Contrasting Evolutionary Forces in the Arabidopsis thaliana Floral Developmental Pathway

Corresponding author: Michael D. Purugganan, Department of Genetics, North Carolina State University, Raleigh, NC 27695., michael_purugganan{at}ncsu.edu (E-mail)

Communicating editor: M. K. UYENOYAMA

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

The floral developmental pathway in Arabidopsis thaliana is composed of several interacting regulatory genes, including the inflorescence architecture gene TERMINAL FLOWER1 (TFL1), the floral meristem identity genes LEAFY (LFY), APETALA1 (AP1), and CAULIFLOWER (CAL), and the floral organ identity genes APETALA3 (AP3) and PISTILLATA (PI). Molecular population genetic analyses of these different genes indicate that the coding regions of AP3 and PI, as well as AP1 and CAL, share similar levels and patterns of nucleotide diversity. In contrast, the coding regions of TFL1 and LFY display a significant reduction in nucleotide variation, suggesting that these sequences have been subjected to a recent adaptive sweep. Moreover, the promoter of TFL1, unlike its coding region, displays high levels of diversity organized into two distinct haplogroups that appear to be maintained by selection. These results suggest that patterns of molecular evoution differ among regulatory genes in this developmental pathway, with the earlier acting genes exhibiting evidence of adaptive evolution.

GENES that control morphogenesis invariably function as interacting components of complex developmental networks (ARNONE and DAVIDSON 1997 ; SHUBIN et al. 1997 ; DAVIDSON 1999 ). Morphological evolution is believed to arise from the diversification of these interacting developmental genes (SHUBIN et al. 1997 ; DOEBLEY and LUKENS 1998 ; PURUGGANAN 1998 ), including both the trans-acting regulatory genes that function within developmental networks and cis-acting promoter sequences that control gene expression patterns (DOEBLEY and LUKENS 1998 ). Despite the central importance of developmental gene evolution for the diversification of morphology, little is known about how developmental gene networks evolve. Previous studies on individual developmental genes indicate that they are subject to a variety of microevolutionary forces (WANG et al. 1999 ; LUDWIG et al. 2000 ; PURUGGANAN 2000 ), and some genes have been shown to be targets for adaptive evolution (WANG et al. 1999 ). However, the evolutionary dynamics across interacting regulatory genetic components remain unclear.

Analysis of patterns of nucleotide variation at different genes within a developmental pathway may allow us to assess whether the genetic components of a developmental network are subject to equivalent evolutionary forces, or whether they differ in their modes of evolution. An evolutionary analysis across genes also allows an examination of how the selective pressures acting on a gene relate to the gene's position and functional role within the developmental pathway. Moreover, this type of integrated analysis may further allow us to examine how the structure of developmental gene networks constrains the types of evolutionary change observed in nature.

Flower development provides an excellent system for studying the evolution of morphogenesis in plants (LAWTON-RAUH et al. 2000 ; CRONK 2001 ). Geneticists have managed to identify and isolate many of the genes that control flower development in the weed Arabidopsis thaliana, which has served as a model plant genetic system (WEIGEL 1995 ; YANOFSKY 1995 ; JACK 2001 ). Moreover, in many cases, geneticists have elucidated the interactions that occur among these genes to control floral development (see Fig 1; YANOFSKY 1995 ; LILJEGREN et al. 1999 ; NG and YANOFSKY 2000 ; JACK 2001 ).

View larger version (9K): In this window In a new window Download PPT slide   Figure 1. Schematic of the genetic interactions in the A. thaliana floral developmental pathway.

Among the regulatory genes first expressed in floral development is the inflorescence architecture gene TERMINAL FLOWER1 (TFL1), whose product is similar to RAF kinase inhibitor proteins and which appears to be required for the maintenance of inflorescence meristem identity (BRADLEY et al. 1997 ; RATCLIFFE et al. 1998 ). In A. thaliana, mutations at TFL1 result in the transformation of the apical inflorescence meristem into a floral meristem, leading to the formation of a single terminal flower instead of an indeterminate inflorescence (SHANNON and MEEKS-WAGNER 1991 ; RATCLIFFE et al. 1998 , RATCLIFFE et al. 1999 ).

Three floral meristem identity genes, LEAFY (LFY), APETALA1 (AP1), and CAULIFLOWER (CAL), act downstream of TFL1 and specify the formation of flowers (see Fig 1). LFY encodes a putative DNA-binding transcriptional activator; mutations in this gene result in the partial transformation of flowers to inflorescence shoots (WEIGEL et al. 1992 ; WEIGEL and MEYEROWITZ 1993 ). Genetic studies indicate that TFL1 acts in part by repressing the expression of LEAFY in the inflorescence meristems (see Fig 1; LILJEGREN et al. 1999 ). Thus, downregulation of TFL1 leads to LFY expression and is one of the first steps in the genetic cascade that leads to flower formation. LFY in turn activates the expression of AP1, which encodes a MADS-box DNA-binding transcriptional factor (MANDEL et al. 1992 ; GUSTAFSON-BROWN et al. 1994 ; LILJEGREN et al. 1999 ). The third floral meristem identity gene, CAL, is a relatively recent duplicate of AP1 (KEMPIN et al. 1995 ). CAL and AP1 share redundant functions in the establishment of floral meristem identity, and double mutants of these two genes result in a massive proliferation of inflorescences resulting from an ontogenetic arrest in the ability to form floral meristems (BOWMAN et al. 1993 ; KEMPIN et al. 1995 ).

