deviation (STDEV). #: P > 0.05 all other: P < 0.01 (see Table S3). See Figure S4 for results using 0% sucrose. (c) Analysis of CHALCONE SYNTHASE expression. RT-PCR was performed on RNA from 3-day-old dark-grown Col-0, cop1-4, mid-2 and rhl2 seedlings. RT-PCR of UBIQUITIN10 (UBQ10) served as control. CHS: 29 and 32 cycles, UBQ10: 26 and 29 cycles. See Figure S5 for expression analysis of CAB and RBCS in mid-2.

Cop1 mutants accumulate anthocyanin in dark-grown seedlings due to a decreased COP1-dependent degradation of transcription factors such as HY5, which regulates CHALCONE SYNTHASE (CHS) expression (Yi and Deng, 2005). We determined the anthocyanin content per gram fresh weight, to compensate for unequal growth phenotypes of Col-0, cop1-4, mid-2 and topoVI (here rhl2) mutants. Here, 7-day-old dark-grown seedlings of wild-type Col-0, mid-2, rhl2 and cop1-4 mutants were analysed. The mid-2 and rhl2 mutant accumulated only roughly a quarter of the cop1-4 anthocyanin levels, but in comparison with the wild-type, significantly increased anthocyanin contents were observed (Figure 4b). Here, a sucrose concentration of 2% was used as for 0% already the positive control is close to the detection limit of this assay (Figure S4). Consistently we found expression of the CHS gene in dark-grown mid-2 and rhl2 mutant seedlings, while in dark-grown wild-type seedling the CHS transcript was not detectable (Figure 4c). Furthermore, enhanced levels of CAB and RBCS in dark-grown mid-2 seedlings underline the activation of the photomorphogenic program in the dark (Figure S5).

Taken together, cop1, mid and rhl2 mutants share a common set of photomorphogenetic defects, suggesting that MID and TOPOVI might act in a COP1 pathway.

MID and COP1 interact genetically

The physical interaction and the phenotypic analyses suggest that MID is a positive regulator of COP1. Ectopic expression of MID in cop1 mutant background, and similarly overexpression of COP1 in mid mutant background, respectively, did not result in any informative aberrant overexpression phenotypes. To analyse a potential genetic interaction of COP1 and MID, we therefore created double loss-of-function mutants. Two weak cop1 alleles, cop1-4 and cop1eid6, were chosen because strong cop1 alleles are adult lethal and exhibit severe germination defects (McNellis et al., 1994). When crossing the strong mid mutant, mid-2, that exhibits severe growth defects as well, to cop1-4, we did not obtain any homozygous double mutant. Consequently, the weak mid-1 mutant was chosen in order to allow statistical analysis. Seven-day-old dark-grown double mutant seedlings homozygous for both mutations were analysed with respect to their morphogenesis in the dark (Figure 5a; phenotypes of 4-day-old seedlings see Figure S8).

Figure 5.

MIDGET and COP1 interact genetically. Phenotypes of 7-day-old dark-grown (a) and light-grown (b) seedlings. Scale bar: 5 mm. (c) Hypocotyl lengths of 7-day-old dark-grown seedlings. (d, e) Opening angles of cotyledons: (d) lamina angle, (e) petiole angle. (f) Hypocotyl length of 7-day-old light-grown seedlings. (g) Root length and (h) root/hypocotyl ratio of 7-day-old dark or light-grown seedlings. All double mutants were verified by PCR. Error bars are standard deviation (STDEV). For the double mutants in comparison with its single mutants, the levels of significance are indicated by asterisks as follows: ***P < 0.001, **P < 0.01, m: only in comparison to mid-1, c: only in comparison to the cop1 mutant. For a detailed statistic analysis see Table S3.

