SQUAMOSA-PROMOTER BINDING PROTEIN 1 initiates flowering in Antirrhinum majus through the activation of meristem identity genes


For correspondence (fax 785 864 5860; e-mail jcpxt8@ku.edu).


The degree to which developmental genetic pathways are conserved across distantly related organisms is a major question in biology. In Arabidopsis thaliana (L.) Heynh., inflorescence development is initiated in response to a combination of external and internal floral inductive signals that are perceived across the whole plant, but are integrated within the shoot apical meristem. Recently, it was demonstrated that SQUAMOSA-PROMOTER BINDING PROTEIN (SBP)-box proteins regulate A. thaliana flowering time by mediating signals from the autonomous and photoperiod pathways, and by directly activating key genes involved in inflorescence and floral meristem identity, including FRUITFULL (FUL), APETALA1 (AP1) and LEAFY (LFY). In the distantly related core eudicot species Antirrhinum majus L., paralogous SBP-box proteins SBP1 and SBP2 have likewise been implicated in regulating the AP1 ortholog SQUAMOSA (SQUA). To test the hypothesis that SBP-box genes are also involved in the floral induction of A. majus, we used a reverse genetic approach to silence SBP1. SBP1-silenced lines are late to nonflowering, and show reduced apical dominance. Furthermore, expression and sequence analyses suggest that the SBP1-mediated transition to flowering occurs through the positive regulation of FUL/LFY homologs. Together, these data outline the utility of virus-induced gene silencing in A. majus, and provide new insight into the conservation of flowering time genetic pathways across core eudicots.


The transition from vegetative to reproductive growth is a critical step in plant development, and is marked by the acquisition of inflorescence identity by the shoot apical meristem (SAM), resulting in the production of flowers and seeds (Baürle and Dean, 2006). In Arabidopsis thaliana (L.) Heynh. (Brassicaceae; rosid), flowering is induced in response to four converging pathways – autonomous, gibberellic acid, photoperiod and vernalization – the signals of which are integrated in the SAM by the floral integrator genes SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1), FLOWERING LOCUS T (FT) and FD (Kardailsky et al., 1999; Samach et al., 2000; Abe et al., 2005; Turck et al., 2008; Zeevaart, 2008; Yant et al., 2009). The floral integrators trigger flowering by positively regulating genes involved in inflorescence and floral meristem identity, including FRUITFULL (FUL), APETALA1 (AP1) and LEAFY (LFY) (Schultz and Haughn, 1991; Mandel et al., 1992; Mizukami and Ma, 1997; Ferrándiz et al., 2000). Interestingly, many Arabidopsis thaliana flower developmental gene homologs are known to function similarly in distantly related species, such as Antirrhinum majus L. (Plantaginaceae; asterid) (Davies et al., 2006; Benlloch et al., 2007). Despite this, little is known about the upstream regulators of these genes in core eudicot species outside of Arabidopsis thaliana.

In Antirrhinum majus, the closely related genes SQUAMOSA-PROMOTER BINDING PROTEIN 1 (SBP1) and SBP2 are strong candidates for direct regulation of the AP1 ortholog SQUAMOSA (SQUA), suggesting a role for these genes in inflorescence development prior to the production of flowers (Klein et al., 1996). SBP1 and SBP2 were first isolated through their direct in vitro interaction with a specific motif in the SQUA promoter; similar interactions have subsequently been identified between SBP-like proteins of other species, including Arabidopsis thaliana and Betula pendula Roth (Lännenpääet al., 2004; Birkenbihl et al., 2005; Wang et al., 2009). Both SBP1 and SBP2 are expressed earlier in the vegetative plant than SQUA, and expression levels increase during inflorescence development (Klein et al., 1996). Unlike SQUA, northern analyses have revealed that although they are highly expressed in petals, SBP1 and SBP2 are expressed only weakly in sepals (Klein et al., 1996). Both genes are also expressed in inflorescence and floral meristems, whereas SQUA expression is restricted to floral meristems. Based on their spatiotemporal patterns of expression and binding affinities, it is hypothesized that SBP1 and SBP2 are functionally redundant at late stages of inflorescence development, and that loss of function will cause a late flowering phenotype (Klein et al., 1996).

