Abstract
Members of the R/B basic helix-loop-helix (bHLH) family of plant transcription factors are involved in a variety of growth and differentiation processes. We isolated a dominant mutation in an R/B-related bHLH transcription factor in the course of studying Arabidopsis tryptophan pathway regulation. This mutant, atr2D, displayed increased expression of several tryptophan genes as well as a subset of other stress-responsive genes. The atr2D mutation creates an aspartate to asparagine change at a position that is highly conserved in R/B factors. Substitutions of other residues with uncharged side chains at this position also conferred dominant phenotypes. Moreover, overexpression of mutant atr2D, but not wild-type ATR2, conferred pleiotropic effects, including reduced size, dark pigmentation, and sterility. Therefore, atr2D is likely to be an altered-function allele that identifies a key regulatory site in the R/B factor coding sequence. Double-mutant analysis with atr1D, an overexpression allele of the ATR1 Myb factor previously isolated in tryptophan regulation screens, showed that atr2D and atr1D have additive effects on tryptophan regulation and are likely to act through distinct mechanisms to activate tryptophan genes. The dominant atr mutations thus provide tools for altering tryptophan metabolism in plants.
PLANTS are capable of integrating a wide range of tissue, developmental, and environmental signals to achieve a complex pattern of gene expression. Immediate effectors of this spatial and temporal regulation include members of the R/B-related basic helix-loop-helix (bHLH) family. The R and B factors are two highly related maize proteins that control transcription of pigment biosynthesis genes. These factors are the founding members of a large class of plant proteins characterized by conserved sequences near the amino terminus including an acidic domain plus a bHLH DNA binding/dimerization domain near the carboxy terminus (Purugganan and Wessler 1994). R/B bHLH family members have been shown to regulate such diverse processes as cell shape determination, tissue-specific pigment accumulation, and adaptation to the environment.
Extensive genetic and biochemical studies have shown that maize R and B factors require a Myb cotranscription factor, C1 or Pl, to activate target gene expression leading to purple anthocyanin pigmentation (reviewed in Molet al. 1998). Genetic studies have revealed a more complex system in petunia, where the AN1 bHLH factor plus the AN2 Myb factor and a WD-repeat factor AN11 are all required for floral pigmentation (de Vettenet al. 1997; Molet al. 1998; Quattrocchioet al. 1999; Speltet al. 2000). In Arabidopsis, the GLABRA3 (GL3) bHLH factor acts along with the GLABRA1 (GL1) Myb-related factor, plus a WD-repeat-containing factor TRANSPARENT TESTA GLABRA1 (TTG1), to promote the formation of trichomes (leaf hairs; Oppenheimeret al. 1991; Walkeret al. 1999; Payneet al. 2000). Furthermore, in vitro studies have implicated the Arabidopsis RD22BP1 bHLH factor and the ATMYB2 Myb factor in activating a drought-regulated gene rd22 (Abeet al. 1997). The emerging view is that various bHLH/Myb heterodimer combinations serve as transcriptional activators for a number of plant biosynthetic and developmental genes, with WD-repeat factors perhaps performing a modifying or stabilizing function (Molet al. 1998; Payneet al. 2000; Speltet al. 2000).
We recovered a dominant mutation, atr2D, in an R/B-related factor in the course of studying tryptophan pathway regulation in Arabidopsis. The atr2D mutation confers constitutively activated expression of several tryptophan pathway genes, as well as perturbed levels of other stress-responsive genes. The mutation changes an aspartic acid to an asparagine at a position near the amino terminus of the predicted ATR2 protein sequence that is highly conserved in related plant transcription factors. Mutation of this conserved residue to other amino acids with uncharged side chains also resulted in similar dominant phenotypes. Overexpression of the atr2D mutant, but not wild-type ATR2, resulted in pleiotropic phenotypes including dark pigmentation and sterility. This result suggests that atr2D is an altered-function allele and that ATR2 is not normally involved in tryptophan and other stress-responsive gene regulation. The conservation of aspartic acid at the atr2D position suggests that mutation of the residue to asparagine might generate activated alleles of other R/B family members. Possible mechanisms for how this single amino acid change can generate activated stress phenotypes are discussed.
Previously, we characterized another dominant atr mutation, atr1D, which was found to encode an overexpression allele of a Myb transcription factor (Bender and Fink 1998). The atr1D plants upregulated the expression of primary tryptophan biosynthetic genes (Bender and Fink 1998) as well as secondary tryptophan biosynthetic genes (Smolen and Bender 2002). The ATR1 Myb transcription factor is likely to be an endogenous regulator of the tryptophan pathway, because increased ATR1 steady-state message levels correlate with tryptophan pathway activation in a recessive atr mutant, atr4 (Smolen and Bender 2002). The atr4 complementation group consists of loss-of-function mutations in the cytochrome P450 gene CYP83B1. The elevated ATR1 message levels in cyp83B1 seedlings are thought to result from perturbations in multiple signaling pathways converging on ATR1 gene expression.
Given the precedent for Myb and bHLH factors to act in combination, we examined whether the ATR1 Myb and the atr2D-activated mutant bHLH might interact to control tryptophan gene expression. Using a variety of approaches, we found that ATR1 and atr2D act independently of each other to activate the tryptophan pathway. In particular, the atr1D atr2D double mutant displays additive activation of several tryptophan primary and secondary metabolism genes. Therefore, the dominant atr mutations provide novel tools for the study and metabolic engineering of tryptophan primary and secondary biosynthetic pathways.
MATERIALS AND METHODS
atr2D mutant isolation: Columbia (Col) M2 seeds (~50,000) generated as described (Niyogiet al. 1993) by EMS mutagenesis were screened for resistance to 5-methyl-tryptophan (5MT) by plating surface-sterilized seeds on 150 × 25-mm petri plates containing 100 ml of minimal plant nutrient sucrose (PNS) medium (Haughn and Somerville 1986) supplemented with 15 μm 5MT (solubilized in 100 mm NaOH). Plates were sealed with Parafilm and were incubated at 22° under continuous illumination with yellow long-pass filters as previously described (Bender and Fink 1998). Seedlings were monitored between 7 and 10 days postgermination for root growth. The same conditions were used to test T2 and T3 generation transgenic plants and the panel of dominant tryptophan mutants for 5MT resistance, except that seeds were plated on 100 × 100-mm square petri plates containing 50 ml of medium, and the plates were screened with clear glass rather than with yellow long-pass filters.
