SPATULA is required for cytokinin signaling output at the medial domain

We previously observed expression of the cytokinin signaling reporter TCS::GFP in the medial tissues, such as CMM, septa primordia, septum, transmitting tract [7], and in cells where the provasculature will arise (Fig 1), but could hardly detect expression in the lateral domain of young gynoecia. Our first question was what determined this spatial pattern of cytokinin signaling. To identify possible regulators of cytokinin signaling in gynoecia, we sought for patterning genes important for early gynoecium development and whose expression pattern overlapped with that of TCS::GFP.

Strikingly, we found that the expression pattern of the regulatory gene SPATULA (SPT) largely mirrored that of TCS::GFP (Fig 1B–1E; S1 Fig) [7,47,48]. SPT encodes a bHLH transcription factor, whose function is key in early gynoecium morphogenesis as it participates in CMM, septum and the transmitting tract development [45–47]. SPT is expressed since early stages in the CMM and its derived structures [47,48]. The single spt mutant shows a reduced number of cells in the CMM, absence of the septum and of the transmitting tract, retarded growth of the gynoecial tube and of vasculature, reduced number of ovules, and apical carpel fusion defects, which finally results in poor seed production [45–47]. In this context, we decided to investigate whether SPT was participating in the cytokinin signaling pathway during early gynoecium development.

To do this we analyzed the activity of the TCS::GFP transgene in spt mutant gynoecia (Fig 1F–1I). We used confocal laser scanning microscopy to observe fluorescence signal in transversely hand-sectioned gynoecia at stages 7, 8, 9, and 12; stages according to [49]. Remarkably, during early gynoecium development (stage 7–9), no fluorescence signal was detected in the CMM or septa primordia of spt mutants (Fig 1F–1H). On the other hand, TCS activity was increased when SPT was constitutively expressed (Fig 1J–1M). Interestingly, whereas the fluorescence signal from the TCS::GFP reporter was increased upon 48 hrs of exogenous cytokinin treatment (Fig 1N–1Q), no GFP signal increase was observed in cytokinin treated spt mutant gynoecia (Fig 1R–1U). Note that in the mature spt gynoecia (stage 12), TCS::GFP fluorescence can be observed at the edges of the defective septa, strongly suggesting that this later signal is non-SPT dependent (Fig 1I).

In summary, these results support a positive role for SPT in the cytokinin signaling pathway at the CMM and septa primordia during early gynoecium formation.

The cytokinin signaling pathway is necessary for proper gynoecium development Taking into consideration that the lack of SPT function causes severe gynoecial developmental defects [45–47] (Figs 2I and 3C) and that, based on our results, it influences cytokinin signaling output (Fig 1), we expected to observe gynoecium morphological alterations when genes in the cytokinin signaling pathway are mutated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phenotypes of the type-B arr mutants and of the spt mutant.

(A) Mature gynoecium size of wild-type, arr1, arr10, arr12, arr1 arr10, arr10 arr12, arr1 arr12, and arr1 arr10 arr12. (B) Mature fruit size of wild-type, arr1, arr10, arr12, arr1 arr10, arr10 arr12, arr1 arr12, and arr1 arr10 arr12. (C-F) Phenotypes of the type-B arr1 arr10 arr12 triple mutant compared to wild-type (WT): fruit length (C), ovule number (D), replum width (E), and replum cell number (F). (G-I) Transverse sections of stage 12 gynoecia of wild-type (G), arr1 arr10 arr12 (with transmitting tract and septum fusion defects) (H), and spt-2 (I). Scale bars: 1 mm (A), 5 mm (B), 50 μm (G-I). Error bars represent SD. *P < 0.05 (Student-t test). Sample numbers: (C, D) WT, n = 14 and arr1 arr10 arr12, n = 19; (E, F) WT, n = 20 and arr1 arr10 arr12, n = 19.

https://doi.org/10.1371/journal.pgen.1006726.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. SPT enables cytokinin responses during early gynoecium development and regulates type-B ARR gene expression.

(A) Phenotypes of wild-type, arr1, arr10, arr12, arr1 arr10, arr10 arr12, arr1 arr12, arr1 arr10 arr12, and spt-2 gynoecia three to four weeks after receiving BAP treatment for five to ten days. (B-E) Scanning electron microscopy image of wild-type and spt-2 stage 12 gynoecia one day after either receiving mock (B, C) or BAP treatment for only 48 hours (D, E). Insets show a transverse section of the ovary. (F) Expression analysis by qRT-PCR of ARR1, ARR10, and ARR12 in wild-type and spt-12 dissected gynoecia. (G-J) In situ hybridization of type-B ARR1 mRNA in wild-type (G, H) and spt-2 (I, J) floral buds at stages 9 and 12. Arrowheads indicate the detected expression in wild-type and the absence in spt-2. (K) Luciferase reporter assay in N. benthamiana leaves co-transformed with 35S::SPT and pARR1::LUC. Ratio of firefly luciferase (LUC) to Renilla luciferase (REN) activity. (L) ChIP experiments against the ARR1 promoter region (indicated by “a” in the scheme above) using a 35S::SPT-HA line and wild-type. ACT2/7 served as a negative control. For the LUC assays and qRT-PCR experiments error bars represent the SD based on three biological replicates. ChIP results of one representative experiment are shown; error bars represent the SD of the technical replicates. *P < 0.05 (LUC: Student-t test; qRT-PCR and qPCR: ANOVA). Scale bars: 500 μm (A), 100 μm (B-E, H, J), 50 μm (insets in B-E, G, I).