Floral organ identity genes control the identity of the organs in the four floral whorls—the sepals, petals, stamens, and carpels (YANOFSKY 1995 )—and act downstream of the meristem identity loci. The AP1 locus has floral organ identity functions in addition to its meristem identity functions described above; mutations at AP1 lead to loss of floral sepals and petals (BOWMAN et al. 1993 ). Other floral organ identity genes include the MADS-box loci APETALA3 (AP3) and PISTILLATA (PI), both of which are required for petal and stamen development (GOTO and MEYEROWITZ 1994 ; JACK et al. 1992 ). Mutations at these loci result in the homeotic transformation of petals and stamens to sepals and carpel-like structures, respectively. AP3 and PI are ancient gene duplicates (PURUGGANAN 1997 ; KRAMER et al. 1998 ). Expression of these two genes involves coregulation as well as regulation by the floral meristem identity genes LFY and AP1 (see Fig 1; SAMACH et al. 1997 ).

The broad functional categories used to classify these floral developmental genes—inflorescence architecture, meristem identity and organ identity—are not absolute, and some genes may be expected to have roles characteristic of more than one functional class, as is evident in AP1. Nevertheless, these categories provide a framework for formulating hypotheses about how evolution may act among the different genetic components of the floral developmental pathway. One such hypothesis involves the action of selection on genes that control evolutionarily conserved vs. variable morphological traits. Like all members of the Brassicaceae, A. thaliana shows strong conservation in the number, positioning, and identities of floral organs. In contrast, adaptive divergence is observed in the numbers and positioning of reproductive shoots, both within A. thaliana and among its close relatives (ENDRESS 1992 ; SHU et al. 2000 ). This difference in evolutionary constraint should be reflected in the evolution of the genes underlying these morphological traits. In particular, we would expect that evidence of adaptive evolution should be more prevalent in the inflorescence architecture and meristem identity loci, which control the evolutionarily more labile reproductive meristem traits, than in the organ identity genes.

In previous studies, we reported on the molecular population genetics of the floral meristem identity gene CAL (PURUGGANAN and SUDDITH 1998 ) and the floral organ identity genes AP3 and PI in A. thaliana (PURUGGANAN and SUDDITH 1999 ). We have now characterized the molecular population genetics of three additional A. thaliana floral developmental genes: the floral meristem identity genes AP1 and LFY and the inflorescence architecture gene TFL1. Our results indicate that these six genes possess distinctive patterns of sequence diversity and that the two early acting genes in the floral developmental pathway—the floral meristem identity gene LFY and the inflorescence architecture gene TFL1—show evidence of adaptive evolution.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Isolation and sequencing of alleles:
The A. thaliana ecotypes were obtained from single-seed propagated material provided by the Arabidopsis Biological Resource Center (ABRC; see Table 1). The Kent, Bretagne, Lisse, and Corsacalla seed stocks were from the population collection of P. H. Williams maintained at ABRC. A. lyrata seed was provided by C. H. Langley and O. Savolainen. The majority of the accessions used in this study were the same as those in previous work (PURUGGANAN and SUDDITH 1998 , PURUGGANAN and SUDDITH 1999 ).

Miniprep DNA was isolated from young leaves as previously described (AUSUBEL 1992 ). PCR was performed with the error-correcting Pwo polymerase (Roche, Indianapolis) to minimize nucleotide misincorporation. The error rate for this polymerase formulation, based on multiple amplification and resequencing of known genes, is less than one in 7–10 kb (our unpublished observations). We estimate that the nonsampling variance of nucleotide diversity due to PCR misincorporation, V_PCR(), is negligible [V_PCR()/V() 0.14] and does not significantly affect the frequency distribution of polymorphisms. In several cases, multiple sequences from independently amplified products were obtained to recheck potential PCR-induced errors. The APETALA1 gene was amplified in three segments in A. thaliana and A. lyrata: the first segment primers AP1-PR1F (5'-AGGCTTATGCAATATATGCCTTAAGC-3') and AP1-X1R (5'-TTGTCTATTGATCTTGTTCTCTATCC-3'); the second segment primers AP1-1F (5'-ATGGGAAGGGGTAGGGTTCA-3') and AP1-2R (5'-AAGGTTGCAGTTGTAAACGGG-3'); and the third segment primers AP1-X2F (5'-ATGGAGAAGATACTTGAACG-3') and AP1-X2R (5'-CTCAGGTGCAATAAGCTGTCT-3'). The LEAFY gene was amplified in two segments: the first segment primers LFY1F (5'-CAGACTCAGAGTGCTGATATTTCT-3') and LFY1R (5'-GTTCCTCAGATAACCCTGTCCA-3') and the second segment primers LFY2F (5'-TGGACAGGGTTATCTGAGGAAC-3') and LFY2R (5'-ATCTTAGTACTTTTGAGTTTGACC-3'). Finally, the TERMINAL FLOWER1 gene was amplified with the primers TFL16F (5'-CCTACTCTGAGCAATAATTGTATCC-3') and TFL1R (5'-GCAGTTTATGACAATCATGAAACTA-3'). Amplification conditions followed the Pwo polymerase manufacturer's protocols (Roche), with annealing temperatures adjusted for each primer pair.

Amplified DNA products were cloned into pCR2.1 using the TOPO TA cloning kit (Invitrogen, San Diego). DNA sequencing for both genes was conducted with an ABI377 automated sequencer using a series of nested internal sense and antisense primers. All sequence polymorphisms were visually rechecked from chromatograms, with special attention to low-frequency polymorphisms (HAMBLIN and AQUADRO 1997 ). The DNA sequences are available from GenBank (accession nos. AF466771, AF466772, AF466773, AF466774, AF466775, AF466776, AF466777, AF466778, AF466779, AF466780, AF466781, AF466782, AF466783, AF466784, AF466785, AF466786, AF466787, AF466788, AF466789, AF466790, AF466791, AF466792, AF466793, AF466794, AF466795, AF466796, AF466797, AF466798, AF466799, AF466800, AF466801, AF466802, AF466803, AF466804, AF466805, AF466806, AF466807, AF466808, AF466809, AF466810, AF466811, AF466812, AF466813, AF466814, AF466815, AF466816, AF466817).

