Characterization of an Unstable Allele of the Arabidopsis HY4 Locus

Corresponding author: Edward P. Bruggemann, Pioneer Hi-Bred International, Inc., Research and Product Development, 7300 N.W. 62nd Ave., Johnston, IA 50131-1004.

Communicating editor: J. CHORY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Arabidopsis HY4 gene encodes the nonessential blue light photoreceptor CRY1. Loss-of-function hy4 mutants have an elongated hypocotyl phenotype after germination under blue light. We previously analyzed 20 independent hy4 alleles produced by fast neutron mutagenesis. These alleles were grouped into two classes based on their genetic behavior and corresponding deletion size: (1) null hy4 alleles that were semidominant over wild type and contained small or moderate-sized deletions at HY4 and (2) null hy4 alleles that were recessive lethal and contained large HY4 deletions. Here we describe one additional fast neutron hy4 mutant, B144, that did not fall into either of these two classes. Mutant B144 was isolated as a heterozygote with an intermediate hy4 phenotype. One allele from this mutant, hy4-B144, contains a large deletion at HY4 and is recessive lethal. The other allele from this mutant, HY4-B144*, appears to be intact and functional but is unstable and spontaneously converts to a nonfunctional hy4 allele. In addition, HY4-B144* is lethal in homozygotes and suppresses local recombination. We discuss genetic and epigenetic mechanisms that may account for the unusual behavior of the HY4-B144* allele.

THE Arabidopsis HY4 gene was originally defined by mutants that had an elongated hypocotyl phenotype after germination under blue light (KOORNNEEF et al. 1980 ). The HY4 gene is nonessential and null hy4 homozygotes are completely viable, yet light-grown seedlings are quite sensitive to the level of HY4 function they express. Defective hy4 alleles are semidominant over wild-type (KOORNNEEF et al. 1980 ; BRUGGEMANN et al. 1996 ), and transgenic plants that overexpress HY4 have severely shortened hypocotyls (LIN et al. 1995B ; LIN et al. 1996 ). When germinated in the dark, hy4 seedlings have a wild-type etiolated phenotype (KOORNNEEF et al. 1980 ; CHORY 1992 ; AHMAD and CASHMORE 1993 ; JACKSON and JENKINS 1995 ). Adult hy4 plants display subtle morphological differences compared to wild-type plants, many of which are related to cell growth and extension (JACKSON and JENKINS 1995 ).

HY4 encodes a protein, CRY1, that is likely a blue light photoreceptor. CRY1 comprises 681 amino acid residues, of which ~500 residues are homologous to photolyase (AHMAD and CASHMORE 1993 ). Two chromophores, flavin adenine dinucleotide and methenyltetrahydrofolate, noncovalently bind to CRY1 (LIN et al. 1995A ; MALHOTRA et al. 1995 ). Mutations that are presumed to interfere with chromophore binding abrogate CRY1 function (AHMAD et al. 1995 ). CRY1 exhibits no detectable photolyase activity (LIN et al. 1995A ; MALHOTRA et al. 1995 ), and a critical tryptophan residue, which is conserved among all known photolyases and implicated in DNA binding, is not conserved in CRY1 (AHMAD and CASHMORE 1993 ). The mechanism by which CRY1 transduces blue light into the appropriate developmental responses is unknown. CRY1 is cytoplasmic and is constitutively expressed (AHMAD and CASHMORE 1993 ; LIN et al. 1996 ). Compared with photolyase, CRY1 contains a carboxy terminus extension of ~180 amino acid residues (AHMAD and CASHMORE 1993 ), which may interact with downstream components of a signal transduction pathway. Mutations in the extension abrogate CRY1 function (AHMAD and CASHMORE 1993 ; AHMAD et al. 1995 ). Genetic evidence suggests that CRY1 interacts with phytochromes PHYA and PHYB (CASAL and BOCCALANDRO 1995 ; AHMAD and CASHMORE 1997 ), whereas other proteins known to play a central role in Arabidopsis photomorphogenesis, such as DET1 and COP1, probably act downstream of CRY1 (CHORY 1992 ; ANG and DENG 1994 ).

We previously isolated 21 independent hy4 mutants that were generated by fast neutron mutagenesis. Twenty of these mutants fell into one of two classes based on their genetic behavior and corresponding deletion size (BRUGGEMANN et al. 1996 ). The first class contained null hy4 alleles that were semidominant over wild type and that generally contained small to moderate-sized deletions at HY4. In all but one allele, at least part of the HY4 gene was still present. The second class contained null hy4 alleles that were recessive lethal. These alleles were lethal in homozygous plants and male gametophytes but did not affect the viability of female gametophytes. All of the recessive lethal hy4 alleles contained large deletions, greater than 8 kb in size, that removed the entire HY4 gene. Because loss of HY4 function does not in itself confer recessive lethality, we proposed that an essential gene of unknown function lies adjacent to HY4 (BRUGGEMANN et al. 1996 ). In this model, small or moderate-sized deletions at HY4 do not affect this adjacent essential gene and confer only the hy4 phenotype, whereas large deletions at HY4 remove both HY4 and the adjacent essential gene and confer recessive lethality along with the hy4 phenotype.

To complete our study of hy4 alleles generated by fast neutron mutagenesis, we describe one additional hy4 mutant, B144, that did not fall into either of the two classes described above. We show that mutant B144 contains two alleles of HY4. One allele, hy4-B144, contains a large deletion at HY4 and is recessive lethal. This allele is identical to the recessive lethal hy4 alleles with large deletions described above. The other allele, HY4-B144*, appears to be structurally intact, transcriptionally active, and completely functional, but it is unstable and can spontaneously convert to a nonfunctional hy4 allele. Furthermore, HY4-B144* is lethal in homozygotes and suppresses local recombination. Although we cannot rule out other explanations, we believe an undetected chromosomal aberration that does not directly affect HY4 function accounts for the genetic behavior of HY4-B144*.

