Phytochromes are the primary red/far-red photoreceptors of higher plants, mediating numerous developmental processes throughout the life cycle, from germination to flowering. In seed plants, phytochromes are encoded by a small gene family with each member performing both distinct and redundant roles in mediating physiological responses to light cues. Studies in both eudicot and monocot species have defined a central role for phytochrome B in mediating responses to light in the control of several agronomically important traits, including plant height, transitions to flowering and axillary branch meristem development. Here we characterize Mutator-induced alleles of PhyB1 and a naturally occurring deletion allele of PhyB2 in Zea mays (maize). Using single and double mutants, we show that the highly similar PhyB1 and PhyB2 genes encode proteins with both overlapping and non-redundant functions that control seedling and mature plant traits. PHYB1 and PHYB2 regulate elongation of sheath and stem tissues of mature plants and contribute to the light-mediated regulation of PhyA and Cab gene transcripts. However, PHYB1 and not PHYB2 contributes significantly to the inhibition of mesocotyl elongation under red light, whereas PHYB2 and to a lesser extent PHYB1 mediate the photoperiod-dependent floral transition. This sub functionalization of PHYB activities in maize has probably occurred since the tetraploidization of maize, and may contribute to flowering time variation in modern-day varieties.
Phytochromes are a family of chromoproteins that display R/FR photoreversible changes in activity (Borthwick et al., 1952; Butler, 1962). Plant phytochromes are conjugates of a PHY apoprotein covalently attached to the linear tetrapyrrole chromophore 3E-phytochromobilin (PΦB; Lagarias and Rapoport, 1980; Terry et al., 1993). PΦB is synthesized from heme in chloroplasts and assembled with nuclear-encoded PHY apoproteins in the cytosol, forming the inactive, R-absorbing Pr conformation of phytochromes. Incident R converts the dimeric phytochromes from the inactive Pr to the active Pfr state. This reversible isomerization results in the import of Pfr into the nucleus, where it functions as a transcriptional regulator (Kircher et al., 1999; Yamaguchi et al., 1999).
In angiosperms, phytochromes exist as small gene families that can be further classified into four subfamilies: PHYA, PHYB/D, PHYC/F and PHYE (Mathews and Sharrock, 1996, 1997; Mathews et al., 1995). The monocot lineage has probably lost the PHYE subfamily that arose near the time of divergence of monocots and eudicots, and contains a single PHYB paralog (Mathews and Sharrock, 1997). Moreover, variation in phytochrome gene-copy number is observed within grasses. In rice and sorghum, phytochromes are single-copy, but in maize each phytochrome is duplicated (Basu et al., 2000; Childs et al., 1997; Sheehan et al., 2004). The duplication of Phy genes in maize is probably due to a allotetraploid event that occurred over 11 Mya in the maize lineage, resulting in large segmental duplications within the present-day maize genome (Gaut, 2001; Gaut and Doebley, 1997; Wilson et al., 1999). Because monocots and eudicots have been evolving independently since their split 140–200 million years ago (Mya), PhyA, PhyB and PhyC functions are not strictly conserved between these two diverse groups.
Here we describe the identification and characterization of phyB1 and phyB2 loss-of-function alleles of maize. In a previous study, we showed that the homologous PhyB1 and PhyB2 genes are predicted to encode functional proteins, and that both are expressed in several seedling tissues (Sheehan et al., 2004). To define the functions of each of these genes, Mutator transposon populations were screened to identify insertion alleles. Loss-of-function phyB1 alleles were then introduced into multiple inbred lines. During the course of these introgressions, a naturally occurring phyB2 deletion allele was discovered in the inbred line France 2 (F2), enabling the construction of single- and double-mutant populations. Here we show that PHYB1 and PHYB2 regulate several aspects of seedling development, including mesocotyl elongation and chloroplast gene expression. They also contribute to a number of adult morphological traits, including plant height, ear node height, culm width, leaf sheath and internode length. We further show that PHYB1 and PHYB2 have non-redundant roles in mediating both seedling and mature plant traits. In particular, PHYB1 plays a predominant role in regulating many seedling traits, whereas PHYB2 is necessary to repress flowering under non-permissive LD photoperiods.
Transposon-induced PhyB1 alleles
To identify transposon-induced PhyB alleles, a public collection of Mu insertion lines was screened (MTMdB; May et al., 2003). These screens identified one unique Mu-insertion allele of PhyB1 (phyB1-7793). Three additional PhyB1 insertions, phyB1-563, phyB1-562 and phyB1-567, were identified in the Pioneer Trait Utility System for Corn (TUSC) collection (Bensen et al., 1995) and kindly provided to us by Dr Michael Muszynski. Fine mapping of the three insertions revealed that all were located upstream of a conserved cysteine residue in the GAF domain of phytochrome (Marchler-Bauer et al., 2005). The Mu insertion in phyB1-7793 was present in multiple lines in the MTM population and mapped to the identical position as phyB1-563 (Figure 1). Thus, this allele was probably segregating in a Mu population that was common to both MTM and TUSC programmes. Alternatively, this insertion site defines a hotspot for Mu insertion (May et al., 2003).
To examine the phenotype of each Mu insertion allele, the three insertion alleles were introgressed into the W22 inbred line (see Experimental procedures). Previous studies have shown that the inhibition of mesocotyl elongation under R is a phytochrome-dependent response in maize (Vanderhoef et al., 1978, 1979). Thus lines homozygous for each allele were grown in darkness (D) or under continuous R, and mesocotyl lengths were measured after 10 days’ growth. All showed partial insensitivity to R when compared with wild-type siblings, suggesting that each insertion allele results in a loss of PHYB1 function (Figure S1). Because the phyB1-563 allele had been back-crossed the greatest number of generations into multiple genetic inbreds, it was used as a representative phyB1 loss-of-function allele for further phenotypic and molecular characterizations.
