AtMYB32 is required for normal pollen development in Arabidopsis thaliana


(fax 61 394791188; e-mail


AtMYB32 gene is a member of the R2R3 MYB gene family coding for transcription factors in Arabidopsis thaliana. Its expression pattern was analysed using Northern blotting, in situ hybridization and promoter-GUS fusions. AtMYB32 is expressed in many tissues, but most strongly in the anther tapetum, stigma papillae and lateral root primordia. AtMYB32-GUS was induced in leaves and stems following wounding, and in root primordia by auxin. T-DNA insertion populations were screened and two insertion mutants were identified, both of which were partially male sterile, more than 50% of the pollen grains being distorted in shape and lacking cytoplasm. AtMYB4 is closely related to AtMYB32 and represses the CINNAMATE 4-HYDROXYLASE gene. Distorted pollen grains were produced in both AtMYB4 insertion mutant and overexpression lines. In an AtMYB32 insertion mutant, the transcript levels of the DIHYDROFLAVONOL 4-REDUCTASE and ANTHOCYANIDIN SYNTHASE genes decreased while the level of the CAFFEIC ACID 0-METHYLTRANSFERASE transcript increased. Change in the levels of AtMYB32 and AtMYB4 expression may influence pollen development by changing the flux along the phenylpropanoid pathways, affecting the composition of the pollen wall.


Transcription factors can be categorized into families based on their structure and target DNA binding sequences (Pabo and Sauer, 1992). The R2R3 MYB family of transcription factors is one of the largest discovered in plants to date and there are approximately 125 individual members in Arabidopsis alone (Riechmann et al., 2000; Stracke et al., 2001). Plant R2R3 MYB proteins are characterized by their N-terminal DNA binding (MYB) domain, which contains two highly conserved imperfect repeats (R2 and R3) of 51–52 amino acids in length. Each repeat contains three α-helices of which the second and third form a helix-turn-helix structure when bound to DNA (Ogata et al., 1994). In contrast, their evolutionary predecessors in animals generally contain three repeats (R1, R2 and R3) (Lipsick, 1996). While the MYB domain shows little variation among all R2R3 MYB proteins, their C-termini differ greatly and are only related through the conservation of small domains (Kranz et al., 1998) which are thought to be responsible for activation and repression of specific target genes (Jin and Martin, 1999). Attempts to classify MYB proteins based on the presence of conserved domains have revealed 22 distinct subgroups (Kranz et al., 1998; Stracke et al., 2001) and it is becoming evident that proteins within subgroups share a degree of functional conservation (Jin et al., 2000; Lee and Schiefelbein, 2001). The large size of the MYB family in plants indicates their importance in the control of plant-specific processes (Martin and Paz-Ares, 1997). Hence, MYB genes have been shown to be involved in the regulation of phenylpropanoid metabolism, identity and fate of plant cells, disease resistance, response to hormones and environmental stimuli (reviewed in Martin and Paz-Ares, 1997; Stracke et al., 2001). Moreover, it is possible that the expansion and diversification of the MYB family facilitated the adaptive radiation of flowering plants (Martin and Paz-Ares, 1997; Rabinowicz et al., 1999).

Plant R2R3 MYB proteins appear to be major regulators of phenylpropanoid metabolism (Figure 1). The maize C1 protein activates the transcription of four anthocyanin biosynthetic genes in the aleurone and scutellum of kernels (Martin and Paz-Ares, 1997) while its functional duplicate (PL) controls anthocyanin biosynthesis in leaves, stems and other vegetative tissues (Cone et al., 1993). Overexpression of AtMYB75/PAP1 in Arabidopsis caused upregulation of the phenylalanine ammonia lyase (PAL), chalcone synthases (CHS), dihydroflavonol reductase (DFR) and glutathione S-transferase (GST) genes and plants had increased levels of anthocyanins, flavonols, hydroxycinnamic acids and lignin (Borevitz et al., 2000). AtMYB90/PAP2 is a functional duplicate of AtMYB75. In seeds, another R2R3-MYB gene, TRANSPARENT TESTA2 (TT2/AtMYB123), has also been found to be an activator of DFR and the BANYULS (BAN) gene, which encodes an anthocyanidin reductase involved in the formation of proanthocyanidins in the testa (Nesi et al., 2001).

Figure 1.

Model of the phenylpropanoid biosynthetic pathway.
Enzymes are indicated in bold upper-case letters and the corresponding genetic loci in lower-case italics. Regulatory genes and mutants are in parentheses. Maize orthologues are indicated in upper-case italics. ANS, anthocyanidin synthase; AR, anthocyanidin reductase; ban, banyuls; CHI, chalcone isomerase; CHS, chalcone synthase; C3H, 4-coumarate 3-hydroxylase; C4H, cinnamate 4-hydroxylase; 4CL3, 4-coumarate CoA ligase 3; DFR, dihydroflavonol 4-reductase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′hydroxylase; F5H, ferulate 5-hydroxylase; FLS1, flavonolsynthase1; 3GT, anthocyanidin 3-glucosyltransferase; GST, glutathione S-transferase; LDOX, leucoanthocyanidin dioxygenase; COMT, caffeic acid o-methyltransferase PAL, phenylalanine ammonia-lyase; tt, transparent testa; ttg1, transparent testa glabra 1; A1, ANTHOCYANLESS1; C2, COLORLESS2; BZ, BRONZE.
Reproduced from diagrams contained in Nesi et al. (2000), Debeaujon et al. (2001), Shirley et al. (1995), Wisman et al. (1998), Bovy et al. (2002), Jaakola et al. (2002), Bartel and Matsuda (2003), Tamagnone et al. (1998), Jin et al. (2000), Schoch et al. (2001), Ruegger et al. (1999), Lehfeldt et al. (2000), Humpheys and Chapple (2002), Hemm et al. (2004) and Walker et al. (1999).

Overexpression of the tomato R2R3 MYB gene ANT1 caused upregulation of genes encoding proteins regulating early and later steps of anthocyanidin synthesis (CHS and DFR) and genes controlling glycosylation and transport of anthocyanins into the vacuole (Mathews et al., 2003). A high degree of conservation occurred in the R2R3 MYB domain of ANT1 and other MYB proteins involved in flavonoid accumulation, namely AtMYB75/PAP1, AtMYB90/PAP2, petunia AN2, and maize P1 and C1.

The P MYB gene in maize regulates phlobaphene biosynthesis (Grotewold et al., 1994), the Antirrhinum majus genes AmMYB305 and AmMYB340 regulate flavonol synthesis (Moyano et al., 1996) while AmMYB308 and AmMYB330 regulate hydroxycinnamic acid and monolignol biosynthesis in tobacco (Tamagnone et al., 1998). AtMYB308 downregulates the cinnamate 4-hydroxylase (C4H), p-coumaroyl 4-CoA ligase (4CL) and cinnamyl alcohol dehydrogenase (CAD) genes. AtMYB4 is an orthologue of AmMYB308 and also functions as a repressor, particularly of C4H (Jin et al., 2000). In the absence of UV-B light, AtMYB4 negatively modulates sinapate ester formation. UV-B light downregulates AtMYB4 expression and the enhanced levels of sinapate esters (hydroxycinnamic acid derivatives) in leaves leads to improved tolerance to UV-B treatment. Moreover, AtMYB4 is capable of regulating the flux along the hydroxycinnamic acid pathway and may represent an important control point. AtMYB4 belongs to the R2R3 MYB subgroup 4 proteins, which share the conserved motif pdLNLD/ELxiG/S and a putative zinc-finger domain at their C-termini (Kranz et al., 1998). The conserved motif in AtMYB4 has been shown to be partly responsible for repression (Jin et al., 2000). While C4H is common to both branches of phenylpropanoid biosynthesis, derepression of C4H by mutation of AtMYB4 did not result in increased production of flavonoid compounds, suggesting that other factors are responsible for their regulation.