https://doi.org/10.1371/journal.pgen.1006726.g003

The reporter line TCS::GFP has a synthetic promoter containing type-B ARR binding sites [50], suggesting that type-B ARRs could be involved in CMM and septum development. We thus analyzed plants with impaired type-B ARR function [13]. Out of the 11 type-B ARR transcription factors present in Arabidopsis, ARR1, ARR10, and ARR12 are considered to have the main roles, based on cytokinin response assays, studies on root meristem development, and the severe reduction in cytokinin signaling [51–54]. Unfortunately, largely due to gene redundancy, single or double loss-of-function mutants in type-B ARRs do not show obvious phenotypic alterations [52]. Indeed, we observed no obvious phenotypic differences between wild type plants and the single loss-of-function mutants arr1, arr10, and arr12, nor with the double loss-of-function mutants arr1 arr10, arr1 arr12, and arr10 arr12 (Fig 2 and S2 Fig). In contrast, the type-B arr1 arr10 arr12 triple mutant plants are severely affected (Fig 2 and S2 Fig). General plant growth is strongly reduced and flower and fruit production is drastically reduced, suggesting that the meristematic activity is affected in this triple mutant (S2 Fig).

We then morphologically characterized the produced gynoecia in the type-B arr triple mutant plants. The arr1 arr10 arr12 background exhibited reproductive defects such as reduced gynoecium and fruit length, reduced replum width, and fewer ovules (Fig 2A–2F), phenotypes not observed in single and double arr mutants (Fig 2A and 2B; S2 Fig). Furthermore, thin sections of arr1 arr10 arr12 triple mutant gynoecia showed septum fusion defects and a reduction of transmitting tract tissue in some gynoecia (Fig 2H), phenotypes that we did not observe in thin sections of gynoecia of single or double arr mutants (S3 Fig), confirming the high level of redundancy among type-B ARR transcription factors.

In summary, the analyzed phenotypes provided further evidence supporting the relevance of cytokinin signaling in gynoecium development, and, since some aspects of the triple mutant were reminiscent of the spt single mutant, also support the connection between SPT and the cytokinin signaling pathway during gynoecium development.

SPT and type-B ARRs are required for the cytokinin induced proliferation

We next investigated the functional relevance and nature of the relationship between SPT and cytokinin signaling in the gynoecium using a pharmacological assay to evaluate the cytokinin response competence of the gynoecium. The repeated application of cytokinin to wild-type Arabidopsis inflorescences results in tissue overproliferation, causing ectopic outgrowths from the medial domain of the gynoecium; observed three to four weeks after the treatment [7] (Fig 3A). However, this response was affected in the type-B arr mutants and in spt (Fig 3A).

Single arr1 and arr12 mutants presented a very reduced response to the exogenous cytokinin treatments, while arr10 presented only a mild reduction (Fig 3A). Some proliferation was observed in the double arr1 arr10, which resembled the single arr1 mutant. Some proliferation was also observed in the double arr10 arr12 mutant, which was a little less than in the single arr12 mutant. However, no ectopic tissues were produced in the double arr1 arr12 mutant nor the triple arr1 arr10 arr12 mutant (Fig 3A). Interestingly, 14 out of 16 spt gynoecia (87.5%) also did not show a cytokinin response in the medial domain (Fig 3A), and only a minor proliferation effect was detected in the other two (12.5%) spt gynoecia examined (S4B Fig). It is worth noting that, as recently observed by others [55], the cytokinin response of the style and stigma of wild-type and spt gynoecia was different to that of the internal ovary. In summary, based on the pharmacological assay, type-B ARR redundancy is observed, with ARR1 and ARR12 playing the major role in the cytokinin response competence of the gynoecium. Moreover, SPT is also a major player in this response. Interestingly, during normal gynoecium development, i.e., no exogenous cytokinin application, morphological defects only become visible when three type-B ARR genes are not functional anymore (Fig 3; S2 and S3 Figs), demonstrating that the developmental program active during early gynoecium development is robust. However, the pharmacological assay indicates that the full competence of the gynoecium to respond to the artificial high level of cytokinin needs all type-B ARR proteins to be active, because a decreased response is already visible by removing one type-B ARR (Fig 3A).

We also asked whether an exogenous cytokinin treatment could rescue the developmental defects observed in the spt mutant. This mutant has two clear fusion defects: at the apex of the gynoecium and in the internal ovary region [45–47] (Fig 3C). Cytokinin was applied to inflorescences during a 48-hour period only. We observed a virtually complete rescue of the apical closure defect in 20 out of 26 spt gynoecia (76.9%), 24 hours after this treatment (Fig 3B–3E; S4C–S4K Fig). However, the spt septum defects were not rescued (Fig 3E; S4K Fig). This suggests that the internal spt fusion defects in the ovary were most likely not due to reduced cytokinin biosynthesis, and that SPT could be acting at a different level of the cytokinin pathway.

In conclusion, the data together clearly indicate that SPT is necessary for positive cytokinin signaling output (both visualized by the TCS reporter, and as the proliferation response to exogenous cytokinin treatments) in the young gynoecium.

SPT regulates type-B ARR expression

Given that the internal fusion defects of spt were not rescued by exogenous cytokinin, together with the resemblances between the spt and the type-B arr1 arr10 arr12 triple mutant phenotypes, and the alteration of cytokinin signaling in the spt mutant, we hypothesized whether one of the ways in which SPT could be connected to the cytokinin pathway could be through the regulation of the type-B ARR genes.

