The Plant Cell, Vol. 7, 1635-1644, October 1995 O 1995 American Society of Plant Physiologists

Conservation of Floral Homeotic Gene Function between Arabidopsis and Antirrhinum

Vivian F. Irish’ and Yuri T. Yamamoto2

Department of Biology, Osborn Memorial Laboratory, P.O. Box 208104, Yale University, New Haven, Connecticut 06520-8104

Severa1 homeotic genes controlling floral development have been isolated in both Antirrhinum and Arabidopsis. Based on the similarities in sequence and in the phenotypes elicited by mutations in some of these genes, it has been proposed that the regulatory hierarchy controlling floral development is comparable in these two species. We have performed a direct experimental test of this hypothesis by introducing a chimeric Antirrhinum Deficiens (DefA)/Arabidopsis APETALA3 (AP3) gene, under the control of the Arabidopsis AP3 promoter, into Arabidopsis. We demonstrated that this transgene is sufficient to partially complement severe mutations at the AP3 locus. In combination with a weak ap3 mutation, this transgene is capable of completely rescuing the mutant phenotype to a fully functional wild-type flower. These observa- tions indicate that despite differences in DNA sequence and expression, DefA coding sequences can compensate for the loss of AP3 gene function. We discuss the implications of these results for the evolution of homeotic gene function in flowering plants.

INTRODUCTION

Arabidopsis (Brassicaceae) and Antirrhinum (Scrophularia- ceae) are widely divergent dicotyledonous species and belong to different subclasses. Based on morphological criteria and evidence from the fossil record, these species are considered to have diverged at least 70 million years ago (Cronquist, 1981, 1988; Muller, 1981). Although both species have the typical dicot arrangement of concentric whorls of floral organs, both the number and forms of these organs are quite different. The radially symmetric Arabidopsis flower, from outer to inner whorls, consists of four sepals, four petals, six stamens, and two fused carpels that form the pistil (Figures 1C and 1D). An- tirrhinum flowers are subtended by a leaflike bract and consist of five sepals, five petals which are fused at the base, four sta- mens (a fifth stamen primordium aborts early in development), and two fused carpels that form the pistil (Figures 1A and 1B). Antirrhinum flowers are zygomorphic, and this bilateral sym- metry is especially evident in the different shapes of the upper and lower petals and in the location of the aborted stamen.

Floral organ identities appear to be established by the ac- tion of similar homeotic genes in both of these species. Genetic and molecular analyses of these homeotic genes have provided the basis for the ABC model of floral development (Coen and Meyerowitz, 1991; Meyerowitz et al., 1991). In both species, mutations in the floral homeotic genes generally affect two ad- jacent whorls of organs; thus, three domains of function can be postulated. The A function is required for the development

’ To whom correspondence should be addressed.

State University, Raleigh, NC 27695. Current address: Department of Genetics, Box 7614, North Carolina

of sepals and petals, the B function for peta1 and stamen development, and the C function for stamen and carpel de- velopment. The ABC model postulates that floral organ identity is specified by the combination of A, B, or C homeotic gene functions expressed in a particular whorl. In addition, the A and C functions are postulated to regulate negatively each other’s expression. Based on similarities in DNA sequence, expression patterns, and phenotypes elicited by mutations in a number of these homeotic genes, it has been suggested that the regulatory capabilities of these homeotic gene products have been conserved through dicot evolution (Coen, 1991).

Homeotic mutations in either the Globosa (Glo) or Deficiens (DefA) loci in Antirrhinum affect the development of petals and stamens (Sommer et al., 1990; Trobner et al., 1992). In both cases, the mutant flowers display transformations of the five second whorl structures into sepaloid organs. In the third whorl, five fused carpelloid structures arise and fuse to form an en- larged gynoecium. 60th the DefA and Glo genes have been cloned, and they encode proteins containing a conserved 58-amino acid region termed the MADS box (Sommer et al., 1990; Trobner et al., 1992). The MADS box-containing domains of transcription factors from severa1 systems have been shown to bind to DNA at a consensus motif, termed the CArG box (Norman et al., 1988; Passmore et al., 1988; Christ and Tye, 1991). In vitro studies have shown that the DefA and Glo gene products can bind to specific CArG motifs as heterodimers (Schwarz-Sommer et al., 1992). One such CArG box is found in the promoter of the DefA gene, suggesting that these MADS box-containing gene products may be autoregulatory (Schwarz-Sommer et al., 1992).

1636 The Plant Cell

B

Figure 1. Antirrhinum and Arabidopsis Wild-Type Flowers.

(A) Wild-type Antirrhinum flower.(B) Diagram of an Antirrhinum flower. Organs are, from outside to inside, the subtending bract, five sepals, five petals, four stamens (a fifth abortedstamen is indicated by an X), and two fused carpels.(C) Wild-type Arabidopsis flower.(D) Diagram of an Arabidopsis flower. Organs are, from outside to inside, four sepals, four petals, six stamens, and two fused carpels.The inflorescence axis is indicated by black dots in (B) and (D).

Mutations in the Arabidopsis APETALA3 (AP3) and PISTIL-LATA (PI) genes also result in the homeotic transformation ofpetals into sepals, and stamens into carpelloid structures(Bowman et al., 1989; Hill and Lord, 1989; Jack et al., 1992).The AP3 gene encodes a MADS box-containing protein witha high degree of similarity with the Antirrhinum DefA gene prod-uct (Jack et al., 1992). Consistent with the mutant phenotype,AP3 transcripts are expressed early in floral development inthe cells that are thought to give rise to petals and stamens(Jack et al., 1992). The PI gene appears to be the homologof the Antirrhinum G/o gene (Goto and Meyerowitz, 1994). Boththe PI and AP3 gene products are required for maintainingexpression of the AP3 gene (Jack et al., 1994). This observa-tion is consistent with the biochemical evidence from theAntirrhinum system showing that the G/o and DefA gene prod-ucts form heterodimers (Schwarz-Sommer et al., 1992).