Five MYB-related genes expressed in anthers have so far been identified, namely AtMYB26 and AtMYB103 in Arabidopsis (Higginson et al., 2003; Steiner-Lange et al., 2003), ZmMYBP2 in maize (Zhang et al., 1997) and NtMYBAS1 and NtMYBAS2 in tobacco (Yang et al., 2001). AtMYB103 is expressed in the tapetum of anthers and downregulation results in early tapetal degeneration and aberrant pollen, possibly by interfering with polyploidization (Higginson et al., 2003). Disruption of AtMYB26 prevents normal anther endothecial cell layer development, resulting in non-dehiscent anthers and male sterility (Steiner-Lange et al., 2003). NtMYBAS1 transactivated two different phenylalanine ammonia-lyase promoters (PALA and gPAL1) in tobacco leaf protoplasts (Yang et al., 2001). NtMYBAS1 may be a positive regulator of phenylpropanoid biosynthesis because the onset of its suppression coincided with the decay of tapetal cells and the suppression of gPAL1 expression.

We report here the characterization of an Arabidopsis R2R3-MYB gene, AtMYB32, which belongs to subgroup 4 (Kranz et al., 1998) and shows high similarity to AtMYB4. AtMYB32 is expressed in most tissues, but is strongest in the tapetum of anthers, the stigmatic papillae and developing lateral roots. Mutation of AtMYB32 leads to aberrant pollen and partial male sterility.


Identification and characterization of AtMYB32

The AtMYB32 gene encodes an R2R3 MYB-like protein comprising 274 amino acids with an estimated molecular weight of 31.5 kDa (Li et al., 1999). AtMYB32 contains two exons (263 and 564 bp) and a single intron (96 bp) in the R3 repeat and is a single copy gene that maps to the short arm of chromosome IV. The cloned 5′ promoter and 3′ non-coding regions extend −1775 bp upstream of the transcription start site, which was found to be 42 bp upstream of the translational start codon (ATG) and +1511 bp downstream of the translational stop codon respectively. Comparison of the entire cloned sequence from the Arabidopsis Ler accession with that from the Col-0 accession revealed 16 separate base substitutions, including two in the coding region, both of which are in the carboxy terminal coding sequence (199 F → L and 212 H → Y).

AtMYB32 has been placed in subgroup 4 by Kranz et al. (1998) under their R2R3 MYB classification scheme. Subgroup 4 members contain the conserved MYB domain motif LLsrGIDPxT/S HRxI/L at the end of R3 whose function is unknown and a putative negative regulatory domain (NRD) pdLHLD/LLxi G/S in their C-termini (Jin et al., 2000). AtMYB32 shares with some subgroup 4 members a putative zinc-finger domain (CX1−2 CX7−12 CX2C) in its C-terminus, suggesting interaction with other regulatory proteins to control transcription of target genes. Amino acid sequence alignment of subgroup 4 proteins (Figure 2) illustrates the high conservation of the MYB domains. In contrast, their C-termini are highly divergent outside of the conserved motifs.

Figure 2.

Amino acid alignment of AtMYB32 and other subgroup 4 MYB proteins (as classified by Kranz et al., 1998).
The alignment was carried out using the ClustalW program (Thompson et al., 1997). The line above indicates the R2 and R3 MYB repeats. Black background indicates 100% conservation; white and black letters on grey background indicate 80 and 60% conservation respectively. Motifs representing putative activation (L1srGIDPxT/SHRxI/L), repression (pdLNLD/ElxiG/S) and zinc-finger domains (CX1−2CX7−12CX2C) common to subgroup 4 MYB proteins are boxed. Upper-case letters in the boxes indicate amino acids found in all members of a subgroup, lower-case letters indicate conservation in more than 50% of the genes. If two amino acids are found at the same position, both are given and these are separated by a slash (Kranz et al., 1998). The serine-rich regions that may be subject to phosphorylation in AtMYB32 and AtMYB7 are underlined and residues that constitute the putative zinc-finger domain are boxed. Am, Antirrhinum majus; At, Arabidopsis thaliana; Hv, Hordeum vulgare; Le, Lycopersicon esculentum; Zm, Zea mays.

Expression patterns of AtMYB32

Total RNA was extracted from organs of 6-week-old wild-type (Col-0) plants and a 453-bp radiolabelled C-terminal coding sequence probe used to detect AtMYB32 expression by Northern blotting. Transcript was detected in all major organs (Figure 3a) and was most abundant in flowers, both young (unopened) and mature (opened). Weaker expression was detected in stems, roots, and rosette and cauline leaves. RT-PCR using intron-spanning primers also detected transcript in the same organs (data not shown). In situ hybridization of flower buds was carried out using the radiolabelled C-terminal coding ribonucleotide probe. Strong hybridization signal was detected in tapetum and the adjacent cell layers (Figure 3b). Hybridization signal also was found in anther connective tissue and petals.

Figure 3.

AtMYB32 expression.
(a) Northern analysis of AtMYB32 expression (top panel). Filter containing approximately 10 μg total RNA isolated from different Arabidopsis thaliana (Col-0) plant tissues probed with a 32P-labelled 453 bp AtMYB32 C-terminal coding region cDNA. Lr, rosette leaves; Lc, cauline leaves; Fy, young flowers unopened; Fm, mature flowers opened; R, roots and St, stems. Ethidium bromide staining of ribosomal RNA (bottom panel) shows equal loading.
(b, c) Localization of AtMYB32 transcript in floral sections using 35S radio-labelled antisense (b) and sense (c) probes. High levels of transcript are evident in the tapetum and the adjacent cell layers. anl, anther locule; pt, petal.

Promoter expression analyses were conducted using the β-GLUCURONIDASE (GUS) reporter gene. The −1775 bp 5′ (promoter) and the approximately 1562 bp 3′ non-coding region were fused to GUS and plants were transformed with the constructs. One construct contains the promoter/GUS (AtMYB32-5′:GUS) and the other promoter/GUS/3′ region (AtMYB32-5′:GUS:3′). There was no significant difference in the expression patterns obtained from the two constructs. Figure 4 shows the GUS expression patterns obtained in AtMYB32-5:GUS and AtMYB32-5:GUS:3′ transformant lines. A series of developing florets with excised sepals and petals demonstrate that GUS is expressed in the gynoecium and the stigmatic surface throughout floral development and expression decreases as the anthers approach dehiscence (Figure 4a). X-gluc and safranin-stained transverse sections show strong GUS expression in the papillae of the stigmatic surface (Figure 4h). A series of developing anthers representing stages 6 through to 13 (Figure 4b) (assigned according to Sanders et al., 1999) show that GUS expression commences between stages 7 and 8 and is induced strongly during the latter stages of anther development (stages 9–12) and declines as the anther approaches dehiscence. GUS expression can also be seen in the vascular tissue of filaments at stages 10 and 11. X-gluc and safranin-stained transverse sections show that GUS induction occurs in the tapetum during the post-meiotic microspore stage (stage 8) and maximal expression occurs during stages 9–12, just prior to tapetal degeneration. When the anther dehisces at stage 13, GUS activity is evident in the endothecium, connective tissue and pollen grains (Figure 4k). The GUS expression pattern in anthers is consistent with the results obtained from in situ hybridization of floral sections. GUS expression in mature cauline and rosette leaves from 5-week-old plants was weak (Figure 4d) and in most cases associated with the region of excision and wounding (Figure 4l,m), suggesting that the AtMYB32 promoter is responsive to wounding. A similar expression pattern was observed in the pedicel following floral excision. In strong AtMYB32-5:GUS:3′ lines (Figure 4e), GUS was expressed strongly both in young and mature leaves. Generally, GUS expression in leaves decreased as they matured and became confined to regions of excision.

