Skip to main content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Ann Bot. 2016 Apr; 117(5): 905–923.
Published online 2016 Apr 20. doi: 10.1093/aob/mcw023
PMCID: PMC4845806
PMID: 27098089

Analysis of the Arabidopsis superman allelic series and the interactions with other genes demonstrate developmental robustness and joint specification of male–female boundary, flower meristem termination and carpel compartmentalization

Associated Data

Supplementary Materials


Background and Aims SUPERMAN is a cadastral gene controlling the sexual boundary in the flower. The gene’s functions and role in flower development and evolution have remained elusive. The analysis of a contrasting SUP allelic series (for which the names superman, superwoman and supersex have been coined) makes it possible to distinguish early vs. late regulatory processes at the flower meristem centre to which SUP is an important contributor. Their understanding is essential in further addressing evolutionary questions linking bisexuality and flower meristem homeostasis.

Methods Inter-allelic comparisons were carried out and SUP interactions with other boundary factors and flower meristem patterning and homeostasis regulators (such as CLV, WUS, PAN, CUC, KNU, AG, AP3/PI, CRC and SPT) have been evaluated at genetic, molecular, morphological and histological levels.

Key Results Early SUP functions include mechanisms of male–female (sexual) boundary specification, flower mersitem termination and control of stamen number. A SUP-dependent flower meristem termination pathway is identified and analysed. Late SUP functions play a role in organ morphogenesis by controlling intra-whorl organ separation and carpel medial region formation. By integrating early and late SUP functions, and by analyzing in one single experiment a series of SUP genetic interactions, the concept of meristematic ‘transference’ (cascade) – a regulatory bridging process redundantly and sequentially co-ordinating the triggering and completion of flower meristem termination, and carpel margin meristem and placenta patterning – is proposed.

Conclusions Taken together, the results strongly support the view that SUP(-type) function(s) have been instrumental in resolving male/female gradients into sharp male and female identities (whorls, organs) and in enforcing flower homeostasis during evolution. This has probably been achieved by incorporating the meristem patterning system of the floral axis into the female/carpel programme.

Keywords: Arabidopsis, allelic series, carpel, evo-devo, flower meristem determinacy, flower homeostasis, flower pattern, meristematic ‘cascade’/transference, pistillody/carpelloidy, placenta, stamen, SUPERMAN gene: superman, clark-kent/superwoman, supersex, AG, CLV, CRC, CUC2, KNU, PAN, SPT, WUS


Arabidopsis flowers produce fixed numbers of sepals, petals, stamens and carpels as distinct and successive whorls. The floral meristem (FM) becomes determinate after the initiation of carpels, when it ceases to maintain its stem cell population. In this process, also known as FM termination, the expression of WUS, a homeodomain protein with a central role in stem cell activity, is controlled through a sequence of processes that initiate and subsequently maintain stem cell arrest at the flower meristem centre (FMC) (see also Prunet et al., 2008); we label the process the ‘triggering and completion of FM termination’.

There are spatial and/or temporal control regulators of FM termination (Prunet et al., 2009): an altered spatial control results in an increased stem cell population, enlarged meristem and supernumerary organs (i.e. an altered homeostasis); a prolonged maintenance of stem cells produces a delay in FM termination with supernumerary whorls and organs (i.e. flower indeterminacy). This study primarily focuses on several factors of determinacy that control reproductive organ identities and the corresponding boundaries in the FMC space. Importantly, meristem termination processes and boundary establishment are interdependent (Rast and Simon, 2008) and trigger female programme development (Prunet et al., 2009). One question this study addresses is how the sequence of meristematic and differentiation processes is orchestrated in space and time. Meristematic modularity or ‘transference of meristematic capacity’ (Mathews and Kramer, 2012) is likely to account for the way meristematic ‘cascading’ is subtended by genetic-level modularity. The concept relies on the co-option of meristematic programmes [termination–(re)initiation)] through identity switches deployed during development, and is reflected in new trait acquisition by organs and tissues.

In this context, the timely separation of the stamen whorl from the carpel whorl at the FMC is the critical early event in a hermaphrodite flower. The homeotic MADS domain family genes AP3, PI (B function) and AG (C function), respectively, determine stamen (B + C) and carpel (C) whorl identities. AG is thus the master player in reproductive organ formation. Loss of AG function results in an indeterminate flower in which sepals and petals are produced repeatedly. Therefore, AG is also required for FM termination, the gene being expressed at high levels in an inner sub-domain of whorl 4 (Prunet et al., 2009). By shutting off the stem cell factor WUS (Lenhard et al., 2001; Lohmann et al., 2001), AG integrates stem cell regulation with the general floral patterning and female programme initiation (Bowman et al., 1989; Yanofsky et al., 1990; Sablowski, 2007a, b; Das et al., 2009; Prunet et al., 2009).

Loss of class B gene function feminizes the stamen whorl by producing carpel-like structures and causes premature FM termination immediately above the third whorl (Bowman et al., 1991; Jack et al., 1992). Conversely, ectopic expression of B genes results in partial indeterminacy, with stamens replacing carpels, a phenotype similar to that resulting from loss of function in the sexual boundary gene SUPERMAN (SUP) (Jack et al., 1994; Krizek and Meyerowitz, 1996). More precisely, mutant flowers of the best characterized allele, sup-1, produce extra whorls of stamens at the expense of the gynoecium (Bowman et al., 1992).

Thus, the correct control of FM termination requires both class B and C gene functions (Davies et al., 2006).

The cadastral gene SUP is necessary for the proper development of reproductive organs (Schultz et al., 1991; Bowman et al., 1992). The gene encodes a transcriptional repressor with a C2H2-type Zn finger motif, a serine/proline-rich domain, a basic domain and a leucine zipper-like domain with a C-terminal EAR-like active repression motif (Sakai et al., 1995; Ito et al., 2003; Table 1). Consistent with its role in separating stamens from carpels, SUP is expressed (Ito et al., 2003) at stages 3–5 in the boundary region between whorls 3 and 4 as a ring, in the inner part of whorl 3 and in stamen primordia. By excluding class B genes from the FMC, SUP is insulating AG, allowing more specific control of determinacy and carpel identity at the FMC.

Table 1

Contrasting SUP allelic series used in this study with ecotype, mutation position, protein domain/motif affected and source of material and information

AlleleEcotypeAssociated mutationSUP putative proteinReference
Class 1, superman
sup-1 (EMS)LerNonsense mutation at amino acid 22*Possibly an N-terminal 22 amino acid truncated proteinSakai et al. (1995); this study
sup-3 (EMS)Col (NASC; CS6227)Missense mutation at amino acid 63 (Gly > Asp; GGT > GAT)Altered protein in conserved QALGGH ZnF motifBowman et al. (1992); Sakai et al. (1995)
sup-6WsMissense mutation at amino acid 108 (Pro > His; CCT > CAT)Altered protein in Ser/Pro-rich domainThis study
Class 2, superwoman (including clark-kent and fon, flower organ number, mutants) (Huang and Ma, 1997; Jacobsen and Meyerowitz, 1997)
sup-eA31LerSUP methylation (Southern blot BglII digestion and sequencing the region 31–354, cf. Jacobsen and Meyerowitz, 1997)No significant reduction in transcript level observed (protein quantity or quality?)Ac/Ds line, Kidner C, CSHL. This study§
Class 3, supersex
sup-5 (EMS)LerDeletion starting before the ATG, from –46 to + 568 (614 nucleotides)A truncated transcript encompassing the 5'-UTR + 88 nucleotide intronic + 47 nucleotides of coiled-coil/leucine zipper (456 nucleotides). Not in frameGaiser et al. (1995); Jacobsen and Meyerowitz (1997)

Three distinct phenotypic classes and the corresponding genetic alterations are presented within the context of already published alleles. SUP alleles are in three distinct ecotypes, the results indicating no major ecotype effects (i.e. SUP modifiers). The structure of the gene is sketched below, with the ORF (blue box), intron (grey box), BglII restriction site and the relative position of the mutations in the gene with the corresponding SUP protein domains indicated.

*sup-4 (Col) and sup-2 (Col) have stop codons at the same position in the gene (Sakai et al., 1995). A missense variant mutation at amino acid 156 (Val > Met; GTG > ATG) in the coiled-coil region has been registered in the Ler background, including sup-1 and sup-eA31. This amino acid substitution does not seem to affect SUP function.

The mutation abolishes SUP DNA binding activity (Dathan et al., 2002);

Reduced amounts of SUP transcript are reported in SUP epi-alleles (clark-kent type; Jacobsen and Meyerowitz, 1997).

§The bisulfite sequencing was targeted to the region surrounding the 5' BglII site (323 nucleotides, including the CT-rich region). Methylation systematically targeted CHH positions. The methylation rate was on average 10 % higher in mutant compared with wild-type samples. In the mutant, nine cytosines were exclusively methylated (including the BglII site; 15–60 % range) and six motifs with up to 3-fold higher methylation rates were registered in the mutant. Four of the motifs contained some of the exclusively methylated cytosines.

Mutant from root transformation regenerants; Ds is not linked to the mutation. BglII restriction fragment of 805 nucleotides spanning the intron to the 3'-end region.

An external file that holds a picture, illustration, etc.
Object name is mcw023ilf1.jpg

This is the simplified picture. As a matter of fact, there are at least three additional layers to this picture that this report is addressing in the frame of the concept of ‘meristematic cascading/transference’ mentioned above, and, ultimately, FM homeostasis. First, the nature of the sup phenotype and functional models have largely been based on one allele, sup-1, a nonsense mutation. Other alleles have been reported (Schultz et al., 1991; Bowman et al., 1992; Gaiser et al., 1995; Jacobsen and Meyerowitz, 1997, and references therein), several of which exhibit phenotypes departing from sup-1 (Table 1). Such alleles mimic CLV pathway alterations affecting RLK (receptor-like kinase) receptors, a HSP90-like chaperone or a kinase-associated protein phosphatase (KAPP) (Clark et al., 1993; Ishiguro et al., 2002; Tichtinsky et al., 2003). The question then is whether and, if so how, SUP integrates the WUS repression system controlled by partly redundant CLV and AG pathways (Prunet et al., 2009). Secondly, SUP is also expressed at later stages: in adaxial and distal parts of anthers at stages 5–7 (in connective, vasculature) and at stages 9–12 (14) in the septum and developing ovules (in funiculi, integuments) (Ito et al., 2003). Thirdly, other members of the SUP gene family, such as RABBIT EARS (petal–stamen boundary; Takeda et al., 2004) and KNUCKES (KNU; carpel–placenta boundary; Payne et al., 2004), represent flower- and whorl/organ-specific boundary regulators. For example, knu mutants produce reiterated stamen–carpel structures at the level of the placenta.

A shortcut view to the above considerations is that WUS regulation at the FMC can be seen as a co-ordination between whorls 3 and 4: in whorl 3, class B genes induce SUP to control cell growth (proliferation rates and differentiation, with models according to Davies et al., 1999; Sakai et al., 2000; Prunet et al., 2009); in whorl 4, AG regulates stem cell activity and cell differentiation via KNU (Sun and Ito, 2010). Class B and C proteins participate in quartet MADS protein complexes where their relative abundance determines the balance of male and female organs/identities (Liu and Mara, 2010; Smaczniak et al., 2012).

Taken together, these data suggest that the analysis of SUP functions should consider both early and late stages of reproductive organ development in an attempt to reveal possible interactions between as yet unlinked processes (FM patterning and termination, and sexual organ identity and differentiation).

To help clarify some of these questions, we (1) made a comparative genetic, morphological, cellular and molecular analysis of the chosen sup alleles and (2) dissected the combined early and late effects of SUP mutations in a spatial–temporal manner. To do so, we performed crosses between allelic classes, between sup alleles and mutations in genes affecting reproductive organ identity and flower determinacy, as well as crosses to reporter genes and in situ hybridization with some of these genes.

We show that SUP contributes to an integrative process of meristematic capacity transference in which FMC, carpel and placenta formation offer new opportunities to question theories on the origin of bisexuality in angiosperms (Frohlich and Parker, 2000; Theissen et al., 2002; Baum and Hileman, 2006) and the making of the robust modern eudicot flower through developmental homeostasis.


