Analyses of GA20ox- and GID1-over-expressing aspen suggest that gibberellins play two distinct roles in wood formation

Authors


*(fax +46 90 786 8165; e-mail thomas.moritz@genfys.slu.se).

Summary

Gibberellins (GAs) are involved in many aspects of plant development, including shoot growth, flowering and wood formation. Increased levels of bioactive GAs are known to induce xylogenesis and xylem fiber elongation in aspen. However, there is currently little information on the response pathway(s) that mediate GA effects on wood formation. Here we characterize an important element of the GA pathway in hybrid aspen: the GA receptor, GID1. Four orthologs of GID1 were identified in Populus tremula × P. tremuloides (PttGID1.1–1.4). These were functional when expressed in Arabidopsis thaliana, and appear to present a degree of sub-functionalization in hybrid aspen. PttGID1.1 and PttGID1.3 were over-expressed in independent lines of hybrid aspen using either the 35S promoter or a xylem-specific promoter (LMX5). The 35S:PttGID1 over-expressors shared several phenotypic traits previously described in 35S:AtGA20ox1 over-expressors, including rapid growth, increased elongation, and increased xylogenesis. However, their xylem fibers were not elongated, unlike those of 35S:AtGA20ox1 plants. Similar differences in the xylem fiber phenotype were observed when PttGID1.1, PttGID1.3 or AtGA20ox1 were expressed under the control of the LMX5 promoter, suggesting either that PttGID1.1 and PttGID1.3 play no role in fiber elongation or that GA homeostasis is strongly controlled when GA signaling is altered. Our data suggest that GAs are required in two distinct wood-formation processes that have tissue-specific signaling pathways: xylogenesis, as mediated by GA signaling in the cambium, and fiber elongation in the developing xylem.

Introduction

Hormones in wood formation

The major components of woody stems are phloem and xylem tissues, which are generated throughout the plant’s life by the meristematic activity of the vascular cambium, which expands in the form of a hollow cylinder between the phloem and xylem. Many factors are involved in various stages of cambial differentiation and wood formation, including several plant hormones that have important regulatory functions. Notably, a steep concentration gradient of IAA (indole-3-acetic acid) occurs across the cambial zone, and has been suggested to provide positional information, a major determinant of cambial growth (Uggla et al., 1996; Tuominen et al., 1997). Moreover, auxin also induces xylogenesis in a concentration-dependent manner (Sundberg and Little, 1990) and controls fiber elongation (Digby and Wareing, 1966). These findings, and various other lines of evidence, have shown that IAA plays key roles in cambium maintenance and vascular development (Jacobs, 1952; Mattsson et al., 1999). Gibberellins (GAs) are also known to be involved in wood formation, specifically in the differentiation of xylem fibers (Wareing, 1958; Digby and Wareing, 1966). In addition, IAA and GAs may act together in xylem differentiation (Digby and Wareing, 1966), in accordance with evidence that they act synergistically in the coordination of growth control during seedling, stem and fruit development (Yamaguchi, 2008). Furthermore, several studies have shown that IAA can stimulate the expression of GA biosynthetic genes in various plants (Ross et al., 2000, 2001; Wolbang and Ross, 2001; O’Neill and Ross, 2002; Wolbang et al., 2004), and GA also apparently mediates increased rates of polar IAA transport in aspen (Bjorklund et al., 2007).

However, some experiments have demonstrated that GA can have pronounced effects on xylem by themselves. First, increases in GA content (due to over-expression of genes encoding GA biosynthesis enzymes) can induce significant increases in levels of xylem lignification in tobacco and changes in syringyl/guaiacyl ratios in poplar (Israelsson et al., 2003; Biemelt et al., 2004, respectively). Second, GA alone can affect the length of xylem fibers, as injecting GA inhibitors into woody stems leads to reductions in fiber length, even though IAA levels remain unchanged (Ridoutt et al., 1996). The xylem-elongating effect of GA is also supported by the GA concentration gradient observed across wood-forming tissues, with a large peak in the elongating xylem (Israelsson et al., 2005).

