Auxin modulates the transition from the mitotic cycle to the endocycle in Arabidopsis

The development of multicellular organisms requires mechanisms that pattern cell proliferation and cell differentiation with temporal and spatial precision. The requirement of such patterning strategies is, for example, evident in plant root meristems, where stem cells undergo cell division and differentiation in a highly ordered fashion (De Veylder et al., 2007; Dello Ioio et al., 2008a). It is now well established that the plant hormone auxin forms concentration gradients in developing plant organs and that different cellular auxin levels are translated into distinct cellular responses, such as cell proliferation, cell expansion and cell differentiation (Benjamins and Scheres, 2008; Vanneste and Friml, 2009). Recent studies have demonstrated that both local auxin biosynthesis and intercellular auxin transport contribute to the establishment and maintenance of active auxin gradients (Vanneste and Friml, 2009; Zhao, 2008). Central players of the auxin polar transport are the group of auxin efflux carrier proteins encoded by the PIN gene family (Galweiler et al., 1998; Paponov et al., 2005). Genetic analyses of multiple pin mutants illustrate that the combinatory actions of several PIN proteins play pivotal roles in producing auxin gradients in the Arabidopsis root meristem, with the maximum concentration being located in the quiescent centre and young columella cells (Blilou et al., 2005). These pin mutants also display smaller root meristems, demonstrating the requirement of active auxin distribution in maintaining the root meristem (Blilou et al., 2005). Other key regulators crucial for the development of Arabidopsis root meristems are the members of the PLETHORA (PLT) gene family, which encode AP2-domain transcription factors, and which also operate in a concentration-dependent manner to maintain the balance between cell proliferation and cell differentiation (Aida et al., 2004; Galinha et al., 2007). The PLT proteins act redundantly to control the expression of the PIN genes (Galinha et al., 2007) while, in turn, several PIN proteins together modulate the expression of the PLT genes (Blilou et al., 2005), pointing to a robust feedback mechanism to sustain the dynamic auxin gradients. Recent studies have also uncovered that another class of plant hormones, the cytokinins, acts antagonistically to auxin to stimulate cell differentiation in the Arabidopsis root meristem (Dello Ioio et al., 2007). The proposed model suggests that this process is mediated through the AHK3/ARR1 two-component signalling pathway downregulating PIN expression and thus normal auxin transport (Dello Ioio et al., 2008b; Ruzicka et al., 2009).

Cellular auxin levels are first perceived by the TRANSPORT INHIBITOR RESPONSE 1 (TIR1)/AUXIN-SIGNALING F BOX (AFB)-class of F-box proteins that function as auxin receptors (Dharmasiri et al., 2005a; Dharmasiri et al., 2005b; Kepinski and Leyser, 2005). At high concentrations, auxin is thought to stabilise the interaction between TIR1/AFB and AUXIN (AUX)/INDOLE-3-ACETIC ACID (IAA) transcriptional repressors that constitute the core auxin signalling pathway together with auxin response factor (ARF) transcription factors (Mockaitis and Estelle, 2008). The auxin-dependent stabilisation of TIR1/AFB-AUX/IAA interaction promotes the ubiquitination of AUX/IAA by SCF^TIR1/AFB E3 ligases and their subsequent proteolysis, resulting in an activation of ARF transcription factors and thus an up- or downregulation of their target gene expression. As auxin concentrations decrease, AUX/IAA become more stable and increase their interactions with ARF to repress their transcriptional activities. What remains to be established is how these relatively short auxin signalling cascades mediate so many different cellular responses, including cell proliferation, cell expansion and cell differentiation, which presumably involve the activation or repression of completely different sets of genes and subsequent cellular activities. The Arabidopsis genome encodes 29 AUX/IAA and 23 ARF (Mockaitis and Estelle, 2008), and the current challenge is to test how far numerous different combinations of AUX/IAA-ARF heterodimers, potentially formed under different developmental or environmental conditions, can account for the diverse transcriptional responses to auxin.