Data analysis:
Sequences used in this study were visually aligned. Phylogenetic analyses were conducted using the heuristic search algorithm (maximum parsimony criterion) in PAUP 3.1 (SWOFFORD 1992 ), with the A. lyrata orthologs specified as the outgroup. Node support was assessed with 500 bootstrap replicates of the data. Nucleotide diversity was analyzed using the DnaSP program (ROZAS and ROZAS 1999 ). Levels of nucleotide diversity were estimated as mean pairwise differences (). The TAJIMA 1989 and FU and LI 1993 tests for selection were conducted without specifying an outgroup. For the TFL1 promoter, Tajima's D was estimated using nucleotide and indel polymorphism data, with the latter coded as single characters. Indels in an interrupted microsatellite near the TFL1 translation start were excluded from the analysis. Significance of Tajima's D was determined in simulations with 10,000 runs using the number of segregating sites (s = 46) and the recombination parameter (R = 2.3) estimated from the data. Contingency tests for independence of mutational categories, referred to as the McDonald-Kreitman test (MCDONALD and KREITMAN 1991 ), were conducted using Fisher's exact test to evaluate significance. The Hudson-Kreitman-Aguade (HKA) two-locus and multilocus tests (HUDSON et al. 1987 ) were conducted using DnaSP (ROZAS and ROZAS 1999 ) and the HKA program available from J. Hey.

	RESULTS AND DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Nucleotide variation in the coding regions of the A. thaliana floral developmental genes:
A total of 15 APETALA1, 15 LEAFY, and 14 TERMINAL FLOWER1 alleles were isolated from a collection of A. thaliana ecotypes, sampled primarily from Europe. Allele sequences from the entire coding region and a portion (1 kb) of the promoter and 5'-untranslated region (5'-UTR) were obtained for each gene. Approximately 4.7 kb was sequenced for each AP1 allele, spanning exons 1–8 and including 1.2 kb of the 5'-untranslated region and the promoter (Fig 2). Approximately 3.9 kb of sequence was obtained from LFY alleles. The LFY sequences include the entire coding region (from exons 1–3 and intervening introns) and 1.1 kb of the 5'-UTR and promoter (Fig 3). About 1.8 kb of the TFL1 allele sequence was obtained, including exons 1–4 and 0.7 kb of promoter and 5'-UTR (Fig 4). The AP1, LFY, and TFL1 genes encode proteins of 255, 424, and 177 amino acids in length, respectively.

View larger version (39K): In this window In a new window Download PPT slide   Figure 2. Sequence of APETALA1 alleles from different A. thaliana ecotypes. All allele sequences are compared to the reference allele from the Burghaun/Rhon-0 Bu-0) ecotype. The position of the polymorphic sites and their locations in introns and exons are indicated at the top. Numbers within indels indicate a repeated nucleotide; the repeated nucleotide at position 2353 is T. Insertions (+) at positions 3663 and 4024 are CTAT and ATA, respectively.

View larger version (36K): In this window In a new window Download PPT slide   Figure 3. Sequence of LEAFY alleles from different A. thaliana ecotypes. All allele sequences are compared to the reference allele from the Burghaun/Rhon-0 (Bu-0) ecotype. The position of the polymorphic sites and their locations in introns and exons are indicated at the top. Insertions (+) at positions 599 and 3871 are AAAA and ATGGCTAATTTGTT, respectively. Polymorphisms not shown: position 576, multiple A repeats; position 1082, multiple GA repeats.

View larger version (29K): In this window In a new window Download PPT slide   Figure 4. Sequence of TERMINAL FLOWER1 alleles from different A. thaliana ecotypes. All allele sequences are compared to the reference allele from the Columbia (Col-0) ecotype. The position of the polymorphic sites and their locations in introns and exons are indicated at the top. Insertions (+) are as follows: position 225, TATATAGG; position 308, TTGAAACCATG; position 350, TGTTGCTGAAGAAGTCAA; position 458, AGCTT; position 485, AGTATTAGAAA; position 510, TACG; position 518, CA; position 552, AA; position 565, multiple GA repeats.

These three A. thaliana floral developmental genes show different levels of coding region nucleotide polymorphism. In AP1, a total of 91 polymorphic nucleotide sites are in the 3.5-kb coding region of this gene (Fig 2). Of these, 10 exon polymorphisms are replacements and 8 are synonymous changes; 73 polymorphisms are found in introns. There are also 8 insertion/deletion (indel) polymorphisms, all in introns, which range in size from 1 to 5 bp. The LEAFY alleles have a total of 20 nucleotide polymorphisms within the 2.9-kb coding region (Fig 3). There are 4 replacement and 2 synonymous polymorphisms in exon sequences, while 14 segregating sites are found in introns. Eight indel polymorphisms occur in LEAFY; like AP1, all of these are found in intron regions. Finally, the sampled alleles from the TFL1 locus have 6 polymorphic nucleotide sites, including 1 site with two mutations, in the 1.1-kb region that spans the exons and introns. Of these polymorphisms, 3 are replacements, 1 is a synonymous polymorphism, and 3 are found in intron sequences. Three indel polymorphisms are observed in the TFL1 coding region, all in introns.

