The Pea DELLA proteins LA and CRY are important regulators of gibberellin synthesis and root growth.

The theory that bioactive gibberellins (GAs) act as inhibitors of inhibitors of plant growth was based originally on the slender pea (Pisum sativum) mutant (genotype la cry-s), but the molecular nature of this mutant has remained obscure. Here we show that the genes LA and CRY encode DELLA proteins, previously characterized in other species (Arabidopsis [Arabidopsis thaliana] and several grasses) as repressors of growth, which are destabilized by GAs. Mutations la and cry-s encode nonfunctional proteins, accounting for the fact that la cry-s plants are extremely elongated, or slender. We use the la and cry-s mutations to show that in roots, DELLA proteins effectively promote the expression of GA synthesis genes, as well as inhibit elongation. We show also that one of the DELLA-regulated genes is a second member of the pea GA 3-oxidase family, and that this gene appears to play a major role in pea roots.

It is well known that primary root growth is strongly influenced by the plant hormone GA (Davies, 2004). For example, the application of bioactive GA to roots treated with the growth inhibitor ancymidol completely restored growth to that of the untreated plants (Tanimoto, 1991). Yaxley et al. (2001) established the importance of GAs for root growth in peas (Pisum sativum) by using a variety of GA-deficient mutant plants. In the na-1 mutant, for example, root GA 1 levels, and root elongation, were significantly reduced compared with wild-type plants, and when the GA 1 content was restored to wild-type levels, so too was root elongation.
The GAs act by destabilizing the growth inhibitory DELLA proteins (Peng et al., 1997;Harberd et al., 1998;Silverstone et al., 2001;Alvey and Harberd, 2005). In other words, GA acts as an ''inhibitor of an inhibitor'' (Harberd et al., 1998). Interestingly, there is also evidence that DELLA proteins promote the biosynthesis of active GAs. For example, in the Arabidopsis (Arabidopsis thaliana) DELLA mutant rga, the expression of the biosynthesis gene GA4 is reduced, indicating that high DELLA protein levels are associated with an up-regulation of GA synthesis genes (Silverstone et al., 2001). More recently Zentella et al. (2007) provided evidence that the Arabidopsis GA synthesis genes GA3ox1 (GA4) and GA20ox2 are direct DELLA targets.
DELLA proteins display conserved amino acid sequences among both dicot (Arabidopsis) and monocot (rice [Oryza sativa], wheat [Triticum aestivum], and barley [Hordeum vulgare]) species (Silverstone et al., 1998;Gubler et al., 2002). However, the available evidence indicates greater redundancy in dicots compared with monocots (Ikeda et al., 2001;Thomas and Hedden, 2006). There have been five DELLA genes isolated from Arabidopsis (GAI, RGA, RGL1, RGL2, and RGL3), yet only one in rice (SLR1), barley (SLN1), and maize (Zea mays; D8; Peng et al., 1997Peng et al., , 1999Silverstone et al., 1998;Ikeda et al., 2001;Chandler et al., 2002;Gubler et al., 2002), with the possibility of another DELLA gene in maize (D9; accession no. ABI84225). It should be noted, however, that DELLAs have been studied in fewer dicot model species than in monocot species. To date, DELLA-encoding genes from pea have not been reported, even though observations on the slender phenotype of pea triggered the early suggestion that GA acts an inhibitor of an inhibitor (Brian, 1957). The elongated slender phenotype, conferred by the gene combination la cry-s, has long been considered to show constitutive GA signaling (Potts et al., 1985), but the exact nature and function of the LA and CRY genes have not been reported before now.
In this investigation we identify DELLA-encoding genes and their associated mutants from pea. We then use those mutants to show that in roots DELLA proteins promote the expression of GA synthesis genes and inhibit the expression of GA deactivation genes. We also report on the discovery of a previously unidentified GA 3-oxidase gene ( Fig. 1) from pea.

LA and CRY Encode DELLA Proteins
Although there has been much work on the involvement of DELLA proteins in shoots, there has been less emphasis on roots. We sought to clone pea DELLA genes to study their involvement with GAs in the regulation of root growth. Two partial sequences from PCR (see ''Materials and Methods'') were used to probe a pea-shoot complementary DNA (cDNA) library. Full-length clones, selected on the basis of a conserved amino-terminus, were obtained in both cases, and sequence analysis showed that they both encode DELLA-like proteins. Sequencing of PCR products from genomic DNA from mutant genotypes showed that the base sequence of one of the two clones was altered in cry-s and cry-c plants, whereas the other was altered in la plants (Fig. 2). The mutant sequence altered in la plants cosegregated with the slender phenotype in a progeny in which LA and la segregated on a cry-s background (Supplemental Fig.  S1). On this background, la segregates were immediately recognizable as slender plants whereas plants carrying at least one LA allele were wild type in appearance. These data suggest that the first of the two cloned DELLA genes is LA.