To test this possibility, we assayed the transcript levels of ARR1, ARR10 and ARR12 using quantitative real-time reverse transcriptase-mediated polymerase chain reaction (qRT-PCR) from dissected wild-type and spt gynoecia, respectively. Whereas this experiment did not reveal clear changes in the expression level of ARR10, transcript abundance for ARR1 and ARR12 was reduced in spt when compared to wild-type (Fig 3F), and both showed a higher relative expression than ARR10 in wild type gynoecia (S5 Fig). Note, the gynoecium is a very complex structure with many different tissues. Therefore, we cannot exclude that the changes in expression levels in specific tissues of the gynoecium are not well reflected in this assay. Therefore, we performed an in situ mRNA hybridization on ARR1 in wild-type and spt gynoecia, because ARR1 transcript abundance showed the most conspicuous reduction in dissected spt gynoecia. In wild-type, ARR1 transcripts are present in the CMM, ovule primordia, and in the style region, overlapping with SPT expression pattern (Fig 3G and 3H; S6 Fig). However, in the spt mutant, ARR1 messenger was either not detected or detected at very reduced levels, suggesting that SPT was required for ARR1 expression (Fig 3I and 3J). These results support a role for SPT positively regulating the cytokinin signaling pathway by modulating the expression of at least two type-B ARR genes, ARR1 and ARR12. On the other hand, since the qRT-PCR experiment did not show a reduction of ARR10 in the spt mutant, we cannot conclude that it is also positively regulated by SPT as ARR1 and ARR12. Therefore, it is highly likely that SPT affects, besides ARR1 and ARR12, other components of the cytokinin signaling pathway.

The ARR1 promoter fragment contains a G-box, a cis-regulatory motif targeted by bHLH transcription factors (as SPT) for gene regulation [39,56,57]. ARR10 and ARR12 have also bHLH binding motifs in their promoters, but do not have the G-box version. It has been reported that SPT binds only to the G-box version [39,57]. Therefore, the positive regulation by SPT on ARR12 expression is most likely indirect or performed by a complex where SPT participates.

To determine whether SPT is able to positively regulate ARR1 directly, we performed luciferase transient reporter assays in Nicotiana benthamiana leaves [58,59]. We observed that transiently expressed SPT protein was able to activate an ARR1::LUC reporter construct (Fig 3K). To further determine whether this regulation could be due to direct binding to ARR1 regulatory regions in the DNA, we performed chromatin immunoprecipitation assays followed by qPCR (ChIP-qPCR) using 35S::SPT-HA and wild-type Arabidopsis inflorescence tissue (Fig 3L). When compared to wild-type, ChIP-qPCR results from the 35S::SPT-HA line showed a significant enrichment of the ARR1 promoter fragment that contains the G-box, reported to be targeted by SPT [39,56,57].

In summary, all these results together are consistent with SPT being able to activate ARR1 expression in gynoecia, possibly in a direct manner. Furthermore, SPT likely regulates ARR12 as well, but in an indirect manner. Moreover, this regulation would also explain the lack of response to exogenous cytokinin application in spt and in the arr1 arr12 double mutant and, at least partly, the reduction of cytokinin-induced signal response in the CMM and septa primordia in spt gynoecia. However, we cannot discard the possibility of indirect effects of SPT on type-B ARR expression and it is highly likely that SPT affects, besides ARR1 and ARR12, also other components of the cytokinin signaling pathway.

Cytokinin signaling activates an auxin biosynthetic gene in a SPT-dependent manner

The next question was whether and how the cytokinin signaling pathway interacts with the auxin pathway in gynoecia. It has been previously described that the cytokinin signaling pathway and SPT can interact with several genes in the auxin signaling pathway [13,28,31,38,39,55,60]. We therefore explored whether interactions between them were also taking place during the formation of medial tissues in the gynoecia.

Interestingly, TCS::GFP, SPT, and the auxin biosynthesis gene TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS 1 (TAA1) are co-expressed in the CMM (Fig 4) [36,61], and mutant combinations of TAA1 with its paralog TAR2 (TRYPTOPHAN AMINOTRANSFERASE RELATED 2) resulted in plants with severely affected gynoecium development, indicating the importance of local auxin biosynthesis for correct gynoecium morphogenesis [61].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Cytokinin signaling activates the auxin biosynthetic gene TAA1 in a SPT-dependent manner.

(A, B) Expression of the translational fusion TAA1::GFP-TAA1 in a transverse section of a stage 9 wild-type gynoecium that either received mock (A) or BAP treatment for 48 hours (B). (C, D) Expression of the translational fusion TAA1::GFP-TAA1 in a transverse section of a stage 9 spt-12 gynoecium that received mock (C) or BAP treatment for 48 hours (D). (E) Luciferase reporter assay in N. benthamiana leaves co-transformed with 35S::ARR1 and pTAA1::LUC. Ratio of LUC/REN activity. (F) ChIP experiments against the TAA1 promoter region (indicated by “a” in the scheme above) using an inducible 35S::ARR1ΔDDK:GR line treated with dexamethasone or mock. ACT2/7 served as a negative control. (G) Luciferase reporter assay in N. benthamiana leaves co-transformed with 35S::SPT and pTAA1::LUC. Ratio of LUC/REN activity. (H) ChIP experiments against the TAA1 promoter region (indicated by “a” in the scheme above) using a 35S::SPT-HA line and wild-type. ACT2/7 served as a negative control. Error bars represent the SD for the LUC assays based on three biological replicates. ChIP results of one representative experiment are shown; error bars represent the SD of the technical replicates. *P < 0.05 (LUC: Student-t test; qPCR: ANOVA). Scale bars: 10 μm (A-D).

https://doi.org/10.1371/journal.pgen.1006726.g004

To test whether cytokinin had an effect on TAA1 expression, we applied cytokinin to TAA1:GFP inflorescences and observed a strong increase in the GFP signal in the medial domain of stage 8 and 9 gynoecia, indicating that cytokinin induces TAA1 expression (Fig 4B). Moreover, a microarray data meta-analysis has also identified TAA1 as a cytokinin-responsive gene [62].