The SBP genes are members of a lineage of plant-specific putative transcription factors, characterized by a highly conserved SBP-box of 76 amino acids involved in DNA binding and nuclear localization (Klein et al., 1996; Birkenbihl et al., 2005; Cardon et al., 1997, 1999). Arabidopsis thaliana has 16 SBP-box genes (known as SPL genes) that can be separated into several distinct clades based on length, exon–intron structure and sequence similarity. The shorter Arabidopsis thaliana SPL genes, SPL3, SPL4 and SPL5, are expressed in inflorescence tissues, and most of them appear to be negatively regulated by miR156 and miR157 microRNAs (Rhoades et al., 2002; Boutet et al., 2003; Kasschau et al., 2003; Chen et al., 2004; Schwab et al., 2005; Wang et al., 2009; Wu and Poethig, 2006). Together, SPL3, SPL4 and SPL5 are the closest homologs of the Antirrhinum majus genes SBP1 and SBP2 (Cardon et al., 1999).

Functional analyses suggest that SPL3, SPL4 and SPL5 promote flowering in Arabidopsis thaliana through both the photoperiod and autonomous flowering genetic pathways, the former of which is independent of negative miR156 regulation (Cardon et al., 1999; Unte et al., 2003; Wu and Poethig, 2006; Gandikota et al., 2007; Fornara and Coupland, 2009; Wang et al., 2009; Yamaguchi et al., 2009). Overexpression of SPL3 causes early flowering under short-day conditions, whereas overexpression of its negative regulator miR156 results in late flowering (Cardon et al., 1997; Wu and Poethig, 2006; Gandikota et al., 2007). In wild-type Arabidopsis thaliana, SPL3 is induced by long days and/or increasing developmental age prior to the onset of flowering (Yamaguchi et al., 2009). In leaves, increased levels of the SPL3 protein promotes flowering through the indirect activation of FT (Wang et al., 2009). By contrast, expression of SPL3, SPL4 and SPL5 in the SAM results in the direct activation of the inflorescence meristem gene FUL, a close relative of AP1, and the floral meristem identity genes AP1 and LFY (Wang et al., 2009; Yamaguchi et al., 2009). These studies highlight the importance of SBP-box genes as facilitators of Arabidopsis thaliana flowering (Fornara and Coupland, 2009). However, it is unclear whether this function is unique to Arabidopsis thaliana.

Here, we present functional and expression data supporting functional conservation of the SPL3–SPL5/SBP1 genes between Arabidopsis thaliana and Antirrhinum majus. We demonstrate that Antirrhinum majus SBP1 is expressed earlier than, and likely regulates, homologs of the Arabidopsis thaliana inflorescence and floral meristem identity genes, AmFUL, DEFH28 and SQUA (collectively known as FUL-like genes) (Litt and Irish, 2003), and the LFY ortholog FLORICAULA (FLO). We also present evidence that the FUL-like genes are transcribed independently of SBP1 and SBP2 in certain floral tissues, and that, unlike its homolog in Arabidopsis thaliana (Mandel and Yanofsky, 1995), AmFUL is not negatively regulated by SQUA in floral meristems.


SBP1 is more closely related to SPL3, SPL4 and SPL5 than to SBP2

Maximum parsimony, maximum likelihood and Bayesian analyses of aligned SBP-box genes placed SBP1 in a well-supported clade with other asterid SBP1-like genes (Figure 1a). Relationships among these genes were strongly supported by parsimony bootstrap and Bayesian posterior probability values, and followed familial relationships for this group. Rosid SBP-box genes from Arabidopsis, Gossypium and Betula formed a grade sister to the asterid SBP1-like genes (Figure 1a). Although there was little support for relationships among the SBP1-like genes from different rosid species, a rosid plus asterid SBP1-like clade, excluding SBP2, was strongly supported. These data suggest that the gene duplication giving rise to SBP1 and SBP2 occurred before the diversification of the core eudicots, much earlier than the duplication events that produced the closely related Arabidopsis paralogs SPL3, SPL4 and SPL5. They also demonstrate that SBP1 is the closest homolog of SPL3–SPL5.

Figure 1.

 Evolutionary relationships and structure of SQUAMOSA-PROMOTER BINDING PROTEIN (SBP)-box genes.
(a) Maximum likelihood tree showing the relationship of SBP1 and SBP2 (in bold) relative to SBP-box genes from other asterid and rosid species. Parsimony bootstrap values of 70 or greater are shown above the branches, Bayesian posterior probabilities of 90 or greater are shown below the branches.
(b) Simplified structure of SBP1 showing the region subcloned into the pTRV2 vector and the region amplified to assess the level of silencing. The gray triangle denotes the intron.