Plant growth, crosses and line isolation

Mutants lines were obtained from the Nottingham Arabidopsis Stock Center (SIGnAL T-DNA-Express,, the INRA deVersailles stock ( and the Meyerowitz lab (California Institute of Technology, CA, USA), unless stated otherwise (see Table 1). The reporter lines pAP3::GUS, pAGi::GUS and pWUS::GUS were provided by the Meyerowitz lab. Plants were grown as reported in Prunet et al. (2008) and Ito et al. (2003).

Desired mutant combinations were identified among phenotypic categories in the F2 segregants. Genotypes were confirmed by monitoring Mendelian ratios, by progeny testing, PCR analysis and sequencing. Observed averages were compared using the t-test.

PCR, RT-PCR and sequencing of amplification products

RNA isolation and amplification conditions were as described in Breuil et al. (2004) and Prunet et al. (2008). The following primers were used: SUP (SUP4, TTGTGAACTTCGAATCCTTTATGG and SUP9, GATATATCTTAGATTTTTCCCAGGGG), AP3 (Jack et al., 1992), CLV1 (Clark et al., 1993), CRC and SPT (Alvarez and Smyth, 1999), CUC2 (Kamiuchi et al., 2014), KAPP (Williams et al., 1997), KNU (Payne et al., 2004), PAN (Das et al., 2009) and WUS (Lohmann et al., 2001).

Sequencing was performed on an ABI PRISM® 3100 Genetic Analyzer sequencer (Perkin Elmer, USA), according to the manufacturer’s instructions. Computer-based sequence analysis was performed using the VectorNTI® Suite (Informax, USA) and Sequencher 4.0 (Gene Codes Corporation, Ann Harbor, MI, USA) software. Multiple alignments were obtained using ClustalX and edited with SeqVu (The Garvan Institute of Medical Research, Australia).

Bisulfite sequencing

Eight sup-eA31 plants with checked phenotype and six wild-type plants were analysed. DNA extraction and quantification were performed with the DNeasy plant mini kit Qiagen 69106 and DropSENSE 96 TRINEAN, respectively. A 500 ng aliquot of DNA was treated with bisulfite according to the ‘Cells-to-CpG’ Applied Biosystems 4445555 instruction kit, with the following time and temperature conditions: 2 min at 95 °C, nine series of 2 h at 75 °C and 2 min at 95 °C. The reaction products were purified according to the Cells-to-CpG protocol. The PCRs were performed on the purified DNA with TAKARA RR006A Extaq with the following designed oligos: SP Sup1 Bis2, GCGTTTAAGAATTAGTTTTTTTTATAGTTTTAAT; and ASP Sup1 Bis2, CAAAAACTACAAATATAAAA TCTTAATAACCATAAAAACC. The PCR products were purified and sequenced with the ASP Sup1 Bis2 oligo by GATC company.

The methylation profiles were determined with the BISMA program (

SUP antibodies

Antigenic specific motifs, eventually discriminating between alleles, were searched including two of the closest SUP homologues (At2g37740 and At4g17810). Prediction of the secondary structure was based on DSC, PHD and SOPM methods, and physico-chemical profiling (antigenicity, solvent accessibility and degree of hydrophilicity) was performed with ClustalWProtein.

Four synthetic antigenic peptides were designed (Prosite data and procedures according to Two were common to the three SUP genes (amino acids 191–204 and 80–100) and two were SUP-specific, one in the ZnFinger motif (amino acids 18–31; RTSPWSYGDYDNCQ) and the second in the LeuZipper domain (amino acids 136–153; XSPKSKHTPENACKTKKSSX). Antisera were raised in rabbits and mice by ChangBioscience, Eurogentec and/or Covalab).

None of the tested antibodies, including the commercial brand (Santa Cruz Biotech), worked with plant extracts, despite consistent reactions in bacterial samples.

Inflorescence (cut flower) culture and bromodeoxyuridine (BrdU) incorporation and immunohistochemical detection were as reported in Breuil-Broyer et al. (2004).

In situ hybridization and GUS staining

Tissue preparation was as described by de Almeida-Engler et al. (2009). Tissue sections were 10 μm thick. Probes for in situ hybridization were transcribed using the digoxigenin labelling mix (Boehringer, Germany) as described in Prunet et al. (2008). β-Glucuronidase (GUS) activity was assayed according to Jefferson and Wilson (1991).


Plants were observed with a Leica MZFLIII stereomicroscope coupled to a DC300F digital camera (Leica Microsystems, Germany). For morphological analyses, imaging of reproductive structures was systematically performed to ensure homogenous and repeated analysis capacity. Images were processed with the FW4000 software (Leica). In situ hybridization signals were visualized using a Nikon epifluorescence microscope (Optiphot-2, Nikon, Japan) coupled with an AxioCam MRc digital camera (Carl Zeiss, Germany). Images were processed with the AxioVision software (Zeiss). All figures were composed with Adobe Photoshop 7.0 (Adobe Systems, USA), and certain pictures of flowers were treated to remove unrelated background.


Most of what is known about SUP has been determined through a systematic analysis of one allele, sup-1 (Bowman et al., 1992; Sakai et al., 1995, 2000; Ito et al., 2003). Therefore, one might miss some key aspects of the gene’s functions because a detailed comparative analysis of allelic types has not been performed. For example, there are sup-2 and sup-3 alleles (with flowers similar to sup-1, i.e. excess of stamens and almost no female organs). sup (epi)alleles and sup-5 showing an excess of carpels have been reported (Gaiser et al., 1995; Huang and Ma, 1997; Jacobsen and Meyerowitz, 1997). Table 1 gives an inventory of chosen alleles by classifying them into three groups: superman, superwoman and supersex. To generate a meaningful comparison of the chosen alleles, we performed one large and comprehensive series of crosses of sup alleles to mutants altered in reproductive organ identity and flower determinacy (Tables 2 and and3).3). Therefore, some of these crosses are repetitions of past results, including ag, clv, pan or class B gene overexpressors.

Table 2

Morphological analysis of single and double mutants assessing male, female and chimeric traits, and their proportions

Allele and crossesNo. of flowers (% male flowers)Stamens/flower (s.d.)Free stamens (%)Fused stamens (%)Pistillody (%) w/m/sCarpels/flower (s.d.)*Genetic interactions, observations
WT50 (0)5·7 (0·7)100002 (0)
sup-139 (84)13·2 (1·6)8548 w0·133 % perfect flowers
sup-332 (94)11·9 (2·1)854·56·7 w0·122 % perfect flowers
sup-630 (96)10·4 (1·9)813·514 w033 % perfect flowers
sup-531 (0)11·5 (2·1)63408 m, s, w3·2 (0·8)Phenocopy 35S::KAPP
sup-eA3133 (0)7·5 (1)7159 m, s, w2·9 (1)Phenocopy clv1-6
ap3-325 (0)2·2 (1·6)0541003 (0·9)(anti-sup-1)
35S::AP3 ap3-331 (87)11·7 (1·4)919·518 w, m0·66 % perfect flowers, phenocopy sup-1
clv1-625 (0)6·1 (0·3)100002·6 (0·7)
35S::KAPP§42 (0)13·6 (1·7)711711 m, s, w3·4 (1·5)Phenocopy sup-5 sup-1; clv1-6 sup-5
clv1-6 sup-138 (0)16·4 (0·4)80173 w (m, s)1·6 (1·2)Synergistic St, additive Ca; enhanced pistillody
clv1-6 sup-531 (0)14·4 (2)68345 m, s4·6 (1·3)Synergistic
sup-5 sup-148 (0)13·6 (2)80122 m, s2·8 (1)Additive
sup-5 sup-357 (0)12·6 (1·4)80176 m, s1·6 (1·2)Weak synerg· St/additive Ca; dose effect??
sup-5 sup-634 (0)10 (1·5)811115 w, m0·6 (0·7)Weak epistatic (to weak additive Ca?)
eA31 sup-556 (0)7·8 (1·8)85910 w,m2·8 (0·9)eA31 epistatic
eA31 sup-159 (0)7·8 (2·5)8683 w, m, s2·1 (0·6)eA31 partly epistatic
eA31 sup-343 (0)6·3 (1)912·52 w2·5 (0·8)‘Near wild-type’
eA31 sup-638 (0)6 (0·3)82002 (0·2)‘Wild type’

Corresponding phenocopy candidates are evaluated· The general trend is a partial, limited degree of indeterminacy. Counts were performed on flowers 3–10 of the main inflorescence. w, weak; m, medium; s, strong; s.d., standard deviation.

All flowers were counted for flower organ numbers. Sepals and petals were removed and each flower was photographed to record the organ phenotypes at the FMC only.

*Female fertility: 0·1–0·5 % of flowers form single carpels in sup-1; sup-3 – 1 silique/500 flowers; 20–30 % of misshapen carpels in sup-5 produce low seed set; 70 % of carpels are normal in epi-supA31, average seed set.

t-test performed between the wild type and mutants, single or double, and in pair-wise comparisons of mutants. No statistically significant differences were registered in the categories corresponding to epistatic, ‘near wild-type’ and ‘wild-type’ genetic interactions (alpha risk of 0·05). Ca, carpel; St, stamen.

Ectopic expression of AP3 results in the persistent expression in whorl 4 of the endogenous PI gene (Jack et al., 1994). Similar figures where produced in p35S::AP3/PI flowers (not shown).

§The published 35S::KAPP line caused phenotypes similar to weak loss-of-function clv1 alleles (cf. Williams et al., 1997).

Table 3

Analysis of genetic interactions between SUP with CRC, SPT, PAN and KNU as evaluated at the flower meristem centre (FMC)

AlleleNo. of flowersStamens/flowerFree St (%)Fused St (%)Pistillody (%) w/m/sCarpels/flowerObservations
Wild-type*505·7 (0·7)100002 (0)
sup-1(E) 4215 (2·4)9126·6 w0·2 (0·5)15 total ReprOrg
(L) 5810·8 (1)872·77·6 w0·1
sup-52113·7 (3·2)604011 s, m, w2·9 (1)16·7 total ReprOrg
knu sup-52817 (2·4)89209·5 s, m, w3·6 (1·2)Internal mass prolif.; carpel reiterations
pan-3 sup-13110·3 (1·7)80 (95)3·716 leafywm2 (1·2)Ca restoration; wh5 structures
pan-3 sup-5(E) 257·3 (1·4)84 (94)210 leafy3·5 (1·4)Ca restoration; wh5 carpel reiter.; most carpels WT
(L) 208·9 (1·6)801602·1 (0·8)
crc-1 knu§255·8 (0·6)100002·6 (0·7)Fifth whorl, reiterated stamens and carpels
crc-1 sup-14914·7 (1·6)903·37·4 leafyvar0·04sup-1 epistatic; leafy pistillody, sterile
crc-1 sup-55310·6 (1·5)55434 leafy0·8 (0·5)Attenuation of sup-5 strength, similar to sup-3/6; sterile
spt-2 sup-1(E) 3011·2 (1·6)882·15·3 leafy0sup-1 epistatic; Sterile
(L) 3610·6 (1·2)872·68·5 leafy0·06
spt-2 sup-5(E) 3110·3 (0·8)762211 leafyspt0·9 (0·9)Attenuation of sup-5 strength, similar to sup-3/6; pistil apically unfused, fewer ovules, sterile
(L) 258·5 (0·6)742715 leafymsw1

Early and late flower phenotypes were observed in certain mutant combinations. Parameters as in Table 2.

Ca, carpel; E, first ten flowers; L, late flowers. Pistilody levels, w, weak; m, medium; s, strong; Prolif, proliferation; ReprOrg, reproductive organs; St, stamen; wh, whorl.

*spt, crc, knu: same flower formula as the wild type; in our conditions in knu, 11 % of siliques were tri-carpelloid, 86 % of pistils contained ectopic stamens and 53 % produced knuckles (carpel-like structures) that were derived from placental tissues (Payne et al., 2004).

knu sup-1 – strong indeterminate male flower (Fig. 7A, ,BB).

pan-3: average of Se5 P5 St5; no changes in carpel number and morphology (Running and Meyerowitz, 1996).

§Gynophore is present on most primary carpels; 64 % of outbreaking structures in whorl 5 show elongated internodes with reiterated stamens and carpels.