We then aimed at exploring the molecular mechanism involved in this induction. Since type-B ARRs are important positive regulators of cytokinin transcriptional response, we investigated whether ARR1 could activate TAA1. We found that transiently expressed ARR1 is able to activate a TAA1::LUC reporter construct in transient assays in N. benthamiana leaves (Fig 4E). We next performed ChIP-qPCR assays using the dexamethasone (DEX) inducible glucocorticoid receptor (GR) fusion line 35S::ARR1ΔDDK-GR. In this line, upon DEX induction, ARR1 is constitutively active in the absence of cytokinin [63,64]. ChIP-qPCR results from DEX-treated 35S::ARR1ΔDDK-GR inflorescences, when compared to those mock-treated, showed significant enrichment of a TAA1 promoter fragment that contains various type-B ARR binding sites (Fig 4F), consistent with ARR1 directly regulating TAA1. This strongly suggests that one of the outputs of the cytokinin signaling pathway is to activate the auxin biosynthetic pathway and that ARR1 is a hub connecting the cytokinin signaling and auxin pathway. Our results further substantiate previous reports suggesting the connection between ARR1 and auxin biosynthesis [65].

Given the fact that SPT enables cytokinin responses in the early gynoecium, and that it likely activates ARR1 directly, we evaluated TAA1 response to cytokinin in spt mutants. As expected, when SPT is mutated, gynoecium cytokinin-dependent TAA1::GFP induction is abolished (Fig 4D). We also analyzed TAA1:GFP expression in untreated spt gynoecia, and observed a moderate reduction in GFP signal (Fig 4C), suggesting that SPT is able to activate TAA1, although additional regulators might contribute to TAA1 expression in developing gynoecia. Interestingly, besides being required for the cytokinin induction of TAA1 expression, based on our luciferase transient reporter assays (Fig 4G) and ChIP-qPCR experiments (Fig 4H), SPT seems to directly regulate TAA1 expression by recognizing a cis-motif present within the TAA1 promoter (Fig 4H).

In summary, these results indicate that both cytokinin and SPT can activate TAA1 at the medial domain of the ovary, probably in a cooperative fashion, integrating a regulatory node for correct cytokinin signaling and auxin biosynthesis in the medial tissues of the gynoecia. This TAA1 activation might be mediated by phosphorylated type-B ARRs and SPT, possibly as direct regulators, as the ARR1 and the SPT activation of TAA1 and ChIP experiments suggest, though indirect regulation cannot be discarded.

The auxin transporter PIN3 is coordinately activated by cytokinin and SPT

Intriguingly, the expression of TAA1 at the medial tissues of the ovary did not coincide with the expression of the auxin reporter DR5rev::GFP, which was not detected in these tissues (CMM, septa primordia, septum and transmitting tract) (S7A–S7D Fig). One possible explanation for this discrepancy is that the auxin synthesized by TAA1 is transported outside these tissues by PIN auxin efflux transporters [66]. To determine whether this was the case, we analyzed the expression pattern of different GFP reporters for PIN genes (PIN1, 3, 4 and 7) in the medial region of the ovary of wild-type gynoecia and observed that PIN1:GFP and PIN3:GFP were expressed in the medial domain (S7 and S8 Figs). Since pin1 mutants do not produce flowers [67], we focussed most of our analyses on PIN3. PIN3:GFP signal was detected in the CMM, septa primordia, septum and transmitting tract (Fig 5A; S7F and S8K–S8T Figs).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The auxin transporter PIN3 is coordinately activated by cytokinin and SPT.

(A-C) PIN3 expression in stage 9 PIN3::PIN3-GFP gynoecia that either received mock (A, transverse section) or BAP treatment for 48 hours (B, transverse section and C, longitudinal view). The inset in (C) shows a magnified view of the proliferating tissue. Arrows indicate the possible auxin flow. (D-F) PIN3 expression in transverse sections of stage 9 PIN3::PIN3-GFP gynoecia in spt-2 (D), 35S::SPT (E), and in spt-2 treated for 48 hours with BAP (F). (G-J) Transverse sections of stage 12 gynoecia of wild-type (G, H) and pin3-4 (I, J). Gynoecia phenotypes after three to four weeks of mock (G, I) or BAP treatment for five days (H, J). Insets show a scanning electron microscopy image of the gynoecium. (K) Luciferase reporter assay in N. benthamiana leaves co-transformed with 35S::ARR1 and pPIN3::LUC. Ratio of LUC/REN activity. (L) ChIP experiments against the PIN3 promoter regions (indicated by “a” and “b” in the scheme above) using an inducible 35S::ARR1ΔDDK:GR line treated with dexamethasone or mock. ACT2/7 served as a negative control. (M) Luciferase reporter assay in N. benthamiana leaves co-transformed with 35S::SPT and pPIN3::LUC. Ratio of LUC/REN activity. (N) ChIP experiments against the PIN3 promoter regions (indicated by “a” and “b” in the scheme above) using a 35S::SPT-HA line and wild-type. ACT2/7 served as a negative control. Error bars represent the SD for the LUC assays based on three biological replicates. ChIP results of one representative experiment is shown and the error bars represent the SD of the technical replicates. *P < 0.05 (LUC: Student-t test; qPCR: ANOVA). Scale bars: 10 μm (A-F), 100 μm (G-J, G-J insets). Ovule primordium (op).

