The role of two LEAFY paralogs from Idahoa scapigera (Brassicaceae) in the evolution of a derived plant architecture


(fax +1 608 262 7509; e-mail


Idahoa scapigera produces solitary flowers in the axils of rosette leaves without elongation of the shoot axis, a rosette-flowering architecture. Previous work with one of the two I. scapigera LFY paralogs, IscLFY1, showed that this gene caused aerial flowering rosettes in Arabidopsis thaliana. In this paper, we report that after three generations IscLFY1 transgenic lines are phenotypically indistinguishable from wild-type Arabidopsis, indicating that IscLFY1 protein is able to replace normal LFY function. Additionally, we found that ectopic LFY expression late in development can phenocopy aspects of the aerial rosette phenotype, suggesting that shoot compression caused by IscLFY1 could be caused by localized overexpression of a functional IscLFY protein. We also characterized the expression and function of the second I. scapigera LFY paralog, IscLFY2, in A. thaliana. In contrast to IscLFY1, this paralog was expressed in floral meristems and the shoot apical meristem (SAM). In I. scapigera, LFY-specific antibodies detected high protein levels in developing flowers but not in the apex, suggesting trans-regulatory differences between I. scapigera and A. thaliana. Most IscLFY2 transgenic A. thaliana plants were indistinguishable from wild type, but in a minority of lines the SAM was converted to a terminal flower as would be expected from the reporter-expression pattern. Taken together these results show that both I. scapigera paralogs have conserved LFY function, both proteins can rescue lfy and both can modify inflorescence architecture in an A. thaliana background: either by affecting internode elongation (IscLFY1) or by causing homeotic conversion of shoots into flowers (IscLFY2).


In recent years, an increasing number of evolutionary developmental genetic studies have traced morphological innovations to their genetic causes. Crosses between closely related lineages have identified genetic loci of large effect that modify plant architecture in Zea (for a review see Doebley, 2004) and adaptive floral traits in both Aquilegia (Hodges et al., 2002) and Mimulus (Bradshaw and Schemske, 2003). In more distantly related lineages that no longer hybridize, direct transformation of Arabidopsis thaliana with genes from other species has provided evidence to support a role for KNOX genes in the evolution of leaf morphology (Hay and Tsiantis, 2006) and of LFY in the evolution of plant architecture (Sliwinski et al., 2006; Yoon and Baum, 2004). Nonetheless, the use of interspecies transformation to study the genetic basis of species is still in its infancy and the interpretation of the results remains challenging.

Idahoa scapigera, a diminutive species in Brassicaceae from northwestern USA, is one of several lineages that have evolved a novel flowering architecture, rosette-flowering, from an inflorescence-flowering ancestor (Figure 1). A. thaliana, a model of the ancestral inflorescence-flowering condition, has been used to elucidate the role of an I. scapigera homolog of LFY, IscLFY1, in modifying plant architecture (Yoon and Baum, 2004). It was shown that the IscLFY1 locus (a genomic fragment including the coding sequences and flanking DNA regions) yielded a number of abnormal phenotypes when placed in an A. thaliana lfy mutant background. Some transgenic plants produced an apetala1-like phenotype (flowers initiated from the periphery of older flowers in place of petals), and showed regions with shortened internodes and bracteate flowers. Occasionally, shoots terminated in aerial flowering rosettes: shoots bearing bracteate flowers and showing a complete lack of internode elongation. Because these aerial flowering rosettes so closely resembled the phenotypes of the donor species, I. scapigera, it seemed plausible that changes at the IscLFY1 locus interacted with other, as yet undetermined, genetic changes to cause the evolution of the rosette-flowering architecture (Yoon and Baum, 2004). Here we conducted follow-up studies of IscLFY1 and a second LFY paralog from I. scapigera, IscLFY2. These two genes are highly divergent, with only 85% nucleotide identity to each other, as contrasted with 87% nucleotide identity between each gene and the Arabidopsis LFY gene (Baum et al., 2005). Similarly, the inferred amino acid sequences share 84% identity with each other, versus 89 and 88% between IscLFY1 and IscLFY2, respectively, and Arabidopsis LFY. These observations raised the possibility that the two LFY paralogs of I. scapigera might have evolved divergent functions.

Figure 1.