Another key cellular event that often accompanies the switch from cell proliferation to cell differentiation in growing plant organs is the transition from the mitotic cell cycle to an alternative cell cycle, called the endoreduplication cycle or endocycle. Cells undergo sequential phases of DNA replication and cell division in the mitotic cycle, whereas in the endocycle cells replicate genomic DNA without intervening cell divisions, thus leading to an increase in the total nuclear DNA content, ploidy, within a cell. In Arabidopsis, most postembryonic cells, except those in meristems, guard cell lineages or meiotic tissues, endoreduplicate several times and their ploidy levels usually reach 16C to 32C (Galbraith et al., 1991). Exact functions of endoreduplication have not been established, but the correlation that is often found between the level of endocycles and cell size has suggested that an increase in ploidy by endocycling plays an important role in cell expansion and/or cell differentiation (Inze and De Veylder, 2006; Nagl, 1978; Sugimoto-Shirasu and Roberts, 2003). Ectopic induction of endocycling in Arabidopsis meristem cells causes severe developmental defects, such as small and abnormally shaped leaves (Verkest et al., 2005), clearly demonstrating that tight controls over the mitotic-to-endocycle transition are prerequisites for normal postembryonic plant development. As in all other eukaryotes, the progression of mitotic cycles in plants is controlled by cyclin-dependent kinases (CDKs) and their interacting partners, cyclins (CYCs), which regulate the kinase activity of CDKs. Plants possess two classes of CDKs that directly control the cell cycle, CDKA, which has functional orthologues in yeasts, and CDKB, which has been identified only in the plant kingdom. Previous studies implicate CDKA for both the G1-to-S and the G2-to-M transition, whereas CDKB appears to function only for the G2-to-M transition (Inze and De Veylder, 2006). The switch from the mitotic cycle to the endocycle usually involves the downregulation of CDK activities at the G2-to-M transition to skip mitoses while still permitting DNA replication (Edgar and Orr-Weaver, 2001; Larkins et al., 2001). Accumulating evidence suggests that the downregulation of both CDKA and CDKB is vital to promote this transition in plants (De Veylder et al., 2007; Inze and De Veylder, 2006). Recent molecular and genetic studies have identified several key regulators, such as E2F/DEL transcription factors and Kip-related proteins (KRPs) that modulate the expression and/or activities of these mitotic CDKs, respectively (De Veylder et al., 2007; Inze and De Veylder, 2006). Given that cells often transit into endocycles at well-defined time points and positions during plant development (Beemster et al., 2005), the decision of endocycle initiation is likely to be developmentally programmed. The candidates that can provide such developmental cues include various plant hormones, and one of the most direct ways in which they might work is that such upstream regulators modulate the expression and/or activities of the core cell cycle regulators to initiate the endocycle. To date, however, very little is established about the identities these upstream regulators and how they function in the cell cycle transition.

In this study we demonstrate that TIR1-AUX/IAA-ARF-mediated auxin signalling acts as a repressor of the mitotic-to-endocycle transition in Arabidopsis. We show that mutations in auxin biosynthesis, transport or signalling all lead to an early transition into endocycling. Our physiological experiments using exogenous auxin and its antagonists demonstrate that lowering auxin levels is sufficient to promote an entry into endocycles both in planta and in vitro cell culture systems. Reduced auxin signalling rapidly downregulates the expression of several mitotic cycle genes and their upregulation can partially override the premature cell differentiation under low auxin conditions. These data strongly suggest that transcriptional repression of the mitotic cycle regulators constitutes an early cellular response to modified auxin signalling with downstream consequences in cell cycle transition and cell differentiation.