Mu populations of maize harbor a high copy number of Mu insertions and are from mixed genetic backgrounds, thus introgressions were performed to converge the phyB1-563 allele into multiple inbred lines. Plants were back-crossed for a minimum of three generations into the B73, Mo17, W22 and F2 inbred lines, and self-pollinated to create lines that were homozygous for the Mu insertion allele. Phenotypic comparisons were made with near-isogenic siblings that lacked the Mu insertion allele (see Experimental procedures). As shown in Figure 2, plants homozygous for the phyB1-563 insertion allele in the B73, Mo17 and W22 inbred backgrounds did not display any obvious seedling phenotypes when grown under glasshouse conditions. However, when introgressed into the F2 inbred the effect was striking: plants homozygous for the phyB1-563 allele displayed pale, narrow and elongated leaf blades and elongated internodes, and contained less anthocyanin in sheath tissues.
We had shown previously that both PhyB1 and PhyB2 are predicted to encode functional proteins and are expressed in several seedling tissues of the B73 inbred line (Sheehan et al., 2004). However, during the course of introgressions we observed a novel RFLP associated with the phyB2 locus of the F2 inbred (phyB2-F2) that was not detected in B73, Mo17 or W22 (data not shown). To examine the nature of this polymorphism in more detail, the genomic sequence of the PhyB2-F2 gene was determined. As shown in Figure 3, sequences near the translation start site were identical for inbreds B73, Mo17 and W22. However, the phyB2-F2 allele was polymorphic throughout the sequenced region and harbored a 10-bp deletion encompassing the predicted translational start site. This deletion is responsible for the loss of an NcoI site that was used in RFLP analysis. Thus, by converging the phyB1-563 allele into the F2 inbred a phyB1 phyB2 double mutant was created. To exploit this naturally occurring phyB2 allele, we created single- and double-mutant combinations of phyB1-563 and phyB2-F2 alleles in a predominantly F2 inbred background (see Experimental procedures). A wild-type PhyB2 allele was introduced from the Mutator parent of unknown ancestry and a wild-type PhyB1 allele from the F2 inbred. Four genotypic classes of plants were propagated for this study. For simplicity, we have used symbols for each genotype as follows: +/+ +/+ (functional PhyB1 and PhyB2 genes); b1/b1 +/+ (single phyB1-563 mutant); +/+ b2/b2 (single phyB2-F2 mutant); and b1/b1 b2/b2 (double mutant).
Although we assumed that the Mu insertion in the phyB1-563 allele resulted in a loss of PhyB1 function, alternative splicing using a splice donor or acceptor site within Mu may have resulted in the production of some functional PHYB1, as reported previously for a Mu-insertion allele of adh1 (Ortiz and Strommer, 1990). To characterize further each of the mutant alleles, Phy transcript accumulation profiles were examined in each of the four genotypic classes. As shown in Figure 4, PhyB transcripts of the expected size (approximately 4 kb) were detected in D- and white light (W)-grown samples of the wild-type and phyB2 single-mutant plants. However, in lines homozygous for the phyB1 mutant allele, transcripts were approximately 2 kb larger, indicating that transcription proceeded through the Mu3-like element inserted in the PhyB1 gene (Figure 4a,c).
Phenotypic characterization of phyB1-563 and phyB2-F2 alleles
To distinguish PhyB1 and PhyB2 transcript pools, we utilized a semi-quantitative RT-PCR assay that permits the discrimination of transcripts encoded by each of the maize phytochrome homologs (Sheehan et al., 2004). PCR amplification reactions were performed using primers that anneal to both PhyB1 and PhyB2 cDNA products. A restriction enzyme digestion was performed that cleaved the 304-bp PhyB1 products into 252- and 52-bp fragments. The 52-bp product was not resolved on the agarose gels. Restriction-site polymorphisms resulted in cleavage of the PhyB2-derived products into 155-, 92- and 52-bp fragments. Thus we were able to examine the relative contribution of PhyB1 and PhyB2 to the total PhyB transcript pool (see Experimental procedures). The assay revealed that PhyB1 transcripts accumulate to higher levels than PhyB2 transcripts in D- and W-grown tissues (Figure 4b,d). This result is consistent with a previous characterization of PhyB transcript accumulation in the B73 inbred (Sheehan et al., 2004). In lines homozygous for the phyB1 mutant allele, phyB1 transcripts accumulated to lower levels than the wild-type, perhaps due to a destabilization of the transcript caused by the Mu insertion. Taken together, these results strongly suggest that the Mu element is not spliced from the phyB1 transcript and thus phyB1-563 represents a non-functional allele.
As the 5′ deletion in the phyB2-F2 allele removed the canonical translational start site, translation may have initiated at the next AUG codon (nucleotide 822 of the PhyB2 sequence). Initiation at this site would delete most of the GAF domain, and probably render the protein non-functional. However, alternative splicing or the use of a non-canonical CUG codon (Christensen et al., 2005) as a translation start in the phyB2-F2 allele may lead to the accumulation of some functional PHYB2 protein. Thus it was important first to determine whether or not the phyB2-F2 allele was transcriptionally active. As shown in Figure 4, low levels of phyB2 transcripts were detected in W- and D-grown tissues of phyB1 phyB2 double mutants using RNA blot and RT-PCR assays. These results indicate that transcripts are generated from the phyB2-F2 allele and are similar in size to wild-type PhyB2 transcript pools. We detected no evidence that alternative or cryptic splice sites are utilized in the phyB2-F2 allele.
To define transcriptional start sites, 5′ RACE products were sequenced from plants carrying mutant and wild-type PhyB alleles (Figure S2). The majority of PhyB1 transcripts initiated at nucleotide positions −158 to −136 relative to the translation start site. The majority of transcripts in the phyB1 mutant initiated at −136 to −89 relative to the translation start site, suggesting that transcription initiates at similar positions in the phyB1 mutant and wild-type alleles. Furthermore, as phyB1 transcripts accumulate in lines homozygous for the phyB1-563 insertion and are detected by the RT-PCR assay that specifically amplifies sequences near the 3′ UTR (Sheehan et al., 2004), it is likely that the phyB1-Mu fusion transcript is polyadenylated and capable of directing translation.
Multiple transcription-initiation sites were also detected at the phyB2 locus. There were two predominant transcript pools encoded by both wild-type and mutant alleles. The first pool initiated at positions −148 to −131, and a second pool initiated at positions −19 to −16 relative to the translation start site (Figure S2). These findings strongly suggest that the deletion of the translational start site in the phyB2 mutant allele does not affect transcript initiation.