Idahoa scapigera.
I. scapigera, a diminutive rosette-flowering Brassicaceae, does not produce an inflorescence but rather solitary flowers extend directly from the basal rosette on elongated pedicels.

In this article, we describe the expression of IscLFY proteins in I. scapigera and describe experiments in A. thaliana aimed at understanding the role of these proteins in rosette flowering. We explored the basis of internode compression in IscLFY1 transgenic A. thaliana lines and generated transgenic lines to examine the effect of IscLFY2 in an A. thaliana background, with or without IscLFY1. We also determined the effects of ectopic LFY expression late in A. thaliana development. We found no evidence of synergy between IscLFY1 and IscLFY2, but this could reflect amelioration of the IscLFY1 phenotypes by inadvertent selection. Interestingly, we found that some IscLFY2 transgenic lines produce terminal flowers on inflorescence stems similar to those induced by the LFY homolog of Leavenworthia crassa (Sliwinski et al., 2006). These results show that the two I. scapigera LFY paralogs have divergent functions and that each is capable of altering plant architecture in an A. thaliana genetic background. These data are compatible with a role for one or both genes in the evolutionary origin of rosette flowering.

Results and discussion

Expression of LFY homologs in I. scapigera and A. thaliana

To elucidate the distribution of LFY proteins in the I. scapigera shoot apex, we performed immuno-localizations with anti-LFY antibodies. Given the close similarity of the two proteins (Baum et al., 2005), it is likely that the signal detected in I. scapigera represents a composite of their combined expression. Hybridizations were conducted on sections of I. scapigera plants soon after the transition to flowering. At this time the shoot apical meristem (SAM) has a broadly conical shape and produces a succession of bracteate flower primordia on its flanks. Consistent with a role in floral meristem identity regulation, the highest protein concentrations were detected in floral anlagen and much lower levels were detected in the SAM (Figure 2a). This pattern resembles that reported in A. thaliana (Sessions et al., 2000). Although some signal above background was detected within the SAM, significantly stronger staining was visible in floral primordia. Weak expression was also visible in floral bracts of stage 1–3 flowers (stages follow Smyth et al., 1990), but this diminished to background levels after stage 4. Uniform IscLFY expression was maintained throughout the flower primordium up to stage 3, but was excluded from the elongating pedicel. By stage 8, stamens expressed IscLFY in anther locules, but expression was not seen in stamen filaments (Figure 2b). At this stage IscLFY was also expressed throughout the gynoecium with the highest signal in the developing ovules (Figure 2c).

Figure 2.

 LFY immuno-localization in Idahoa scapigera and reporter constructs in Arabidopsis thaliana.
(a) Longitudinal section through the I. scapigera apex showing strongest IscLFY expression in young flowers (f) with lower levels in bracts (b) and the shoot apical meristem (*).
(b) Stage-8 flowers showing IscLFY expression in stamens restricted mainly to anther locules (arrow).
(c) IscLFY is expressed in the gynoecium, with highest levels in developing ovules (arrow).
(d) IscLFY2 5′cis-regulatory DNA drives strong reporter expression in both young flowers and the shoot apical meristem (SAM).

In order to compare the expression driven by the 5′cis-regulatory regions of both I. scapigera paralogs, we transformed reporter constructs into A. thaliana. It was reported previously that a 2989-bp genomic fragment including the IscLFY1 5′cis-regulatory region drives reporter expression primarily in the anther locules and carpels, with no detectable expression in floral anlagen, perianth, or the SAM (Yoon and Baum, 2004). To determine the expression pattern of the other I. scapigera LFY paralog, a 3136-bp genomic fragment including the 5′ intergenic region of IscLFY2 was translationally fused in the first exon to the uidA reporter gene, and GUS expression was monitored in transgenic A. thaliana (Figure 2d). In contrast to the IscLFY1 reporter fusion, but in accordance with the immuno-localization results, we observed strong staining in floral anlagen and throughout young flowers. However, strong staining was also noted in the SAM, in contrast to the immuno-localization results and the expression of LFY in wild-type A. thaliana (Sessions et al., 2000). This IscLFY2 expression pattern resembles that obtained with the cis-regulatory regions of another LFY ortholog, LcrLFY (Yoon and Baum, 2004).