Figure 4.

Expression pattern in Arabidopsis Col-0 obtained with the AtMYB32 5′-GUS and AtMYB32 5′-GUS-3′ (a, b, i, l, m) fusion constructs.
(a) Series of developing florets with sepals and petals removed. GUS expression occurs in the stigma and anthers.
(b) Series of developing anthers with approximate stages of development numbered according to Sanders et al. (1999). GUS induction commences between stages 7 and 8 and persists as the anther approaches dehiscence (stage 12).
(c) GUS expression in florets of various developmental stages (strongly expressing line).
(d, e) Seventeen-day-old seedlings from weak and strongly expressing lines. In both lines GUS expression is strong in developing lateral root primordia. In the strongly expressing line (e), GUS activity persists in the growing lateral roots and is strong in the leaves.
(f) Series of developing lateral roots. GUS induction occurs early during the commitment to lateral root initiation and is then confined to the growing root tip.
(g) GUS induction associated with lateral root initiation in a strong AtMYB32 5′-GUS expressing line.
(h) Longitudinal section of a gynoecium. Strong GUS expression can be seen in the papillae on the stigma surface.
(i–k) Cross sections of anthers at varying stages of development show that GUS expression is absent in the tapetum (tp) at stage 6 (i), and is maximal at stages 10 and 11 (j). After tapetal degeneration at stage 13 (k), GUS activity appears in pollen (arrow). GUS expression is also evident throughout the connective tissue from stages 7 to 13.
(l) GUS expression is associated with wounding in stems (arrows) and excision.
(m) Leaf wounded by compression with forceps. Strong GUS expression is associated with the area wounded and with excision. Bars = 50 μm.

In roots, localization of GUS expression correlates tightly with the commitment to lateral root initiation (Figure 4f,g). GUS induction is evident in pericycle cells prior to emergence of lateral roots and maximal GUS expression occurs during lateral root tip emergence, after which it becomes confined to the growing tip and cells at the lateral–primary root junction. GUS expression was not observed in the primary root tip.

Characterization of atmyb32 T-DNA insertion mutants

T-DNA-transformed Arabidopsis lines (WS) (Arabidopsis Knockout Facility, Wisconsin-Madison, USA) (Sussman et al., 2000) were screened for insertions in AtMYB32. An activation-tagged T-DNA insertion was identified in the C-terminal coding region of AtMYB32 (Figure 5, top) approximately 85 bp upstream of the translational stop codon (TGA), which correlates to a 28 amino acid truncation of the protein. The insertion mutant was designated atmyb32-1. While activation tagging can lead to overexpression of an adjacent gene (Weigel et al., 2000), the T-DNA insertion was positioned such that the 4x 35S enhancers and T7 promoter were oriented away (3′) from AtMYB32, thus AtMYB32 overexpression is highly unlikely (Figure 5 top). The translational start site of the gene immediately downstream of AtMYB32 (SUBTILISIN PROTEASE) is a considerable distance away from the T-DNA insertion (approximately 2231 bp). The initial insertion plant isolated was found to be heterozygous and displayed no observable phenotype. Plants from the subsequent generation were selected for BASTA resistance and segregated at a ratio of 3:1. The atmyb32 mutant was found to be homozygous recessive. Homozygous lines displayed a partially male sterile phenotype with more than 50% aberrant pollen (Figure 5c). Most of the collapsed pollen grains are devoid of cytoplasmic contents (Figure 5a) and form aggregates.

Figure 5.

Comparison of anthers and pollen from wild-type and atmyb32-1 T-DNA insertion mutant flowers. Position of the T-DNA insertion in the AtMYB32 gene (top).
(a, b) Light micrographs showing Alexander's staining of wild-type (b) and mutant (a) anthers. Wild-type pollen stains red inside and the outer wall (exine) stains yellow/green. Arrow (a) points to aberrant pollen that appears devoid of internal contents and consists of exine only. These aberrant pollen grains stain green. Bars = 30 μm.
(c, d) Scanning electron micrographs of pollen from mutant (c) and wild-type plants (d). Bars = 10 μm.
(e–j) Wild-type (f, h, j) and mutant (e, g, i) anthers were fixed and embedded in LR-white plastic resin, sliced into 10 μm transverse sections and stained with toluidine blue following methods outlined by Sanders et al. (1999). Developmental stages were assigned according to Sanders et al. (1999). No difference was detected between the wild-type and mutant tapetal layers. Stage 6 (e, f), the microspore mother cells enter meiosis and the tapetum becomes vacuolated. Stage 8 (g, h), the callose wall surrounding tetrads has degenerated and the individual microspores have been released. The tapetum has thickened significantly. Stage 10 (i, j), microspores in the mutant contain less cytoplasm than wild type. Msp, microspores; tp, tapetum. Bar = 50 μm.

Flowers from the coding region T-DNA insertion mutant were fixed, embedded and sectioned for analysis of anther development. While AtMYB32 expression is strong in the tapetum and several pollen mutants have been reported to exhibit tapetal abnormalities, the tapetum of atmyb32 mutants appeared normal and fully developed by stage 8 (stages defined according to Sanders et al., 1999) (Figure 5e–j). Comparison of mutant and wild-type anther sections at stage 6, when the microspore mother cell enters meiosis, stage 8, when the microspores are released from the tetrads and stage 10, when microspores become enlarged and tapetum degeneration commences, failed to reveal differences in tapetum morphology. However, at stage 10, the locules of the mutant contained fewer microspores, many appearing irregular in shape and containing less cytoplasm (i).

A second atmyb32 T-DNA insertion mutant (atmyb32-2) was obtained from the Salk collection. The insertion creates a 40 amino acid truncation in the AtMYB32 protein and the mutant produced collapsed pollen grains similar to the first mutant (data not shown).

Pollen morphology in AtMYB4 mutants

AtMYB4 and AtMYB32 are both included in the MYB subgroup 4 (Kranz et al., 1998). Hence, the pollen morphology of AtMYB4 insertion mutant and overexpression lines was examined and compared with that of atmyb32. The pollen grains of both insertion and overexpression lines exhibited an irregular and collapsed phenotype (Figure 6b,c). The abnormal pollen grains were partially or completely devoid of cellular contents as indicated by the reduced staining inside the pollen grains (Figure 6a). Hence, the pollen morphology of the atmyb4 plants and the atmyb32-1 mutant was similar. However, when pollen grains from the atmyb4 insertion mutant were treated with Alexander stain, they became ‘donut’ shaped (Figure 6a) whereas those from the atmyb4 overexpression lines and the atmyb32 mutant failed to do so.

Figure 6.

Pollen from wild-type and atmyb4 insertion and overexpression mutant plants.
(a) Light micrograph showing Alexander's staining of aberrant pollen grains from the atmyb4 T-DNA insertion mutant. Green staining indicates absence of cytoplasm; staining treatment result in donut-shaped pollen grains. Bars = 10 μm.
(b–d) Scanning electron micrographs of pollen from atmyb4 T-DNA insertion mutant (b), AtMYB4 overexpression mutant (c) and wild-type plants (d). Bars = 10 μm.

