The PlanlJournal(l991) l(l), 37-49
Control of anthocyanin biosynthesis in flowers of Antirrhinurn majus Cathie Martin*, Andy Prescott, Steve Mackay, Jeremy Bartlett and Eli Vrijlandt Department of Genetics, John lnnes Institute, John lnnes Centre for Plant Science Research, Colney Lane, Norwich NR4 7UH, UK Summary The intensity and pattern of anthocyanin biosynthesis in Antirrhinum flowers is controlled by several genes. We have isolated six cDNA clones encoding enzymes in the pathway committed to flavonoid biosynthesis and used these to assay how the regulatory genes that modify colour pattern affect the expression of biosynthetic genes. The biosynthetic genes of the later part of the pathway appear to be co-ordinately regulated by two genes, Delila (Del), and Eluta ( N ) ,while the early steps (which also lead to flavone synthesis) are controlled differently. This division of control is not the same as control of anthocyanin biosynthesis by the regulatory genes R(S) and C1 in maize aleurone, and may result from the adaptive significance of different flavonoids in flowers and seeds, reflecting their attractiveness to insects and mammals respectively. €1 and de/ are probably involved in transcriptional control and both genes appear to be able to repress expression of some biosynthetic genes and activate expression of others. Introduction
Plant development depends to a large extent on regulated patterns of gene expression. These in turn dictate the differences in production of enzymes and other proteins that distinguish cell and tissue types structurally and functionally. An enormous amount of recent research has examined environmental and developmental regulation of a wide range of genes, including those encoding enzymes in metabolic pathways. However, the control of entire metabolic pathways is of greater functional significance to an organism, because this determines where and when the end products of the pathways are made. The expression of structural genes encoding the enzymes of secondary metabolic pathways generally appears to be under tight co-ordinate regulation, although pathways may be divided into sections in which sub-groups of structural genes are regulated co-ordinately (Cramer et a/.. 1985; Grisebach Received 28 February 1991; revised 8 April 1991 For correspondence (fax +44 603 56844)
*
and Hahlbrock, 1974; Hahlbrock, 1981; Hahlbrock et a/., 1971: Kuhn et a/., 1984; Lawton and Lamb, 1987; Schroder et a/.,1979). The control of metabolic pathways can be analysed using mutations. The synthesis of anthocyanin lends itself well to this type of analysis because loss of production is non-lethal and mutations can readily be found. Two genes have been identified in maize, R(S) and C1, which are primary regulators of anthocyanin biosynthetic genes in the aleurone cells of the kernel (Dooner, 1983; Dooner and Nelson, 1977, 1979; Kirby and Styles, 1970; McCormick, 1978: Reddy and Coe, 1962; Styles and Ceska, 1977). These genes appear to induce biosynthesis of all the structural genes encoding the enzymes committed to flavonoid biosynthesis (including chalcone synthase [CHS] (Dooner, 1983: Ludwig et a/., 1989), dihydroflavonol 4-reductase [DFR] (Cone et a/.. 1986; Klein et a/., 1989; Ludwig eta/.,1989; Reddy et al., 1978) and UDP glucose: flavonoid 3-0-glucosyltransferase [UFGT] (Cone et a/.. 1986; Dooner, 1983; Dooner and Nelson, 1977, 1979; Goff et a/., 1990; Klein et a/., 1989). R contains sequences homologous to the myc transcriptional activators in animals (Ludwig et a/., 1989) and C1 contains sequences homologous to myb transcriptional activators (Paz-Ares et a/., 1989).Their control of the structural genes would therefore appear to be by transcriptional activation, a conclusion supported by functional analysis (Goff eta/.,1990).Homologues of these genes (R(P), B , L c , PI) activate gene expression in other parts of the maize plant (Chandler et a!., 1989; Coe et a/., 1988; Dooner and Kermicle. 1971; Emerson, 1921; Gerats eta/., 1984; Stadler, 1946; Stadler and Neuffer. 1953), and the specificity of activation in different tissues is achieved through the specificity of expression of the regulatory genes themselves (Ludwig and Wessler, 1990). In addition, other regulatory genes such as Vp-I may regulatethese regulatory genes (McCarty et a/., 1989) and environmental inducers such as light and developmental inducers such as abscisic acid may operate (at least in part) through regulation of expression of these regulatory genes (McCarty eta/., 1991; Taylor and Briggs, 1990). Even in maize it is not clear whether the regulatory genes induce structural gene expression in the same way in all tissues. Although B , L c , and R(S) and R(P) are functionally homologous (Chandler eta/.,1989; Goff eta/.: 1990) in R-g plants (which carry a mutation of R(P) that blocks anthocyanin biosynthesis in the maize seedling), CHS expression may occur independently of R(P),whereas in kernels it is controlled by R(S) (Dooner, 1983: Ludwig et a!., 1989; Taylor and Briggs, 1990). 37
38
Cathie Martin et at.
In flowers, the genes encoding the anthocyanin biosynthetic enzymes are also co-ordinately expressed (Beld et a/., 1989; Coen et a/., 1986; Gerats et a/., 1985; van Tunen eta/.,1988).Because the increase in transcripts closely mirrors the accumulation of anthocyanin during flower development, the regulation of expression of the structural genes would appear to be the major means of controlling the production of anthocyanin. Several genes that influence the expression of biosynthetic genes in flowers have also been detected, including the genes contributing to the ‘red star’ phenotype in Petunia, which affect expression of CHS in a spatially specific manner (Mol et a/., 1983), the A and A 2 loci of pea that control CHS gene expression in flowers (Harker eta/., 1990), and the Delila (Del) locus of Antirrhinum majus that controls DFR expression in the flower tubes (Almeida eta/., 1989). We have been interested in how genes that modify the pattern of colour in flowers control the whole metabolic pathway. To study this we have isolated cDNA clones corresponding to six anthocyanin biosynthetic genes from Antirrhinum majus by differential cDNA cloning, transposon tagging or by homology to previously isolated genes. We have used these clones to assay the effects of two genes, Del (de Vries, 1903) and Huta (N) (Stubbe, 1966), which influence the pattern of anthocyanin production in flowers
Figure 1. Phenotypic effects of genes controlling the pattern and intensity of anthocyanin biosynthesis in Antirrhinurn flowers. (a) Full red flower of wild-type (el/e/:De//De/). (b) del- flower showing absence of anthocyanin in tube which is ivory coloured (e//e/:del/del). (c) Eluta flower showing restriction of anthocyanin production to the central region of the flower face, the central region of the two back petals, and a ring of anthocyanin at the base of the tube (€//€/:De//De/). (d) €luta:de/- flower. No anthocyanin is made in the flower tube. The pigmentation in the lobes is more restricted than in e/:De/+, and is concentrated around the central region of the face and in the central part of the two back petals (€//€/:del/del).There is therefore a synergistic interaction between N a n d d e l - .