Searches of PhyB1 and PhyB2 upstream sequences failed to reveal consensus TATA-binding sites or Inr elements, typical of many eukaryotic promoters (Kadonaga, 2004), near the mapped transcription start sites. This finding is consistent with the mapping of multiple transcription start sites, as no consensus sequences appear to direct transcription initiation. Interestingly, although the majority of transcripts initiated at PhyB1 and PhyB2 result in a 5′ UTR of approximately 130–150 nt, a pool of transcripts initiated at PhyB2 carry much shorter leader regions (16–19 nt). As total seedling tissue was utilized in 5′ RACE, these discrete transcript pools may reflect tissue-specific differences in PhyB2 transcript initiation.
PHYB protein accumulation
To examine PHYB protein accumulation, D- and R-grown seedlings were examined by protein-blot analysis. Proteins were isolated from plants grown under R or D for 10 days and phytochrome-enriched protein extracts were selectively precipitated (see Experimental procedures). Blots were challenged either with an oat PHYA monoclonal antibody (PhyA1.9B5A; Daniels and Quail, 1984) or with a monoclonal antibody that was raised against the rice phyB protein (Rice 123; Wagner et al., 1991). The PhyA1.9B5A monoclonal antibody is specific to monocot PHYA, whereas the Rice 123 monoclonal recognizes both PHYA and PHYB proteins from maize, as shown in Figure 5. Antibody against a lid component of the 26S proteosome (RPN6) was used as a protein-loading control (Yang et al., 2005).
Although the predicted molecular weight of PHYA (125 kDa) is less than PHYB (128 kDa), PHYB proteins migrate more quickly through the gel, perhaps due to different post-translational modifications. PHYA protein accumulates in all genotypes of D-grown plants and was not detectable in R-grown plants. In contrast, PHYB protein accumulated to similar levels in both D- and R-grown plants. In the absence of a functional PhyB1 allele, PHYB protein was not detectable in D- and R-grown seedling tissues. These results indicate that most of the PHYB protein that accumulates in seedling tissues is encoded by PhyB1. However, we cannot exclude the possibility that low levels of PHYB2 protein accumulate in seedling tissues of phyB1 single mutants that are below the limit of detection of our protein-blot assay. Indeed, the single phyB1 mutant phenotype can be clearly differentiated from the double phyB1 phyB2 mutant in seedling tissues (Figure 2), strongly suggesting that PHYB2 protein accumulates in the phyB1 single mutant. The absence of detectable protein in the phyB1 phyB2 double mutant, together with the results from RNA-blot analysis, RT-PCR assays and 5′ RACE, discussed above, strongly suggest that both the phyB1-563 allele and the phyB2-F2 alleles are complete loss-of-function alleles.
PHYB is not essential for R-mediated inhibition of mesocotyl elongation
To examine the contribution of each PhyB gene to the repression of mesocotyl elongation, we grew wild-type, phyB1 single, phyB2 single and phyB1 phyB2 double mutants in D, W, R, FR or B. Mesocotyl lengths were measured after 10 days. As shown in Figure 6, some variation was observed among the genotypes of D-grown plants. This may be due to residual heterozygosity present in the genome of these lines (estimated at approximately 6.25% for the self-pollinated fourth-generation back-cross families). Nevertheless, the mesocotyl length of all genotypes under all light treatments were significantly different from D-grown seedlings. Under W or R, phyB1 single mutants were significantly longer than wild-type plants (P < 0.0001), whereas mesocotyl lengths of the phyB1 single and phyB1 phyB2 double mutants were similar (P > 0.05). These results suggest that PHYB2 cannot compensate for the loss of PHYB1 under W or R.
There was no significant difference in mesocotyl length between wild-type and phyB1 single, phyB2 single or phyB1 phyB2 double mutants under FR, suggesting that neither PHYB1 nor PHYB2 contributes significantly to mesocotyl length inhibition under FR. Mesocotyl elongation was significantly inhibited in phyB1 phyB2 double mutants under R relative to D. Mean mesocotyl length of the double mutants was approximately 30% of the length of D-grown seedlings. For comparison, the mesocotyl length of elm1, a maize mutant lacking spectrophotometrically active pools of phytochromes, is not significantly different between R- and D-grown plants (Sawers et al., 2002). This indicates that PHYA, and possibly PHYC, contribute significantly to the inhibition of mesocotyl elongation under R.
The expression of PHYA is regulated by PHYB
Previous studies in Arabidopsis have shown that PhyA transcript accumulation is dependent on the activity of phyB (Canton and Quail, 1999). To determine if a similar mechanism of control exists in maize, RNA-blot and semi-quantitative RT-PCR assays were used to monitor PhyA and PhyC transcript levels in the four genotypic classes. Plants were grown under D or W (8 μmol m−2 sec−1) and total RNA was extracted after 10 days’ growth. In D, PhyA1, PhyA2, PhyC1 and PhyC2 transcript pools accumulated to similar levels in each of the four genotypic classes (Figure 7a,b). Thus PhyA and PhyC expression is not altered in D by the absence of PHYB proteins. As shown previously, PhyA transcripts accumulate to low levels in light-grown maize seedlings (Christensen and Quail, 1989), and our results confirm this result in W-grown wild-type plants (Figure 7c,d). PhyA transcripts accumulate to higher levels in the light in phyB1 phyB2 double mutants relative to wild-type or single phyB1 or phyB2 mutant plants. Interestingly, PhyA2 transcript pools appear to show a stronger dependence on PHYB than PhyA1 in light-grown tissues (Figure 7d). As both 5′ and 3′ UTR sequences of PhyA1 and PhyA2 are highly similar (Sheehan et al., 2004), differential rates of transcription at PhyA1 and PhyA2 may drive the gene-specific accumulation profiles of PhyA transcript pools. In summary, neither PhyA nor PhyC transcript accumulation is dependent on PHYB in D-grown tissues. PHYB negatively regulates transcript accumulation of PhyA2, and to a lesser extent PhyA1, in light-grown tissues. However, PHYB does not appear to regulate PhyC transcript accumulation.