Identification of putative transcriptional targets of mutant AtMYB32

The similarity of AtMYB32 with several R2R3 MYB proteins shown to be involved in regulating genes encoding enzymes of phenylpropanoid metabolism, in particular AtMYB4 and AmMYB308 (Jin et al., 2000; Tamagnone et al., 1998), suggests that AtMYB32 may also regulate genes of this pathway. To test this hypothesis, we used quantitative RT-PCR to assess transcript accumulation in mutant and wild-type flowers for genes encoding enzymes controlling some of the major biochemical steps in hydroxycinnamic acid, lignin and flavonoid biosynthesis. Figure 7 shows duplicate PCR products that were detected by DNA gel-blot analysis and hybridization with their respective radiolabelled probes. Compared with the wild-type, transcript levels of PAL2 and C4H, the first two genes in the phenylpropanoid pathway (Figure 1), were unchanged in the mutant. C4H is negatively regulated by the putative homologue of AtMYB32, AtMYB4 (Jin et al., 2000). Transcript levels of COMT increased marginally in the mutant. COMT is thought to catalyse two steps in hydroxycinnamic acid biosynthesis (Figure 1) (Jin et al., 2000; Tamagnone et al., 1998; Zhang et al., 1997), but might also function as a novel flavonoid 3′-OMT (Muzac et al., 2000). There were no changes to the transcript levels of 4CL3, a 4-coumarate CoA ligase specific to the flavonoid branch of phenylpropanoid metabolism (Ehlting et al., 1999), or the flavonoid-specific CHS, F3H and F3H genes. The transcript levels of FLS1, which catalyses the synthesis of the flavonols quercetin and kaempferol from dihydroquercetin and dihydrokaempferol (Figure 1), were also unchanged in the mutant. In addition, the transcript levels of CCoAOMT (caffeoyl CoA O-methyl transferase) were also unchanged. However, the genes encoding the dihydroflavonol 4-reductase (DFR) and anthocyanidin synthase (ANS) enzymes showed substantially lower transcript accumulation in the mutant. DFR catalyses the conversion (Figure 1) of dihydroquercetin and dihydrokaempferol to leucocyanidin and leucopelargonidin (leucoanthocyanidins) respectively. ANS encodes a dioxygenase that operates downstream of DFR and catalyses the conversion of leucocyanthocyanidins to anthocyanidins (Figure 1).

Figure 7.

Quantitative RT-PCR analysis of transcript accumulation for AtMYB32 and genes involved in phenylpropanoid biosynthesis in flowers of wild-type WS (WT WS) and the atmyb32-1 T-DNA insertion mutant with WS genotype (atmyb32). Duplicate PCR products were detected by DNA gel-blot analysis and hybridization with their respective probes. The expression of β-TUBULIN transcript was used as an internal control. Transcript levels decreased in the mutant for the late-flavonoid biosynthetic genes DFR and ANS. Levels increased in the mutant for the COMT gene. Full-length AtMYB32 transcript was not detected in the mutant indicating homozygosity for the T-DNA insertion.

Quantitative RT-PCR was also used to assess the effect of the C-terminal T-DNA insertion on AtMYB32 transcript accumulation. Full-length AtMYB32 transcript was not detected in the mutant (atmyb32), thus confirming the mutant was homozygous for the T-DNA insertion.

Auxin enhances AtMYB32 promoter activity in developing lateral roots

The plant hormone auxin indole-3-acetic acid (IAA) is a principle inducer of lateral root formation (Boerjan et al., 1995; Celenza et al., 1995; Gray et al., 1998) and exogenous application of IAA stimulates excessive lateral root production (Blakely et al., 1982; Evans et al., 1994). Roots of transgenic lines containing the AtMYB32-5:GUS and AtMYB32-5:GUS:3′ constructs were tested for induction of GUS by auxin (IAA) and the auxin analogues 2,4-dichlorophenoxyacetic acid (2,4-D) and napthol acetic acid (NAA). Plants were grown on GM (see Experimental procedures) for 14 days, then transferred to GM containing the relevant compound and grown for a further 48 h. IAA, 2,4-D and NAA all triggered dramatic increases in GUS expression throughout the root and particularly in laterals (Figure 8a). However, GUS expression in aerial tissues was unaffected. Fluorometric quantification of GUS activity (Jefferson et al., 1987) in roots of transgenic lines treated with IAA, 2,4-D and NAA was conducted (Figure 8b). Greater than 10-fold increases in activity were stimulated by 25 μm IAA and NAA after 48 h and an almost fivefold increase was stimulated by 0.5 μm 2,4-D. The endogenous AtMYB32 transcript levels in roots were also examined following auxin induction treatment (Figure 8d). The transcript levels were increased in response to external auxin application.

Figure 8.

The effect of auxin on AtMYB32 expression.
Transgenic AtMYB32 5′-GUS plants were selected on kanamycin and transferred to GM media containing indole-3-acetic acid (IAA), 2,4-dichlorophenoxy acetic acid (2,4-D), napthol acetic acid (NAA) or no auxin (control). Controls were transferred to GM without kanamycin. Fourteen-day-old seedlings were treated for 48 h prior to GUS analysis.
(A) Seedlings were stained for GUS expression. (a) control, (b, e) 50 μm IAA, (c) 0.5 μm 2,4-D and (d) 50 μm NAA.
(B) GUS activity was measured as a function of the rate of nmol substrate cleaved per minute per gram fresh weight of roots. Vertical bars indicate standard deviation, n = 4.
(C) Plants were selected on kanamycin and transferred to GM media containing 50 μm indole acetic acid (IAA). GUS was analysed at intervals of 0, 1, 4, 8, 24 and 48 H.
(D) Northern analysis of AtMYB32 expression induced by naphthalene acetic acid (NAA). mRNA was isolated from root tissue of 16-day-old seedlings grown on GM for 2 weeks then transferred to GM media (1–3) or GM media containing 50 μm NAA (4–6) for 48 H. Each lane contains 10 μg of total RNA. The membrane was probed with (top) a 32P-labelled AtMYB32-specific probe and (bottom) a 32P-labelled β-tubulin-specific probe as loading control.

To assess the speed of the auxin response, a time-course assay was conducted for GUS activity in roots incubated with 50 μm IAA at 0-, 1-, 4-, 8-, 24- and 48-h intervals (Figure 8c). A significant increase in GUS activity was first detected after 8 h treatment and activity continued to rise, the greatest increase being between 24 and 48 h.


AtMYB32 expression patterns

AtMYB32 belongs to the subgroup 4 R2R3 family of plant MYB transcription factors (Kranz et al., 1998). Northern analysis and RT-PCR detected AtMYB32 transcript in all major organs, the highest levels being in flowers. GUS reporter gene analysis showed the 5′ promoter region directed particularly strong expression in anthers during the latter stages of their development (stages 7–12) and was most active in the tapetum. GUS activity in pollen grains is likely to reflect adsorption of the enzyme released from remnants of the tapetum following degeneration, as microarray analysis failed to detect AtMYB32 expression in mature pollen (J. Becker, Centre of Plant Biotechnology, University of Lisboa, Portugal, personal communication).In situ hybridization confirmed the strong tapetum expression. GUS expression in leaves was relatively weak and decreased as leaves matured. Wounding could induce it. GUS expression in roots was associated with the cells initiating lateral root formation and subsequently the lateral root tip. However, activity was not present in the primary root tip.