Positional cloning of the atr2D locus: The atr2D mutation was mapped to the lower arm of chromosome 5 near the DFR marker using standard cleaved amplified polymorphic sequence (CAPS) analysis (Konieczny and Ausubel 1993). Analysis with additional standard Col vs. Landsberg erecta (Ler) polymorphic markers (http://www.arabidopsis.org) localized the mutation several centimorgans centromere-distal to DFR. Two novel markers, MZ2L and MZ2R, were used to localize atr2D to the MZA15 bacterial artificial chromosome clone. MZ2L, at the centromere-proximal end, was a 1088-bp fragment PCR amplified with the primers MZ2LF 5′-CTCCGATCGCTTCTTAATCC-3′ and MZ2LR 5′-CAATTATGGGCCTGTAGTAC-3′. The Col product cleaves with TaqI into fragments of 157, 169, 242, and 520 bp whereas the Ler product cleaves with TaqI into fragments of 157, 411, and 520 bp. MZ2R, at the centromere-distal end, was a 1190-bp fragment PCR amplified with the primers MZ2RF 5′-TTGCTTCCATGGAAGGTCTC-3′ and MZ2RR 5′-TCCCACCACAGAGGTGTTCG-3′. The Col product cleaves with AluI into 810- and 380-bp fragments whereas the Ler product is uncleaved.
BclI cleaves at sites flanking the ATR2 gene to yield a 12-kb fragment. A library was constructed from BclI-digested atr2D mutant genomic DNA by ligating into BamHI-cut λDASH (Stratagene, La Jolla, CA) arms and packaging with Gigapack II Gold (Stratagene) in vitro packaging extracts. The library was probed with an ATR2 cDNA probe to identify clones carrying the atr2D fragment. A wild-type genomic ATR2 clone was obtained by hybridization screening of a wild-type Col genomic λ library provided by J. Mulligan and R. Davis (Stanford University). A full-length expressed sequence tag (EST) clone of ATR2 was obtained from the Arabidopsis Biological Resource Center at Ohio State University.
Plant transformation: The isogenic genomic ATR2 and atr2D constructs consisted of 6.0-kb SacI to EcoRV fragments cloned into the SacI to SmaI sites of the pBIN19 vector (Bevan 1984). To make the 35S-ATR2 and 35S-atr2D constructs, site-directed mutagenesis (Kunkelet al. 1987) was performed on a genomic clone of each allele (1) to create a BamHI site at 8–13 bp upstream of the translational start codon, removing the upstream out-of-frame ATG, and (2) to create a BamHI site at a position corresponding to the 3′ end of the EST cDNA sequence, 296–301 bp downstream of the translational stop codon. The resulting 2.09-kb BamHI fragment of each allele was then cloned into the BamHI site of vector pCaMV35S (Hullet al. 2000). Clones were transformed into wild-type Col (ATR2) plants by an Agrobacterium tumefaciens-mediated in planta method (Clough and Bent 1998). Primary T1 transformants were scored for transgene copy number by preparing genomic DNA from a single leaf and performing Southern blot analysis with digests and probes that could distinguish the endogenous ATR2 gene from the transgene signal. Lines that displayed a transgene band intensity one-half that of the endogenous locus were identified as putative single-copy isolates. Copy number was confirmed by 3:1 segregation of the kanamycin resistance transgene marker in the next generation and by additional Southern blot analysis on homozygous progeny lines.
Mutagenesis of ATR2 codon 94: The 6.0-kb SacI-EcoRV ATR2 genomic fragment in a pBlueScript KS II+ (Stratagene) vector was subjected to site-directed mutagenesis (Kunkelet al. 1987) to alter the GAT aspartic acid codon 94 to AAT asparagine, GAA glutamic acid, CAA glutamine, TCG serine, or GCT alanine. To facilitate screening, each mutagenesis at codon 94 was designed to also create a nearby novel AvrII site by alteration of codon 90 CTC leucine to CTA leucine. Mutant variants of ATR2 were confirmed by sequencing and then subcloned into pBIN19 and transformed into wild-type Col plants as described above.
Yeast two-hybrid assays: The ATR1, ATR2, and atr2D segments tested in yeast two-hybrid analysis were amplified by PCR and subcloned as SalI to NotI fragments into the vectors pDBLeu and pEXP-AD502 from the ProQuest kit (Life Technologies). Full-length ATR1 was amplified using ATR1F (5′-GTCGACCATGGTGAGGACACCATGTTGC-3′) and ATR1R1 (5′-GCGGCCGCTCAGACAAAGACTCCAACCATATTG-3′). Truncated ATR1 (1–122) was amplified using ATR1F and ATR1R2 (5′-GCGGCCGCTCAATCGATGCCTTTTTGCTTCAACC-3′). An ATR1 cDNA isolated from a Ler strain seedling library (Minetet al. 1992) was used as template for both ATR1 amplifications. Full-length ATR2 and atr2D were amplified with ATR2F (5′-GTCGACCATGAACGGCACAACATCATCAATC-3′) and ATR2R1 (5′-GCGGCCGCTCAATAGTTTTCTCCGACTTTCGTC-3′). Truncated ATR2 and atr2D (1–371) were amplified with ATR2F and ATR2R2 (5′-GCGGCCGCTCATGATCTAACCACAGTACTAAACG-3′). Since the ATR2 locus is encoded by a single exon, the genomic clones of ATR2 and atr2D were used as amplification templates.
5MT resistance of tryptophan regulation multiple mutants
All two-hybrid analyses were done in the yeast strain MaV230 according to a protocol provided with the ProQuest system. The interactions were assessed by scoring three independent reporter genes (HIS3, URA3, and lacZ). All interactions tested yielded internally consistent results between the three reporter genes.
Expression of fusion proteins was monitored by immunoblot analysis using antibodies against the Gal4 DNA-binding domain (SC-510, Santa Cruz) or the Gal4 transcription activation domain (SC-1663, Santa Cruz).
Dominant tryptophan mutant plant strains: The S115F mutation in the ASA1fbr transgene was constructed using site-directed mutagenesis (Kunkelet al. 1987) on a genomic clone of ASA1 extending from ~4.8 kb upstream of the start codon to 5.7 kb downstream of the stop codon. This mutant genomic clone was then subcloned into the SalI site of the plant transformation vector pBIN19 (Bevan 1984). The transgene was introduced into wild-type Columbia plants (Clough and Bent 1998), and a transgenic line that displayed 3:1 segregation of the kanamycin resistance transgene marker plus heritable 5MT resistance was selected for further analysis. Southern blot assays indicated that this line contains a four-copy array of the ASA1fbr transgene.