Extensive organ fusions occurred.

So far, it is not clear, mainly due to the absence of effective SUP antibodies (either commercial or home-designed; see the Materials and Methods), whether the effects of the various alleles within the series are due to variant SUP proteins and/or to the levels of their production and/or degradation. Furthermore, the formation of numerous and variable stamen/carpel chimeric structures raises questions about their origins and precise natures. The dominant idea is that stamen–carpel mosaics in sup-1 ‘result from congenital fusion of carpel primordia and innermost presumptive stamen primordia’ (Bowman et al., 1992; for a contrasting view, see Ronse De Craene et al., 2011).

sup alleles define three distinct mutant classes with partial indeterminacy

By comparing the alleles at the transcriptional and morphological levels (i.e. the levels of indeterminacy and sexual boundary effects, and typology and nature of chimeric organs), three contrasting classes have been defined. Table 1 summarizes the position and nature of the chosen mutations. Tables 2 and and33 present the genetic and morphological analyses of five sup alleles, alone and in various crosses.

All sup alleles produce SUP transcripts.

The generated cDNA samples were visualized by RT-PCR, cloned and sequenced. SUP cDNA fragments from the wild type and all alleles, except for sup-5, migrate as two different size products of 700 and 800 nucleotides, respectively (Fig. 1; also see Ito et al., 2003). The largest transcript containing the intronic sequence is the most abundant form, indicating facultative splicing and a putative role for SUP mRNA in the global SUP functions. Translation of SUP sequences (tBlastN in TAIR and CBI) identified a putative product (in direct 2 frame) (IYLRFFPGEKSEFIFSDLTNTLLLTDVPQIITSTILIIFTIFLSSIAKKKELAH) in the 5'-untranslated region (UTR) that had no hits with TAIR data resources. PCR amplification of sup-5 produced a 289 nucleotide truncated transcript in low amounts (<10 % of wild-type levels). We checked whether the sup-5 transcript sequence and stretches with 50 bp overlaps (Table 1; Supplementary Data Information S1) could generate (cryptic) stem–loop precursors or mature microRNAs, but no such evidence was found when blasting on TAIR (

An external file that holds a picture, illustration, etc.
Object name is mcw023f1p.jpg

Transcript analysis in the sup allelic series. RT-PCR products with primers SUP 9 and the 3'-UTR; the amplification products have been sequenced and correspond to spliced (lower band) and non-spliced (upper band) transcripts in all alleles, except sup-5. The latter produces a single transcript as a result of a large deletion in the gene (Table 1). In quantitative terms, similar levels of transcripts are generated, except for sup-5, with strongly reduced amounts of cDNA. In the transgenic line 35S::KAPP, essentially no SUP transcript is observed.

Contrasting morphologies in sup allelic classes superman, superwoman and supersex indicate distinct mechanisms of FM termination.

The results of morphological analysis are summarized in Table 2. They focus on changes of reproductive organ number and morphology. sup heterozygotes for the various alleles in the wild-type background were scored phenotypically as wild type, indicating no co-suppression and negative dominant effects or haploinsufficiency. These results do not entirely agree with those of Jacobsen and Meyerowitz (1997) reporting semi-dominance in clk-3 and sup-5 heterozygotes, but are in conformity with sup-5 data from Gaiser et al. (1995) and Huang and Ma (1997).

Class 1, sup-1, sup-3, sup-6 (superman)

These alleles produce partially indeterminate (imperfect) male flowers (on average, a doubling of stamen number as compared with the wild type) (Table 2). Extranumerary stamens form 1–2 additional sub-whorls (i.e. stamens are not formed simultaneously in the mutants). In the first (sub-)whorl, 7–8 stamens are produced instead of six, corresponding to an increase in lateral stamen number. Depending on the allele, 84–94 % of class 1 sup flowers have stamens or slightly aberrant stamens filling up the entire FMC space (i.e. are male flowers, Fig. 2A–H; in B, the very centre space remains unspecified), the rest producing carpelloid structures at their centre (see below). Overall, the fertility of these mutants is very low (0·1–0·2 % of the flowers set seeds).

An external file that holds a picture, illustration, etc.
Object name is mcw023f2p.jpg

Morphological analysis of SUP alleles – mutations of SUP produce a limited degree of indeterminacy. The alleles are: sup-1 (A–D), sup-3 (E, F), sup-6 (G, H), sup-eA31 (I–K) and sup-5 (L–Q). Only structures at the FMC are shown, with most of the stamens removed. Stamen abscission zones and nectaries can be seen (A, D, H, K, M, O). The centre-most region of the FMC can produce normal stamens, i.e. perfect male flowers (A), remain free, i.e. residual apex (B), or exhibit feminized stamens (see below). Only centre-most stamens feminize (i.e. pistillody, sensu Ronse De Craene et al., 2011; see text). The weak to strong feminization series is illustrated: the identity of distinct stamen parts can be traced to colour codes, i.e. silverfish for filaments (D, F, G, H, O1), greenish for connective, brown-yellowish for pollen sacs (D, F, I–K, M-Q). Thecae (pollen sacs) are less susceptible to feminization (C, E, G–K, M–Q). Group 1 alleles (sup-1, -3 and -6). Weak to medium pistillody is observed. The weakest phenotypes show stamens tending to enlarge and stay greenish, with the apical region of the stamen bulging (G). In weak phenotypes, stamens produce stigmatic-like tissue apically (E, F, H) and ectopic ovules can be seen in various positions (C, D, F, H). The series illustrates the remarkable feminization potential of the connective (C–H) and of the filament that becomes wider and shorter (B–H). Groups 2 and 3 (sup-eA31 and sup-5). The carpel-like structures in (J–L, M–O, Q) have a narrow base and strongly expand apically. Medium to strong pistillody is the rule. The stamen-derived structures contain recognizable pollen sacs and filaments (I, J, M–P). An irregularly shaped pistil with extra valves and proliferating medial ridge in (L). Considering organ polarity, on the adaxial side, expanded connectives produce rows of ovules along the vasculature (C, J), resembling placental arrangements. Sometimes style–stigma structures (E, H, I) are observed. On the abaxial side, connectives and filaments can produce one to several valve-like structures (K, M, O, Q). The (O) series represents the same stamen observed from lateral (O1), adaxial (O2) and abaxial (O3) positions. On (strongly) expanded connectives, pollen sacs are visible as disporangiate (E, H, J, M, N, P) and monosporangiate arrangements (C, D, H, M, O). The latter indicate asymmetric transformations. In (P), the adaxial side of a feminized connective produced a secondary ‘carpel’. Scale bars = 500 μm.

Phenocopy of class 1 phenotypes. The p35S::AP3 ap3 genotype (Table 2; also compare with ap3 and with flowers in p35S::AP3/PI plants (Krizek and Meyerowitz, 1996; Yang et al., 2003) has a similar phenotype, including pistillody in stamens produced at the FMC (see below; Bowman, 1994). The same phenotype has been reported in pAP3::AG ag-3 (Jack et al., 1997).

Taken together, results with class 1 support the SUP ‘boundary model’: SUP prevents the spread of AP3 expression into whorl 4 (Sakai et al., 2000; Yun et al., 2002).

In summary, class 1 alleles have a common trend: an indistinct male–female boundary (carpel formation is occasional), suggesting that indeterminacy is linked to the systematic expanding of class B gene function at the FMC, i.e. FM termination is delayed (see also Schultz et al., 1991, Bowman et al., 1992, Breuil-Broyer et al., 2004). The main differences concern the relative strength of indeterminacy, with sup-1 > sup-3 > sup-6, and indicate a quantitative effect on determinacy. The above suggests that the putative defective SUP proteins produced in sup-3 and sup-6 (Table 1) have some biological activity. Amino acid substitutions in sup-3 and sup-6 and their similar phenotypes indicate that SUP ZnFinger-like and Serine/Proline-rich domains are involved in sexual boundary function (Table 1). We conclude that the boundary function requires an intact SUP protein.

Class 2, sup-epiA31 (sup-eA31; superwoman; Table 1)

This class is phenotypically similar to the clark-kent epiallele type and exhibits an altered SUP genomic DNA methylation profile (Table 1; Supplementary Data Information S1). All flowers are bisexual, with stamens on a single whorl (average of 7·5) and whorl 4 indeterminate (approx. 3 carpels per flower) (Table 2; similar figures were reported in the fon1 and clk series of mutants, Huang and Ma, 1997; Jacobsen and Meyerowitz, 1997). Seventy per cent of the carpels were wild type, and thus partly phenocopy weak clavata alleles (Diévart et al., 2003). However, carpels were frequently open, and fusion of stamens and carpels has been observed (Table 2; Fig. 2I–K).

This phenotype indicated that the sexual boundary was only slightly affected as compared with class 1 alleles, with flower indeterminacy resulting in increased carpel number. Such a phenotype could result from a reduction by > 50 % of SUP expression (remember, there was no phenotype in plants heterozygous for SUP) and/or from an altered protein. sup-eA31 is the only allele to produce an increased level of indeterminacy in whorl 4 mainly, i.e. after sexual boundary setting and during carpel initiation.

Class 3, sup-5 (supersex)

All flowers in class 3 are bisexual and exhibit whorl 3 and 4 indeterminacy, i.e. more stamens (average of 12·5) and carpels (average of 3) than in the wild type (Table 2). Similar figures for sup-5 were reported in Jacobsen and Meyerowitz (1997). The majority of carpels were short, misshapen and exhibited various alterations such as club-shaped pistils. Fifty per cent of the carpels were apically open, suggesting medial region defects with frequent callus-like proliferation masses along an uneven and split replum (Fig. 2L, Q).

sup-5 produces a truncated transcript and is a strong allele: the sexual boundary function was delayed and diminished, there being a pronounced carpel number increase.

Phenocopy of sup-5. The sup-5 phenotype appears to combine most of the class 1 and class 2 traits (stamen number increase, i.e. extended class B + C gene domain, but also carpel number increase in whorl 4, i.e. an extended class C gene domain). sup-5 is largely phenocopied by 35S::KAPP (Table 2; Fig. 5B–D).

An external file that holds a picture, illustration, etc.
Object name is mcw023f5p.jpg

FM termination alterations in sup-5 phenocopies and crosses of sup mutants with clv and pan. Stamen, carpel and carpelloid structures in sup-5 (A), p35S::KAPP (B, C), clv1-6 sup-1 (D–G) and clv1-6 sup-5 (H, I). Note the broad range of carpelloid morphologies at the FMC that combine pistillody (medium to strong), organ fusion and strong expansion of apical stigmatic tissues. clv1-6 enhances pistillody strength in the sup-1 background. In clv1-6 sup-5, the medial region was reduced or absent. Organ fusions were similar to those observed in the sup-5 single mutant. Partial restoration of whorl 4 and formation of whorl 5 carpels or carpelloid tissues in pan1 sup-1 (J–L) and pan1 sup-5 (M–O). Note that most primary carpels lack medial ridges (L–N), the leafy type pistillody (K), and seed setting in a whorl 5 gynoecium (N, O). sqn-4 sup-5 exhibited strong indeterminacy, with reiterated and elongating axes bearing carpelloid structures (P). Scale bars = 500 μm.

KAPP is a 2C type kinase-associated protein phosphatase capable of modulating CLV1–CLV3 complex endocytosis through dephosphorylation (Trotochaud et al., 1999). Plants overexpressing 35S:KAPP showed stronger indeterminacy compared with sup-5 (for both stamen and carpel numbers; Table 2), while chimeric organs and organ fusions were similar in sup-5 and 35S:KAPP flowers. Such phenotypes are reminiscent of strong clv alleles (Dievart et al., 2003). This suggests that this genotype opposes, or at least dampens, CLV signalling (Williams et al., 1997).

We subsequently investigated SUP expression in these ectopic KAPP plants and show that SUP transcription was abolished in the corresponding 35S:KAPP flowers (Fig. 1). SUP appears therefore as one of multiple and redundant regulatory networks of stem cell signalling and homeostasis operating in the FM (Tichtinsky et al., 2003) that require KAPP function. This is supported by the phenotypic indeterminacy strength scale 35S:KAPP > sup-5sup-1 > clv1-6 (Table 2), cautiously acknowledging the limits of ectopic gene expression data interpretation.

