STIMPY mediates cytokinin signaling during shoot meristem establishment in Arabidopsis seedlings

Flowering plants initiate most of their organs post-embryonically from groups of pluripotent cells residing in the meristems. The primary meristems originate from embryogenesis, when the shoot apical meristem (SAM) and the root meristem are patterned and set aside (Steeves and Sussex, 1989). The patterning of the meristems has been extensively studied in Arabidopsis in the last two decades (reviewed in Baurle and Laux, 2003; Bowman and Floyd, 2008; Carles and Fletcher, 2003; Petricka and Benfey, 2008; Veit, 2004). Apart from the meristem-specific mechanisms, it has been demonstrated that general cell cycle regulators are involved in the maintenance of both the shoot and the root meristems (Blilou et al., 2002; Cockcroft et al., 2000; Riou-Khamlichi et al., 1999; Wildwater et al., 2005; Willemsen et al., 1998).

Although patterned at the embryonic stage, the meristems are only fully established during seedling development. The SAM enters active proliferation after germination, reaching its full size by the end of vegetative development (Medford et al., 1992). In the root, the fast-dividing meristematic zone also goes through a similar expansion following germination. Compared with the patterning aspect, mechanisms underlying the growth and establishment of the primary meristems in the emerging seedlings are less well understood.

One possible regulator of this process is the plant hormone cytokinin, which refers to a family of adenine derivatives with zeatin being the most abundant form. Cytokinins were first discovered for their ability to promote cell division in tissue culture (Miller et al., 1955) and have been linked to a wide range of developmental events (reviewed in Kyozuka, 2007; Mok and Mok, 2001). In intact plants, whereas the endogenous levels of cytokinins promote cell division, exogenously applied cytokinins in higher concentrations result in growth inhibition. This concentration-dependent effect is also reflected in how the primary meristems respond to changes in cytokinin signaling levels. Earlier studies have shown that in plants with partial loss of cytokinin availability or sensing, the two primary meristems often respond in opposite manners, leading to the hypothesis that cytokinins promote shoot growth while negatively regulating the size of the root meristem (Riefler et al., 2006; Werner et al., 2003). However, seedlings lacking all three cytokinin receptors show reduction in both the shoot and the root meristems, suggesting that low levels of cytokinin signaling positively regulate meristematic cell proliferation in both the above- and under-ground tissues (Higuchi et al., 2004; Nishimura et al., 2004), and the root meristem is inhibited by a lower threshold of cytokinins than the SAM (To and Kieber, 2008). Recently, it has been demonstrated that cytokinins regulate the size of the root meristems by controlling differentiation (Dello Ioio et al., 2007).

At the molecular level, cytokinin signal transduction is initiated by cytokinin binding to its two-component element receptors, the ARABIDOPSIS HISTIDINE KINASES (AHKs), triggering a cascade of phosphorelay events involving the ARABIDOPSIS RESPONSE REGULATORS (ARRs). The ARR proteins are divided into two groups based on their amino acid sequences and functional domains (D'Agostino and Kieber, 1999). The type-B ARR proteins, which are activated by changes in their phosphorylation state, positively regulate cytokinin responses by activating the transcription of their downstream targets, including the type-A ARR genes. The type-A ARR proteins then downregulate the cytokinin responses through a negative-feedback loop (reviewed in To and Kieber, 2008). More recently, a group of AP2 transcription factors, the CYTOKININ RESPONSE FACTORS (CRFs), have been found to be regulated by cytokinin signaling and involved in embryonic and cotyledon development (Rashotte et al., 2006).

The cytokinin signaling network interacts with known regulators of meristem development. In the shoot meristem, WUSCHEL (WUS) (Laux et al., 1996; Lenhard and Laux, 1999) represses the transcription of several type-A ARRs, thereby enhancing cytokinin activities (Leibfried et al., 2005). In the root, cytokinin signaling interacts with the auxin pathway in the establishment of the embryonic root meristem (Muller and Sheen, 2008). However, mutations in known cytokinin signaling targets do not phenocopy the defects seen in cytokinin perception mutants in their above-ground development (Argyros et al., 2008; Hutchison et al., 2006; Mason et al., 2005), implying additional mechanisms linking cytokinin signaling to shoot meristem development.