PHYB1 and PHYB2 regulate accumulation of Cab gene transcripts
To examine PHYB control of transcript accumulation further, a light-shift experiment was conducted. RNA was extracted from plants that were grown in D for 10 days and from plants shifted to R for 6 h on day 10 (R-shift). RNA pools were examined by RNA-blot analysis and the levels of PhyA and Cab determined for each genotype using densiometry (see Experimental procedures). The levels of PhyA and Cab relative to a Ubiquitin control were calculated for each genotype. The ratios of gene expression in R-shift plants relative to D controls for each genotype were compared to account for genotype-specific differences in gene expression in D. For example, PhyA transcript levels were approximately fivefold lower in wild-type plants after 6 h R relative to D controls (Figure 8a; wild-type, R/D = 0.19). PhyA transcript pools were also reduced, but to a lesser extent in the phyB1 single mutant, whereas PhyA transcript levels accumulated to a level similar to the wild type in the phyB2 mutant (phyB1 mutant, R/D = 0.29; phyB2 mutant, R/D = 0.14). However, in the double mutant, PhyA transcripts pools were reduced to approximately half the level of D controls (R/D = 0.52). Taken together, these findings suggest that PHYB1 and PHYB2 act redundantly to negatively regulate PhyA transcript accumulation in the light. Furthermore PHYA, and possibly PHYC, also negatively regulate PhyA transcript accumulation. As shown in Figure 2, the loss of functional PHYB protein clearly affects the photosynthetic development of maize seedlings. Thus we also examined the expression of Cab, encoding Chl a/b binding protein (LHCPII), in each of the four genotypic classes. As shown in Figure 8(b), Cab transcript pools increased in leaf tissues of R-shifted wild-type, phyB1 single, phyB2 single, and phyB1 phyB2 double mutants relative to D controls (2.7-, 4.7-, 3.1- and 2.1-fold, respectively). The increase in Cab transcript pools in double mutants suggests that PHYA, and possibly PHYC, mediate changes in Cab transcription under R light.
PHYB mediates morphological development of field-grown plants
To examine the role of PHYB in the control of adult plant traits, 292 individuals were genotyped in segregating families to identify 15 wild-type, 15 phyB1 single-, 10 phyB2 single- and 15 phyB1 phyB2 double-mutant individuals, as described in Experimental procedures. Traits measured included the distance of the primary (topmost) ear from the ground (ear node height), the full height of the plant (full height), diameter of the culm (mean stem diameter), leaf sheath and internode length between the ear node and the node above, and distance between the flag leaf and lowest branch of the tassel (tassel stem length). As shown in Figure 9, there was no significant difference between wild-type, phyB1 and phyB2 single and double mutants for tassel stem length and internode length (Figure 9e,f). However, ear node height, full height and mean stem diameter and the leaf sheath internode length difference were significantly reduced in the double-mutant plants relative to the other three genotypic classes (Figure 9a–d). In wild-type and phyB single mutants, leaf sheath tissues generally extended above the nodes. However, in the phyB1 phyB2 double mutants, the nodes of the culm were often exposed. Although we did not detect significant differences in internode length among genotypes, the mean internode length was greatest in the double mutant and the sheath length was shortest in the double mutant. Thus the difference between leaf sheath and internode length in the double mutant varied significantly from the other three genotypes (Figure 9d). These results suggest that PHYB1 and PHYB2 act redundantly in the control of a number of agronomically important traits. In particular, reduced plant height and lower ear node placement are correlated with early maturity, and reduced stem diameter is correlated to an increase in stem lodging.
PHYB contributes to flowering time variation
Although generally considered day-neutral, maize is a domesticated form of the SD plant teosinte (Zea mays ssp. parviglumis) and, as such, still retains some photoperiod sensitivity (Thornsberry et al., 2001). To examine the contribution of PHYB1 and PHYB2 to the control of flowering time, segregating populations were grown in the field under both LD and SD conditions, and flowering time measured as days to anthesis after sowing (DAS). During summer 2003 and winter 2004, male flowering-time data were collected from a population homozygous for the phyB2-F2 allele but segregating phyB1-563. These plants were genotyped to identify phyB2 single-mutant and phyB1 phyB2 double-mutant plants, and flowering time was measured (Figure 10a,b). anova tests of these data showed that the effect of photoperiod, genotype, and photoperiod × genotype were all highly significant (P < 0.0001; Table S1). Mean flowering time among plants that carried both the phyB1-563 and the phyB2-F2 alleles was significantly earlier under LD relative to phyB2 single mutants (P < 0.0001). In the winter nursery (SD), PHYB1 also played a small but significant role in repressing flowering time (P = 0.0047). These data indicate that PHYB represses flowering under LD and has a small effect on inhibiting flowering under SD photoperiods.
To determine the functions of PHYB1 and PHYB2 in the control of flowering time variation, days to anthesis was measured in each of the 292 individuals grown in our summer 2004 nursery (LD). Double mutants flowered, on average, 10 days earlier than lines that contained functional copies of PHYB1 and PHYB2. Interestingly, when a functional allele of PhyB2 was introduced into the F2 inbred line, plants flowered significantly later than lines carrying the phyB2-F2 deletion allele (P < 0.0001). This finding strongly suggests that the absence of a functional PhyB2 allele in the F2 inbred contributes to the precocious flowering of this inbred line.
To explore further the contribution of PhyB to flowering time variation, we examined PhyB-expression levels using the RT-PCR assay in a number of inbred lines (Figure S3). These inbred lines represent a diverse germplasm collection with representatives from each of the three major heterotic groups of maize (Markelz et al., 2003). Although PhyB2 expression was detected in all inbreds examined, the phyB2 deletion allele was detected only in early flowering varieties (marked with asterisks). Furthermore, in screens of approximately 250 diverse inbreds, we were unable to detect the phyB2 deletion allele in any tropical or semi-tropical lines. These results suggest that PhyB2 variation may contribute significantly to flowering time variation in maize.
To obtain transposon-induced lesions in the PhyB genes, we performed reverse-genetic screens of large Mutator populations. Several Mutator insertions were identified and fine-mapped to the GAF domain of PhyB1, representing three independent alleles, phyB1-562, phyB1-563 and phyB1-567. Each allele conditioned a similar phenotype in seedling assays, thus we utilized phyB1-563 (phyB1) in the majority of our phenotypic analysis. Although there have been reports of Mu elements being alternatively spliced from transcripts (Ortiz and Strommer, 1990), RNA gel-blot analyses indicated that transcription proceeds through the Mu3-like insertion in phyB1-563 and is retained in the mature transcript, as evidenced by an approximately 2 kb shift in the predicted PhyB1 transcript size. Furthermore, we did not detect any PHYB protein in phyB1-563 mutant seedlings using a monoclonal PHYB antibody, suggesting that little functional PHYB1 protein accumulates in lines homozygous for the phyB1-563 allele.