SBP and FLO genes are expressed earlier in development than FUL genes

The RT-PCR analyses of gene expression from leaf tissues at different stages of development revealed a similar pattern of upregulation for SBP1 and SBP2. The expression of both genes increased in leaves around the seven-leaf-node stage; low levels of expression were evident for SBP1 prior to this stage (Figure 2a). Transcripts of both genes were detectable in inflorescence tissues following the transition to flowering, at around 17 leaf nodes, and in flowers, both genes were expressed in petals and gynoecia, with low expression in sepals and stamens (Figure 2a).

Figure 2.

 Expression of putative inflorescence and floral meristem identity genes, and their upstream regulators, in wild-type Antirrhinum majus tissues.
(a) SBP1, SBP2 and FLO are upregulated at similar stages during vegetative development (measured by leaf node number) prior to the upregulation of FUL-like genes. All genes are transcribed at high levels in the flowering inflorescence (inf). SBP1, SBP2 and DEFH28 have similar floral expression patterns in the petals (pe) and gynoecium (gy). SQUA and AmFUL are also expressed in sepals (se), but expression is low in stamens (st). –rt, inflorescence RNA minus reverse transcriptase control.
(b–c) In situ hybridization of an antisense AmFUL probe showing expression (blue staining) in the inflorescence meristem (im) and floral meristems (fm) of Antirrhinum majus.
(d) Sense control AmFUL probe showing minimal staining in Antirrhinum majus inflorescences.

Patterns of expression were similar between AmFUL, DEFH28 and SQUA, but not FLO, during vegetative development. Whereas FLO was upregulated in leaves at a similar time to SBP1 and SBP2, little to no gene expression in leaves was detectable for AmFUL, DEFH28 and SQUA (Figure 2a). By contrast, all four putative meristem identity genes were expressed in inflorescence tissues following the transition to flowering. In mid-stage flowers, FLO mRNA was undetectable. AmFUL and SQUA were similarly expressed in sepals, petals and gynoecia, but not in stamens, whereas DEFH28 mRNA was only detectable in petals and gynoecia, similar to SBP1 and SBP2 (Figure 2a).

More detailed in situ hybridization analyses of AmFUL revealed expression in both the apical inflorescence meristem and lateral floral meristems (Figure 2b,c); little to no staining was detected using a sense control (Figure 2d). This is similar to patterns of DEFH28, SBP1 and SBP2 expression (Klein et al., 1996; Müller et al., 2001). Expression of both AmFUL and DEFH28 is different from SQUA, which is only expressed in floral meristems (Huijser et al., 1992). Furthermore, these data suggest that, unlike Arabidopsis thaliana FUL, AmFUL is not excluded from floral meristems by SQUA.

Virus-induced gene silencing (VIGS) is an effective reverse genetic approach in Antirrhinum majus

As SBP1 is orthologous to the previously characterized SPL3, SPL4 and SPL5 genes of Arabidopsis thaliana (Figure 1a), we selected this gene for a comparative functional study using pTRV2-mediated VIGS (Dinesh-Kumar et al., 2003). Of the approximately 700 seedlings vacuum infiltrated with the pTRV2-SBP1 or pTRV2-Empty vectors, between 46.8 and 57.0% survived to the screening stage (Table S1). Screening for both pTRV2 and pTRV1 vectors revealed an initial infection rate as high as 16% at the five- to six-leaf-node stage. However, this rate dropped to 1.4–2.6% by the 20-leaf-node stage (the stage at which all uninfected or infected with pTRV2-Empty plants were flowering), resulting in a total of nine pTRV2-SBP1 and five pTRV2-Empty positive plants available for phenotypic comparison with treated but uninfected plants at the 15-leaf-node stage and beyond (Table S1).

In order to verify that our VIGS treatments resulted in gene-specific silencing, we carried out ACTIN-standardized RT-PCR analysis on cDNA from the youngest leaves of each pTRV2/pTRV1 positive plant and compared them with uninfected plants at the same developmental stage (assessed by leaf node number). In each case, the presence of pTRV2-SBP1, but not pTRV2-Empty, was significantly correlated with a reduction in the SBP1 transcript, according to a one-tailed Wilcoxon paired sample test (Figure 3). Furthermore, no significant reduction in SBP2 gene expression levels was detected in plants that were positive for pTRV2-SBP1 or in plants containing the pTRV2-Empty vector control (Figure 3). This indicates that silencing was gene specific and was not related to infection per se.