Arabidopsis mutants and transgenic lines used in this study were provided by Eva Benkova [monopteros (mp) mutants (Hardtke and Berleth, 1998), bodenlos (bdl) mutants (Hamann et al., 2002), auxin resistant 2-1 (axr2-1) mutants (Nagpal et al., 2000), HS:axr3-1 lines (Knox et al., 2003), and pinoid (pid) mutants (Friml et al., 2004)], Hiroo Fukuda [vascular network 3 (van3) mutants (Koizumi et al., 2005)], Yunde Zhao [yucca 1 (yuc1) yuc4 yuc10 yuc11 mutants (Cheng et al., 2007)], Ben Scheres [Histone 2B (H2B)-YFP and LTI6B-GFP lines (Campilho et al., 2006)], Dirk Inze [CDKA;1pro:GUS lines (Hemerly et al., 1993)], Takashi Aoyama [CYCA2;3pro:GUS and CYCA2:3-GFP lines (Imai et al., 2006)], and Tatsuo Kakimoto [isopentenyltransferases 3 (ipt3) ipt5 ipt7 mutants (Miyawaki et al., 2006)]. The transgenic lines with CDKB2;1pro:GUS or CYCB1;1-GUS were previously described by Adachi et al. (Adachi et al., 2006) and Breuer et al. (Breuer et al., 2007). The mp, bdl and van3 mutants were in Lansdberg erecta (Ler) ecotype and axr2-1, HS:axr3-1, pid, yuc1 yuc4 yuc10 yuc11 and ipt3 ipt5 ipt7 mutants were in Columbia (Col) ecotype. Seeds were surface sterilized in 70% ethanol for 1 minute, then in 20% (v/v) sodium hypochlorite for 5 minutes, and rinsed three times in sterile water. Sterilized seeds were plated on Murashige and Skoog (MS) media supplemented with 1% (w/v) sucrose and 0.5% Gelrite. After cold treatment in the dark for 2 days, seeds were incubated under continuous light at 22°C. For the ploidy and cell-size analysis of leaves, seedlings were grown on horizontally placed plates. For the ploidy and cell-size analysis of roots, seedlings were grown on vertically placed plates.

Arabidopsis MM2d cells were cultured in MS media supplemented with 200 mg/l potassium dihydrogenphosphate, 100 mg/l myo-inositol, 1 mg/l thiamine hydrochloride, 3% (w/v) sucrose, and 0.2 mg/l 2,4-D (pH 5.8). The MM2d cell culture was incubated at 27°C in continuous darkness and subcultured every 7 days.

For chemical treatments, 1-naphthaleneacetic acid (NAA), BH-IAA [also referred to as probe-8 in Hayashi et al. (Hayashi et al., 2008b)], PEO-IAA and 6-Benzylaminopurine (6-BA) were dissolved in dimethyl sulfoxide (DMSO) as a stock solution of 1 mM, 100 mM, 50 mM and 0.1 mM, respectively, and diluted into the MS medium.

To induce transient expression of axr3-1 from the heat shock promoter, transgenic plants harbouring the HS:axr3-1 construct were first incubated at 22°C on the MS media. At 4 days after germination, plants were transferred to the 40°C growth chamber for 1 hour and returned to the 22°C growth chamber thereafter. This 1-hour heat-shock treatment was applied every 12 hours and plants were harvested for observation at various time points from 2 days after the first heat-shock treatment.

Ploidy levels were measured using the ploidy analyzer PA-I (Partec) as described previously (Sugimoto-Shirasu et al., 2002). Briefly, nuclei were released in Cystain extraction buffer (Partec) from fresh cotyledons or true leaves by chopping lightly with a razor blade, filtered through a CellTrics filter (Partec), and stained with Cystain fluorescent buffer (Partec). At least 5000 nuclei isolated from ~30 cotyledons or true leaves were used for each ploidy measurement. Flow cytometry experiments were repeated at least three times for each genotype using independent biological replicates.

For the visualization of nuclei in cotyledons, 7-day-old seedlings were fixed following the protocol described by Sugimoto et al. (Sugimoto et al., 2000) and stained with 4′,6-diamidino-2-phenylindole (DAPI). Root nuclei werevisualised as above or by using transgenic plants harbouring H2B-YFP fusion protein constructs. Nuclei in the Arabidopsis MM2d cell culture were fixed in methanol:acetic acid (3:1) and stained with DAPI. Root meristem organisations were visualised using seedlings stained with 5 mg/ml propidium iodide or in transgenic plants harbouring LTI6B-GFP fusion protein constructs. Samples were observed using either an Olympus BX51 fluorescence microscope or a Carl Zeiss LSM510 META confocal laser microscope. To estimate the initiation of endocycling and cell expansion in the root meristem, three to six confocal optical sections, which were collected at approximately 1-μm intervals using identical confocal settings, were merged. Surface areas of individual nuclei and lengths of individual cells within same epidermal cell files were measured using ImageJ software (NIH).