https://doi.org/10.1371/journal.pgen.1006726.g005

Given that cytokinin signaling was also detected there, we investigated whether cytokinin was influencing PIN3 expression. Interestingly, when cytokinin was applied, the PIN3:GFP reporter was strongly induced in the medial domain of the ovary, while it seemed to be localized in a non-polar fashion in the cells (Fig 5B; S9M–S9O Fig). This clearly indicated that PIN3 is responsive to cytokinin in these tissues. Accordingly, we observed a similar induction of PIN3 and PIN1 expression by cytokinin in the ectopic outgrowths produced from the medial region of the ovary (Fig 5C; S9N Fig). In these tissues, PIN3:GFP and PIN1:GFP were polarly localized towards the emerging outgrowths (Fig 5C; S9X Fig), which also showed a high DR5rev::GFP signal at their tips [7].

To determine whether PIN3 is relevant for the overproliferation of the medial tissue after cytokinin treatment, we applied cytokinin to pin3 inflorescences (pin1 could not be tested due to the lack of inflorescences). If PIN3 is required, the ectopic outgrowths would not be produced. Indeed, a minor ectopic medial outgrowth was observed in cytokinin-treated pin3 gynoecia (S10D Fig) and only apical-basal defects were detected in 78.2% of the cases (i.e., alterations in the size of the ovary, gynophore and style with respect to each other) (n = 330) (Fig 5G–5J and insets) [32]. This suggests that observed medial tissue responses to exogenously applied cytokinin require a functional PIN3.

To explore the role of PIN3 in medial tissue development, we analyzed thin sections of pin3 untreated gynoecia and observed mild alterations in transmitting tract development (Fig 5G–5J), characterized by reduced blue staining of the cells. A possible explanation for the pin3 mild phenotype in the medial tissue is that the related PIN7 can partially compensate for the PIN3 function, as it has been reported in other developmental programs [68]. Accordingly, the double pin3 pin7 mutant has severe floral defects and none of the gynoecia form correctly (S10E–S10H Fig) [66]. However, we were not able to detect PIN7 signal in wild type ovaries, perhaps due to low signal of the reporter line. On the other hand, another explanation is that PIN1 partially compensates for the PIN3 loss, because PIN1 signal is clearly detected in the medial tissues and the reporter line responds to cytokinin (S8 and S9 Figs).

To obtain insights about the possible molecular mechanism by which cytokinin activates PIN3 expression at the medial domain, and since type-B ARRs are important effectors of cytokinin signaling, we explored whether type-B ARR activity could be involved, using ARR1 to test this. We found that ARR1 was able to activate PIN3 in luciferase transient reporter assays (Fig 5K). Moreover, ChIP-qPCR data were compatible with the possibility of ARR1 directly activating PIN3 expression in the medial domain of the gynoecia (Fig 5L). The data suggests that ARR1 is able to bind two regions within the PIN3 promoter containing putative cis-regulatory motifs for type-B ARR transcription factors.

Next, we investigated whether SPT could also regulate PIN3 expression and whether the PIN3 cytokinin response was SPT-dependent. Expression of PIN3 in the CMM and septa primordia appears to be dependent on SPT given that no PIN3 expression in these medial tissues was observed in a spt mutant background (Fig 5D; S9E–S9H Fig), whereas it increased when SPT was constitutively expressed (Fig 5E; S9I–S9L Fig). qRT-PCRs using dissected gynoecia also showed a decrease in PIN3 expression in a spt background and an increase in a 35S::SPT background (S9W Fig). Moreover, induction of PIN3 expression by cytokinin is also dependent on the presence of SPT, as no PIN3 cytokinin-dependent activation was seen in spt ovaries (Fig 5F; S9P–S9R Fig). Though indirect regulation cannot be discarded, luciferase transient reporter assays (Fig 5M) together with ChIP-qPCR data (Fig 5N) were consistent with direct regulation of PIN3 by SPT. Specifically, and based on our results, SPT bound to two regions (`a`and preferentially `b`) within the PIN3 promoter (Fig 5N).

In summary, all these results together support that SPT and the cytokinin-signaling pathway, probably in a cooperative fashion, are connected to auxin biosynthesis and auxin transport at the medial domain of the ovary.

Cytokinin signaling repressors are expressed in the lateral domain After finding that the localization of cytokinin signaling in the medial domain depends on SPT, we still wondered whether other factors could be repressing cytokinin signaling in the lateral domain. Besides barely detecting TCS::GFP expression at the lateral domain, we also had observed that exogenous cytokinin treatments could not activate this marker in the lateral domain (presumptive valves). Moreover, TCS::GFP signal could also not be detected in the lateral domain of 35S::SPT gynoecia, not even when cytokinin was added to this line (S11 Fig). Finally, the outgrowths promoted by cytokinin treatments were observed to arise from the medial and not the lateral domain of the gynoecium. Together, these observations indicate that gynoecia lateral tissues (presumptive valves) are not responsive to the exogenously applied cytokinin or ectopic SPT expression, and suggest that repression is taking place in these tissues.