The double- and triple-mutant strains combining atr1D, atr2D, and ASA1fbr were made by screening F2 progeny of crosses for resistance to 50 μm 5MT, a concentration at which the parental single mutants are sensitive (Table 1). Candidates from this screen were subsequently tested for their genotypes using molecular markers. The atr1D genotype was confirmed by a PCR-based assay for a restriction site polymorphism created by the atr1D mutation (Bender and Fink 1998). The ASA1fbr genotype was confirmed by scoring transgene kanamycin resistance. The atr2D mutation was confirmed by a PCR-based assay that converts the base change into a restriction site change by combining it with a nearby mismatch in a PCR primer, which results in PCR amplification of a novel DNA species (Neffet al. 1998). For this assay, we used primers atr2DF 5′-CTTACGCTATCTTCTGGCAG-3′ and atr2DR 5′-CTCTTTATCTTCCTCTCCTTTGTAGTAACCAG-3′, where the underlined base is mismatched with the genomic sequence. This primer set amplifies a 110-bp fragment. For ATR2 template DNA, the fragment cleaves with AluI into 78- and 32-bp products, whereas for atr2D template DNA the fragment does not cleave with AluI.
RNA gel blot analysis: Total RNA was prepared from whole seedlings grown aseptically on PNS medium for 10 days postgermination before RNA extraction or from the indicated adult plant tissues sources in Figure 2B. RNA gel blot analysis was performed by formaldehyde gel electrophoresis, transfer to nylon membranes, and hybridization with radiolabeled probes as previously described (Melquistet al. 1999). Probes were full-length or nearly full-length cDNA fragments for ATR1 (At5g60890), ATR2 (At5g46760), ASA1 (At5g05730), ASA2 (At2g-29690), ASB1 (At1g25220), CYP79B2 (At4g39950), CYP83B1 (At4g31500), CHS (At5g13930), PDF1.2 (At5g44420), and TUB4 (At5g44340) and partial cDNA fragments for TSB1 (At5g54810), PR1 (At2g19990), LOX2 (At3g45140), and CYP79B3 (At2g22330). A β-tubulin (TUB) cDNA probe was used as a control to correct for loading differences. Band intensities were quantitated using a Fuji Phosphoimager and MACBAS 2.2 software. Results were reproduced in two or three independent experiments.
Assessment of mutant plant morphologies: The morphological consequences of dominant tryptophan regulatory mutations were scored by germinating seedlings in potting medium (Scott's Metromix 360) and growing them under continuous illumination at ~22°. The altered morphologies of the atr1D atr2D double mutant and the atr1D atr2D ASA1fbr triple mutant were consistently observed in several independent experiments.
Soluble tryptophan measurements: For soluble tryptophan measurements, seedlings were grown aseptically on PNS medium for 10 days postgermination. The extraction procedure was performed as previously described (Li and Last 1996). Tryptophan measurements were performed by the Bioresource Center of Cornell University (Ithaca, NY) using previously published HPLC methods (Heinrikson and Meredith 1984). Three independently prepared samples were analyzed in parallel for each strain tested.
RESULTS
Isolation and characterization of the atr2D mutant: The atr2D mutant was isolated from a previously described screen for altered tryptophan regulation (atr) Arabidopsis mutants resistant to the toxic tryptophan analog, 5MT (Bender and Fink 1998). The screen exploits the feedback inhibition of anthranilate synthase (AS), the first enzyme of the pathway, by the end product of the primary pathway, tryptophan. Feedback inhibition limits the flow through the pathway under conditions of abundant tryptophan. 5MT acts by triggering feedback inhibition of AS activity without substituting for the nutritional role of tryptophan. Mutants that are resistant to 5MT will thus include plants with feedback resistance mutations in AS catalytic subunit genes (Niyogi 1993; Krepset al. 1996; Li and Last 1996) and mutants with increased expression of AS and other tryptophan metabolism genes (Bender and Fink 1998; Smolen and Bender 2002). To isolate such resistant mutants, EMS-mutagenized M2 seeds of the Col strain were plated on agar medium containing 15 μm 5MT, and seedlings were scored at 10–14 days for root growth. Under these conditions, wild-type seedlings have strongly inhibited root development whereas atr mutants have unimpaired root growth (Figure 1). The 5MT resistance phenotype of the atr2D mutant isolated from this screen was found to be dominant in F1 heterozygous plants (Figure 1) and segregated at 75% (533 resistant out of 709 total seedlings scored) in an F2 population from a backcross to wild-type Col. Only a single atr2D allele was identified in a screen of 50,000 seedlings. The atr2D mutant displayed no obvious morphological alterations at any stage of development.
The atr2D mutation confers dominant 5MT resistance. Seedlings of wild-type Col (ATR2/ATR2), a Col atr2D homozygote (atr2D/atr2D), a Col atr2D/ATR2 F1 heterozygote made by crossing Col atr2D as the male with wild-type Col as the female, a representative homozygous T3 transgenic line of Col transformed with a single-copy wild-type ATR2 genomic clone, ATR2/ATR2(ATR2/ATR2), and a representative homozygous T3 transgenic line of Col transformed with a single-copy mutant atr2D genomic clone, ATR2/ATR2 (atr2D/atr2D), are shown after being grown on PNS medium containing 15 μm 5MT for 7 days postgermination.
Gene expression patterns in ATR2 and atr2D plants. (A) Replicate gel blots of seedling RNAs were probed with cDNA fragments of the indicated genes. The TUB probe was used to normalize for differences in loading. (B) RNA samples prepared from leaves (leaf), flowers and buds (flower), or green siliques (silique) of adult wild-type Col plants grown in soil, plus 10-day-postgermination aseptically grown whole Col seedlings (seedling), were used for gel blot analysis with an ATR2 probe. The ethidium bromide (EtBr)-stained gel is shown as a loading control. (C) Wild-type Col seedlings were grown aseptically on PNS medium under glass plates for 10 days postgermination and then transferred to liquid PNS medium for a 6-hr induction before RNA extraction. The following concentrations of inducers were used: 20 μm methyl jasmonate (MeJA), 20 μm abscisic acid (ABA), 500 μm salicylic acid (SA), 20 μm IAA, 20 μm 6-benzylaminopurine (BAP), 20 μm gibberellic acid A3 (GA), 20 μm 1-aminocyclopropane 1-carboxylic acid (ACC), and 1 μm brassinolide (BR). The RNA samples were analyzed by gel blot with an ATR2 probe.