Despite these many similarities between DefA and APS, an-other member of this orthologous gene family appears to berequired only for petal identity. A null mutation in green petals(gp), a petunia homolog of DefA, affects the development ofpetals but does not affect stamens (Kush et al., 1993; van derKrol et al., 1993). Furthermore, ectopic expression of gp in-duces the formation of petaloid organs in place of sepals(Halfter et al., 1994). One explanation for these observationsis that the regulatory function encoded by the petunia homo-log has diverged from that of Antirrhinum and Arabidopsis.

Here, we describe a direct test of the hypothesis that similarAntirrhinum and Arabidopsis homeotic genes are functionallyconserved. We generated transgenic Arabidopsis plants con-taining a chimeric AP3-DefA coding sequence driven by theArabidopsis APS promoter. We also generated transgenicplants carrying a control construct consisting of the AP3

Conservation of Floral Homeotic Gene Function 1637

pBam21

H H E I , 44 Y//Y////////A k H

RB AP3 promoter ATG AP3 cDNA ter LB -

100 bD

pDef7

H H E , I 4 Y/A\\\\\\\Y k H RB AP3 promoter ATG DefA cDNA ter LB

~

100 bD

Figure 2. Plasmid Constructs Used in This Study

pBam21 contains a 9-kb AP3 genomic region, with the position of the exons and introns within the transcription unit indicated by a jagged arrow. pla19 and pDet7 contain AP3 cDNA and DefA sequences, respec- tively, which are fused to genomic sequences that include the promoter and part of the 5‘ most exon derived from AP3. A nopaline synthase translational terminator (ter) was fused to the 3’ end. AP3 coding se- quences are indicated by rightward hatching, and DefA coding sequences are indicated by leftward hatching. Selected restriction sites are BamHl (E), EcoRl (E), and Hindlll (H). Left border (LB) and right border (RB) sequences of the T-DNA are indicated by arrowheads.

promoter driving the AP3 coding sequence. We observed that the AP3-DefA transgene can partially complement severe mu- tations at the AP3 locus. Furthermore, the AP3-DefA transgene is able to restore completely a weak ap3 mutant phenotype to the wild type. These observations demonstrate that the DefA coding sequences are sufficient to compensate for a muta- tion in the AP3 locus. We propose several models to explain these complementation results and discuss the evolutionary implications of the functional conservation we observed.

RESULTS

Generation of Transgenic Arabidopsis Plants Containing AP3 and AP3-DefA Chimeric Genes

Three constructs were generated in the pBinl9 binary trans- formation vector (Bevan, 1984) and are illustrated in Figure 2. pBam21 contains a 9-kb Arabidopsis genomic insert con- taining the entire AP3 locus, with -6 kb 5’ and 4 . 5 kb 3’to the coding region. pla19 consists of a translational fusion of a 1.9-kb Arabidopsis genomic fragment to an AP3 cDNA at amino acid position 53, followed by a nopaline synthase trans- lational terminator. pDef7 is similar to pla19, except that a DefA

cDNA was fused to the AP3 genomic fragment at the same site. This fusion resulted in a chimeric gene that contains the AP3 promoter, followed by the AP3 MADS box coding region fused to the C-terminal portion of the DefA coding sequence.

Although the coding regions of DefA and AP3 are highly con- served, especially in the MADS box domain, the promoters have diverged significantly (Figure 3). The AP3 promoter does not contain the large inverted repeat found in the DefA pro- moter, nor does it contain the putative MADS box binding site CCTTTTTAGG (Schwarz-Sommer et al., 1992). Instead, avery similar sequence, CCTTTTTGGG, is found at approximately the same position upstream of the translational start site. We identified several other sequences that may serve as poten- tia1 MADS box binding sites in theAP3 promoter and that may function as regulatory elements (Figure 3A; S. Carr, C.D. Day, T. Hill, and V.F. Irish, unpublished data). Within the coding regions, the deduced amino acid sequences of AP3 and DefA are 92% identical (53 of 58 amino acids) through the MADS box domain and only 51% identical (87 of 169 amino acids) beyond the MADS box domain (Figure 36).

Transgenic Arabidopsis plants were generated with all three constructs by using Agrobacterium-mediated transformation of root explants (Valvekens et al., 1988). For each construct, we generated >40 independent transgenic lines and charac- terized three transgenic lines for each construct (Table 1). The number of independently segregating transgene inserts was estimated by progeny tests. Despite the segregation data sug- gesting that all tested lines contained a single locus of insertion, DNA gel blot analyses demonstrated that most lines contained multiple, independent insertions of the transgene (Table 1). The relative leve1 of transgene expression was assessed by RNA gel blot analysis and indicated that transgene expres- sion in each line was roughly correlated with the number of transgene inserts (data not shown).