Previously, we reported that STIMPY [STIP; also known as WUSCHEL HOMEOBOX 9 (WOX9)] (Haecker et al., 2004; Wu et al., 2005), a homeodomain transcription factor that is related to WUS, plays a unique role in maintaining pluripotency and proliferation in meristematic tissue in Arabidopsis. At the seedling stage, the loss of STIP function does not cause any disruption in meristem patterning. Instead, the meristematic tissues in stip mutant seedlings exit the cell cycle and enter premature differentiation, resulting in the arrest of seedling development. This phenotype can be fully rescued by the addition of sucrose in the growth media. Genetically, STIP exerts a positive influence on WUS in regulating the stem cell population in the SAM (Wu et al., 2005). Here, we present evidence that cytokinin signaling is essential for early seedling development and that STIP acts downstream of cytokinin-sensing in the establishment of the SAM during this period.

Plants were grown in long days (16 hours light/8 hours darkness) under about 120 μE m^–2 sec^–1 light at 22°C. To observe seedling phenotypes, seeds were germinated on 1/2 Murashige Minimal Organics Medium (MS; Phytotechnology Lab) with 0.6% agar after two days of stratification at 4°C. Sucrose (1.5%) was added to the media when the sugar effect was assayed.

stip-1 and stip-2 were described previously (Wu et al., 2005). The ARR5::GUS reporter was obtained from the Arabidopsis stock center, and crossed to stip-2/+. The phenotypes were analyzed in F2.

Cytokinin sensitivity was assayed by transferring two-day-old Col-0 and stip-1 seedlings germinated on MS agar to MS agar with 1.5% sucrose and various concentrations of trans-zeatin (Sigma Z0876). The phenotypes were recorded after seven days.

In zeatin induction experiments, two-day-old seedlings were treated with 250 nM trans-zeatin on MS agar (3 ml per 100 mm plate). The control plates received 3 ml of water with 0.3 μl of acetic acid (same amount as in 3 ml of 500 nM trans-zeatin) each. Samples were incubated for 30 minutes before harvesting. For the zeatin-induction time course, two-day-old seedlings were treated with 250 nM trans-zeatin on MS agar, and the 0-minutes plates received no trans-zeatin.

pAS1 was generated by inserting the 5.3 kb EcoRI/BamHI plasmid rescued genomic fragment from stip-D (Wu et al., 2005) into the transformation vector pJHA212G (Yoo et al., 2005). The segregating population of ahk2-2 ahk3-3/+ cre1-12 was transformed using floral dipping (Weigel and Glazebrook, 2002). Transgenic plants were selected on MS agar with gentamicin (50 μg/ml). Seven independent single-insertion lines in ahk2-2 ahk3-3/+ cre1-12 were selected from a total of 24 in the T2 generation and grown to T3, when plants homozygous for the pAS1 insertion were analyzed for their ability to rescue the ahk2-2 ahk3-3 cre1-12 mutants. All seven lines showed the same partial rescue effect.

The ethanol-inducible XW119 was generated in two steps. First, full-length STIP cDNA was introduced into the EcoRI site of pBJ36-AlcA. Then the NotI fragment of AlcA::STIP:OCSt was inserted into the transformation vector pBart-35S::AlcR (Roslan et al., 2001). Transgenic plants were selected on soil using Finale (Bayer).

Wholemount in situ analysis was performed as previously described (Friml et al., 2003), except 10 μg/ml of proteinase K (Sigma-Aldrich # P2308) was used for digestion. For in situ hybridization on tissue sections, two-day-old seedlings were fixed in 4% paraformaldehyde and 0.1% Tween-20 in PBS for 5 hours, then dehydrated and infiltrated with paraffin using an H2850 Microwave Processor (Energy Beam Sciences) as previously described (Schichnes et al., 1999) with the following modifications: sections were digested with 20 μg/ml of proteinase K (Sigma-Aldrich # P2308) for 15 minutes, washed with 0.2× SSC for 1 hour at 55°C and the RNase treatment step was omitted. The STIP antisense RNA probe was described in Wu et al. (Wu et al., 2005). The antisense ARR5 RNA probe was generated using the last exon of ARR5 as the template.