The discrepancy between the immuno-localization and the transcription-reporter results has several possible causes. The 5′cis-regulatory DNA of IscLFY2 may include a cis-regulatory element outside the cloned fragment (e.g. within one of the two introns) that inhibits expression in the SAM. Alternatively, there could be a trans-regulatory interaction in I. scapigera (lacking in A. thaliana) that causes IscLFY protein to be repressed in the SAM. A third possibility is that IscLFY2 is not expressed at detectable levels in I. scapigera, in which case the protein detected by immuno-localization in young flower primordia could be attributed to IscLFY1 having broader expression than predicted based on the transcription-reporter experiments in A. thaliana.

IscLFY1 transgenic A. thaliana lines

Previous work showed that IscLFY1 is able to rescue the floral defects of lfy-6, but that transgenic plants produced modified inflorescences with compressed internodes, bracteate flowers and aerial flowering rosettes (Yoon and Baum, 2004). Additionally, it was found that most of these lines produced fewer petals and some ap1-like flowers, which was attributed to a lack of IscLFY1 expression in the perianth causing inadequate activation of ap1 and other downstream genes. We characterized additional independent transgenic lines in a segregating lfy-6 population for a total of 97 independent IscLFY1 insertions, and found phenotypic variability not only between independent transgenic lines, as would be expected because of variation in genomic position of the transgene, but also within individual plants, suggesting variable expression through the course of development.

Among the 97 T1 lines examined, 24 were homozygous for lfy-6. Of these only two lines showed full floral rescue, where all flowers produced petals and functional stamens. The remaining 22 lines produced some lfy mutant flowers, ranging in degree from a complete lack of rescue (all flowers were lfy-like) to almost full rescue (<10% of flowers were lfy-like). We found a continuum in floral phenotypes between lfy mutant, ap1-like and wild type, but most ap1-like flowers were produced early in development (the first few flowers on the main stem). As in lfy-6, some of these 22 lines also produced bracteate flowers, although unlike lfy-6, bracts subtended wild-type flowers, suggesting that, even when conferring floral meristem identity, IscLFY1 is not as capable as LFY at suppressing bract formation in A. thaliana.

The only phenotype observed in these IscLFY1 transgenic lines that cannot be explained by a partial lack of LFY function is internode compression. This was noted primarily late in development, towards the tips of inflorescence shoots. In cases where the internodes between bracteate flowers became compressed, aerial flowering rosettes where formed that resembled the flowering habit of I. scapigera. However, severe internode compression also occurred on stems bearing shoots (Figure 3a) and stems bearing ebracteate flowers (Figure 3b). Most plants showed internode compression in fewer than half of their inflorescence branches, but in rare cases we found severe compression throughout the entire inflorescence (Figure 3c), although these plants were sterile.

Figure 3.

 Phenotypes of IscLFY1 transgenic Arabidopsis lines.
The range of phenotypes produced by IscLFY1 in Arabidopsis thaliana lfy-6 was variable.
(a) Severe internode compression was occasionally visible between shoots on paraclades.
(b) Severe internode compression was also found between ebracteate flowers at the paraclade apex.
(c) Rarely, severe compression occurred at all apices throughout the inflorescence. These plants generally did not produce seeds.
(d) Bulk seed collections selected IscLFY1 transgenic plants that were indistinguishable from wild-type A. thaliana. This T3 transgenic plant of line 7-2 showed full rescue of lfy-6. Petals were present on all flowers and the first flower produced on the main axis was fertile (silique indicated by arrow).
(e) Plants within one transgenic line (7-1) still showed partial rescue of lfy-6 in the T6 generation. The first flower produced on the main axis (arrow) and a few subsequent flowers were lfy-like, whereas fertile flowers were produced later in development. This developmental pattern was repeated in the paraclades.

To limit the variability associated with the IscLFY1 transgene, we established true breeding lines homozygous for IscLFY1 in a lfy-6 background. We selected three independent T1 lines that were homozygous for lfy-6, based on cleaved amplified polymorphic sequences (CAPS) genotyping, and were inferred to each contain a single insert based on segregation of the resistance marker in the T2 generation. These lines included one full floral rescue line (23-2) and two partial rescue lines (7-1 and 7-2 produced some lfy-like flowers). Most of the individual T2 plants in these lines produced aerial rosettes (10 of 10 T2 plants of 7-1; 6 of 6 T2 plants of 7-2; 11 of 16 T2 plants of 23-2) and some produced ap1-like flowers (9 of 10 plants of 7-1; 1 of 6 plants of 7-2; 10 of 16 plants of 23-2).

