Cytokinin response factors integrate auxin and cytokinin pathways for female reproductive organ development

In Arabidopsis, ovules emerge as lateral organs from the placenta, a meristematic tissue that originates after the fusion of the carpel margin meristem (CMM) (Reyes-Olalde et al., 2013; Schneitz et al., 1995). Placenta formation and ovule growth require auxins. Reduced local auxin biosynthesis or transport causes severe defects in pistil development with a consequent loss of placental tissue and ovules (Nemhauser et al., 2000; Nole-Wilson et al., 2010). The auxin efflux carrier PIN-FORMED 1 (PIN1) is one of the main elements modulating auxin accumulation during all phases of ovule development (Benková et al., 2003; Ceccato et al., 2013). Although pin1-201 does not develop any flowers, in the pin1-5 mutant the gynoecium has shorter valves and contains a few ovules (Bencivenga et al., 2012; Sohlberg et al., 2006).

Cytokinins (CKs) positively regulate ovule formation and pistil development. Indeed, mutants that have a reduced capacity for CK production or perception exhibit a dramatic reduction in ovule number and pistil size, and compromised female fertility (Kinoshita-Tsujimura and Kakimoto, 2011; Riefler et al., 2006; Werner et al., 2003). By contrast, increased CK levels result in a bigger pistil with a greater number of ovules compared with wild type, confirming a positive correlation between CK levels and ovule numbers (Bartrina et al., 2011; Bencivenga et al., 2012; Galbiati et al., 2013). It has been shown that CK treatment positively influences the number of ovules per pistil via a strong increase in PIN1 expression (Bencivenga et al., 2012; Galbiati et al., 2013; Zúñiga-Mayo et al., 2014).

Conversely, in roots, CKs modulate organogenesis by downregulating PIN1 expression (Dello Ioio et al., 2012; Ruzicka et al., 2009) and PIN1 protein endocytic recycling (Marhavý et al., 2011).

Cytokinin response factors (CRFs) are encoded by closely related members of the Arabidopsis AP2 gene family and mediate a large proportion of the CK transcriptional response that functionally overlaps with the B-type ARR-mediated response (Rashotte et al., 2006). Recently, Šimášková and colleagues (2015) found that CRFs bind directly to the PIN1 cytokinin response element (PCRE1) in the PIN1 promoter and thus modulate PIN1 expression in response to CKs. Deletion of the PCRE1 cis-regulatory element uncouples PIN1 transcription from CRF regulation and affects root sensitivity to CKs (Šimášková et al., 2015). Here, we show that CRF2, CRF3 and CRF6 redundantly induce the expression of PIN1, which is required for ovule development, supporting the crucial and general role of CRF factors as mediators of auxin-CK crosstalk guiding plant organogenesis.

In the placenta, CKs promote PIN1 expression , which is needed for the establishment of the auxin gradient that leads to ovule primordia development (Bencivenga et al., 2012; Benková et al., 2003; Ceccato et al., 2013; Galbiati et al., 2013). Recently, it has been shown that three members of the CRF family, CRF2, CRF3 and CRF6, directly regulate PIN1 expression upon CK signalling in roots (Šimášková et al., 2015). CRF2 and CRF6 promoters were able to drive reporter gene expression in stage 9 and 10 of pistil development, whereas the CRF3 promoter did not show any activity (Fig. S1). These results are consistent with recently published transcriptomic data of the gynoecial medial domain, which show high expression of CRF2 and CRF6 and low expression of CRF3 (Villarino et al., 2016).

To investigate whether these three CRFs control PIN1 expression during early stages of pistil development, we have analysed crf2, crf3 and crf6 single, double and triple mutants. Ovule counts were performed on ovules from stage 1-II (primordia) to stage 2-I (finger-like), which corresponds to stages 9 and 10 of pistil development [according to Schneitz et al. (1995) and Roeder and Yanofsky (2006)]. Analysis of single crf3 and crf6 mutants, as well as the crf3 crf6 double mutant, did not reveal any significant difference in ovule number compared with wild type, whereas the single crf2 and the double crf2 crf3 mutant showed a small but significant decrease in ovule number (Fig. S2). Instead, the crf2 crf6 double mutant presented ovule numbers comparable to wild type, suggesting a compensatory mechanism between crf2 and crf6 (Fig. S2). Finally, in the crf2 crf3 crf6 (crf2/3/6) triple mutant, a reduction of 31.68% in ovule number was observed with respect to the wild type (Fig. 1A). Wild-type Col-0 plants grown under long-day conditions developed on average (mean±s.e.m.) 46.36±1.24 ovules per pistil whereas 31.67±2.01 ovules were formed in the crf2/3/6 triple mutant pistils (Fig. 1A).