(Figure 1). We have shown that the later part of the pathway committed to flavonoid biosynthesis is regulated differently by these genes compared with the earlier steps. Both regulatory genes appear to be able to stimulate expression of some genes and repress others, indicating that the action of such regulatory genes is complex and probably realized through changing interactions with other regulatory genes. Results
/solation of structural genes involved in anthocyanin biosynthesis In Antirrhinum, at least nine enzymes are involved in the anthocyanin biosynthetic pathway (Figure 2). To investigate how regulatory genes control the pattern of floral pigmentation it was necessary to isolate the structural genes for these enzymes. Two of them, CHS (Sommer and Saedler, 1986; Wienand eta/., 1982) and DFR (Coen eta/., 1986; Martin et a/., 1985) had previously been isolated from Antirrhinum. Other genes encoding biosynthetic enzymes had been isolated as cDNA clones from other species, including phenylalanine ammonia-lyase (PAL) from bean and parsley (Edwards eta/., 1985; Kuhn et a/., 1984), chalcone isomerase (CHI) from Petunia (van Tunen et a/., 1988) and UFGT from maize (Fedoroff et a/., 1984). Several other biosynthetic genes had not previously been isolated, including the flavanone 3-hydroxylase (F3H),the rhamnosyl transferase (RT) and some biochemically illdefined enzymes operating between DFR and UFGT (Coe, 1985; Heller and Forkmann, 1988). We isolated cDNA clones to PAL, CHI and UFGT using the heterologous cDNA clones from bean, Petunia and maize as probes. To confirm that these cDNA clones encoded the respective enzymes we sequenced each of them and compared their putative products to the published amino acid sequences for these enzymes: all three matched closely. The homology between the Antirrhinum UFGT sequence and the maize UFGT was the most patchy, but it extended along the whole amino acid sequence and gave a value of 58% similarity on the GAP program (Devereaux eta/., 1984). The isolation of previously uncloned cDNA types was achieved by differential cDNA cloning. From previous experiments we knew that the del- mutation caused a major reduction in DFR steady-state transcripts in the colourless tubes but only a relatively minor effect on CHS transcript levels (Almeida eta/., 1989). Feeding experiments using the product of the DFR gene, 2,3-cis-leucocyanidin, had revealed that del- did not block DFR expression alone, because feeding the compound to Antirrhinum flowers could not overcome the del- block and allow anthocyanin formation in del- tubes. We reasoned that
Control of flower colour 39
I
del- was a mutation affecting the steady-state level of transcripts of a number (but not all) of the anthocyanin biosynthetic genes. It should therefore have been possible to isolate genes regulated by Del by differential cDNA cloning of transcripts present in Del+ but absent in de/-tube tissue. After extensive analysis of cDNA clones isolated by differential screening we identified two groups of cDNA clones (group I and group II)that were differentially expressed in Del+ and del- tissue (Figure 3).
1 CHS
OH
J ' FLAVONES
To determine the identity of the genes corresponding to group I and group II transcripts we used Antirrhinurn
m'ti
OH
0
OH
0
"oJ & FlAVONOLS
ldentification of group I and group I/ transcripts
3
HO &OH
_.-_ F3'H 3 -
OH
OH
O'H
,
RHAMNOSE
Figure 2. Enzymic steps committed to flavonoid biosynthesis in Antirrhinum majus. Apart from glycosylation, no further modifications are made to the anthocyanins in flowers. The branch routes to flavones and flavonols are shown to illustrate the biochemical effects of inhibition of some enzymic steps and not others. (1) CHS condenses three molecules of malonyl-CoA and one of p-coumaroyl-CoA to form chalcononaringenin. (2) CHI catalyses a step that also has an appreciable spontaneous rate, converting chalcononaringenin to naringenin. (3) Flavonoid 3'-hydroxylase. The hydroxylation of the B ring dictates whether cyanidin or pelargonidin is the final product. This step may occur using either naringenin or dihydrokaempferolas substrates and is catalysed by a P450-type hydroxylase (Forkmann and Stotz, 1981). (4) F3H hydroxylates naringenin (4' hydroxylated) or eriodyctiol (3',4' hydroxylated) to dihydrokaempferol or dihydroquercetin respectively.
Figure. 3 Differential expression of group I and group II cDNA clones in De/+ and del- lobes and tubes. Northern blots showing the signal from the largest differentially hybridizing clones from group I (pJAM239)and group II (pJAM221) hybridizing to total RNA (25 kg) from lobes and tubes of De/+ and del- flowers.
(5) Dihydroflavonols are reduced by DFR to form leucocyanidin (3',4' hydroxylated) or leucopelargonidin (4' hydroxylated). (6/7) Colour is given to leucoanthocyanidin by introducing charge on the ring. The enzymic steps are uncharacterized, but a hydroxylation plus dehydration are suggested biochemically (Heller and Forkmann, 1988). The Candigene described here and the A 2 gene of maize (Menssen eta/., 1990) probably encode the hydroxylase. (8) Anthocyanidins are stabilized by glycosylation. The first step, catalysed by UFGT, adds glucose to the 3 position. (9) A second glycosylation, catalysed by rhamnosyl transferase, adds rhamnose to the glucose to form the rutinoside. In Antirrhinum further modifications such as methylation do not appear to occur.
40
Cathie Martin et al.