Figure 3.

 Verification of gene silencing in virus-induced gene silencing (VIGS)-treated plants at different developmental stages.
(a) Example RT-PCR gels for uninfected wild type (WT), pTRV2-Empty positive (E) and pTRV2-SBP1 positive plants grouped according to developmental stage, as measured by leaf node number. pTRV2-SBP1 positive plants showed silencing of SBP1, but not SBP2, indicating that silencing was gene specific. ACTIN was used as an internal control.
(b) Boxplots showing summarized SBP1 (left) and SBP2 (right) expression data for multiple pTRV2-SBP1 and pTRV2-Empty infected plants, based on RT-PCR standardized against ACTIN. Gene expression is plotted relative to expression from uninfected plants (dotted line) at the same developmental stage. Asterisks indicate a significant deviation (at the 0.025 level) from uninfected plants according to a one-tailed Wilcoxon paired sample test (Zar, 1984). Solid horizontal line, median; box, interquartile range; whiskers, extending to the furthest point that is within 1.5 times the interquartile range from the box; NS, not significant at the 0.05 level.

SBP function in flowering time is broadly conserved

Of the nine individual plants that were consistently positive for pTRV2-SBP1 throughout vegetative development, four individuals failed to flower, even after 6 months in the growth chamber (with more than 60 leaf nodes), and two individuals died before flowering (with 23 and 35 leaf nodes). All of these individuals developed significantly more leaf nodes than VIGS-treated uninfected or pTRV2-Empty controls at flowering, the latter of which produced approximately 17 leaf nodes prior to flowering (Figure 4). The remaining three pTRV2-SBP1 positive plants all flowered eventually at the 26-, 28- and 35-leaf-node stage. However, all flowered significantly later and after developing more leaf nodes than uninfected or pTRV2-Empty plants (Figure 4). By contrast, none of the pTRV2-Empty plants flowered later or with significantly more leaf nodes than VIGS-treated uninfected plants (Figure 4a,e).

Figure 4.

 Silencing of SBP1 causes late- to non-flowering phenotypes.
(a) Plants positive for pTRV2-SBP1 (right) produce more leaves and lateral branches than pTRV2-Empty controls (left) or uninfected plants. The pTRV2-SBP1 plant on the right failed to flower, even after 6 months in the growth chamber.
(b) Example primary branch of an uninfected plant showing 13 decussately arranged leaves, five spirally arranged leaves lacking flowers, two flowers in the axils of spiral leaves (arrows) and flowers subtended by bracts.
(c) Lateral branch of a plant positive for pTRV2-SBP1 showing a typical number of decussately arranged leaves, but a significantly increased number of spirally arranged leaves, relative to control plants.
(d) Diagram of wild-type (left) versus non-flowering pTRV2-SBP1 positive plants (right). pTRV2-SBP1 positive plants develop more lateral branches and spirally arranged leaves per branch than wild-type plants. Filled circles denote floral buds.
(e) Cumulative flowering of uninfected, pTRV2-Empty and pTRV2-SBP1 plants over developmental time, as measured by leaf node number. All uninfected wild-type (black circles) and pTRV2-Empty plants (open diamonds) flowered prior to the production of 20 leaf nodes. In contrast, none of the pTRV2-SBP1 (gray triangle) plants flowered prior to the 20-leaf node stage, with almost half of the plants remaining in the pre-flowering stage at leaf node 65.

The late flowering of three pTRV2-SBP1 treated plants was correlated with infection in the youngest leaf pairs. However, at the time of flowering, none of the floral tissues were still positive for pTRV1 or pTRV2-SBP1, and exhibited normal levels of SBP1 in the same tissues (data not shown). This suggests that flowering only occurred when the SAM was no longer infected with pTRV2-SBP1 in these three plants. Together with the non-flowering pTRV2-SBP1 infected individuals, these data strongly support the hypothesis that expression of SBP1 at wild-type levels is critical for the transition to flowering in Antirrhinum majus.