Transgenic plants were used as the polien donors in crosses to plants homozygous f3r either ap3-7, ap3-3, or ap3-4. These mutations differed in their severity, with ap3-7 plants showing a weaker phenotype and ap3-3 and ap3-4 mutations resulting in more extreme second and third whorl homeotic transfor- mdions (Tables 2 and 3, and Figure 4; Bowman et al., 1989; Jack et al., 1992). The ap3-7 allele is temperature sensitive, and at 22OC, ap3-7 homozygous flowers display four sepals in place of petals and a conversion of the six stamens into car- pelloid stamens (Figures 4B and 5C). The ap3-7 lesion is correlated with a single nucleotide change in the K-box, result- ing in a conversion of lysine-153 to methionine (Jack et al., 1992). The K-box is predicted to form several amphipathic a-helices that may be required for protein-protein interactions (Ma et al., 1991; Pnueli et al., 1991; Schwarz-Sommer et al., 1992). Flowers homozygous for ap3-3 have four sepals in place of petals. The third whorl organs of ap3-3 plants have a vari- able phenotype, ranging from free-standing filaments to carpels fused to the central two fourth whorl carpels (Jack et al., 1992). We have rarely observed unfused carpelloid structures in the third whorl of ap3-3 plants (Tables 2 and 3, and Figure 4C). The phenotype of homozygous ap3-4 flowers is similar to that

1638 The Plant Cell

A

AAGCTTCTTAAGAATTATAGTAGCACTTGTTGATATCGGG GTTCTTTTTTCTTAACATAGGTTTTGGTGGGTAGATCAAC AGAAGACCTCGCTGTGGACATTGATTTGGGAAGAGAAAAG CGGGGTAGCA AAATATCTCG ACGACAGGTC GGTCAATAGT AGATACTTCT ATTTGTATTT TAGGTCTTTA AGTTGTATGA GAAGCAGCAGCCCAGGATCTGTAATGGTTGTTGTGTATGT T T T T T T m m T T A C T G A A T G C A G G C G A T C A T A C G GCTGGGTGATGACGGTTCGTTTCATATAAAAAACCCTGGG GAAGTATTCAATCTCAGTAAATGAGAAGGAAGTAGATCCT GGACAGAGTTTAA'ICCTCAAATCCGATTGTCTAGTTGAGG TTTGATCCGG GATACTCTTT TATGACTCAT TACTTACGTT CTAGGTCCGT ACTTAAACCC GATATATCGT GCGCGCAGAT ACGGGGAATGCCTTTTATATTTGAAACAAACCAAAGTTGC ATGCAAGAGTACCTGAAGAGAAGAGGGAAAGTGAACTGAG ACCAGATCAAGAGTGCGTGTTGTTTGTGTAGATGAATGAT GCATCTCTTGTGTAATGTACCTTAGCCATAGGAACACGTC TTGTAGATCT TTTAATACAT CTTTAGTTCC GCATCATGCA TAGTTGACCCTGTTTTAAGGCGTTGAAATGAAAATACAAG TCTCTTGTATCTGAATTTGTGT'ITTAAGCGAAGAATGATT GTTCTTGTGA AGTTGATACA CAAGTTCTTT GGATATCCTA TCAGTATAAAGGATAGGTTTCCATTTTCGTGACTCACTCA CTGATTTCCATTGCTGAAAATTGATGATGAACTAAGATCA ATCCATGTAGTTCAAACAACAGTAACTGTGCCACTAGTTT GAACAACACTAACTGGTCGAGCAAAAGAAAAAGAGTTCAT CATATATCTGATTTGATGGACTGTTTGGAGTTAGGACCAA ACATTATCTACAAACAAAGACTTTTCTCCTAACTTGTGAT TCCTTCTTAAACCCTAGGGGTAATATTCTATTTTCCAAGG ATCTTTAGTTAAAGGCAAATCCGGGAAATTATTGTAATCA TTTGGGGCCACATATAAAAGATTTGAGTTAGATGGAAGTG ACGATTAATCCAAACATATATATCTCTTTCTTCTTATTTC CCAAATTAACAGACAAAAGTAGAATATTGGCTTTTAACAC CAATATAAAA ACTTGCTTCA CACCTAAACA CTTTTGTTTA CTTTAGGGTAAGTGCAAAAAGCCAACCAAAWCACCTGCA CTGATTTGACGTTTACAAACGCCGTTAAGTTTGTCACCGT CTAAACAAAAACAAAGTAGAAGCTAACGGAGCTCCGTTAA TAAATTGACGAAAAGCAAACCAAGTTTTTAGCTTTGGTCC CCCTCTTTTACCAAGTGACAATTGATTTAAGCAGTGTCTT GTAATTATACAACCATCGATGTCCGTTGATTTAAACAGTG TCTTGTAATT AAAAAAATCA G T T T W =AAAATT TATCACTTAGTTTTCATCAACTTCTGAACTTA- G A T T A G G C A A T A C T T T D m T A A CTCAAGTGGA CCCTTTACTTCTTCAACTCCATCTCTCTCTTTCTATTTCA CTTCTTTCTTCTCATTATATCTCTTGTCCTCTCCACCAAA TCTCTTCAACAAAAAGATTAAACAAAGAGAGAAGAATATG

B

DefA

A??3

-1682 -1642 -1602 -1562 -1522 -1482 -1442 -1402 -1362 -1322 -1282 -1242 -1202 -1162 -1122 -1082 -1042 -1002 -982 -922 -882 -842 -802 -762 -722 -682 -642 -602 -562 -522 -482 -442 -402 -3 62 -322 -2 82 -2 42 -2 02 -162 -122 -82 -42 -2 +39

of ap3-3 (Tables 2 and 3, and Figure 4D). The ap3-3 and ap3-4 mutations appear to be caused by single nucleotide changes resulting in stop codons within the MADS box and presum- ably abolish function (Jack et al., 1992).

The F1 plants heterozygous for both an ap3 allele and a transgene were allowed to self-fertilize, and kanamycin-resis- tant progeny were recovered. For each transgenic construct, three independent lines were analyzed for their ability to com- plement the ap3 mutations. In ali cases, plants homozygous for the ap3 allele and heterozygous for the transgenic insert were assayed for their phenotype, and the genotype was con- firmed by progeny testing.