As mentioned in the Introduction, some components of the cytokinin signaling pathway execute a repressing effect on cytokinin-dependent outputs. We therefore wondered whether these repressors were present in the lateral domain of the gynoecium. The gene AHP6 encodes a histidine phosphotransfer protein that inhibits cytokinin signaling responses [69] and it has been also shown to participate in auxin-cytokinin communication (i.e., it is induced by auxin) [41,69,70]. In gynoecia, AHP6::GFP reporter activity is strong in the lateral domains of stage 7, 8, and 9 gynoecia (Fig 6A–6D), suggesting that AHP6 is negatively regulating cytokinin signaling in the valves. As mentioned above, the TCS::GFP reporter is not active in the lateral domains of gynoecia (Fig 1). However, this marker was ectopically active in the lateral domains of ahp6 gynoecia (Fig 6E and 6F). Moreover, repeated cytokinin applications to ahp6 inflorescences caused ectopic tissue proliferation in a radial pattern in the apical part of the gynoecium, a morphological effect not seen in wild-type treated gynoecia (Fig 6G and 6H).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The cytokinin signaling repressors AHP6 and ARR16 likely block cytokinin responses in lateral tissues.

(A-D) Expression of the transcriptional reporter AHP6::GFP in transverse sections of stage 7, 8, 9, and 12 gynoecia. (E, F) Expression of the cytokinin response reporter TCS::GFP in transverse sections of stage 9 and 12 gynoecia in an ahp6-1 mutant background. Arrowheads indicate the absence of GFP signal in the epidermis of the valves. (G, H) Phenotypes of wild-type (G) and ahp6-1 (H) gynoecia one week after receiving BAP treatment for two weeks. (I-L) Expression of the transcriptional reporter ARR16::GUS (type-A ARR) in transverse sections of stage 7, 8, 9, and 12 gynoecia. Scale bars: 10 μm (A-C, E), 20 μm (D, F), 1 mm (G, H), 100 μm (I-L).

https://doi.org/10.1371/journal.pgen.1006726.g006

Interestingly, we also observed that the GUS-reporter construct (ARR16::GUS) for the type-A ARR16 gene, which encodes a regulatory protein that negatively regulates cytokinin signaling pathway responses [13], was active in the lateral domain of stage 7, 8, 9, and 12 gynoecia (Fig 6I–6L).

Altogether, these data support a scenario in which the cytokinin signaling pathway is negatively regulated in the lateral domain of the gynoecium by AHP6, and perhaps by ARR16 as well. This negative regulation can confine cytokinin signaling to the medial domain of the gynoecium (Fig 7).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Model of the regulatory network in early gynoecium development integrating SPT, cytokinin signaling, auxin biosynthesis, and auxin transport.

Model of the regulatory network in early gynoecium development. This regulatory network integrates the results that SPT, an important player of gynoecium development, enables cytokinin signaling in the medial domain of the young gynoecium by activating the transcription of type-B ARR genes (at least ARR1 and ARR12; likely ARR1 directly and ARR12 indirectly), which proteins become active upon phosphorylation because of a phosphorelay cascade initiated when cytokinin is present, and then together activate auxin biosynthesis (TAA1) and transport important (PIN) for growth. It is likely that SPT also affects other components of the cytokinin signaling pathway (indicated by gray arrows). Solid black arrows indicate a positive regulation and a T-bar indicates a repression function, a broken black arrow indicates possible positive regulation by auxin, a double arrowhead indicates phosphorylation, purple arrows indicate possible auxin flow; CK, cytokinin; P, phosphate group.

https://doi.org/10.1371/journal.pgen.1006726.g007

A dynamic GRN Boolean model during early gynoecium development In this work, we found that SPT enables cytokinin signaling at the medial domain. Moreover, both SPT and cytokinin signaling, probably together, positively regulate auxin biosynthesis and transport genes in this domain. Based on these findings we developed a preliminary gene regulatory network (GRN; Figs 7 and 8) that acts during early gynoecium development. To verify that this network fits the experimental data, we made a GRN Boolean model using the computational tools BioTapestry [71] and GeNeTool [72] (Fig 8A and 8B), and we confirmed that the wiring of this network gives a stable output (i.e., fixed steady state for each gene) (Fig 8A). Note, we modeled TAA1 and PIN3 regulation by SPT and ARR1 in a cooperative manner (i.e., meaning that both are necessary). The possible cooperative regulation could be through the formation of a protein complex. Support for this is the observation that SPT and ARR1 bind to the same fragments of the TAA1 and PIN3 promoters in ChIP assays. We already explored whether these two transcription factors interact directly. However, we could not detect any protein-protein interaction in yeast two-hybrid (between SPT and ARR proteins) nor in a bimolecular fluorescence complementation assays (between SPT and ARR1) (S12 Fig). Nevertheless, SPT and ARR1 could be part of a higher-order complex where both factors are present but do not interact directly. Note, cytokinin should be as well present, to start the phosphorelay cascade that finally leads to phosphorylation of the type-B ARR proteins so that they are functional. Moreover, in silico perturbations of the regulatory links in our model produces the observed phenotypes (Fig 8C–8F). The experimental evidence supports direct regulatory links between the genes, but the value of the identified GRN holds even though if regulatory links would be indirect. The topology of the network presents interesting features. For example, the regulatory interactions between SPT, ARR1, and TAA1 as well as between SPT, ARR1, and PIN3 are wired as coherent feed-forward subcircuits [73,74]. This type of subcircuit has been detected in other plant developmental processes (e.g., [75–77]), and can integrate upstream spatial regulatory inputs, cause high expression of the target gene, and temporal delay in switching the target gene on or off [73,74].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Dynamic GRN Boolean model active during early gynoecium development.