mutants with acyanic phenotypes which were not allelic to CHS (nivea) or DFR (pallida) mutations. One of these mutations, incolorata (incj. was available in several acyanic lines of independent origin (stocks JI:19, Jl:598, JI:700). Although allelic, these mutations probably arose from independent mutational events so we reasoned that consistent differences in transcripts involved in anthocyanin biosynthesis in these lines were likely to reflect the effects of the inc mutations rather than effects of modifying genes in the genetic background or genes closely linked to inc. Probing of RNA extracted from niv (JI:44 a null mutation of the CHS gene). pa/ (JI:56 a null mutation of the DFR gene) and two independent inc lines (JI:19, JI:700) with pJAM239 (the largest representative of group I transcripts) revealed a normal transcript in niv and pal- lines, but no transcript in either of the inc lines (Figure4).This suggested that group I transcripts encoded F3H which is the product of the Inc gene (Forkmann and Stotz. 1981). ~
Figure 5 . RFLPs for pJAM322 in Candi and midi lines Genorntc Southern blot showtnq DNA from wild-type candi and two inc lines (A JI 19 B JI 598) digested with Hindill and probed with pJAM322 Note lack of hvbridization to the cDNA probe in the candi line
The candica (candi) mutation gives a completely acyanic flower and is not allelic to niv, pa/ or inc (R. Carpenter, personal communication). Feeding with 2,3-cis-leucocyanidin produced no anthocyanin (although feeding inc nivor pallines led to anthocyanin production), suggesting that the gene encodes an enzyme for a late step in anthocyanin biosynthesis (i.e. after DFR) (Bartlett. 1989). Enzyme assays showed appreciable levels of UFGT activity in candiplants(Martin eta/.,1987),ruling out a block in UFGT activity. When DNA from candl plants was probed with pJAM322 (the longest representative of group II cDNA clones) no DNA homologous to the cDNA was observed, but a conserved fragment was apparent in other lines (Figure 5). This suggested that group II transcripts might correspond to the candi gene, representing a late step in anthocyanin biosynthesis. Confirmation that group I cDNAs encode the products of the inc (F3H)gene and group / I cDNAs encode the product of the candi gene
Figure 4. LOSS of transcripts homologous to pJAM239 in inclines Northern blots showing hybridization of poly(A+)RNA (5 k g ) from flowers of n i b and pa/ and two independent inc lines to the lonqest represent ative of group I cDNA clones pJAM239 (incAwas JI 19 incB was JI 700 j Note the lack of hybridization to either inc line (right-hand panel) To confirm that RNA had been loaded in each track the blot was rehybridizeo to the PAL cDNA clone (left-hand panel)
Genomic DNA was extracted from independent inc mutants (JI:19 and Jl:598) and compared to DNA from Inc' lines (Figure 6). Each mutant line showed different restriction fragment length polymorphisms (RFLPs)compared to the Inc' type. Both inc- lines were crossed to the I n c ' line (JI:7) and the segregation of RFLPs homologous to pJAM239 reiative to segregation of the inc phenotype was scored in 24 Inc ' or inc homozygotes in each cross. Both RFLPs showed 100% segregation with the Inc ' or inc phenotype and no heterozygotes were observed. ~
-
Control of flower colour
41
that the candimutation does not include other neighbouring genes. Comparison of the segregation of the acyanic phenotype to that of the Hindlll RFLP caused by this deletion showed 100°/~linkage, confirming that pJAM322 encodes the candi gene product. Sequence comparison of candito the cDNA-derived amino acid sequence of the A2 gene from maize (Menssen et a/., 1990) shows very high homology (70% similarity on the GAP program) between these two genes, indicating that they probably encode the same enzyme.
Effect of del on expression o f anthocyanin biosynthetic genes The del mutation was known to affect the steady-state levels of DFR transcripts in flower tubes dramatically while having a relatively minor quantitative effect on CHS transcripts (Almeida et a/., 1989). Differential cDNA cloning revealed that delalso had a major effect on F3H transcripts (lnc)and Canditranscripts. We completed this analysis by examining the effect of delon PAL, CHI and UFGT expression (Figure 8, Table 1). In colourless del- tubes the levels of PAL and CHI were normal compared to the wild-type Del controls. The UFGT transcripts were dramatically reduced in del tubes, supporting earlier enzyme assays (Bartlett, 1989; Martin et a/., 1987). In summary, del does not affect the activity of PAL, a relatively early step in phenylpropanoid metabolism, before the pathway is committed to flavonoid biosynthesis. We found no significant effect of del on CHS or CHI in the Antirrhinum tubes. There was no detectable quantitative effect of delon CHS or CHI expression in the Antirrhinum lobes. Although we have not yet isolated the full complement of anthocyanin biosynthetic genes (rhamnosyl-transferase is lacking and it is likely that at least one more gene, besides Candi, is +
Figure 6. RFLPs for pJAM239 in Inc and inc lines. Genomic Southern blot showing DNA from wild-type, candi and two inc ,ines (A:JI:19, B:JI:598) digested with Hindlll and probed with pJAM239. Note the polymorphisms for both inc lines which were used in RFLP mapping pJAM239 to inc. +
confirming that pJAM239 encoded the inc (F3H) gene product. Similarly, mapping of genomic DNA from the candi line with pJAM322 revealed a major rearrangement of the DNA of this locus compared to all other lines examined. The Candi coding region appears to have been deleted in the mutant (Figure 7) (Bartlett, 1989). Comparison of the restriction maps for the gene in candi- and Candi' lines revealed that the rearrangement includes most of the coding sequence, but restriction sites lying 3 kb from the coding sequence are unaltered (Figure 7). This suggested
@ reglonot IOc"5 conta,n,ng cendl Ccdlng sequence
A .
Figure 7. Restriction map of genomic clone of Antirrhinum DNA homologous to pJAM322. This map and the genomic clone were used to map sites in the candi line. A loss of hydridiring sequences in the coding region was observed in candi. However, sequences flanking the coding sequences appeared unaltered as determined by the hybridization of flanking probes, the maintenance of restriction sites, and the complete linkage of sequences on each side of the rearrangement. The candi mutation appears therefore to be limited to a discrete region of about 5 kb. B = BamHI, E = EcoRI, H = Hindlll, Hp = Hpal, P = Pstl, S 7 SmaI.
Table 1. Relative hybridization of PAL, CHS, CHI, F3H, DFR, Candi, UFGT and actin cDNA probes on Northern blots measured by scanning densitometry
N L
Del
N L
del T
el L
Del T
el L
del T
4.8 1.1
2.6
1.0 1.0
1.6
0.9
1.1
1.2 0.1
0.3
0.2
0.1
0.0 0.1
1.0 1.0 1.0 1.0 1.0
0.8 1.1
0.1 0.4
2.2 0.5 0.1
1.0 1.0 1.0
1.1
0.8
1.0
1.0
T
PAL CHS CHI F3H DFR Candi UFGT
0.8
0.9
0.7 2.2 0.6 0.1 0.6 0.2
0.7
Actin
0.8
1.0
0.2
0.7
0.0
1.0 1.0 1.0 1.0
1.0 1.1 1.0 1.2 0.8
0.0 0.0 0.0
1.0
0.8
1.0
0.0
Figures for hybridization of RNA from lobes (L) or tubes 0of different genotypes were compared to the hybridization in lobes or tubes of el:De/' plants standardized to 1.O.