AP3 Genomic and cDNA Constructs Rescue the ap3 Mutant Phenotype

Of four transgenic lines containing the pBam21 construct tested, three lines showed complete rescue to a wild-type flo- ral phenotype of plants homozygous for the ap3-7, ap3-3, and ap3-4 alleles. One pBam21-containing line did not complement the ap3 mutations and was not analyzed further. Complemen- tation of the weak ap3-7 allele by a 6-kb genomic AP3 fragment has also been demonstrated by Okamoto et al. (1994). The three lines containing the pla19 construct also showed complete res- cue of these mutant phenotypes to the wild type (Figure 4E). The transgenic complementation was confirmed by segrega- tion analysis of progeny from candidate rescued plants.

The pla19 construct contains 1.7 kb 5'of the transcriptional start site and 0.24 kb of 3' untranslated sequence, but it does not contain any intron sequences. Therefore, it appears that essentially all of the spatial and temporal regulatory signals required for proper function of the AP3 gene are contained within this 5' region. Surprisingly, a fusion of the AP3 coding sequence to the constitutive cauliflower mosaic virus 35s

I T

I

Figure 3. Sequence Analysis of the AP3 Promoter.

(A) Sequence of theAP3 region 5'to the start of translation. Sequences showing similarity with the CArG box motif are underlined. A putative TATA box is shown in boldface letters. The initiation of transcription as defined by Jack et al. (1992) is at position +l. The 5' ends of the three AP3 cDNAs isolated in this study are at positions +9, +20, and +22. (B) Sequence comparison of theAP3 and DefA genes. Promoter regions are denoted by a thin bar and exons by a thick bar. lntrons are shown as lines. The position of introns is conserved between LkfA and AP3, but the length is not. The degree of nucleotide sequence similarity is shown for the promoter regions and exons, with black indicating >80% identity, dark hatching 40 to 80% identity, light hatching 20 to 40%, and no hatching <20% identity. Positions of CArG box homolo- gous sequences are shown by arrowheads.

Table 1. Transnenic Lines Used in This Studv

Transgenic Kanr/KanS Linea Seqreqationb

~~ ~

Number of lnsertsC

B6 160:58 (3:l) 4 88 66:27 (3:l) 1 B16 72:27 (3:l) 1 A6 26:9 (3:l) 2 A7 45:18 (3:l) 3 A1 O 88:36 (3:l) 1 D3 71:26 (3:l) 2 D6 94:36 (3:l) 2 D11 73:17 (3:l) 1

a Lines designated with B are transgenic for the pBam21 construct; with A, for the pla19 construct; and with D, for the pDef7 construct.

The best fit (x2 . P > 0.05) segregation of progeny derived from a self of heterozygous plants is given in parentheses. Kan', kanamy- cin resistant; Kans, kanamycin sensitive.

Number of inserts is estimated from number of bands hybridizing on DNA gel blots.

Conservation of Floral Homeotic Gene Function 1639

Table 2. Summarv of Second Whorl Homeotic Transformations in Control and Transgenic Plantsa

Genotype (N)b

ap3-7 (18) ap3-3 (16) ap3-3 D6 (21) ap3-3 D11 (15) ap3-4 (1 7) ap3-4 D3 (21) ap3-4 D6 (22) a~3-4 D11 f18)

Second Whorl Sepaloid Petals W)

O O

52 O O O

73 O

Second Whorl Petaloid Sepals ( O 4

O O

48 67

O 58 27 67

Second Whorl Sepals I010 )

Average Number of Second Whorl Organs per Flower

1 O0 1 O0

O 33

1 O0 42

O 33

~~ ~ ~~ ~ ~

a The percentage of different organ types recovered and the average number of organs per whorl scored in dissected flowers from each geno- type are listed.

N, number of flowers scoced; the number IS given in parentheses.

promoter does not completely restore the ap3-3 phenotype to the wild type (Jack et al., 1994). This observation suggests that the 35s promoter may lack some of the regulatory ele- ments present in theAP3 promoter that are necessary to drive expression effectively in these floral tissues.

Complementation of ap3 Mutant Phenotypes by the Chimeric AP3-DefA Construct

Transgenic plants containing the pDef7 construct were also crossed to plants homozygous for the ap3-7, ap3-3, or the ap3-4 allele. For all three of the transgenic lines examined, similar results were obtained. The pDef7 construct is sufficient to res- cue plants homozygous for ap3-7 to a wild-type phenotype (Figure 4F). These pDef7 ap3-7 flowers are fertile and have normal seed set.

The pDef7 transgene partially complements the strong ap3-3 or ap3-4 lesions. The phenotype produced by ap3-3 plants car- rying the pDef7 transgene is less severe than that of ap3-3 homozygous mutants. The pDef7 ap3-3 plants have second whorl organs that are generally petaloid sepals or sepaloid petals (Table 2 and Figure 4G). The epidermal cells in these

second whorl organs have a shape and size very similar to these of wild-type peta1 cells but do not appear quite as round or regular (Figure 5E). Furthermore, these transgenic second whorl organs do not appear to be mosaic organs of different cell types. In pDef7 ap3-3 flowers, the most frequent third whorl phenotype consists of freestanding carpelloid stamens, often with distinct locules, that occasionally have ovules along the lateral edges of each organ (Table 3 and Figures 5G to 51).

In the pDef7 ap3-4 flowers, we observed a phenotype that is milder than that of ap3-4 flowers (Tables 2 and 3, and Figures 4H and 5J). These phenotypes are similar to those produced by pDef7 ap3-3 plants. In addition, we ObseNed a slight differ- ence in the frequency of these phenotypes with different pDef7 lines (Tables 2 and 3). Plants homozygous for ap3-3 or ap3-4 carrying the D6 transgene showed slightly less severe pheno- types than those carrying the D3 or D11 transgene.