Placenta length was measured at the same developmental stages. In the wild type, the average length of the placenta was found to be 351±12 µm at stage 9 and 517±12 µm at stage 10, whereas in the crf2/3/6 mutant it was significantly shorter (269±20 µm at stage 9 and 436±19 µm at stage 10) (Fig. 1B). Ovule density (number of ovules per µm placenta) was also reduced in the crf2/3/6 mutant (Fig. 1C).

CK treatment results in an increase in pistil size and ovule number (Bencivenga et al., 2012; Galbiati et al., 2013). Because CRFs regulate the transcriptional response to cytokinins (Rashotte et al., 2006), we tested the CK response in wild type and the crf2/3/6 triple mutant. Wild-type plants treated with the synthetic cytokinin 6-benzylaminopurine (BAP) yielded 60% more ovules and a 58% longer placenta than untreated plants (Fig. 1D,E). The crf2/3/6 triple mutant treated with BAP produced 19% more ovules and an increase in placenta length of 32% (Fig. 1D,E), indicating that the capacity to respond to CKs is strongly reduced in the absence of CRF2, CRF3 and CRF6 activities.

To investigate whether the pistil phenotypes observed in the crf2/3/6 mutant were due to changes in PIN1 expression, we performed real-time qPCR experiments. In the crf2/3/6 triple mutant, PIN1 expression was significantly lower than in the wild type (Fig. 2A). As previously reported (Bencivenga et al., 2012; Galbiati et al., 2013), PIN1 expression was at least twofold higher in BAP-treated wild-type inflorescences. Interestingly, the level of PIN1 mRNA in the crf2/3/6 mutant did not increase upon CK application, suggesting that CRFs are required for CK-dependent PIN1 expression (Fig. 2A).

In roots, CRFs regulate PIN1 expression by binding the PCRE1 sequence in the PIN1 promoter (Šimášková et al., 2015); therefore, we analysed plants carrying a ΔPIN1::PIN1-GFP construct in which a PIN1 promoter lacking the PCRE1 element drives the expression of a fully functional PIN1-GFP fusion protein. Real-time qPCR experiments were performed on GFP instead of PIN1 in order to avoid the detection of endogenous PIN1. The level of PIN1-GFP transcripts in ΔPIN1::PIN1-GFP inflorescences was lower than that in plants carrying the same fusion protein construct under the control of a wild-type version of the PIN1 promoter (Fig. 2B). The reduction in PIN1-GFP expression under control of the ΔPIN1 promoter was also evident by confocal microscopy in placenta cells and ovule primordia at stages 1-I and 1-II (compare Fig. 2C,D with Fig. 2E,F). Although PIN1 expression was dramatically reduced (Fig. 2B), PIN1-GFP protein in ΔPIN1::PIN1-GFP plants was correctly localized at the membrane of placenta cells (Fig. 2E,F).

To understand whether PCRE1 is the only element in the PIN1 promoter required for CK-mediated PIN1 expression in inflorescences, we also analysed GFP expression after treatment with CKs in PIN1::PIN1-GFP and ΔPIN1::PIN1-GFP plants. Interestingly, GFP expression increased in both PIN1::PIN1-GFP and ΔPIN1::PIN1-GFP inflorescences compared with the control (mock treatment) (Fig. 2B), suggesting that CRFs might bind to other regions of the PIN1 promoter besides PCRE1. The possibility that other CK-induced transcription factors regulate PIN1 expression is unlikely as PIN1 expression remains unchanged in CK-treated crf2/3/6 inflorescences (Fig. 2A). The same reduction of GFP expression in ΔPIN1::PIN1-GFP compared with PIN1::PIN1-GFP was observed in a second independent ΔPIN1::PIN1-GFP line (Fig. S3). Also, in the independent line ΔPIN1::PIN1-GFP_2, GFP expression increased after BAP treatment, reconfirming the results obtained with line 1 (Fig. S3). These results confirm that CRFs are required to regulate the expression of PIN1 in the pistil. The possibility of other CRF regulatory regions needs to be investigated as the lack of PCRE1 does not cause complete CK insensitivity. It is important to recall that in roots PIN1::PIN1-GFP expression is reduced by CKs and that ΔPIN1::PIN1-GFP is completely CK insensitive (Šimášková et al., 2015), indicating that there might be a specific regulation of PIN1 expression depending on the developmental context.