42
Cathie Martin et al. Eluta is a semi-dominant gene that restricts pigmentation in the flower, concentrating anthocyanin biosynthesis to the central face, the inner edges of the two back petals, and a ring at the base of the tube (Stubbe, 1966) (Figure 1). Overall anthocyanin accumulation in the EIIN homozygote is reduced 8- to 10-fold compared to a full red ellel plant. Expression of each biosynthetic gene was assayed in EIIEI and ellel homozyogtes segregating in an F2 population (Figure 8, Table 1). The expression of CHS was not affected by the El mutation. The steady-state level of CHI transcript was enhanced in the lobes in EIIEI plants, although no effect was observed in the tubes. This result was surprising because El reduces the amount of anthocyanin in flowers. Elalso affected the steady-state levels of F3H, DFR, Candi and UFGT, decreasing them all in both lobes and tubes. The effect of Elon mRNA levels was greatest for DFR and UFGT, where only about 10-20% of transcript was found in El plants compared to el plants. These effects probably account for the reduction in anthocyanin in €/flowers. The effect of El on F3H transcript was less dramatic (60%). However, because inc is semi-dominant, F3H may be the major rate-limiting step for anthocyanin accumulation in flowers (compared to DFR which does not become limiting until transcript levels drop below 30% of normal (Coen et a/., 1986)) and so the effect of El on inc may also be significant in modifying the amount of anthocyanin made. The influence of El on Candi transcripts was not as great as that on DFR and UFGT transcripts; whether this reduction could be sufficient to affect the biosynthesis of anthocyanin is not clear.
Figure 8. Northern blots showing hybridization of PAL, CHS, CHI, DFR, Candiand UFGT cDNA probes. Hybridization of probes to 5 pg poly(A+j RNA from lobes (Lj and tubes (T) of €/:De/+, €/:de/-, e/:De/' and e/:de/- flowers. The phenotypes are shown above. The hybridizationwas quantified using scanning densitometty (Table 1j.
required for the conversion of leucoanthocyanidin to anthocyanidin (Heller and Forkmann, 1988)), from those genes that we have analysed, Del would appear to influence the accumulation of steady-state transcripts of the structural genes from the later part of the anthocyanin biosynthetic pathway, so giving rise to the acyanic tubes. By analogy to the situation for DFR we suggest that this effect operates through a failure to transcribe these genes in the upper part of the tube (Almeida et a/., 1989).
Effect of Eluta on expression of anthocyanin biosynthetic genes We used the clones of the anthocyanin biosynthetic genes to analyse the effects of another potential regulatory gene, Eluta , that modifies the pattern of floral pigmentation.
Combined effect of El and del on anthocyanin biosynthesis and gene expression E1:del- plants were identified in the segregating F2 population of an EI/El:Del/Del x el/el:del/del cross, and compared to El:Del+, el:Del+ and e1:del- plants. E1:del- plants produce less anthocyanin than El:Del+. There is no anthocyanin in the tubes but synthesis is also more restricted in the lobes (about 40% of that in El:Del+ flowers). The double mutant therefore shows a synergistic interaction between El and del (Figure 1). In Elrdel- plants, a significantly higher level of PAL transcript was detected in both lobes and tubes, compared to all other genotypes (Figure 8, Table 1). No effect was observed on CHS transcript levels. The expression of CHI was the same in E1:del- and El:Del+ plants, with elevated levels of CHI transcript in the lobes. There was a slight decrease in the expression of F3H, Candi and UFGT in elrdel- lobes compared to El:Del+ plants, although we cannot be sure that these changes would have resulted in a decrease in anthocyanin production of over 50%. In tubes there was reduced transcript for all four genes,
Control of flower colour 43 reflecting the influence of del, but the levels of F3H and UFGT transcripts were not as low as in el:del-. plants (Figure 8, Table 1). Both El and del show major effects on the levels of steady-state transcripts of F3H, DFR, Candi and UFGT and these are in accordance with their effects on anthocyanin biosynthesis. It therefore seems likely that the phenotypes of these genes arise through their control of the transcript levels of the biosynthetic genes of the later part of the pathway. Expression of PAL, CHS and CHI genes may be influenced by del and El but the type of control is different from that exerted over F3H, DFR, Candi and UFGT; CHI expression in lobes appears to be enhanced by El and PAL expression is enhanced in E1:del- plants. However, neither of these last two effects are reflected in the phenotype of the flower.
Sequence comparison of the upstream regions of genes regulated by del and El The most likely means for del and N to control structural gene expression is through transcriptional activation andlor repression. We therefore aligned the regions immediately upstream of the coding sequences of F3H, DFR (Almeida et al., 1989; Coen et al., 1986) and Candi and compared them to the upstream sequences of CHS (Sommer and Saedler, 1986) and CHI from Antirrhinum to search for target motifs that might be binding sites for these regulators (Figure 9). Apart from the TATA and CAAT motifs found in most eukaryotic genes (Breathnach and Chambon, 1981) we were able to identify a number of motifs that were similar to the binding sites reported for c-myb or C1 in maize (which also carries the myb-binding domain (Biedenkapp etal., 1988; Wienand et a/., 1989)). There were also motifs that were similar to the c-myc binding motif (Sen and Baltimore, 1986) which may represent the binding sequence of R in maize (Goff et a/., 1990). No motifs common to F3H, DFR and Candi, but absent from CHS and CHI, were found which might have revealed the recognition sequences for delandlor El. However, because Elaffects CHI expression, albeit differently from the way it affects F3H, DFR and Candi, the promoter of the CHI gene could also contain binding motifs for this regulator. In the DFR promoter, a 79 bp region lying around the CAAT box (-202 to -123), called box C, has been proposed to bind the del gene product, and deletion of this region has a dramatic effect on DFR gene expression in the tube (Almeida et a/., 1989). The sequences most homologous to this region in the CHS gene were -153 to -85 (with 86% similarity), in the CHI gene were -34 to +27 (with 59% similarity), in the F3H gene were +79 to +122 (with 71 YOsimilarity),and in the candigene -71 to -25 (with 70% similarity), (all calculated by GAP with a gap weight of 3.0
-200 CHS CHI
ACCAAAAGSA AAATAAlGTA CATGGTTTGA GASGTCAGTA CTCTAAGA1'A C t i C G A C A A A A A C A T G A A T TAATATGTTG GTACCTAACC T C T G C T E - C L
F3H DPR Candi
AASTATAArA 4TTTATTTCT ATTGCAAGAC AACAA?TAC-T-hTMS CACTTATYA_G TTATTACAAC AGAATCMTT TCTACC$CLMTCI~ACGA CTTTAAASTA ~@Z~&V.GAG AAAGTTACTA AATTAAATAT CTCTTGGCAC
CHS CHI
~ A A T T C C A A C CCAISTCACGT~CC?Z??%??A CCCGTAGCTA AAGTTGTTGG C7ACGTACTA ATAATAATAA T A A T A A T M T AACAATACTA CCCAQC