Compared to other genetic systems such as Drosophila or maize, little is known about the effects of ionizing radiation on Arabidopsis chromosomes and the pairing, recombination, and segregation of affected chromosomes during Arabidopsis gametogenesis. Because the HY4 gene is nonessential and hy4 alleles are semidominant over wild type, our collection of hy4 alleles generated by fast neutron mutagenesis contains a broad range of defects. The results described here, combined with our previous characterization of fast neutron-generated hy4 deletions (BRUGGEMANN et al. 1996 ), provide the most complete account to date of the effects of fast neutron mutagenesis in Arabidopsis. In addition, further characterization of the unique properties of mutant B144 should provide insights into Arabidopsis chromosome biology.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Arabidopsis lines and growth conditions:
Mutant B144 (hy4-B144/HY4-B144*) was isolated from M₂ seed generated by fast neutron mutagenesis (Lehle Seeds #M2F-1A-1, Round Rock, TX). Mutant B36 (hy4-B36/hy4-B36) was isolated during the same screen from a different M₂ family of fast neutron mutagenized seed (BRUGGEMANN et al. 1996 ). The hy4-B36 allele contains a deletion that leaves the 3' end of HY4 intact but removes the entire HY4 coding sequence and at least 4 kb of untranscribed 5' sequence (BRUGGEMANN et al. 1996 ). The wild-type genetic background for both mutants was Columbia with a gl1 marker (Lehle Seeds #WT-1A). To separate the B144 alleles, mutant B144 (hy4-B144/HY4-B144*) was backcrossed to one of two wild-type lines, either Columbia [Arabidopsis Biological Resource Center (ABRC) #CS-908, Columbus, OH] or Landsberg with an er marker (ABRC #CS-20). Both wild-type lines carry the wild-type GL1 allele. The hy4 reference allele, hy4-2.23N (ABRC #CS-70), is in a Landsberg er genetic background. Mutant screening, growth conditions, and phenotype scoring were described previously (BRUGGEMANN et al. 1996 ).

Southern and Northern blot analysis:
Genomic DNA preparations and Southern blots were performed as described previously (BRUGGEMANN et al. 1996 ). Two micrograms of genomic DNA was digested by each restriction endonuclease and separated by electrophoresis through gels ranging from 0.5 to 1.0% agarose. Enzymes used included ApaI, BamHI, BclI, BglII, BsaAI, ClaI, DraI, EcoRI, EcoRV, HincII, HindIII, HpaII, KpnI, MscI, MspI, NdeI, PstI, PvuI, SalI, ScaI, SphI, SspI, StuI, XbaI, and XhoI. For Northern blots, total RNA was prepared by phenol/sodium dodecyl sulfate (SDS) extraction and LiCl precipitation (AUSUBEL et al. 1987 ) from ~10 g (fresh weight) of 5-day-old seedlings grown under continuous blue light. Ten micrograms of total RNA was separated by electrophoresis through 1.0% agarose/formaldehyde (AUSUBEL et al. 1987 ). For both Southern and Northern blots, mutant B144 (hy4-B144/HY4-B144*) was either M₆ or M₈ generation, wild type was Columbia gl1, and mutant B36 was previously backcrossed to Columbia gl1 four times.

The HY4 probes were described previously (BRUGGEMANN et al. 1996 ). Briefly, the 5' HY4 probe extends upstream ~6 kb from the middle of exon 3 and includes ~4 kb of untranscribed 5' sequence. The 3' HY4 probe extends ~2.5 kb downstream from the middle of exon 3 and includes ~1 kb of untranscribed 3' sequence. Together, the 5' and 3' HY4 probes span ~8 kb. The HY4 exon 3 probe is 1452 bp and includes nearly all of exon 3. The chalcone synthase (CHS) probe was prepared from plasmid pCHS3.9 (FEINBAUM and AUSUBEL 1988 ), and the chlorophyll a/b binding protein (CAB) probe was prepared from plasmid pAB165 (LEUTWILER et al. 1986 ).

Blots were quantified by using a phosphorimager with software provided by the manufacturer (Molecular Dynamics, Sunnyvale, CA). Band intensities were corrected for background by quantitating an identical area directly above and directly below the band of interest, averaging the background signals obtained, and subtracting the average background from the signal of interest. After correcting for background, HY4 signals were normalized to an internal control, CHS for Southern blots or CAB for Northern blots. For Northern blots, the results from three independent RNA preparations of each genotype were averaged together and presented in Figure 4B.

View larger version (111K): In this window In a new window Download PPT slide   Figure 1. Comparison of wild-type and hy4 phenotypes. Seedlings were germinated and grown for 5 days under continuous blue light. Shown from left to right are Columbia wild type (+/+), short, wild-type phenotype; hy4 deletion homozygote B36 (hy4-B36/hy4-B36), tall hy4 phenotype; mutant B144 (hy4-B144/HY4-B144*), medium hy4 phenotype; putative HY4-B144* homozygote (HY4-B144*/HY4-B144*?), short, phenotypically wild-type segregant from self-pollination of mutant B144; hy4-B144 F1 heterozygote (hy4-B144/+), medium hy4 phenotype, from B144 (hy4-B144/HY4-B144*) x Landsberg er (+/+); HY4-B144* F1 heterozygote (HY4-B144*/+), short, wildtype phenotype, from the same cross.

View larger version (93K): In this window In a new window Download PPT slide   Figure 2. Southern blot to compare the HY4 locus in mutant B144 with wild type. Genomic DNA was prepared from mutant B144 (hy4-B144/HY4-B144*) and Columbia wild type (+/+) and digested with the indicated restriction endonucleases. The blot was probed simultaneously with the HY4 5' and HY4 3' probes.

View larger version (27K): In this window In a new window Download PPT slide   Figure 3. Quantitative Southern blot to determine HY4 copy number in mutant B144. (A) Genomic DNA was prepared from mutant B144 (hy4-B144/HY4-B144*), Columbia wild type (+/+), and hy4 deletion homozygote B36 (hy4-B36/hy4-B36). A Southern blot similar to the one shown in Figure 2 was probed simultaneously with the HY4 exon 3 probe and a chalcone synthase probe (CHS). A representative blot using SspI is shown. (B) For each digest, the signals from both probes were quantitated by phosphorimaging, the HY4 signal was normalized to the CHS signal, and the normalized HY4 signal from wild type was divided by the normalized HY4 signal from mutant B144.