GUS activity staining was carried out as previously described (Sessions et al., 1999), using 2 mM potassium ferro and ferri cyanide. The GUS-stained seedlings were mounted in 30% glycerol for analysis. The SAM sectioning and staining was done according to Roeder et al. (Roeder et al., 2003).

Tissue sections, in situ hybridization and GUS-stained samples were photographed on a Zeiss Axio Imager equipped with an AxioCam HRc camera. Seedling samples were photographed using an Olympus SZ61 stereo microscope with a Micrometrics camera. All comparative samples were taken under identical magnifications.

For the ethanol induction of STIP, four-day-old XW119 seedlings were treated with 2% ethanol on MS agar (3 ml/100 mm plate) and incubated at 22°C in light for 0, 1, 2 and 4 hours before harvesting. The 0-hour sample received no ethanol. Total RNA was extracted using the Spectrum Total Plant RNA (Sigma, STRN50-1KT). Biotin-labeled cDNA probes were generated using the One-Cycle Target Labeling and Control Reagents (Affymetrix) and hybridized to the ATH1 expression arrays. Each timepoint was done in triplicate.

Raw intensity data was pre-processed and normalized using gcRMA (Wu et al., 2004) and genes with unreliably low expression levels (<60) were filtered out. Differentially expressed genes were then identified using RankProd (Hong et al., 2006) at a P-value of 0.01. The raw data for the XW119 ethanol induction experiment is available at ArrayExpress, accession number E-MEXP-2229.

Total RNA was extracted using Spectrum Total Plant RNA (Sigma). The first-strand cDNA was obtained using SuperScript III First Strand (Invitrogen, 18080-051). Real-time PCR was carried out with the SYBR Green (Molecular Probes) method on Opticon-2 MJ machines. The relative changes in gene expression levels were determined using the 2^–ΔΔCT method. Each treatment was done in at least two biological replicates, each with technical replicates. The primers used in this study are: ARR5, 5′-tcgata^*tagatttagagggtattgtttg-3′ and 5′-ttaacggtaactttatcacaaatcctaa-3′; ARR6, 5′-tttgag^*gatttgaaggttaatttaatt-3′ and 5′-caccggtacttctctaaaagctga-3′; ARR7, 5′-cgtatacaaga^*atgtctcaaagaagga-3′ and 5′-taagattttgcattcctcagctt-3′; ARR9, 5′-actgatagtaacgacccaaatgc-3′ and 5′-atatctctaaaagctgatgattc^*cttg-3′; CRF2, 5′-tttccacctagagaagaagaagagtatc-3′ and 5′-gtttataacatagagattagggaaagcaaa-3′; CRF5, 5′-tgagtggtttgatgacaacg-3′ and 5′-cagataaataa^*aaactccgttgcg-3′; At4g27410, 5′-tctggtcttttacgccggaaaa-3′ and 5′-cccaatcatccaa^*cttggagct-3′; At1g68670, 5′-aggagcgtgta^*atttgcaacc-3′ and 5′-agctttaatcaaatcacttttaccacc-3′; ARR1, 5′-cggaagaac^*gatggacttacctg-3′ and 5′-gttactccctttagaaccacactcttt-3′; ARR10, 5′-gactatggagcaagaaattgaagtct-3′ and 5′-ttcgttgttgtaa^*cgtgatattgg-3′; ARR12, 5′-ggatatgaga^*aggctatgcctaaa-3′ and 5′-ctgattagccacaccactgatc-3′; STIP, 5′-tcaagttcaagcggacgca-3′ and 5′-caacctcttccccaaatgca-3′; Act8, 5′-ttccagcag^*atgtggatctcta-3′ and 5′-agaaagaaatgtgatcccgtca-3′. The Act8 primers have been previously described (Puthiyaveetil and Allen, 2008) and the asterisk (^*) indicates exon-intron boundaries.