Estimates of silent nucleotide variation within the coding region of the APETALA1 locus are comparable to the three other floral homeotic genes previously studied (PURUGGANAN and SUDDITH 1998 , PURUGGANAN and SUDDITH 1999 ; see Table 2 and Fig 5). The estimate of AP1 species-wide nucleotide diversity for coding region silent sites, , is 0.0047. This level of variation is slightly lower than that of CAL ( = 0.0069), which is a recent duplicate of AP1; both possess redundant meristem identity functions in flower development. The level of silent site coding region variation for LFY, however, is 0.0019 (see Table 2 and Fig 5), which is less than half that of AP1. An even lower level of silent variation is observed in the TFL1 coding region, with an estimate of for silent sites at 0.0007.

View larger version (11K): In this window In a new window Download PPT slide   Figure 5. Comparison of silent site nucleotide diversity levels in the coding regions of genes in the A. thaliana floral developmental pathway. The genes are arranged from earlier acting (left) to later acting (right) genes.

Excess of low-frequency and replacement polymorphisms at the AP1 transcriptional unit:
The frequency distribution of polymorphisms provides information on the relative roles of neutral drift vs. selection at a locus. The skewness of frequency distributions for nucleotide polymorphisms can be evaluated with both the TAJIMA 1989 and FU and LI 1993 tests for selection. Since A. thaliana may have experienced a recent population expansion, these two tests should be interpreted with caution when inferring selection. However, they may still provide information on deviations in molecular diversity patterns from predictions of the neutral-equilibrium model. When the nucleotide diversity in the AP1 coding region is examined, the distribution of polymorphic sites is significantly skewed toward rare alleles. Seventy-three of the 92 nucleotide polymorphisms occur in only a single allele (singleton mutations). The Tajima test statistic, D, is -2.101 for AP1 (P < 0.05); the negative value of this statistic indicates that sampled alleles show an excess of rare polymorphisms over that expected in a population at mutation-drift equilibrium (Table 2). The Fu and Li test statistic, D*, is also significantly negative for this gene (, P < 0.02), again indicating an excess of singletons. This excess of low-frequency polymorphisms for AP1 is similar to that observed for several other Arabidopsis nuclear genes, including the floral organ identity genes AP3 and PI, as well as the AP1 paralogue CAL (PURUGGANAN and SUDDITH 1998 , PURUGGANAN and SUDDITH 1999 ; see Table 2). This pattern of variation may reflect the inbreeding associated with this selfing plant and/or population disequilibrium caused by the rapid post-Pleistocene range expansion in this species (PURUGGANAN and SUDDITH 1999 ; KUITTINEN and AGUADE 2000 ; AGUADE 2001 ).

Selective sweeps at the LFY and TFL1 transcriptional units:
Like AP1, the TFL1 locus also possesses an excess of low-frequency polymorphisms in its coding region. The Tajima test statistic, D, is -2.032 for this gene (P < 0.05; see Table 2); the Fu and Li test statistic D* is also significantly negative (, P < 0.02). Negative values for Tajima and Fu and Li test statistics are also observed for LFY; this excess of rare alleles, however, is not significantly different from the distribution expected under a neutral-equilibrium model (, P > 0.05; , P > 0.05; Table 3).

The excess of low-frequency polymorphisms observed at TFL1 (and to some extent LFY) is associated with a reduction in overall levels of nucleotide variation relative to other genes in the floral developmental pathway. Whereas the mean silent-site nucleotide diversity for the coding regions of these two genes is 0.0013, the mean value for the four other genes examined here is 0.0063. Compared to 15 previously examined Arabidopsis nuclear loci, TFL1 and LFY have a 5- to 10-fold reduction in polymorphism (KAWABE et al. 1997 ; MIYASHITA et al. 1998 ; KAWABE and MIYASHITA 1999 ; KAWABE et al. 2000 ; KUITTINEN and AGUADE 2000 ; SAVOLAINEN et al. 2000 ; AGUADE 2001 ; MIYASHITA 2001 ).

A multilocus HKA test comparing intraspecific polymorphism with interspecific divergence in the coding regions of all six floral developmental genes is significant (, P < 0.035), with LFY and TFL1 contributing most to the deviation from neutral expectations (Fig 6). This pattern suggests that the evolutionary dynamics of the coding regions of these two regulatory genes differ significantly from the other genes of the floral developmental pathway. Specifically, the patterns of evolution for the coding regions of TFL1 and LFY suggest a recent selective sweep, either at these loci or at closely linked genes. Of 15 genes in A. thaliana analyzed thus far, only one other locus—CHALCONE ISOMERASE—shows evidence of a recent selective sweep (KUITTINEN and AGUADE 2000 ).

View larger version (13K): In this window In a new window Download PPT slide   Figure 6. Contribution of each gene in the A. thaliana floral developmental pathway to the 2 statistic for the multilocus HKA test of selection on coding region sequences. The total 2 is significantly greater than expected (P < 0.035) in the multilocus HKA.

Patterns of regulatory protein evolution:
An excess of intraspecific replacement polymorphisms has previously been documented for the floral developmental genes AP3, PI, and CAL (see Table 3), as well as several other nuclear loci (KAWABE et al. 1997 ; PURUGGANAN and SUDDITH 1998 , PURUGGANAN and SUDDITH 1999 ). While not observed at either AP1 or LFY (Table 3), this pattern is found at TFL1. This inflorescence architecture gene has three replacement polymorphisms but only a single synonymous polymorphism within A. thaliana, compared to three replacements and 13 synonymous changes fixed between A. thaliana and A. lyrata (Table 3). Nonetheless, this observed elevation in within-species replacement polymorphism is not statistically significant (McDonald-Kreitman test, P = 0.0609).

Patterns of replacement and synonymous variation at the six floral genes examined here suggest that many of the nonsynonymous polymorphisms may be slightly deleterious. A hierarchical Bayesian model of protein evolution indicates negative selection intensities against amino acid replacements for all of these floral developmental genes except LFY, a pattern consistent with that found in the majority of A. thaliana nuclear loci (BUSTAMANTE et al. 2002 ). This prevalence of deleterious mutations may reflect the effect of inbreeding in this predominantly selfing species (BUSTAMANTE et al. 2002 ).