To determine whether the 5MT resistance phenotype of the atr2D mutant involved increased expression of tryptophan genes, RNA was prepared from aseptically grown whole seedlings of wild-type ATR2 vs. atr2D and analyzed by gel blot with tryptophan gene probes (Figure 2A). The probes tested included those detecting two AS catalytic α-subunit-encoding genes ASA1 and ASA2 (Niyogi and Fink 1992); a probe that cross-hybridizes to three highly related AS glutamine amidotransferase β-subunit-encoding genes ASB1, ASB2, and ASB3 (Niyogiet al. 1993); and a probe detecting the tryptophan synthase β-subunit-encoding gene TSB1 (Berlynet al. 1989). This analysis revealed that steady-state transcript levels of the ASA1 gene, the ASB genes, and the TSB1 gene were elevated in atr2D plants. These tryptophan genes have been shown to be induced by a variety of stress conditions including pathogen attack and amino acid starvation (Niyogi and Fink 1992; Niyogiet al. 1993; Zhao and Last 1996; Zhaoet al. 1998; Reymondet al. 2000), although the stress responsiveness of the individual ASB genes has not been determined (Niyogiet al. 1993). In contrast, the steady-state levels of the ASA2 gene, which was previously found to be unresponsive to stress induction, were not affected by the atr2D mutation. Because this pattern suggested that stress-inducible genes might be generally upregulated in the atr2D mutant background, we tested several other known stress gene probes in this assay, including the flavonoid compound synthesis gene chalcone synthase (CHS; Feinbaum and Ausubel 1988), the salicylic acid-responsive gene PR1 (Penninckxet al. 1996), methyl jasmonate-responsive genes PDF1.2 (Penninckxet al. 1996) and LOX2 (Bell and Mullet 1993), and the ATR1 Myb gene, which can be either up- or downregulated in response to various signaling molecules (Smolen and Bender 2002). This analysis showed that CHS, PDF1.2, and LOX2 were upregulated in the atr2D mutant, while PR1 was downregulated and ATR1 was not affected. Note that although PR1 expression is not normally detectable by RNA gel blot in adult plant tissues (Clarkeet al. 2000), it does have a detectable basal level of expression in seedlings grown under our aseptic conditions (Smolen and Bender 2002). Collectively, this sampling of probes revealed that a subset of stress response genes as well as inducible tryptophan genes are activated by atr2D.
The previously isolated atr1D mutation conferred a seedling phenotype diagnostic of resistance to high levels of exogenous tryptophan (Bender and Fink 1998). This phenotype assay uses the blue fluorescent trp1-100 mutant, which carries a missense mutation in the PAT1 gene encoding the second enzyme of the tryptophan pathway (Rose et al. 1992, 1997). The blue fluorescence of trp1-100 results from the accumulation of anthranilate in a sugar-conjugated form, particularly in the cotyledons of seedlings. When the trp1-100 mutant is grown on agar medium containing high levels of exogenous tryptophan, the cotyledon blue fluorescence is suppressed, presumably due to feedback inhibition of AS activity and reduced production of anthranilate (Bender and Fink 1998). However, in the trp1-100 atr1D double mutant, cotyledons remain fluorescent when grown on high tryptophan medium. To determine whether atr2D behaves similarly to atr1D in this assay, the trp1-100 atr2D double mutant was constructed and tested for fluorescence on high tryptophan medium. Unlike the trp1-100 atr1D double mutant, the trp1-100 atr2D double mutant displayed suppressed cotyledon fluorescence similarly to the trp1-100 parental strain (data not shown). This result suggests that atr1D and atr2D perturb tryptophan metabolism in different tissues and/or via different mechanisms.
Positional cloning of the atr2D mutation: To understand the molecular basis of the atr2D mutation, we cloned the ATR2 gene on the basis of its map position. Standard procedures were used to map the mutation to the lower arm of chromosome 5. The atr2D mutant was then crossed to a polymorphic strain carrying visible marker mutations on the upper and lower ends of chromsome 5 flanking the atr2D locus (Landsberg erecta ttg yi). The visible marker mutations were used to identify 502 F2 progeny that had crossed over between the markers and thus had recombination break points around atr2D. The recombinants with break points closest to the atr2D 5MT resistance phenotype defined the minimal atr2D locus as being contained between the ends of a cloned and sequenced 79,995-bp bacterial artificial chromosome MZA15 (GenBank accession no. AB016882).
Inspection of the MZA15 sequence revealed that it contained an open reading frame for a bHLH transcription factor. This gene was an attractive candidate for the atr2D tryptophan gene and CHS-activating locus because the related R gene in maize is known to encode a transcriptional activator of phenylalanine-derived flavonoid compound biosynthetic genes such as CHS (reviewed in Molet al. 1998). We confirmed that the bHLH open reading frame is in fact the ATR2 gene, using three approaches. First, we found that a polymorphic PstI restriction enzyme site in the central coding sequence of this gene cosegregated perfectly with the mutant phenotype in the mapping population. Second, we cloned and sequenced the bHLH gene from the atr2D mutant vs. wild-type ATR2 and found a single G to A transition mutation in the gene that creates an aspartic acid to asparagine change in the predicted amino acid sequence at codon 94 (D94N). Third, we transformed isogenic 6.0-kb genomic regions containing the transcription factor gene but no other open reading frames from either wild-type ATR2 or atr2D into the wild-type Col genome. The inserts in these transgene constructs were completely sequenced and found to differ only at the single base that creates the D94N mutation. All of 36 transformants of the wild-type clone ATR2(ATR2) were sensitive to 5MT, whereas 25 of 27 transformants of the mutant clone ATR2(atr2D) were resistant to 5MT (Figure 1). The two exceptional ATR2(atr2D) lines that were sensitive to 5MT were also sensitive to the transgene selectable marker kanamycin in the second generation after transformation and contained multiple cytosine methylated copies of the transgene, suggesting that these lines carry silenced transgene arrays. The transformation experiment thus demonstrated that the atr2D locus lies in the 6.0-kb clone and that the activated tryptophan phenotype is conferred by the D94N mutation.