(A) The topology of the Gene Regulatory Network (GRN) model visualized using the computational and graphical platform BioTapestry [71]. Regulatory relations among genes are based on the experimental evidence (this work). The two coherent feed-forward subcircuits are formed by starting with SPT regulating ARR1 and both together regulate TAA1, and also both together regulate PIN3. TAA1 also has other positive regulators, but for simplicity only SPT and ARR1 are depicted. ARR1 is depicted, but ARR12 is likely part of the network too. CK: cytokinin; Aux: auxin; P: phosphorylation. (B) All regulatory interactions were fed to the computational tool GeNeTool [72] to create the Boolean vector equations and for modeling of the GRN. The coherent feed-forward subcircuits are both configured as an AND-gate, i.e., cooperate regulation. Boolean output for gene active = 1 and for gene inactive = 0. (C-F) Alterations of the topology of the GRN model after perturbations. Boolean output was calculated by GeNeTool for each perturbation and visualized with BioTapestry. Gray color of lines and genes means it is inactive. The GRN is affected after the following perturbation experiments: SPT off (C), CK signaling off (D), PIN3 off (E), and TAA1 off (F). In all cases the GRN is altered, which predicts that gynoecium development will also be altered, and this happens in the mutants, validating our GRN model.

https://doi.org/10.1371/journal.pgen.1006726.g008

Plant materials and growth conditions

Seeds were obtained for spt-2 (CS275), arr1-3 (CS6971), arr10-5 (CS39989), arr12-1 (CS6978), arr1-3 arr10-5 (CS39990), arr1-3 arr12-1 (CS6981), arr10-5 arr12-1 (CS39991), arr1-3 arr10-5 arr12-1 (CS39992), and DR5rev::GFP (CS9361) from the Arabidopsis Biological Resource Center (Ohio State University, Columbus), TCS::GFP from Bruno Muller, pSPT-6253:GUS from David Smyth, spt-12, 35S::SPT, and 35S::SPT-HA from Karen Halliday, pin3-4 and pin3 pin7 from Eva Benková, PIN3::PIN3-GFP and TAA1::GFP-TAA1 spt-12 from Lars Østergaard, PIN1::PIN1-GFP and PIN7::PIN7-GFP from Luis Herrera-Estrella, PIN4::PIN4-GFP from Elena Alvarez-Buylla, TAA1::GFP-TAA1 from Anna Stepanova, AHP6::GFP from Ykä Helariutta, TCS::GFP in the ahp6-1 background from Teva Vernoux, 35S::ARR1ΔDDK-GR from Takashi Aoyama, and ARR16::GUS from Takeshi Mizuno. Arabidopsis thaliana, Nicothiana tabacum, and Nicotiana benthamiana were grown in soil under normal greenhouse conditions or in a growth chamber (~22°C, long day light regime).

Cytokinin treatments

Inflorescences were treated with cytokinin 6-Benzylaminopurine (BAP) as previously described [32]. In summary, one week after bolting, BAP solution drops were placed on the inflorescences once a day for 2 (48-hour period) or 5 to 10 (repeated applications) consecutive days. To observe ectopic outgrowths from the medial domain of the pistil, a BAP treatment for five days is given and after three to four weeks observations are made. The BAP solution contains 100 μM 6-benzylaminopurine (BAP; Duchefa Biochemie) and 0.01% Silwet L-77 (Lehle Seeds) in distilled water. Mock treatments contained only 0.01% Silwet L-77 in distilled water. All treated plants with their respective controls were cultivated simultaneously under the same growth conditions.

Gene expression analysis

For qRT-PCR analysis, stage 8–10 gynoecia or inflorescence with only floral buds were collected and total RNA was extracted using TRIzol (Invitrogen). After DNAse I (Invitrogen) treatment, cDNA was prepared using SuperScript III Reverse Transcriptase (Invitrogen) according to manufacturer’s instructions and using reverse specific primers for each of the corresponding genes under test (S1 Table). The cDNA was analyzed in an ABI PRISM 7500 sequence detection system (Applied Biosystems) with SYBR Green Master Mix (Applied Biosystems) according to the manufacturer’s instructions. Three biological replicates and four technical replicates were done for each assay. Data was analyzed using the 2^-ΔΔCT method [109]. Target gene expression levels were normalized to ACTIN2/7. Primer sequences are listed in S1 Table.

In situ hybridization

In situ hybridization was carried out as previously described [110]. The template for the DIG-labeled antisense and sense probe synthesis for ARR1 mRNA was generated by PCR using specific primers (S1 Table) and inflorescence wild-type cDNA. The resulting PCR fragment was purified, sequenced and used as template to transcribe the antisense probe with the T7 RNA polymerase (Promega) and the sense probe with the SP6 polymerase (Promega).

ChIP assay

ChIP experiments were performed as previously described [111]. Between 0.5 g and 1 g of inflorescences were collected for each experiment. The 35S::SPT-HA homozygous transgenic line [112] and wild-type Ler (as relative control) were used for ChIP on SPT. A monoclonal mouse anti-HA (Sigma; H3663) (2 μg per sample) was used to immunoprecipitate SPT-HA complexes. We additionally tested 35S::SPT-HA line ChIP enrichment with no anti-HA antibody as a preliminary test, to ensure specificity of the ChIP reaction. ChIP assays for ARR1 were performed using the DEX (dexamethasone)-inducible 35S::ARR1ΔDDK-GR line [64] after DEX induction. Mock treated plants were employed as controls. In short, DEX-treated 35S::ARR1ΔDDK-GR inflorescences were collected 24 hours after two DEX applications (each one separated by 12 hours) with 10 μm DEX solution with 0.015% Silwet in distilled water, frozen in liquid nitrogen, and stored at -80°C till enough material was collected. Glucocorticoid Receptor alpha polyclonal antibody (Thermo Scientific; PA1-516) (2 μg per sample) was used to immunoprecipitate ARR1ΔDDK-GR complexes. Results from qPCR experiments were analyzed using the 2^-ΔΔCT method [109]. Each biological sample was assayed for relative enrichment with respect to its input sample (fragments were normalized using ACTIN2/7). Binding was concluded if PCR enrichment was detected in at least three out of five independent biological replicates. Primers used for ChIP-qPCR analysis are listed in S1 Table.