Previous studies found that Arabidopsis seedlings lacking all three cytokinin receptors, AHK2, AHK3 and CRE1 (AHK4), have reduced shoot and root meristems and impaired growth (Higuchi et al., 2004; Nishimura et al., 2004). However, in both cases, the seedlings were initially grown on sucrose-containing media, which can potentially promote meristem growth. We therefore examined the phenotype of the cytokinin triple receptor mutants without exogenously supplied sugar. One week after germinating on sugarless media, the wild-type seedlings had already produced true leaves and lateral roots (Fig. 1A). By contrast, the ahk2-2 ahk3-3 cre1-12 (Higuchi et al., 2004) seedlings were still similar in size to two-day-old wild-type seedlings and the cotyledons failed to fully expand and turn green (Fig. 1B, compare with 1A inset). This phenotype indicates that cytokinin signaling is essential for early seedling development, and is reminiscent of that of the stip mutants, which enter complete growth arrest following germination without exogenous sugar (Fig. 1C) (Wu et al., 2005). The growth-arrested ahk2-2 ahk3-3 cre1-12 seedlings could resume limited growth when transferred to sucrose-containing media, producing the phenotypes described previously (data not shown) (Higuchi et al., 2004; Nishimura et al., 2004).

The similarity of stip and ahk2-2 ahk3-3 cre1-12 seedling phenotypes led us to examine the relationship between STIP and cytokinin signaling. To identify molecular events downstream of STIP, we generated transgenic plants (XW119) in which STIP expression throughout the plant was under the control of the ethanol-inducible alc system (Caddick et al., 1998; Roslan et al., 2001; Salter et al., 1998). After 1 hour of treatment with 2% ethanol, the STIP expression level increased by 73-fold. When we examined the mRNA profile of these plants with ATH1 expression arrays, we found that approximately 800 genes were differentially regulated compared with those of the uninduced plants (P<0.01).

Previously, in an effort to characterize the cytokinin signaling pathway, Brenner and colleagues (Brenner et al., 2005) expression-profiled wild-type seedlings treated with 5 μM benzyladenine (BA), a cytokinin analog, for 15 and 120 minutes. Using the same criteria as for the XW119 arrays, we identified genes responsive to BA treatments in their experiment. Of the 109 genes that were differentially expressed after the 15 minute BA treatment, 31 were also affected by elevated STIP expression (P=0.00087 based on the hypergeometric test), including 13 transcription factor genes (Table 1). Unfortunately, most of these genes have not been studied for their involvement in meristem development. Furthermore, 107 of the 475 genes that were affected after the 120 minute BA treatment were also found to respond to STIP induction (P=2.6e-54, based on the hypergeometric test; see Table S1 in the supplementary material), suggesting partial overlaps between the STIP and cytokinin pathways.

As sucrose can restore stip mutant seedlings to normal growth (Wu et al., 2005), but fails to completely rescue the ahk2-2 ahk3-3 cre1-12 mutants (Higuchi et al., 2004; Nishimura et al., 2004), we hypothesized that STIP acts downstream of cytokinin perception during seedling development, and that part of the cytokinin triple-receptor mutant phenotype might be due to the lack of STIP activity. To test this model, we examined STIP expression in two-day-old wild-type and ahk2-2 ahk3-3 cre1-12 seedlings using in situ hybridization. In wild-type seedlings, STIP RNA was detected in the shoot apical meristem (SAM), the emerging leaf (Fig. 1F) and the upper root meristematic zone (Fig. 1D). By contrast, although the cytokinin receptor mutants are morphologically similar to the wild type at this stage, STIP expression was much reduced in their shoot apex (Fig. 1G) and there was very little detectable STIP expression in the roots (Fig. 1E). We also found similar reductions in STIP mRNA levels in the roots of the type-B arr1-3 arr10-5 arr12-1 triple-mutant seedlings (Fig. 1H), which are nearly completely insensitive to cytokinin stimulation in the roots (Argyros et al., 2008; Yokoyama et al., 2007). Although we cannot rule out the possibility that the observed reductions in STIP expression are caused by morphological changes due to the lack of cytokinin signaling, these findings are consistent with our hypothesis that STIP acts downstream of the cytokinin signaling pathways.