Promoter variation in floral meristem identity and inflorescence architecture genes:
Comparison of promoter/5'-UTR regions of AP1, LFY, and TFL1 using a multilocus HKA test does not indicate statistically significant differences in patterns of evolution among these three genes (, P > 0.4). However, examination of patterns and levels of nucleotide variation within each of these genes suggests that they do differ in their evolutionary dynamics. For two of the genes, AP1 and LFY, nucleotide variation in the promoter/5'-UTR is very similar to that observed in the coding region of the gene. Individual HKA tests for both of these loci indicate no significant difference between the promoter/5'-UTR region and coding region in levels of within-species polymorphism vs. between-species divergence (Table 4). These patterns suggest that for AP1 and LFY, evolutionary forces have acted comparably in the two portions of the gene.

In contrast to AP1 and LFY, TFL1 shows a dramatic difference in nucleotide variation between its promoter/5'-UTR and coding region. Whereas nucleotide diversity in the coding region is extremely low (; see above), that of the promoter/5'-UTR is very high (; Table 4). The HKA test confirms that these two portions of the gene differ in their patterns of evolution (, P < 0.012). Moreover, whereas Tajima's D and Fu and Li's D* estimates are negative for the promoter/5'-UTR and coding regions of AP1 and LFY as well as for the coding region of TFL1, these two polymorphism measures are positive for the TFL1 promoter/5'-UTR (see Table 2 and Table 4). Thus, the promoter/5'-UTR and coding region of TFL1 have differed markedly in their patterns of evolution.

Nucleotide variation in the TFL1 promoter characterizes two distinct classes of alleles (haplogroups), which are distinguished by 20 nucleotide polymorphisms and 10 indels (Fig 4). This pattern of allelic dimorphism is not unprecedented in A. thaliana, having been reported in several loci including FAH (AGUADE 2001 ) and ChiB (KAWABE and MIYASHITA 1999 ). The origins and evolutionary forces maintaining such allelic dimorphism are still unclear (KAWABE and MIYASHITA 1999 ; AGUADE 2001 ). What distinguishes the TFL1 promoter sequence dimorphism from other genes, however, is (i) the close proximity of the dimorphic region to sequences that appear to have undergone a recent selective sweep and (ii) the significantly positive value of Tajima's D. Both of these observations are consistent with selection to maintain TFL1 promoter dimorphism in the face of selective sweeps in the adjacent coding region of this gene.

Variation of TFL1 is expected to affect the number of floral meristems established by the shoot apical meristem. Indeed, in a sample of 78 ecotypes, there is a difference in the number of main axis flowers produced in plants that have TFL1 promoter haplogroups A and B—14.29 ± 3.93 and 16.08 ± 3.86, respectively (our unpublished observations). This difference is only marginally nonsignificant after correcting for the effects of population structure (P < 0.064).

The molecular population genetics of the floral developmental pathway:
We have examined the molecular population genetics of six regulatory genes that are found at different positions in the floral developmental pathway. These loci have various roles in floral development, including the maintenance of inflorescence meristem identity (TFL1), the specification of floral meristem identity (LFY, AP1, and CAL), and the development of sepals, petals, and stamens (AP1, AP3, and PI). On the basis of the broad functional classes, one gene is an inflorescence architecure gene (TFL1), two are categorized as floral meristem identity genes (LFY and CAL), while two are floral organ identity loci (AP3 and PI). AP1 is both a meristem specification and a floral patterning gene.

Two of these loci show evidence of having experienced a recent adaptive sweep. These are the inflorescence architecture gene TFL1 and the floral meristem gene LFY, both of which are early acting genes that regulate the identities of inflorescence and floral meristems. Coding regions of both of these loci show reduced levels of polymorphism compared to other floral developmental loci, consistent with recent positive selection. It is unclear whether these loci were the actual targets of selection; it is possible that the selective target may have been a closely linked gene (BARTON 2000 ). However, the chromosomal locations of these genes do not suggest that they would be tightly linked to other loci; comparison of physical and genetic map distances in the regions flanking both of these loci indicates that they are not regions of low recombination (K. M. OLSEN and M. D. PURUGGANAN, unpublished observation). Moreover, the data on TFL1 indicate that the selective sweep was confined to the coding region and did not include the promoter sequences, which suggests that recombination was sufficient to restrict the physical impact of the selective sweep.

In addition to selective sweeps, other selective forces are also evident in these floral genes. At TFL1, the two promoter haplogroups occur at roughly equal frequency in wild ecotypes and haplogroup differentiation does not extend to the 5' proximal gene rpS28 (our unpublished observations). Moreover, the TFL1 promoter haplogroup types are weakly associated with the developmental decision to form flowers. The allele structure of TFL1 appears to have been shaped by two contrasting selective forces: an adaptive sweep in the coding region and selection to maintain variation in the promoter/5'-UTR. At the meristem identity gene CAL, genetic analysis indicates that naturally occurring alleles differ in their ability to regulate floral meristem identity specification; however, unambiguous evidence for positive or diversifying selection has not been observed for this locus (PURUGGANAN and SUDDITH 1998 ). In contrast to LFY, TFL1, and CAL, the three genes with exclusively organ identity functions (AP1, AP3, and PI) show no evidence of either positive or balancing selection (see also PURUGGANAN and SUDDITH 1999 ).