Control of anthocyanin biosynthesis in flowers of Antirrhinurn majus Cathie Martin*, Andy Prescott, Steve Mackay, Jeremy Bartlett and Eli Vrijlandt Department of Genetics, John lnnes Institute, John lnnes Centre for Plant Science Research, Colney Lane, Norwich NR4 7UH, UK Summary The intensity and pattern of anthocyanin biosynthesis in Antirrhinum flowers is controlled by several genes. We have isolated six cDNA clones encoding enzymes in the pathway committed to flavonoid biosynthesis and used these to assay how the regulatory genes that modify colour pattern affect the expression of biosynthetic genes. The biosynthetic genes of the later part of the pathway appear to be co-ordinately regulated by two genes, Delila (Del), and Eluta ( N ) ,while the early steps (which also lead to flavone synthesis) are controlled differently. This division of control is not the same as control of anthocyanin biosynthesis by the regulatory genes R(S) and C1 in maize aleurone, and may result from the adaptive significance of different flavonoids in flowers and seeds, reflecting their attractiveness to insects and mammals respectively. €1 and de/ are probably involved in transcriptional control and both genes appear to be able to repress expression of some biosynthetic genes and activate expression of others. Introduction
Plant development depends to a large extent on regulated patterns of gene expression. These in turn dictate the differences in production of enzymes and other proteins that distinguish cell and tissue types structurally and functionally. An enormous amount of recent research has examined environmental and developmental regulation of a wide range of genes, including those encoding enzymes in metabolic pathways. However, the control of entire metabolic pathways is of greater functional significance to an organism, because this determines where and when the end products of the pathways are made. The expression of structural genes encoding the enzymes of secondary metabolic pathways generally appears to be under tight co-ordinate regulation, although pathways may be divided into sections in which sub-groups of structural genes are regulated co-ordinately (Cramer et a/.. 1985; Grisebach Received 28 February 1991; revised 8 April 1991 For correspondence (fax +44 603 56844)
*
and Hahlbrock, 1974; Hahlbrock, 1981; Hahlbrock et a/., 1971: Kuhn et a/., 1984; Lawton and Lamb, 1987; Schroder et a/.,1979). The control of metabolic pathways can be analysed using mutations. The synthesis of anthocyanin lends itself well to this type of analysis because loss of production is non-lethal and mutations can readily be found. Two genes have been identified in maize, R(S) and C1, which are primary regulators of anthocyanin biosynthetic genes in the aleurone cells of the kernel (Dooner, 1983; Dooner and Nelson, 1977, 1979; Kirby and Styles, 1970; McCormick, 1978: Reddy and Coe, 1962; Styles and Ceska, 1977). These genes appear to induce biosynthesis of all the structural genes encoding the enzymes committed to flavonoid biosynthesis (including chalcone synthase [CHS] (Dooner, 1983: Ludwig et a/., 1989), dihydroflavonol 4-reductase [DFR] (Cone et a/.. 1986; Klein et a/., 1989; Ludwig eta/.,1989; Reddy et al., 1978) and UDP glucose: flavonoid 3-0-glucosyltransferase [UFGT] (Cone et a/.. 1986; Dooner, 1983; Dooner and Nelson, 1977, 1979; Goff et a/., 1990; Klein et a/., 1989). R contains sequences homologous to the myc transcriptional activators in animals (Ludwig et a/., 1989) and C1 contains sequences homologous to myb transcriptional activators (Paz-Ares et a/., 1989).Their control of the structural genes would therefore appear to be by transcriptional activation, a conclusion supported by functional analysis (Goff eta/.,1990).Homologues of these genes (R(P), B , L c , PI) activate gene expression in other parts of the maize plant (Chandler et a!., 1989; Coe et a/., 1988; Dooner and Kermicle. 1971; Emerson, 1921; Gerats eta/., 1984; Stadler, 1946; Stadler and Neuffer. 1953), and the specificity of activation in different tissues is achieved through the specificity of expression of the regulatory genes themselves (Ludwig and Wessler, 1990). In addition, other regulatory genes such as Vp-I may regulatethese regulatory genes (McCarty et a/., 1989) and environmental inducers such as light and developmental inducers such as abscisic acid may operate (at least in part) through regulation of expression of these regulatory genes (McCarty eta/., 1991; Taylor and Briggs, 1990). Even in maize it is not clear whether the regulatory genes induce structural gene expression in the same way in all tissues. Although B , L c , and R(S) and R(P) are functionally homologous (Chandler eta/.,1989; Goff eta/.: 1990) in R-g plants (which carry a mutation of R(P) that blocks anthocyanin biosynthesis in the maize seedling), CHS expression may occur independently of R(P),whereas in kernels it is controlled by R(S) (Dooner, 1983: Ludwig et a!., 1989; Taylor and Briggs, 1990). 37
38
Cathie Martin et at.
In flowers, the genes encoding the anthocyanin biosynthetic enzymes are also co-ordinately expressed (Beld et a/., 1989; Coen et a/., 1986; Gerats et a/., 1985; van Tunen eta/.,1988).Because the increase in transcripts closely mirrors the accumulation of anthocyanin during flower development, the regulation of expression of the structural genes would appear to be the major means of controlling the production of anthocyanin. Several genes that influence the expression of biosynthetic genes in flowers have also been detected, including the genes contributing to the ‘red star’ phenotype in Petunia, which affect expression of CHS in a spatially specific manner (Mol et a/., 1983), the A and A 2 loci of pea that control CHS gene expression in flowers (Harker eta/., 1990), and the Delila (Del) locus of Antirrhinum majus that controls DFR expression in the flower tubes (Almeida eta/., 1989). We have been interested in how genes that modify the pattern of colour in flowers control the whole metabolic pathway. To study this we have isolated cDNA clones corresponding to six anthocyanin biosynthetic genes from Antirrhinum majus by differential cDNA cloning, transposon tagging or by homology to previously isolated genes. We have used these clones to assay the effects of two genes, Del (de Vries, 1903) and Huta (N) (Stubbe, 1966), which influence the pattern of anthocyanin production in flowers
Figure 1. Phenotypic effects of genes controlling the pattern and intensity of anthocyanin biosynthesis in Antirrhinurn flowers. (a) Full red flower of wild-type (el/e/:De//De/). (b) del- flower showing absence of anthocyanin in tube which is ivory coloured (e//e/:del/del). (c) Eluta flower showing restriction of anthocyanin production to the central region of the flower face, the central region of the two back petals, and a ring of anthocyanin at the base of the tube (€//€/:De//De/). (d) €luta:de/- flower. No anthocyanin is made in the flower tube. The pigmentation in the lobes is more restricted than in e/:De/+, and is concentrated around the central region of the face and in the central part of the two back petals (€//€/:del/del).There is therefore a synergistic interaction between N a n d d e l - .