Histochemical β-glucuronidase (GUS) staining of roots was performed as described by Ishida et al. (Ishida et al., 2009).

Auxin forms active concentration gradients in developing plant organs to promote various cellular processes, including cell proliferation, cell expansion and cell differentiation (Benjamins and Scheres, 2008; Vanneste and Friml, 2009). To investigate an involvement of auxin in regulating the switch from the mitotic cycle to the endocycle, we first tested whether a disruption of auxin signalling results in an impaired mitotic-to-endocycle transition. The ARF5/MP transcription factor promotes the expression of auxin response genes, and loss-of-function mutations in the MP gene severely disrupt auxin signalling (Hardtke and Berleth, 1998). Our flow cytometry analysis showed that at 7 days after germination, the ploidy level of approximately 40% of nuclei in cotyledons remained at 2C or 4C, whereas all other nuclei progressed through several rounds of endocycling, allowing them to reach 8C or 16C (Fig. 1A). By contrast, the proportion of both 2C and 4C nuclei was severely reduced in 7-day-old mp mutant cotyledons (Fig. 1A), implying an early exit from the mitotic cycle. The mp mutants also displayed significantly increased 16C and 32C ploidy peaks (Fig. 1A), suggesting the promotion of successive endocycling in mp mutants. By 14 days, mp mutant cells had undergone further rounds of endocycling, giving rise to 64C and 128C nuclei that are not normally present in wild-type plants at the same developmental stage (Fig. 1B). Furthermore, the ploidy distribution resulting from gain-of-function mutations of IAA7/AXR2, IAA17/AXR3 or IAA12/BDL genes, which cause constitutive repression of auxin signalling (Hamann et al., 2002; Nagpal et al., 2000; Rouse et al., 1998), was similar to that observed in mp mutants, demonstrating a decrease in 2C and 4C peaks and an increase in 32C, 64C and, in some cases, 128C and 256C peaks (Fig. 1B). Consistent with these altered ploidy distributions, visualisation of DAPI-stained nuclei revealed the presence of >32C nuclei in 7-day-old cotyledon epidermal cells of mp, bdl, axr2-1 and axr3-1 mutants (Fig. 1C).

To further substantiate the role of auxin in mediating the switch from mitotic cycles to endocycles, we examined the ploidy distribution of plants defective in auxin biosynthesis or its transport. The flavin monooxygenases encoded by the YUC gene family are key enzymes involved in tryptophan-dependent auxin production (Zhao et al., 2001), and correspondingly yuc1 yuc4 yuc10 yuc11 quadruple mutants display pleiotropic auxin deficient phenotypes (Cheng et al., 2007). Our flow cytometry revealed that yuc1 yuc4 yuc10 yuc11 mutants displayed reduced 2C and 4C peaks and increased 64C and 128C peaks (see Fig. S1 in the supplementary material). Consistently, loss-of-function mutations in PID and VAN3, both of which result in impaired auxin distribution and response (Friml et al., 2004; Koizumi et al., 2005), caused similar ploidy alterations (Fig. 1B).