The selective forces inferred for these genes are consistent with our predictions based on patterns of morphological evolution. Organ identity genes (AP1, AP3, and PI), which control evolutionarily conserved floral organ traits, show no evidence of adaptive evolution. The significant excess of rare polymorphisms in these three genes, as reflected in the negative Tajima's D estimate for these loci, appear to result from the persistence of deleterious mutations in Arabidopsis nuclear genes (BUSTAMANTE et al. 2002 ). In contrast, the inflorescence architecture and meristem identity genes (TFL1 and LFY, respectively), which control traits associated with inflorescence morphology in A. thaliana and its allies, bear molecular signatures suggesting adaptive selection. Both TFL1 and LFY are positioned at critical points in the developmental pathway, integrating signals from multiple genes and regulating developmental decisions on the initiation of the floral developmental program (BRADLEY et al. 1997 ; LILJEGREN et al. 1999 ). This suggests that the adaptive evolution of inflorescence morphology may arise through selection on the timing and spatial patterning of flower and inflorescence branch formation. It will be interesting to determine how selection has operated even earlier than TFL1 in the floral developmental pathway, including those genes that specify natural variation in flowering time (JOHANSON et al. 2000 ).

These results suggest that detailed understanding of the organization of developmental gene pathways and the precise functional roles and interactions of their component loci may provide information on the patterns of diversification of genes that control morphogenesis. Evolutionary analyses of enzymatic pathways, such as those of the glycolytic pathways in Drosophila (EANES 1999 ; VERRELLI and EANES 2001 ) and the anthocyanin biosynthesis pathway in Ipomoea (RAUSHER et al. 1999 ), have already provided insights into the position- and function-dependent action of selective forces. In the study of Drosophila glycolytic enzymes, evolutionary analysis was informed by metabolic control theory, which provides a theoretical basis for advancing hypotheses on the evolution of pathway components (EANES 1999 ). A similar theoretical framework is possible for developmental gene networks (MENDOZA et al. 1999 ; HASTY et al. 2001 ), although quantitative models permitting precise evolutionary hypotheses have yet to be advanced. Nevertheless, the ability to correlate patterns of adaptive evolution with the functional roles of regulatory genes in a developmental gene network allows greater understanding of the mechanisms of morphological evolution.

	ACKNOWLEDGMENTS

The authors thank K. A. Shepard for insightful discussions and members of the Purugganan laboratory for a critical reading of this manuscript. This work was funded with a grant from the National Science Foundation to M.D.P., T. F. C. Mackay, and J. Schmitt and an Alfred P. Sloan Foundation Young Investigator Award in Molecular Evolution to M.D.P.

Manuscript received October 11, 2001; Accepted for publication January 25, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

AGUADE, M., 2001  Nucleotide sequence variation at two genes of the phenylpropanoid pathway, the FAH and F3H genes in Arabidopsis thaliana.. Mol. Biol. Evol. 18:1-9[Abstract/Free Full Text].

ARNONE, M. I. and E. H. DAVIDSON, 1997  The hardwiring of development: organization and function of genomic regulatory systems. Development 124:1851-1864[Abstract].

AUSUBEL, F., 1992 Short Protocols in Molecular Biology. John Wiley & Sons, New York.

BARTON, N. H., 2000  Genetic hitchhiking. Philos. Trans. R. Soc. London B Biol. Sci. 355:1553-1562[Medline].

BOWMAN, J. L., J. ALVAREZ, D. WEIGEL, E. M. MEYEROWITZ, and D. R. SMYTH, 1993  Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development 119:721-743[Abstract].

BRADLEY, D., O. RATCLIFFE, C. VINCENT, R. CARPENTER, and E. S. COEN, 1997  Inflorescence commitment and architecture in Arabidopsis. Science 275:80-83[Abstract/Free Full Text].

BUSTAMANTE, C. D., R. NIELSEN, S. A. SAWYER, K. M. OLSEN, and M. D. PURUGGANAN et al., 2002  The cost of inbreeding in Arabidopsis. Nature in press.

CRONK, Q. C. B., 2001  Plant evolution and development in a post-genomic context. Nat. Rev. Genet. 2:607-619[Medline].

DAVIDSON, E. H., 1999  A view from the genome: spatial control of transcription in sea urchin development. Curr. Opin. Genet. Dev. 9:530-541[Medline].

DOEBLEY, J. and L. LUKENS, 1998  Transcriptional regulators and the evolution of plant form. Plant Cell 10:1075-1082[Free Full Text].

EANES, W. F., 1999  Analysis of selection on enzyme polymorphisms.. Annu. Rev. Ecol. Syst. 30:301-326.

ENDRESS, P. K., 1992  Evolution and floral diversity—the phylogenetic surroundings of Arabidopsis and Antirrhinum.. Int. J. Plant Sci. 153:S106-S122.

FU, Y.-X. and W. H. LI, 1993  Statistical tests of neutrality of mutations. Genetics 133:693-709[Abstract].

GOTO, K. and E. M. MEYEROWITZ, 1994  Function and regulation of the Arabidopsis floral homeotic gene PISTILLATA.. Genes Dev. 13:1548-1560.

GUSTAFSON-BROWN, C., B. SAVIDGE, and M. F. YANOFSKY, 1994  Regulation of the Arabidopsis floral homeotic gene APETALA1.. Cell 76:131-143[Medline].

HAMBLIN, M. T. and C. F. AQUADRO, 1997  Contrasting patterns of nucleotide sequence variation at the Gld locus in different populations of Drosophila melanogaster.. Genetics 145:1053-1062[Abstract].

HASTY, J., D. MCMILLEN, F. ISAACS, and J. J. COLLINS, 2001  Computational studies of gene regulatory networks: in numero molecular biology. Nat. Rev. Genet. 2:268-279[Medline].

HUDSON, R. R., M. KREITMAN, and M. AGUADE, 1987  A test of neutral molecular evolution based on nucleotide data. Genetics 116:153-159[Abstract/Free Full Text].