AtMYB32 and other subgroup 4 proteins

There are five subgroup 4 R2R3 MYB genes in Arabidopsis, namely AtMYBs 3, 4, 6, 7 and 32. The conserved motif involved in gene repression by AtMYB4 (Jin et al., 2000) is also present in AtMYB7 and AtMYB32. The putative zinc finger domain also occurs in the three proteins but mutation of its cysteine residues had no effect on gene repression by AtMYB4 (Jin et al., 2000). A serine-rich sequence is only found in AtMYBs 7 and 32. AtMYBs 3 and 6 possess none of the three sequences. AtMYB4 and the A. majus subgroup 4 genes AnMYB308 and AnMYB330 all downregulated the C4H, 4CL1 and CAD genes of hydoxycinnamic acid/monolignol biosynthesis when overexpressed in transgenic tobacco (Jin et al., 2000; Tamagnone et al., 1998). However, while AnMYB308 possesses the ‘repressor’ sequence, AnMYB330 does not, indicating different mechanisms of downregulation. Jin et al. (2000) suggest AtMYB4 may act as both a direct repressor and a competitive repressor, in the latter case displacing MYB proteins binding to the MYB motifs found in many genes of the phenylpropanoid pathway. Thus deletion of the entire AtMYB4 terminal region failed to completely prevent repression.

Although AtMYB4 represses the expression of the C4H gene, it does not seem to affect flavonoid biosynthesis, while the hydroxycinnamic acid/monolignol pathway is downregulated. Sinapoyl malate accumulates in atmyb4 plants and the mutant showed increased resistance to UV-B light. In wild-type plants exposed to UV-B, AtMYB4 expression was downregulated. C4H gene expression was unaffected in the Atmyb32 insertion mutant.

GUS reporter gene experiments showed transcription from the AtMYB32 promoter is activated by wounding. In contrast, AtMYB4 transcript levels decreased in leaves 4–6 h after wounding and systemically in non-wounded leaves from the same rosette (Jin et al., 2000). All the genes of the phenylpropanoid pathway are induced by wounding (Bell-Lelong et al., 1997; Ehlting et al., 1999; Mizutani et al., 1997) and the induction of AtMYB32 tends to support an activating rather than a silencing role.

To date, COMT is the only gene we have identified in the hydroxycinnamic acid pathway that is upregulated in the atmyb32 mutant. The DFR and ANS genes of the flavonoid pathway were downregulated. Both genes possess MYB-binding sites in their promoters, however this does not necessarily mean they are directly activated by AtMYB32 or that AtMYB32 functions as a transcriptional activator rather than a repressor. Jin et al. (2000) found that the caffeoyl CoA O-methyl transferase (CCoAOMT) gene was upregulated in overexpressing and downregulated in knockout mutants of atmyb4. However, AtMYB4 was unable to activate or repress the CCoAOMT promoter in trans-activation assays suggesting that changes in AtMYB4 expression indirectly affected CCoAOMT expression. The authors suggest the CCoAOMT gene responds to feedback regulation by hydroxycinnamic acid derivatives. We are currently assessing the ability of AtMYB32 to act on the DFR and ANS promoters using trans-activation assays. DFR is known to be activated by R2R3 MYBs in maize (C1, PL and P) (Cone et al., 1993; Grotewold et al., 1994), Petunia (AN2 and AN4) (Quattrocchio et al., 1998; Spelt et al., 2000) and thought to be activated in Arabidopsis by AtMYB75/PAP1 and AtMYB90/PAP2 (Borevitz et al., 2000) and TT2/AtMYB123 (Nesi et al., 2001). In Arabidopsis, AtMYB75/PAP1 and AtMYB90/PAP2 appear to be regulators of DFR, PAL and CHS throughout the whole plant (Borevitz et al., 2000). TT2/AtMYB123 is believed to be a seed- specific activator of DFR and plays a role in the regulation of proanthocyanidin biosynthesis (Nesi et al., 2001).

Pollen development is affected in the AtMYB32 mutant

Pollen development is aberrant in both the atmyb32 and atmyb4 plants. Pollen grains are partially or completely devoid of cytoplasm (deflated). The effect is more pronounced in atmyb32 than atmyb4 pollen. The atmyb32-1 insertion is 28 amino acids from the C-terminus and the inserted DNA would add four amino acids (derived from the T-DNA) to the truncated protein. RT-PCR confirmed that the full-length mRNA was no longer present. Back crossing of the insertion mutant with wild-type plants indicated that only the AtMYB32 insertion and no other is present.

Recently, two other R2R3 MYB genes have been implicated in male fertility in Arabidopsis. Mutation of the floral-specific AtMYB26 was found to cause male sterility through a defect in anther dehiscence (Steiner-Lange et al., 2003). However, pollen formation in atmyb26-2 mutants was not affected and pollen grains remained viable following mechanical release from anthers. It was discovered that in mutant anthers, endothecial cell walls failed to undergo the necessary lignification required for dehiscence. In the second case, downregulation of a tapetally expressed R2R3 MYB gene AtMYB103, using both co-suppression (sense) and antisense silencing approaches caused the production of aberrant pollen (deflated pollen) (Higginson et al., 2003). The phenotype was associated with premature degeneration of the tapetum and in addition, overbranched trichomes. AtMYB103 may regulate cell division events during tapetum and trichome development.

To date, the majority of pollen mutants characterized in Arabidopsis have been found to affect either cell division and/or tapetum development and stability. The defective pollen 3 mutant generates fragmented chromosomes during meiosis resulting in aberrant microspores (Sanders et al., 1999). Thereafter, the anthers develop fully and undergo a normal dehiscence process. Like atmyb32 mutant pollen, the grains have normal exine wall sculpturing but exhibit a collapsed morphology. Mutation of the SIDECAR POLLEN (SCP) gene also results in collapsed pollen grains similar to those of atmyb32 mutants (Chen and McCormick, 1996). SCP is involved in cell division during pollen development, as is the MALE STERILITY 1 (MS1) gene, which encodes a PHD-finger transcription factor that regulates meiosis and microspore release (Wilson et al., 2001). The microspores of the ms1 mutant adhere to each other, collapse and their cytoplasm degenerates. In addition, the tapetal cells become abnormally vacuolated and, as the anther approaches dehiscence, the aberrant pollen and tapetum degenerate to form a mass of undifferentiated cells. Similarly, silencing of the TAPETUM DEVELOPMENT ZINC FINGER PROTEIN 1 (TAZ1) gene from Petunia, which also exhibits tapetum-specific expression, results in abortion of the pollen (collapsed pollen) at the microspore stage and premature degeneration of the tapetum (Kapoor et al., 2002). Aberrant pollen grains from taz1 mutants also possess a lower flavonol content. The ABORTED MICROSPORES (AMS) gene coding for a MYC class transcription factor plays a crucial role in tapetal cell development and the post-meiotic transcriptional regulation of microspore development in Arabidopsis (Sorensen et al., 2003). Disruption of the gene causes premature tapetum and microspore degeneration. The TAPETUM DETERMINANT 1 (TPD1) gene and the EXCESS MICROSPOROCYTES/EXTRA SPOROGENOUS CELLS (EMS1/EXS) gene control somatic and reproductive cell fates in the Arabidopsis anther (Canales et al., 2002; Yang et al., 2003; Zhao et al., 2002). Disruption of TPD1 and EMS1/EXS genes caused the precursors of tapetal cells to differentiate into microsporocytes. Their anthers lack a tapetum and contain extra microsporocytes which do not undergo cytokinesis.

While the tapetum-specific expression of AtMYB32 correlates closely with the expression patterns of AtMYB103, TAZ1, MS1, MS2 and AMS, aberrant pollen in atmyb32 mutants was not associated with premature tapetal degeneration. The function of the tapetum is to supply specific compounds to the developing pollen grains such as phenolic and lipid derivatives, which form important components of the pollen wall (Piffanelli et al., 1998; Wiermann and Gubatz, 1992). Failure by the tapetum to produce these compounds can lead to the formation of aberrant pollen. The MS2 gene encodes a fatty acid reductase involved in the synthesis of the lipid component of sporopollenin, which is the major constituent of the outer layer or exine of the pollen wall (Aarts et al., 1997). In ms2 mutants, microspores collapse and eventually degenerate. The pollen grains fail to develop an exine (outer wall). In the Arabidopsis dex1 mutant, pollen grains collapsed totally because normal primexine development is disrupted, ultimately affecting microspore plasma membrane conformation and sporopollenin deposition (Paxson-Sowders et al., 2001). Apparently, the integrity of the pollen wall is affected in atmyb32 and these other pollen mutants. Given the tapetum is intact in atmyb32 mutants, it is possible that a particular tapetum-derived compound required for normal pollen formation is depleted.