In a separate cross segregating for both la and cry-s, each of the seven slender (la cry-s) F 2 plants carried the mutant form of the second DELLA gene, as well as the mutant form of the first DELLA gene (LA). The remaining 52 F 2 plants were tall (wild type) and thus carried LA and/or CRY. Because of the duplicate nature of LA and CRY, segregation for cry-s cannot be followed phenotypically in plants carrying LA. However, in this cross, cry-s was segregating in coupling phase with a tightly linked and easily seen leaf-wax marker (wa). Within the tall F 2 plants, an RFLP in the second sequence cosegregated with wa in 50 of the 52 plants. These data, taken together, strongly indicate that the second DELLA gene corresponds to CRY. Of 59 plants genotyped in this F 2 , clearly two plants were recombinant between CRY and WA. This corresponds to a recombination frequency of approximately 3% (Stevens, 1939). In turn, this recombination frequency agrees well with the previously reported value for linkage between WA and CRY of 2.0 6 1.1 (maximum likeli-hood method; Lamm, 1947). In a third cross, there was 100% cosegregation of wa and the second sequence (33 plants genotyped).
The nature of the la and cry-s mutations was established by comparing the LA and CRY DNA sequences with those of the la and cry-s mutants. The la mutant was found to result from a 190-bp insertion at position Gln-85 (Fig. 2), and the cry-s mutation involves a frame-shift deletion at position 152 ( Fig. 2). Both of these mutations are, therefore, predicted to encode nonfunctional proteins as a result of the outof-frame stop codons. Another mutation in CRY, cry-c, involves a G to A substitution at base 583, which results in a Gly to Gln substitution in the predicted protein. This, in turn, results in a reduced (but not abolished) capacity to inhibit growth; la cry-c plants are shorter than la cry-s plants (Reid et al., 1983).
As in the case of other multigene families, it would be expected that the gene pair, LA and CRY, arose from gene duplication. In fact, this gene pair is one of the first described examples of duplicate genes (Rasmusson, 1927). Interestingly, however, the duplication event  appears to be quite ancient, having occurred prior to the divergence of Arabidopsis and pea (Fig. 3).
Our molecular studies were performed using mutant alleles derived from lines that date from the foundation reports on cryptodwarf (genotype la cry-c le-1; Rasmusson, 1927) and slender (de Haan, 1927) peas, and on their relationship (Lamm, 1937). Full sequences were obtained for the la allele from Rasmusson's 'Gray Dwarf' line (HL2, Lamm line 2), cry-c from a Rasmusson cryptodwarf (HL8, Lamm line 8a), and cry-s from Lamm line 7 (HL7, Vilmorin's acacia line). In addition, the cry-s allele from de Haan line 204.1 (HL6, Lamm line 6) was shown to have the frame-shift deletion at position 152.
To investigate the roles of LA and CRY on root development, we recombined the severely GA-deficient mutant na-1 with la and/or cry-s. The na-1 mutant has been crucial for establishing a role for GAs in root development (Yaxley et al., 2001). This mutant has an extremely short shoot phenotype, termed nana, and roots that are shorter, thicker, and less ramified than those of wild-type plants (Yaxley et al., 2001). We found (Fig. 4) that recombining na-1 with la and cry-s essentially rescues the root phenotype of na-1. Interestingly, the la mutation on its own is able to largely rescue the na-1 root phenotype ( Fig. 4; Supplemental Table S1), whereas cry-s on its own does not. This suggests that LA is the main functioning DELLA protein in the roots of pea. The la mutation is also more effective at rescuing the shoot phenotype of na-1 plants, compared with cry-s (Fig. 4). It was previously noted that gene LA is a more effective inhibitor of shoot elongation than is CRY (de Hann, 1927;Reid et al., 1983). This difference is pronounced in the shoots of na-1 plants, and is especially clear in na-1 roots (Fig. 4).