Adult lethal strong cop1 mutants differ from weak cop1 mutants by shorter hypocotyls and wider open cotyledons with clearly separated petioles (e.g. Ang and Deng, 1994). As compared with the single mutants, both double mutants (mid-1cop1-4 and mid- 1cop1eid6) exhibited significantly decreased hypocotyl lengths suggesting genetic interaction of the weak alleles (Figure 5a,c and Table S3). Similarly, the double mutants differed from the single mutants with respect to the opening of cotyledons. For reasons of simplicity, opening of cotyledons usually is characterized by the cotyledon angle, formed between the tips of the cotyledons and the SAM (Neff and Chory, 1998). However, weak cop1 mutants such as cop1-4 or cop1-6 differ from strong cop1 mutants (cop1-1) and for example from the det1-1 mutant mainly in the angle formed between the cotyledonary petioles (Ang and Deng, 1994). Because this difference might not be captured appropriately, when measuring the overall cotyledon angle, we quantified the opening by separately assessing the lamina angle formed between the petiole and the lamina, and the petiole angle (Figure 5d,e). Col-0, mid-1, cop1-4 and cop1eid6 showed significantly different lamina angles (P < 0.001) with on average 22, 60, 286 and 225 degrees, respectively (Figure 5d, Table S3). Comparing the lamina angles of single and double mutants, there was no significant difference between the mid-1cop1-4 double mutant and cop1-4 (P > 0.05), while the mid-1cop1eid6 double mutant significantly differed from the single mutants and Col-0 (P < 0.01; Figure 5d and Table S3). When analysing the petiole angle, however, we found a highly significant increase in both double mutants as compared to the respective single mutants (P < 0.001; Figure 5e and Table S3). While in Col-0, mid-1 and cop1-4 the petiole angles were on average 0, 1 and 7°, respectively, the double mutant mid-1cop1-4 exhibited a petiole angle of 45° (Figure 5e). Similarly, the mid-1cop1eid6 mutant seedlings showed an enhanced cop1 mutant phenotype with a petiole angle of 15° (Figure 5e and Table S3). These results support the view that COP1 and MID cooperate in suppressing photomorphogenic development.

Weak cop1 alleles have only minor phenotypic aberrations when grown in the light. Strong cop1 mutant alleles, however, show growth arrest early in development even when exposed to continuous light (Deng and Quail, 1992; McNellis et al., 1994). This phenotype indicates that COP1 function is not confined to an inhibitory activity in the dark. Mutants of COP1 regulators/co-factors can therefore be expected to enhance morphogenetic defects of weak cop1 alleles in the light. Indeed, when compared with the single mutants and to Col-0, both double mutants showed a significantly reduced hypocotyl length and an even more pronounced reduction of root length when grown for 7 days in the light (P < 0.001; Figure 5b,f,g and Table S3).

As the root lengths in the dark do not vary as much as in the light (Figure 5g), we wondered if the ratio of root to hypocotyl length varies in our mutants. Col-0 seedlings elongate their hypocotyl at the expense of the root in darkness with a root/hypocotyl ratio of 0.22 (Figure 5h). In the light, this differential elongation of root and hypocotyl is reversed with a corresponding ratio of 15.7 (Figure 5h). Although less pronounced, all single mutants showed the wild-type response to light (Figure 5h). In darkness, cop1-4 displayed the strongest phenotype of the single mutants with the root length almost equaling the hypocotyl length (ratio = 1.26) (Figure 5h). Dark-grown mid-1 and cop1eid6 seedlings showed a less pronounced reduction of the root/hypocotyl ratio than the wild-type (Figure 5h). Interestingly, in the light, the double mutant seedlings are not able to differentially determine the root and hypocotyl length. This finding indicates that both functional MID functional COP1 proteins are necessary for this particular light-dependent developmental response.

We were not able to obtain adult soil-grown mid-1cop1eid6 plants, indicating that similar to strong cop1 mutants (McNellis et al., 1994), this double mutant might be adult lethal. In contrast with the single mutants, mid-1cop1-4 seeds exhibit a severe germination defect (Table S4) as was described previously for cop1-16 and cop1-17 (McNellis et al., 1994).

In summary, the enhanced cop1 mutant phenotypes of double mutants of weak cop1 and mid alleles indicate a genetic interaction of COP1 and MID.