To further test the genetic relationship between STIP and cytokinin signaling, we examined the cytokinin sensitivity of stip mutants. Exogenous cytokinin normally inhibits seedling growth. The wild-type shoots stopped developing after being transferred to media containing 100 nM trans-zeatin, with the cotyledons and the existing leaves accumulating anthocyanin pigment, an indicator of stress response (Fig. 2C). This became even more apparent on higher concentrations of zeatin (Fig. 2D). By contrast, stip-1 seedlings continued to develop in response to the sucrose in the media, and had noticeably less anthocyanin in their leaves (Fig. 2G,H). Similar results were obtained using cytokinin N⁶-(Δ²-isopentenyl)-adenine (2-iP; see Fig. S1 in the supplementary material). However, stip-1 showed wild-type responses to cytokinins in root inhibition, callus greening, and senescence assays in the presence of sucrose. In additional experiments, stip mutants showed normal responses to other hormones, including auxin, gibberellins, brassinosteroids and abscisic acid, when grown on sucrose-containing media. These results indicate that STIP acts downstream of cytokinin perception during early seedling development.

At the molecular level, a plant's ability to respond to cytokinin stimulation can be measured by the expression of the immediate cytokinin response genes, a group of type-A ARR genes including ARR5. To further analyze the role for STIP in cytokinin signaling, we compared ARR5 expression patterns in wild-type and stip mutant seedlings using in situ hybridization. It has been previously reported that, in five-day-old seedlings, ARR5 is expressed weakly in the SAM, and stronger expression can be detected in the root meristem region (D'Agostino et al., 2000). We detected a similar ARR5 mRNA pattern in two-day-old wild-type seedlings (Fig. 3A,C). In comparison, ARR5 expression levels were much reduced in both the shoot and the root of stip-1 mutants (Fig. 3B,D). We confirmed the reduction of ARR5 expression in the shoot apex using a β-glucuronidase (GUS) reporter under the control of a 1.6 Kb ARR5 promoter, which has been shown to respond to cytokinin induction (D'Agostino et al., 2000). Two days after germination, strong ARR5::GUS activity could be detected in the SAM of the wild-type seedlings (Fig. 3E) and was significantly reduced in stip-2 seedlings (Fig. 3F).

We also compared the expression levels of additional cytokinin signaling components between two-day-old wild-type and stip-1 seedlings, using reverse transcription followed by real-time PCR analysis (qRT-PCR). In this experiment, we included three other type-A ARR genes (ARR6, 7 and 9), three type-B ARR genes (ARR1, 10 and 12), and two CRF genes (CRF2 and CRF5), which are involved in cytokinin signaling during early seedling development (Rashotte et al., 2006). Both of the type-A ARR genes and the CRF genes tested showed significant reduction in the stip-1 background, with CRF2 being most strongly affected (Table 2; Fig. 4A), indicating that STIP is involved in their transcriptional activation. As the two-day-old stip-1 seedlings are morphologically similar to the wild-type, and still have cell cycle activities in both meristems (Wu et al., 2005), we believe that the observed changes in gene expression profiles in stip mutants are caused by the lack of STIP activities, instead of resulting indirectly from the onset of the seedling growth arrest. By contrast, we found no significant change in the expression levels of the type-B ARR1, 10 and 12, in stip mutant seedlings (Table 2), suggesting that STIP does not affect type-B ARR transcription.

The above findings suggest that STIP acts upstream of the type-A ARR and the CRF genes, all of which are also transcriptionally activated by the type-B ARR proteins. This result can be explain by one of the two following mechanisms: a linear pathway where STIP is required for type-B ARR function in activating these genes, or STIP acts in parallel to the type-B ARRs in this process. To distinguish these two possibilities, we analyzed the ability of cytokinins to induce the expression of the above-mentioned target genes, which is known to require type-B ARR protein activities, in stip mutants. If the linear model was correct, we would expect to find reduced induction of these genes in stip mutants. In comparison, the parallel model would predict normal induction in stip background. As shown in Table 2, after 30 minutes of 250 nM zeatin treatment, both the type-A ARR6, 7, 9 and CRF2, 5 genes showed a similar to or higher than wild-type level of induction in stip-1, suggesting that STIP acts in parallel to the type-B ARRs to activate these genes. However, consistent with our hypothesis that STIP plays a positive role in the activation of cytokinin response genes, the absolute post-induction transcript levels of four of the five genes remained below that of the wild type in stip-1 (Fig. 4A). Only ARR7 expression in stip-1 after cytokinin induction reached similar levels as those in the wild type, which could be attributed to the reduced expression of its repressor WUS in stip mutants (Leibfried et al., 2005; Wu et al., 2005).