Our data suggest that low levels of auxin signalling promote entry into endocycles. Given the major role that auxin plays in the cell fate specification during the Arabidopsis embryogenesis, however, it is also possible that impaired cell specification and/or pattern formation in mutant embryos indirectly leads to a premature exit from the mitotic cycle. To distinguish between these possibilities, we determined whether post-embryonic manipulation of auxin levels could directly modify the cell cycle transition. Externally added auxin is known to promote cell proliferation in plants. Consistently, more than 90% of 14-day-old wild-type leaf cells grown in the presence of synthetic auxin NAA (1 μM) contained 2C or 4C nuclei, whereas at least 20% of cells grown without NAA switched into the endocycle, reaching 8C or 16C (Fig. 2A). Within 2 days of removing NAA from the growth media, however, the proportion of 2C and 4C nuclei started decreasing, and by 4 days more than 30% of leaf cells possessed 8C or 16C nuclei, whereas only ~10% of cells reached 8C in leaves that remained exposed to NAA (Fig. 2B,C). These observations are consistent with our hypothesis that high auxin levels repress the transition into the endocycle; however, they do not exclude the possibility that the delayed entry into endocycling is an indirect consequence of other developmental changes caused by auxin. To test whether blocking auxin signalling directly converts mitotic cells into endocycling cells, we used Arabidopsis MM2d suspension culture cells, which normally possess 6C nuclei in G1 phase (Menges and Murray, 2002). As shown in Table 1, 100% of mock-treated MM2d culture cells remained in the mitotic cycle for up to 5 days, carrying only 6C and 12C nuclei. By contrast, ~10% of cells treated with an auxin antagonist BH-IAA (50 μM), which specifically blocks TIR1-mediated auxin signalling (Hayashi et al., 2008b), entered the endocycle and increased their ploidy to 24C. Consistently, our microscopic observation demonstrated that all of the cells we examined in mock-treated MM2d cell culture possessed nuclei of a relatively uniform size, which presumably corresponds to 6C or 12C, whereas a small but significant proportion (approximately 13%, n=240) of BH-IAA-treated cells possessed 24C nuclei (Fig. 3). Arabidopsis MM2d culture cells tend to aggregate to form large cell clusters (Menges and Murray, 2002), and cells with 24C nuclei in BH-IAA-treated cell culture were found only at the periphery of these cell clusters (Fig. 3), suggesting that the intrinsic heterogeneity of these cell culture might contribute to the relatively low frequency of 24C nuclei. Together, these results strongly suggest that lowering auxin signalling directly initiates endocycling in Arabidopsis postembryonic cells.

The final step of cell differentiation in plants is often marked by a rapid increase in cell size by cell expansion (De Veylder et al., 2007; Dello Ioio et al., 2008a). We, therefore, tested whether auxin-mediated progression of endocycling is coupled with such postmitotic cell expansion. Application of 1 μM NAA strongly repressed the expansion of Arabidopsis leaf mesophyll cells, and within 2 days of removing NAA from the growth media cells started to increase their size (Fig. 2D). By 6 days, cells grown in the absence of NAA were approximately three times larger than those that were left exposed to 1 μM NAA (Fig. 2D-F). Similarly, we found that BH-IAA-treated MM2d cells that had 24C nuclei had a significantly larger cell size (1037±234.5 μm², n=10) compared with cells with 6C or 12C nuclei (266.4±76.7 μm², n=42; Fig. 3). These data clearly demonstrate that auxin-dependent endocycling is tightly coupled with postmitotic cell expansion in Arabidopsis.

Auxin concentration gradients direct pattern formation in Arabidopsis roots (Benjamins and Scheres, 2008; Vanneste and Friml, 2009). A maximum auxin concentration in the root stem-cell niche promotes cell proliferation, whereas in the region distal to the maximum, where lower auxin levels are predicted, cells expand and differentiate (Blilou et al., 2005; Sabatini et al., 1999). To test whether auxin-mediated modulation of the mitotic-to-endocycle switch accompanies these developmental transitions, we determined whether lowering auxin levels triggers an early entry into endocycling at the root meristem. Compromised auxin gradients reduce the root meristem size (Blilou et al., 2005) and, correspondingly, our confocal microscopy revealed that the first endocycle initiated closer to the quiescent centre in wild-type roots with H2B-YFP (Campilho et al., 2006) treated with either 50 μM BH-IAA (data not shown) or another auxin antagonist PEO-IAA (10 μM) which also exhibits specific inhibition of TIR1-mediated auxin signalling (Hayashi et al., 2008a) (Fig. 4). Consistently, we found that root meristem cells transited into endocycling prematurely in axr3-1 roots (Fig. 4). In addition, visualisation of the propidium iodide (PI)-stained root meristem structures showed that blocked auxin signalling induced an early postmitotic cell expansion in both PEO-IAA-treated wild-type roots and axr3-1 roots (see Fig. S2 in the supplementary material). These cellular responses are strictly dependent on auxin deprivation, as the transient induction of dominant-negative axr3-1 mutations by a heat shock promoter (Knox et al., 2003) was sufficient to induce premature endocycling and cell expansion (Fig. 5). These results indicate that auxin is required to prevent an early onset of endocycling and accompanying cell expansion in the Arabidopsis root meristem.