(Figure 1). We have shown that the later part of the pathway committed to flavonoid biosynthesis is regulated differently by these genes compared with the earlier steps. Both regulatory genes appear to be able to stimulate expression of some genes and repress others, indicating that the action of such regulatory genes is complex and probably realized through changing interactions with other regulatory genes. Results
/solation of structural genes involved in anthocyanin biosynthesis In Antirrhinum, at least nine enzymes are involved in the anthocyanin biosynthetic pathway (Figure 2). To investigate how regulatory genes control the pattern of floral pigmentation it was necessary to isolate the structural genes for these enzymes. Two of them, CHS (Sommer and Saedler, 1986; Wienand eta/., 1982) and DFR (Coen eta/., 1986; Martin et a/., 1985) had previously been isolated from Antirrhinum. Other genes encoding biosynthetic enzymes had been isolated as cDNA clones from other species, including phenylalanine ammonia-lyase (PAL) from bean and parsley (Edwards eta/., 1985; Kuhn et a/., 1984), chalcone isomerase (CHI) from Petunia (van Tunen et a/., 1988) and UFGT from maize (Fedoroff et a/., 1984). Several other biosynthetic genes had not previously been isolated, including the flavanone 3-hydroxylase (F3H),the rhamnosyl transferase (RT) and some biochemically illdefined enzymes operating between DFR and UFGT (Coe, 1985; Heller and Forkmann, 1988). We isolated cDNA clones to PAL, CHI and UFGT using the heterologous cDNA clones from bean, Petunia and maize as probes. To confirm that these cDNA clones encoded the respective enzymes we sequenced each of them and compared their putative products to the published amino acid sequences for these enzymes: all three matched closely. The homology between the Antirrhinum UFGT sequence and the maize UFGT was the most patchy, but it extended along the whole amino acid sequence and gave a value of 58% similarity on the GAP program (Devereaux eta/., 1984). The isolation of previously uncloned cDNA types was achieved by differential cDNA cloning. From previous experiments we knew that the del- mutation caused a major reduction in DFR steady-state transcripts in the colourless tubes but only a relatively minor effect on CHS transcript levels (Almeida eta/., 1989). Feeding experiments using the product of the DFR gene, 2,3-cis-leucocyanidin, had revealed that del- did not block DFR expression alone, because feeding the compound to Antirrhinum flowers could not overcome the del- block and allow anthocyanin formation in del- tubes. We reasoned that
Control of flower colour 39
I
del- was a mutation affecting the steady-state level of transcripts of a number (but not all) of the anthocyanin biosynthetic genes. It should therefore have been possible to isolate genes regulated by Del by differential cDNA cloning of transcripts present in Del+ but absent in de/-tube tissue. After extensive analysis of cDNA clones isolated by differential screening we identified two groups of cDNA clones (group I and group II)that were differentially expressed in Del+ and del- tissue (Figure 3).
1 CHS
OH
J ' FLAVONES
To determine the identity of the genes corresponding to group I and group II transcripts we used Antirrhinurn
m'ti
OH
0
OH
0
"oJ & FlAVONOLS
ldentification of group I and group I/ transcripts
3
HO &OH
_.-_ F3'H 3 -
OH
OH
O'H
,
RHAMNOSE
Figure 2. Enzymic steps committed to flavonoid biosynthesis in Antirrhinum majus. Apart from glycosylation, no further modifications are made to the anthocyanins in flowers. The branch routes to flavones and flavonols are shown to illustrate the biochemical effects of inhibition of some enzymic steps and not others. (1) CHS condenses three molecules of malonyl-CoA and one of p-coumaroyl-CoA to form chalcononaringenin. (2) CHI catalyses a step that also has an appreciable spontaneous rate, converting chalcononaringenin to naringenin. (3) Flavonoid 3'-hydroxylase. The hydroxylation of the B ring dictates whether cyanidin or pelargonidin is the final product. This step may occur using either naringenin or dihydrokaempferolas substrates and is catalysed by a P450-type hydroxylase (Forkmann and Stotz, 1981). (4) F3H hydroxylates naringenin (4' hydroxylated) or eriodyctiol (3',4' hydroxylated) to dihydrokaempferol or dihydroquercetin respectively.
Figure. 3 Differential expression of group I and group II cDNA clones in De/+ and del- lobes and tubes. Northern blots showing the signal from the largest differentially hybridizing clones from group I (pJAM239)and group II (pJAM221) hybridizing to total RNA (25 kg) from lobes and tubes of De/+ and del- flowers.
(5) Dihydroflavonols are reduced by DFR to form leucocyanidin (3',4' hydroxylated) or leucopelargonidin (4' hydroxylated). (6/7) Colour is given to leucoanthocyanidin by introducing charge on the ring. The enzymic steps are uncharacterized, but a hydroxylation plus dehydration are suggested biochemically (Heller and Forkmann, 1988). The Candigene described here and the A 2 gene of maize (Menssen eta/., 1990) probably encode the hydroxylase. (8) Anthocyanidins are stabilized by glycosylation. The first step, catalysed by UFGT, adds glucose to the 3 position. (9) A second glycosylation, catalysed by rhamnosyl transferase, adds rhamnose to the glucose to form the rutinoside. In Antirrhinum further modifications such as methylation do not appear to occur.
40
Cathie Martin et al.
