SUPERMAN prevents stamen formation and promotes stem cell termination in the fourth whorl of the Arabidopsis flower

Summary

The molecular and genetic networks underlying the determination of floral organ identity are well studied, but much less is known about how the flower is partitioned into four developmentally distinct whorls. The SUPERMAN gene is required for proper specification of the boundary between stamens in whorl 3 and carpels in whorl 4, as superman mutants exhibit supernumerary stamens but usually lack carpels. However, it has remained unclear whether extra stamens in superman mutants originate from an organ identity change in whorl 4 or the overproliferation of whorl 3. Using live confocal imaging, we show that the extra stamens in superman mutants arise from cells in whorl 4, which change their fate from female to male, while floral stem cells proliferate longer, allowing for the production of additional stamens.

Introduction

Most of the adult body of flowering plants is generated post-embryonically from apical meristems, which are pools of undifferentiated cells at the tips of stems and roots. Divisions in the shoot apical meristem (SAM) allow for the continuous production of lateral organs on the flanks of the stem: leaves during the vegetative phase, then floral meristems (FMs) after the SAM transitions to the reproductive phase. FMs in turn generate the floral organs that comprise the flower.

Partition of the organism into distinct tissues and organs is a fundamental process of development in both animals and plants, yet it relies on different mechanisms in each kingdom. In animals, tissue separation is determined by cell surface cues that influence the adhesive properties of cells and their ability to interact with each other¹. Unlike animal cells, however, plant cells are surrounded and connected to their neighbors by contiguous cell walls that prevent them from migrating. As new organs form, they are separated from surrounding tissues by a boundary, which consists of a group of cells with restricted growth that act as a physical barrier separating two different developmental programs².

Unlike the SAM, which gives rise to lateral organs one at a time, in an iterative, spiral pattern, the FM semi-synchronously produces 16 floral organs, with four different identities, in four adjacent whorls. Floral organ identity is determined by the combinatorial action of four classes of MADS-box transcription factors, which form distinct complexes in the four floral whorls^3,4. For instance, a combination of APETALA3 and PISTILLATA (AP3 and PI, class B), together with AGAMOUS (AG, class C) and SEPALLATA3 (SEP3, class E) specifies stamens in whorl 3, while complexes composed solely of AG and SEP3 trigger carpel development in whorl 4. Targets of these MADS-box transcription factors have been extensively studied, and downstream regulatory networks have been partially deciphered⁴. However, the mechanisms that underlie the patterning of the FM, with the generation of four distinct types of organs in such a constrained space and time, remain poorly understood. In particular, how boundaries between the floral whorls are established is still unclear. Here, we analyze the role of SUPERMAN (SUP) in defining the boundary between whorl 3 stamens and whorl 4 carpels.

SUP encodes a transcriptional repressor with a C2H2 zinc-finger DNA-binding domain and an EAR repression domain^5–8, and is expressed at the boundary between whorls 3 and 4 ^6,9. sup mutant flowers have numerous extra stamens, while carpel tissue is usually reduced or missing^10,11. This phenotype is associated with the expansion of AP3 and PI expression closer to the center of the FM compared to the wild type¹⁰. Overall, floral organ number is higher in sup flowers than in the wild type, indicating an increase in cell proliferation in developing sup flower buds. Although SUP was first characterized a quarter century ago, there are still two conflicting models to explain SUP function and the developmental origin of the sup phenotype. Here, we refer to these two models as “whorl 3” and “whorl 4” models, based on the whorl where the extra stamens in sup mutant flowers hypothetically form. The whorl 4 model proposes that SUP functions to prevent ectopic expression of AP3 and PI in whorl 4. According to this model, ectopic AP3/PI expression in whorl 4 of developing sup flowers triggers the formation of stamens instead of carpels, and prolongs cell proliferation in the FM^10,11. Conversely, the whorl 3 model proposes that SUP controls the balance of cell proliferation between whorl 3 and 4, and suggests that production of extra stamens in sup mutant flowers results from increased cell proliferation in whorl 3 at the expense of whorl 4^6,12. In this study, we used live confocal imaging to investigate the developmental basis of the sup phenotype. We show that extra stamens in sup mutant flowers arise from a subset of whorl 4 cells that switch identity from female to male, as predicted by the whorl 4 model.