MID and COP1 jointly regulate endoreduplication

We and others have shown previously that MID is involved in the regulation of endoreduplication (Breuer et al., 2007; Kirik et al., 2007). Likewise, cop1 mutants are affected in endoreduplication of hypocotyl cells (Gendreau et al., 1998). In the hypocotyl, light-dependent repression of endoreduplication beyond the second cycle is regarded as a characteristic of photomorphogenesis (Gendreau et al., 1998; Traas et al., 1998). We therefore tested the genetic interactions of MID and COP1 with respect to the regulation of endoreduplication in hypocotyls of dark-grown seedlings focusing on cells with a DNA content above 8C. To exclude cell type-specific variation in DNA content, only epidermal cells were analysed excluding the stomata. The cotyledon's stomata were used as a reference for 2C as these do not undergo endocycles. Dark-grown seedlings of mid-1 and cop1 single mutants exhibited significantly reduced DNA contents apparent in a greatly reduced number of cells having completed more than three endocycles (>16C, Figure 6). Cop1eid6 mutants on the other hand, are comparable with wild-type, with the exception of slightly less cells with a DNA content of 32C and an increase in 4C-containing cells. Analysis of mid-1cop1-4 and mid-1cop1eid6 double mutants revealed significantly repressed endoreduplication. Mid-1cop1-4 cells differed from the respective single mutants by a complete absence of cells with a DNA content higher than 8C (Figure 6); and in mid-1cop1eid6 no hypocotyl epidermal cell had completed its fourth endocycle (Figure 6). The highest class of DNA content that was present in the hypocotyl epidermis of both single mutants was absent in the corresponding double mutant. This finding indicates that MID is dependent on a fully functional COP1 protein to promote more than two endocycles (DNA content above 8C) or to overcome the suppression of the third endocycle in the dark. Vice versa, the mutant COP1-4 and COP1EID6 proteins can fulfill its partial functions only if a functional MID protein is present. This synergistic phenotype of two weak mutants and our other findings strongly suggests that MID and COP1 act together in the regulation of endoreduplication.

Figure 6.

MID and COP1 regulate endoreduplication. Relative DNA content of 7-day-old dark-grown hypocotyl epidermal cells were determined by measuring the fluorescence of DAPI-stained nuclei. Stomata served as a reference for 2C. The number of analysed epidermal cells and seedlings is given below the columns. The relative proportions of cells with a DNA content >8C (more than two endocycles) or ≤8C are separated by a black line.

Discussion

COP1 suppresses photomorphogenesis in the dark by ubiquitylating target proteins, thereby labeling them for degradation. A major aspect of COP1-controlled skotomorphogenesis is hypocotyl elongation, which is achieved mainly by cell expansion rather than cell proliferation. In roots on the other hand, COP1 suppresses cell elongation in the dark presumably by triggering the proteasomal degradation of the actin-regulating SCAR proteins (Dyachok et al., 2011). The mechanism by which COP1 promotes hypocotyl cell elongation in the dark, however, is still largely unknown. Cell expansion correlates with amplification of DNA content by several cycles of endoreduplication. COP1 function might therefore include either the suppression of so far unknown growth repressors, or alternatively the induction of endoreduplication which then may be the trigger for cell expansion. In any case, there is a need to coordinate suppression of photomorphogenesis with the regulation of endoreduplication.

Both, TOPOVI and MID function have been specifically attributed to regulation of endoreduplication in A. thaliana (Hartung et al., 2002; Sugimoto-Shirasu et al., 2002; Breuer et al., 2007; Kirik et al., 2007). MID is an essential regulator of the TOPOVI core complex that can disentangle DNA strands by inducing DSB and re-ligating the DNA after the passage is completed (Hartung and Puchta, 2001; Sugimoto-Shirasu et al., 2002; Bates and Maxwell, 2005; Breuer et al., 2007; Kirik et al., 2007).

Our finding of a physical COP1/MID interaction in vivo and in vitro provides a direct molecular connection of the suppressor of photomorphogenesis to the machinery facilitating endoreduplication. We provide evidence supporting a positive regulation of COP1-function by MID with respect to both, the activity of COP1 as a regulator of photo- and skotomorphogenetic development and its endoreduplication-promoting function.