Although pTRV2-SBP1 infected plants had significantly more leaf nodes prior to flower development than control plants, for each vegetative branch the difference was only observed for spirally arranged inflorescence leaf nodes (Figure 4b–d). The number of decussately arranged vegetative leaf nodes was ∼12 for both pTRV2-SBP1 infected and control plants, demonstrating that SBP1 functions after the switch to inflorescence development, but before the development of floral meristems, at least under long-day growth conditions. Aside from an increased leaf node number per branch, pTRV2-SBP1 infected plants also had increased vegetative branches relative to control plants (Figure 4a,d). Whereas uninfected and pTRV2-Empty plants very rarely developed lateral vegetative branches prior to flower development on the main vegetative shoot, lateral vegetative branches of pTRV2-SBP1 elongated when their main shoot had produced just slightly more leaf nodes than flowering wild-type individuals. Thus, there appeared to be a loss of apical dominance associated with the downregulation of SBP1.

SBP1 positively regulates inflorescence and floral meristem identity genes

In Arabidopsis thaliana, SPL3–SPL5 proteins regulate flowering time through the direct upregulation of FUL, AP1 and LFY in the SAM and leaves (Wang et al., 2009; Yamaguchi et al., 2009). To determine if SQUA, AmFUL, DEFH28 and FLO are similarly positively correlated with SBP1 expression, we compared expression levels of these genes in leaf tissues of pTRV2-SBP1 infected (with 23 or more leaf nodes) and control plants (with 18 leaf nodes). As pTRV2-SBP1 infected plants were always developmentally older than pTRV2-Empty or uninfected plants, this represents a conservative test of the predicted correlation. In leaves, SBP1, SQUA, AmFUL, DEFH28 and FLO were expressed at a much lower level when harvested from pTRV2-SBP1 infected versus 18-leaf-node stage uninfected plants (Figure 5a). These results were consistent for two pTRV2-SBP1 individuals examined, despite similar levels of SBP2 (Figure 5a).

Figure 5.

 Evidence for positive regulation of inflorescence and floral meristem identity genes by SBP1.
(a) Gene expression in the youngest leaves (nodes 23–27) of non-flowering pTRV2-SBP1 positive plants relative to the youngest leaves of pre-flowering (node 18) control plants (dotted line). Expression is normalized against UBIQUITIN 5; the means and standard deviations for two biological replicates and four technical qRT-PCR replicates are shown.
(b) Putative SBP-box protein binding motifs (outer box) in the promoters or intron (CAL only) of FUL-like genes from Antirrhinum majus, Arabidopsis thaliana and Mimulus guttatus. The AtFUL promoter has two putative SBP-box binding motifs, designated #1 and #2. The core binding motif (inner box) is in bold.

The direct regulation of FUL-like genes by SBP proteins is likely to be conserved

A conserved motif (GTCCGTACAA), within 500 bp upstream of the SQUA, FUL and AP1 start codons, was previously identified and reported to be the direct binding site of SBP1/2 and SPL3, respectively (Klein et al., 1996; Cardon et al., 1997, 1999; Yamaguchi et al., 2009). This conserved motif contains the core sequence CCGTAC, known to be a consensus-binding site for SBP-box proteins (Cardon et al., 1997; Liang et al., 2008) (Figure 5b). To determine if similar binding sites exist in other FUL-like gene homologs, we surveyed 1200 bp of the 5′cis regulatory sequence and the first intron of Arabidopsis thaliana CAULIFLOWER (CAL) (a Brassicaceae-specific paralog of AP1), and newly identified Mimulus guttatusFUL-like genes, designated MgFUL, MgSQUA-like and MgDEFH28-like. Phylogenetic analyses suggest orthology between MgFUL, AtFUL and AmFUL, MgDEFH28-like and AmDEFH28, and MgSQUA-like, AP1/CAL and SQUA (data not shown).

Of the gene sequences investigated, all but MgDEFH28-like contained the core SBP-box protein binding domain CCGTAC (Figure 5b). With the exception of CAL, this motif was found within 700 bp upstream of the start codon; the CAL motif was found in the first intron and was absent from the upstream regulatory sequences. Although not a direct assessment, these results suggest that SBP-box proteins are likely to directly regulate all FUL genes in both rosids and asterids.