GA biosynthesis and signaling

GAs consist of a large group of tetracyclic diterpenes, a few of which are highly bioactive and play roles in diverse aspects of plant growth and development, not only in wood formation, but also numerous processes involving cell division and/or elongation, including seed germination, trichome formation, stem elongation, leaf expansion, flowering and pollen maturation (Kende and Zeevaart, 1997; Olszewski et al., 2002). GA biosynthesis is mainly controlled by feedback regulation, which provides an important link between GA metabolism and GA signaling pathways (Phillips et al., 1995; Hedden and Phillips, 2000; Olszewski et al., 2002; Weston et al., 2008). However, a wide range of other factors, reviewed by Yamaguchi (2008), are also involved in the control of GA biosynthesis, including light (Oh et al., 2006), temperature (Yamauchi et al., 2004), various stresses (Magome et al., 2004) and the hormones auxin (Ross et al., 2000), ethylene (Achard et al., 2007) and abscisic acid (ABA) (Seo et al., 2006). Much progress, reviewed in Hirano et al. (2008), Schwechheimer (2008) and Ueguchi-Tanaka et al. (2007b), has recently been made towards elucidating the mechanisms of GA signaling and responses using rice and A. thaliana mutants. DELLA proteins and the soluble GA receptor (GID1) are key mediators of GA response pathways. DELLA proteins act as negative regulators (Peng and Harberd, 1997; Ikeda et al., 2001) and are degraded in response to GA treatments (Dill et al., 2001; Silverstone et al., 2001). The recent discovery of GID1 (Ueguchi-Tanaka et al., 2005; Griffiths et al., 2006; Nakajima et al., 2006) has led to revelations regarding several key events in GA perception and subsequent degradation of DELLA (Ueguchi-Tanaka et al., 2007a). In the absence of GA, the DELLA protein represses the GA response pathway, whereas a GA/GID1 complex is formed and binds to DELLA proteins in the presence of GA. This GID1/GA/DELLA complex is stable, and targets DELLA for degradation through the SCFGID2/SLY1 proteasome pathway (McGinnis et al., 2003; Sasaki et al., 2003), thus allowing GA signaling to proceed.

Increasing GA levels in hybrid aspen through over-expression of a key gene in the GA biosynthesis pathway (GA 20-oxidase) has been previously shown to affect wood formation. The resulting plants show increased rates of xylogenesis and elongated xylem fibers in comparison to wild-type counterparts (Eriksson et al., 2000). Nevertheless, there is currently little information on the response pathway(s) that mediate these effects on wood formation.

In order to obtain further details of the role of GA signaling in the regulation of wood formation, we studied the potential involvement of the GID1 receptor in this developmental process in aspen by cloning and characterizing members of the GID1 gene family. In addition, we explored their functions in wood formation by analyzing their expression patterns, and examining effects of their over-expression using the CaMV 35S promoter and a xylem-specific promoter (LMX5).

Results and discussion

Orthologs of OsGID1 in poplar

We found four sequences orthologous to rice OsGID1 in the poplar genome using the BLAST program of the Joint Genome Institute database (Tuskan et al., 2006). ESTs from various organs (in the Populas EST database (Sterky et al., 2004)) have been found to correspond to these gene models (Table 1).

Table 1.   Poplar orthologs of OsGID1
NameGene modelcDNA libraries
  1. The gene models are presented with the P. trichocarpa Jamboree gene model names. The ESTs were found in cDNA libraries included in Populus DB.

PtGID1.1estExt_fgenesh4_pg.C_700172Female catkins/imbibed seeds
PtGID1.2estExt_fgenesh4_pm.C_LG_XIII0124Roots
PtGID1.3estExt_Genewise1_v1.C_LG_XIV2782Female catkins/senescing leaves
PtGID1.4eugene3.00001948Roots

An alignment of the deduced amino acid sequences of these four aspen genes with known GID1 sequences (from Oryza sativa and A. thaliana) is presented in Figure 1(a). Most of the residues known to be involved in GA binding by OsGID1 and its interactions with SLR1 (the rice ortholog of DELLA– SLENDER RICE1) (Ueguchi-Tanaka et al., 2007a) are conserved in the orthologs; three of these amino acids are not conserved in the PtGID1.3 sequence, only one of them is not conserved in both PtGID1.2 and PtGID1.4 sequences, and all of the residues are present in the PtGID1.1 sequence.

Figure 1.

 Protein sequence alignment of rice, aspen and A. thaliana GID1 proteins.
(a) Alignment by Clustal W of the protein sequences of the four PtGID1s, OsGID1 and the AtGID1s. The important residues for GA binding and SLR1 interactions in OsGID1 are shaded in gray. The arrows indicate the catalytic triad of Hormone-sensitive Lipase.
(b) Unrooted neighbor-joining phylogenetic tree of the same protein sequences.