Typical targets of COP1 bind to the COP1 WD40 domain and exhibit light-dependent regulation of protein abundance based upon COP1-triggered degradation in darkness. MID, by contrast, interacts with an N-terminal domain of COP1 (COP11–67), and can accumulate in darkness. These findings provide evidence that MID is not a target of COP1, but rather a potential regulator.

In line with this, mid mutants exhibit constitutive photomorphogenesis reminiscent of a COP1 loss-of-function phenotype, whereas knock-out mutants of COP1 targets usually show the opposite phenotype, i.e. reduced photomorphogenesis. Interestingly, constitutive photomorphogenesis has previously been observed in topoVI mutants (Yin et al., 2002). These phenotypes, however, have not been associated with COP1-dependent photomorphogenesis, but have been attributed to a role of the TOPOVI complex in modulating the expression of brassinosteroid-dependent genes (Yin et al., 2002). Our finding of a physical interaction of the TOPOVI-activating MID protein with COP1 suggests that the link to the regulation of photomorphogenesis is direct rather than indirect. Consistent with the idea that MID is a positive regulator of COP1 function rather than a target, double mutant analysis of weak mid and weak cop1 mutants revealed enhancement of the cop1 mutant phenotype in the absence of a functional MID protein.

COP1 activity is critically dependent on regulated nucleo-cytoplasmic shuttling and specific accumulation in nuclear bodies (von Arnim and Deng, 1994; Stacey and von Arnim, 1999; Stacey et al., 1999). We have shown that accumulation of COP1 in nuclear bodies is enhanced by the presence of MID, which may be a consequence of MID-induced increased amounts of COP1 protein in the nucleus. The two parameters determining COP1 accumulation in the nucleus are protein stability and the ratio between nuclear import and export. COP1 protein stability requires the function of its RING domain, which is involved in auto-ubiquitylation and degradation (Seo et al., 2003). The MID-interacting region (COP11–67) overlaps with the first Zn binding motif of the RING-finger motif. Our observation of an increased prevalence of COP1 nuclear bodies in the presence of MID suggests that the interaction with MID might slow down nuclear export and, by covering loop I of the RING domain, might interfere with auto-ubiquitylation, leading to a stabilization of COP1 in the nucleus. This conclusion is consistent with our results from analysis of the weak cop1 allele cop1eid6. While being hypersensitive to light, cop1eid6 does not exhibit a constitutive photomorphogenetic phenotype when grown for 4 days in darkness. The mutant phenotype presumably is based upon more rapid nuclear export and/or enhanced auto-ubiquitylation of the COP1EID6 protein (Dieterle et al., 2003). The single amino acid exchange in COP1EID6 neither disrupts the interaction with MID nor abolishes its MID-dependent sub-nuclear accumulation. The mid-1cop1eid6 double mutants exhibit phenotypes resembling those of stronger cop1 loss-of-function mutants, consistent with the idea that absence of the MID protein function in cop1eid6 mutant background further destabilizes the already labile nuclear COP1 EID6 protein.

In the cop1eid6 mutant endoreduplication proceeds comparable to wild-type, and cells with DNA contents up to 64C can be found. This finding indicates that this particular mutation of COP1 does not significantly affect its endoreduplication-controlling function. The weak mid-1 mutant is mildly affected in endoreduplication, apparent mainly in the absence of cells with a DNA content of 64C. However, introduction of the mid-1 allele into cop1eid6 leads to significantly reduced endoreduplication, with only a small percentage of cells completing the third endoreduplication cycle (16C) and a complete absence of 32C and 64C. These double mutants prove that at least part of the function of COP1 in promoting endoreduplication in the skotomorphogenic development of A. thaliana seedlings (Gendreau et al., 1998) is dependent on the presence of a functional MID protein and therefore, presumably on a signaling process involving TOPOVI. The cop1-4 mutant, in contrast with cop1eid6, exhibits strongly reduced endoreduplication in dark-grown hypocotyl cells, which is only mildly enhanced by additional loss-of-function of MID in the mid-1cop1-4 double mutant. The interpretation of this phenotype has to take into account that in the cop1-4 mutant an N-terminal part of the COP1 protein is expressed. The mutant, therefore, combines loss-of-function phenotypes due to the missing C-terminus with a proven dominant negative function based on the presence of the truncated N-terminus (McNellis et al., 1996). The endoreduplication phenotype in cop1-4 can thus be interpreted on the one hand as a function of the COP1 protein that independently of MID promotes endoreduplication, or on the other hand, a dominant negative repression of regulators of endoreduplication. The partial COP1 protein expressed in the cop1-4 mutant contains the complete MID-interaction domain and, therefore, it is easily conceivable that one such regulator of endoreduplication might be the MID protein. Future studies are needed to clarify the exact mechanisms of these two possibilities.