Some Transgenic Lines Show a Cosuppressed Phenotype

Approximately 10% of the transgenic lines that we generated displayed a dominant cosuppressed phenotype. These included

~

Table 3. Summary of Third Whorl Homeotic Transformations in Control and Transgenic Plantsa

Genotype (N)

Third Whorl Third Whorl Third Whorl Third Whorl Carpeloid Stamens Stamenoid Carpels Fused Carpels Filaments ( 0 4 ( 0 4 ( O 4 (04

ap3- 7 (1 8) ap3-3 (1 6) ap3-3 D6 (21) ap3-3 D11 (15) ap3-4 (1 7) ap3-4 D3 (21) ap3-4 D6 (22) a~3-4 D11 (1 8)

84 O

1 O0 89

O 30 53 21

16 3 O O O

70 44 51

O 89

O 1

61 O O 2

O 8 O

10 39

O 3

26

Average Number of Third Whorl Organs per Flower

5.6 3.9 5.6 4.7 4.1 4.7 5.5 5.4

~~

a Notations are as given in Table 2.

1640 The Plant Cell

Figure 4. Phenotypes of ap3 Mutant and Transgenic Arabidopsis Flowers.(A) Wild-type flower.(B) to (D) ap3-7, ap3-3, and ap3-4 homozygous mutant flowers, respectively.(E) Flower from a homozygous ap3-3 mutant plant carrying the A6 transgene shows complete rescue of the mutant phenotype.(F) Flower from a homozygous ap3-7 mutant plant carrying the D3 transgene also shows complete rescue.(G) Flower from a homozygous ap3-3 mutant carrying the D6 transgene shows partial rescue, as compared with the flower in (C).(H) Flower from a homozygous ap3-4 mutant carrying the D3 transgene also shows partial rescue, as compared with the flower in (D).Scale bar in (A) = 0.5 mm for (A) to (H).

three pBam21, four pla19, and five pDef7 lines. The severityof the cosuppressed phenotype varied between lines as wellas within individual plants of any one line. In the most severelyaffected lines, the mutant phenotype resembled that producedby a strong ap3 aliele. Lines showing moderate phenotypesdisplayed sepaloid petals and partially transformed stamens.Some lines were weakly fertile, had nearly normal organ de-velopment, and showed some defects only in the size andshape of the stamens. These were characterized as havingweakly cosuppressed phenotypes. In addition, most of thecosuppressed lines had flowers that showed a reversion to awild-type phenotype. This reversion would often extend throughseveral flowers along an inflorescence.

DISCUSSION

Arabidopsis and Antirrhinum Homeotic GeneFunctions Are Conserved

We showed that coding sequences from the Antirrhinum DefAgene can functionally complement the Arabidopsis ap3-1 mu-tation. This functional test demonstrated that the chimeric DefAgene product that we used can interact appropriately with var-ious Arabidopsis proteins to confer a normal developmentalpattern to the Arabidopsis flower. However, the pDef7 trans-gene did not completely complement the strong ap3-3 and

ap3-4 mutations. Plants carrying both the pDef? transgene anda strong ap3 mutation showed defects in both second and thirdwhorl organs, but these transformations were much milder thanthose produced by strong ap3 mutations alone. In fact, onepDef7-containing line showed almost complete rescue of strongmutations at the AP3 locus (D6; see Tables 2 and 3). The D6transgenic line produced an almost full restoration of the sec-ond whorl to a wild-type phenotype and partial restoration ofthird whorl organs (Figures 4G and 5G to 51).

The chimeric pDef7 construct that we used to rescue mu-tant ap3 plants contains essentially all of the MADS box domainderived from AP3, with other sequences, including the K-box,derived from the DefA coding region. Thus, the lack of com-plete rescue seen in the complementation experiments withstrong ap3 alleles is due to sequence differences outside theMADS box. This observation indicates that specificity of MADSbox gene function is modulated by sequences outside theMADS box. However, the regulatory specificity of homeobox-containing genes in Drosophila appears to depend primarilyon the specificity of the homeodomain (Kuziora and McGinnis,1989; Mann and Hogness, 1990). Thus, the mechanisms bywhich the regulatory specificities of MADS box-containing andhomeodomain-containing homeotic genes are specified ap-pear to be distinct.

The partial rescue of strong ap3 mutations by the pDef7transgene can be explained in at least two ways. The pDef7transgenic product may be able to activate completely onlya subset of downstream genes. This model is supported by

Conservation of Floral Homeotic Gene Function 1641

Figure 5. Cellular Phenotypes of Wild-Type, Mutant, and Transgenic Arabidopsis Flowers.

(A) to (E) Abaxial surface view of first or second whorl organs. In (A), wild-type (Landsberg erecfa) second whorl petal cells are shown. In (B),wild-type first whorl sepal cells are shown. In (C), ap3-1 second whorl cells at 22°C resemble sepal cells. Note elongated epidermal cells andstomata. In (D), ap3-3 second whorl cells with sepal-like characteristics are shown. In (E), second whorl cells produced by an ap3-3 plant carryingthe 06 transgene are shown. The size and shape of epidermal cells are similar to petal cells; note the lack of stomata.(F) to (J) Third whorl organs. In (F), a wild-type stamen during anthesis is shown. In (G) to (I), third whorl organs with increasingly carpelloidcharacteristics from transgenic ap3-3 plants containing the D6 transgene are shown. In (J), a filamentous third whorl organ from a transgenicap3-4 plant containing the D3 transgene is shown.Scale bar in (A) = 10 urn for (A) to (E). Scale bar in (F) = 100 urn for (F) to (J).

the observation of intermediate cellular phenotypes in boththe second and third whorls. These mosaic organs do not ap-pear to be a mixture of different tissue types, as seen with someof the floral homeotic mutations (Bowman et al., 1989; Irishand Sussex, 1990); rather, the cells themselves appear to havean intermediate shape and size. One interpretation of this in-termediate cellular phenotype is that the pDef7 transgene canappropriately regulate some but not all of the downstreamgenes required to elaborate the cellular architecture. Anotherpossible explanation of the partial rescue results is that thechimeric transgene is not as active as the endogenous AP3gene, so there is a threshold effect in the ability of the DefAsequences to participate in activating downstream targe!genes. This threshold effect could explain why the chimerictransgene can fully complement the ap3-7 mutation; the chi-meric transgene and the defective ap3-1 protein producttogether could be sufficient to effect normal regulatoryfunctions.