In contrast to the roots, both of the null mutations la and cry-s are required to fully rescue the shoot phenotype of na-1 plants, and the shoot phenotype of na-1 la cry-s is slender ( Fig. 4; Potts et al., 1985). The GAsaturated shoot phenotype of la cry-s plants indicates that LA and CRY are the only DELLA genes operative in the shoot. Results from genomic Southern blots probed with PsLA and PsCRY also did not provide evidence for more than two DELLA genes in pea. Here we found that na-1 la plants heterozygous for CRY/cry-s had internodes much longer than nana plants, and fitting within the range of Mendel's le-1 dwarf. The dwarf phenotype of na la CRY/cry-s plants was not apparent in the previous investigation (Potts et al., 1985) because le-1 was also segregating. In contrast, we found homozygous na-1 la CRY plants had internodes only around half as long as the CRY/cry-s heterozygotes and a phenotype consistent with the upper end of the nana phenotypic range.
A Second GA 3-Oxidase Gene, PsGA3ox2, Is Expressed Primarily in the Roots and Is Responsible for the Conversion of GA 20 to GA 1 Before investigating the effects of DELLA proteins on GA synthesis gene expression, we sought to clone additional 3-oxidase genes from pea. The reasoning for this was that the roots of the GA biosynthesis mutants le-1 and le-2 (null) are phenotypically similar to wild type and contain similar levels of endogenous GA 20 and GA 1 to wild type, in contrast to their dwarf shoot phenotype (Yaxley et al., 2001). It was therefore expected that another GA 3-oxidase must carry out substantial 3-oxidation in the roots (Yaxley et al., 2001). Indeed, another pea GA 3-oxidase gene, PsGA3ox2 (Supplemental Fig. S2), was isolated using PCR primers based on Medicago sequence, 3# RACE, and cDNA  clones. The expression product of PsGA3ox2 converted [ 14 C]GA 20 to [ 14 C]GA 1 , as shown by HPLC and gas chromatography-mass spectrometry-selected ion monitoring, demonstrating its 3-oxidase activity (Supplemental Fig. S3).
The expression levels of the 3-oxidase genes PsGA3ox1 (also known as Mendel's LE; Lester et al., 1997;Martin et al., 1997) and PsGA3ox2 were measured in the shoot and root tissue of 6-d-old pea seedlings using real-time PCR. PsGA3ox1 was more highly expressed in the shoot compared with root tissue, with an approximate 2-fold difference (P , 0.001; Fig. 5). Conversely, PsGA3ox2 showed approximately 2-fold higher expression in the root tissue than the shoot tissue (P , 0.02; Fig. 5).

Pea DELLA Proteins Promote the Expression of GA Synthesis Genes and Inhibit That of GA Deactivation Genes
In Arabidopsis shoots, DELLA proteins feed-back regulate the GA biosynthesis genes AtGA3ox1 and AtGA20ox1 King et al., 2001;Silverstone et al., 2001). To investigate whether the pea DELLA proteins are involved in the feed-back regulation of key GA biosynthesis genes in pea roots, we undertook real-time PCR on LA and the mutant la on a cry-s background. In the la mutant, there was a 4-fold and 6-fold down-regulation of PsGA3ox1 (P , 0.001; Fig. 6A) and PsGA3ox2 (P , 0.05; Fig. 6A), respectively. The greatest effect on gene expression was seen for PsGA20ox1, which was down-regulated 14-fold in the mutant (P , 0.01; Fig. 6A). In contrast, a more than 2-fold up-regulation of the 2-oxidase genes PsGA2ox1 (P , 0.02; Fig. 6B) and PsGA2ox2 (P , 0.01; Fig. 6B) was seen in the mutant roots. Similar results were obtained for CRY and the mutant cry-s on a la background (Supplemental Fig. S4).

DISCUSSION
DELLA proteins play a pivotal role in GA signal transduction (Fu and Harberd, 2003). However, despite the historical importance of pea for physiological and genetic studies, DELLA genes have not previously been isolated from this species. In pea, the mutations la and cry-s together produce the slender phenotype, which played a pivotal role in the development of early theories on GA action. In 1957, Brian published the prescient theory that GAs, rather than positively promoting stem elongation, actually inhibit an inhibitor of that process. This was supported by genetic studies using GA-deficient mutants (Potts et al., 1985). Next came the discovery of growth-repressing DELLA proteins, and the finding that DELLAs are destabilized by GA (Harberd et al., 1998;Silverstone et al., 2001). Here we complete the picture by showing that LA and CRY encode DELLA proteins in pea. The mutant alleles la and cry-s both appear to encode nonfunctional proteins, and the stem elongation of la cry-s plants is similar to that of wild-type plants given a saturating dose of bioactive GA (Brian, 1957;Potts et al., 1985). It appears, therefore, that at least with respect to shoot elongation, LA and CRY are the only DELLA-encoding genes in pea.