In summary, the sup-5 data support the ‘cell proliferation control’ model in which SUP function consists of controlling the balance of cell proliferation and differentiation at the FMC (whorl 3 vs. whorl 4) (Sakai et al., 2000).

In conclusion, in the analysed allelic series, the allelic strength of indeterminacy is sup-1/sup-5 > sup-3 > sup-6 > sup-eA31. The large deletion of sup-5 and the N-terminal truncation in sup-1 produce the strongest phenotypes. Of note, sup-1 produced significant transcript levels, suggesting that sup-1 transcripts with an early stop codon are not degraded as often occurs by nonsense-mediated decay, and that sup-1 translation might be re-initiated from a downstream methionine. This suggests that sup-1 may not be null. In addition, sup-5 and sup-eA31 show that in the corresponding mutants the boundary activity can be (partly) dissociated from other SUP functions.

Crosses between allelic classes inform on structural and functional SUP features

Heteroallelic combinations supa/supb were generated, and the results of the morphological analysis are presented in Table 2 and Supplementary Data Fig. S2. The results indicated that sup-5 truncated transcripts do not operate through a co-suppression mechanism or through a dominant-negative effect in diallelic contexts.

Crosses between class 1 and class 3 alleles showed a range of situations, from additivity to epistasis, confirming the allelic force series for class 1: sup-1 > sup-3 > sup-6. The following characteristics of diallelic combinations have been observed:

  1. sup-5 sup-1 showed additive effects with, as a result, the highest indeterminacy levels in the sup context (and phenotypically the closest to the transgenic 35S::KAPP line). Comparing total stamen number in sup-1 and sup-5, sup-1 produced more stamens (+2), but the overall number of reproductive organs in the two alleles was rather similar in terms of number of primordia formed at the FMC (Tables 2 and and3),3), suggesting that the two alleles have an equivalent strength of overall indeterminacy. However, based on their contrasting phenotypes and corresponding DNA alterations, sup-1 and sup-5 are candidates for alleles lacking a functional SUP protein (Table 1). Therefore, sup-5 sup-1 could represent the equivalent of a null ‘allele’. If this is the case, we cannot exclude the fact that the altered SUP transcripts in sup-5 could operate through effects at the RNA level. Finally, the heteroallelic heterozygote phenotype suggests that boundary formation and FM termination represent variations on the same theme. This remains to be further explored.
  2. sup-5 sup-3 showed weak synergistic effects on stamen number (whorl 3) at the expense of whorl 4 carpel number, suggesting sexual boundary sliding towards the FMC (expansion of B activity, delay in B exclusion) as compared with sup-5 alone. One likely hypothesis is that one dose of a putatively altered SUP-3 protein in the ZnFinger motif could compensate for some of the whorl 4 defects in a sup-5 context and leave more room for stamen production.
  3. sup-5 sup-6 showed partial sup-6 epistasis, indicating that one dose of the putatively altered SUP-6 protein in the Ser/Pro-rich domain could compensate for most sup-5 defects in whorl 4. Interestingly, siliques in the double mutant were frequently helicoidal, indicating substantial asymmetric valve growth (Supplementary Data Fig. S2E, F) and a role which SUP could have in the formation of the carpel medial zone. In comparison with sup-5 sup-3, the results indicate that a functional ZnFinger motif is determinant for a functional SUP.
  4. e-supA31 (class 2) showed (partial) epistasis over sup-5 (class 3) and sup-1 (class 1) in the corresponding double allelic combinations, and almost full complementation to the wild type when crossed into class 1 sup-3 and sup-6 alleles (Table 2; Supplementary Data Fig. S2C, D, G). The latter cases suggest that, in contrast to the monoallelic contexts, these diallelic combinations produced sufficient amounts of functional SUP protein to complement most of the specific allelic defects. Although we cannot exclude enhanced instability of the epi-allele in inter-ecotype crosses (reduction of methylation levels), we interpret this as suggesting that functional SUP acts as a dimer. The corresponding diallelic (hetero)dimers with sup-3 and sup-6 are likely to produce distinct levels of inter-molecular complementation.

In the light of the above considerations, we focused the analysis on sup-1 and sup-5.

Early SUP functions – linking sexual boundary and FM processes

To investigate the dual role SUP plays in sexual boundary formation and FM termination, crosses to a series of reporters (including AP3, AG and WUS reporters), to a clv null allele and to several other regulators of FM patterning and termination were performed. In addition, WUS and a general boundary marker, CUC2, were analysed by in situ hybridization, and BrdU labelling patterns were observed in parallel.

(i) Linking sexual boundary and FM termination

AP3 is expressed at the FMC in sup mutants.

Exclusion of class B genes from whorl 4 by SUP is necessary for floral determinacy; their ectopic expression at the FMC somehow opposes FM termination. AP3 expression is altered in sup-1 (Bowman et al., 1992). We wanted to compare sup-1 and sup-5 by observing pAP3:GUS and pAGi::GUS expression in the two alleles. pAP3::GUS early patterns showed that the gene was not expressed in whorl 4 in the wild type, but was systematically present at the FMC in sup-1 and sup-5 at flower developmental stages 4–6(7) (Fig. 3A–F). In comparison, the pAGi::GUS pattern was unchanged in sup alleles at these early stages (not shown), suggesting that AG expression per se was not altered in sup flowers as compared with the wild type (also shown by in situ hybridization; Sakai et al., 1995). This means that altered levels of class B vs. class C genes are present at the FMC at stages at which the acquisition of whorl identity and organ initiation occur. The data are in agreement with the morphology of class 1 alleles, which clearly indicated that most flowers produced stamens at their FMC. Therefore, this pattern of class B gene expression can fully explain the class 1 phenotype, but not class 2 and class 3 phenotypes.

An external file that holds a picture, illustration, etc.
Object name is mcw023f3p.jpg

Class B and C gene-driven GUS reporter expression analysis in sup alleles. pAP3::GUS patterns in early stages of the wild type (A; stages 4–5/6) and sup-5 (B; stages 6–8), and late stages in the wild type (C; stage 12), sup-1 (D, E; stages 10–12/13) and sup-5 (F; stage above 12). Note that whorl 4 is not stained in the wild type. pAGi::GUS pattern in the wild type (G, H; stages 8/9 and 12, respectively), sup-1 (I; stages 10–12) and sup-5 (J, K; stage 12). Note that patterns differ exclusively in structures developing at the FMC; stamens exterior to the region express a wild-type reporter profile. Connective tissues in the wild type and connective- and filament-derived structures in the mutants are sites with persistently high GUS expression levels. Pistillody ranges from weak to medium. Asymmetric staining is frequently observed (I, K), most carpelloid tissues in adaxial positions (asterisk) showing no reporter expression. The arrow in (E) indicates stigmatic tissues. Scale bars = 50 μm (A, B), 100 μm (H), 200 μm (C, E–G, I–K) and 400 μm (D).

New redundancy at the FMC: a SUP-dependent mechanism of FM termination.

The FM stem cell system depends on two independent pathways, the CLV–WUS and the AG–WUS pathways (Sun and Ito, 2015): the former spatially restricts and maintains the stem cell population, while the latter controls the temporal arrest of the stem cell activity. Thus the level and timing of WUS repression is paramount for FM homeostasis. Since SUP is primarily pointing to class B gene activity alterations, it is not clear whether and how the SUP–class B gene system is targeting WUS. To that end, we have analysed WUS expression in the mutants and made crosses of sup alleles into mutants of the CLV signalling system.

WUS expression is temporally, but not spatially, altered in sup-1 and sup-5 mutants. WUS expression patterns inform on spatial (extent of cell proliferation) and temporal (delayed termination) control of the FM. To evaluate the situation in sup, in situ hybridization with WUS as probe showed that WUS expression is prolonged in the FM to late stage 6 in sup-1 and sup-5 as compared with the wild type, where the signal was detected up to stage 5/6 (Fig. 4A–H). We concluded that FM termination is slightly delayed in sup mutant flowers due to an altered WUS signalling. Prolonged WUS expression can explain that extra organs are sequentially produced in sup alleles, but it does not allow understanding of the morphological differences between sup-1 and sup-5 alleles.

An external file that holds a picture, illustration, etc.
Object name is mcw023f4p.jpg

WUS expression patterns in early and late stages of flower development in sup alleles. In situ hybridization in the wild type (A–D), with two distinct sections at stage 5–6 (B, C), and in sup-5 (E–H), with two serial sections at late stage 6 (G, H). The following structures are visible: in (A), SAM, stage 1 and 2, and stage 9/10 buds; in (D), stages 6/7 and 12, the latter with the dehiscence zone labelled; in (E) SAM, stages 3, 4 and 10, the latter with the dehiscence zone labelled; in (F) FM and stage 6. Similar patterns have been observed in sup-1 (not shown). pWUS::GUS expression in ovules of a wild-type pistil (I; stage 9/10), sup-1 (J) and sup-5 (K) both at stage 10. Note that such ovules are produced in stamens with medium pistillody strength. The sketch (L) illustrates stamen morphology and pistillody leading to the simultaneous production of ovules and pollen sacs in mono- and bi-sporangiate arrangements (M). Scale bars = 50 μm (A, I) and 500 μm (J, K, M).

sup crosses to clavata1 and squint mutants show strong synergic interactions. CLV1 and SQN are more general regulators of FM stem cell homeostasis loops (Prunet et al., 2008). The morphology of single mutant/transgenic line flowers indicated the following phenotypic strength series: 35S::KAPP > sup-5sup-1 > clv1-6 (Table 2).

sup-1 and sup-5 crossed into the clv1-6 null allele resulted in more severe phenotypes than any of the single mutants (Table 2). In these mutants the apical region of carpel or carpelloid structures was strongly expanded, with abundant stigmatic tissues being produced (Fig. 5A–I). However, some important differences need to be acknowledged: the double clv1-6 sup-1 mutant showed synergistic effects on stamen number (with stamens at the FMC exhibiting stronger levels of pistillody), while carpel number was intermediate (additive) with respect to single mutant phenotypes. In contrast, the morphology of clv1-6 sup-5 was consistent with synergistic effects for both stamen and carpel numbers. The carpelloid structures had a lamellar shape with few if any medial ridge tissues. In the sup-5 context, the strength of the observed indeterminacy was sup-5 clv1-6 > 35S::KAPP > sup-1 sup-5 (Table 2), indicating a combined effect of (partly) independent pathways operating through SUP and CLV1 regulatory networks.

We refined the analysis by performing crosses with the clv3-1 mutant. The resulting double mutant phenotypes were extreme indeterminate flowers (loss of determinacy), the phenotypic strength series being clv3-1 sup-1clv3-1 sup-5 > clv3-1 35S::KAPP (Supplementary Data Fig. S3). The result suggests that CLV, KAPP and SUP overlap in qualitatively distinct manners. Concerning the AG–SUP interaction, sup ag double mutants show synergistic effects including fasciation at the FMC (i.e. loss of determinacy; Bowman, 1994; Huang and Ma, 1997; Prunet et al., 2009; data not shown), pointing to an AG-independent mechanism of FM termination control by SUP.

That sup-5 acts in a broader regulatory context than sup-1 was further substantiated by phenotypes generated in crosses with mutants altered in SQUINT. SQN is expected to activate AG and modulate stem cell termination through the CLV pathway (Prunet et al., 2008). In sqn-4 sup-1 and sqn-4 sup-5, the mild effects of sqn-4 on flower morphology were strongly enhanced by sup alleles, with sup-1 impacting on stamen number (Prunet et al., 2008) and sup-5 on carpel number and carpel reiterating axes (extra whorls of carpels; Fig. 5P).

Taken together, these results point to the fact that SUP works in synergy with more general factors of FM termination, indicating indirect genetic interactions between those factors and a SUP–class B gene-dependent mechanism of FM termination. Thus, two FM-specific termination pathways are at work: a class C gene- and a class B gene-dependent pathway.

(ii) Linking sexual boundary and FM patterning

SUP, CUC2 and PAN.