A putative phyB2 deletion allele was identified during the course of converging the phyB1-563 mutation into the early flowering northern flint line F2. Sequence analysis of the 5′ UTR and coding regions of the PhyB2-F2 allele suggests that the loss of the initiating codon (Kozak, 1986) prevents accumulation of PHYB2 protein. Importantly, our comparison of 5′ RACE products derived from the phyB2-F2 and PhyB2-B73 alleles indicates that similar transcription initiation sites are used in both alleles. Finally, protein-blot analysis failed to detect any PHYB protein in the phyB1 phyB2 double mutant. These results demonstrate that both phyB1-563 and phyB2-F2 are complete loss-of-function alleles.
Phytochrome control of seedling de-etiolation
Previous studies of seedling de-etiolation have defined a central role for PHYB in mediating R-inhibition of hypocotyl elongation in tomato (Weller et al., 2000) and Arabidopsis (Neff and Chory, 1998). However, elongation of seedling tissues is significantly inhibited under R in phyB mutants of pea (Weller et al., 1995), sorghum (Childs et al., 1995) and rice (Takano et al., 2005), suggesting that, in these species, additional phytochromes play a role in R-mediated growth inhibition. Our analysis of mesocotyl elongation in phyB1 phyB2 double mutants under R indicates that PHYA, and perhaps PHYC, contribute to this response. However, in the absence of a functional PHYB gene, phyC pools are significantly reduced in Arabidopsis and rice (Hirschfeld et al., 1998; Takano et al., 2005). Thus, in the maize phyB1 phyB2 double mutants, it is likely that PHYA mediates the inhibition of mesocotyl elongation under R light.
Given that PHYA pools rapidly decline when plants are exposed to light (Vierstra, 1994), a role for PHYA in mesocotyl growth inhibition under continuous R is not inherently obvious. As we have shown, no PHYA protein is detectable in R-grown seedling tissues in wild-type or phyB1 phyB2 mutant plants, thus any response mediated by PHYA must require relatively low levels of PHYA pools. A role for PHYA in mesocotyl or hypocotyl growth inhibition in response to R is often observed only in the absence of PHYB (Mazzella et al., 1997; Reed et al., 1994; Takano et al., 2005; van Tuinen et al., 1995a), suggesting an antagonism between the actions of PHYA and PHYB. Interestingly, PhyA transcript levels in light-grown plants are elevated in the phyB1 phyB2 double mutants relative to wild-type controls, suggesting that PhyA protein levels may also be slightly elevated in light, despite our inability to detect PHYA in immunoblot analysis. This PHYB-dependent regulation of PhyA transcript accumulation, observed here and in Arabidopsis (Canton and Quail, 1999; Somers and Quail, 1995), could account for some of the observed antagonism between PHYA and PHYB action in higher plants.
Our measurements of mesocotyl length revealed little difference between single phyB1 and double phyB1 phyB2 mutants. The simplest interpretation of these data is that maize PHYB2 cannot compensate for a loss of PHYB1. However, as we did not examine the kinetics of R-mediated inhibition of mesocotyl elongation, it is possible that there is a developmental window where PHYB2 action is required for growth inhibition. In rice, for instance, the inhibition of coleoptile elongation under R is biphasic, where phyB is required only during the first stage of seedling growth (Takano et al., 2005). In tomato, PHYB1 mutants are temporarily insensitive to R for the first 2 days of seedling growth, but recover to resemble wild-type plants in adulthood, possibly through the action of PHYB2 (van Tuinen et al., 1995a). Despite these caveats, the relatively low levels of PhyB2 expression in seedling tissues, and our failure to detect PHYB protein in the phyB1 single mutants, strongly suggest that PHYB2 contributes little to seedling responses under R.
Control of gene expression by PHYB1 and PHYB2
Although there was no PHYB-dependent regulation of either PhyC1 or PhyC2 transcripts in plants grown in either D or W, PHYB was required for the reduced accumulation of PhyA transcript pools in W and following an R-shift. The accumulation of PhyA transcript pools was greatest in the phyB1 phyB2 double mutant, suggesting that both PHYB1 and PHYB2 contribute to either the downregulation of PhyA transcription or an increased rate of PhyA transcript degradation. As a result, PhyA transcripts in the phyB1 phyB2 double mutant were reduced to approximately half the levels found in D-grown controls. This suggests that additional phytochromes are required to reduce PhyA transcript pools in the light.
Interestingly, PhyA2 transcripts accumulated to higher levels in double-mutant plants relative to either the single mutants or wild-type plants in W, while PhyA1 transcript levels were independent of PHYB activity. Sequence analysis of the PhyA1 and PhyA2 genes indicates a strong conservation of promoter and UTR sequences, and both genes display similar tissue-specific and light-regulated patterns of expression (Sheehan et al., 2004). However, there is extensive sequence divergence between the first intron of PhyA1 and PhyA2 (Sheehan et al., 2004). Thus sequences present in the first intron of PhyA2 that are absent in PhyA1 may mediate transcriptional regulation by PHYB. In Arabidopsis, both phyA and phyB appear to regulate the transcription of PHYA (Canton and Quail, 1999), and it is likely that PHYA, and possibly PHYC, contribute to the regulation of PhyA transcript accumulation in maize. In summary, PHYB1, PHYB2 and an additional PHYs negatively regulate the accumulation of PhyA transcripts in R light. Furthermore, these PHYs act selectively to negatively regulate the accumulation of PhyA2 transcript pools in W-grown seedling tissues of maize.