Cytokinins regulate the root meristem size through modulating polar auxin transport (Dello Ioio et al., 2008b; Ruzicka et al., 2009). The application of cytokinins promotes an early transition from cell proliferation to cell differentiation in the Arabidopsis root meristem, whereas mutations in cytokinin biosynthesis or signalling delay this transition. To test whether cytokinins also impact the mitotic-to-endocycle transition, we examined whether the application of exogenous cytokinins (e.g. 6-BA) modifies the timing of endocycle onset. As shown in Fig. S3A-C in the supplementary material, 0.1 μM 6-BA induced endocycling closer to the quiescent centre compared with mock-treated wild-type roots. Conversely, entry into endocycling was significantly delayed in atipt3 atipt5 atipt7 triple mutants, which have mutations in ATP/ADP isopentenyltransferases which are involved in the biosynthesis of cytokinins (Miyawaki et al., 2006) (see Fig. S3D-F in the supplementary material). These results suggest that cytokinins act antagonistically to auxin in mediating the mitotic-to-endocycle transition in the Arabidopsis root.

To gain insight into the molecular basis underlying the auxin-mediated endocycle transition, we tested whether a transient reduction in auxin signalling modifies the expression of mitotic cell cycle genes. Transgenic plants carrying CDKA;1pro: GUS constructs (Hemerly et al., 1993) or CDKB2;1pro:GUS constructs (Adachi et al., 2006) displayed strong GUS activities in 6-day-old Arabidopsis root meristems (Fig. 6A,B), reflecting their high promoter activities. We found that reducing auxin signalling by the application of 20 μM PEO-IAA led to a rapid reduction of these GUS signals. The reduced GUS activities were detectable within 1 hour of PEO-IAA treatment, and by 3 hours the signal intensities were greatly weakened throughout the meristem (Fig. 6A,B). Importantly, the root meristem size was not yet reduced by 3 hours (Fig. 6A,B), indicating that these modified GUS activities are not due to the compromised meristems. The GUS activities driven by the CYCA2;3 promoter (Imai et al., 2006) were also strong in the root meristems but were significantly reduced within 3 hours of PEO-IAA treatment (Fig. 6C).

We also examined the effect of PEO-IAA on the accumulation level of mitotic cyclin CYCB1;1 by using transgenic lines carrying CYCB1;1-GUS constructs in which the expression of CYCB1;1-GUS fusion proteins is driven by the CYCB1;1 promoter (Colon-Carmona et al., 1999). As shown in Fig. 6D, CYCB1;1-GUS signals are normally detected in a punctate pattern across the root meristem. Both the signal intensity and the number of CYCB1;1-GUS-positivecells appeared to be comparable between mock-treated roots and PEO-IAA-treated roots within the first 1 hour of PEO-IAA treatment but by 3 hours the number of CYCB1;1-GUS-positive cells dropped substantially in PEO-IAA-treated roots (Fig. 6D). Our quantitative RT-PCR analysis, using RNA isolated from 6-day-old roots, further supported that the inhibition of auxin responses by PEO-IAA causes significant reduction in the expression of CYCA2;3 and CYCB1;1 genes by 3 hours (Fig. 6E). These results strongly suggest that the transcriptional and/or post-transcriptional repression of these cell cycle regulators is a part of the early cellular responses to reduced auxin signalling in the Arabidopsis root meristem.

Recent studies using transgenic lines harbouring the β-oestradiol-inducible CYCA2;3-GFP overexpression construct show that CYCA2;3 is one of the key mitotic cycle regulators that suppress endocycle onset (Imai et al., 2006; Boudolf et al., 2009). Consistently, we found that both entry into endocycling and the initiation of postmitotic cell expansion were moderately delayed in 6-day-old CYCA2;3-GFP roots grown in the presence of 10 μM β-oestradiol from germination (see Fig. S4 in the supplementary material). To test whether the downregulation of mitotic cycle genes was responsible for the early transition into cell differentiation under low auxin conditions, we transiently induced CYCA2;3-GFP expression by 1-day exposure to β-oestradiol in the absence or presence of PEO-IAA. As shown in Fig. 7, 1-day incubation of 5-day-old CYCA2;3-GFP roots with 10 μM β-oestradiol did not modify meristem size in the absence of PEO-IAA. By contrast, 1-day co-treatment of CYCA2;3-GFP roots with both β-oestradiol and PEO-IAA partially overrode the premature initiation of postmitotic cell expansion induced by PEO-IAA (Fig. 7). These results support that an early transcriptional change of cell cycle genes can provide an instructive signal for cell differentiation in the Arabidopsis root.