It was shown previously that although the na-1 mutation dramatically reduces GA levels and leads to the very short nana phenotype, the la cry-s gene combina- For each gene, the shoot value was set to 1, and the level in the root sample was calculated relative to the corresponding shoot value. Shown are means with SE (n 5 4). It should be noted that direct comparison between the expression levels of the two genes is not valid. Plants were grown in pasteurized potting mix. tion is completely epistatic to na-1 in shoots (Potts et al., 1985). Here we show that la cry-s also rescues (Fig. 4) the distinctive na-1 root phenotype (Yaxley et al., 2001). Interestingly, la on its own largely rescues the na-1 root phenotype, at least in terms of elongation of lateral roots (Fig. 4). This indicates that LA may be the main functioning DELLA protein in roots.
We then used the pea DELLA mutations to examine the effects of these proteins on the expression of GA synthesis and deactivation genes in roots. In the DELLA slender mutants sln in barley , slr1 in rice (Itoh et al., 2002), and la cry-s in pea (Potts et al., 1985;Martin et al., 1996), the synthesis of bioactive GAs is reduced, but this has only been shown for shoots. We therefore monitored the expression of GA biosynthesis and deactivation genes in the roots of wild-type and slender pea plants. When both DELLA genes were null (la cry-s), there was a strong reduction in expression of the biosynthesis genes PsGA20ox1, PsGA3ox1, and PsGA3ox2, and a strong promotion of the deactivation genes PsGA2ox1 and PsGA2ox2, compared with LA cry-s or la CRYplants. Therefore, DELLA proteins promote the expression of GA synthesis genes and inhibit that of GA deactivation genes, indicating that in roots, DELLAs are an integral part of the feedback and feed-forward phenomena, whereby bioactive GA reduces GA synthesis and speeds up GA deactivation Zentella et al., 2007).
Another key GA gene from pea is Mendel's LE, also known as the 3-oxidase gene PsGA3ox1 (Lester et al., 1997;Martin et al., 1997). Mendel exploited the dwarf stature of mutant le-1 shoots in his original genetics experiments, but it is interesting to note that the roots of le-1 plants (and of other le mutants) are indistinguishable from the wild type, and contain normal GA levels (Yaxley et al., 2001). The identification of a second GA 3-oxidase gene (PsGA3ox2) from pea provides an explanation for these observations. It appears that PsGA3ox2, which is relatively strongly expressed in roots, can compensate for the reduction in PsGA3ox1 activity in le-1 roots, and even for the complete loss of that activity in roots of the null mutant le-2 (Martin et al., 1997;Lester et al., 1999). The capacity for compensation by PsGA3ox2 (and possibly by other, as yet unknown, 3-oxidase genes) is clearly reduced in the shoots and consequently le-1 and le-2 shoots are dwarfed. However, it appears that there is some compensation even in the very short le-2 shoots because they do produce a trace amount of GA 1 .
In conclusion, we have isolated the LA and CRY genes of pea and have shown that they encode DELLA proteins. This provides valuable support for the inhibitorof-an-inhibitor model of GA action, which was based originally on the slender la cry-s mutant (Brian, 1957). Of the two DELLA genes, LA appears to be the major one operating in roots, as indicated by the capacity of la on its own to largely rescue the root phenotype of the GA-deficient na-1 mutant. We have used the la and cry-s mutations to show that in pea roots, DELLA proteins can be viewed as positive regulators of the expression of GA biosynthesis genes, including the second pea 3-oxidase gene, PsGA3ox2. Our studies on the pea DELLA proteins LA and CRY further demonstrate the importance of GA signaling in the regulation of root growth.