In the phylogenetic tree derived from the sequences shown in Figure 1(a), two sub-groups, both including two of the PtGID1s and one or two AtGID1s, can be distinguished at approximately the same distances from OsGID1 (Figure 1b). OsGID1 has around 60% identity with all of the other sequences, while the AtGID1s have approximately 70–80% similarity with the PtGID1s in their respective sub-groups. The similarity between the two sub-groups of PtGID1 is quite high (78%), and between the two sequences in each sub-group it is even higher (94%). These findings indicate that the PtGID1 pairs originated from recent gene duplication events. The presence of A. thaliana and P. tremula GID1s in both sub-groups also indicates that the two types of GID1 may have different functions (Nakajima et al., 2006).

Ubiquitous expression of GID1s in aspen

In the present study, we examined two types of aspen: P. tremula (aspen) and P. tremula × tremuloides (hybrid aspen). The former was grown outside, while the latter, which can be easily transformed (Nilsson et al., 1992), was rapidly grown in the greenhouse. We designated the GID1 genes detected in these plants PtGID1 and PttGID1, respectively.

To elucidate the roles of each aspen GID1 in developmental processes, we examined their spatial expression patterns in 3-month-old hybrid aspens by quantitative RT-PCR analysis of samples from their main organs. Expression of PttGID1 genes was detected in every sample (Figures 2 and S1), in accordance with abundant indications that GAs are active throughout plants during many developmental processes (Fleet and Sun, 2005), and similar to the spatial expression pattern of AtGID1 genes (Griffiths et al., 2006; Nakajima et al., 2006).

Figure 2.

 Relative expression levels of the PttGID1 genes measured by quantitative RT-PCR. Expression levels in the following organs of 3-month-old hybrid aspen: A, apex; yL, young leaf; mL, medium-aged leaf (fully elongated); oL, old leaf (mature); yN, young internode; mN, medium-aged internode; oN, old internode; R, roots. The expression levels of each gene are relative to that of the At4g34270 ortholog, and to their value in the apex (set at 1) (ND, not detected). Values are means ± SE of three technical replicates. The results for three biological replicates are shown in Figure S1.

The two genes of the first sub-group (PttGID1.1 and 1.2 in Figure 1b) showed almost identical expression patterns, but those of the two genes of the second sub-group (PttGID1.3 and 1.4 in Figure 1b) differed considerably. The expression pattern of PttGID1.3 followed the pattern of the first sub-group of genes. PttGID1.4 is expressed in old leaves and internodes (but was not always detected in some biological replicates) and at a high level in roots. The expression levels of the other three genes increased during the development of organs, especially the leaves. Similar increases in the expression of AtGID1a during development have been detected in A. thaliana (Griffiths et al., 2006), with leaves of 24-day-old plants showing higher AtGID1 levels than 11-day-old rosettes, and inflorescences showing higher levels at 31 days than at 24 days (Griffiths et al., 2006). PttGID1.4 is expressed to a higher level in the roots, in comparison with the other three genes, which are expressed more highly in the leaves, suggesting a specific role for PttGID1.4.

PttGID1over-expression enhances GA signaling inA. thalianaand aspen

To identify the functions of the PttGID1 genes, we studied one gene from each sub-group: PttGID1.1, which has the same expression pattern as PttGID1.2 and encodes a protein containing all the important conserved residues (Figure 1a), and PttGID1.3, which encodes the protein with the most numerous changes within the conserved motifs (Figure 1a).

We transformed A. thaliana and hybrid aspen with PttGID1.1 and PttGID1.3 under the control of the 35S promoter. Seven 35S::PttGID1.1 and eight 35S::PttGID1.3 A. thaliana lines were obtained, each with only one copy of the respective inserts. All of these lines expressed the respective transgene, but at very different levels (with a 5000-fold range), so the line most strongly expressing each transgene was selected. Both selected transgenic lines presented similar phenotypic traits (Figure S2), being taller and more responsive (in terms of hypocotyl elongation) to bioactive GA than wild-type when grown in vitro in the presence of GA4 (data not shown). In addition, their flowering time was shorter under short-day conditions, and their rosettes were larger in diameter (Table 2 and Figure S2). Their phenotypic deviations from wild-type were stronger than those described for AtGID1a over-expressors (Willige et al., 2007). There were probably several reasons for this, including the fact that the PttGID1 over-expressors that we examined were grown under short photoperiods, while the plants observed by Willige et al. (2007) were grown under long-day conditions, in which the requirement of GAs for floral induction is less important (Putterill et al., 2004). However, the phenotypic traits that we observed were consistent with altered GA signaling, suggesting that PttGID1.1 and 1.3 genes can function heterologously in A. thaliana as GA receptors.