Results

SUP is expressed on both sides of the boundary between whorl 3 and 4

We generated a gSUP-3xVenusN7 translational SUP reporter that complements the sup-1 mutant phenotype. The SUP protein is first detected at stage 3, in cells adjacent to the boundary between whorl 3 and 4, inside of lateral sepal primordia (Fig. 1A-B; stages as described in¹³), and quickly expands to form an oblong ring approximately 3-4-cell wide, and longer medially than laterally (Fig. 1, A and C). At early stage 5, SUP is detected on both sides of the boundary between whorls 3 and 4, which at this stage forms a groove between the developing stamen primordia and the center of the flower (Fig. 1D). By late stage 5, SUP expression becomes restricted to a narrower band of cells at the boundary (Fig. 1A). gSUP-3xVenusN7 fluorescence appears to peak at stage 4, before decreasing in intensity during stage 5 and becoming undetectable by late stage 6 (Fig. 1A). Overall, the SUP expression pattern resembles that of AP3, but SUP appears to be expressed closer to the center of the flower (Fig. 1, compare A and E). To determine more precisely where SUP is expressed relative to the boundary between whorl 3 and 4, we monitored the expression of SUP and class B genes simultaneously, using the gSUP-3xVenusN7 reporter together with a gAP3-GFP (Fig. 1F) or a gPI-GFP (Fig. S1) translational reporter, or a pAP3-CFPN7 transcriptional reporter (Fig. S2). SUP expression initiates shortly after that of AP3 at stage 3 (Fig. 1F), and the first cells to express SUP also express AP3 (Fig. 1G), indicating that SUP is initially expressed in whorl 3. However, from late stage 3 on, we observed cells that express SUP but do not express AP3 or PI (Fig. 1H-I; Fig. S1; Fig. S2), demonstrating that SUP expression expands into whorl 4. At stages 4-5, SUP is clearly found on both sides of the boundary between whorls 3 and 4 (Fig. 1I; Fig. S1, B-D; Fig. S2, A1-A3). SUP expression overlaps with AP3/PI in whorl 3 at the boundaries between stamen primordia (Fig. S1D Fig. S2, B, D and F1-F3; Movie S1), as well as in a narrow, 1-2-cell wide band on the adaxial side of stamen primordia (Fig. 1I; Fig. S1D; Fig. S2, C and E1-E3; Movie S1). SUP is also expressed without AP3/PI in another narrow, 1-2-cell wide band in the outer part of whorl 4 (Fig. 1I; Fig. S1D; Movie S1). Together, these data clearly show that, contrary to earlier interpretations⁶, SUP is expressed on both sides of the boundary between whorl 3 stamens and whorl 4 carpels, and is not confined solely to whorl 3. Double fluorescence in situ hybridization experiments for SUP and AP3 confirmed that this is also the case at the mRNA level (Fig. S3). Indeed, SUP protein levels appear higher in whorl 4, where class B genes are not expressed (Fig. S2, G1 and H). Similarly, AP3 expression appears stronger in whorl 3 cells that do not express SUP (Fig. S2, G2 and I).

Figure 1. Expression of SUP and AP3 in wild-type flowers

Expression of the gSUP-3xVenusN7 (A-D) and gAP3-GFP (E) reporters separately (A-E), or together (F-I). (A, E and F) whole inflorescences; numbers indicate floral stages. (B-D and G-I) flower buds at early stage 3 (B and G), late stage 3 (H), stage 4 (C) and stage 5 (D and I). A, E, F and left panels in G and H show maximum intensity projections (MaxIPs). Lower panel in D, right panels in G and H and bottom left panel in I show slice views along horizontal planes. Upper panel in D, top left and bottom right panels in I show slice views along vertical planes. Yellow arrowheads indicate the position of the boundary between whorls 3 and 4, blue arrowheads mark cells that express both AP3 and SUP, and white arrowheads mark cells that express SUP but not AP3. Bars = 25 µm.