Role of SBP1 in the transition to flowering

Direct assays of gene function in Antirrhinum majus have been hampered by the lack of efficient transformation procedures (but see Cui et al., 2003; Shang et al., 2007). To circumvent this problem we optimized a reverse genetics approach in Antirrhinum majus using VIGS. We identified nine individuals infected with the pTRV2-SBP1 construct, all of which showed significant, but incomplete, silencing of SBP1. Incomplete silencing is consistent with reports from other species, including Papaver somniferum, Aquilegia vulgaris and Solanum esculentum (Liu et al., 2002; Hileman et al., 2005; Drea et al., 2007; Kramer et al., 2007). Unlike wild-type and control plants, individuals consistently infected with pTRV2-SBP1 were either significantly late flowering or failed to flower completely. Individual differences in flowering time were probably caused by quantitative and/or temporal differences in gene silencing or by the loss of the pTRV2-SBP1 vector, similar to previously reported variation (Drea et al., 2007; Kramer et al., 2007). These data demonstrate that SBP1 is critical for the transition to flowering.

Positive regulation of meristem identity genes by Antirrhinum majus SBP-box proteins

Previous studies hypothesized a role for SBP1 in inflorescence and flower development, based on the observation that SBP1 expression is most abundant in inflorescence tissues, and that in vitro its protein product directly binds to the promoter of the floral meristem identity gene SQUA (Klein et al., 1996; Huijser et al., 1992). We found a similar pattern of upregulation for SBP1 and SBP2 during vegetative and inflorescence development. In addition, the upregulation of both genes occurred before that of the FUL-like genes. In combination with promoter analyses, demonstrating the presence of consensus binding sites in the closely related M. guttatus genes, MgFUL and MgSQUA-like, these data are consistent with an early role for SBP1 and SBP2 in the regulation of all three FUL-like genes. However, differential expression patterns in the flower suggest different or additional regulators of FUL-like genes in late inflorescence and flower development.

If SBP1 is upstream of FLO and FUL-like genes, silencing of SBP1 should result in the downregulation of FLO and FUL-like gene expression in plant tissues where they are normally expressed. To test this prediction we compared gene expression in leaves of late flowering SBP1-silenced and pre-flowering control plants. Transcript levels of SQUA, DEFH28, AmFUL and FLO were all consistently reduced in leaves of SBP1-silenced plants relative to control plants, despite similar levels of SBP2. In addition to promoter analyses, these data support the hypothesis that SBP1 directly regulates all three FUL-like genes in Antirrhinum majus. Furthermore, the fact that DEFH28 and AmFUL are expressed in both inflorescence and floral meristems (Müller et al., 2001; this study), and that SQUA is expressed only in floral meristems (Huijser et al., 1992), is consistent with SBP1 functioning at different levels or developmental stages to initiate and maintain the development of the Antirrhinum majus inflorescence. Unlike the FUL-like genes, no SBP-protein binding site has been found in FLO (Cardon et al., 1999). This either suggests an indirect interaction between SBP1 and FLO, or the presence of trans-acting SBP1 regulatory elements that regulate FLO expression.

Conservation of the inflorescence genetic pathway across core eudicots

In Arabidopsis thaliana, constitutive expression of the SBP-box gene SPL3 results in early flowering (Cardon et al., 1997; Gandikota et al., 2007). A similar phenotype has been found for species overexpressing other developmental genes, such as the SEPALLATA (SEP), AGAMOUS (AG) and AP1 MADS-box genes, suggesting a general role for these genes in inflorescence development (Mizukami and Ma, 1997; Pelaz et al., 2001; Peña et al., 2001). Further evidence supporting a role for SPL3 in the early stages of the floral transition is based on the constitutive expression of its negative regulator miR156 (Wu and Poethig, 2006). Although miR156 targets multiple SBP-box genes (Birkenbihl et al., 2005; Wang et al., 2008), the constitutive expression of a miR156-insensitive form of SPL3 is able to correct the late-flowering phenotype of 35S:miR156 individuals (Wu and Poethig, 2006).

Our functional characterization of the SPL3–SPL5 ortholog from Antirrhinum majus supports a similar role for SBP1 in floral induction. However, there is clearly higher functional redundancy among Arabidopsis thaliana SPL3–SPL5 compared with Antirrhinum majus SBP1, as silencing of SPL3 alone results in no mutant phenotype, and SPL3–SPL5 silenced individuals flower eventually (Wu and Poethig, 2006; Wang et al., 2009). Furthermore, secondary inflorescence development in SBP1-silenced individuals suggests an additional role for SBP1 in maintaining apical dominance. SBP1-silenced lines grown under long-day conditions were either non-flowering or significantly late flowering, but showed no abnormal inflorescence phenotypes when individuals did flower. Spatiotemporal changes in infection suggest that flowering commenced because of a loss of the pTRV2-SBP1 virus, rather than redundancy with other flowering time genes. Thus, there is little evidence to support the hypothesis of Klein et al. (1996) that SBP1 and SBP2 function redundantly in the floral induction of Antirrhinum majus. The functional role of SBP2 remains to be tested.