Table 2.   The flowering times and rosette diameters of A. thaliana PttGID1 over-expressors
GenotypeFlowering time (day)Flowering time (number of leaves)Rosette diameter (mm)
  1. Values are means ± SE.

Col-057.6 ± 1.5157.4 ± 1.1999.2 ± 1.74
35S:PttGID1.151.2 ± 150.1 ± 0.82135.6 ± 2.57
35S:PttGID1.352.5 ± 0.9549 ± 0.94147.9 ± 1.76

For the hybrid aspen analyses, we selected the three lines of PttGID1.1 and PttGID1.3 over-expressors that showed the highest global PttGID1 expression levels (Figure 3a). The over-expressors grew faster than the wild-type, being 50 cm taller after 9 weeks in soil in the greenhouse (Figure 3b–d). This increase in height was not correlated with an increased number of nodes, but was instead due to longer internodes, a characteristic that is also seen in hybrid aspen AtGA20ox1 over-expressors (Eriksson et al., 2000). We also observed several other phenotypic characteristics that were similar to those seen in AtGA20ox1 over-expressors, including longer leaves and petioles (Figure 3e,f). However, we did not notice any problems in rooting the regenerated PttGID1 over-expressors, although such problems have been reported for AtGA20ox1 over-expressors (Eriksson et al., 2000). This suggests that over-expression of GID1 in hybrid aspen leads to phenotypic changes that are similar to, but slightly less severe than, those associated with increases in GA levels.

Figure 3.

 Phenotypic and molecular characterization of hybrid aspen 35S:PttGID1 over-expressors.
(a) PttGID1 expression levels measured by quantitative RT-PCR in elongated internodes of wild-type hybrid aspen, three lines over-expressing PttGID1.1 (82.2, 5, 13), and three lines over-expressing PttGID1.3 (87.5, 6, 12). The expression levels are relative to UBQ, and to the wild-type value (set at 1). Values are means ± SE of three biological replicates.
(b) Phenotype of three transgenic plants, with a control plant on the left from the same day of sampling.
(c,d) Cumulative shoot elongation of the same lines shown in (a), grown in the greenhouse. Time point 0 corresponds to 4-week-old plants.
(e) Images of parts of the stem of a representative control plant (left) and transgenic plant (right).
(f) Image of a mature leaf of each of the two plants shown in (e).
(g) GA responsiveness of 35S::PttGID1 over-expressors. Height increment after 4 weeks of growth in vitro in the presence of 10 μm paclobutrazol or GA4. The results are mean values of three replicates ± SE.

To confirm the GA receptor functions of PttGID1.1 and PttGID1.3, we compared their responsiveness to GAs with that of controls by applying GA and paclobutrazol, an inhibitor of GA synthesis, to plants growing in vitro. As expected from the previous observations on soil-grown plants, the transgenic plants grew more strongly than the wild-type under the control conditions (Figure 3g). Both PttGID1.1 and 1.3 over-expressors showed more pronounced responses to GA than the wild-type (Figure 3g). However, surprisingly, they differed in their responses to paclobutrazol, with PttGID1.3 over-expressors being less sensitive to the inhibitor than PttGID1.1 over-expressors and wild-type. In contrast, no clear rooting differences were observed, and both wild-type and over-expressors responded in a similar manner to in vitro additions of GA3 or GA4 (data not shown).

Collectively, these results clearly demonstrate that GA signaling is enhanced in A. thaliana and aspen PttGID1 over-expressors, indicating that both the PttGID1.1 and 1.3 genes (and probably the other two PttGID1 genes given their high sequence identity) encode proteins that function as GA receptors.