We selected individual T2 plants from each line that were found to yield 100% Basta resistance in the subsequent generation, indicating homozygosity at the transgene locus. Seeds were collected from these true breeding T3 populations and later generations in bulk from multiple plants. Interestingly, from the T3 to the T6 generation, we observed an increase in the number of wild-type flowers produced and a decrease in the frequency of internode compression. By the T6 generation, the 23-2 and 7-2 lines were indistinguishable from wild type. Internode length was no longer suppressed at any point in inflorescence development and no lfy-like flowers were produced (Figure 3d). Most plants within line 7-1 were also indistinguishable from wild type, but even in the T6 generation a small proportion (<10%) still produced lfy-like flowers, generally early in development (Figure 3e). Even though these 7-1 plants produced lfy-like flowers, the internodes throughout the inflorescence were indistinguishable from wild type.

We reconfirmed homozygosity at lfy-6 for these later generation plants using CAPS markers, and confirmed the presence of IscLFY1 transcripts by RT-PCR in selected T4 lfy-6/lfy-6 plants, the phenotypes of which were indistinguishable from wild type (data not shown). Thus, seed contamination neither explains the improved rescue of lfy-6 nor the absence of compressed internodes. Rather, it seems likely that inadvertent selection for fecundity yielded lines with genetic or epigenetic characteristics that result in phenotypes closer to wild type. Given the limited number of generations and the consistent pattern of reversion towards wild type in multiple independent lines, novel mutations that rescued protein function seem unlikely. Rather, we conclude that IscLFY1 is able to replace LFY function when expressed in an appropriate manner. The observation that a majority of T1 transgenic lines exhibited abnormal phenotypes would be explained by variability between lines, and even shoots within a plant, in the pattern of gene expression. We hypothesize that selection for expanded IscLFY1 expression or function is responsible for the improved rescue of lfy-6. Although these lines provide an opportunity for future studies of epigenetic effects, they are no longer suitable for studying how IscLFY1 caused the internode compression phenotype.

The effect of late overexpression of LFY on plant architecture

The phenotype of the IscLFY1 transgenic Arabidopsis lines with the greatest potential for explaining the evolution of rosette flowering is internode compression, because this resembles the flowering portion of the I. scapigera stem. This phenotype led us to ponder how changes in the expression or function of LFY, a meristem identity gene, could lead to compressed internodes. We noted that, from an evolutionary point of view, a flower is a shoot with very compressed internodes separating lateral organs (sepals, petals, stamens and carpels; see Baum and Hileman, 2006 and references therein). Because LFY confers floral identity on a meristem, it can be seen as formally upstream of a genetic program that prevents internode elongation within the flower. This led us to wonder whether LFY-dependent suppression of internodes could be genetically dissociated from other aspects of floral identity, most notably meristem determinacy and floral organ initiation.

To examine whether aspects of LFY function could be isolated, we used an inducible P35S:LFY:GR construct (Wagner et al., 1999) in a wild-type A. thaliana background. This construct enabled us to activate LFY function with dexamethosone (DEX) treatment at various stages during normal development. We reasoned that there might be a developmental window when inflorescence meristems are canalized to maintain indeterminacy and to continue producing floral meristems on their flanks, yet are susceptible to LFY-dependent suppression of internode elongation. Indeed, we found that ectopic LFY activity in plants that had already produced flowers, and had inflorescence meristems that were still active, yielded inflorescence apices with severely compressed internodes (compare Figure 4a and 4b). This LFY-mediated internode compression occurred at all inflorescence shoot apices (Figure 4c), which contrasts with plants exhibiting IscLFY1-mediated internode compression, which varied from slight compression between nodes bearing wild-type flowers to severe compression throughout an inflorescence without fertile flowers (Figure 3c). Examination of the P35S:LFY:GR flowers produced in the zone of internode compression showed that gynoecium development was severely altered. Specifically, we noted a proliferation of unfused carpels and the conversion of young flowers into carpels (Figure 5a). Embryo development was also abnormal, including the appearance of possible stigmatic tissue on embryos within the carpel (Figure 5b and c), resulting in sterility after DEX treatment.