Luciferase activity assay

Promoter regions of PIN3 (4.3 kb, -4310 to ATG), ARR1 (2.1 kb, -2116 to ATG), and TAA1 (2 kb, -2047 to ATG) were amplified from Arabidopsis Col-0 genomic DNA with specific primer pairs (S1 Table), cloned into pGEM-T vector (Promega), digested with SmaI and NcoI restriction enzymes, and ligated into pGREEN-LUC [59] to generate pPIN3::LUC, pARR1::LUC, and TAA1::LUC reporters, respectively, for transient expression assays in N. benthamiana leaves. The 35S::SPT effector construct that was used in the coinfiltrations with the corresponding LUC reporters, was generated by transferring the SPT ORF into the pEARLY100 vector [113] through Gateway reactions, which was previously cloned in the pDONR221 vector (Invitrogen). The 35S::HA-ARR1 has been previously described [114]. The transient Luciferase expression assays were performed by transient transformation of N. benthamiana leaves by Agrobacterium infiltration, which was performed as previously described [115] with minor modifications [58]. At least three plants at the same developmental stage were used for each treatment, and the experiments were repeated at least three times.

Tissue preparation and confocal microscopy analysis To observe fluorescence signal, gynoecia were dissected and observed as previously described [116]. In summary, gynoecia were observed longitudinally or cut transversely using a scalpel and mounted in glycerol. Propidium iodide (PI; Fluka), 50 μM PI for 30–60 seconds, was used as counterstain. Imaging was done using a LSM 510 META inverted confocal microscope (Carl Zeiss) with either a 20X or 40X air objective. GFP was excited with a 488 nm line of an Argon laser and PI with a 514 laser line. GFP emission was filtered with a BP 500–520 nm filter and PI emission was filtered with a LP 575 nm filter.

Scanning electron microscope analysis Fresh tissue samples were visualized in a Zeiss scanning electron microscope EVO40 (Carl Zeiss) using the VPSE G3 or the BSD detector with a 15–20 kV beam.

GUS analysis

Gynoecia were dissected and pre-fixed with cold acetone for 20 min, rinsed, and transferred into GUS substrate solution: 50 mM sodium phosphate pH 7, 5 mM K3/K4 FeCN, 0.1% (w/v) Triton X-100, and 2 mM X-Gluc (Gold BioTechnology, Inc). After application of vacuum for 5 min, SPT::GUS and ARR16::GUS samples were incubated at 37°C for 12 hrs.

Histology

Tissues were fixed in FAE (3.7% formaldehyde, 5% glacial acetic acid and 50% ethanol) with vacuum (15 min, 4°C) and incubated for 60 min at room temperature. The material was rinsed with 70% ethanol and incubated overnight at 4°C, followed by dehydration in a series of alcohol solutions (70, 85, 95, and 100% ethanol) for 60 min each and embedded in Technovit as previously described [117]. Pictures were taken using a Leica DM6000B microscope coupled with a DFC420 C camera (Leica).

Transmitting tract analysis Transmitting tract staining was performed as previously described [118].

Yeast two-hybrid assay

The SPT cDNA was cloned in the pENTR/D TOPO vector (Invitrogen), verified by sequencing, and introduced into the LexA DNA-binding domain vector (pBTM116c-D9) by Gateway LR recombination. The ARRs fused to the Gal4 activation domain in pACT2 (Clontech, Mountain View, CA, USA) are previously described [119]. Yeast transformations were performed as previously described [120] using the L40ccaU strain (MATa his3D200 trp1-901 leu2-3 112 LYS::(lexAop)4-HIS3 URA3::(lexAop)8-lacZ, ADE2::(lexAop)8- URA3 GAL4 gal80 can1 cyh2) [121]. The assay was done on SD-Gluc medium lacking Leucine, Tryptophan, and Histidine complemented with 3 mM 3-Amino-1,2,4-triazole. Interactions were scored after growing yeast at 25°C for 5 days.

Bimolecular fluorescence complementation assay

SPT and ARR1 coding sequences in Gateway entry vectors were recombined with pYFC43 and pYFN43 to generate C- and N-terminal YFP fusion constructs, respectively [122]. BiFC in young N. tabacum leaves was previously described [117]. YFP signal was assayed 3 days after infiltration using a confocal microscope.

GRN Boolean model

The topology of the Gene Regulatory Network (GRN) model was visualized using the computational and graphical platform BioTapestry [71]. Regulatory relations among genes are based on the experimental evidence (this work). The two coherent feed-forward subcircuits are formed by starting with SPT regulating ARR1 and both regulate TAA1, and also both regulate PIN3. TAA1 also has other positive regulators, but for simplicity only SPT and ARR1 are depicted. ARR1 is depicted, but other type-B ARR genes are part of the network. All regulatory interactions were fed to the computational tool GeNeTool [72] to create the Boolean vector equations and for modeling of the GRN. The two coherent feed-forward subcircuits are both configured as an AND-gate. Boolean output for gene active = 1 and for gene inactive = 0. Alterations of the topology of the GRN model after perturbations were calculated by GeNeTool and visualized with BioTapestry.