The surprising difference in paclobutrazol sensitivity between the over-expressors (Figure 3g) warrants further consideration. Intriguingly, the plants least sensitive to this inhibitor were those over-expressing PttGID1.3, which shows the weakest conservation of the residues required for GID1/GA/DELLA binding (Ueguchi-Tanaka et al., 2007a). Furthermore, the unconserved amino acids are located in the ‘lid’ of the receptor protein, corresponding to the postulated region of contact with the DELLA protein rather than that with GA (Ueguchi-Tanaka et al., 2007a; Hirano et al., 2008). This suggests that the unconserved amino acids of PttGID1.3 may allow it to bind to DELLA in the absence of GA, and so, reduces the sensitivity of pttG1D1.3 over-expressors to a reduction in GA levels.

There is also evidence to suggest at least some degree of sub-functionalization among members of the PttGID1 family. For instance, PttGID1.4 could be involved in root development, as the two 35S::PttGID1 over-expressors (which do not over-express PttGID1.4) did not show the characteristic rooting problems of the 35S::AtGA20ox1 over-expressors (Eriksson et al., 2000). The rooting difficulties of the latter are presumably due to enhanced GA responses mediated by a GA receptor, but the GA receptor involved appears to be neither PttGID1.1 nor PttGID1.3. Hence, PttGID1.4 may be the most active GA receptor in the roots, as it is the one that shows its highest expression in this organ (Figure 2).

Hence, the activities of the four PttGID1 genes may be less redundant than those of the three AtGID1 genes, which have some specificity of action even though they can also substitute for one another (Iuchi et al., 2007). However, analyses of specific over-expressors/knockout plants are required to confirm such a hypothesis. Next, we characterized the molecular basis of the GA signaling enhancement in PttGID1 over-expressors in more detail by quantifying the levels of the molecules, the four PttGID1 transcripts and various GAs involved in the signaling.

PttGID1 over-expression leads to transcript level negative feedback regulation of PttGID1 paralogs

The expression levels of the four PttGID1 genes were assessed in quantitative RT-PCR analyses, which detected a more than 100-fold decrease in PttGID1.3 transcript levels in PttGID1.1 over-expressors (not detected, Figure 4a), and a more than fivefold decrease in PttGID1.1 transcript levels in PttGID1.3 over-expressors, in comparison to wild-type plants. However, reductions in the expression levels of PttGID1.2 were less clear (Figure 4a) in both types of over-expressors, and no expression of PttGID1.4 was detected, which is not surprising as tissues in which no PttGID1.4 transcripts were previously detected were used in the expression analysis (Figure 2). These results indicate that expression of the PttGID1.1, PttGID1.3 and (possibly) PttGID1.2 genes is subject to feedback regulation at the transcriptional level. The regulatory interactions may induce reductions in GA responsiveness, but if so they are not sufficient to compensate for the substantial enhancement of GA responsiveness generated by strong 35S-mediated PttGID1.1 or PttGID1.3 over-expression.

Figure 4.

 Level of PttGID1 gene expression and GA biosynthesis in hybrid aspen over-expressing 35S:PttGID1.
(a,c) Relative expression levels of (a) PttGID1 genes and (c) PttGA20ox1 measured by quantitative RT-PCR in elongated internodes of wild-type hybrid aspen and three lines of each PttGID1 over-expressor. The expression levels of each gene are relative to UBQ, and to the value of each gene in the wild-type (set at 1). No expression of PttGID1.4 was detected in these plants (ND, not detected). Values are means ± SE of three biological replicates.
(b) GA contents.

PttGID1 over-expression leads to negative feedback of GA biosynthesis pathways

It is interesting to compare the levels of GAs in PttGID1 over-expressors to those in AtGA20ox1 over-expressors, as substantial modifications in the GA profiles of 35S::AtGA20ox1 plants have previously been found (Eriksson et al., 2000), with high levels of bioactive GAs (GA1 and GA4), their immediate precursors (GA20 and GA9), and their catabolites (GA8 and GA34). Here we examined GA profiles in the elongating internodes of both PttGID1.1 and 1.3 over-expressors and wild-type (Figure 4b). We measured distant C20 GA precursors (GA53, GA19 and GA24, respectively), C19 GAs including immediate precursors (GA20 and GA9, respectively) and bioactive GAs (GA1 and GA4, respectively) and their C2-hydroxylated catabolites (GA8 and GA34, respectively) for both the early and non-early 13-hydroxylation pathways. Both types of PttGID1 over-expressors showed similar perturbations; lower levels of bioactive GAs and other C19 GAs, and higher levels of C20 GAs.