To better understand where SUP is expressed relative to the positions where stamen and carpel primordia initiate, we examined plants expressing both the gSUP-3xVenusN7 reporter and the DORNROSCHEN-LIKE (DRNL) pDRNL-erGFP reporter, which marks floral organ founder cells (Fig. S4, A-B)¹⁴. In particular, in whorl 3, DRNL is expressed in a ring that is reminiscent of AP3 and SUP expression patterns (Fig. S4A), while in whorl 4, DRNL is expressed in two foci, which correspond to the sites of carpel initiation, as well as in two narrow arcs of cells connecting these foci, (Fig. S4, B and C)¹⁴. SUP and DRNL expression partially overlap in stamen primordia in whorl 3 (Fig. S4, A-B), while a narrow ring of SUP expression in whorl 4 directly surrounds DRNL expression in carpel founder cells in the center of whorl 4 (Fig. S4B).

Extra stamens in sup-1 flowers arise from whorl 4 cells

To determine whether the extra stamens in sup mutant flowers arise from whorl 3 or whorl 4, we compared the expression of class B genes in wild-type and sup-1 flowers using a pAP3-3xVenusN7 transcriptional reporter (Fig. 2) and the gAP3-GFP (Fig. S5) and gPI-GFP translational reporters (Fig. S6). At stages 3-4, AP3 expression appears similar in the wild type and in sup-1 (Fig. 2, compare A and B; Fig. S5, compare A and B). However, by stage 5, both AP3 and PI are expressed closer to the center of the flower in sup-1 than in the wild type (Fig.2, compare A and B; Fig. S5, compare A and B; Fig. S6, compare A and B)¹⁰. While the fourth whorl of wild-type flowers shows no AP3/PI expression (Fig. 2, C and E; Fig. S6C), a narrow, 2-cell wide band of AP3/PI expression can be seen inside of the boundary between stamen primordia and the center of sup-1 flowers at stage 5 (Fig. 2, D and F; Fig. S6D). At stage 6, the whole fourth whorl of wild-type flowers develops into carpel primordia (Fig. 2C)¹³. Conversely, in sup-1 flowers, extra stamen primordia only start forming within the ring of extra AP3-expressing cells at stage 7, with a slight delay compared to wild-type carpels (Fig. 2G; stages for sup-1 flowers were determined based on time elapsed after stage 5, which is the last stage at which wild-type and sup-1 flowers are morphologically identical). As these extra stamens develop, AP3 expression spreads again beyond the boundary of the primordia towards the center of sup-1 flowers, forming another narrow ring of AP3-expressing cells, which later gives rise to more stamen primordia (Fig. 2K). This iterative process allows for the formation of several consecutive rings of stamens, sometimes resulting in flowers with more than 20 stamens. It is worth noting that AP3 is never expressed throughout the center of sup-1 flowers, which eventually develop into stunted, misshapen carpels or chimeric stamen/carpel organs^10,11. Accordingly, organ primordia composed both of cells that express AP3 and cells that do not, can often be seen in the center of developing sup-1 flowers (Fig. S7).

Figure 2. Expression of AP3 in wild-type and sup-1 flowers

Expression of the pAP3-3xVenusN7 reporter in the wild type (A, C and E) and sup-1 (B, D, F-K). (A-B) whole inflorescences; numbers indicate floral stages. (C-D) stage 5 flowers after removal of medial sepals; ca: carpel; ls: lateral sepal, covering lateral stamen; ms: medial stamen; dotted blue lines mark the boundary between whorl 3 stamens and the center of the flower. (E-F) stage 5 flowers, slice views along horizontal planes (bottom panels) and vertical planes (top panels); white arrowheads mark the boundary between whorl 3 stamens and the center of the flower. (G) 4-day time lapse of a single sup-1 flower between stage 7 and stage 9; white asterisks mark extra stamen primordia. (H-I) 2-day time lapse of an individual sup-1 flower between stage 4 (H) and 5 (I); left panels show a lateral view of the flower, with a segmented projection of the L1 layer in the center; right panels show a close-up of the same area on each day; red numbers mark cells that do not express AP3 at stage 4, but express AP3 at stage 5; asterisks indicate divisions that occurred between stage 4 and 5. (J) MaxIP (left panel) and slice view along vertical planes (right panels) of the flower shown in J; white arrowheads mark the boundary between whorl 3 stamens and the center of the flower. (K) MaxIP (left panel) and slice view along the vertical planes (right panel) of a stage 8 sup-1 flower; white asterisks mark extra stamen primordia; white arrowheads and dashed blue lines mark the boundary between extra stamen primordia and the center of the flower. Bars = 20 µm.