JACK, T., 2001  Relearning our ABCs: new twists on an old model. Trends Plant. Sci. 6:310-316[Medline].

JACK, T., L. L. BROCKMANN, and E. M. MEYEROWITZ, 1992  The homeotic gene APETALA3 of Arabidopsis thaliana encodes a MADS-box and is expressed in petals and stamens. Cell 68:683-697[Medline].

JOHANSON, U., J. WEST, C. LISTER, S. MICHAELS, and R. AMASINO et al., 2000  Molecular analysis of FRIGIDA, a major determinant of natural variation in Arabidopsis flowering time. Science 290:344-347[Abstract/Free Full Text].

KAWABE, A. and N. T. MIYASHITA, 1999  DNA variation in the basic ChiB region of the wild plant Arabidopsis thaliana.. Genetics 153:1445-1453[Abstract/Free Full Text].

KAWABE, A., H. INNAN, R. TERAUCHI, and N. T. MIYASHITA, 1997  Nucleotide polymorphism in the acidic chitinase locus (ChiA) region of the wild plant Arabidopsis thaliana.. Mol. Biol. Evol. 14:1303-1315[Abstract].

KAWABE, A., K. YAMANE, and N. T. MIYASHITA, 2000  DNA polymorphism at the cytosolic PGI locus of the wild plant Arabidopsis thaliana.. Genetics 156:1339-1347[Abstract/Free Full Text].

KEMPIN, S. A., B. SAVIDGE, and M. F. YANOFSKY, 1995  Molecular basis of the cauliflower phenotype in Arabidopsis.. Science 267:522-525[Abstract/Free Full Text].

KRAMER, E., R. DORIT, and V. IRISH, 1998  Molecular evolution of genes controlling petal and stamen development: duplication and divergence within the APETALA3 and PISTILLATA MADS-box gene lineages. Genetics 149:765-783[Abstract/Free Full Text].

KUITTINEN, H. and M. AGUADE, 2000  Nucleotide variation at the CHALCONE ISOMERASE locus in Arabidopsis thaliana.. Genetics 155:863-872[Abstract/Free Full Text].

LAWTON-RAUH, A., E. ALVAREZ-BUYLLA, and M. D. PURUGGANAN, 2000  Molecular evolution of flower development. Trends Ecol. Evol. 15:144-149[Medline].

LILJEGREN, S., C. GUSTAFSON-BROWN, A. PINYOPICH, G. S. DITTA, and M. F. YANOFSKY, 1999  Interactions among the APETALA1, LEAFY and TERMINAL FLOWER1 specify meristem fate. Plant Cell 11:1007-1018[Abstract/Free Full Text].

LUDWIG, M. Z., C. BERGMAN, N. H. PATEL, and M. KREITMAN, 2000  Evidence for stabilizing selection in a eukaryotic enhancer element. Nature 403:564-567[Medline].

MANDEL, M. A., C. GUSTAFSON-BROWN, B. SAVIDGE, and M. F. YANOFSKY, 1992  Molecular characterization of the Arabidopsis floral homeotic gene APETALA1.. Nature 360:273-277[Medline].

MCDONALD, J. and M. KREITMAN, 1991  Adaptive protein evolution at the Adh locus in Drosophila. Nature 351:652-654[Medline].

MENDOZA, L., D. THIEFFRY, and E. R. ALVAREZ-BUYLLA, 1999  Genetic control of flower morphogenesis in Arabidopsis thaliana: a logical analysis. Bioinformatics 15:593-606[Abstract/Free Full Text].

MIYASHITA, N., 2001  DNA variation in the 5' upstream region of the Adh locus of the wild plants Arabidopsis thaliana and Arabis gemmifera.. Mol. Biol. Evol. 18:164-171[Abstract/Free Full Text].

MIYASHITA, N. T., A. KAWABE, H. INNAN, and R. TERAUCHI, 1998  Intra- and interspecific DNA variation and codon bias of the alcohol dehydrogenase (Adh) locus in Arabis and Arabidopsis species. Mol. Biol. Evol. 15:1420-1429[Free Full Text].

NG, M. and M. F. YANOFSKY, 2000  Three ways to learn the ABCs.. Curr. Opin. Plant Biol. 3:47-52[Medline].

PURUGGANAN, M. D., 1997  The MADS-box floral homeotic gene lineages predate the origin of seed plants: phylogenetic and molecular clock estimates. J. Mol. Evol. 45:392-396[Medline].

PURUGGANAN, M. D., 1998  The molecular evolution of development. Bioessays 20:700-711[Medline].

PURUGGANAN, M. D., 2000  The molecular population genetics of regulatory genes. Mol. Ecol. 9:1451-1461[Medline].

PURUGGANAN, M. D. and J. SUDDITH, 1998  Molecular population genetics of the Arabidopsis CAULIFLOWER regulatory gene: non-neutral evolution and wild variation in floral homeotic function. Proc. Natl. Acad. Sci. USA 95:8130-8134[Abstract/Free Full Text].

PURUGGANAN, M. D. and J. I. SUDDITH, 1999  Molecular population genetics of floral homeotic loci: departures from the equilibrium-neutral model at the APETALA3 and PISTILLATA genes of Arabidopsis thaliana. Genetics 151:839-848[Abstract/Free Full Text].

RATCLIFFE, O. J., I. AMAYA, C. A. VINCENT, S. ROTHSTEIN, and R. CARPENTER et al., 1998  A common mechanism controls the life cycle and architecture of plants. Development 125:1609-1615[Abstract].

RATCLIFFE, O. J., D. J. BRADLEY, and E. S. COEN, 1999  Separation of shoot and floral identity in Arabidopsis.. Development 126:1109-1120[Abstract].