Recent evidence suggests that SPL3–SPL5 genes facilitate flowering through the photoperiod and autonomous pathways via the positive regulation of the inflorescence meristem gene FUL, and the floral meristem identity genes AP1 and LFY (Fornara and Coupland, 2009; Wang et al., 2009; Yamaguchi et al., 2009). In combination with the identification of conserved SBP-box protein binding sites in asterid FUL-like genes (Klein et al., 1996; this study), our data demonstrate a similar regulatory pathway in Antirrhinum majus. Specifically, reduced levels of SBP1 resulted in significantly reduced expression of FUL-like and FLO genes. Unlike FUL from Arabidopsis thaliana, we saw no exclusion of AmFUL from floral meristems, suggesting differential regulation and partitioning of function between FUL-like genes of Arabidopsis thaliana and Antirrhinum majus. Although the exact role of AmFUL in inflorescence development is yet to be functionally assessed, its expression, in combination with the established role of SQUA and FLO in floral meristem identity specification (Huijser et al., 1992; Coen et al., 1990), suggests that FUL-like genes and FLO are essential to both establish and maintain inflorescence development. Taken together, these data suggest functional conservation of SBP-box proteins via the regulation of FUL- and LFY-like genes across the major asterid-rosid lineage of core eudicots.

Experimental procedures

Plant material

Antirrhinum majus ssp. majus seeds, accession number ANTI 11 (D2836), were obtained from the Gatersleben collection (Leibniz Institute of Plant Genetics and Crop Research, http://www.ipk-gatersleben.de). This was the sole Antirrhinum majus accession in which VIGS was attempted. J.K. Kelly provided seeds of M. guttatus DC ‘Point Reyes’. Plants were grown in a growth chamber under long-day conditions (16-h light/8-h dark) at 21–25°C.

Phylogenetic analysis

SBP1 and SBP2 were used in a BLAST search of the provisional M. guttatus genome sequence to identify homologs. A single gene sequence with high sequence identity to SBP1 was identified from M. guttatus; cloning and sequencing from a different (‘Point Reyes’) population was carried out to confirm this result. Full-length SBP-like amino acid sequences were aligned in MacClade (Maddison and Maddison, 2003), and phylogenetic relationships for their corresponding nucleotide sequences were estimated using maximum-likelihood methods in garli (Zwickl, 2006), and Bayesian methods in MrBayes 3.1.2 (Ronquist and Huelsenbeck, 2003). Both analyses were run using the best-fitting model of molecular evolution (GTR + I + Γ), as recommended by ModelTest 3.7 (Posada and Crandall, 1998). The maximum-likelihood analysis was run with 10 random-addition sequences, and the Bayesian analysis was run twice for 3 000 000 generations, sampling every 1000th generation. Parsimony bootstrap values were obtained using 1000 replicates in paup 4.0b10 (Felsenstein, 1985; Swofford, 2001).

Gene expression analyses

RNA was extracted from Antirrhinum majus material using the RNeasy Plant Mini kit and RNase-Free DNase Set (Qiagen, http://www.qiagen.com), and 1 μg was reverse transcribed using the iScript Select cDNA Synthesis kit (Bio-Rad, http://www.bio-rad.com). To confirm that there was no contaminating DNA, negative controls lacking the reverse transcriptase enzyme were generated for representative RNAs, and used in subsequent experiments. For RT-PCR, amplifications for each primer pair (Table S2) were run for different numbers of cycles to confirm linearity. ACTIN was amplified as a cDNA control for 26–28 cycles. All other genes were amplified for 32–35 cycles. Expression levels were normalized against ACTIN in ImageJ (Abramoff et al., 2004), and significant differences between infected and control plants were assessed using a one-tailed Wilcoxon paired sample test (Zar, 1984). All RT-PCRs were carried out twice to validate each result. One biological sample was used to determine the relative expression in wild-type leaves, inflorescences and flowers. Between six and nine biological replicates were used to test VIGS-mediated gene silencing.