The DefA domain used in these experiments includes theK-box, which is thought to be required for heterodimer forma-tion (Schwarz-Sommer et al., 1992; Trobner et al., 1992). Thelack of complete rescue of strong mutant ap3 alleles by thepDef7 transgene suggests that this transgene may not be ableto participate as readily in heterodimer formation. Becausethe AP3 protein is thought to be stabilized by heterodimeriz-ing with PI through the K-box domain (Goto and Meyerowitz,1994; Jack et ai., 1994), it seems possible that the sequencedifferences we have introduced have attenuated the stabilityof such heterodimeric complexes.

Conservation of AP3 and DefA Target Sequences

The AP3 and DefA gene products are thought to autoregulateand so presumably bind to sequences present in the respec-tive promoter regions (Schwarz-Sommer et al., 1992; Jack et

1642 The Plant Cell

al., 1994). In vitro studies have shown that the DefA CArG box at position -1207 effectively binds DefA-Glo heterodimers (Trobner et al., 1992). The conservation of a putative CArG box at approximately the same distance upstream of the start of AP3 and DefA translation, embedded in an otherwise noncon- served promoter region, suggests that these gene products may bind to this sequence to effect appropriate autoregula- tion. Additional evidence that this autoregulatory circuit is conserved within dicots is based on observations that thisAP3 promoter drives apparently normal B gene expression of a reporter gene in tobacco (Day et al., 1995).

The complementation assay we have used is conceptually different from that used to examine the functional equivalence of orthologous homeotic genes in other systems. Conserva- tion of mammalian and Drosphila homeotic gene functions has been assessed by comparing the phenotypes produced by ectopic overexpression of mammalian Hox genes with that produced by ectopic overexpression of the orthologous endogenous genes in Drosophila. In some cases, these ex- periments suggest that the orthologous genes are functionally similar, whereas other comparisons indicate differences in function (Malicki et al., 1990; McGinnis et al., 1990; Bachiller et al., 1994). However, it is not clear whether these differences are due to unexpected interactions between the ectopically expressed genes and spatially restricted gene products that do not normally participate in such regulatory cascades (Bachiller et al., 1994). Similarly, the fact that ectopically in- duced expression of Brassica AGAMOUS (BAG7) and tobacco AGAMOUS (NAG7) in tobacco produces somewhat different phenotypes can be explained by nove1 and different interac- tions of these ectopic gene products with the endogenous regulatory machinery (Mande1 et al., 1992; Kempin et al., 1993). In our experiments, the heterologous Antirrhinum sequences were expressed in the appropriate AP3 domain so that we could investigate the extent to which this gene can replace the normal regulatory functions of its ortholog. Thus, these experiments provide a more sensitive assay for assessing the equivalence of orthologous gene functions.

lmplications for the Evolution of Regulatory Pathways

Despite the observations of similarities in the sequences of and mutant phenotypes conferred by the floral homeotic genes, explicit models for how these developmental pathways might be similar have not been detailed (Coen, 1991). How are the extreme differences in floral morphology generated?

At least four possibilities explain the wide variety of species- specific differences in peta1 and stamen morphology. First, differences in the expression of homeotic genes could poten- tially confer some of these unique morphologies. The slight differences in expression patterns of apparently similar floral homeotic genes from different species provide some support for such a model. Second, altering the specificity of heterodi- meric pairing partners of the AP3 and DefA proteins could lead to new regulatory cascades. Although the K-box, which is

thought to be required for heterodimerization of these proteins, has diverged at the sequence level, these K-box domains ap- pear to share a conservation of secondary Structure (Ma et al., 1991; Pnueli et al., 1991). One could imagine a coevolu- tion of pairing partners that could then lead to activation of new suites of target genes. However, we demonstrated here that the DefA K-box can partially substitute forAP3 sequences and so presumably can interact with the PIgene product. Third, differences in CArG box binding sites could be responsible for differences in target gene activation. In vitro studies have suggested that different MADS box-containing genes have somewhat different target site specificities (Trobner et al., 1992; Huang et al., 1993; Shiraishi et al., 1993), but the extent to which these differences play a role in modulating target gene expression in vivo is unclear. Our comparison of promoter se- quences suggests that such differences may be subtle (Figure 2). Finally, differences in target gene function could be largely responsible for defining species-specific morphologies. Our evidence showing that the DefA coding sequences can compen- sate for mutations in the AP3 gene suggests that differences in target gene function probably are responsible in large part for the significant morphological changes we see in flower form.

METHODS

Cloning and Sequencing

DNA manipulations were performed according to standard techniques. Four APE TALA3 (AP3) genomic clones were recovered by screening a XFlX genomic Landsberg erecta library (a gift from 8. Hauge, Mas- sachusetts General Hospital, Boston, MA) with pDeflB (kindly provided by Z. Schwarz-Sommer, Max-Planck-lnstitut für Züchtungsforschung, Cologne, Germany) (Sommer et al., 1990). Hybridizations were performed at Iow stringency with washes at 60% in 0.3 M NaCI, 20 mM sodium phosphate, 2 mM EDTA, 0.5% SDS, pH 7.0. A 1.9-kb Hindlll fragment corresponding to the MADS box and 5’ upstream region of AP3 was subcloned into pBluescribe (Stratagene) to give pV15. This 1.9-kb re- gion was sequenced using a combination of Xbal or Sacl subclones in pBluescript II SK+ (Stratagene) and specific primers. 60th strands were sequenced in double-stranded sequencing reactions using Se- quenase (U.S. Biochemical). The AP3 promoter sequence has GenBank accession number of U30729.