Sporopollenin is a bipolymer made from polymerized lipid and phenolic derivatives (Piffanelli et al., 1998; Wiermann and Gubatz, 1992). The phenolic polymer is composed of coumaric acid and ferulic acid, which are synthesized on the hydroxycinnamic acid branch of the phenylpropanoid pathway (Figure 9). The enzymes of this pathway are highly active in the tapetum and phenylalanine is incorporated into sporopollenin (Wiermann and Gubatz, 1992). Flavonoids are also important components of the pollen coat (Figure 9) and are involved in pigmentation, structure, and protection against UV radiation and pathogen attack (Piffanelli et al., 1998; Wiermann and Gubatz, 1992). In tobacco, PAL activity in the tapetum is required for normal pollen formation, as the expression of sense and antisense PAL transgenes driven by a tapetum-specific promoter resulted in the production of irregularly shaped pollen grains (Matsuda et al., 1996). These pollen grains were lacking in flavonols. Given that phenylalanine is incorporated into sporopollenin and the pollen of pal mutants has reduced flavonol content, PAL is involved in pollen development via both the hydroxycinnamic acid/monolignol and flavonoid branches of the phenylpropanoid pathway (Figure 9). Other studies have shown that in tobacco, petunia and maize, flavonols are essential for pollen development and fertility, and disruption of CHS causes male sterility (Franken et al., 1994; Mo et al., 1992; Napoli et al., 1999; Vogt et al., 1994; Ylstra et al., 1992). However, in Arabidopsis flavonols are not essential for pollen formation or fertility (Burbulis and Winkel-Shirley, 1999; Ylstra et al., 1996). Indeed, downregulation of genes encoding flavonol-specific enzymes (CHS, F3H, F3′H and FLS) was not observed in the atmyb32 mutant. Preliminary staining of atmyb32 mutant pollen grains according to Ylstra et al. (1996) failed to detect changes in flavonol content (data not shown). The genes encoding DFR and ANS were substantially downregulated, but the DFR deletion mutant tt3 (Shirley et al., 1995) has normal pollen (results not shown). Hence, the aberrant pollen function in the atmyb32 mutant cannot be due solely to the downregulation of the DFR gene.

Figure 9.

Scheme showing phenylpropanoid biosynthetic pathways in the tapetum and their relationship to pollen wall formation. The phenylpropanoid pathway splits into two main branches leading to hydroxycinnamic acids and flavonoids. Tapetally synthesized 4-coumaric acid and ferulic acid polymerize to form the phenolic component of sporopollenin which is the major structural component of the exine. Flavonoids are synthesized from both phenylalanine and malonyl-CoA and are deposited on the developing pollen grain as part of the pollen coat. The role of flavonoids is most likely one of protection against UV, pathogen attack and could also be structural.

Why do changes in AtMYB32 and AtMYB4 expression affect pollen development?

Although changes in the expression levels of AtMYB32 and AtMYB4 lead to a similar pollen phenotype, namely loss of cytoplasm and collapse of pollen, they affect the expression of different genes in the phenylpropanoid pathway. In the atmyb4 insertion mutant C4H gene expression was increased and CCoAOMT transcript levels decreased (Jin et al., 2000). In AtMYB4 overexpression lines the CCoAOMT gene had increased transcript levels while those of PAL2, F5H, COMT and CAD1 were unaffected (Jin et al., 2000). However, the CHS, 4CLI and 4CL3 genes were downregulated by the higher activity of AtMYB4. Hence, new genes appear to be repressed by AtMYB4 when the transcription factor is at high concentrations. Nevertheless, both downregulation and upregulation of AtMYB4 levels interfere with normal pollen development, resulting in a similar pollen phenotype. The only obvious difference was that atmyb4 mutant pollen grains became donut shaped when treated with Alexander stain whereas those of the AtMYB4 overexpression mutant did not. AtMYB32 downregulation increased transcript levels of the COMT gene and reduced the levels of the DFR and ANS genes. However, the expression of the CCoAMT and C4H genes was unaffected.

An explanation of the results could be that variations in the levels of MYB proteins significantly influence the flux along the phenylpropanoid pathways. Changes in the concentrations of the different metabolites then interfere with normal pollen development. The structural components of the pollen wall are probably changed (Figure 9). As sporopollenin consists of polymerized phenols and fatty-acid derivatives, its composition is the most likely to be affected. The majority of sporopollenin in the pollen wall originates in the tapetum where AtMYB32 is strongly expressed, and the pollen grains of the two Arabidopsis mutants defective in sporopollenin (ms2 and dex1) eventually collapse (Aarts et al., 1997; Paxson-Sowders et al., 2001).

AtMYB32 expression and lateral root formation

GUS reporter gene studies suggest AtMYB32 expression occurs in lateral root primordia. The GUS activity appears very early during lateral root initiation, probably even before stage VIII when periclinal divisions of the pericycle cells create a three-layered primordium (data not shown).

Auxin is a key signal for lateral root initiation and many lateral root mutants also exhibit auxin-related defects (Casimiro et al., 2001). Exogenous application of auxin stimulates lateral root formation (Evans et al., 1994) and roots treated with the auxin transport inhibitor N-1-naphthylphthalamic acid (NPA) fail to initiate laterals (Casimiro et al., 2001). GUS expression studies showed the AtMYB32 promoter is strongly activated in roots following treatment with IAA and the auxin analogs 2,4-D or NAA. Increased GUS activity was first detected 8 h after adding auxin and the strongest increase was between 24 and 48 h. Following 24 h incubation with IAA, most developing lateral root primordia had only undergone one to three periclinal divisions, generating primordia of two to four cell layers (Laskowski et al., 1995). The first new lateral roots appear 13 h after the commencement of auxin treatment and by 36 h, 50% of new laterals are initiated (Taylor and Scheuring, 1994). The maximum rate of induction was between 24 and 48 h (0.7 initials mm−1 h−1).

Whether the auxin effect on AtMYB32 expression is direct or indirect remains to be seen. However, the AtMYB32 promoter contains a putative auxin-responsive element (GCATGTCGCCATT) similar to the GCA box sequence and located −338 bp from the transcriptional start site. Studies have shown that the GCA box (GCACGTATCCATT) is tightly linked to auxin induction of the NOS gene of agrobacterium T-DNA (An et al., 1985) and the tomato RSI-1 gene (Taylor and Scheuring, 1994).

No obvious root phenotype was observed in the atmyb32 mutant. Lateral root initiation mutants are rare and Malarny and Ryan (2001) suggest there may be only a few master genes regulating root branching while many are involved in pathways that inhibit or stimulate branching according to environmental signals. Such pathways would allow the plant to fine-tune root system morphology and interference with them would only be detectable under certain specific conditions. AtMYB32 may be involved in such a pathway. Alternatively, it may be functionally redundant in roots.

Flavonoids accumulate at secondary root primordia during root development in Arabidopsis (Peer et al., 2001). Basipetal auxin transport is enhanced in the absence of flavonoids (tt4 mutant) and reduced in the presence of excess kaempferol and quercetin (tt3 and tt7 mutants) (Peer et al., 2004). However, root growth and other auxin-related growth phenotypes are barely affected in tt3 and tt7, possibly because excess flavonols cause expanded PIN1 protein distribution, which, in part, compensate for inhibition of auxin transport (Peer et al., 2004).