Overexpression of atr2D but not ATR2 confers pleiotropic effects: During the construction of the ATR2(ATR2) and ATR2(atr2D) transgenic lines we found that all of the ATR2(ATR2) lines were morphologically identical to untransformed Columbia plants regardless of transgene copy number, whereas the majority of multiple-insert ATR2(atr2D) lines had increased purple pigmentation, reduced size, and reduced fertility. In the most extreme cases, ATR2(atr2D) multiple-insert plants were completely sterile. These observations suggested that overexpression of atr2D, but not wild-type ATR2, perturbs gene expression in a way that leads to 5MT resistance, increased pigmentation, and morphological changes.
ATR2 and atr2D overexpression phenotypes. Representative 5-week-old single-copy primary transformant hemizygous T1 plants of ATR2/ATR2(35S-ATR2/o) (left, 35S-ATR2) and ATR2/ATR2(35S-atr2D/o) (right, 35S-atr2D) are shown.
To explicitly test the effects of overexpression, we made additional transgenic lines in which the ATR2 or atr2D D94N coding sequence was expressed from the strong constitutive cauliflower mosaic virus 35S promoter (35S-ATR2 and 35S-atr2D). In the wild-type Columbia (ATR2) background, three independent 35S-ATR2 single-copy primary transformants were phenotypically normal, whereas four independent single-copy 35S-atr2D primary transformants were small, bushy, darkly pigmented, and completely sterile (Figure 3). Second-generation ATR2(35S-ATR2) lines were found by RNA gel blot analysis to have approximately fivefold elevated steady-state transcript levels of ATR2, but the plants were not resistant to 5MT and did not display upregulation of ASA1 expression (data not shown). The sterility of the ATR2(35S-atr2D) lines precluded detailed analysis. However, these results confirm that overexpression of the wild-type ATR2 factor is not sufficient to alter 5MT resistance or morphology, whereas overexpression of the atr2D mutant factor has pleiotropic consequences, presumably due to enhanced activation of target gene expression.
The atr2D mutation alters a highly conserved amino acid position in a related group of plant transcription factors: The ATR2 gene consists of a single exon encoding a predicted 592-amino-acid polypeptide. A full-length Columbia cDNA isolate of this gene was available in the Arabidopsis EST collection (GenBank accession no. N96108), and we cloned an additional full-length cDNA from a Ler cDNA library (Minetet al. 1992) by hybridization. In both cases, the transcribed region contained an upstream out-of-frame ATG sequence just 8 bp upstream of the correct ATR2 translational start codon, which could impair optimal translation of the gene. RNA gel blot analysis of ATR2 expression revealed that the gene is well expressed in seedlings, adult leaves, and flowers, with reduced expression in green siliques (Figure 2B). ATR2 steady-state transcript levels were not altered in the atr2D mutant (Figure 2A), indicating that the activated atr2D allele does not affect its own expression. The ATR2 gene displayed only modest changes in steady-state message levels when wild-type seedlings were subjected to exogenous treatment with a variety of plant signaling molecules, in contrast to the more dramatic transcriptional regulation of the ATR1 Myb-encoding gene previously detected under these conditions (Smolen and Bender 2002).
Database comparisons showed that the ATR2 gene is related to a group of plant bHLH transcription factor genes (Figure 4). These genes are characterized by conserved sequences including an acidic region near the amino terminus plus a basic helix-loop-helix domain near the carboxy terminus (Purugganan and Wessler 1994). The codon 94 aspartic acid residue in the ATR2 gene occurs in a highly conserved sequence motif. Of 11 closely related cDNA and genomic sequences identified in the Arabidopsis nucleic acid sequence database using a BLAST search with the full-length ATR2 amino acid sequence as a query, 9 sequences also carried an aspartic acid at the analogous position (Figure 4B). Moreover, the majority of related cDNA sequences cloned from other plant species carried an aspartic acid at the analogous position. One additional Arabidopsis genomic sequence, a maize cDNA, and two Perilla frutescens cDNAs carried the conservative amino acid substitution of a glutamic acid at the analogous position. However, one exceptional Arabidopsis sequence, the TT8 gene that controls pigment production in the seed coat, carried an asparagine at the analogous position (Nesiet al. 2000). The TT8 protein has relatively low sequence identity with ATR2 and lacks a large block of amino acids between the conserved acidic and bHLH domains. Thus, it is difficult to predict whether the activated atr2D D94N factor might act in part by mimicking the effects of TT8 over a broader expression range.
Several mutations at ATR2 codon 94 yield atr2D-like phenotypes: The GAT aspartic acid to AAT asparagine atr2D mutation was the only change predicted to be recovered at codon 94 of ATR2 using EMS mutagenesis, which primarily generates C to T and G to A transitions. Therefore, an interesting question was whether activated atr2D-like alleles of ATR2 could be generated by alteration of codon 94 to other amino acids. To address this question we explicitly mutagenized ATR2 codon 94 to several other amino acids including glutamic acid, glutamine, serine, and alanine. Glutamic acid was selected because it is found at the analogous position in some related proteins (Figure 4). The other three changes were selected as having uncharged side chains of various lengths. Mutagenesis to asparagine was also included as a positive control. The resulting five isogenic constructs were transformed into wild-type ATR2 plants and tested in the T2 generation for 5MT resistance. This analysis showed that all of 8 D94N, 14 D94Q, 6 D94S, and 4 D94A lines tested were 5MT resistant regardless of transgene copy number (data not shown). For each of these constructs the highest-copy-number lines were bushy, darkly pigmented, and had reduced fertility similarly to the atr2D94N overexpression lines. However, all five D94E lines tested were 5MT sensitive and all of these lines were morphologically normal. This result implies that the D94E mutation has a neutral effect, although it is also possible that the mutation renders the protein inactive. Taken together, the mutagenesis experiments suggest that loss of a negatively charged residue at codon 94 of ATR2 yields an altered-function atr2D protein.