View larger version (18K): In this window In a new window Download PPT slide   Figure 4. Quantitative Northern blot to determine the HY4 mRNA level in mutant B144. (A) Total RNA from hy4 deletion homozygote B36 (hy4-B36/hy4-B36), Columbia wild type (+/+), and mutant B144 (hy4-B144/HY4-B144*) was prepared from 5-day-old seedlings grown under continuous blue light. The blot was probed with the 5' HY4 probe, then stripped and probed with a chlorophyll a/b binding protein (CAB) probe. (B) For each RNA preparation, the signals from both probes were quantitated by phosphorimaging, and the HY4 signal was normalized to the CAB signal. The signal from wild type was set at 100%, and the signal from mutant B36 was set at 0%. Data from three independent RNA preparations of each genotype were averaged together.

Genetic analysis, HY4 genotype assay, and mapping:
For genetic analysis of the B144 alleles, mutant B144 (hy4-B144/HY4-B144*) was backcrossed to either Columbia or Landsberg er. Both of these wild-type lines carry the wild-type GL1 allele and display wild-type trichomes, whereas mutant B144 carries a gl1 allele and lacks trichomes. The gl1 allele is completely recessive to the wild-type GL1 allele, and the GL1 locus is unlinked to the HY4 locus. Crosses were confirmed by scoring trichomes in either the F₁ or F₂ generation, depending on whether mutant B144 was used as the female or as the male in the cross.

A molecular assay for HY4 parental genotype was developed based on the CAPS (Cleaved Amplified Polymorphic Sequences) strategy (KONIECZNY and AUSUBEL 1993 ). We designed HY4 PCR primers (forward primer: 5'-CTAGAAGCTGCTTCAAGAGC-3'; reverse primer: 5'-AATCGTTGTCAAAGGTAACG-3') that amplified the 3' region of HY4. A HaeIII site present in the amplified fragment from Columbia, but absent from the Landsberg fragment, allowed us to distinguish HY4 alleles from the two wild-type lines.

To perform the assay, seedlings were grown for 5 days under continuous blue light, then placed in 1.5-ml polypropylene microfuge tubes and frozen on dry ice. Mini-prep DNA suitable for PCR was prepared by a protocol modified from one previously described (KLIMYUK et al. 1993 ). Briefly, the frozen plants were crushed with a microfuge pestle. After addition of 10 µl 0.5 M NaOH, the samples were heated at 100° for 1 min, then neutralized with 100 µl 200 mM Tris pH 8.0, 1 mM EDTA. PCR reactions were performed in 40 µl total volume with 10 mM Tris pH 8.3, 50 mM KCl, 1.5 mM MgCl₂, 200 µM dNTPs, 0.5 U Taq polymerase, 0.5 µM of each primer, and 2 µl mini-prep DNA. The reactions were denatured for 2 min at 94°, followed by 40 cycles of amplification: 1 min at 92°, 1 min at 55°, and 3 min at 72°. HaeIII digests were performed by adding 14 µl water, 6 µl 10x reaction buffer, and 1 µl (10 U) enzyme to each sample, then incubating the samples at 37° for 2 hr. Finally, 2 µl 10x sample buffer was added to each reaction, the volume was reduced to ~20 µl by evaporation at 65°, and the reactions were analyzed by electrophoresis through 0.8% agarose stained with ethidium bromide.

CAPS assays for GA1 and AG have been previously described (KONIECZNY and AUSUBEL 1993 ) and were performed using the above protocol. The calculation of genetic distances and standard errors presented in Figure 6 was performed using the Kosambi mapping function as previously described (KOORNNEEF and STAM 1992 ).

View larger version (26K): In this window In a new window Download PPT slide   Figure 5. Assay to distinguish between Columbia and Landsberg HY4 alleles. Seedlings were scored for phenotype after germination and growth for 5 days under continuous blue light. Mini-prep DNA was then amplified by PCR using the HY4 CAPS primers and digested with HaeIII. An agarose gel stained with ethidium bromide is shown. From left to right: Columbia gl1 (+/+); Landsberg er (+/+); Columbia gl1 x Landsberg er F1 (+/+); mutant B144 (hy4-B144/HY4-B144*), medium hy4 phenotype; hy4-B144 F1 heterozygote (hy4-B144/+), medium hy4 phenotype, from B144 (hy4-B144/HY4-B144*) x Landsberg er (+/+); HY4-B144* F1 heterozygote (HY4-B144*/+), short, wild-type phenotype, from the same cross.

View larger version (19K): In this window In a new window Download PPT slide   Figure 6. Recombination in regions flanking the HY4 locus. (A) A diagram of chromosome 4 with CAPS marker loci is shown (KONIECZNY and AUSUBEL 1993 ). The approximate location of the centromere is indicated with a dot. (B) F2 populations of three HY4 alleles described in Table 4 were genotyped by CAPS assay at two loci flanking HY4, GA1, and AG. The genetic distance in cM was calculated for two intervals: GA1 to HY4 and HY4 to AG.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

To identify Arabidopsis genes required for normal photomorphogenic development under blue light, we screened M₂ seed derived from fast neutron mutagenesis for seedlings that displayed an elongated hypocotyl phenotype after germination under blue light (BRUGGEMANN et al. 1996 ). From this screen we obtained 21 independent hy4 mutants, 20 of which contained hy4 alleles that could be classified either as semidominant or as recessive lethal. When a semidominant hy4 heterozygote with a medium hy4 phenotype is allowed to self-pollinate, the progeny segregate tall hy4 homozygotes, medium hy4 heterozygotes, and short, wild-type homozygotes in the familiar 1:2:1 Mendelian ratios. Recessive lethal hy4 alleles are lethal in homozygous plants and male gametophytes, but female gametophytes are unaffected and viable. When a recessive lethal hy4 heterozygote with a medium hy4 phenotype is allowed to self-pollinate, the progeny segregate medium hy4 heterozygotes and short, wild-type homozygotes in a 1:1 ratio.

The remaining hy4 mutant, B144, was isolated as an M₂ individual with a medium hy4 phenotype identical to a hy4 heterozygote. However, unlike the progeny from other self-pollinated hy4 heterozygotes described above, nearly all of the progeny from self-pollinated mutant B144 have a medium hy4 phenotype, identical to the parent, whereas only a few progeny have a short, wild-type phenotype (Figure 1). We have allowed B144 individuals with a medium hy4 phenotype to self-pollinate through the M₁₀ generation, and this segregation pattern has not varied significantly from the original M₂ mutant individual (Table 1). Unlike the other hy4 mutants we isolated, B144 adults are rather weak and partially infertile compared to wild-type plants.