The type-A ARR genes modulate cytokinin response through a feedback mechanism, which results in the downregulation of the cytokinin primary response genes after a brief peak in their RNA levels (D'Agostino et al., 2000). To determine whether STIP is involved in this regulation, we examined the cytokinin induction profiles of two type-A ARR genes, ARR6 and ARR7, in two-day-old wild-type and stip-1 seedlings over the course of three hours. Consistent with the previous report (D'Agostino et al., 2000), both ARR6 and ARR7 expression peaked after 15 minutes of induction in wild type, and both were downregulated after about 30 minutes. stip-1 seedlings showed a more dramatic initial induction of ARR6 and ARR7. Their expression levels continued to increase until 30 minutes of treatment, after which both were downregulated, in a manner similar to what was seen in the wild type. In addition, ARR9, another type-A ARR gene, displayed similar induction patterns between the wild-type and stip-1 samples (Fig. 4B). These observations indicate that the loss of STIP activities does not interfere with the negative feedback by the type-A ARR genes. The slight shift observed in ARR6 and ARR7 induction profiles might be owing to the fact that both genes were expressed at much lower levels in stip-1 prior to zeatin treatment, and therefore required a longer time to reach the threshold for triggering the negative feedback responsible for attenuating the induction responses.

To assess how much the lack of STIP contributes to the seedling arrest phenotype of the cytokinin triple-receptor mutants, we over-expressed STIP in a ahk2-2 ahk3-3 cre1-12 background, taking advantage of the activation-tagged stip-D allele (Wu et al., 2005). As STIP expression in this context is driven by the 35S enhancers, it bypasses the requirement for cytokinin signaling in STIP activation. Two weeks after germination, the ahk2-2 ahk3-3 cre1-12 seedlings that also express an activation-tagged STIP (pAS1) were fully green and continued to develop without exogenous sugar (Fig. 5C), although they grew much slower than the wild type and the leaves did not fully expand (compare with Fig. 5A). A limited root system was also able to develop in ahk2-2 ahk3-3 cre1-12 in response to pAS1. Phenotypically, these partially rescued seedlings resembled the cytokinin receptor mutants germinated on sucrose-containing media (Fig. 5D) (Higuchi et al., 2004; Nishimura et al., 2004).

The partial rescue by pAS1 is reflected in the size of the SAM. At this stage, the wild-type SAM has reached its mature size and transitioned from vegetative to inflorescence meristem (Fig. 5E). Consistent with its arrested growth, the SAM in ahk2-2 ahk3-3 cre1-12 seedlings failed to develop after germination and remained flat (Fig. 5F). In comparison, the pAS1-rescued cytokinin triple-receptor mutants developed a dome-shaped SAM, although it was smaller than in the wild type (Fig. 5E,G). This finding strongly supports the notion that STIP mediates cytokinin signaling in promoting SAM growth. Consistent with our earlier finding that sucrose acts downstream of STIP in promoting meristem development (Wu et al., 2005), the sucrose-rescued ahk2-2 ahk3-3 cre1-12 mutants had a flatter SAM compared with that of the pAS1-rescued plants (Fig. 5H).

As pAS1 was also able to stimulate limited root growth in the cytokinin receptor mutants, we examined its effect on the root meristems in these seedlings. Without cytokinin perception, the roots of ahk2-2 ahk3-3 cre1-12 seedlings fail to develop. A closer examination revealed that the ahk2-2 ahk3-3 cre1-12 root morphology was largely normal, with the exception that the root meristematic zone was much reduced compared with that of the wild type (Fig. 5I,J). Interestingly, increased STIP expression driven by pAS1 not only failed to improve, but also led to slight reductions in the size of the root meristematic zones in ahk2-2 ahk3-3 cre1-12 (Fig. 5K). A similar phenotype was seen in the roots of the sugar-grown ahk2-2 ahk3-3 cre1-12 seedlings (Fig. 5L). Both the changes in root cellular morphologies and the length of the root vasculatures point to earlier differentiation of the meristematic cells in the pAS1 ahk2-2 ahk3-3 cre1-12 roots (Fig. 5K), which is similar to the result of exogenous cytokinin application (Dello Ioio et al., 2007).