Interestingly, our results on the light-dependent differentiation of root and hypocotyl growth suggest that the COP1-4 protein, in the presence of MID seems to be more functional than COP1EID6. Apparently, it is not the WD40 domain that is essential for this COP1 function but possibly a functional RING domain.

On the molecular level, our findings suggest a role for the MID protein in the sub-nuclear accumulation and stability of COP1 and thereby in enhancing COP1 activity. A general impact on COP1 function can explain the aberrant photomorphogenesis phenotypes of the mid loss-of-function mutants. With respect to endoreduplication, the MID protein might act as an adaptor or mediator between COP1 and the TOPOVI complex.

However, we cannot exclude alternative pathways such as COP1-independent control of photomorphogenesis by MID, and MID-independent control of endoreduplication by COP1. These pathways could include regulation of photomorphogenesis based upon changes in gene expression influenced by the activity of the MID/TOPOVI complex, and the ubiquitylation of inhibitors of the TOPOVI complex or other endoreduplication-promoting factors by COP1, respectively.

The plant TOPOVI complex is essential for the decatenation of replicated DNA during endocycles (Sugimoto-Shirasu et al., 2002), thereby preventing replication-induced DNA damage, which accumulates in TOPOVI as well as in MID loss-of-function mutants (Hartung et al., 2002; Sugimoto-Shirasu et al., 2005; Breuer et al., 2007; Kirik et al., 2007). Similarly, cop1 mutants accumulate DSBs in dark-grown root tissue (Dohmann et al., 2008). These roots have not been exposed to potentially damaging light/UV. Therefore, DNA damage is probably caused by unfavorable topological situations during replication, transcription or endoreduplication. Prevention of these situations is precisely the proposed function of MID and TOPOVI (Breuer et al., 2007; Kirik et al., 2007). It remains to be investigated whether the endoreduplication-promoting function of COP1 is indirect by preventing DNA damage, possibly in co-operation with MID/TOPOVI or whether COP1 acts more directly in endocycle initiation and progression.

It is tempting to speculate that MID and COP1 act in concert to form a molecular switch or signal integration platform linking endoreduplication and photomorphogenesis. COP1 is expressed ubiquitously throughout development (Hruz et al., 2008) and is inactivated by light, while MID is expressed in young endoreduplicating tissue (Breuer et al., 2007; Kirik et al., 2007). The suggested switch may therefore be active only in tissues and under conditions when both complexes are present in the nucleus, sensing each other's integrity via direct protein–protein interaction. This is the case in the hypocotyl, where far-red and white light-induced photomorphogenesis involves the suppression of endoreduplication exceeding the second cycle (>8C) (Gendreau et al., 1997). This suppression of endoreduplication is released in darkness in a COP1-dependent fashion (Gendreau et al., 1998) and is also dependent on a functional MID protein. Co-operation of MID and COP1 might therefore allow proper skotomorphogenic development including elongation of the hypocotyl, and the proceeding through several additional rounds of endoreduplication.

In order not to accumulate replication-induced DNA damage it may be critical for the hypocotyl cells to sense or ensure the presence and integrity of TOPOVI before proceeding to rapid endoreduplication. In this context, the COP1–MID interaction might be the molecular link that coordinates signaling pathways in darkness, with the machinery preventing replication-induced DNA damage.