Introducing ΔPIN1::PIN1-GFP in a wild-type A. thaliana does not lead to any abnormalities in pistil and ovule development (Fig. 3A-C). To examine the functional significance of the CRF regulatory regions in the PIN1 promoter (PCRE1), we introgressed ΔPIN1::PIN1-GFP into the pin1-5 mutant. pin1-5 is a hypomorphic mutant that has shorter pistils and develops an average of nine ovules per pistil (Bencivenga et al., 2012; Sohlberg et al., 2006). Confirmation of the presence of ΔPIN1::PIN1-GFP construct in the pin1 mutant is shown in Fig. S4.

PIN1::PIN1-GFP completely rescued the pin1-5 mutant phenotype whereas ΔPIN1::PIN1-GFP was unable to rescue the pistil growth phenotype of pin1-5 (Fig. 3A-C). The placenta length of pin1-5 ΔPIN1::PIN1-GFP remained the same as in pin1-5 (Fig. 3A,B). Placenta length in pin1-5 ΔPIN1::PIN1-GFP was similar to that of the crf2/3/6 mutant (Fig. 3A,B). This suggests that PCRE1-mediated transcriptional regulation of PIN1 is necessary for correct elongation of the pistil. Furthermore, ovule density in pin1-5 ΔPIN1::PIN1-GFP (0.0902±0.008 ovules/µm placenta) was similar to that of crf2/3/6 (0.0926±0.004 ovules/µm placenta). By contrast, ΔPIN1::PIN1-GFP did rescue the ovule number phenotype of pin1-5, raising the ovule count of pin1-5 from an average of 8.5±1.7 to 28.67±1.84 (Fig. 3C).

These results suggest that PCRE1-mediated control of PIN1 expression is required for determining the correct size of the pistils, whereas it seems to be less relevant for ovule formation. However, it should be taken into account that transcription of pin1-5 (which encodes a partially functional protein) could be induced by CKs. For this reason, we also analysed the phenotype of ΔPIN1::PIN1-GFP in pin1-201 mutant. This mutant fails to develop any lateral organs due to a loss-of-function mutation (Fig. S5). Pistil length in ΔPIN1::PIN1-GFP pin1-201 is similar to that in ΔPIN1::PIN1-GFP pin1-5 and crf2/3/6 (Fig. 2A,B). Regarding the ovule number, ΔPIN1::PIN1-GFP pin1-201 showed a reduction in comparison with ΔPIN1::PIN1-GFP pin1-5 and crf2/3/6 (Fig. 3C). The reduction in ovule number highlighted in ΔPIN1::PIN1-GFP pin1-201 compared with ΔPIN1::PIN1-GFP pin1-5 might be explained by residual function of the PIN1-5 mutant protein. The analysis of ΔPIN1::PIN1-GFP in both pin1-5 and pin1-201 allelic backgrounds confirmed that pistil elongation is affected when PIN1 expression is uncoupled from regulation of CRFs. Finally, we also tested the capacity of ΔPIN1::PIN1-GFP pin1-5 and ΔPIN1::PIN1-GFP pin1-201 to respond to CK by checking the number of ovules after BAP treatment. Interestingly, both lines are still able to respond to CK showing an increase in ovule density of 27% and 21%, respectively (Fig. 3D). This result is in agreement with the fact that PIN1-GFP expression level increases in ΔPIN1::PIN1-GFP after BAP treatment (Fig. 2B; Fig. S3), confirming the importance of CRF-mediated PIN1 expression for pistil elongation.