The atr2D mutation affects a conserved aminoterminal motif in members of the R/B bHLH factor family. (A) R/B bHLH factor conserved structural domains. A diagram of a generic R/B bHLH factor protein structure is shown. N indicates the amino terminus, C indicates the carboxy terminus, and the solid box indicates the atr2D-proximal region with the asterisk indicating the position of the mutated residue. (B) Sequence alignment of the atr2D mutant region, shown as the solid box in A, with homologs from Arabidopsis and other plant species. Arabidopsis bHLH-predicted amino acid sequences are shown in the top group and other plant species are shown in the bottom group. All aligned bHLH proteins are indicated by their GenBank protein accession numbers. Except for protein BAA97217, the Arabidopsis bHLH genes either have been assigned a function by mutation (GL3 and TT8) or have been cloned as cDNAs and/or ESTs. The bHLH-predicted amino acid sequences from other plant species are all based on cDNA sequences. An asterisk above the sequences indicates the position of the atr2D mutation. Pv indicates Phaseolus vulgaris (bean), Zm indicates Zea mays (corn), Os indicates Oryza sativa (rice), Pf indicates Perilla frutescens, Am indicates Antirrhinum majus (snapdragon), Ph indicates Petunia hybrida, and Gh indicates Gerbera hybrida.
The atr2D bHLH mutant factor and the ATR1 Myb factor do not interact in a yeast two-hybrid assay: Because several characterized plant bHLH factors act in combination with partner Myb proteins (see Introduction), we investigated whether the atr2D bHLH might act together with the ATR1 Myb factor previously identified in atr mutant screens (Bender and Fink 1998). As one approach to this question, we tested for physical interactions using a yeast two-hybrid system. This approach has previously been used to demonstrate interactions between the maize R/B proteins and C1 and the Arabidopsis GL3 and GL1 bHLH/Myb combinations (Goffet al. 1992; Grotewoldet al. 2000; Payneet al. 2000). These studies showed that the regions critical for interaction lie in the amino-terminal regions of both partner proteins. Thus, for our studies, we tested full-length proteins and amino-terminal (NT) constructs of ATR1, ATR2, and atr2D as DNA-binding fusion “baits” or activation-domain fusion “preys.” We found that full-length ATR1, and full-length and NT ATR2 or atr2D constructs, were all self-activating when used as baits. However, the ATR1 NT construct was not self-activating as a bait, and the ATR2 and atr2D constructs were not self-activating as preys. Immunoblot detection of fusion protein expression for these constructs revealed that protein products of the predicted sizes were expressed for the ATR1 NT bait and the ATR2 and atr2D NT preys. However, the full-length ATR2 and atr2D preys were not detectably expressed. Therefore, the ATR1 NT bait construct was used together with ATR2 NT or atr2D NT prey constructs to assess activation of three different reporter genes—lacZ, HIS3, and URA3. Neither of the combinations tested showed activation of any of the reporter genes (data not shown), suggesting that neither the ATR1 NT/ATR2 NT pair nor the ATR1 NT/atr2D NT pair has a significant physical interaction.
The atr2D and atr1D mutations have additive effects on tryptophan pathway deregulation: As another approach to understanding the relationship between atr2D and atr1D, we constructed a double mutant of the two dominant alleles and analyzed it for 5MT resistance and patterns of tryptophan gene expression. Furthermore, because we were interested in the potential for using the dominant atr mutations as tools to engineer tryptophan metabolism, we analyzed their effects in combination with a transgenic feedback resistance allele of ASA1 (ASA1fbr) as double or triple mutants. The ASA1fbr transgene carries a genomic clone of ASA1 with the feedback resistance mutation S115F. This mutation was predicted to confer feedback resistance on the basis of the phenotype of an analogous mutation in the conserved position of the Salmonella AS α-subunit-encoding gene trpE (Caligiuri and Bauerle 1991). The ASA1 S115F mutation was previously engineered into a 35S-ASA1 cDNA transgene and found to confer 5MT resistance in transgenic Arabidopsis (Niyogi 1993). Transgenic plant extracts contained feedback-resistant AS activity in in vitro assays and had ~1.5-fold elevated levels of soluble tryptophan. For comparison, we also assayed the genetically isolated feedback-resistant ASA1 mutant amt-1, which carries a D341N missense mutation (Krepset al. 1996).
5MT resistance in tryptophan regulation multiple mutants. Representative seedlings of the indicated genotypes are shown after being grown on PNS medium supplemented with either (A) 15 μm 5MT or (B) 300 μm 5MT for 7 days postgermination.
In root growth assays for 5MT resistance, the double- and triple-mutant strains displayed root resistance stronger than that of the single mutants (Table 1). At 300 μm 5MT, only the triple atr1D atr2D ASA1fbr mutant had resistant root growth (Figure 5, Table 1). These results show that the individual mutations confer additive effects on the 5MT resistance phenotype.
To elucidate the genes that contribute to additive 5MT resistance in double- and triple-mutant plants, we performed RNA gel blot analysis of steady-state transcript levels for various tryptophan metabolism genes. Because the ASA1fbr transgenic strain carried four extra copies of the ASA1 gene at an ectopic locus (materials and methods), we expected that this strain would display elevated ASA1 transcript levels. ASA1 expression was indeed increased approximately sevenfold in the ASA1fbr strain (Figure 6). In double-mutant combination strains, ASA1 expression was upregulated to a higher degree than in the parental single mutants. For example, the atr2D ASA1fbr strain had strongly elevated ASA1 levels (Figure 6). This activation is presumably due to the transcriptional activation effect of atr2D on the four ASA1fbr transgene sequences as well as on the endogenous ASA1 gene. Surprisingly, the atr1D atr2D ASA1fbr triple mutant had reduced ASA1 transcript levels compared to the two atr ASA1fbr double mutants. The reduced expression in the triple mutant might reflect a downregulation mechanism that is activated when either ASA1 transcripts or tryptophan pathway metabolites exceed a threshold level (see discussion). The tryptophan synthase gene TSB1 was upregulated in all strains carrying the atr2D mutation, without significant additional activation in response to the atr1D or ASA1fbr mutations. Similar results were obtained for ASB (data not shown).
Tryptophan gene expression in tryptophan regulation multiple mutants. RNA samples prepared from seedlings of the indicated genotypes were subjected to RNA gel blot analysis with the indicated probes.
Adult plant morphologies of tryptophan regulation multiple mutants. Representative 4-week-old plants of the indicated genotypes are shown. Other strains used in this study were morphologically indistinguishable from the wild-type Col (WT) control.
Two Arabidopsis cytochrome P450 enzymes, CYP79B2 and CYP79B3, catalyze the conversion of tryptophan into indole-3-acetaldoxime (Hullet al. 2000). Because overexpression of CYP79B2 leads to modest 5MT resistance, we investigated the CYP79B2 and CYP79B3 steady-state message levels in the panel of dominant mutant strains. Both genes were upregulated in all the strains carrying the atr1D mutation (Figure 6). Both genes were also slightly upregulated in the atr2D mutant background. In the atr1D atr2D double-mutant and the triple-mutant backgrounds, there was an additive effect of the two atr mutations on CYP79B2 and CYP79B3 expression.