The above results prompted us to investigate the interaction of SUP contrasting alleles with more general meristem patterning and boundary regulators, such as CUC2, and with FM patterning factors, such as PAN. The question is whether SUP, in addition to specifying the sexual boundary and stamen number in whorl 3, is also affecting FMC patterning.

Differential expression of CUC2 in sup-1 and sup-5

CUC2 is a NAC gene family member expressed in all flower organ boundaries, in the medial region of stage 8 carpels and in septum tissues (Kamiuchi et al., 2014, and references therein). The CUC genes are redundantly required for carpel margin formation and fusion (Nikovics et al., 2006). In sup-1, the CUC2 expression pattern is similar to that of the wild type, except that the gene is transiently misexpressed at the FMC up to flower stage 6 (Breuil-Broyer et al., 2004; Fig. 6M–T). Alternatively, the signal marked the boundary of a strongly diminished whorl 4. In sup-5, the CUC2 signal was faint at the perianth level and erratic to absent in presumptive central whorls of the FM. Thus, CUC2 expression is altered differently in sup-1 and sup-5, and represents an important indication that the observed patterns affect, albeit in distinct ways, FM indeterminacy.

An external file that holds a picture, illustration, etc.
Object name is mcw023f6p.jpg

Cell proliferation patterns in the FM and at inter-whorl and inter-organ boundaries in sup-5 observed by BrdU incorporation and CUC2-1 in situ hybridization. BrdU labelling was monitored on (A) shoot apex, (B–D) three consecutive sections of stage 4–5 flowers, (E, F) two consecutive sections of stage 6 flowers, (G, H) two consecutive sections of stage 8 flowers and (I–L) older flower buds (from stages 10 to 12). CUC2-1 expression was characterized on (M) inflorescence meristems, (N–Q) stage 3–5 flowers and (R–T) stage 8–12 flowers. (L) Close-up view of (K). Arrowheads indicate non-dividing cells organized as a true boundary, stars indicate non-dividing cells islets at the base of stamens and arrows point to small groups of cells at the base of, and between, stamens that weakly express CUC2-1. All flowers are oriented with the apex of the flower towards the top. im, inflorescence meristem; fm, floral meristem; se, sepal, pe, petal; st, stamen; cs, chimeric carpel–stamen structure. Scale bars = 60 μm.

To see whether these CUC2 patterns correlate with changes in cell division patterns at the FMC, we have analysed the topographical distribution of BrdU-labelled cells in the wild type, sup-1 and sup-5. In sup-1, at early stages 4–6, BrdU labelling profiles were similar to those of the wild type: clear-cut stretches of unlabelled cells (boundaries) separate successive whorls of stamen initials (Breuil-Broyer et al., 2004). Subsequently, during organ morphogenesis (stages 8–14), such boundaries became narrower and more irregular (Breuil-Broyer et al., 2004). Occasionally, BrdU-labelled cells could be observed in epidermal and sub-epidermal layers of such boundaries.

In sup-5, only sepal–FM and petal–FM boundaries were easy to distinguish at the early stages 4–6, the rest of the meristem consisting of proliferating cells spread across the FM (almost all cells of the FM have incorporated BrdU during the pulse). Subsequently, growing stamen, carpel and carpelloid primordia (strong label) were not separated by stretches of unlabelled cells, as shown in the analysed serial sections (Fig. 6A–L).

In brief, BrdU labelling patterns confirmed the observed CUC2 hybridization profiles. The results showed that sup-5 flowers lacked proper general and specific boundary functions at the level of whorls 3 and 4, with, as a consequence, no apparent cell division domains at the FMC, i.e. rather homogenous BrdU incorporation/cell (over)proliferation patterns on both sides of the putative whorl 3–whorl 4 boundary.

Taken together, CUC2 and BrdU patterns represent so far the most refined molecular and cellular results that best discriminate between sup-1 and sup-5 alleles.

PERIANTHIA (PAN)–SUP interactions in FM homeostasis

Because SUP controls cell growth redundantly with AG and CLV, interactions with other FM patterning factors, such as PAN, could contribute to better understand the role that class B and C genes play in FMC homeostasis. PAN, a bZIP transcription factor, is a flower (pre-)pattern gene that controls flower organ number (Running and Meyerowitz, 1996) and regulates stem cell fate by positively controlling AG expression (Das et al., 2009). This activity is spatially restricted to the centre-most region of the AG expression domain. PAN also exhibits enhanced indeterminacy in double mutants with SEUSS (fifth whorl formation) and contributes to the development of the carpel margin meristem from the medial region (Wynn et al., 2014). pan mutations restore fourth whorl carpels to flowers of the sup-1 single mutant (Running and Meyerowitz, 1996). Reduction in AG expression caused by the absence of PAN could lead to increased or prolonged WUS expression, and thus to an increase in the size of this region (Das et al., 2009).

In crosses with pan-3, pan-3 sup doubles had fewer extra stamens and the innermost stamens showed stronger levels of pistillody with leafy appearance (Table 3; Fig. 5J–O). They produced extra carpels at the FMC, frequently positioned on a gynophore. Thus, pan-3 counteracted most of the sup-1 effects by restoring a male–female boundary (i.e. somehow diminishing B expansion), increasing carpel number and consequently female fertility. In pan-3 sup-5, the same tendencies were observed. In 66 % of the flowers, the restored carpels were located inner to a fourth whorl of primary open carpels, many with a narrow base and a leafy appearance (Fig. 5M–O), i.e. they corresponded to an additional whorl 5. While sup-5 showed expanding proliferation capacity in whorl 3 and 4, pan-3 sup-5 flowers suffered a reduction in whorl 3 to the benefit of whorl 4 (i.e. a likely impact on class B gene function levels). The results indicated that the pan-3 mutation can restore a bona fide, albeit loose, carpel boundary in sup-5. Furthermore, pan-3 also counteracted sup-5 effects on organ fusion, suggesting additional links to the CUC2 boundary system explored above.

The results revealed contrasting roles of PAN and SUP in stamen and carpel whorls. Sexual boundary restoration in the double mutant was associated with extra whorls of carpels formed in the intercarpelary space (also see Prunet et al., 2009). This is important because the effects of PAN and SUP mutant combinations at the centre-most region of whorl 4 allow separation of the boundary setting from FM termination processes. The results further indicate that regulators of FM patterning (PAN), boundary (SUP) and reproductive organs/meristem identity (B/C) are components of FMC patterning and flower homeostasis mechanisms. They also contribute to proper carpel identity, considered as a developmental unit homologous to the leaf (Ferrandiz et al., 2010).

Linking early and late SUP functions – co-ordinating meristematic territories at the FMC

Since SUP activity interferes with both FM patterning and homeostasis, and with general boundary and whorl 4 functions, some kind of integration between these processes is expected to occur in co-ordinating FM termination, carpel initiation and carpel margin meristem/placenta formation. The boundary factors SUP and KNUCKLES (KNU), belonging to the same gene family, might play such a role. KNU has a carpel–placenta boundary function, i.e. the next boundary in space and time at the FMC. The early spatial–temporal expression profiles of SUP and KNU are complementary. KNU expression is initiated at stage 6 at the FMC (Ito et al., 2003; Payne et al., 2004; Sun et al., 2009). SUP and KNU expression patterns overlap during carpel morphogenesis, in the carpel margin meristem region (placenta, septum).

In the context of carpel initiation and development, a SUP–KNU boundary system might depend on interactions with early carpel-specific factors, such as CRABS-CLAW (CRC) and SPATULA (SPT), known to promote lateral vs. medial zones patterning in the gynoecium (Alvarez and Smyth, 2002).

SUP and KNU are bridging the meristematic capacity in space and time.

So far, the analysis concentrated on processes that initiate FM termination. We wanted to understand how FM termination is maintained. The question is whether SUP and KNU boundary factors are central players in the initiation/maintenance of FM termination. Genetic relationships of AG, KNU and WUS suggest that these genes function in the same pathway in floral meristem control, KNU being a key link in the feedback regulation of WUS by AG (Sun et al., 2009).

Crosses of knu1 to sup-1 and sup-5 resulted in synergistic effects. knu1 mutant flowers produced ectopic stamens and carpels in a reiterating pattern inside fourth whorl carpels (Payne et al., 2004). In knu1 sup-1, strong indeterminate male flowers were observed (Fig. 7A, B) corresponding to an enhanced sup-1 phenotype. In knu1 sup-5, carpel number was increased (Table 3), and reiterated carpelloid structures, increased internal mass and callus-like tissues were produced inside the primary carpels (Fig. 7C, D). The synergistic effects in knu1 sup-5 indicated independent but co-ordinated mechanisms linking the control of FM termination and subsequent development events leading to medial region formation. The production of abundant internal tissues and callus in knu1 sup-5 further indicated that SUP and KNU jointly control that region. One can hypothesize that SUP and KNU are bridging over early (sexual boundary and FM termination initiation as illustrated by knu1 sup-1) and late (FM termination maintenance and medial region formation as illustrated by knu1 sup-5) meristematic functions. In other words, SUP and KNU redundantly overlap to initiate and maintain FM determinacy.

An external file that holds a picture, illustration, etc.
Object name is mcw023f7p.jpg

The meristematic continuity process revealed in morphological traits of double mutant combinations of SUP, KNU, CRC and PAN. The double mutatnts are knu1 sup (A–D), crc1 knu1 (E, F) and crc1 pan-3 (G–I). In knu1 sup-1 (A, B), the sup-1 phenotype was strongly enhanced, stamens being produced in extravagant numbers. Similar trends of indeterminacy consisting of reiteration of leafy carpelloid structures inside supra-nummerary primary carpels were observed. In knu1 sup-5 (C, D), fused or unfused primary carpels can form with proliferative masses along medial ridges. In the other combinations, carpels are not fused. Reiterated carpel whorls can form on elongated axes, with occasional stamen formation. Scale bars = 1 mm.

Linking FM termination and carpel compartmentation – unravelling the FMC meristematic continuity.

The above hypothesis questions whether proper termination of the FM is required for formation of carpel primordia and carpel margin meristem. To understand the possible role played by the carpel polarity genes CRC and SPT in FM termination maintenance, we crossed sup and knu to crc and spt mutants. Of note, CRC and SPT were shown to pattern the gynoecium through indirect genetic interactions (Alvarez and Smyth, 2002): CRC is an early component of carpel identity and abaxial polarity (Alvarez and Smyth, 2002), while SPT is essential in carpel adaxial polarity and carpel margin meristem formation (Alvarez and Smyth, 2002; Nahar et al., 2012). Similar results were obtained with the respective double mutant combinations (Table 3; Supplementary Data Fig. S4). Only results concerning crc sup are presented below.

In crc1 sup-1 double mutants, sup-1 was epistatic to crc1 (Table 3; Supplementary Data Fig. S4B, C), which is an expected outcome when considering the sup-1 phenotype and that the spatial–temporal expression profiles of SUP and CRC do not overlap. However, in crc sup/+ backgrounds, indeterminacy resulted in both an increase in carpel number and mild whorl 5 phenotypes (Fig. S4I–K).

In crc1 sup-5, mutant flowers exhibited reduced indeterminacy for stamens and, to a larger extent, for carpels as compared with sup-5 (Table 3; Supplementary Data Fig. S4D, E), i.e. the attenuation of sup-5 allelic strength, the phenotype being similar to a class 1 weak sup allele. Note the enhanced leafy-type pistillody traits. Thus, the crc1 sup-5 phenotype indicated that (late) SUP and CRC functions overlap and participate in the FM termination–carpel patterning switch. An active KNU might confer the bridging context.

To explore this possibility, we analysed the knu1 crc1 double mutants: strong indeterminacy was observed, with supra-numerary (whorls of) stamens and carpelloids inside the primary carpels (Table 3; Fig. 7E, F). Thus, crc enhanced the knu phenotype, indicating synergistic effects between CRC and KNU in defining meristematic territories within the carpel through genetically independent and complementary genetic pathways. Of note, a similar phenotype was produced in crc1 pan-3 double mutant flowers (Fig. 7G–I).

SUP and KNU have an apparently opposite relationship to CRC and SPT. We hypothesize that SUP and KNU are primary FMC components of FM termination initiation, KNU operating through the AG pathway and SUP through an independent route. Since sup-1, but not knu-1, is epistatic to crc, KNU is likely to bridge over FM termination initiation and maintenance, with CRC and SPT having joint and redundant effects in the maintenance process (note that crc spt double mutants are indeterminate, exhibiting extra carpels and stamens inside the primary carpel whorl (Alvarez and Smyth, 1999).