We sought to establish whether the extra AP3-expressing cells in stage 5 sup-1 flowers derive from whorl 4 cells that change identity, or from whorl 3 cells that overproliferate. The ring of extra AP3-expressing cells in sup-1 flowers looks similar to the ring of SUP-expressing cells in whorl 4 of wild-type flowers (compare Fig. 2F to Fig. 1, D and I), suggesting that the loss of SUP function might cause ectopic expression of AP3 in these cells. Using time-lapse imaging of sup-1 pAP3-3xVenusN7 flower buds, we identified numerous individual cells at the boundary between whorls 3 and 4 that do not express AP3 at stage 4 but begin to express AP3 de novo at stage 5 (Fig. 2, compare H and I). These cells that switch identity from female-fated, non AP3-expressing cells to male-fated, AP3-expressing cells are situated inside of the boundary between whorl 3 stamens and the center of the flower, indicating that they belong to whorl 4 (Fig. 2J). These data clearly show that the extra AP3-expressing cells in sup-1 flowers originate from whorl 4 cells that switch fate from female to male, rather than from whorl 3 cells that overproliferate, and supports the whorl 4 model.

Stem cell termination is delayed in sup-1 flowers

The respecification of a small ring of cells in the fourth whorl of sup-1 flowers at stage 5 is not sufficient to explain the formation of so many supernumerary stamens. The iterative production of rings of extra stamens in the fourth whorl of sup-1 flowers requires an increase or prolongation of cell proliferation compared to the wild type. To test if cells in the floral meristem are the source of overproliferation in sup-1 mutants, we monitored the expression of stem cell marker CLAVATA3 (CLV3) and stem cell-promoting gene WUSCHEL (WUS) using pCLV3-erGFP¹⁵ and pWUS-erGFP transcriptional reporters. CLV3 expression persists in wild-type flowers through stage 6 (Fig. 3A)¹⁶, but is no longer detectable at stage 7, as stem cells are incorporated into developing carpels (Fig. 3B). Conversely, we observed CLV3 expression as late as stage 10 in a small dome at the center of sup-1 flowers, after several extra stamens have formed (Fig. 3C). Stem cell termination is thus clearly delayed in sup-1 flowers compared to the wild type. Similarly, WUS expression stops by stage 5 in wild-type flowers¹⁷, but is maintained much longer in some sup-1 flowers (Fig. 3D), indicating that a bona fide FM remains functional in sup-1 flowers longer than it does in the wild type. AG is responsible for triggering stem cell termination in wild-type flowers by turning off the expression of WUS^18–21, and most mutants with a delay or loss of floral stem cell termination have defects in AG expression⁴. We thus used the gAG-GFP reporter²² to compare the expression of AG in wild-type and sup-1 flowers, and AG expression appears unaffected in sup-1 flowers (Fig. 3, compare E and F and G and H)¹⁰, suggesting that an AG-independent mechanism is responsible for the delay in stem cell termination in sup-1 flowers. SUP affects floral stem cells non cell-autonomously, as SUP and CLV3 expression domains are separated by a narrow, 1-2-cell wide ring (Fig. 3I; Fig. S8A). Indeed, this ring of cells separating the SUP and CLV3 expression domains expresses DRNL and likely corresponds to the carpel founder cells (Fig. S4, D-E). However, the SUP expression domain tightly surrounds WUS expression, with a few cells expressing both genes, suggesting that the effect of SUP on stem cells may be mediated by WUS (Fig. S8B; Movie S2).

Figure 3. SUP promotes stem cell termination non-cell autonomously, and independently of AG expression

(A-C) expression of the pCLV3-erGFP reporter in stage 6 (A) and 7 (B) wild-type flowers, and in a stage 10 sup-1 flower (C); top left panels show MaxIPs, with GFP fluorescence detected with Imaris; bottom left and top right panels show slice views along the xz and yz planes, respectively; s: whorl 3 stamen, c: carpel. (D) expression of the pWUS-erGFP reporter in a stage 8 sup-1 flower. Asterisks in C and D mark extra stamen primordia. (E-H) expression of the gAG-GFP reporter in stage 4 (E and F) and 5 (G and H) wild-type (E-F) and sup-1 (G-H) flowers. (I) expression of the gAP3-GFP (green), gSUP-3xVenusN7 (red) and pCLV3-dsRedN7 (blue)³⁰ reporters in a wild type inflorescence; numbers indicate floral stages. Bars = 20 µm.