We have demonstrated that auxin has a physiological role in coordinating the developmental transition from the mitotic cycle to the endocycle in Arabidopsis. Our model (see Fig. S5 in the supplementary material) predicts that high auxin levels, e.g. at the proximal root meristem, serve to repress endocycles, thus promoting the mitotic phase. By contrast, as cells depart from the meristem, cellular auxin levels are lowered and thus the repression is lifted, allowing cells to switch into endocycles. Our genetic and physiological data suggest that the mitotic-to-endocycle transition is mediated by the TIR1-AUX/IAA-ARF-dependent auxin signalling pathway. That auxin promotes the mitotic cycle is well established but what is striking in our finding is that depleting auxin signals does not simply block the mitotic cycle but also leads to the initiation of an alternative cell cycle, i.e. the endocycle, thus still permitting plant organs to grow as a whole. This role of auxin as a repressor of endocycling has not been demonstrated in the plant organ context before; the only previous study that has reported similar effects of auxin is in the cell culture context, where depletion of 2,4-D from the culture media induced endocycling and a corresponding elongation of tobacco Bright Yellow-2 cells (Magyar et al., 2005). Our study not only supports their finding but further demonstrates that this function of auxin is vital for programming postembryonic plant development. Interestingly, Magyar et al. (Magyar et al., 2005) found that auxin positively regulates the stability of E2FB transcription factors but that this auxin action does not depend on the TIR1-mediated auxin signalling.

Physiological roles of auxin in promoting cell proliferation have been extensively studied in various plant species but surprisingly little is established about the molecular mechanisms underlying these responses. For example, the application of exogenous auxin promotes the expression of many cell cycle genes (John, 2007), but it is not known whether some members of the AUX/IAA-ARF transcription factors directly control their expression in response to auxin. Previous studies have shown that auxin is also involved in regulating the stability of E2F proteins (del Pozo et al., 2002; Magyar et al., 2005), which suggests that part of the auxin-dependent cell cycle control might rely on their post-transcriptional modification. However, how direct this regulation is in relation to the existing auxin signalling pathway is not established. How auxin controls the mitotic-to-endocycle transition at the molecular level also remains unclear but our data suggest that transient inhibition of auxin signalling by PEO-IAA rapidly downregulates the expression of key mitotic cycle genes. Importantly, reduced activities of these genes have been implicated for endocycle initiation (Andersen et al., 2008; Verkest et al., 2005; Imai et al., 2006; Boudolf et al., 2009) and the promoter sequence of CDKB2;1 contains a putative auxin responsive element (Vanneste et al., 2005), which raises the possibility that transcriptional repression of these cell cycle regulators by AUX/IAA-ARF constitutes a part of the primary responses to altered auxin levels. We have recently identified a novel SUMO E3 ligase HIGH PLOIDY2 (HPY2) that functions in the PLT-dependent pathway to mediate endocycle onset and meristem development in Arabidopsis (Ishida et al., 2009). It is thus plausible that HPY2-dependent sumoylation constitutes part of the developmental programme linking auxin signalling and downstream cell cycle progression.

It is intriguing that all of the auxin mutants we have tested also exhibit enhanced endoreduplication phenotypes with their ploidy reaching up to 256C, indicating that the termination of endocycling is an active process requiring regulatory mechanisms and that auxin signalling is involved in this control. The hpy2 mutants also display similar ploidy phenotypes, i.e. reduced 2C and 4C peaks and increased 32C, 64C and 128C peaks (Ishida et al., 2009), suggesting that the entry into and exit from endocycling might be linked processes. Finally, we need to add that our data are also consistent with the idea that endoreduplication occurs ‘by default’ in plant organs and that cell proliferation is promoted only where stimulating signals, such as auxin, are provided (e.g. at the root meristem). Further elucidation of the molecular basis governing the mitotic-to-endocycle transition should help us to uncover how plants use these two alternative cell cycles during their postembryonic development.