PHYB1 and PHYB2 act redundantly in the control of several agronomic traits
PHYB has been shown to regulate several traits of agronomic value in the grasses, including seed germination, shade avoidance, axillary stem (tiller) development, and time to flowering (Childs et al., 1997; Finlayson et al., 1999; Hanumappa et al., 1999; Izawa et al., 2000). Thus we examined the role of PHYB on several morphological characters in a segregating population in the field. Four of the six traits measured (ear node height, full plant height, mean stem (culm) diameter, and leaf sheath minus internode length difference) showed complete functional redundancy for PHYB1 and PHYB2. The reduced plant height, earlier emergence of the ear shoot and acceleration of flowering are consistent with the notion that phyB1 phyB2 mutants make the transition from the vegetative to reproductive phase of development more rapidly. The narrowing of the stem diameter in phyB1 phyB2 double mutants is indicative of a constitutive shade-avoidance response (Smith, 1995), and suggests a central role for PHYB in mediating responses to vegetative shade in maize. In rice, elongation of the coleoptile in a phyB mutant is largely due to increased cell expansion, rather than an increase in cell number (Takano et al., 2005), suggesting a primary role for phyB in mediating growth through changes in cell wall extensibility, rather than through changes in rates of cell division. It is interesting to note that sheath tissues of double mutants were shorter, whereas culm tissues were longer in the phyB1 phyB2 double mutants relative to wild-type plants. It is likely that PHYB is acting to inhibit elongation growth in both culm and sheath tissues. However, it is possible that PHYA accumulates to higher levels in sheath tissues than in stem tissues, and acts antagonistically to PHYB. Thus, in the absence of PHYB, PHYA mediates an inhibition of sheath elongation in an FR-rich canopy, whereas its absence from stem tissues allows elongation growth. Future examination of PhyA and PhyB transcript accumulation in our four genotypic classes of plants in culm and sheath tissues will address these interactions.
PHYB2 regulates flowering time in maize
The manipulation of flowering time has been under intense selection pressure in crop-breeding programmes (Lin et al., 1995; Paterson, 1995; Paterson et al., 1995; Sawers et al., 2005). Furthermore, previous studies in barley, rice, sorghum and maize have revealed that lesions in phytochrome signaling result in precocious flowering (Childs et al., 1995, 1997; Hanumappa et al., 1999; Izawa et al., 2000; Sawers et al., 2002). Here we have shown that a loss of PHYB function results in early flowering under both LD (summer) and SD (winter) photoperiod conditions. These results are consistent with previous studies indicating that PHYB represses flowering time in both SD and LD (Childs et al., 1995; Halliday et al., 1994; Takano et al., 2005). Interestingly, a loss of PHYB2 resulted in a statistically significant acceleration of flowering relative to the phyB1 single-mutant and wild-type plants. The F2 inbred line is an early flowering northern flint variety carrying a naturally occurring phyB2 deletion allele, and by introgression of a functional copy of PhyB2, we were able to delay flowering by up to 1 week under an LD photoperiod. Recent surveys of a diverse maize germplasm collection have revealed that many early flowering flint varieties also carry a similar PhyB2 deletion allele (Figure S3). Thus PHYB2 may play a predominant role in regulating flowering time in response to photoperiod in maize.
The relatively recent duplication of maize Phy gene paralogs, approximately 11–16 Mya (Sheehan et al., 2004), has provided a unique opportunity to explore the consequences of polyploidization in a major crop plant. Here we have shown that, although PHYB1 and PHYB2 act redundantly in the control of a number of mature plant traits, there are a number of traits in which one of the two PhyB genes exerts greater control. This subfunctionalization of PHYB activities is particularly apparent in seedling tissues where the loss of PHYB2 function has no significant effect on mesocotyl elongation under R. However, the loss of PhyB1 function results in an increase in mesocotyl elongation under W and R. The loss of PhyB2 function results in a significant acceleration of flowering, whereas flowering time is not significantly different between phyB1 mutants and wild-type plants. A partitioning of PHYB activities between relatively recent PhyB duplications is also evident in tomato and Nicotiana plumbaginifolia (tobacco). In tomato, phyB1 plays a predominant role in early seedling development, mediating the accumulation of anthocyanins, chlorophyll, cotyledon mass and hypocotyl length. However, phyB1 and phyB2 have largely redundant functions in mature plant tissues (van Tuinen et al., 1995a; Weller et al., 2000). In tobacco, the hlg mutant displays elongated hypocotyls under R, and a slight delay in flowering time under LD growth conditions, but displays a shade-avoidance response similar to wild-type plants (Hudson and Smith, 1998; Hudson et al., 1997). These findings suggest that a duplicated copy of phyB in N. plumbaginifolia may mediate responses to R in mature plants. Thus in maize, tomato and possibly tobacco, duplicated PhyB genes show diversification of function. In maize, both PHYB1 and PHYB2 perform unique functions in the control of both seedling and mature plant traits. Higher-order mutant combinations with loss-of-function alleles of PhyA and PhyC should provide additional insights into the diversification of the phytochrome gene family in this important cereal crop.
Plant materials and mutant alleles
Inbred lines of Zea mays L. (maize) used include B37, B73, B97, CM37, CML333, CMV3, EP1, F2, H99, IA2132, IDS28, IL101, KUI2007, KUI21, Mo17, N192, N28Ht, NC260, NC348, OH43, P39, PA91 and W22. Screens of MTMdB (Cold Spring Harbor Laboratory, NY, USA) were performed with primers TB27 (5′ TTCCGCCTCCTCGCCTTCTC 3′) and TB28 (5′ CCGTGGAGGCGGGAGATGGCG 3′) resulting in three unique insertions. One insertion from MTMdB (identified twice as 7784 and 7793) was mapped to the identical position to the phyB1-563 allele from TUSC. Two additional alleles from TUSC (562 and 567) were also mapped to unique positions. The 563, 562 and 567 insertions map to positions 2591, 2058 and 2541, respectively, relative to GenBank accession AY234827. Introgression of each phyB1 allele was monitored through each round of genetic back-crossing by DNA-blot analysis. Stock lines of the phyB1-563 allele were back-crossed into B73 (three generations), Mo17 (four generations), F2 (four generations) or W22 (five generations). Additionally, phyB1-562 and phyB1-567 were back-crossed into W22 for three generations prior to self-pollination, to generate homozygous materials.
To generate the segregating population used for seedling and morphological experiments, an individual of the genotype PhyB1 phyB-F2 was crossed by a phyB1-563 PhyB2 plant. The progeny of this cross was self-pollinated to generate a segregating population. A population of 400 seeds was planted in the summer 2004 nursery (Aurora, NY, USA). The surviving 292 individuals were genotyped by DNA-blot analysis using the PhyB-CAR probe as described previously (Sheehan et al., 2004). A χ2 test for goodness of fit established that the observed genotypic ratios were within the expected range.