The cytochrome P450 enzyme CYP83B1 acts on the indole-3-acetaldoxime product of the CYP79B2/B3-catalyzed reactions to convert it into 1-aci-nitro-2-indolylethane, a precursor for indole glucosinolate synthesis (Baket al. 2001; Hansenet al. 2001). As previously reported (Smolen and Bender 2002), the CYP83B1 gene was upregulated in atr1D mutant backgrounds. However, there was no significant additional activation via atr2D or ASA1fbr. Moreover, it is unlikely that the CYP83B1 gene product contributes significantly to 5MT resistance, because loss-of-function alleles in the gene are 5MT resistant (Smolen and Bender 2002). This resistance is presumably due to perturbations in expression levels of other tryptophan genes in the cyp83B1 background.
Beyond the altered patterns of 5MT resistance and tryptophan gene expression, the atr1D atr2D double-mutant and the atr1D atr2D ASA1fbr triple-mutant strains had distinct adult plant (4-week-old) phenotypes (Figure 7). The atr1D atr2D double mutants displayed a modest reduction in size and flowered a few days earlier than did wild-type plants. The atr1D atr2D ASA1fbr triple mutants displayed more severe defects, including strongly reduced size and a bushy morphology. None of the other single or double mutants in the panel displayed any obvious morphological effects (data not shown). These results suggest that the morphological transition created in the atr1D atr2D double mutants is accentuated by the ASA1fbr-induced increased flow of metabolites through the tryptophan pathway.
Soluble tryptophan levels in tryptophan regulation multiple mutants. Soluble tryptophan levels were measured in triplicate samples for seedlings of the indicated genotypes.
Taken together, these analyses suggest that atr1D and atr2D have distinct effects on tryptophan gene regulation, implying independent mechanisms of gene activation. The tryptophan gene expression profiles suggest that in the triple mutant, the strong upregulation of primary pathway genes (ASA1, ASB, and TSB1), strong upregulation of secondary pathway genes (CYP79B2 and CYP79B3), and a feedback resistance allele of ASA1 combine to uniquely deregulate the pathway and confer strong 5MT resistance and altered morphology.
Soluble tryptophan levels are perturbed in dominant tryptophan regulation mutants: Feedback resistance mutations in Arabidopsis ASA1 increase soluble tryptophan levels (Niyogi 1993; Krepset al. 1996; Li and Last 1996), but the atr mutations had not been previously characterized by this assay. We therefore measured soluble tryptophan levels in our complete panel of mutant plants (Figure 8). This analysis showed that only the feedback resistance ASA1 alleles conferred increased soluble tryptophan levels relative to wild-type plants: ~3-fold elevated for amt-1 and ~1.5-fold elevated for ASA1fbr. These levels are similar to those previously reported for each feedback resistance mutation (Niyogi 1993; Li and Last 1996). In contrast, the atr1D and atr2D single mutants both displayed reduced soluble tryptophan levels relative to wild-type plants. This reduction might reflect the increased expression of the tryptophan-metabolizing cytochrome P450 genes CYP79B2 and CYP79B3 (Figure 6) and/or activation of other as-yet-unidentified tryptophan metabolism genes. Especially in the case of atr2D, it seems likely that other tryptophan secondary metabolism genes are activated because the slight upregulation of CYP79B2 and CYP79B3 does not correlate with the depletion of soluble tryptophan levels, especially in light of the coordinate upregulation of primary pathway genes (Figures 2A and 6). When the atr mutations were combined with ASA1fbr, wild-type or higher levels of soluble tryptophan were observed. This pattern suggests that ASA1fbr is able to counteract the tryptophan-depleting effects of the atr mutations.
We also assayed the panel of mutant plants for perturbations in the levels of two other tryptophan pathway metabolites. First, we performed a thin layer chromatography assay on seedlings for the presence of the pathogen defense compound camalexin (Zhao and Last 1996; Zhaoet al. 1998). Neither wild-type nor mutant plants displayed any detectable camalexin by this assay, indicating that none of the mutant combinations deregulates camalexin synthesis. Second, we inspected the panel of mutant plants for blue fluorescence under ultraviolet light diagnostic of accumulation of anthranilate, the first compound in the tryptophan pathway. None of the mutants was fluorescent, indicating that there is not a significant accumulation of anthranilate in these strains.
DISCUSSION
By screening for mutants with deregulated tryptophan pathway phenotypes, we have identified a novel dominant allele of the Arabidopsis ATR2 bHLH transcription factor. This factor is a member of a family of plant transcriptional activators that share structural homology with the maize R and B factors. The R/B group is characterized by a conserved amino-terminal region with an acidic patch and the bHLH domain located near the carboxy terminus of the protein (Purugganan and Wessler 1994). Significantly, the region where the activating mutation occurs in ATR2 is highly conserved in other members of the group, with the majority of factors having an aspartic acid residue and a minority having a glutamic acid residue at this position (Figure 4). Therefore, similarly to the atr2D allele, altering the charge at this position could generate activated alleles of other R/B family members.
Given the precedent for R/B-related factors to work together with Myb factors (see Introduction), it is possible that ATR2 and atr2D have partner Myb proteins. Thus, the activated phenotypes in the atr2D mutant could be conferred through altered atr2D protein/protein interactions with partner Mybs. Consistent with this possibility, the Myb interaction domains of GL3, R, and B bHLH factors have been mapped to amino-terminal regions of these proteins (Goffet al. 1992; Grotewoldet al. 2000; Payneet al. 2000), where the atr2D mutation nisms, including alterations in protein stability, nuclear localization, or DNA binding for the mutant atr2D factor vs. wild-type ATR2. Furthermore, because the atr2D mutation stimulates expression of a number of stress-inducible genes such as CHS and PDF1.2 in addition to inducible tryptophan biosynthetic genes (Figure 2), the mutant phenotypes could occur indirectly due to a general stress response caused by the altered form of the protein rather than by direct interaction of atr2D with target genes. A specific possibility is that the atr2D mutation creates a dominant negative protein that interferes with the normal function of ATR2 or other bHLH factors, thus indirectly causing stress-response phenotypes. More extensive genetic and molecular analysis of atr2D should allow us to distinguish among these scenarios. For example, an interesting question is whether the analogous mutation in ATR2-related Arabidopsis bHLH genes will yield a similar profile of stress-responsive gene activation as observed in atr2D. This result would suggest that the effect of the dominant mutation is to make a general structural change in bHLH proteins that provokes an activated stress response in the host plant cell.