Plasmid Constructions

pBam21 was generated by subcloning a BamHl genomic fragment containing theAP3 gene into pBin19 (Bevan, 1984). Three AP3 cDNA clones were obtained by screening a Landsberg erecta inflorescence cDNA library (Yamamoto et al., 1995) at high stringency with washes at 0.1 x SSC (1 x SSC is 0.15 M NaCI, 0.015 sodium citrate), 0.1% SDS at 65OC. One full-length AP3 cDNA, pla, was completely se- quenced, and the coding region is identical to that published by Jack et al. (1992). The pla19 construct was generated by subcloning a Hindlll- Smal pla fragment corresponding to the 3’ end of the AP3 cDNA into a Hindlll-Smal cut pBinl9/cauliflower mosaic virus 35s promoterl

Conservation of Floral Homeotic Gene Function 1643

nopaline synthase terminator vector (kindly provided by M. Conkling, North Carolina State University, Raleigh. NC). This intermediate con- struct was then cut with Hindlll, and the 1.9-kb genomic Hindlll fragment corresponding to the promoter and 5' end of the AP3 gene was in- serted to give pla19. pDef7 was generated in a similar manner, except that a Hindlll-EcoAV fragment of the pDeflB clone was used for sub- cloning into the pBinl9/35Slnopaline synthase terminator vector.

Generation and Analysis of Transformants

Constructs generated in the binary pBinl9 vector were transferred to Agrobacterium tumefaciens LBA4404 by electroporation. Arabidopsis thaliana wild-type (Nossen ecotype) roots were transformed using stan- dard procedures (Valvekens et al., 1988). Kanamycin-resistant plantlets were rooted, and seed were collected from putative transformants. The presence of the transgene was confirmed by segregation of the kanamy- cin resistance trait to progeny, as well as by DNA gel blotting with probes corresponding to AP3 or Deficiens (DefA) sequences. RNA gel blot analyses were performed using standard methods, with total RNA ex- tracted using Trizol (Gibco BAL). Transgenic plants were used as pollen donors in crosses to ap3 mutant plants. Tissue was prepared for scan- ning electron microscopy as previously described (Irish and Sussex, 1990).

ACKNOWLEDGMENTS

We thank Zsuzsanna Schwarz-Sommer for providing the pDeflB clone, Brian Hauge for the Arabidopsis genomic library, and Mark Conkling for the pBinl9/35S/nopaline synthase terminator vector. We also ap- preciate the technical assistance of Roberta Miller and Roshani Anandappa in plasmid generation and sequence analysis. We thank members of fhe lrish laboratory for critical comments on the manu- script. This work was supported by U.S. Department of Agriculture Grant No. 92-37304-7716 and National Science Foundation Grant No. IBN- 9220536 to V.F.I., and a McKnight Foundation Grant to Yale University.

Received April 28, 1995; accepted August 7, 1995.

REFERENCES

Bachiller, D., Macias, A., Duboule, D., and Morata, G. (1994). Con- servation of a functional hierarchy between mammalian and insect Hox-Hom genes. EMBO J. 13, 1930-1941.

Bevan, M. (1984). Einary Agrobacterium vectors for plant transforma- tion. Nucleic Acids Res. 12, 8711-8721.

Bowman, J.L., Smyth, D.R., and Meyerowitz, E.M. (1989). Genes directing flower development in Arabidopsis. Plant Cell 1, 37-52.

Christ, C., and Tye, B.-K. (1991). Functional domains of the yeast tran- scriptionlreplication factor MCMI. Genes Dev. 5, 751-763.

Coen, E.S. (1991). The role of homeotic genes in flower development and evolution. Annu. Rev. Plant Physiol. Plant MOI. Biol. 42,241-279.

Coen, E.S., and Meyerowitz, E.M. (1991). The war of the whorls: Genetic interactions controlling flower development. Nature 353, 31-37.

Cronquist, A. (1981). An lntegrated System of Classification of Flowering Plants. (New York: Columbia University Press).

Cronquist, A. (1988). The Evolution and Classification of Flowering Plants. (New York: New York Botanical Garden).

Day, C.D., Galgoci, B.F.C., and Irish, V.F. (1995). Genetic ablation of petal and stamen primordia to elucidate cell interactions during floral development. Development 121, 2887-2895.

Goto, K., and Meyerowitz, E.M. (1994). Function and regulation of the Arabidopsis floral homeotic gene PISTILLATA. Genes Dev. 8,

Halfter, U., Ali, N., Stockhaus, J., Ren, L., and Chua, N.-H. (1994). Ectopic expression of a single homeotic gene, the petunia gene green petal, is sufficient to convert sepals to petaloid organs. EMBO J.

Hill, J.P., and Lord, E.M. (1989). Floral development in Arabidopsis thaliana: A comparison of the wild type and the homeotic pistillata mutant. Can. J. Bot. 67, 2922-2936.

Huang, H., Mizukami, Y., Hu, Y., and Ma, H. (1993). lsolation and characterization of the binding sequences for the product of the Arabidopsis floral homeotic gene AGAMOUS. Nucleic Acids Res. 21, 4769-4776.

Irish, V.F., and Sussex, I.M. (1990). Function of the apetala-7 gene during Arabidopsis floral development. Plant Cell 2, 741-753.

Jack, T., Brockman, L.L., and Meyerowitz, E.M. (1992). The homeotic gene APETALA3 of Arabidopsis thaliana encodes a MADS box and is expressed in petals and stamens. Cell 68, 683-697.