Experimental procedures

Plant materials and growth

Arabidopsis thaliana accessions Columbia (Col-0), Landsberg erecta (Ler) and Wassilewskija (WS) were used as wild-type controls where appropriate. Ler tissue was used for in situ hybridization experiments, Col-0 for all gene transfer experiments as it can be transformed more efficiently than Ler. An AtMYB32 T-DNA insertion mutant line (atmyb32-1) was isolated from the BASTA populations (Sussman et al., 2000). Seed was obtained from the Arabidopsis Knockout Facility (super pool 36, plate 66, BIO, University of Wisconsin-Madison, USA). Another AtMYB32 T-DNA insertion line (atmyb32-2, Salk 132874, was obtained from the SALK population (Alonso et al., 2003). Seed was supplied from the Arabidopsis Biological Resource Center (Ohio State University, USA). AtMYB4 insertion (atmyb4-) and overexpression (35S-AtMYB4)lines (Jin et al., 2000) were supplied by Cathy Martin (John Innes Centre, UK). Plants were grown at 22°C under constant light in either soil or on germination media (GM) containing the appropriate selective antibiotic.

Stock solutions (50 mm) of photohormones IAA and 2,4-D were dissolved in 100% ethanol. NAA was dissolved in 1 m NaOH to a final concentration of 1 m NAA. Stock solutions were added to GM to a final concentration of 50 μm IAA, 0.5 μm 2,4-D or 50 μm NAA. Control GM plates containing the appropriate amount of either 100% ethanol or 1 m NaOH were made for each photohormone treatment. Fourteen-day-old seedlings grown on GM plates with kanamycin were transferred to either GM plates containing a photohormone (IAA, 2,4-D or NAA) or the appropriate control GM plate containing either ethanol or NaOH. After 48 h plants were either stained for GUS expression or total RNA was extracted for northern analysis. Germination medium [(4.606 g murashige and Skoog salt mixture, 10 g of sucrose and 0.5 g 2(N-Mopholino)ethanesulfonic acid (MES) in 1 l of water pH5.7 with 1 ml of potassium hydroxide, 2 g phytogel] was then autoclaved. Prior to pouring 1 ml of filter-sterilized 100× vitamin stock: 100 g l−1 inositol, 0.5 g l−1 nicotinic acid, 0.5 g l−1 pyridoxine and 1.0 g l−1 thiamine HC1 was added.

Plasmid construction

AtMYB32-5′:GUS.  The AtMYB32-5:GUS reporter gene construct was produced by fusing approximately 1775 bp -Bbs1 AtMYB32 5′ promoter fragment to the GUS coding sequence in the pBI101 binary vector (Clontech, Palo Alto, CA, USA) at the HindIII and SmaI restriction sites. The resultant plasmid produces a fusion protein containing 24 N-terminal amino acid of AtMYB32 fused to the GUS protein.

AtMYB32-5′:GUS:3′.  An approximately 1562 bp AtMYB32 3′ non-coding region fragment was amplified from Arabidopsis Col-0 genomic DNA using the primers 3′SacI-FWD (5′-CGATGAGCTCGGTCTAAACAACACT-3′) and 3′SacI-REV (5′-CTAGGAGCTCGTCGACACAATTAC-3′) which added 5′ and 3′SacI sites, respectively. The fragment was then cloned into the SacI restriction site of the AtMYB32-5:GUS plasmid, immediately downstream of the GUS gene. Correct orientation was confirmed by restriction digest and sequencing.

Plant transformation

Arabidopsis thaliana (accession Col-0) was transformed by vacuum infiltration with Agrobacterium tumefaciens strain GV3101. T0 seeds were sown on germination medium containing 50 μg ml−1 kanamycin sulphate and grown under constant light (80 μE m−2 s−1) at 22°C. At least 20 transformant lines carrying each construct were analysed. Successful incorporation of the constructs in each line was confirmed by PCR amplification of the insert from genomic DNA using primer pairs specific for each construct. The primer pair pBI101-FOR (5′-TGTGGAATTGTGAGCGGATA-3′) and pBI101-REV (5′-ATTCCACAGTTTTCGCGATC-3′) located 5′ and 3′ of the pBI101 binary vector multicloning site were used for confirmation of AtMYB32-5:GUS transformants. The NOS-REV primer (5′-GAATCCTGTTGCCGGTCTTG-3′) and 3′SacI-FWD were used to detect the 3′ non-coding region fragment of the AtMYB32-5:GUS:3′ construct.

Screening for an insertion mutant of AtMYB32

Arabidopsis T-DNA insertion populations (Arabidopsis Knockout Facility, University of Wisconsin-Madison, WI, USA; were screened by PCR using the gene-specific primers 32-5′ (5′-AGCCATATAGCGTTTTCTCATTGAAACTG-3′) and 32-3′ (5′-AGAAAAGTTGAGAACCATGAACGAGTCTG-3) in combination with the T-DNA left border primer JL-202 (5′-CATTTTATAATAACGCTGCGGACATCTAC-3′). Positive hits were confirmed by Southern blotting using an α-32P-labelled AtMYB32 cDNA probe amplified by PCR using the 32-3′ and 32-5′ primers.

RNA isolation, Northern analysis and in situ hybridization

Total RNA was isolated from tissue of mature soil grown Arabidopsis plants and root tissue grown hydroponically for 14 days using an RNeasy plant mini kit (Qiagen Germantown, MO, USA) as per the manufacturer's instructions. For Northern analysis, aliquots (10 μg) of RNA were separated on a 1% agarose gel containing 2.2 m formaldehyde and transferred to a Hybond-N membrane (Amersham, Buckinghamshire, UK). A 453 bp α-32P dCTP-labelled cDNA fragment corresponding to the AtMYB32 C-terminal coding region prepared according to the manufacturer's instructions (Amersham) was used a probe. Following 2–3 h pre-hybridization, membranes were hybridized overnight in Church buffer [0.3 m Na2HPO4 at pH 7.2 with H3PO4, 1% BSA (w/v), 7% SDS (v/v), 0.83 mm EDTA pH 8.0] (Church and Gilbert, 1984) at 65°C. Following incubation, membranes were rinsed in 2 × SSC/0.1% SDS (1x SSC = 0.15 m NaCl, 0.015 m sodium citrate), and then washed under low stringency conditions (2x SSC/0.1% SDS) at 65°C for 20 min, and twice under stringent conditions (0.2x SSC/0.1% SDS) for 10 min at 65°C. Detection was performed using autoradiography and X-OMAT AR film (Kodak, Melbourne, Australia). The in situ hybridizations were carried out essentially as described by Drews et al. (1991) and Cox and Goldberg (1988). Flowers from soil grown plants were fixed in 50% ethanol, 5% acetic acid and 3.7% formaldehyde. Fixed flowers were dehydrated and embedded in paraffin (Sigma, Saint-Louis, MO, USA) then sliced into 10 μm sections and attached to Superfrost*/plus slides (Menzel-Glaser, Braunschweig, Germany). The sections were treated with xylene followed by hydration, proteinase K treatment, acetylation and dehydration. The 35S-labelled probes were hydrolysed to about 100 nt in length and hybridized to the sections at 42°C for 17 h. The sections were then treated with ribonuclease A and washed, followed by emulsion of the slides. A 370-bp PvuII-VspI fragment encoding the C-terminus of AtMYB32 was cloned into pBluescript II for the production of a riboprobe. RNA probes were synthesized from the linearized vector using T7 and T3 RNA polymerases, respectively, in the presence of α-35S UTP.