Figure 4.

 Phenotypes induced by LFY activation late in development.
LFY was activated by dexamethasone (DEX) treatment of transgenic P35S:LFY:GR Ler plants after between two and five open flowers were produced.
(a) The apex of a LFY-activated plant shows severe compression between floral nodes.
(b) The control Ler plants showed normal internode elongation following irrigation with DEX.
(c) Severe compression was apparent at all inflorescence apices of the P35S:LFY:GR plant irrigated with DEX (left) compared with the DEX-treated non-transgenic Ler plant (right).

Figure 5.

 SEM images of compressed apices caused by P35S:LFY:GR and IscLFY1.
(a) Image of the shoot apex after activation of LFY late in the development of a P35S:LFY:GR transgenic plant. Young floral primordia have been transformed into gynoecium and some older flowers are producing unfused carpels. Scale bar = 1 mm.
(b) Embryo development in activated P35S:LFY:GR plants was disrupted in the zone of compression. Scale bar = 0.1 mm.
(c) Some embryos of activated P35S:LFY:GR plants show abnormal growth at the embryonic shoot apical meristem while still within the carpel, including the development of stigmatic papillae (arrow). Scale bar = 0.1 mm; r, root.
(d) A non-transgenic Arabidopsis thaliana Ler plant treated with dexamethasone (DEX) showing normal development. Scale bar = 0.5 mm.
(e) Embryos within this control plant did not produce visible abnormalities. Scale bar = 0.1 mm; c, cotyledon; r, root.
(f) At the compressed apex of this transgenic IscLFY1 plant, anthers (a) are rare and neither sepals nor petals are visible. Similar to the activated P35S:LFY:GR plants, the youngest flowers have been transformed into gynoecium and many unfused carpels are present.

If IscLFY1 is functionally equivalent to LFY, then it is probable that the internode suppression phenotype of IscLFY1 lines is the result of overactivation of LFY function late in development – with different degrees of internode compression resulting from different degrees of IscLFY upregulation. To evaluate this hypothesis, we examined the flowers from regions of the strongest internode suppression in IscLFY1 transgenic lines to determine if they resembled the flowers produced by P35S:LFY:GR plants following DEX application. At the apex of these severely compressed paraclades (secondary inflorescence branches), most flowers were sterile and produced unfused claw-like carpels (Figure 5f) similar to the gynoecial abnormalities seen in treated P35S:LFY:GR plants. The youngest floral primordia were also transformed into carpels and did not produce other floral organs. These observations explain the reduced fertility inferred for IscLFY1 plants exhibiting internode compression, and can account for inadvertent selection for flowers and shoots that showed more tightly regulated IscLFY1 expression.

These P35S:LFY:GR experiments confirm that a change in LFY regulation is capable of compressing an inflorescence axis, although this change must be tightly regulated to cause internode compression without disrupting proper floral development. This result demonstrates the possibility that IscLFY1 could be acting in I. scapigera to prevent inflorescence elongation, thus contributing to the evolution of the rosette flowering habit.

IscLFY2 transgenic A. thaliana lines

To explore whether IscLFY2 played a role in the evolution of rosette flowering, we constructed IscLFY2 transgenic A. thaliana lines following the same strategy used previously for IscLFY1 (Yoon and Baum, 2004). A genomic fragment containing the IscLFY2 coding sequence and cis-regulatory sequences (2.8-kb 5′ and 0.9-kb 3′ of the coding region) was introduced into a segregating lfy-6 population by floral-dip transformation. In contrast to IscLFY1 transgenic lines, most of the IscLFY2 primary transformants (18 of 23 T1 plants) showed wild-type flower and inflorescence development with no detectable floral defects. Genotyping of sib-groups from these lines with CAPS markers identified some plants homozygous for lfy-6 that were indistinguishable from non-transgenic wild-type plants, indicating IscLFY2 is able to rescue LFY function.