Mutant B144 is heterozygous for a deletion at HY4:
We previously showed that most fast neutron hy4 alleles contain deletions that are easily detectable by Southern blot (BRUGGEMANN et al. 1996 ). To determine if mutant B144 contains a deletion at HY4, we performed Southern blot analysis with B144 and wild-type DNA digested by a variety of restriction enzymes. We probed the blots with HY4 probes that cover the HY4 transcription unit and span ~8 kb. The results revealed no obvious differences between mutant B144 and wild type at the HY4 locus (Figure 2). However, close examination of these blots revealed that the HY4 signal strength of most B144 bands was somewhat weaker than those of wild type. This observation suggested that B144 might be heterozygous for a deletion at the HY4 locus.

To test this idea, we performed a quantitative Southern blot. We simultaneously probed a blot with a HY4 probe and with a chalcone synthase (CHS) probe (Figure 3A). CHS is a single copy gene unlinked to HY4 (CHANG et al. 1988 ; FEINBAUM and AUSUBEL 1988 ). The HY4 signals from each digest were quantitated and normalized to CHS, then the normalized HY4 signals from B144 and wild type were compared. The results showed that the HY4 signal from B144 was half the signal from wild type (Figure 3B). From these results we concluded that mutant B144 contains only one copy of the HY4 gene and is heterozygous for a deletion at the HY4 locus. We designated this deletion allele hy4-B144. Returning to the original Southern blot analysis (Figure 2), we concluded that the remaining HY4 allele in B144, which we designated HY4-B144*, is structurally intact.

We also examined the steady-state HY4 mRNA levels in mutant B144 by performing a quantitative Northern blot with B144 and wild-type RNA prepared from seedlings. The blot was probed with a HY4 probe, then stripped and probed with chlorophyll a/b binding protein (CAB) probe (Figure 4A). The HY4 signals were quantitated and normalized to CAB, and then the normalized HY4 signals from B144 and wild type were compared. Data from three independent RNA preparations of each genotype were averaged together. The results showed that the HY4 mRNA signal from B144 was half the signal from wild type (Figure 4B). This result was consistent with the presence of only one copy of the HY4 gene in mutant B144 and showed that the HY4-B144* allele is transcriptionally active, providing one full gene dose of HY4 mRNA. To show that CAB mRNA levels are unaffected by a hy4 mutation, we also probed RNA prepared from a hy4 deletion homozygote, mutant B36 (Figure 4A).

These results accounted for the phenotype of mutant B144. Mutant B144 has a medium hy4 phenotype, identical to a hy4 heterozygote, because it contains only one structurally intact HY4 allele that is transcriptionally active, HY4-B144*. The deletion allele, hy4-B144, provides no HY4 transcript and no HY4 function. However, these results could not explain the genetic behavior of mutant B144, which after self-pollination yields almost entirely heterozygotes (Table 1). What could account for the absence of both homozygous classes in the progeny of a self-pollinated heterozygote? To address this question we isolated the two alleles and studied them separately.

The hy4-B144 allele is recessive lethal:
The absence of tall hy4 phenotypes from mutant B144 populations suggested that the hy4-B144 allele might be a recessive lethal hy4 allele, similar to other large deletions at HY4 that we previously analyzed (BRUGGEMANN et al. 1996 ). To isolate the hy4-B144 allele, we crossed mutant B144 (hy4-B144/HY4-B144*) to Columbia wild type (+/+) and recovered F₁ heterozygotes (hy4-B144/+) with a medium hy4 phenotype. We allowed these F₁ heterozygotes to self-pollinate and performed reciprocal backcrosses to Columbia wild type.

When allowed to self-pollinate, HY4-B144 F₁ heterozygotes yielded F₂ populations that segregated medium hy4 phenotypes and short, wild-type phenotypes in a 1:1 ratio (Table 2). When HY4-B144 F₁ heterozygotes were backcrossed to wild type using the heterozygotes as the female, the backcross progeny also segregated medium hy4 phenotypes and short, wild-type phenotypes in a 1:1 ratio (Table 2). This result indicated that the hy4-B144 allele does not affect the viability of female gametophytes. However, the progeny from the reciprocal backcross, in which HY4-B144 F₁ heterozygotes were used as the male, segregated ~2% medium hy4 phenotypes and 98% short, wild-type phenotypes (Table 2). This result indicated that the hy4-B144 allele is poorly transmitted through male gametophytes, probably because this allele is usually lethal in male gametophytes.

View this table: In this window In a new window   Table 2. Genetic behavior of the hy4-B144 allele

We also crossed HY4-B144 F₁ heterozygotes to a hy4 reference line (hy4-2.23N/hy4-2.23N). The hy4-2.23N allele is semidominant, and homozygotes have a tall hy4 phenotype (KOORNNEEF et al. 1980 ). The progeny of this cross segregated tall hy4 phenotypes (hy4-B144/hy4-2.23N) and medium hy4 phenotypes (+/hy4-2.23N) in a 1:1 ratio (Table 2). The tall hy4 phenotype of the compound heterozygote (hy4-B144/hy4-2.23N) suggested that a hy4-B144 homozygote would also have a tall hy4 phenotype, if it was viable. Because we never observed a tall hy4 phenotype among thousands of F₂ progeny from self-pollinated HY4-B144 F₁ heterozygotes (Table 2), we concluded that the hy4-B144 allele is lethal in homozygous plants.

Taken together, these results demonstrated that hy4-B144 is a recessive lethal hy4 allele, identical to other recessive lethal hy4 alleles with large deletions that we previously analyzed (BRUGGEMANN et al. 1996 ). However, these results still could not account for the unusual genetic behavior of mutant B144 and suggested that the HY4-B144* allele might influence the genetic behavior of mutant B144.

The HY4-B144* allele is functional:
Mutant B144 was derived from a Columbia gl1 genetic background. To isolate the B144 alleles and study them separately in parallel, we crossed mutant B144 (hy4-B144/HY4-B144*) to Landsberg er (+/+), a polymorphic wild-type strain. Both medium hy4 phenotypes and short, wild-type phenotypes occurred in the F₁ generation (Figure 1 and Table 3). We also designed HY4 PCR primers that allowed us to distinguish the Columbia HY4 allele from the Landsberg HY4 allele. The PCR-amplified fragment from both wild-type alleles is ~1.6 kb, but upon digestion with HaeIII, the Columbia HY4 fragment yields two fragments of ~1.4 and 0.2 kb, and the Landberg HY4 fragment remains 1.6 kb (Figure 5). Using this assay, we genotyped individual F₁ and F₂ seedlings at the HY4 locus after scoring their phenotype. We also genotyped seedlings for two markers flanking HY4, GA1, and AG by an identical assay (KONIECZNY and AUSUBEL 1993 ).