Our explicit mutagenesis of ATR2 codon 94 revealed that changes of the aspartic acid to asparagine, glutamine, serine, or alanine result in dominant transgene constructs that confer 5MT resistance to wild-type ATR2 plants. In contrast, conversion of the codon to glutamic acid results in a construct that does not confer 5MT resistance or any other obvious phenotypic effects. These results suggest that an uncharged amino acid side chain at codon 94 yields an activated atr2D-like factor whereas a negatively charged amino acid side chain at codon 94 (either aspartic acid or glutamic acid) yields a version of ATR2 with normal functions. Because overexpression of wild-type ATR2 is not sufficient to activate 5MT resistance or other atr2D phenotypes (Figure 3), it seems unlikely that tryptophan gene regulation is the normal role of this protein. However, we are currently pursuing isolation of an atr2 loss-of-function allele to clarify whether wild-type ATR2 has any involvement in tryptophan and other stress-responsive gene regulation.
Several lines of evidence argue that atr1D, a Myb overexpression allele previously isolated in the atr genetic screen (Bender and Fink 1998), and the atr2D bHLH allele activate the tryptophan pathway through distinct mechanisms. First, atr1D was originally recovered by screening for a cotyledon phenotype diagnostic of resistance to high levels of exogenous tryptophan (Bender and Fink 1998), but atr2D does not confer this phenotype. Second, ATR1 does not display detectable interaction with either wild-type ATR2 or mutant atr2D in a yeast two-hybrid assay. Third, the atr1D atr2D double mutant displays additive 5MT resistance (Figure 5, Table 1), additive activation of tryptophan gene expression (Figure 6), and a synthetic reduced size phenotype (Figure 7), suggesting that each single mutation confers its effects independently of the other locus. The ability of each dominant atr mutation to enhance the 5MT resistance of ASA1fbr suggests that the feedback resistance of this allele is incomplete, as is also suggested by the lower accumulation of soluble tryptophan that this allele confers relative to the amt-1 allele (Figure 8; Niyogi 1993; Li and Last 1996). It is also possible that 5MT has toxic effects beyond feedback inhibition of AS that are ameliorated by the atr mutations. For example, 5MT might be misincorporated into proteins, and the stimulation of tryptophan secondary metabolism enzymes that directly deplete 5MT levels might limit this misincorporation. However, studies in both yeast and plants suggest that this compound is not significantly used in protein synthesis (Miozzariet al. 1977; Sasseet al. 1983).
When used in combination, the dominant atr mutations allow deregulation of multiple tryptophan primary and secondary metabolism genes (Figure 6). These two dominant mutations thus provide a much more facile means of manipulating the tryptophan pathway than does the alternative strategy of overexpressing multiple biosynthetic genes from separate transgene constructs. Another advantage of the atr mutations as tryptophan metabolism tools is that they activate tryptophan gene expression from native promoters, so that gene products are expressed in appropriate tissues. Although the atr-activated expression of tryptophan genes is not as high as the expression levels that could be achieved by expression from a strong viral promoter such as 35S, this is not necessarily a disadvantage because extremely high levels of expression can sometimes cause RNA silencing (reviewed in Vance and Vaucheret 2001), defeating the purpose of an overexpression construct as a means of metabolic engineering. In fact, the downregulation of ASA1 expression we observed in the triple-mutant atr1D atr2D ASA1fbr strain (Figure 6) could be due to this strain expressing ASA1 at high enough levels to provoke weak ASA1-directed RNA silencing. Alternatively, accumulation of a tryptophan-derived metabolite could trigger downregulation of ASA1 in atr1D atr2D ASA1fbr plants. This downregulation of ASA1 could represent a natural threshold for tryptophan pathway activation.
Tryptophan secondary metabolism in plants provides a number of important agricultural and medicinal compounds including the growth regulator indole acetic acid (IAA), the defense compound DIMBOA in maize (Freyet al. 1997), indole glucosinolate defense compounds in Brassicas (Chappleet al. 1994), and anticancer alkaloids vincristine and vinblastine in Catharanthus roseus (Kutchan 1995). There is thus considerable interest in understanding and manipulating the biosynthetic enzymes that convert tryptophan into secondary compounds. Recent genetic studies in Arabidopsis have identified some of the components of these secondary pathways, such as the cytochrome P450 enzymes CYP-79B2, CYP79B3, and CYP83B1 (Hullet al. 2000; Baket al. 2001; Hansenet al. 2001). The double atr1D atr2D mutant displays significant upregulation of these three genes, as well as upregulation of key genes in the primary tryptophan pathway (Figure 6). These patterns of gene expression suggest that flux into secondary metabolic pathways might be significantly perturbed in this strain. Consistent with this view, when a tryptophan feedback-resistant allele of ASA1, which further increases flux through the primary tryptophan pathway, is combined with the atr1D and atr2D mutations in a triple-mutant strain, this strain displays a unique altered morphology (Figure 7). The altered morphology could result from increased production of IAA, increased production of indole glucosinolates, and/or increased production of other tryptophan pathway metabolites. Genetic suppressor analysis of this morphological transition could identify key biosynthetic genes responsible for the effect. Furthermore, metabolic labeling studies with the triple-mutant strain could elucidate the major routes of secondary metabolism that are deregulated in this background. The triple dominant mutant strain therefore serves as a novel tool for studies of tryptophan secondary metabolism in plants.
Acknowledgments
We thank the Arabidopsis Biological Resource Center for the Ler ttg yi mapping strain and the ATR2 EST cDNA clone, Dr. Chris Town for seeds of the amt-1 strain, and Dr. Krishna Niyogi for a genomic clone of ASA1 and reagents for generating the S115F mutation. We also thank Lisa Bartee, Juan Quiel, Christine Prater, and Julie Blum for technical assistance. This material is based upon work supported by the National Science Foundation under grant IBN 9723172 to J.B.
Footnotes
-
Communicating editor: C. S. Gasser
- Received February 15, 2002.
- Accepted April 4, 2002.
- Copyright © 2002 by the Genetics Society of America