As expected from the observed reporter expression pattern for IscLFY2 showing strong expression at the SAM (Figure 2d), some IscLFY2 transgenic lines (5 of 23 T1 plants) converted the SAM to a terminal flower (Figure 6a and b). The timing of these terminal flowers was variable, from immediately following the transition to the reproductive phase (Figure 6b) to after the production of 23 flowers on the main stem. This variation was seen between sibling plants and between shoots of a single plant, suggesting that IscLFY2 function within the A. thaliana SAM is not tightly controlled by genotype. To determine whether IscLFY1 and IscLFY2 might produce phenotypes not found in the single transgene lines, we generated double transgenic lines (IscLFY1/2) by dip-transforming the three IscLFY1 homozygous lines described above with the IscLFY2 genomic construct, and selecting for a second resistance marker (kanamycin). The resulting 25 primary transformants all produced phenotypes previously seen in the IscLFY2 single transgene lines, yielding no evidence of a synergistic effect. As was the case for the single transgenic IscLFY2 lines, a majority (20 lines) of the IscLFY1/2 double transgenic T4/T1 plants produced a wild-type inflorescence, whereas a minority (five lines) produced terminal flowers (Figure 6c and d). This is not a significantly different proportion of terminal flower lines to that generated by IscLFY2 alone (χ2 test, P = 0.83).

Figure 6.

 The range of terminal flower phenotypes on the primary axis of transgenic Arabidopsis lines containing IscLFY2.
(a) Terminal flowers were occasionally produced on the primary inflorescence axis of IscLFY2 transgenic plants later than tfl1-2 plants (after more than six floral nodes had been produced).
(b) In contrast, some IscLFY2 transgenic plants formed a terminal flower much earlier in development.
(c) A similar range of terminal flower phenotypes was observed in the double transgenic IscLFY1/2 plants including terminal flowers late in development.
(d) Terminal flowers were formed early in the development of some IscLFY1/2 plants.
(e) In our growth conditions tfl1-2 plants consistently produced a terminal flower after between one and six floral nodes had been produced.
(f) In our growth conditions, lfy-6 plants consistently produced a terminal flower (inset) at proliferative arrest after more than 30 nodes appeared on the main stem.

The terminal flowers produced in IscLFY2 and IscLFY1/2 transgenic lines resemble those produced by loss-of-function mutations in the TFL1 gene (Figure 6e) (Alvarez et al., 1992; Shannon and Meeks-Wagner, 1991), and also those induced by the L. crassa LFY homolog, LcrLFY (Sliwinski et al., 2006; Yoon and Baum, 2004), in that they lacked petals and usually contained two or more deformed carpels. Similar flowers are also observed at the apex of inflorescence shoots that have been forced to produce many more flowers than is typical before the onset of proliferative arrest, for example by continual removal of fruits (Hensel et al., 1994). Indeed lfy-6 mutants, which are sterile, produce such terminal flowers after at least 30 nodes on the primary axis in our growth conditions (Figure 6f). This mechanism does not seem to be responsible for the terminal flowers in IscLFY2, IscLFY1/2 and LcrLFY transgenic lines, as terminal flowers in these plants are generally produced well before proliferative arrest. A more specific genetic mechanism is likely to be acting in these lines, for example an inability of TFL1 to repress these orthologs in the Arabidopsis SAM. However, the Arabidopsis SAM is at least partially protected from IscLFY2 (either by TFL1 activity or by an independent mechanism) because more floral nodes can be produced in the IscLFY2 transgenic lines, which range from 1 to >23 floral nodes before a terminal flower, than in tfl1-2 mutants, which never produced more than six floral nodes on the primary axis in our growth conditions.