Experimental procedures

Plant material, growth conditions and phenotypic characterization

The following A. thaliana mutants were used: mid-1, mid-2 (Kirik et al., 2007), rhl2, hyp6, (Santoni et al., 1997; Sugimoto-Shirasu et al., 2002) cop1-4 (McNellis et al., 1994) and cop1eid6 (Dieterle et al., 2003). Seeds were sterilized with 70% (v/v) ethanol, 2% NaOCl, placed on MS agar without sucrose and stratified (2–3 days, 4°C). Germination was induced with white light and the plates were kept in the dark at 21°C for 7 days, if not stated otherwise. Double mutants were generated as described earlier (Kirik et al., 2007), genotypes were verified by PCR (Table S5). pEarleyGate201–MID (Kirik et al., 2007), pEarleyGate104–MID and pNmR–COP1 were stably transformed via A. tumefaciens-mediated floral dip transformation (Clough and Bent, 1998) into Col-0, cop1-4, mid-1 or mid-2 for rescue and MID stability experiments. The Binocular Leica MZ FL III was used for fluorescence analyses. Pictures for Figures 4 and 5 were taken with a digital camera Canon EOS 5D Mark. Hypocotyl and root length were measured using Image J 1.41o (Wayne Rasband, NIH, rsb.info.nih.gov/ij/). Petiole and lamina angles of the cotyledons were measured manually using printouts of appropriate size. The petiole angle was defined as the angle between two lines drawn through the centers of the two petioles. The lamina angle was defined as the angle between two lines drawn through the base and tip of the lamina. Plants from 2 to 4 independent plates were analysed.

Determination of anthocyanin accumulation

For this procedure, 160–240 mg of 7-day-old dark-grown seedlings (2% sucrose) were extracted with 18% (v/v) 1-propanol, 1% (v/v) concentrated HCl, and measured with a NovaSpecII spectrophotometer (Pharmacia LKB Biotech, www.gelifesciences.com) as described by Lange et al. (1971) and Schmidt and Mohr (1981).

RT-PCR

Total RNA was isolated from 3-day-old dark-grown seedlings using innuPREP Plant RNA Kit (Analytik Jena, www.analytik-jena.de/). cDNA was prepared with the RevertAid™ H Minus First Strand cDNA Synthesis Kit (Fermentas, http://www.thermoscientific.com/fermentas) and treated with RNase H (Martel et al., 2002). A reaction mix that lacked the reverse transcriptase was used as a negative control. The exponential phase of amplification was determined for CHS and UBQ10 (Table S5). UBQ10 served as a control (Harari-Steinberg et al., 2001).

Constructs

The CDS of MID, HY5 (both Col-0 cDNA), COP1, COP11–67, GFP (pBat-TL–GFP, JFU, unpublished), RFP–HA–attB1 and YFP–attB1 were cloned into pDONR207 (Invitrogen, www.invitrogen.com). To obtain pDONR207–COP1K550E, an inverse PCR was applied using pDONR207–COP1 and primers listed in Table S5. The PCR product was DpnI digested and ligated. Destination vectors used in this work are pAS2-1–attR, pACT–attR (Clontech, http://www.clontech.com/, modified), pCL112, pCL113, pBatTL-B-p35s, pNmR (A. Schrader and J.F. Uhrig, unpublished data), pEarleyGate104 (Earley et al., 2006), and pENSG–CFP (Wenkel et al., 2006). pBatTL-B-p35s was created by cutting (SpeI) and re-ligating pBatTL-B-sYFPN (J.F. Uhrig, unpublished data). To generate pNmR, RFP–HA was amplified from pGJ2811 (Jach et al., 2006), cut with BamHI and BglII, and ligated into the BglII site pBatTL-B-p35s (Table S5).

Yeast two-hybrid screening

COP1 protein (Hoecker and Quail, 2001) was used as bait to screen an A. thaliana cDNA (Nemeth et al., 1998). Yeast two-hybrid screening was performed according to Soellick and Uhrig (2001). Interaction candidates were verified individually. Auto-activation tests and YTH assays were done by co-transformation with GFP as described previously (Gietz and Schiestl, 2007).

BiFC and particle bombardment

Epidermal cells of Allium porrum leaves or mid-2 rosette leaves were biolistically transformed with PDS-1000/He instrument (Bio-Rad, http://www.bio-rad.com). CFP–TALIN served as transformation control (Saedler et al., 2004). Samples were protected from light (0.002 μmol) and analysed after 24 h of incubation in the dark at room temperature (leek) or 2 days at 21°C under LD conditions (mid-2). Finally, 80–200 transformed cells from at least two independent transformations were analysed.