mutants with acyanic phenotypes which were not allelic to CHS (nivea) or DFR (pallida) mutations. One of these mutations, incolorata (incj. was available in several acyanic lines of independent origin (stocks JI:19, Jl:598, JI:700). Although allelic, these mutations probably arose from independent mutational events so we reasoned that consistent differences in transcripts involved in anthocyanin biosynthesis in these lines were likely to reflect the effects of the inc mutations rather than effects of modifying genes in the genetic background or genes closely linked to inc. Probing of RNA extracted from niv (JI:44 a null mutation of the CHS gene). pa/ (JI:56 a null mutation of the DFR gene) and two independent inc lines (JI:19, JI:700) with pJAM239 (the largest representative of group I transcripts) revealed a normal transcript in niv and pal- lines, but no transcript in either of the inc lines (Figure4).This suggested that group I transcripts encoded F3H which is the product of the Inc gene (Forkmann and Stotz. 1981). ~
Figure 5 . RFLPs for pJAM322 in Candi and midi lines Genorntc Southern blot showtnq DNA from wild-type candi and two inc lines (A JI 19 B JI 598) digested with Hindill and probed with pJAM322 Note lack of hvbridization to the cDNA probe in the candi line
The candica (candi) mutation gives a completely acyanic flower and is not allelic to niv, pa/ or inc (R. Carpenter, personal communication). Feeding with 2,3-cis-leucocyanidin produced no anthocyanin (although feeding inc nivor pallines led to anthocyanin production), suggesting that the gene encodes an enzyme for a late step in anthocyanin biosynthesis (i.e. after DFR) (Bartlett. 1989). Enzyme assays showed appreciable levels of UFGT activity in candiplants(Martin eta/.,1987),ruling out a block in UFGT activity. When DNA from candl plants was probed with pJAM322 (the longest representative of group II cDNA clones) no DNA homologous to the cDNA was observed, but a conserved fragment was apparent in other lines (Figure 5). This suggested that group II transcripts might correspond to the candi gene, representing a late step in anthocyanin biosynthesis. Confirmation that group I cDNAs encode the products of the inc (F3H)gene and group / I cDNAs encode the product of the candi gene
Figure 4. LOSS of transcripts homologous to pJAM239 in inclines Northern blots showing hybridization of poly(A+)RNA (5 k g ) from flowers of n i b and pa/ and two independent inc lines to the lonqest represent ative of group I cDNA clones pJAM239 (incAwas JI 19 incB was JI 700 j Note the lack of hybridization to either inc line (right-hand panel) To confirm that RNA had been loaded in each track the blot was rehybridizeo to the PAL cDNA clone (left-hand panel)
Genomic DNA was extracted from independent inc mutants (JI:19 and Jl:598) and compared to DNA from Inc' lines (Figure 6). Each mutant line showed different restriction fragment length polymorphisms (RFLPs)compared to the Inc' type. Both inc- lines were crossed to the I n c ' line (JI:7) and the segregation of RFLPs homologous to pJAM239 reiative to segregation of the inc phenotype was scored in 24 Inc ' or inc homozygotes in each cross. Both RFLPs showed 100% segregation with the Inc ' or inc phenotype and no heterozygotes were observed. ~
-
Control of flower colour
41
that the candimutation does not include other neighbouring genes. Comparison of the segregation of the acyanic phenotype to that of the Hindlll RFLP caused by this deletion showed 100°/~linkage, confirming that pJAM322 encodes the candi gene product. Sequence comparison of candito the cDNA-derived amino acid sequence of the A2 gene from maize (Menssen et a/., 1990) shows very high homology (70% similarity on the GAP program) between these two genes, indicating that they probably encode the same enzyme.
Effect of del on expression o f anthocyanin biosynthetic genes The del mutation was known to affect the steady-state levels of DFR transcripts in flower tubes dramatically while having a relatively minor quantitative effect on CHS transcripts (Almeida et a/., 1989). Differential cDNA cloning revealed that delalso had a major effect on F3H transcripts (lnc)and Canditranscripts. We completed this analysis by examining the effect of delon PAL, CHI and UFGT expression (Figure 8, Table 1). In colourless del- tubes the levels of PAL and CHI were normal compared to the wild-type Del controls. The UFGT transcripts were dramatically reduced in del tubes, supporting earlier enzyme assays (Bartlett, 1989; Martin et a/., 1987). In summary, del does not affect the activity of PAL, a relatively early step in phenylpropanoid metabolism, before the pathway is committed to flavonoid biosynthesis. We found no significant effect of del on CHS or CHI in the Antirrhinum tubes. There was no detectable quantitative effect of delon CHS or CHI expression in the Antirrhinum lobes. Although we have not yet isolated the full complement of anthocyanin biosynthetic genes (rhamnosyl-transferase is lacking and it is likely that at least one more gene, besides Candi, is +
Figure 6. RFLPs for pJAM239 in Inc and inc lines. Genomic Southern blot showing DNA from wild-type, candi and two inc ,ines (A:JI:19, B:JI:598) digested with Hindlll and probed with pJAM239. Note the polymorphisms for both inc lines which were used in RFLP mapping pJAM239 to inc. +
confirming that pJAM239 encoded the inc (F3H) gene product. Similarly, mapping of genomic DNA from the candi line with pJAM322 revealed a major rearrangement of the DNA of this locus compared to all other lines examined. The Candi coding region appears to have been deleted in the mutant (Figure 7) (Bartlett, 1989). Comparison of the restriction maps for the gene in candi- and Candi' lines revealed that the rearrangement includes most of the coding sequence, but restriction sites lying 3 kb from the coding sequence are unaltered (Figure 7). This suggested
@ reglonot IOc"5 conta,n,ng cendl Ccdlng sequence
A .
Figure 7. Restriction map of genomic clone of Antirrhinum DNA homologous to pJAM322. This map and the genomic clone were used to map sites in the candi line. A loss of hydridiring sequences in the coding region was observed in candi. However, sequences flanking the coding sequences appeared unaltered as determined by the hybridization of flanking probes, the maintenance of restriction sites, and the complete linkage of sequences on each side of the rearrangement. The candi mutation appears therefore to be limited to a discrete region of about 5 kb. B = BamHI, E = EcoRI, H = Hindlll, Hp = Hpal, P = Pstl, S 7 SmaI.
Table 1. Relative hybridization of PAL, CHS, CHI, F3H, DFR, Candi, UFGT and actin cDNA probes on Northern blots measured by scanning densitometry
N L
Del
N L
del T
el L
Del T
el L
del T
4.8 1.1
2.6
1.0 1.0
1.6
0.9
1.1
1.2 0.1
0.3
0.2
0.1
0.0 0.1
1.0 1.0 1.0 1.0 1.0
0.8 1.1
0.1 0.4
2.2 0.5 0.1
1.0 1.0 1.0
1.1
0.8
1.0
1.0
T
PAL CHS CHI F3H DFR Candi UFGT
0.8
0.9
0.7 2.2 0.6 0.1 0.6 0.2
0.7
Actin
0.8
1.0
0.2
0.7
0.0
1.0 1.0 1.0 1.0
1.0 1.1 1.0 1.2 0.8
0.0 0.0 0.0
1.0
0.8
1.0
0.0
Figures for hybridization of RNA from lobes (L) or tubes 0of different genotypes were compared to the hybridization in lobes or tubes of el:De/' plants standardized to 1.O.