A model for the evolution of rosette flowering in I. scapigera

Two developmental models for the evolution of the rosette-flowering architecture from an inflorescence-flowering ancestor have been described (Yoon and Baum, 2004). The first model involved changes to the reproductive phase of development such that inflorescence internodes no longer elongate and leaf (=bract) repression is diminished. The second model involved changes to the vegetative (rosette) phase of development, such that axillary shoot meristems are converted into ectopic flowers and the vegetative phase is extended, delaying the transition to the reproductive phase. If IscLFY1 played a role in the evolution of rosette flowering, it was most likely via the first developmental model, as indicated by the capacity of this gene to suppress internodes and derepress bracts in an A. thaliana background. In contrast, if IscLFY2 played a role in the evolution of rosette flowering, it was most likely via the second developmental model, as indicated by the capacity of this gene to convert shoot meristems to flower identity. It should be noted that aspects of the two developmental models are not intrinsically incompatible. Thus, one can imagine a composite model (Figure 7) in which one set of genetic changes (including IscLFY1) suppressed internodes in the inflorescence phase, whereas a second set of genetic changes (including IscLFY2) caused paraclades in the inflorescence to be converted to solitary flowers. Thus, our data are compatible with the two IscLFY genes having distinct but complementary roles in the evolution of rosette flowering. Although this may be so, we must critically ask how well supported is this hypothesis and what could one do to more rigorously test it?

Figure 7.

 Model of the evolution of rosette flowering in Idahoa scapigera involving both IscLFY paralogs. In the I. scapigera inflorescence flowering ancestor (left), ebracteate flowers developed acropetally and paraclades (colored in gray) developed basipetally after the reproductive transition. In the lineage leading to I. scapigera (right), IscLFY1 may have evolved to compress the primary inflorescence and derepress bracts (flowers in black), whereas IscLFY2 may have evolved to transform paraclades into solitary terminal flowers (flowers in gray). This model predicts that IscLFY1 would be active acropetally in I. scapigera (suppression of internodes at the shoot apical meristem and bract derepression subtending the youngest flowers) and that IscLFY2 would be active basipetally (homeotic transformation of axillary shoots into flowers). To provide further evidence either in favor or against this model, additional in situ hybridizations and functional studies in I. scapigera are necessary. Filled oval, leaf; open circle, flower; arrow, meristem.

Interspecies transformation studies have both strengths and weaknesses when used to study the genetic basis of species differences. On the positive side, studying genes from different species can reveal fundamental mechanisms that are obscured in traditional mutant or reverse-genetic studies. For example, the addition of petals in LcrLFY lines might indicate a previously unforeseen role of LFY in corolla merosity (Yoon, 2003; Yoon and Baum, 2004). Likewise, our studies of IscLFY1 and P35S:LFY:GR lines suggests that in parallel with its well-known role as a regulator of the floral organ identity program, LFY may function in suppressing internodes within the floral axis. On the negative side, transgenic experiments that do not yield perfect conversion of the recipient species into the phenotype of the donor species may provide neither rigorous rejection nor support of a prior hypothesis. Therefore, to more clearly evaluate the hypothesized role of LFY in rosette flowering our results require additional experiments. Functional studies and gene-specific studies of expression in I. scapigera or another rosette-flowering plant are still necessary to elucidate the role of IscLFY genes in the development of rosette flowering. Although such work would certainly not be easy to initiate, if one moved the A. thaliana LFY gene into a rosette flowering plant and observed, for example, shoot elongation and conversion of flowers into shoots, then the evolutionary hypothesis for the role of LFY in rosette flowering would gain stronger support. Similarly, the introduction of IscLFY genes into other inflorescence-flowering Brassicaceae besides A. thaliana (especially closer relatives of Idahoa) could help rule out the possibility that the IscLFY results are developmental artifacts specific to A. thaliana. A further class of interspecific transformation experiments, introducing the LFY loci from other inflorescence-flowering Brassicaceae into A. thaliana, could then be used to rule out the possibility that terminal flowers/suppressed internodes are a non-specific consequence of subtle misregulation of exogenous LFY genes. Thus, although evolutionary transgenic research is not as simple as perhaps it once appeared, it certainly has the potential to help elucidate many aspects of plant development and evolution.