The reduction in pistil size observed in crf mutants could be due to defective cell division or cell expansion processes or a combination of both. Auxin plays a prominent role in controlling cell expansion. For example, elongation of the primary root and the hypocotyl require specific auxin transport to determine their expansive growth rates (Rayle and Cleland, 1992; Spartz et al., 2012). Interestingly, a reduction in pistil and anther elongation has been reported for tir1 afb1 afb2 afb3, a quadruple mutant with compromised auxin signalling (Cecchetti et al., 2008). Our understanding of the influence of auxin on the cell cycle is still fragmentary, but primary evidence indicates that auxin acts on several targets involved in the control of cell cycle (Perrot-Rechenmann, 2010). On the other hand, the ability of CKs to promote cell division, in particular through their action on D-type cyclins, was described several years ago (Dewitte et al., 2007; Riou-Khamlichi et al., 1999), and it has been recently been shown that the transcript levels of several cell cycle-related genes were decreased in roots of the crf1,3,5,6 quadruple mutant (Raines et al., 2016).

In summary, we propose that PIN1 expression mediated by CRFs is required for the determination of pistil size. The greater number of ovule primordia in CK-treated pistils correlates with the increased pistil size. Therefore, it is likely that when enough space occurs between two ovules, CRFs and/or other CKs-dependent factors induce PIN1 expression to create a new auxin maximum.

Arabidopsis wild-type and mutant plants were grown at 22°C under long-day conditions (16 h light, 8 h dark) in a greenhouse. crf2-2 seeds (Schlereth et al., 2010) were provided by Dolf Weijers (Laboratory of Biochemistry, Wageningen University, 6703 HA Wageningen, The Netherlands). PIN1::PIN1-GFP (Benková et al., 2003), pin1-5 mutant (Bencivenga et al., 2012; Sohlberg et al., 2006), pin1-201 (Furutani et al., 2004), crf3-1, crf6-S2, crf3-1 crf6-S2, crf2-2 crf3-1 crf6-S2, ΔPIN1::PIN1-GFP, ΔPIN1-GFP pin1-201 and PIN1::PIN1-GFP pin1-201 (Šimášková et al., 2015) lines have been described previously. BAP treatment was performed on inflorescences as detailed by Bencivenga et al. (2012).

Total RNA was extracted from inflorescences at pre-fertilization stages using the Macherey-Nagel Nucleospin RNA Plant Kit and then reverse transcribed using the Promega ImProm-II RT System. Gene expression analysis was performed using the Bio-Rad iQ5 Multicolor RT-PCR Detection System with the GeneSpin SYBR Green PCR Master Mix. ACTIN 2-8 and UBIQUITIN 10 were used as reference genes for normalization of transcript levels. RT-PCR primers used in this work were: RT2017fw 5′-TGTTCCATGGCCAACACTTG-3′ and RT2018rev 5′-AAGTCGTGCCGCTTCATATG-3′ for GFP, RT509fw 5′-TGGTCCCTCATTTCCTTCAA-3′ and RT510rev 5′-GGCAAAGCTGCCTGGATAAT-3′ for PIN1, RT147fw 5′-CTGTTCACGGAACCCAATTC-3′ and RT148rev 5′-GGAAAAAGGTCTGACCGACA-3 for UBI10, and RT861fw 5′-CTCAGGTATTGCAGACCGTATGAG-3′ and RT862rev 5′-CTGGACCTGCTTCATCATACTCTG-3′ rev for ACT2-8.

Inflorescences were fixed with ethanol/acetic acid (9:1) overnight, rehydrated with 90% and 70% ethanol and cleared in a chloral hydrate/glycerol/water solution (8 g: 1 ml: 3 ml) for at least 2 h before dissection under a stereomicroscope. Pistils were observed using a Zeiss Axiophot D1 microscope equipped with DIC optics. Images were recorded using a Zeiss Axiocam MRc5 camera with Axiovision software version 4.1. Only ovules of pistils in which both carpels remained intact after slide preparation and where all four rows of ovules were visible and distinguishable were counted.

For confocal laser scanning microscopy (CLSM), fresh material was collected, mounted in water and analysed immediately. CLSM analysis was performed using a Leica TCS SPE microscope with a 488 nm argon laser line for excitation of GFP fluorescence. Emissions were detected between 505 and 580 nm. Images were collected in multi-channel mode and overlay images were generated using Leica analysis software LAS AF 2.2.0.

We would like to thank Andrew MacCabe and Edward Kiegle for editing the paper.