Jack, T., Fox, G.L., and Meyerowitz, E.M. (1994). Arabidopsis homeotic gene APETALA3 ectopic expression: Transcriptional and posttranscriptional regulation determine floral organ identity. Cell

Kempin, S.A., Mandel, M.A., and Yanofsky, M.F. (1993). Conversion of perianth into reproductive organs by ectopic expression of the tobacco floral homeotic gene NAG7. Plant Physiol. 103, 1041-1046.

Kush, A,, Brunelle, A., Shevell, D., and Chua, N.-H. (1993). The cDNA sequence of two MADS box proteins in petunia. Plant Physiol. 102, 1051-1052.

Kuziora, M.A., and McGinnis, W. (1989). A homeodomain substitu- tion changes the regulatory specificity of the Deformed protein in Drosophila embryos. Cell 59, 563-571.

Ma, H., Yanofsky, M.F., and Meyerowitz, E.M. (1991). AGL7-AGLG, an Arabidopsis gene family with similarity to floral homeotic and transcription factor genes. Genes Dev. 5, 484-495.

Malicki, J., Schughart, K., and McGinnis, W. (1990). Mouse Hox 2.2 specifies thoracic segmenta1 identity in Drosophila embryos and larvae. Cell 63, 961-967.

Mandel, M.A., Bowman, J.L., Kempin, S.A., Ma, H., Meyerowitz, E.M., and Yanofsky, M.F. (1992). Manipulation of flower structure in transgenic tobacco. Cell 71, 133-143.

Mann, R.S., and Hogness, D.S. (1990). Functional dissection of Ultrabithorax proteins in D. melanogastel: Cell 60, 597-610.

McGinnis, N., Kuziora, M.A., and McGinnis, W. (1990). Human Hox 4.2 and Drosophila Deformed encode similar regulatory specifici- ties in Drosophila embryos and larvae. Cell 63, 969-976.

Meyerowitz, EM. , Bowman, J.L., Brockman, L.L., Drews, G.N., Jack, T., Sieburth, L.E., and Weigel, D. (1991). A genetic and mo- lecular model for floral development in Arabidopsis thaliana. Development 1, 157-167.

1548-1560.

13, 1443-1449.

76, 703-716.

1644 The Plant Cell

Muller, J. (1981). Fossil pollen records of extant angiosperms. Bot. Rev. 47, 1-140.

Norman, A., Runswick, M., Pollock, R., and Treisman, R. (1988). lsolation and properties of fos cDNA clones encoding SRF, a tran- scription factor that binds to the c-fos serum response element. Cell

Okamoto, H., Yano, A., Shiraishi, H., Okada, K., and Shimura, Y. (1994). Genetic complementation of a floral homeotic mutation, apetala3, with Arabidopsis thaliana gene homologous to DEFlClENS of Antirrhinum majus. Plant MOI. Biol. 26, 465-472.

Passmore, S., Maine, G.T., Elble, R., Christ, C., and Tye, B.-K. (1988). Saccharomyces cerevisiae protein involved in plasmid maintenance is necessary for mating of MATa cells. J. MOI. Biol. 204, 593-606.

Pnueli, L., AbuAbeid, M., Zamir, D., Nacken, W., Schwarz-Sommer, Z., and Lifschitz, E. (1991). The MADS box gene family in tomato: Temporal expression during floral development, conserved second- ary structures and homology with homeotic genes from Antirrhinum and Arabidopsis. Plant J. 1, 255-266.

Schwarz-Sommer, Z., Hue, I., Huijser, P., Flor, P.J., Hansen, R., Tetens, F., Lonnig, W.-E., Saedler, H., and Sommer, H. (1992). Characterization of the Antirrhinum floral homeotic MADS-box gene deficiens: Evidence for DNA binding and autoregulation of its per- sistent expression throughout flower development. EMBO J. 11,

55, 989-1003.

251-263.

Shiraishi, H., Okada, K., and Shlmura, Y. (1993). Nucleotide se- quences recognized by the AGAMOUS MADS domain of Arabidopsis fhaliana in vitro. Plant J. 4, 385-398.

Sommer, H., Beltran, J.-P., Huijser, P., Pape, H., Lonnig, W.-E., Saedler, H., and Schwarz-Sommer, 2. (1990). Deficiens, a homeotic gene involved in the control of flower morphogenesis in Antirrhinum majus: The protein shows homology to transcription factors. EMBO J. 9, 605-613.

Trobner, W., Ramirez, L., Motte, P., Hue, I., Huijser, P., Lonnig, W.E., Saedler, H.; Sommer, H., and Schwarz-Sommer, 2 . (1992). Globosa-A homeotic gene which interacts with deficiens in the con- trol of Antirrhinum floral organogenesis. EMBO J. 11, 4693-4704.

Valvekens, D., Van Montagu, M., and Van Lijsebettens, M. (1988). Agrobacterium tumefaciens-mediated transformation of Arabidop- sis fhaliana root explants by using kanamycin selection. Proc. Natl. Acad. Sci. USA 85, 5536-5540.

van der Krol, A.R., Brunelle, A., Tsuchimoto, S., and Chua, N.-H. (1993). Functional analysis of petunia floral homeotic MADS box gene pMADS1. Genes Dev. 7, 1214-1228.

Yamamoto, Y.T., Conkling, M.A., Sussex, I.M., and Irish, V:F. (1995). An Arabidopsis cDNA related to animal phosphoinosttide-specific phospholipase C genes. Plant Physiol. 107, 1029-1030.

DOI 10.1105/tpc.7.10.1635 1995;7;1635-1644Plant Cell

V F Irish and Y T YamamotoConservation of floral homeotic gene function between Arabidopsis and antirrhinum.

 This information is current as of April 10, 2019

 

Permissions 98X

https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw15322

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

ADVANCING THE SCIENCE OF PLANT BIOLOGY © American Society of Plant Biologists