RAUSHER, M. D., R. E. MILLER, and P. TIFFIN, 1999  Patterns of evolutionary rate variation among genes of the anthocyanin biosynthetic pathway. Mol. Biol. Evol. 16:266-274[Abstract].

ROZAS, J. and R. ROZAS, 1999  DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174-175[Abstract/Free Full Text].

SAMACH, A., S. E. KOHALMI, P. MOTTE, R. DATLA, and G. W. HAUGHN, 1997  Divergence of function and regulation of class B floral organ identity genes. Plant Cell 9:559-570[Abstract].

SAVOLAINEN, O., C. H. LANGLEY, B. P. LAZZARO, and H. FREVILLE, 2000  Contrasting patterns of nucleotide polymorphisms at the Adh locus in the outcrossing Arabidopsis lyrata and the selfing Arabidopsis thaliana.. Mol. Biol. Evol. 17:645-655[Abstract/Free Full Text].

SHANNON, S. and D. R. MEEKSWAGNER, 1991  A mutation in the Arabidopsis TFL1 gene affects inflorescence meristem development. Plant Cell 3:877-892[Abstract/Free Full Text].

SHU, G. P., W. AMARAL, L. C. HILEMAN, and D. A. BAUM, 2000  LEAFY and the evolution of rosette flowering in violet cress. Am. J. Bot. 87:634-641[Abstract/Free Full Text].

SHUBIN, N., C. TABIN, and S. CARROLL, 1997  Fossils, genes and the evolution of animal limbs. Nature 388:639-648[Medline].

SWOFFORD, D., 1992 Phylogenetic Analysis Using Parsimony 3.1. Illinois Natural History Survey, Champaign, IL.

TAJIMA, F., 1989  Statistical methods for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585-595[Abstract/Free Full Text].

VERRELLI, B. C. and W. F. EANES, 2001  Clinal variation for amino acid polymorphisms at the Pgm locus in Drosophila melanogaster.. Genetics 157:1649-1663[Abstract/Free Full Text].

WANG, R. L., A. STEC, J. HEY, L. LUKENS, and J. DOEBLEY, 1999  The limits of selection during maize domestication. Nature 398:236-239[Medline].

WEIGEL, D., 1995  The genetics of flower development: from floral induction to ovule morphogenesis. Annu. Rev. Genet. 29:19-39[Medline].

WEIGEL, D. and E. M. MEYEROWITZ, 1993  Activation of floral hometic genes in Arabidopsis.. Science 261:1723-1726.

WEIGEL, D., J. ALVAREZ, D. R. SMYTH, M. F. YANOFSKY, and E. M. MEYEROWITZ, 1992  LEAFY controls floral meristem identity in Arabidopsis.. Cell 69:843-859[Medline].

YANOFSKY, M. F., 1995  Floral meristems to floral organs—genes controlling early events in Arabidopsis flower development. Annu. Rev. Plant Physiol. Plant Mol. Biol. 46:167-188.

This article has been cited by other articles:

				E. G. Bakker, M. B. Traw, C. Toomajian, M. Kreitman, and J. Bergelson Low Levels of Polymorphism in Genes That Control the Activation of Defense Response in Arabidopsis thaliana Genetics, April 1, 2008; 178(4): 2031 - 2043. [Abstract] [Full Text] [PDF]

				D. L. Filiault, C. A. Wessinger, J. R. Dinneny, J. Lutes, J. O. Borevitz, D. Weigel, J. Chory, and J. N. Maloof Amino acid polymorphisms in Arabidopsis phytochrome B cause differential responses to light PNAS, February 26, 2008; 105(8): 3157 - 3162. [Abstract] [Full Text] [PDF]

				T. Hernandez-Hernandez, L. P. Martinez-Castilla, and E. R. Alvarez-Buylla Functional Diversification of B MADS-Box Homeotic Regulators of Flower Development: Adaptive Evolution in Protein-Protein Interaction Domains after Major Gene Duplication Events Mol. Biol. Evol., February 1, 2007; 24(2): 465 - 481. [Abstract] [Full Text] [PDF]

				I. Simko, K. G. Haynes, and R. W. Jones Assessment of Linkage Disequilibrium in Potato Genome With Single Nucleotide Polymorphism Markers Genetics, August 1, 2006; 173(4): 2237 - 2245. [Abstract] [Full Text] [PDF]

				S. C. Gonzalez-Martinez, E. Ersoz, G. R. Brown, N. C. Wheeler, and D. B. Neale DNA Sequence Variation and Selection of Tag Single-Nucleotide Polymorphisms at Candidate Genes for Drought-Stress Response in Pinus taeda L. Genetics, March 1, 2006; 172(3): 1915 - 1926. [Abstract] [Full Text] [PDF]

				K. J. Schmid, S. Ramos-Onsins, H. Ringys-Beckstein, B. Weisshaar, and T. Mitchell-Olds A Multilocus Sequence Survey in Arabidopsis thaliana Reveals a Genome-Wide Departure From a Neutral Model of DNA Sequence Polymorphism Genetics, March 1, 2005; 169(3): 1601 - 1615. [Abstract] [Full Text] [PDF]

				R. C. Moore, S. R. Grant, and M. D. Purugganan Molecular Population Genetics of Redundant Floral-Regulatory Genes in Arabidopsis thaliana Mol. Biol. Evol., January 1, 2005; 22(1): 91 - 103. [Abstract] [Full Text] [PDF]

K. K. Shimizu, J. M. Cork, A. L. Caicedo, C. A. Mays, R. C. Moore, K. M. Olsen, S. Ruzsa, G. Coop, C. D. Bustamante, P. Awadalla, et al. Darwinian Selection on a Selfing Locus Science, December 17, 2004; 306(5704): 2081 - 2084. [Abstract]