View this table: In this window In a new window   Table 3. Cosegregation of the medium hy4 phenotype with the hy4-B144 allele

The results showed that the medium hy4 phenotype of mutant B144 cosegregated with the hy4-B144 allele. According to the PCR assay, all of the F₁ plants with a medium hy4 phenotype contained only the Landsberg wild-type HY4 allele (Figure 5), yet these plants were heterozygous for the flanking markers GA1 and AG (data not shown), indicating that the absence of the Columbia HY4 allele is due to a deletion of the HY4 locus. These results confirmed our earlier conclusion, based on the Southern blot analysis, that mutant B144 is heterozygous for a deletion at HY4 and that this deletion is responsible for the medium hy4 phenotype of mutant B144.

All of the F₁ plants with a short, wild-type phenotype contained two HY4 alleles, one of which was a Columbia allele, HY4-B144*, and one of which was the Landsberg wild-type HY4 allele (Figure 5). This result demonstrated that the HY4-B144* gene product is functional, because all F₁ heterozygotes containing this allele displayed a short, wild-type phenotype, indicating that they contained two functional HY4 alleles.

Surprisingly, the distribution of genotypes in the F₁ population was not the 1:1 ratio expected from crossing a heterozygote to wild type. Approximately 75% of the F₁ individuals contained the hy4-B144 allele, whereas only ~25% contained the HY4-B144* allele (Table 3). Because the cross was performed using mutant B144 (hy4-B144/HY4-B144*) as the female, this result suggested that the HY4-B144* allele is transmitted somewhat less efficiently through female gametophytes than the hy4-B144 allele. This was an indication that the HY4-B144* allele is not a true wild-type allele. To further characterize the HY4-B144* allele, we examined its segregation and recombination behavior in the F₂ population.

The HY4-B144* allele is lethal in homozygotes:
To generate segregating F₂ populations, we allowed both hy4-B144 and HY4-B144* F₁ heterozygotes from the cross to Landsberg to self-pollinate. The hy4-B144 F₂ population segregated medium hy4 phenotypes and short, wild-type phenotypes in a 1:1 ratio, typical of a recessive lethal hy4 allele (Table 3). The PCR assay indicated that all of the individuals in the hy4-B144 F₂ population contained at least one Landsberg wild-type HY4 allele (data not shown). This result implied that none of the F₂ individuals was homozygous for the Columbia hy4-B144 allele, which confirmed that hy4-B144 is homozygous lethal.

The HY4-B144* F₂ population yielded only short, wild-type phenotypes (Table 3). The PCR assay showed that the HY4-B144* F₂ population, like the hy4-B144 F₂ population, lacked Columbia homozygotes (Table 4). From this result we concluded that the HY4-B144* allele, like the hy4-B144 allele, is homozygous lethal. This conclusion in turn explained the unusual segregation pattern of mutant B144 (hy4-B144/HY4-B144*). Self-pollination of mutant B144 yields almost entirely heterozygous progeny because both homozygous classes are inviable. In addition, this result also provided strong evidence that the HY4-B144* allele is not truly wild type.

The distribution of HY4 genotypes in the F₂ populations revealed another difference between the two B144 alleles (Table 4). The 1:1 ratio in the hy4-B144 F₂ population is typical of a recessive lethal hy4 allele that is lethal in male gametophytes but does not affect female gametophytes. In the HY4-B144* F₂ population, the ratio of heterozygotes to Landsberg homozygotes was very close to 2:1. Although the F₂ population used here was small, this result suggested that the Columbia homozygotes were absent from the HY4-B144* F₂ population, primarily due to a lack of viability of HY4-B144* homozygotes rather than due to any major effect this allele might have on the viability of gametophytes. Unlike the backcross results described in the previous section, the F₂ results described here provided no evidence that the HY4-B144* allele affects female gametophytes. This apparent contradiction is discussed below.

The HY4-B144* allele suppresses local recombination:
We also genotyped both F₂ populations at two flanking loci, GA1 and AG, and calculated the genetic distance between these loci and the HY4 locus (Figure 6). The medium hy4 phenotype conferred by the hy4-B144 allele mapped to the HY4 locus, ~4 cM from GA1 and 25 cM from AG. These distances were not significantly different from the wild-type HY4 allele values in the small mapping populations used and again confirmed that the medium hy4 phenotype of mutant B144 is due to the HY4 deletion. The HY4-B144* allele also mapped to a locus between GA1 and AG, ~2 cM from GA1 and 6 cM from AG. The 2-cM distance measured for the GA1 to HY4 interval was not significantly different from the wild-type HY4 allele, but the 6-cM distance measured for the HY4 to AG interval was significantly less than the wild-type value. From this result we concluded that the HY4-B144* allele suppresses local recombination, specifically in the HY4 to AG interval. This result was yet another indication that the HY4-B144* allele is not truly wild type.

The HY4-B144* allele is unstable:
Although the HY4-B144* allele appears to be structurally intact by Southern blot analysis, transcriptionally active by Northern blot analysis, and functional by the phenotype it confers, it is clear that a defect is associated with this allele. HY4-B144* is lethal in homozygotes and suppresses local recombination. Further investigation of this allele revealed that it could spontaneously convert to a nonfunctional hy4 allele with high frequency.

The instability of the HY4-B144* allele was first revealed by the genetic behavior of the short, phenotypically wild-type segregants from mutant B144. After self-pollination, mutant B144 (hy4-B144/HY4-B144*) segregates ~4% short, wild-type phenotypes in its progeny (Figure 1 and Table 1). We inferred that these short segregants contain two functional HY4 alleles because nonfunctional hy4 alleles are semidominant over wild-type HY4 alleles. We allowed 41 short segregants to self-pollinate and scored the phenotypes in each progeny family. Eight individuals bred true and yielded only short, wild-type phenotypes in their progeny (Table 5). The remaining 33 individuals segregated ~1/3 medium hy4 phenotypes and 2/3 short, wild-type phenotypes after self-pollination (Table 5). From their phenotype we inferred that the medium hy4 progeny contained one functional HY4 allele and one nonfunctional hy4 allele, and we concluded that functional HY4 alleles had spontaneously converted to nonfunctional hy4 alleles in these 33 families.