Experimental procedures


Fresh plant tissue was fixed in 4% paraformaldehyde (solid in 1 ×  PBS) under vacuum for 4 h. Tissue was dehydrated in an ethanol series, moved to Hemo-De (Fisher Scientific,, and embedded in paraffin (Paraplast Plus; Fisher Scientific). Embedded tissue was sectioned longitudinally with a microtome (Leica RM 2145 Rotary Microtome; Leica, Sections (8-μm thick) were affixed to Probe-On Plus slides (Fisher Scientific). Slides were deparaffinized in Hemo-De, rehydrated in an ethanol series and treated in Coplin jars as follows: 10 min in 20 μg ml−1 proteinase K (Roche,; diluted in TE 0.1 m Tris/0.05 m EDTA), two times for 5 min in PBS, 30 min BTX (100 mm Tris–HCl, pH 7.5, 400 mm NaCl, 1% BSA, 0.3% Triton X-100). Slides were transferred into a humid box and treated as follows: 3-h incubation in blocking solution (10% goat serum in BTX; Sigma,, 90-min incubation in 1 : 300 dilution of LFY antibody (Sessions et al., 2000) at 25°C, 15-min rinse three times in BTX (in Coplin jars), 60-min incubation in 1 : 1500 dilution of goat anti-rabbit Alkaline Phosphatase-conjugated secondary antibody (Promega,, 15-min rinse three times in BTX (in Coplin jars), 20-min incubation in detection buffer (100 mm Tris–HCl, pH 9.5, 100 mm NaCl, 50 mm MgCl2). NBT/BCIP stock solution (Roche) was diluted in detection buffer (NBT, 0.15 mg ml−1; BCIP, 0.075 mg ml−1). Staining times varied between 45 min and 2.5 h. Staining was stopped with 1x TE, dehydrated in an ethanol series, incubated in Hemo-De and mounted with Permount (Fisher Scientific). Sections were imaged using an Olympus BX-60 Optical Microscope and Olympus DP70 Digital Camera. Images were cropped, reoriented and contrasted using Adobe Photoshop (Adobe Systems Inc.,

GUS fusion

An IscLFY2 reporter was constructed and stained following the IscLFY1 reporter protocol (Yoon and Baum, 2004). Briefly, a 3135-bp I. scapigera genomic fragment including 2794-bp 5′ of the start codon and extending into the first IscLFY2 exon was generated with PCR using Pfu (Stratagene, and the primers IsLFY2-5′F2-NheI 5′-AAAAAGCTAGCCACGCCCAGAGATGACGGAGAGATTACAAC-3′ and IsLFY2-exon1R 5′-CGTTCAGCTCTAACGGCAGCTTTAATA-3′. The forward primer added an NheI site (underlined) for cloning. The vector, pBI101.3, was cut with XbaI and SmaI for directional cloning. Athaliana Ler plants were transformed via the floral-dip method (Clough and Bent, 1998).

Ectopic induction of LFY activity

Transgenic Athaliana Ler seeds containing a constitutively expressed translational fusion of LFY to the rat glucocorticoid receptor hormone binding domain, P35S:LFY:GR, were obtained from D. Wagner (Wagner et al., 1999). Plants were grown in long-day conditions (16-h light) under fluorescent lights at 22–25°C. Application of the steroid hormone DEX switches on LFY activity post-translationally by allowing the entrance of the LFY:GR fusion protein into the nuclei from the cytoplasm. Beginning approximately 4 weeks after germination (between two and five open flowers), 5 μm DEX was applied by irrigation continuously. Control non-transgenic Ler plants received the same DEX treatment.


Living specimens were fixed in Formalin: Acetic Acid: Alcohol (FAA) and subsequently stored in 70% ethanol. Tissue samples were then dehydrated in an ethanol series, critical-point dried, sputter-coated with gold, and studied at 5 kV in a Hitachi S570 SEM (Hitachi,

IscLFY2 construct

An I. scapigera genomic fragment including 2794 bp upstream of the start codon and 901 bp downstream of the stop codon was cloned following the GUS fusion protocol described above with the following modifications. The reverse primer was IsLFY2-(WI)R 5′-CCAGTAGAACCGAACCGATGATTATCCC-3′. The resulting PCR fragment was cloned into pCAMBIA2300 and transformed into A. thaliana segregating for lfy-6 or into true breeding IscLFY1 transgenic T3 lines (7-1-3, 7-2-3, 7-2-4, 23-2-8). For CAPS genotyping of lfy-6, the modified protocol of Yoon et al. was used in this study (Yoon and Baum, 2004).


Funding was provided by a grant from the NSF (IBN 0 234 118). MVB was supported by a postdoctoral fellowship from the Forschungskommission of the University of Zurich, Switzerland. The authors would like to thank Michael Frohlich and one anonymous reviewer for comments on an earlier draft of this manuscript and the BBPIC Laboratory of the University of Wisconsin-Madison for the use of their scanning electron microscope.