42
Cathie Martin et al. Eluta is a semi-dominant gene that restricts pigmentation in the flower, concentrating anthocyanin biosynthesis to the central face, the inner edges of the two back petals, and a ring at the base of the tube (Stubbe, 1966) (Figure 1). Overall anthocyanin accumulation in the EIIN homozygote is reduced 8- to 10-fold compared to a full red ellel plant. Expression of each biosynthetic gene was assayed in EIIEI and ellel homozyogtes segregating in an F2 population (Figure 8, Table 1). The expression of CHS was not affected by the El mutation. The steady-state level of CHI transcript was enhanced in the lobes in EIIEI plants, although no effect was observed in the tubes. This result was surprising because El reduces the amount of anthocyanin in flowers. Elalso affected the steady-state levels of F3H, DFR, Candi and UFGT, decreasing them all in both lobes and tubes. The effect of Elon mRNA levels was greatest for DFR and UFGT, where only about 10-20% of transcript was found in El plants compared to el plants. These effects probably account for the reduction in anthocyanin in €/flowers. The effect of El on F3H transcript was less dramatic (60%). However, because inc is semi-dominant, F3H may be the major rate-limiting step for anthocyanin accumulation in flowers (compared to DFR which does not become limiting until transcript levels drop below 30% of normal (Coen et a/., 1986)) and so the effect of El on inc may also be significant in modifying the amount of anthocyanin made. The influence of El on Candi transcripts was not as great as that on DFR and UFGT transcripts; whether this reduction could be sufficient to affect the biosynthesis of anthocyanin is not clear.
Figure 8. Northern blots showing hybridization of PAL, CHS, CHI, DFR, Candiand UFGT cDNA probes. Hybridization of probes to 5 pg poly(A+j RNA from lobes (Lj and tubes (T) of €/:De/+, €/:de/-, e/:De/' and e/:de/- flowers. The phenotypes are shown above. The hybridizationwas quantified using scanning densitometty (Table 1j.
required for the conversion of leucoanthocyanidin to anthocyanidin (Heller and Forkmann, 1988)), from those genes that we have analysed, Del would appear to influence the accumulation of steady-state transcripts of the structural genes from the later part of the anthocyanin biosynthetic pathway, so giving rise to the acyanic tubes. By analogy to the situation for DFR we suggest that this effect operates through a failure to transcribe these genes in the upper part of the tube (Almeida et a/., 1989).
Effect of Eluta on expression of anthocyanin biosynthetic genes We used the clones of the anthocyanin biosynthetic genes to analyse the effects of another potential regulatory gene, Eluta , that modifies the pattern of floral pigmentation.
Combined effect of El and del on anthocyanin biosynthesis and gene expression E1:del- plants were identified in the segregating F2 population of an EI/El:Del/Del x el/el:del/del cross, and compared to El:Del+, el:Del+ and e1:del- plants. E1:del- plants produce less anthocyanin than El:Del+. There is no anthocyanin in the tubes but synthesis is also more restricted in the lobes (about 40% of that in El:Del+ flowers). The double mutant therefore shows a synergistic interaction between El and del (Figure 1). In Elrdel- plants, a significantly higher level of PAL transcript was detected in both lobes and tubes, compared to all other genotypes (Figure 8, Table 1). No effect was observed on CHS transcript levels. The expression of CHI was the same in E1:del- and El:Del+ plants, with elevated levels of CHI transcript in the lobes. There was a slight decrease in the expression of F3H, Candi and UFGT in elrdel- lobes compared to El:Del+ plants, although we cannot be sure that these changes would have resulted in a decrease in anthocyanin production of over 50%. In tubes there was reduced transcript for all four genes,
Control of flower colour 43 reflecting the influence of del, but the levels of F3H and UFGT transcripts were not as low as in el:del-. plants (Figure 8, Table 1). Both El and del show major effects on the levels of steady-state transcripts of F3H, DFR, Candi and UFGT and these are in accordance with their effects on anthocyanin biosynthesis. It therefore seems likely that the phenotypes of these genes arise through their control of the transcript levels of the biosynthetic genes of the later part of the pathway. Expression of PAL, CHS and CHI genes may be influenced by del and El but the type of control is different from that exerted over F3H, DFR, Candi and UFGT; CHI expression in lobes appears to be enhanced by El and PAL expression is enhanced in E1:del- plants. However, neither of these last two effects are reflected in the phenotype of the flower.
Sequence comparison of the upstream regions of genes regulated by del and El The most likely means for del and N to control structural gene expression is through transcriptional activation andlor repression. We therefore aligned the regions immediately upstream of the coding sequences of F3H, DFR (Almeida et al., 1989; Coen et al., 1986) and Candi and compared them to the upstream sequences of CHS (Sommer and Saedler, 1986) and CHI from Antirrhinum to search for target motifs that might be binding sites for these regulators (Figure 9). Apart from the TATA and CAAT motifs found in most eukaryotic genes (Breathnach and Chambon, 1981) we were able to identify a number of motifs that were similar to the binding sites reported for c-myb or C1 in maize (which also carries the myb-binding domain (Biedenkapp etal., 1988; Wienand et a/., 1989)). There were also motifs that were similar to the c-myc binding motif (Sen and Baltimore, 1986) which may represent the binding sequence of R in maize (Goff et a/., 1990). No motifs common to F3H, DFR and Candi, but absent from CHS and CHI, were found which might have revealed the recognition sequences for delandlor El. However, because Elaffects CHI expression, albeit differently from the way it affects F3H, DFR and Candi, the promoter of the CHI gene could also contain binding motifs for this regulator. In the DFR promoter, a 79 bp region lying around the CAAT box (-202 to -123), called box C, has been proposed to bind the del gene product, and deletion of this region has a dramatic effect on DFR gene expression in the tube (Almeida et a/., 1989). The sequences most homologous to this region in the CHS gene were -153 to -85 (with 86% similarity), in the CHI gene were -34 to +27 (with 59% similarity), in the F3H gene were +79 to +122 (with 71 YOsimilarity),and in the candigene -71 to -25 (with 70% similarity), (all calculated by GAP with a gap weight of 3.0
-200 CHS CHI
ACCAAAAGSA AAATAAlGTA CATGGTTTGA GASGTCAGTA CTCTAAGA1'A C t i C G A C A A A A A C A T G A A T TAATATGTTG GTACCTAACC T C T G C T E - C L
F3H DPR Candi
AASTATAArA 4TTTATTTCT ATTGCAAGAC AACAA?TAC-T-hTMS CACTTATYA_G TTATTACAAC AGAATCMTT TCTACC$CLMTCI~ACGA CTTTAAASTA ~@Z~&V.GAG AAAGTTACTA AATTAAATAT CTCTTGGCAC
CHS CHI
~ A A T T C C A A C CCAISTCACGT~CC?Z??%??A CCCGTAGCTA AAGTTGTTGG C7ACGTACTA ATAATAATAA T A A T A A T M T AACAATACTA CCCAQC
![[Competition in Antirrhinum majus L].](/assets/img/hold.png)