We believe that the HY4 allele undergoing this spontaneous conversion to hy4 is HY4-B144*. The HY4-B144* allele is the only functional HY4 allele that mutant B144 (hy4-B144/HY4-B144*) contains, so it seems likely that the short segregants from self-pollination of mutant B144 inherited two copies of HY4-B144* (which implies that the homozygous lethality of HY4-B144* is somewhat leaky). However, because we were uncertain of the genotype of these plants, we sought to observe the same effect in plants of known genotype.

We constructed HY4-B144* heterozygotes (HY4-B144*/+) by crossing mutant B144 (hy4-B144/HY4-B144*) to Columbia wild type (+/+). In the F₁ generation we identified 221 HY4-B144* heterozygotes with a short, wild-type phenotype, allowed each to self-pollinate, and scored the phenotypes in the F₂ progeny families. We inferred that the HY4-B144* allele was providing full HY4 function in the F₁ heterozygotes because they had a short, wild-type phenotype. Most of the HY4-B144* heterozygotes (196) bred true, yielding only short, wild-type phenotypes in the F₂ generation (Table 5). However, 25 HY4-B144* heterozygotes segregated ~1/3 medium hy4 phenotypes and 2/3 short, wild-type phenotypes in their F₂ progeny (Table 5). The medium hy4 phenotype of these F₂ plants was identical to that of hy4 heterozygotes and implied that the HY4-B144* allele was no longer providing HY4 function. To confirm that these F₂ plants contained a defective hy4 allele, we crossed them to the hy4 reference line (hy4-2.23N/hy4-2.23N). The progeny from this cross segregated ~50% tall hy4 phenotypes and 50% medium hy4 phenotypes, indicating that the F₂ parent was heterozygous for a defective hy4 allele (Table 5). We concluded from these results that HY4-B144* is unstable and could spontaneously convert to a nonfunctional hy4 allele.

These results differ from those presented in Table 3 due to the numbers of HY4-B144* F₁ heterozygotes involved. From the results presented above, we calculate that only ~11% (25/221) of HY4-B144* F₁ heterozygotes can be expected to segregate medium hy4 phenotypes in the F₂ generation. To generate the data in Figure 3, four randomly chosen HY4-B144* F₁ heterozygotes were allowed to self-pollinate to generate the F₂ population. By chance, none of these four segregated medium hy4 phenotypes in their F₂ progeny.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Among the 21 independent hy4 mutants that we isolated from fast neutron mutagenesis was the unusual mutant B144. This mutant has a medium hy4 phenotype, identical to a hy4 heterozygote. Unlike other hy4 heterozygotes, when mutant B144 is allowed to self-pollinate, nearly all the progeny have a medium hy4 phenotype identical to the parent. To understand the molecular basis of this segregation pattern, we examined the genetic behavior and molecular biology of both HY4 alleles from mutant B144.

The hy4-B144 allele cannot alone explain the genetic behavior of mutant B144:
Southern blot analysis demonstrated that mutant B144 is heterozygous for a deletion at the HY4 locus. We designated this deletion allele hy4-B144. By backcrossing mutant B144 to wild type, we found that the medium hy4 phenotype of mutant B144 cosegregated with the hy4-B144 allele and concluded that the deletion accounts for the medium hy4 phenotype of mutant B144.

The hy4-B144 allele was lethal in homozygous plants and in male gametophytes but had no effect on female gametophytes. These results indicated that hy4-B144 is a recessive lethal hy4 allele, identical to other recessive lethal hy4 alleles we have studied (BRUGGEMANN et al. 1996 ). Presumably, hy4-B144 contains a large deletion that removes both HY4 and an adjacent essential gene, which accounts for the hy4 phenotype and the recessive lethality. In fact, the deletion in hy4-B144 extends past marker locus nga8 (data not shown), which is ~100 kb distant from the HY4 locus (SCHMIDT et al. 1996 ).

The genetic characteristics of the hy4-B144 allele were familiar to us from our previous analysis of recessive lethal hy4 alleles. Yet mutant B144 does not behave like a recessive lethal hy4 heterozygote. These considerations suggested that the remaining HY4 allele in mutant B144, designated HY4-B144*, might be influencing the genetic behavior of mutant B144. Consequently, we turned our attention to the HY4-B144* allele to determine if it could account for the unusual genetic behavior of mutant B144.

The HY4-B144* allele accounts for the genetic behavior of mutant B144:
Southern blot analysis showed that the HY4-B144* allele is structurally intact. Northern blot analysis showed that this allele is transcriptionally active and provides one complete gene dose of HY4 mRNA. After backcrossing mutant B144 to wild type, HY4-B144* F₁ heterozygotes had a short, wild-type phenotype. This showed that the HY4-B144* coding sequence is functional because nonfunctional hy4 alleles are semidominant over wild-type HY4 alleles. Nothing in these results suggested a defect associated with the HY4-B144* allele.

Detailed genetic analysis, however, revealed that HY4-B144* is not wild type. HY4-B144* homozygotes were absent from the F₂ population, which indicated that the HY4-B144* allele is homozygous lethal. From this result we concluded that mutant B144 (hy4-B144/HY4-B144*) is actually a compound heterozygote and that the properties of the two alleles together account for its unusual genetic behavior. In particular, both the hy4-B144 allele and the HY4-B144* allele are lethal in homozygous plants, yet they complement each other so that the compound heterozygote, mutant B144, is viable. When mutant B144 is allowed to self-pollinate, only the heterozygous progeny are viable, and progeny that are homozygous for either allele are inviable. If the homozygous lethality of each allele is due to the loss of an essential gene, then the successful complementation implies that hy4-B144 and HY4-B144* affect different essential genes.