Quantitative RT-PCR analysis

Total RNA (10 μg) isolated from florets of mature, soil-grown plants was treated with DNase I (Promega, Madison, WI, USA) to remove residual DNA and cleaned using an RNeasy plant mini kit (Qiagen). First strand cDNA was synthesized from 5 μg of RNA using poly (dT12−18) primer, Superscript II reverse transcriptase and accompanying reagents according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA) for 50 min at 42°C. Genes of interest were amplified from cDNA using the following conditions: 94°C/1 min, (94°C/30 sec, 52°C/1 min, 72°C/1 min) × viable number of cycles. Aliquots of 5 μl were taken out after the 72°C/1 min extension at the following number of cycles for the following genes: PAL2, 22; C4H, 17; OMT1, 18; CCoAOMT, 19; 4CL3, 17; CHS, 14; F3H, 20; F3H, 20; FLS1, 18; DFR, 15; ANS, 22; AtMYB32, 27. These numbers of cycles corresponded to approximately six cycles earlier than when the product is normally visible on gels. The samples were separated on 1% agarose gels and subjected to Southern analysis [as per Northern Analysis, except hybridization was performed using APH solution (5 × SSC, 5× Denhardt solution, 1% SDS, 100 μg ml−1 sheared salmon sperm DNA) at 68°C]. The genes were probed with their corresponding cDNA fragments labelled with α-32P dCTP (as previously described). For mRNA detection of the genes of interest, the primers used were as follows: for PAL2 (GenBank accession number L33678) PAL2-FWD (5′-GAGGCAGCGTTAAGGTTGAG-3′) and PAL2-REV (5′-TTCTCGGTTAGCGATTCACC-3′), for C4H (U71080) C4H-FWD (5′-TTCACCGGATCTAACCAAGG-3′) and C4H-REV (5′-CGTTGATTTCTCCCTTCTGC-3′), for OMT1 (U70424) OMT1-FWD (5′ ACGGCGGAGACACAATTAAC-3′) and OMT1-REV (5′-GCATCTTCGATGACATGTGG-3′), for CCoAOMT (L40031) COMT-FWD (5′-TCAGCGTATCCAAGAGAGCA-3′) and COMT-REV (5′-GCGTGATACCATCTCCAATG-3′), for 4CL3 (AF106087) 4CL3-FWD (5′-CGCTGATCACTACCGATGAA-3′) and 4CL3-REV (5′-AAGGATGGCTTGAGGGAGAC-3′), for CHS (M20308) CHS-FWD (5′-GGCTATTGGCACTGCTAACC-3′) and CHS-REV (5′-GGGTTTCTCTCCGACAGATG-3′), for F3H (U33932) F3H-FWD (5′-AAAAAGAGGAGAGATCTGCCG-3′) and F3H-REV (5′-TGAGGGCATTTTGGGTAATAA-3′), for F3H (AH009204) F3′H-FWD (5′-CATGGCAACTCTATTTCTCAC-3′) and F3′H-REV (5′-CGTCACCGTCAAGATCAGTTCC-3′), for FLS1 (U84258) FLS-FWD (5′-ATTCTTCGAGCTTCCTTCGTC-3′) and FLS-REV (5′-CCAGGAACTTCGTTAGGAACA-3′), for DFR (AY042880) DFR-FWD (5′-GTCGGATCCAGTTTCATCGT-3′) and DFR-REV (5′-CTTCATTGTTTGGGTTGCCT-3′), for ANS (AY062643) AS-FWD (5′-ACCAAACCGCTATGTGAAGC-3′) and AS-REV (5′-ATTTAATGGTGACCCAGCCA-3′), β-TUBULIN BTUB-FWD (5′-GGACACTACACTGAAGGTGCTGAG-3′) and BTUB-REV (5′-GGCTCTGTATTGCTGTGATCCACG-3′) and for AtMYB32 full-length 32-RT-FWD (5′-ATAAAGCCCTAATTTCTTC-3′) and 32-RT-REVII (5′-TCACAATTCACAGCTATAAATT-3′).

Floral sectioning

Whole flowers were excised and submerged in 1% glutaraldehyde in 50 mm sodium phosphate buffer (pH 7.4) and placed under vacuum for 2–5 min. Flowers were then washed in the same buffer and either submerged in X-gluc for GUS detection in transgenic plants (see below), or post-fixed in 6% glutaraldehyde with 4% paraformaldehyde in 50 mm sodium phosphate buffer for 2 h. Flowers were then bleached for 15 min through the following ethanol series: 30, 60, 70, 90 and 100% and then infiltrated with LR White medium grade resin for 1 h each of 30% LR White/70% ethanol, 70% LR White/30% ethanol and then 100% LR White overnight. Flowers were placed into capsules and incubated in 100% resin overnight at 65°C. Sections (approximately 10 μm) were cut using a microtome and counterstained with either safranin (for GUS detection) or toluidine blue (for tapetum analysis) (Sanders et al., 1999). Toluidine blue is a polychromatic dye that stains different cell components a different colour.

Histochemical staining of transgenic plants

For detection of GUS expression, transgenic plant material was submerged in 0.5 mg ml−1 5-bromo-4-chloro-3-indoyl glucuronide (X-gluc) in 50 mm phosphate buffer (pH 7.0) in 0.05% Triton X-100 (w/v) and incubated overnight at 37°C (Jefferson et al., 1987). Chlorophyll was removed by incubating the tissue for several hours in increasing concentrations of ethanol. The tissue was examined using an Olympus SZ-PT dissecting microscope (Olympus, Tokyo, Japan).

Pollen analysis

Pollen grains from soil grown plants were examined using a JOEL JSM 6340F field emission scanning electron microscope (Tokyo, Japan). Dry pollen grains were examined at ambient temperature. Images were digitally captured at a working distance of 27 mm at 2.0 kV. For light microscopic analyses, pollen grains were mounted in a drop of Alexander's (1969) stain, a vital stain for pollen, under a coverslip. Floral sections were mounted in several drops of clearene (Surgipath, Richmond, IL, USA). All specimens were viewed using an Olympus BH-Z compound light microscope. Both compound and dissector microscope images were captured on Kodak Ektachrome 64T slide film using an Olympus SC35 camera. Digital Images were processed using Adobe Photoshop 6.0 (Adobe Systems Inc.)

Fluorometric quantification of GUS activity

The detection of the GUS enzyme in roots of transgenic lines was carried out according to procedures outlined in Jefferson et al. (1987). Roots from AtMYB32-GUS plants grown in solid or liquid GM containing one of the hormones IAA, NAA, 2,4-D or none (controls) were snap-frozen in liquid nitrogen, homogenized and GUS protein extracted in 1 ml of GUS extraction buffer (0.1 m K2HPO4/KH2PO4 pH 7.0, 1 mm DTT, 1 mg ml−1 BSA) per 0.2 g of tissue. The homogenate was spun at 15 000 g for 5 min. The supernatant was further diluted in GUS extraction buffer in a total volume of 180 μl and added to wells of a microtitre plate containing 20 μl of 4-methyl umbelliferyl glucuronide (MUG) assay buffer (10 mm MUG in GUS extraction buffer). The fluorescence levels of four replicates for each extract were measured in a Fluoroskan (Labsystems, Vantaa, Finland) utilizing the Genesis II software package (Labsystems) and a kinetic measurement protocol. Calibrations were carried out using sodium methyl-umbelliferone standards. Standard deviation was calculated for replicates.


Sequence analyses were performed using the blast facility at the National Center for Biotechnology Information (NCBI;


We thank Cathie Martin of the John Innes Centre, Norwich, for the AtMYB4 insertion and overexpression mutants. The project was supported by the Grains Research and Development Corporation.