Growth conditions and light sources
For all seedling measurements, plants were grown in soil. Percival E-30 LED chambers (Percival Scientific, Boone, IA, USA) were used to provide monochromatic light. Plants were grown for 10 days at 28°C under B (2.5), R (3.0), FR (3.7) or W (8.5 or 100 μmol m−2 sec−1) light prior to measurements or extractions. The W source was a combination of incandescent and cool white fluorescent lamps. Whole seedlings were collected and two to three plants were pooled for RNA and protein extractions.
RNA extraction and RNA gel-blot analyses
RNA extractions and semi-quantitative RT-PCR were performed as described previously (Sheehan et al., 2004). To distinguish each homolog, a restriction enzyme digestion was performed to exploit restriction-site polymorphisms between each pair of homologs. Following cleavage of the 354-bp PhyA1 PCR products with MaeIII, 215-, 98- and 50-bp fragments are generated, whereas PhyA2 is cleaved into 304- and 50-bp fragments. Digestion of the 304-bp PhyB1 PCR products with AciI results in 252- and 52-bp products, whereas PhyB2 products were cleaved into 155-, 92- and 52-bp fragments. Finally, the 444-bp PhyC1 PCR products digested with AvaII generates 281- and 163-bp fragments, and PhyC2 is cleaved into 333- and 111-bp fragments. Products smaller than 52 bp were not resolved on 2% agarose gels. RNA-blot analysis was performed using approximately 20 μg (Phy blots) or 5 μg (Cab blots) total RNA. The Cab probe was an α−32P-labeled fragment generated from LHCP1020 (Roth et al., 1996), and phytochrome-specific fragments were as described previously (Sheehan et al., 2004). Densiometric analyses were performed using the 1D Array function of imagequant TL ver. 2005 (Amersham Bioscience Corp., Piscataway, NJ, USA) according to the manufacturer's specification. The relative expression was calculated as the densiometry value of the gene of interest normalized to Ubiquitin relative to the normalized value of the gene of interest in D.
5′ rapid amplification of cDNA ends (5′ RACE)
Polyadenylated mRNA was isolated from approximately 250 μg of total RNA from multiple extractions using the Oligotex mRNA Spin-Column kit (Qiagen, Valencia, CA, USA) and 5′ RACE performed using the GeneRacer kit (Invitrogen, Carlsbad, CA, USA) according to manufacturer's recommendation. First-strand cDNA synthesis was performed using SuperScript III (Invitrogen) with either N6 random primers supplied by manufacturer, or with a gene-specific primer (GSP) cocktail of two PhyB primers. The random primers were used at a final concentration of 2.7 μm, and each GSP in the cocktail was at a final concentration of 0.25 μm. The GSPs used for the PhyB1-F2 allele were 17061R (5′-GTGGCATGGCAATCGGCAATCATTCGC-3′) and 12417R (5′-GGAGTGACACCCAATGGGTAGTATTTC-3′). The GSPs used for the phyB1-563 allele were 17161R (5′-GGCTCAAGGTTGTCGCGCCGGCTCTCG-3′) and 17061R. The GSPs used for PhyB2-WT and phyB2-F2 alleles were 10915R (5′-GGATTGTGTACAATAGCTTTGTAGTC-3′) and 12417R.
Amplification reactions were conducted using nested PCR. Reactions contained 0.5 U Platinum Taq High Fidelity (Invitrogen), 1X Platinum Taq buffer, 500 m m Betaine, 2 mm magnesium sulfate, 4% (v/v) DMSO, 150 μm dNTPs and 0.3 μm primer. Five μl fivefold-diluted cDNA template was added to first-round amplification (20 μl reaction volume) and 5 μl 1000-fold diluted first-round PCR template was added to second-round PCR. In the first round, a touchdown PCR protocol was used with the following conditions: 94°C for 2 min, 10 cycles of 94°C for 15 sec, 65°C for 1 min, 68°C for 3 min, 25 cycles of 94°C for 15 sec, 58°C for 1 min, 68°C for 3 min and a 10-min extension at 68°C. For the second-round PCR, parameters were as follows: 94°C for 2 min, 35 cycles at 94°C for 15 sec, 65°C for 1 min, 68°C for 3 min with a final extension of 10 min at 68°C.
Several PhyB1- and PhyB2-specific primers were used in amplification reactions. Primers used in the first round of amplification included: 19340R (5′-GGGAAACGTGGGGCGGCGCAGAGGAGTC-3′), 19568R (5′-GCCCGATTGCTCGAACACCGCGTGTAG-3′), B1_5′RACE_261R (5′-TGTAGGCGCGCGTCTAGGGTGTACTGG-3′), or B1_5′′RACE_117R (5′-ACTCCGTGGCCGCGGCTCCCCCGCCCG-3′). To amplify PhyB2 sequences primers 19356R (5′-AGTCGAGCGAGGGAACGGAGTGG-3′), B2_5′RACE_119R (5′-CGCAGCGCCCCCGCCACCTCCTC-3′), B2_5′RACE_286R (5′-TGAGAACGCCGCCGACCTGCTCGACCTG-3′), or B2_5′RACE_392R (5′-CAGGTCGAGCAGGTCGGCGGCGTTCTCA-3′) were used. For the second round of amplification primers 19356R, B1_5′RACE_261R, B1_5′RACE_117R, 19706R (5′-GCGCGGGACGTGCTCCCGCCCGACGAC-3′), or B1_5′RACE_15R (5′-GGAGGAGGGGGAGCGCGTGGGCGTGGC-3′) were used to amplify PhyB1 sequences and primers 19356R, B2_5′RACE_286R, B2_5′RACE_119R, 19706R, or F2startShortR (5′-GGAAGCGACTGTCAGACGCTGGGGAGC-3′) were used to amplify PhyB2 sequences. Multiple PCR products were cloned and sequenced from each nested PCR reaction. We defined major transcription initiation sites as those where two or more independent clones had identical sequence start sites (Figure S2, red arrows). Minor transcription initiation sites were defined by the recovery of a single clone with a unique start sequence (Figure S2, black arrows). All sites identified are numbered relative to the predicted wild-type PhyB1 and PhyB2 translation start sites in Figure S2. All sequence data were compiled into four allele-specific contigs with annotated transcription start sites and submitted to GenBank under the following accession numbers: PhyB1-F2 (DQ307579), phyB1-563 (DQ307575), PhyB2-WT (DQ307618) and phyB2-F2 (DQ307620).