When mutant B144 was backcrossed to wild type, the deficiency of HY4-B144* heterozygotes in the F₁ generation suggested that the HY4-B144* allele is not efficiently transmitted through female gametophytes. In contrast, the segregation pattern of HY4-B144* in the F₂ generation suggested that the HY4-B144* allele is efficiently transmitted through both male and female gametophytes. The basis of this apparent discrepancy is not clear. We speculate that interactions between HY4-B144* and its homolog during gametogenesis may be responsible. Thus, when the HY4-B144* allele is paired with a large hy4 deletion, as it is in mutant B144 (hy4-B144/HY4-B144*), the HY4-B144* allele is not efficiently transmitted through female meiosis, perhaps due to some problem in homolog pairing, recombination, or segregation. But when the HY4-B144* allele is paired with a wild-type HY4 allele, as it is in an F₁ heterozygote (HY4-B144*/+), the problem does not occur, and the HY4-B144* allele is transmitted faithfully during female meiosis.

In the F₂ population we also found that recombination in the interval HY4 to AG was significantly less than recombination in the same interval adjacent to a wild-type HY4 allele. This result suggested that recombination was suppressed adjacent to the HY4-B144* allele, specifically in the interval from HY4 to AG.

The instability of HY4-B144* is an additional characteristic of this allele:
During our analysis of the genetic behavior of HY4-B144*, we discovered that this allele is unstable and spontaneously converts to a nonfunctional hy4 allele. When the mutant B144 (hy4-B144/HY4-B144*) is allowed to self-pollinate, the progeny always segregate a few plants with a short, wild-type phenotype. We believe that these short segregants represent rare HY4-B144* homozygotes (HY4-B144*/HY4-B144*?). When we allowed these short segregants to self-pollinate, many segregated medium hy4 phenotypes in their progeny. We also constructed HY4-B144* F₁ heterozygotes (HY4-B144*/+) with a short, wild-type phenotype and allowed them to self-pollinate. Several segregated medium hy4 phenotypes in the F₂ generation. We believe that in both cases the appearance of medium hy4 phenotypes after self-pollination is due to the spontaneous conversion of HY4-B144* alleles to nonfunctional hy4 alleles.

In our experience, hy4 alleles are always semidominant over wild-type HY4 alleles, and hy4 heterozygotes always have a medium hy4 phenotype easily distinguishable from the short phenotype of wild-type plants. During our previous genetic analysis of 20 independent hy4 alleles generated by fast neutron mutagenesis, we never observed a short, phenotypically wild-type plant that contained a nonfunctional hy4 allele and segregated medium hy4 phenotypes after self-pollination. In contrast, plants with a short, wild-type phenotype that contained at least one HY4-B144* allele often segregated medium hy4 phenotypes after self-pollination. We concluded that functional HY4-B144* alleles had spontaneously converted to nonfunctional hy4 alleles.

Two further details may provide clues into the underlying mechanism of conversion. First, regardless of the parental genotype, medium hy4 phenotypes and short, wild-type phenotypes always occur in a ratio of 1:2. This small number ratio suggests that normal Mendelian segregation follows the spontaneous conversion of a HY4-B144* allele to a nonfunctional hy4 allele. The constant ratio further suggests that somatic conversion of the HY4-B144* allele is not involved. Somatic conversion would produce hy4 sectors of different sizes among the parental plants, which after self-pollination would produce a wide range of frequencies of medium hy4 phenotypes. Second, the genetic behavior of individuals with a medium hy4 phenotype that results from the spontaneous conversion of a HY4-B144* allele to a nonfunctional hy4 allele suggests that they contain a recessive lethal hy4 allele (data not shown). Although our data are limited, they suggest that the conversion produces a large hy4 deletion.

A chromosome aberration might account for the HY4-B144* allele:
The unusual genetic behavior of the HY4-B144* allele may be summarized as three significant effects: (1) HY4-B144* is lethal in homozygotes, (2) HY4-B144* suppresses recombination in the interval from HY4 to AG, and (3) HY4-B144* is unstable and spontaneously converts to a nonfunctional hy4 allele. Clearly, the HY4-B144* allele, or the chromosome carrying this allele, must carry a heritable defect to account for these various genetic effects. In our Southern blot analysis, we detected no obvious structural rearrangement in or near HY4-B144*. The suppression of recombination in the interval from HY4 to AG suggests that the defect lies somewhere in this interval, whereas the homozygous lethality suggests that an essential gene function has been affected by this defect. The spontaneous conversion to hy4 suggests that the defect is unstable and frequently resolves to another state that affects HY4 function.

We believe that the chromosome that carries HY4-B144* contains an undetected chromosome aberration. In this model, a structural rearrangement of the chromosome in the interval HY4 to AG, perhaps involving a duplication, inversion, and/or translocation of a chromosome segment, is responsible for the genetic characteristics of the HY4-B144* allele. This structural rearrangement could account for the suppression of recombination by interfering with homolog pairing during meiosis and, if it disrupted an essential gene, it would account for the homozygous lethality. To account for the instability of HY4-B144*, we suggest that this rearrangement easily undergoes further rearrangements during meiosis, generating a large hy4 deletion that is recessive lethal. Because the original aberration does not directly affect the HY4 locus, the HY4-B144* allele is actually a wild-type HY4 allele. Many examples of well-characterized chromosomal aberrations exist in Drosophila, some of which are homozygous lethal and/or suppress recombination (ASHBURNER 1989 ). Duplications at the Bar locus are unstable due to unequal crossing over (ASHBURNER 1989 ).

We have also considered epigenetic mechanisms in which an alteration of chromatin structure, perhaps due to a distant chromosome rearrangement, is directly responsible for the various genetic effects observed with HY4-B144*. An altered chromatin environment could interfere with recombination during meiosis, and it could affect the transcription of an essential gene, causing the homozygous lethality. In this model, the instability of the HY4-B144* allele would reflect the unstable nature of the altered chromatin state. However, epigenetic models seem unlikely to be correct given the constant 1:2 ratio segregation pattern following conversion, and the recessive lethal genetic behavior of the converted allele. Attempts to identify changes in methylation status at HY4 that could account for the behavior of HY4-B144* were not successful (data not shown).