Tobacco infiltration

Leaves of Nicotina benthamiana were transiently co-infiltrated with supervirulent A. tumefaciens strain LBA4404.pBBR1MCS.virGN54D (pEGATE104–MID) or LBA4404pBBR1MCS-5.virGN54D (van der Fits et al., 2000) and an Agrobacterium strain that expressed the silencing suppressor TBSV19K (Voinnet et al., 1999) as described earlier (Gigolashvili et al., 2007). Plants were kept for 4 days at 24°C at LD conditions prior to analysis.

Co-immunoprecipitation (Co-IP)/protein stability

Protein extracts were prepared from infiltrated leaves of N. benthamiana and used for Co-IP with anti-GFP MicroBeads (Miltenyi Biotec, http://www.miltenyibiotec.com) with subsequent protein gel blotting and immunodetection (rat anti-HA or mouse anti-GFP (Roche, http://www.roche-applied-science.com), goat anti-rat or goat anti-mouse (Jackson ImmunoResearch, http://www.jacksonimmuno.com).

A rescued mid-2 line expressing YFP–MID was used to test MID stability in 7-day-old seedlings grown under LD conditions (16 h light, ~120 μmols−1 m−2/8 h dark, 21°C) on MS plates that lacked sucrose. Samples were harvested at the end of the 16 h light period and after an additional 8 h of darkness. The same line was crossed with cop1-4 to analyse MID stability in adult plants. Lines were grown on MS plates [0% sucrose, 20 mg/mL glufosinate-ammonium (Dr Ehrenstorfer, http://www.analytical-standards.com/)] under LD conditions (16 h light, ~80 μmols−1 m−2/8 h dark, 21°C). Three plants were pooled at the end of a light cycle and after additional 8 h of darkness. Plant material was analysed by western blotting with standard methods. As a loading control, blots were stripped [0.1 m glycine, pH 2.4 (HCl)] after GFP/HA detection and probed with mouse anti-Hsc70 [mAb (5B7)] (Enzo Life Sciences, http://www.enzolifesciences.com/) and goat anti-mouse (Jackson ImmunoResearch, http://www.jacksonimmuno.com).

Ploidy measurements

The DNA content of a cell is referred to as ‘C’, with 2C representing the diploid genome. Completed endoreduplication cycles result in duplication of the genome leading to DNA contents of 4C, 8C, 16C and so on. DNA content of hypocotyl epidermal cells excluding stomata of 7-day-old dark-grown seedlings was determined after DAPI staining using a Leica DMRB fluorescence microscope (Leica, http://www.leica-microsystems.com). DAPI staining was done as described earlier (Gendreau et al., 1997). Seedlings were transferred for 45 min to 70% ethanol prior to analysis. Four representative dark-grown seedlings from two independent plates were analysed. The mean of three measurements per nucleus was calculated. Fluorescence of 15–22 stomata guard cells of the cotyledons was determined per plant and served as a 2C reference.

CLSM

Confocal laser-scanning microscopy (CLSM) was performed using a Leica TCS-SP2 confocal microscope (DMRE7). For two different fluorescence channels sequential scanning was applied. Z-stacks were merged.

Statistical analysis

For statistical analysis, the Kolmogorov–Smirnov test was applied to test for normal distribution (threshold: 5%). In case of normal distribution, the Welch test was applied for datasets with equal and unequal variance as suggested before (Ruxton, 2006). When the assumption of normal distribution had to be rejected, the Wilcoxon rank sum test was used. For germination analysis, Fisher's exact test was applied. Results are listed in Table S3.

Acknowledgements

We thank Ulrike Temp for excellent technical assistance and Siegfried Werth for help with photography. The work was supported by grants from the Deutsche Forschungsgemeinschaft and the Bundesministerium für Bildung und Forschung (BIODISC-WIZPLANT).

Competing interest

The authors declare no competing financial interests.

Ancillary