We next compared the expression levels of selected cytokinin and STIP common target genes in two-day-old cytokinin perception mutants with or without pAS1 rescue. For this experiment, we included ARR6, ARR7, both type-A ARR genes, CRF2, and two of the transcription factors identified as potential common downstream factors between STIP and cytokinin induction in our microarray analysis, At4g27410 and At1g68670 (Table 1; see also Table S1 in the supplementary material). As shown in Table 3 and Fig. 6, all six genes were expressed at lower levels in ahk2-2 ahk3-3 cre1-12 mutants than in wild-type seedlings. In pAS1 ahk2-2 ahk3-3 cre1-12, increased STIP levels affected these genes in different manners. Both ARR6 and ARR7 were further repressed, possibly owing to an overactive negative-feedback loop. Consistent with the requirement for the type-B ARRs in CRF2 activation (Rashotte et al., 2006), pAS1 was only able to partially rescue CRF2 expression. Conversely, pAS1 restored At4g27410 and At1g68670 expression to the wild-type levels. These results strongly support our hypothesis that pAS1 partially rescues the cytokinin triple-receptor mutants by activating components of the cytokinin signaling network.

Although specified during embryogenesis, primary meristems only become fully established during early vegetative growth, which requires STIP activity. We have shown that cytokinin signaling is essential for this process, and that it does so partly through the function of STIP.

Based on results from tissue culture experiments, it has been proposed that cytokinin signaling is essential in meristem establishment and size regulation (Mok and Mok, 2001). However, previous reports showed that loss of cytokinin perception only results in the reduction of meristem size and seedling growth rate (Higuchi et al., 2004; Nishimura et al., 2004). We have found that the phenotypic severity of the cytokinin receptor mutant ahk2-2 ahk3-3 cre1-12 is affected by exogenous sucrose in the growth media. Without supplemented sugar, the cytokinin perception mutants completely fail to initiate growth after germination (Fig. 1B), indicating a lack of active meristems (Fig. 5F,J). In addition to their inability to initiate new organs, the cotyledons in ahk2-2 ahk3-3 cre1-12 mutants fail to fully green and expand (Fig. 1B). A similar cotyledon phenotype has been reported in plants carrying mutations in the CRF genes, which are regulated by cytokinin stimulation at both the transcription and protein localization levels (Rashotte et al., 2006). These new findings underscore the importance of cytokinin signaling in seedling development and provide in vivo evidence that cytokinins are essential for the establishment of both the shoot and root meristems after germination.

Previously, we showed that the homeodomain transcription factor gene STIP is essential for maintaining the meristematic fate and proliferation through Arabidopsis development (Wu et al., 2005). The striking similarities between stip and ahk2-2 ahk3-3 cre1-12 seedling phenotypes (Fig. 1B,C) and the common downstream targets (Table 1) suggested a possible link between STIP and the cytokinin signaling network. Several lines of evidence place STIP downstream of cytokinin perception: the ahk2-2 ahk3-3 cre1-12 and the type-B arr1-3 arr10-5 arr12-1 seedlings have significantly lower levels of STIP expression (Fig. 1E,G,H); exogenous sucrose can fully rescue stip seedlings (Wu et al., 2005), but it is only able to partially restore the development of cytokinin triple-receptor mutants (Fig. 5D,H); stip mutants show reduced cytokinin sensitivity in selected assays (Fig. 2); STIP is involved in the activation of several known cytokinin signaling components (Fig. 3; Fig. 4A); and most importantly, ectopic STIP expression leads to the partial establishment of the shoot meristem and continued growth in cytokinin receptor mutants (Fig. 5C,G). However, it is worth noting that unlike the B-type ARR genes (Mason et al., 2005), STIP is only involved in limited aspects of cytokinin signaling, namely cotyledon and meristem establishment after germination, and that STIP also functions in additional pathways at later stages of development.