Quantitative (q)-RT-PCR was carried out on a DNA Engine Opticon 2 real-time PCR machine (MJ Research, Ramsey, MN, USA) using SYBR Green I (Invitrogen, http://www.invitrogen.com) and DyNAzyme II Hot Start DNA Polymerase (Finnzymes, http://www.finnzymes.com) according to guidelines set out by Udvardi et al. (2008). Gene-specific primers were designed in Primer3, resulting in amplicons of approximately 150 bp in length (Rozen and Skaletsky, 2000) (Table S2). Prior to mRNA quantification, optimal annealing temperatures and transcriptional stabilities were determined for each primer pair (Table S2). In each case, amplicon length and sequence was verified by electrophoretic separation and sequencing. Three housekeeping genes, EF1alpha, UBIQUITIN5 (UBQ5) and ACTIN, were tested for transcriptional stability across tissues via amplification of a dilution series in triplicate. UBQ5 showed little transcriptional variation across different tissues, and was therefore selected as the qRT-PCR reference gene. After correcting for transcriptional stability, cycle threshold (Ct) values in target tissues were normalized against UBQ5, and the fold change was calculated by dividing the normalized values of the infected plants with that of uninfected plants. The mean and standard deviation was determined for two biological replicates, with four technical replicates each.

In situ hybridizations were carried out with AmFUL sense and antisense riboprobes, as previously described (Jackson, 1991; Jackson et al., 1994; Malcomber and Kellogg, 2004; Preston and Kellogg, 2007). The AmFUL nucleotide sequence was obtained from GenBank (AY306139.1) (Litt and Irish, 2003).


A 187-bp gene-specific region of SBP1 was cloned into pGEM-T, and subcloned in the sense orientation into the pTRV2 vector using BamHI and XhoI (Figure 1b; Table S1). Electrocompetent Agrobacterium was then transformed with pTRV2-SBP1, pTRV2-Empty or pTRV1. Whereas pTRV2 is involved in generating double-stranded RNA of the target gene, pTRV1 is involved in the replication of both pTRV2 and pTRV1 (Robertson, 2004). Single Agrobacterium colonies containing each pTRV vector were used to inoculate 5 ml of Luria Broth (LB), and after 26 (TRV2) and 48 h (TRV1), these cultures were used to further inoculate 500 ml LB cultures, grown overnight to an OD260 of 0.6–0.9 (Hileman et al., 2005). Overnight, 500-ml cultures were screened for the presence of the appropriate vector using the primers pYL156F and pYL156R for pTRV2, and OYL195 and OYL198 for pTRV1, pelleted for 20 min at 4000 g, and resuspended in infiltration solution to an OD260 of 2.0, as previously described (Hileman et al., 2005).

Batches of 50 Antirrhinum majus seedlings at the two- to four-leaf pair stage were vacuum infiltrated for 10 min with Silwet and a 1:1 mixture of pTRV1 and either pTRV2-SBP1 or pTRV2-Empty, as previously described (Dinesh-Kumar et al., 2003; Hileman et al., 2005). Infiltrated seedlings were washed in infiltration solution and planted in soil. After 4 weeks, the surviving plants were screened for the presence of viral vectors (Hileman et al., 2005), and for flowering time and phenotypic defects. Flowering time was measured as both leaf node number [sum of decussate (vegetative) and spirally (inflorescence) arranged nodes; following Bradley et al., 1995] prior to flower production and days to flowering. Plants infected with an empty vector (pTRV2-Empty) were used as experimental controls (Hileman et al., 2005).

Identification of SBP-box protein consensus binding sites

Genbank-obtained Arabidopsis thaliana AP1 and FUL genes were used in a BLAST search of the M. guttatus genome sequence to identify homologs. Phylogenetic analyses were carried out on an alignment of FUL-like genes to determine orthology, as previously described, and putative SBP binding sites were identified for available genomic sequences by searching for the consensus sequence CCGTAC (Birkenbihl et al., 2005; Liang et al., 2008).


We thank M.A. Kost and L. Baldridge for preliminary VIGS experiments, L. Baldridge for help with qRT-PCR, J.K. Kelly for M. guttatus seeds, and V.F. Irish, I. Jiménez and two anonymous reviewers for helpful comments on an earlier version of this manuscript. D. Rokhsar and J. Schmutz of the Department of Energy Joint Genome Institute and J.H. Willis at Duke University provided access to the draft M. guttatus whole-genome sequence. This work was supported by the National Science Foundation (grant IOS-0616025 to LCH).

The sequences reported in this paper have been deposited in the Genbank database (accession nos HM011587 and HM011588).