Discussion

It is worth noting that several studies have shown that ectopic expression of SUP causes a decrease in cell proliferation^5,23–25, which was interpreted as evidence in support of the whorl 3 model. It is not surprising, however, for a boundary gene to control cell proliferation, as cell division rates are lower at boundaries, including the boundary between stamens and carpels, than in developing organs^2,26. For instance, RABBIT EARS (RBE), a gene closely related to SUP, specifies the boundary between whorls 2 and 3 by excluding AG from whorl 2^27,28, and also specifies the intersepal boundaries by regulating cell proliferation in whorl 1 via the miR164/CUP-SHAPED COTYLEDON (CUC) module²⁹. Similarly, a role for SUP in the control of cell proliferation does not exclude the possibility that SUP also affects AP3/PI expression. Moreover, rates of cell proliferation on either side of the boundary between whorls 3 and 4 appear unaffected in sup-1 flowers compared to the wild type²⁶, contrary to the predictions of the whorl 3 model.

Our data confirm, instead, the predictions of the whorl 4 model. Specifically, we show that the extra stamens in sup mutant flowers arise from a narrow ring of cells in the outer part of whorl 4, adjacent to the boundary with whorl 3, which change identity from female to male at the transition between stages 4 and 5, and start expressing AP3 de novo (Fig. 2). Cells in this ring then divide, allowing for the formation of extra stamens. The sup phenotype was initially described as heterochronic, sup flowers being “stuck in developmental time”^10,11. The sup phenotype is indeed iterative: as extra stamen primordia arise, the lack of functional SUP at the inner boundary of these stamens causes AP3 expression to spread again toward the center of the flower (Fig. 2K), allowing for the formation of additional extra stamens. Even as several rings of extra stamens form one after the other, the center of the flower, which is still devoid of AP3 expression, is replenished by the floral stem cells, which are maintained longer in sup flowers than in the wild type (Fig. 3). Eventually, the center of sup flowers differentiates into stunted carpels or mosaic, stamen-carpel organs (Fig. S7). The fact that SUP is expressed in the fourth whorl of wild-type flowers, in the same cells that express AP3 in the fourth whorl of sup flowers (Fig. 1, D and I; Fig. 2, D and F), suggests that SUP cell-autonomously represses AP3 expression in the outer part of whorl 4. Whether such a repression is direct or indirect, however, remains unknown. Conversely, SUP affects floral stem cells non cell-autonomously, and independently of AG expression (Fig. 3).

Over the last two decades, tremendous effort has been put into understanding the mechanisms that underlie the formation of boundaries between different organs and between organs and the meristem, both in the SAM and the FM ^2,4. Numerous genes have been characterized, with some, like the CUC genes, involved in the formation of all boundaries, and some, like SUP or RBE, involved in the formation of specific boundaries in the flower. However, most of these genes have been associated with the growth suppression aspect, and not the identity separation aspect of boundary formation ^2,4. This study provides new insight into how a boundary gene separates two different developmental programs in adjacent organs.

Author contributions

NP, WY and TPJ performed the experiments (NP and TPJ generated the reporter lines; NP did the live confocal imaging and WY did the in situ hybridizations); NP, PD and TPJ designed the experiments; NP wrote the manuscript; PD, EMM and TPJ edited the manuscript.

Competing interests

The authors declare no competing financial interests.

Data availability

The authors declare that all the data supporting the findings of this study are available within the paper and its supplementary information files.

Supplemental Materials

Methods

Figures S1-S8

Table S1

Movies S1-S2

Acknowledgments

The authors wish to thank John Chandler and Paul Tarr for providing reporter lines, Richard Immink for providing the unpublished gAP3-GFP and gPI-GFP reporter lines, Toshiro Ito and Frank Wellmer for critical reading of the manuscript, and Ann Lavanway, a.k.a. Dartmouth Biology Department’s Superwoman, for her tremendous help with live confocal imaging. Funding in the Jack lab was provided by the US National Science Foundation through grant IOS-0926347. Funding in the Meyerowitz Laboratory was provided by the Howard Hughes Medical Institute, the US National Institutes of Health through grant R01 GM104244 and the Gordon and Betty Moore Foundation through Grant GBMF3406.