In summary, the results reveal that the proposed SUP–KNU sequence controlling the transition ‘FMC termination to carpel patterning’ is orchestrated in time and space with the differential contribution of carpel patterning factors.

Pistillody and chimeric stamen–carpels: CUC2 differential expression and altered class B/C gene balance in sup-1 and sup-5.

Stamens that produce ectopic ovules and/or stigmatic tissues undergo a feminization process called pistillody (sensu Ronse De Craene et al., 2011). In the study, sup-1 stamens produced ectopic ovules and/or stigmatic tissues, showing various degrees of filament, connective and/or apical region expansion. Connective and filaments (single-veined tissues; Hufford, 1996) were the most responsive to feminization. Laminar filaments, bulging of the apical region of the connective and disporangiate stamens illustrate some basal angiosperm traits (Hufford, 1996). Feminization was observed at stage 8 or later. The study concluded that in this sup class background the primary identity of supernumerary organs at the FMC was stamen.

Similar results were obtained in the analysed allelic series. Approximately 10 % of the stamens in class 1 exhibited weak pistillody (Table 2; Fig. 2A–H). Such bisexual organs producing mega- and microspores at the same time were positioned at the FMC only. In class 2 and class 3 mutants, pistillody was estimated at 5 and 8 %, respectively, of the total number of stamens, ranging from weak to strong (Table 2; Fig. 2I–Q). Again, pistillody occurred in the innermost (whorl of) stamens. In medium to strong cases, connective and filament tissues formed (partial) valve structures. Those structures were frequently aberrant stamen–pistil chimeras with proliferative medial tissues (Fig. 2; Supplementary Data Fig. S5L–P). For example, in sup-5, two-thirds of whorl 4 organs were stamen–carpel mosaics that were present in all flowers.

We conclude that pistillody and stamen–carpel fusions are primarily late developmental events and that it is rather difficult to distinguish between cases of strong pistillody and stamen–carpel fusions. Post-genital fusion of stamens and carpels is well documented in cases of synorganization (Endress, 2006; Ronse De Craene et al., 2011).

On the basis of these results, we wanted to investigate inter- and intra-whorl boundary defects further in stamen–carpel chimeras and class B gene expression patterns at later stages of flower development. The role of the epidermal cell layer in post-genital fusions has been described (Endress, 2006).

Altered and differential CUC2 expression correlates with organ fusion in sup-5.

As reported in an earlier section, no or erratic CUC2 signal at boundaries between and within reproductive organ whorls was observed in sup-5 flower tissues during all stages of flower development. Proper signals were registered in stamens (locule boundaries) and in carpel internal tissues (Fig. 6, late stages), suggesting that sup-5 did not affect CUC2 during organ formation, i.e. during proper determination and differentiation of stamen and gynoecium compartments. The inter-organ and inter-whorl boundary defects in CUC2 expression could explain the observed extensive organ fusions in sup-5 because cell proliferation was maintained in boundary regions in the mutant and thus affected proper spacing between whorls and between organs within a whorl (Fig. 6I–L, S, T). Mainly epidermal cell layers were not labelled, suggesting that organ fusions are essentially post-genital events, as illustrated in Supplementary Data Fig. S5; wild-type in (A–C), sup-5 in (H–P) and sup-1 (D–G) for comparison.

AP3 and AG genes are misexpressed in chimeric organs at FMC.

The profiles of pWUS::GUS, pAP3::GUS and pAGi::GUS reporters were observed in sup mutants at a range of developmental stages. pWUS::GUS patterns in ovule primordia and in stamen dehiscence zones were normal in all mutants, including ectopic and chimeric tissues or organs (Fig. 4, late stages). With pAP3::GUS and pAGi::GUS, while normal patterns were observed in stamens produced in whorl 3 and in most extra stamens, enhanced expression of AP3 and AG reporters occurred in all feminized stamens (i.e. in expanding connective and filaments and in abnormal chimeric anther parts) and were easily detected beyond developmental stage 8 (Fig. 3, late stages) (see the Discussion regarding quartet complex balance/dosage).

Such aberrant patterns could be at the origin of various levels of pistillody produced at the FMC in both sup-1 and sup-5 mutants. Pistillody and chimeric organs occur in 35S:AP3 and p35S:AP3/PI (Krizek and Meyerowitz, 1996; this work). We concluded that an altered balance of class B/C gene activity was present during organ formation in the mutants, affecting reproductive organ traits alone or in combination with other carpel development factors expressed at the centre-most region of the FM.

In summary, the combined effects of altered B and C balance during reproductive organ initiation/maintenance with differential patterns of CUC2 expression between the wild type and each of the two alleles are likely to explain most of the morphological particularities of sup-1 and sup-5 alleles.


Over the last decade, the known functions of SUP have been cadastral (sexual boundary), FM termination and ovule integument differentiation (Schultz et al., 1991; Bowman et al., 1992; Gaiser et al., 1995; Sakai et al., 2000; Prunet et al., 2009). Our results add to the list of described SUP functions, which have included phenotypes range from superman, to superwoman to supersex. This range of morphologies and double mutant phenotypes mimic some traits of basal angiosperm and even more archaic features (Hufford, 1996; Endress, 2001, 2004; Rudall and Bateman, 2010), and are essential in further addressing evolutionary questions of bisexuality, FM termination and flower homeostasis.

The allelic series we analyse – five alleles with differing strengths, due to defined alterations in SUP protein domains or in the SUP transcript – shows that SUP fulfils multiple functions during flower development, starting with stage 6 (also see Huang and Ma, 1997; Sakai et al., 2000) and extending to the medial region of the gynoecium and to ovule development (see Gaiser et al., 1995). SUP helps to determine the sequence of reproductive organ identities and meristematic specialization events at the FMC by the following functions.

  1. Early SUP functions: linking mechanisms in FM termination (a) to male–female (sexual) boundary setting (i.e. B function exclusion, resulting in stamen–carpel identity specification) and (b) to stamen number control (i.e. lateral stamen reduction from four to two) and organ separation. SUP control of FM termination is linked to general meristem genes (such as WUS, CLV, CUC2 and SQN) and to FM patterning genes (such as PAN, AG and class B genes).
  2. Late SUP functions: contributing to medial region formation through carpel/placenta boundary control (in conjunction with KNU, CRC and SPT) and class B/C gene balance for the proper differentiation of carpel meristematic regions and stamen region differentiation.

Models to explain SUP functions that refer to ‘boundary’ vs. ‘cell proliferation’ control processes (Sakai et al., 2000; Prunet et al., 2009) only address the early set of SUP functions. Our results explore a more inclusive view of SUP actions. They are supported by espression patterns of WUS, AP3 and AG, and by BrdU cell division profiles on both sides of the sexual boundary. Accordingly, SUP sets the correct level of cell allocation between whorls 3 and 4 and contributes to the control of the class B/C MADS-box gene balance during both early (FM termination) and late (stamen and carpel tissue determination and differentiation) stages. According to the gradient model of Liu and Mara (2010), in the quartets of MADS protein complexes, the effects of class B and C genes are strongly dose dependent (Smaczniak et al., 2012), because class B and C MADS-box proteins compete for the same partners. The relative abundance of class B and C proteins in such complexes in both early and late stage flowers was demonstrated by in planta formation of organ-specific quaternary complexes (Smaczniak et al., 2012). The results showed a flexible composition of MADS complexes, including the AP3, PI, AG and SEP3 complex (see also Chanderbali et al., 2010). This can explain the large morphological variations observed between and within sup alleles (see below). Of note, ectopic WUS expression in whorl 3 (resulting in out of balance B/C expression due to increased AG in a domain where it is normally present) produces carpelloid stamens (Lenhard et al., 2001; Lohmann et al., 2001).

In this context, the AG, SUP and CLV pathways appear to function at least partially independently to repress WUS (the end-point of regulatory loops required for stem cell production, maintenance/homeostasis and arrest), because the effects of ag, sup and clv1 mutations on floral meristem determinacy are additive to synergic in double mutant combinations (Clark et al., 1993; Lohmann et al., 2001; Bowman et al., 1992; Prunet et al., 2008; this work; also see Liu et al., 2011). In such combinations, the partial indeterminacy observed in the single mutants becomes a more dramatic loss of determinacy.

For example, sup-5 and 35S::KAPP affect meristematic functions in at least partly distinct ways compared with those involving the CLV cascade. Interestingly, it has been proposed that strong clv alleles alter additional RLK signalling cascades within meristems (Diavert et al., 2003), while KAPP has been shown to interact in vitro with sub-sets of plant RLKs (Braun et al., 1997). SUP therefore may act on FM termination through RLK signalling cascades different from CLV, implying redundant signalling control of FM termination. Thus, the results of the SUP allelic series provide further evidence that the determination of floral stem cell fate is a highly redundant process, requiring interactions and integration between domain- and/or stage-specific factors with diverse floral patterning functions, resulting in developmental homeostasis.

The comparison of sup-1 and sup-5 alleles has been very informative on early vs. late SUP functions, but the question of which allele is the neomorph remains open. sup-1 and sup-5 have similar effects on WUS, but cause different changes in the expression of CUC2, a meristem and general boundary factor. In sup-5, in contrast to sup-1, indeterminacy affects both whorls 3 and 4, pointing to a partial dissociation of sexual boundary from FMC growth control, with extension to carpel medial region control.

In sup-1, strong sexual boundary effects with direct consequences on indeterminacy link class B genes and WUS expression control: SUP control of FM termination is mediated by class B genes. Of note, in p35S::AP3 p35S::PI (Krizek and Meyerowitz, 1996), AP3 early expression was present in the whole of the FM, with the exception of a small region at the very centre of the FM. A similar pattern has been reported in sup-1 (Bowman et al., 1992; also see Sakai et al., 1995, 2000), which leaves room for occasionally producing carpel structures in class 1 alleles (Tables 2 and and3).3). An expansion/maintenance over time of this FMC region lacking class B gene expression could explain class 2 and 3 phenotypes. In addition, in several mutants, the centre-most sub-domain of whorl 4 was shown to correspond to an AG-free sector with phenotypes showing indeterminacy (Prunet et al., 2008; Das et al., 2009).

We argue that the sub-domain of whorl 4 is the domain where a reprogramming of meristematic activities can take place. Therefore, the completion of WUS repression by KNU, CRC and SPT is imperative for the proper development of carpel margin meristems (their initiation, positioning and patterning; Liu and Franks, 2015). Of note, carpel margin meristems share regulatory mechanisms with FMs, including CUC, PAN and STM (Mathews and Kramer, 2012; Liu and Franks, 2015). In brief, the whorl 4 centre-most region may have an open developmental fate characterized by a co-ordinated meristematic transference process: from floral meristem to carpel margin meristem to placenta. The process is illustrated in Fig. 8.

An external file that holds a picture, illustration, etc.
Object name is mcw023f8p.jpg

Sketch of early and late SUP functions encompassing meristematic transference and class B/C gene balance control. The left side of the diagram illustrates the modularity of meristematic capacity (sensu Mathews and Kramer, 2012). The co-option (relocation/reactivation) of meristematic genetic module(s) in distinct identity contexts is illustrated by the sequence FMC–medial ridge/placenta (–and ultimately ovule). We posit that SUP and KNU are bridging agents in the way class B and C genes, WUS, CUC and others are deployed during reproductive organ formation. For example, FM termination is a necessary condition for carpel initiation followed by the onset of the meristematic programme on the adaxial carpel blade. This corresponds to an axilary meristem, CUC 2, KNOX, etc. genes (cf. Mathews and Kramer, 2012) operating in conjunction with carpel module-specific identity factors (CRC, SPT). The right side of the diagram introduces other facets of SUP functions operating through the class B and C balance calibration in setting the male–female boundary, in properly specifying the connective and contributing to other mechanisms such as organ polarity control (+ on the figure). The cross-talk between these processes remains to be understood. Flower development is shown through stages 3 to beyond stage 9. At stages 3–4, SUP expression is induced in whorls 3 and 4, the FM homeostasis being under the control of the CLV–WUS feedback loop. The sexual boundary and the pattern of stamen number are being established. At stage 6, class B and C gene territories are refined, the FM termination being initiated/triggered (Ti) through independent mechanisms by AG, SUP and KNU, together with other FM patterning factors (PAN, SQN, CRC, etc) to extinct WUS. Subsequently FM termination needs to be maintained (Tm). This is consistent with SUP expression starting at stage 3 and KNU expression being initiated at stage 6. WUS extinction is gradual in the wild type and arrests at stages 6. At stage 8, in the centre-most region of whorl 4, FM termination is enforced by KNU and CRC/SPT genes (Tc, termination reaches completion), a condition required for carpel early patterning.At stage 9 and later, KNU and SUP contribute to carpel compartmentalization by controlling the formation of the carpel margin region and meristem. Placenta formation can now proceed. Ca, carpel, FMC, flower meristem centre; St, stamen. T, flower meristem termination triggering/initiation, maintenance, completion.