Plant Material
Experiments involving gene expression studies and quantification of endogenous GAs were conducted with the tall (wild type) Hobart line HL2051 (LA CRY; . Progenies segregating for LA/la and/or CRY/cry-s were derived from the following crosses: HL133 (la cry-s NA) 3 NGB1766 (LA CRY na-1; Potts et al., 1985); HL6 (LA cry-s wa) 3 HL73 (la CRY WA); and HL6 (LA cry-s wa) 3 line K524 (LA CRY WA). Genotypes LA cry-s and la cry-s (Supplemental Fig. S2) were derived from the same F 4 plant from cross HL133 3 NGB1766. Other genotypes were selected from a cross between HL107 (LA CRY NA) and HL188 (la cry-s na-1; HL188 was selected from cross HL133 3 NGB1766). The foundation lines HL2 (Lamm line 2), HL6 (Lamm line 6), HL7 (Lamm line 7), and HL8 (Lamm line 8a) were kindly provided in 1957 by Dr. Robert Lamm.

Plant Growth and Chemical Treatments
Plants to be raised to maturity for genetic studies were grown in a 1:1 mixture of dolerite chips and vermiculite, topped with pasteurized peat/sand potting mix. Plants for gene expression experiments were grown in 100% potting mix for 4 to 5 d. Gene expression material was immediately immersed in liquid nitrogen and stored in a 270°C freezer.
A partial PsCRY sequence was obtained using nested PCR on cDNA, using primers based on the gene LS, a GRAS gene from tomato (Solanum lycopersicum; DELLAs belong to the GRAS protein family). The primary PCR was conducted with primers 5#-ATTCAACTGAACGGTTAGTCCA-3# and 5#-GCAAT-GTAGCTTCCAGTGAATC-3#, followed by a secondary PCR with primers 5#-GTTTACTCAATTAACCGCTAATCA-3# and 5#-AATGTAGCTTCCAGT-GAATCAAA-3#. Using a CAPS marker between the lines Terese and Torsdag , PsCRY was mapped between the two RAPD markers Q4_450 (at 2 cM) and X17_500 (10 cM) on LGVII, not far from RMS4 . The remainder of the sequences of PsLA and PsCRY was obtained by screening a cDNA library made from the shoots of 7-d-old deetiolated pea (Pisum sativum) shoots (Clontech).
The initial portion of PsGA3ox2 was isolated using primers based on a partial Medicago bacterial artificial chromosome sequence (gi89514974). The 3# end of the PsGA3ox2 sequence was isolated by 3# RACE (Frohman et al., 1988). The partial sequence thereby obtained was used as a probe to screen approximately 350,000 clones of a pea seedling shoot library. A single clone containing the 5# end of the gene was isolated, sequenced, and then ligated into pGem-T Easy (Clontech) and expressed in Escherichia coli. The functional activity of the expression product was tested as before (Lester et al., 1997), using [ 14 C]GA 20 as a substrate.

Segregation Studies
The segregation of PsLA was followed using two PCR primers that flanked the deletion in the la mutant, 5#-CTTAGCTGTATTAGGTTATAAGGTTCGTT-3# and 5#-TCTTCACGAGTCTATCAGCAATCTT-3#, giving a 542-bp band in the wild type, and a 727-bp band in the la mutant. The segregation of PsCRY was followed using two PCR primers, 5#-CTTGAACAAGCTATGGGTAATTT-TCA-3# and 5#-ATCCCTTTCTCCTGCGTT-3#, which amplified a PCR product containing several BccI sites near the cry-s mutation, one of which was polymorphic between cry-s and the CRY gene in line 107 (Torsdag).
RNA Extraction, cDNA Synthesis, and Quantitative PCR Plant material was ground to a fine powder with a mortar and pestle in liquid N 2 . Approximately 100 mg of ground tissue was used for RNA extraction, as carried out by Wolbang et al. (2004). One to 2 mg of total RNA was used to synthesize single-strand cDNA using the QuantiTect reverse transcription kit (QIAGEN). cDNA samples were diluted to a total volume of 100 mL.
Two microliters of cDNA was used for quantitative real-time PCR using Bio-Rad iQ Sybr master mix (Bio-Rad) following the manufacturer's recommendations and run on a Rotorgene 2000 dual-channel machine (Corbett Research). Mean expression levels of the gene of interest were calculated relative to the expression of actin.
Sequence data from this article can be found in the GenBank data libraries under accession number(s) DQ845340 (PsCRY), DQ848351 (PsLA), and DQ864759 (PsGA3ox2).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Cosegregation of the slender mutant and the LA/la gene pair.
Supplemental Figure S4. Effects of cry on expression of GA genes.
Supplemental Table S1. Length of lateral roots of pea genotypes.