Auxin signalling intersects with signalling pathways mediated by other plant hormones and, in particular, recent elegant studies have begun to unravel the molecular basis underlying the multilevel crosstalk between auxin and cytokinin in the Arabidopsis root meristem. The current model suggests that cytokinins counteract the effects of auxin by downregulating the expression of auxin transport genes, thus promoting cell expansion/differentiation (Dello Ioio et al., 2008b; Ruzicka et al., 2009). Consistently, our data show that cytokinins act antagonistically to auxin in mediating the endocycle transition. Although this model seems to fit well for Arabidopsis root development, existing data also suggest that auxin and cytokinin are not the only developmental signals mediating endocycle control in other plant organs. For example, ethylene and gibberellin have been suggested to control endocycle progression in the Arabidopsis hypocotyl (Gendreau et al., 1999). We found that several of the auxin mutants we have examined, including yucca quadruple mutants that give strong ploidy phenotypes in cotyledons, do not display major ploidy alterations in hypocotyls (T.I. and K.S., unpublished), suggesting that endocycle control in Arabidopsis hypocotyls does not require inputs from auxin signalling. Therefore, the upstream signalling pathways involved in endocycle control might differ in various plant organs and it is likely that different combinations of plant hormones take part in this control.

This study shows that the auxin-mediated switch into endocycling is tightly coupled with the developmental transition from cell proliferation to cell expansion and/or cell differentiation. Our data suggest that the downregulation of mitotic activities precedes by far the modification of cell expansion/differentiation under low auxin conditions and that these cell cycle changes can have downstream consequences in cell expansion/differentiation. There are also other examples reported where cell cycle genes provide instructive signals for cell expansion/differentiation (Walker et al., 2000; Verkest et al., 2005; Churchman et al., 2006; Boudolf et al., 2009). It is thus tempting to speculate that the dose-dependent control of the two alternative DNA replication cycles, mitotic cycles and endocycles, is part of the regulatory mechanisms that translate auxin gradients into distinct cellular responses — cell proliferation and cell expansion — in developing plant organs. We need to note that the correlations between ploidy levels and plant cell size are not absolute (Inze and De Veylder, 2006; Sugimoto-Shirasu and Roberts, 2003), and we cannot rule out the possibilities that low auxin triggers endoreduplication and cell expansion independently, or that endoreduplication is a downstream consequence of cell expansion. However, in the context of endocycle or cell expansion initiation, we think these are both less likely situations, as existing evidence points to a very tight coupling between endoreduplication onset and the initiation of cell expansion in various plant cell types (Inze and De Veylder, 2006; Sugimoto-Shirasu and Roberts, 2003), although there are some cases in which final ploidy levels are uncoupled from final cell sizes (Beemster et al., 2002; Gendreau et al., 1998). Likewise, there are numerous experimental data reported for ploidy-dependent cell expansion (e.g. Sugimoto-Shirasu et al., 2005; Breuer et al., 2007; De Veylder et al., 2002), whereas almost no examples have been found so far for cell expansion-induced endoreduplication.

In summary, this study unveils a novel function of auxin in plant cell cycle control and provides a new molecular framework in which to further investigate how growth factors mediate cell differentiation during postembryonic development of multicellular organisms.

We are grateful to Eva Benkova for various auxin mutants, Ben Scheres for the H2B-YFP and LTI6B-GFP lines, Hiroo Fukuda for van3 mutants, Yunde Zhao for yuc1 yuc4 yuc10 yuc11 mutants, Tatsuo Kakimoto for ipt3 ipt5 ipt7 mutants, Dirk Inze for CDKA;1pro:GUS lines, Takashi Aoyama for CYCA2;3pro:GUS and CYC2;3-GFP lines, Kenichiro Hayashi for BH-IAA and PEO-IAA, and Bayer BioScience N.V. for the MM2d cells. We thank Yoko Okushima for helpful discussions, and Keith Roberts, Michael Lenhard and members of the Sugimoto Lab for their critical comments on the manuscript. This work was supported by the Japan Society for the Promotion of Science, as grants to K.S. (grant numbers 20061028 and 20687004) and as a postdoctoral fellowship and a grant to T.I. (grant number 20-8597)