Furthermore, the allelic sup series shows that the centre-most stamens, under altered (higher) B + C gene expression levels, alone or in combination with other factors, can express two sexual fates at the same time. Pistillody in Arabidopsis SUP mutants consists of a redifferentiation process of stamen parts (i.e. a late SUP control process; Ronse De Craene et al., 2011). Carpeloid stamens not only produce ectopic ovules, but also express a broad range of carpel traits. Laminate stamens, as reported here, mimic stamen morphologies known in basal angiosperms (Hufford, 1996; Endress, 2001). Since carpels are recognized as laminar or originally urceolate structures, stamens and carpels are often considered homologous (homoplastic) structures (Endress, 2006).

With weak to strong levels of pistillody as a recurrent allelic trait, SUP offers an experimental opportunity to test hypotheses on the origin of bisexuality in angiosperms. Among such theories, Mostly Male Theory (MMT) and Out of Male/Out of Female Theory (OOM/OOF) posit the resolution of B and C regulatory functions as central for the evolutionary change from uni- to bisexuality (Frohlich and Parker, 2000; Theissen et al., 2002; Baum and Hileman, 2006; Tavares et al., 2010). For example, male gymnosperm cones share genetic features with perianth organs and stamens, supporting the MM and OOM hypotheses (Chanderbali et al., 2010). Importantly, both hypotheses can be tested experimentally.

Based on SUP data, we propose an evo-devo scenario that is a combination of MMT and OOM/OOF models, reconciled by the fact that the control of the class B/C homeotic gene balance is required during both early and late stages of flower development. There are at least four steps in the process. Steps 1 and 2 fulfil MMT conditions (i.e. ectopic female identity expression, with carpels originating through stamen transformation), while step 3 rather corresponds to OOM/OOF (i.e. a homeosis process).

  1. In the ancestor, the male ‘flower’ is indeterminate (identity controlled by continuing B + C expression) and the female ‘flower’ is indeterminate (due to continuing C expression), corresponding to simple uniaxial strobilus type axes (Endress, 2004; Rudall and Bateman, 2010). In basal angiosperms, sporophylls are elongated axes: laminar stamens with thick connectives and incompletely closed carpels lacking synorganization are predominant (Hufford, 1996; Endress, 2004). Features intermediate between inflorescence and flower highlight the role of heterotopy (i.e. gender expression) and heterochrony (i.e. internode length) in flower evolution (Ruddal and Bateman, 2010).Mutations in SUP exhibit several of the features described above. Genetic interactions of SUP with FM patterning and homeostasis genes probably indicate some of the contributors to those changes (see step 4 below).
  2. Pistillody appears, ectopic ovules being produced on stamens. In alloplasmic wheat, for example, pistillody is caused by alterations to the class B gene expression patterns in appropriate genetic backgrounds (Hama et al., 2004). More generally, the heterotopic expression of female gender on the axis of the opposite sex suggests that the centre-most region of the FMC has acquired new gene expression capabilities, resulting in the novel structures produced there (Mathews and Kramer, 2012).
  3. SUP function sets up a sexual boundary (class B gene expression exclusion). Despite the fact that imperfect bisexual flowers can develop in the absence of SUP (as in our allelic series), variable male/female gradients are transformed by SUP into two separate states, male and female. A fluctuating, labile sexual boundary becomes sharp and precisely positioned.Alterations in SUP activity in Arabidopsis probably reflect vestiges of the weaker developmental homeostasis of early angiosperm flowers. Studied ANITA grade plants exhibit overlapping ‘fading borders’ in the flower (Chanderbali et al., 2010). Canalization of ancestral transcriptional networks (of class B gene function in particular) has been considered of paramount importance for flower diversification (Chanderbali et al., 2010), and also for the evolution of enforced flower homeostasis. SUP is a candidate canalization factor for increased developmental robustness (Mestek Boukhibar and Barkoulas, 2015) of the male–female boundary resulting in spatially discrete and strictly controlled B function expression at the FMC. Studies in ANITA grade plants would be very important to observe expression patterns of SUP family genes.
  4. From precise boundary to FM homeostasis, i.e. the optimization of the reproduction process by further reinforcing developmental robustness. A SUP (-type) control of precise sexual boundary is likely to have facilitated additional refinements of flower development, i.e. ‘co-opting the meristem patterning system’ at the centre of the floral axis into the gynoecium programme(Parcy et al., 1998; Endress, 2006).This has most probably been achieved by integrating regulatory mechanisms of organ identity with FMC and flower whorl patterning and determinacy (thus with both temporal and spatial changes in the activity of the FM). The result is the production of robust male and female modules. The early and late control functions of SUP, as reported here, support a scenario in which the formation of reproductive meristematic tissues and their boundaries, and their developmental fates are co-ordinated from the flower to the placenta to the ovule. We show that developmental robustness in Arabidopsis flowers depends on the combined effects of at least two regulatory gene networks. As a result, double mutant combinations are necessary and sufficient to produce extreme (synergic) morphological aberrations.

It is possible that alterations in SUP activity (or SUP-type functions) might have been involved in the origin of flowers, in particular, in flower bisexuality, and/or SUP might have been important only in the increased developmental homeostasis observed in eudicots and many monocots, as compared with ANITA grade flowers.


Supplementary data are available online at and consist of the following. Supplementary information 1: SUP gene structural features and the sup-5 deletion. Figure S1: SUP methylation pattern in wild-type and mutant epi-A31. Figure S2: morphology of reproductive structures produced at the FMC in sup interallelic crosses. Figure S3: FM termination analysed by flower morphology in clv3-1 sup doubles. The trend is loss of floral meristem determinacy. Figure S4: role of SUP in carpel compartmentalization. Figure S5: alterations in male–female boundary and intra- and inter-whorl organ separation in sup mutants.

Supplementary Data:


We thank Elliot Meyerowitz for having hosted part of the project, and we are much indebted to Frank Welmer, Hao Yu, Nathanael Prunet, and Marcio Alves-Ferreira. For technical help we acknowledge the work of Frédérique Rosier, Anne-Marie Thierry, and Isabelle Courseau for her internship contribution. Last, but not least, the authors are grateful to the anonymous reviewers for their careful and consistent contribution to the final version of the manuscript. This work was supported by the Gordon and Betty Moore Foundation and Institut universitaire de France to I.N.