It will be challenging to identify the precise chromosomal rearrangement that is involved. Cytogenetic analysis in Arabidopsis is primitive and resolution is poor, so direct observation of chromosomes that carry HY4-B144* is unlikely to yield satisfactory results. Several of our results, however, suggest further avenues of exploration. Characterization of the interactions between HY4-B144* and other alleles of HY4 during gametogenesis will help define the homologous pairing and recombination behavior of the putative rearrangement. Further characterization of the suppression of recombination adjacent to HY4-B144* may help localize the putative rearrangement. Molecular characterization of the converted HY4-B144* allele will determine whether a large deletion is the result. We are currently pursuing all of these approaches and expect the results to provide important insights into the functional organization of Arabidopsis chromosomes, the effects of ionizing radiation in Arabidopsis, and the meiotic behavior of chromosome aberrations during Arabidopsis gametogenesis.

	ACKNOWLEDGMENTS

We thank Dr. J. BENDER, who provided protocols and advice concerning the plant DNA mini-preps and CAPS assays, and Dr. E. RICHARDS for suggesting the quantitative Southern blot analysis. We also thank Dr. J. BENDER and Dr. J. KENNISON for reviewing the manuscript and providing helpful comments. This work was supported by the National Institute of Child Health and Human Development Division of Intramural Research (Bethesda, MD).

Manuscript received January 30, 1998; Accepted for publication March 23, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AHMAD, M. and A. R. CASHMORE, 1993  HY4 gene of A. thaliana encodes a protein with characteristics of a blue-light photoreceptor. Nature 366:162-166[Medline].

AHMAD, M. and A. R. CASHMORE, 1997  The blue-light receptor cryptochrome 1 shows functional dependence on phytochrome A or phytochrome B in Arabidopsis thaliana.. Plant J. 11:421-427[Medline].

AHMAD, M., C. LIN, and A. R. CASHMORE, 1995  Mutations throughout an Arabidopsis blue-light photoreceptor impair blue-light-responsive anthocyanin accumulation and inhibition of hypocotyl elongation. Plant J. 8:653-658[Medline].

ANG, L.-H. and X.-W. DENG, 1994  Regulatory hierarchy of photomorphogenic loci: allele-specific and light-dependent interaction between the HY5 and COP1 loci. Plant Cell 6:613-628[Abstract/Free Full Text].

ASHBURNER, M., 1989 Drosophila: A Laboratory Handbook. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

AUSUBEL, F. M., R. BRENT, R. K. KINGSTON, D. D. MOORE, J. G. SEIDMAN et al., (Editors), 1987 Current Protocols in Molecular Biology. John Wiley and Sons, New York.

BRUGGEMANN, E., K. HANDWERGER, C. ESSEX, and G. STORZ, 1996  Analysis of fast neutron-generated mutants at the Arabidopsis thaliana HY4 locus. Plant J. 10:755-760[Medline].

CASAL, J. J. and H. BOCCALANDRO, 1995  Co-action between phytochrome B and HY4 in Arabidopsis thaliana.. Planta 197:213-218[Medline].

CHANG, C., J. L. BOWMAN, A. W. DEJOHN, E. S. LANDER, and E. M. MEYEROWITZ, 1988  Restriction fragment length polymorphism linkage map for Arabidopsis thaliana.. Proc. Natl. Acad. Sci. USA 85:6856-6860[Abstract/Free Full Text].

CHORY, J., 1992  A genetic model for light-regulated seedling development in Arabidopsis.. Development 115:337-354[Abstract].

FEINBAUM, R. L. and F. M. AUSUBEL, 1988  Transcriptional regulation of the Arabidopsis thaliana chalcone synthase gene. Mol. Cell. Biol. 8:1985-1992[Abstract/Free Full Text].

JACKSON, J. A. and G. I. JENKINS, 1995  Extension-growth responses and expression of flavonoid biosynthesis genes in the Arabidopsis hy4 mutant. Planta 197:233-239[Medline].

KLIMYUK, V. I., B. J. CARROLL, C. M. THOMAS, and J. D. G. JONES, 1993  Alkali treatment for rapid preparation of plant material for reliable PCR analysis. Plant J. 3:493-494[Medline].

KONIECZNY, A. and F. M. AUSUBEL, 1993  A procedure for mapping Arabidopsis mutations using co-dominant ecotype-specific PCR-based markers. Plant J. 4:403-410[Medline].

KOORNNEEF, M., and P. STAM, 1992 Genetic analysis, pp. 83–99 in Methods in Arabidopsis Research, edited by C. KONCZ, N.-H. CHUA and J. SCHELL. World Scientific Publishing Co., Singapore.

KOORNNEEF, M., E. ROLFF and C. J. P. SPRUIT, 1980 Genetic control of light-inhibited hypocotyl elongation in Arabidopsis thaliana (L.) Heynh. Z. Pflanzenphysiol. Bd. 100. S.: 147–160.

LEUTWILER, L. S., E. M. MEYEROWITZ, and E. M. TOBIN, 1986  Structure and expression of three light-harvesting chlorophyll a/b-binding protein genes in Arabidopsis thaliana.. Nucleic Acids Res. 14:4051-4064[Abstract/Free Full Text].

LIN, C., D. E. ROBERTSON, M. AHMAD, A. A. RAIBEKAS, and M. S. JORNS et al., 1995a  Association of flavin adenine dinucleotide with the Arabidopsis blue light receptor CRY1. Science 269:968-970[Abstract/Free Full Text].

LIN, C., M. AHMAD, D. GORDON, and A. R. CASHMORE, 1995b  Expression of an Arabidopsis cryptochrome gene in transgenic tobacco results in hypersensitivity to blue, UV-A, and green light. Proc. Natl. Acad. Sci. USA 92:8423-8427[Abstract/Free Full Text].

LIN, C., M. AHMAD, and A. R. CASHMORE, 1996  Arabidopsis cryptochrome 1 is a soluble protein mediating blue light-dependent regulation of plant growth and development. Plant J. 10:893-902[Medline].

MALHOTRA, K., S.-T. KIM, A. BATSCHAUER, L. DAWUT, and A. SANCAR, 1995  Putative blue-light photoreceptors from Arabidopsis thaliana and Sinapis alba with a high degree of sequence homology to DNA photolyase contain the two photolyase cofactors but lack DNA repair activity. Biochemistry 34:6892-6899[Medline].

SCHMIDT, R., J. WEST, G. CNOPS, K. LOVE, and A. BALESTRAZZI et al., 1996  Detailed description of four YAC contigs representing 17 Mb of chromosome 4 of Arabidopsis thaliana ecotype Columbia. Plant J. 9:755-765[Medline].