The cytokinin and auxin signaling networks interact extensively in modulating plant growth (Moubayidin et al., 2009). In early embryos, STIP might act partially redundantly with its closest homolog in Arabidopsis, STIMPY-LIKE (STPL or WOX8), in modulating auxin distribution (Breuninger et al., 2008). Therefore, it is formally possible that the cytokinin insensitivity of stip mutants results from disrupted auxin transport or signaling. However, experimental evidence argues against this mechanism: the DR5::GFP reporter (Friml et al., 2003) is expressed in the wild-type pattern and intensity in stip mutant embryos (Wu et al., 2007) and seedlings (data not shown), indicating normal auxin biosynthesis and localization; when treated with exogenous auxin on sugarless media, stip mutant roots display wild-type levels of auxin uptake; and once rescued by sucrose, the stip mutants display normal responses to auxin.

The much-reduced STIP expression in the cytokinin triple-receptor mutants and the type-B arr1, 10, 12 triple mutants (Fig. 1) suggests that cytokinin signals are involved in STIP transcriptional activation. However, we did not detect an immediate increase in STIP transcription upon exogenous cytokinin induction using qRT-PCR, suggesting that STIP might not be a direct transcriptional target of the type-B ARR proteins. An additional layer of STIP regulation by cytokinin signals might occur post-translationally. STIP protein contains a relatively high percentage of serines and threonines. Similar to the CRF proteins (Rashotte et al., 2006), a functional STIP:GFP fusion protein expressed under the native STIP promoter is both nuclear and cytoplasmic (Wu et al., 2007), raising the possibility that STIP protein activity is modulated by phosphorylation. We are currently in the process of studying the mechanisms by which cytokinin signaling affects STIP expression and activity.

Based on our findings, we propose that STIP acts in parallel to the previously studied cytokinin signaling pathway downstream of the type-B ARR proteins during the establishment of the shoot apical meristem. This hypothesis is supported by our finding that ectopic STIP can restore At4g27410 and At1g68670 expression to the wild-type levels in the cytokinin receptor mutants, bypassing the need for activated type-B ARR proteins (Fig. 6). And as the slightly smaller SAM in pAS1 ahk2-2 ahk3-3 cre1-12 seedlings cannot account for the reduced growth rate and leaf size (Fig. 5C,G), it is probable that a combination of STIP and the previously known cytokinin signaling targets is required for mediating the full range of cytokinin responses during vegetative development (Fig. 7).

The root meristem responds to changes in cytokinin levels in a more complex fashion than the SAM. Although the cytokinin triple-receptor mutants develop root meristem arrest (Fig. 5J; Higuchi et al., 2004), increased root meristem activities have been observed in ahk2 ahk3 double mutants (Riefler et al., 2006) and seedlings with increased cytokinin degradation (Werner et al., 2003). Interestingly, whereas STIP is essential in root meristem maintenance in plants with an otherwise intact cytokinin signaling pathway (Wu et al., 2005), ectopic STIP not only failed to rescue, but also resulted in further reductions of, the root meristems in the ahk2-2 ahk3-3 cre1-12 triple mutants (Fig. 5K), mimicking the effect of exogenous cytokinin treatment (Dello Ioio et al., 2007). One possible explanation for this phenotype is that the elevated STIP levels in these mutants led to higher cytokinin signaling that is sufficient to inhibit root meristem. Another possibility is that a balance of activities between STIP and other cytokinin effectors is required for the proper development of the root meristem, as the pAS1 transgene did not cause any visible root defect in wild-type or ahk2-2 ahk3-3/+ cre1-12 seedlings. The precise involvement of STIP in cytokinin signaling during root meristem establishment remains to be investigated.

We thank Eric Schaller for comments and suggestions; Liang Chen for her help in statistical analysis; the Arabidopsis stock center for providing the ARR5::GUS reporter line; Tatsuo Kakimoto for the ahk2-2 ahk3-3/+ cre1-12 seeds; Takafumi Yamashino for the arr1-3 arr10-5 arr12-1 seeds; Wolfram Brenner and Thomas Schmülling for the CEL files of the BA treatments; Jihoon Ahn for the pJHAG transformation vector; Norm Arnheim and the Arnheim lab for their help in qPCR; and Steve Ruzin and Denise Schichnes at the UC Berkeley CNR Biological Imaging Facility for their help in microwave processing. Work at the Salk Institute was supported by NIH grant GM62932. J.C. is an investigator of the Howard Hughes Medical Institute and D.W. is a Director of the Max-Planck Institute. Deposited in PMC for release after 6 months.