  • Alvarez J, Smyth DR. 1999. CRABS CLAW and SPATULA, two Arabidopsis genes that control carpel development in parallel with AGAMOUS. Development 126: 2377–86. [PubMed] [Google Scholar]
  • Alvarez J, Smyth DR. 2002. CRABS CLAW and SPATULA genes regulate growth and pattern formation during gynoecium development in Arabidopsis thaliana. International Journal of Plant Sciences 163: 17–41. [Google Scholar]
  • Baum DA, Hileman LC. 2006. A developmental genetic model for the origin of the flower In: Ainsworth C, ed. Flowering and its manipulation. Oxford: Blackwell Publishing Ltd. [Google Scholar]
  • Bowman JL. 1994. Arabidopsis: an atlas of morphology and development. New York: Springer-Verlag. [Google Scholar]
  • Bowman JL, Smyth DR, Meyerowitz EM. 1989. Genes directing flower development in Arabidopsis. The Plant Cell 1: 37–52. [PMC free article] [PubMed] [Google Scholar]
  • Bowman JL, Smyth DR, Meyerowitz EM. 1991. Genetic interactions among floral homeotic genes of Arabidopsis. Development 112: 1–20. [PubMed] [Google Scholar]
  • Bowman JL, Sakai H, Jack T, Weigel D, Mayer U, Meyerowitz EM. 1992. SUPERMAN, a regulator of floral homeotic genes in Arabidopsis. Development 114: 599–615. [PubMed] [Google Scholar]
  • Braun DM, Stone JM, Walker JC. 1997. Interaction of the maize and Arabidopsis kinase interaction domains with a subset of receptor-like protein kinases: implications for transmembrane signaling in plants. The Plant Journal, 12: 83–95. [PubMed] [Google Scholar]
  • Breuil-Broyer S, Morel P, de Almeida-Engler J, Coustham V, Negrutiu I, Trehin C. 2004. High-resolution boundary analysis during Arabidopsis thaliana flower development. The Plant Journal 38: 182–192. [PubMed] [Google Scholar]
  • Chanderbali AS, Yoo MJ, Zahn LM, et al. 2010. Conservation and canalization of gene expression during angiosperm diversification accompany the origin and evolution of the flower. Proceedings of the National Academy of Sciences, USA 107: 22570–22575. [PMC free article] [PubMed] [Google Scholar]
  • Clark SE, Running MP, Meyerowitz EM. 1993. CLAVATA1, a regulator of meristem and flower development in Arabidopsis. Development 119: 397–418. [PubMed] [Google Scholar]
  • Das P, Ito T, Wellmer F, et al. 2009. Floral stem cell termination involves the direct regulation of AGAMOUS by PERIANTHIA. Development 136: 1605–1611. [PubMed] [Google Scholar]
  • Dathan N, Zaccaro L, Esposito S, et al. 2002. The Arabidopsis SUPERMAN protein is able to specifically bind DNA through its single Cys(2)-His(2) zinc finger motif. Nucleic Acids Research 30: 4945–4951. [PMC free article] [PubMed] [Google Scholar]
  • Davies B, Cartolano M, Schwarz-Sommer Z. 2006. Flower development: the Antirrhinum perspective. Advances in Botanical Research 44: 279–321. [Google Scholar]
  • Davies B, Motte P, Keck E, Saedler H, Sommer H, Schwarz-Sommer Z. 1999. PLENA and FARINELLI: redundancy and regulatory interactions between two Antirrhinum MADS-box factors controlling flower development. EMBO Journal 18: 4023–4034. [PMC free article] [PubMed] [Google Scholar]
  • de Almeida-Engler J, De Veylder L, De Groodt L, et al. 2009. Systematic analysis of cell-cycle gene expression during Arabidopsis development. The Plant Journal 59: 645–660. [PubMed] [Google Scholar]
  • Dievart A, Dalal M, Tax FE, et al. 2003. CLAVATA1 dominant-negative alleles reveal functional overlap between multiple receptor kinases that regulate meristem and organ development. The Plant Cell 15: 1198–1211. [PMC free article] [PubMed] [Google Scholar]
  • Endress PK. 2001. The flowers in extant basal angiosperms and inferences on ancestral flowers. International Journal of Plant Sciences 162: 1111–1140. [Google Scholar]
  • Endress PK. 2004. Structure and relationships of basal relictual angiosperms. Australian Systematic Botany 17: 343–366. [Google Scholar]
  • Endress PK. 2006. Angiosperm floral evolution: morphological developmental framework. Advances in Botanical Researc 44: 1–61. [Google Scholar]
  • Ferrandiz C, Fourquin C, Prunet N, et al. 2010. Carpel development. Advances in Botanical Research 55: 1–73. [Google Scholar]
  • Frohlich MW, Parker DS. 2000. The mostly male theory of flower evolutionary origins: from genes to fossils. Systematic Botany 25: 155–170. [Google Scholar]
  • Gaiser JC, Robinson-Beers K, Gasser CS. 1995. The Arabidopsis SUPERMAN gene mediates asymmetric growth of the outer integument of ovules. The Plant Cell 7: 333–345. [PMC free article] [PubMed] [Google Scholar]
  • Hama E, Takumi S, Ogihara Y, Murai K. 2004. Pistillody is caused by alterations to the class-B MADS-box gene expression pattern in alloplasmic wheats. Planta 218: 712–720. [PubMed] [Google Scholar]
  • Huang H, Ma H. 1997. FON1, an Arabidopsis gene that terminates floral meristem activity and controls flower organ number. The Plant Cell 9: 115–134. [PMC free article] [PubMed] [Google Scholar]
  • Hufford L. 1996. The origin and early evolution of angiosperm stamens In: D’Arcy WG, Keating RC, eds. The anther: form, function, and phylogeny. Cambridge: Cambridge University Press, 58–91. [Google Scholar]
  • Ishiguro S, Watanabe Y, Ito N, et al. 2002. SHEPHERD is the Arabidopsis GRP94 responsible for the formation of functional CLAVATA proteins. EMBO Journal 21: 898–908. [PMC free article] [PubMed] [Google Scholar]
  • Ito T, Sakai H, Meyerowitz EM. 2003. Whorl-specific expression of the SUPERMAN gene of Arabidopsis is mediated by cis elements in the transcribed region. Current Biology 13: 1524–1530. [PubMed] [Google Scholar]
  • Jack T, Brockman LL, Meyerowitz EM. 1992. The homeotic gene APETALA3 of Arabidopsis thaliana encodes a MADS box and is expressed in petals and stamens. Cell 68: 683–697. [PubMed] [Google Scholar]
  • Jack T, Fox GL, Meyerowitz EM. 1994. Arabidopsis homeotic gene APETALA3 ectopic expression – transcriptional and posttranscriptional regulation determine floral organ identity. Cell 76: 703–716. [PubMed] [Google Scholar]
  • Jack T, Sieburth L, Meyerowitz E. 1997. Targeted misexpression of AGAMOUS in whorl 2 of Arabidopsis flowers. The Plant Journal 11: 825–839. [PubMed] [Google Scholar]
  • Jacobsen SE, Meyerowitz EM. 1997. Hypermethylated SUPERMAN epigenetic alleles in Arabidopsis. Science 277: 1100–1103. [PubMed] [Google Scholar]
  • Jefferson RA, Wilson KJ. 1991. The GUS gene fusion system In: Gelvin SB, Schilperoort RA, eds. Plant molecular biology manual. Vol. B14. Dordrecht: Kluwer Academic Publishers, 1–33. [Google Scholar]
  • Kamiuchi Y, Yamamoto K, Furutani M, Tasaka M, Aida M. 2014. The CUC1 and CUC2 genes promote carpel margin meristem formation during Arabidopsis gynoecium development. Frontiers in Plant Science 5: 1–9. [PMC free article] [PubMed] [Google Scholar]
  • Krizek BA, Meyerowitz EM. 1996. The Arabidopsis homeotic genes APETALA3 and PISTILLATA are sufficient to provide the B class organ identity function. Development 122: 11–22. [PubMed] [Google Scholar]
  • Lenhard M, Bohnert A, Jurgens G, Laux T. 2001. Termination of stem cell maintenance in Arabidopsis floral meristems by interactions between WUSCHEL and AGAMOUS. Cell 105: 805–814. [PubMed] [Google Scholar]
  • Liu X, Kim YJ, Müller R, et al. 2011. AGAMOUS terminates floral stem cell maintenance in Arabidopsis by directly repressing WUSCHEL through recruitment of Polycomb Group proteins. The Plant Cell 23: 3654–3670. [PMC free article] [PubMed] [Google Scholar]
  • Liu Z, Franks RG. 2015. Molecular basis of fruit development. Frontiers in Plant Science 6: 28. doi:10.3389/fpls.2015.00028. [PMC free article] [PubMed] [Google Scholar]
  • Liu Z, Mara C. 2010. Regulatory mechanisms for floral homeotic gene expression. Seminars in Cell and Developmental Biology 21: 80–86. [PubMed] [Google Scholar]
  • Lohmann JU, Hong RL, Hobe M, et al. 2001. A molecular link between stem cell regulation and floral patterning in Arabidopsis. Cell 105: 793–803. [PubMed] [Google Scholar]
  • Mathews S, Kramer EM. 2012. The evolution of reproductive structures in seed plants: a re-examination based on insights from developmental genetics. New Phytologist 194: 910–923. [PubMed] [Google Scholar]
  • Mestek Boukhibar LM, Barkoulas M. 2016. The developmental genetics of biological robustness. Annals of Botany 117: 699–707. [PMC free article] [PubMed] [Google Scholar]
  • Nahar MA-U, Ishida T, Smyth DR, Tasaka M, Aida M. 2012. Interactions of CUP-SHAPED COTYLEDON and SPATULA genes control carpel margin development in Arabidopsis thaliana. Plant and Cell Physiology 53: 1134–1143. [PMC free article] [PubMed] [Google Scholar]
  • Nikovics K, Blein T, Peaucelle A, et al. 2006. The balance between the MIR164A and CUC2 genes controls leaf margin serration in Arabidopsis. The Plant Cell 18: 2929–2945. [PMC free article] [PubMed] [Google Scholar]
  • Opik H, Rolfe SA. 2005. The physiology of flowering plants. Physicochemical and environmental plant physiology. Cambridge: Cambridge University Press. [Google Scholar]
  • Parcy F, Nilsson O, Busch MA, Lee I, Weigel D. 1998. A genetic framework for floral patterning. Nature 395: 561–566. [PubMed] [Google Scholar]
  • Payne T, Johnson SD, Koltunow AM. 2004. KNUCKLES (KNU) encodes a C2H2 zinc-finger protein that regulates development of basal pattern elements of the Arabidopsis gynoecium. Development 131: 3737–3749. [PubMed] [Google Scholar]
  • Prunet N, Morel P, Thierry A-M, et al. 2008. REBELOTE, SQUINT, and ULTRAPETALA1 function redundantly in the temporal regulation of floral meristem termination in Arabidopsis thaliana. The Plant Cell 20: 901–919. [PMC free article] [PubMed] [Google Scholar]
  • Prunet N, Morel P, Negrutiu I, Trehin C. 2009. Time to stop: flower meristem termination. Plant Physiology 150: 1764–1772. [PMC free article] [PubMed] [Google Scholar]
  • Rast MI, Simon R. 2008. The meristem-to-organ boundary: more than an extremity of anything. Current Opinion in Genetics and Development 18: 287–294. [PubMed] [Google Scholar]
  • Ronse De Craene L, Trehin C, Morel P, Negrutiu I. 2011. Carpeloidy in flower evolution and diversification: a comparative study in Carica papaya and Arabidopsis thaliana. Annals of Botany 107: 1453–1463. [PMC free article] [PubMed] [Google Scholar]
  • Rudall PJ, Bateman RM. 2010. Defining the limits of flowers: the challenge of distinguishing between the evolutionary products of simple versus compound strobili. Philosophical Transactions of the Royal Society B: Biological Sciences 365: 397–409. [PMC free article] [PubMed] [Google Scholar]
  • Running MP, Meyerowitz EM. 1996. Mutations in the PERIANTHIA gene of Arabidopsis specifically alter floral organ number and initiation pattern. Development 122: 1261–1269. [PubMed] [Google Scholar]
  • Sablowski R. 2007a. Flowering and determinacy in Arabidopsis. Journal of Experimental Botany 58: 899–907. [PubMed] [Google Scholar]
  • Sablowski R. 2007b. The dynamic plant stem cell niches. Current Opinion in Plant Biology 10: 639–644. [PubMed] [Google Scholar]
  • Sakai H, Krizek BA, Jacobsen SE, Meyerowitz EN. 2000. Regulation of SUP expression identifies multiple regulators involved in arabidopsis floral meristem development. The Plant Cell 12: 1607–1618. [PMC free article] [PubMed] [Google Scholar]
  • Sakai H, Medrano LJ, Meyerowitz EM. 1995. Role of SUPERMAN in maintaining Arabidopsis floral whorl boundaries. Nature 378: 199–203. [PubMed] [Google Scholar]
  • Schultz EA, Pickett FB, Haughn GW. 1991. The FLO10 gene-product regulates the expression domain of homeotic genes AP and PL in Arabidopsis flowers. The Plant Cell 3: 1221–1237. [PMC free article] [PubMed] [Google Scholar]
  • Smaczniak C, Immink RGH, Muiño JM, et al. 2012. Characterization of MADS-domain transcription factor complexes in Arabidopsis flower development. Proceedings of the National Academy of Sciences, USA 109: 1560–1565. [PMC free article] [PubMed] [Google Scholar]
  • Sun B, Ito T. 2010. Floral stem cells: from dynamic balance towards termination. Biochemical Society Transactions 38: 613–616. [PubMed] [Google Scholar]
  • Sun B, Ito T. 2015. Regulation of floral stem cell termination in Arabidopsis. Frontiers in Plant Science 6 :17. doi:10.3389/fpls.2015.00017. [PMC free article] [PubMed] [Google Scholar]
  • Sun B, Xu Y, Ng K-H, Ito T. 2009. A timing mechanism for stem cell maintenance and differentiation in the Arabidopsis floral meristem. Genes and Development 23: 1791–1804. [PMC free article] [PubMed] [Google Scholar]
  • Takeda S, Matsumoto N, Okada K. 2004. RABBIT EARS, encoding a SUPERMAN-like zinc finger protein, regulates petal development in Arabidopsis thaliana. Development 131: 425–434. [PubMed] [Google Scholar]
  • Tavares R, Cagnon M, Negrutiu I, Mouchiroud D. 2010. Testing the recent theories for the origin of the hermaphrodite flower by comparison of the transcriptomes of gymnosperms and angiosperms. BMC Evolutionary Biology 10: 240–247. [PMC free article] [PubMed] [Google Scholar]
  • Theissen G, Becker A, Kirchner C, et al. 2002. How land plants learned their floral ABCs: the role of MADS-box genes in the evolutionary origin of flowers In: Cronk QCB, Bateman RM, Hawkins JA, eds. Developmental genetics and plant evolution. London: Taylor & Francis, 173–205. [Google Scholar]
  • Tichtinsky G, Vanoosthuyse V, Cock JM, Gaude T. 2003. Making inroads into plant receptor kinase signalling pathways. Trends in Plant Science 8: 231–237. [PubMed] [Google Scholar]
  • Trotochaud AE, Hao T, Wu G, Yang Z, Clark SE. 1999. The CLAVATA1 receptor-like kinase requires CLAVATA3 for its assembly into a signaling complex that includes KAPP and a Rho-related protein. The Plant Cell 11: 393–406. [PMC free article] [PubMed] [Google Scholar]
  • Williams RW, Wilson JM, Meyerowitz EM. 1997. A possible role for kinase-associated protein phosphatase in the Arabidopsis CLAVATA1 signaling pathway. Proceedings of the National Academy of Sciences, USA 94: 10467–10472. [PMC free article] [PubMed] [Google Scholar]
  • Wynn AN, Seaman AA, Jones AL, Franks RG. 2014. Novel functional roles for PERIANTHIA and SEUSS during floral organ identity specification, floral meristem termination, and gynoecial development. Frontiers in Plant Science 5: 130. doi:10.3389/fpls.2014.00130. [PMC free article] [PubMed] [Google Scholar]
  • Yang YZ, Fanning L, Jack T. 2003. The K domain mediates heterodimerization of the Arabidopsis floral organ identity proteins, APETALA3 and PISTILLATA. The Plant Journal 33: 47–59. [PubMed] [Google Scholar]
  • Yanofsky MF, Ma H, Bowman JL, Drews GN, Feldmann KA, Meyerowitz EM. 1990. The protein encoded by the Arabidopsis homeotic gene AGAMOUS resembles transcription factors. Nature 346: 35–39. [PubMed] [Google Scholar]
  • Yun JY, Weigel D, Lee I. 2002. Ectopic expression of SUPERMAN suppresses development of petals and stamens. Plant and Cell Physiology 43: 52–57. [PubMed] [Google Scholar]

Articles from Annals of Botany are provided here courtesy of Oxford University Press