Approximately 5.0 g 10-day-old seedling tissue was frozen and ground in liquid nitrogen. The frozen powder was mixed in 2/3X HB (1X = 50% ethylene glycol (v/v), 100 mm Tris–HCl pH 8.3, 140 mm ammonium sulfate (w/v), 10 mm EDTA, 0.3 g per 50 ml sodium metabisulfite) with the addition of 1X complete protease inhibitor (Roche Applied Science, Indianapolis, IN, USA). The extracts were filtered through three layers of cheesecloth (Fisher Scientific, Pittsburgh, PA, USA) and two layers of Mira cloth (Calbiochem Corp., San Diego, CA, USA) prior to centrifugation at 28 000 g for 15 min. Polyethyleneimine (0.5% v/v final) was added to the supernatant, and samples were incubated on ice for 5 min then centrifuged at 28 000 g for 15 min. Phytochromes were precipitated by the addition of 0.26 g ml−1 ammonium sulfate and collected by centrifugation at 28 000 g for 25 min. Pellets were resuspended in 0.66X HB and 1X protease inhibitor and stored at −20°C.
Protein gel-blot analyses
Polyacrylamide gels were loaded by mass of starting tissue and equal loading confirmed by GelCode Blue staining (Pierce Biotechnology, Rockford, IL, USA). To generate two replicate blots, protein extracts were mixed with an equal volume of 2X SDS loading buffer, boiled for 10 min and fractionated by SDS–PAGE (7.5 % w/v acrylamide gel). Proteins were transferred to nitrocellulose membrane (Schleicher & Schull, Keene, NH, USA) by electroblotting. PHYB protein was detected on one blot using the monoclonal antibody Rice 123 at a 1:100 dilution (Wagner et al., 1991). PHYA protein was detected on the replicate blot with the PhyA1.9B5A antibody at a 1:1000 dilution. Aliquots of both antibodies were kindly provided by Dr Peter Quail (Plant Gene Expression Center, Albany, CA, USA). Visualization of the proteins was performed with a secondary goat anti-mouse antibody conjugated to horseradish peroxidase and the Supersignal West Femto Maximum Sensitivity Substrate kit (Pierce Biotechnology) according to the manufacturer's recommendations. Molecular masses (kDa) were determined using pre-stained chemiluminescent Blue Ranger marker (Pierce Biotechnology). The Kodak ImageStation 440CF (Eastman Kodak, Rochester, NY, USA) and Kodak id 3.5 imaging software (Scientific Imaging Systems, New Haven, CT, USA) were used to capture and view images.
To control for variations in protein loading, blots were re-challenged with Arabidopsis polyclonal RPN6 antibodies at a 1:1000 dilution (kindly provided by H. Wang, Boyce Thompson Institute, Ithaca, NY, USA) to detect a lid component of the 26S proteosome (Yang et al., 2005). Visualization of signal was performed with 1:1000-diluted secondary goat anti-rabbit antibody conjugated to horseradish peroxidase followed by detection with the Opti-CN4 kit (Biorad, Hercules, CA, USA) according to the manufacturer's specification. Blots were scanned at 600 dpi on a flat-bed scanner.
Quantification of morphological characters and flowering time
Traits selected for quantification were ear height, length of tassel stem, full height, length of leaf sheath (LS) above ear, length of internode (IL) above ear, difference in length between sheath and internode (equal to LS–IL), and two measurements of stalk diameter (SD) at 90° angles to each other. Fifteen wild-type, 15 phyB1 single-, 10 phyB2 single- and 15 phyB1 phyB2 double-mutant individuals were selected randomly from 292 segregating individuals for morphological characterization.
Flowering time of field-grown plants was determined over two summer seasons and one winter season. Male flowering was determined when 50% of the main tassel spike was actively shedding pollen. The time from sowing to 50% male flowering was termed days to anthesis after sowing (DAS). Both summer 2003 and 2004 trials were conducted in Aurora, NY, USA. The winter 2004 trial was carried out in Kaunakakai, Hawaii. Morphological characters were quantified in the 2004 summer nursery. All trials were performed under low planting densities of 30 inches (76.2 cm) between rows and approximately 8 inches (20.3 cm) between plants. In summer 2003 and winter 2004, flowering time was determined for all plants grown in a population segregating phyB1-563 and homozygous for the phyB2-F2 allele. For summer 2003 studies, phyB1 alleles were determined from a random sampling of early, mid- and late-flowering individuals (36 total). Winter 2004 flowering data were tabulated on 81 phyB2 single mutants and 109 double mutants. The DAS was tabulated on the entire population of summer 2004 individuals (292 plants).
The program PROC GLM in sas (sas/stat Software ver. 9.1 for windows, SAS Institute Inc., Cary, NC, USA) was used to analyze the mesocotyl length data, adult plant morphological data and flowering time data. In all analyses genotype is a fixed effect, as are light environment (mesocotyl experiment) and season (field data on flowering time in SD and LD environments). Mesocotyl lengths were natural log-transformed to more closely approximate normality. Tests of differences among genotype × light environment effects on mesocotyl lengths were carried out using least-squares means with the P-values adjusted for multiple comparisons using the Tukey–Kramer method. For morphological traits, differences among genotypes were examined using Tukey's Studentized range (HSD) tests. anova tables for all experiments analyzed are provided in Table S1.
We would like to thank Dr Mike Muszynski for providing PhyB1 insertion alleles and for helpful discussions. We would like to thank Dr Haiyang Wang for the RPN6 antibody and Dr Peter Quail for the PhyA and PhyB antibodies. We would also like to thank Dr Sarah Matthews for helpful discussions and Patrice Dubois for critical reading of the manuscript. This work was funded by a National Science Foundation grant IBN-0110297 to T.P.B. Funding for M.J.S. was provided by a Plant Cell and Molecular Biology training grant (Cornell University).