The decreased levels of bioactive gibberellins suggest that an increase in GA signaling resulting from GID1 over-expression induces feedback mechanisms that repress the GA biosynthesis pathway. This has been observed many times, and is often mediated by inhibition of the GA20ox and GA3ox biosynthetic genes (Croker et al., 1990; Martin et al., 1996; Hedden and Kamiya, 1997). Furthermore, the key step in aspen GA biosynthesis is 20-oxidation (Israelsson et al., 2004). Here, the observed switch in the PttGID1 over-expressors to higher concentrations of the C20 GA substrates of GA 20-oxidase (GA24 and GA53) and lower concentrations of C19 GAs (GA20 and GA9) suggests that PttGA20ox expression could be inhibited due to increased GA responsiveness. Hence, the expression levels of PttGA20ox1 (the only PttGA20ox characterized to date; Eriksson and Moritz, 2002) and putative P. tremula × tremuloides orthologs of PtGA20ox were assessed to test this hypothesis. Only PttGA20ox1 and one of the PtGA20ox putative orthologs appeared to be expressed in the studied tissues (elongated internodes). PttGA20ox1 expression levels were reduced in all lines (Figure 4c), whereas the levels of PtGA20ox3 were only reduced in some lines (Figure S3). These findings are consistent with studies showing that up-regulation of OsGA20ox2 in gid1-1 and gid1-2 rice mutants leads to elevated levels of bioactive GA1 (Ueguchi-Tanaka et al., 2005). However, the feedback-mediated down-regulation of bioactive GA levels that we detected does not appear to be sufficient to compensate for the increases in GA responsiveness of the PttGID1 over-expressors.

PtGID1 genes show different expression patterns within the wood-forming tissues

Several studies have shown that GA strongly influence the number and length of fibers produced during wood formation in trees (Digby and Wareing, 1966; Eriksson et al., 2000). To improve our understanding of the role of GAs during wood formation, we examined the spatial expression patterns of PtGID1 genes by quantitative RT-PCR in longitudinal tangential cryosections of the cambial zone of a 12-year-old P. tremula tree growing outside (as current techniques mean that cryosections of tree stems can only be obtained from thick stems).

In accordance with previous results (Figure 2), PtGID1.1 and 1.2 were expressed with very similar patterns throughout the wood-forming tissues (Figures 5a and S4). PtGID1.4 expression was not detected, but PtGID1.3 was expressed ten times more strongly in the phloem than in the xylem. The differences in expression patterns displayed by the PtGID1 genes suggest that they may play distinct roles in wood formation.

Figure 5.

PttGID1 expression in the wood-forming region.
(a) Relative expression levels of PtGID1s in the wood-forming zone. The indicated area corresponds to the cambium (Sections 4 and 5), phloem (Sections 1–3) and xylem (Sections 6–8). The expression levels of each gene are relative to UBQ, and to their lowest value (set at 1). No expression of PttGID1.4 was detected. Values are means ± SE of three technical replicates. The results for two biological replicates are shown in Figure S4.
(b–d) Wood characteristics of PttGID1 over-expressors. (b) Pith/xylem ratios in tangential sections of fully elongated internodes. (c, d) Diameter and length of fibers in the same internodes.

PttGID1 over-expression affects some but not all aspects of wood formation

We examined the wood of the 35S::PttGID1.1 and 1.3 over-expressors to test whether PttGID1 genes play a role in wood formation. Transverse sections of old internodes showed that the pith/xylem ratio was lower in both types of PttGID1 over-expressor than in the wild-type (Figure 5b). As the diameter of the fibers was not larger in the transgenic plants (Figure 5c), the observed increase in xylem was presumably correlated with higher cell numbers. This is consistent with previously observed correlations between GA levels and the number and length of fibers in GA and/or GA biosynthesis inhibitor experiments (Wareing, 1958; Digby and Wareing, 1966; Ridoutt et al., 1996), and in transgenic tobacco plants expressing AtGA20ox1 and AtGA2ox2 (Biemelt et al., 2004) and transgenic hybrid aspen expressing AtGA20ox1 (Eriksson et al., 2000). However, in contrast to the cited hybrid aspen study, we did not observe a significant difference in fiber length between 35S::PttGID1 over-expressors and wild-type (Figure 5d). In conclusion, increasing the levels of either PttGID1.1 or PttGID1.3 in aspen stimulates cambial activity to produce more xylem, but has no apparent effect on fiber elongation during wood formation.

GAs could act in two distinct areas in the wood-forming region

The differences observed in fiber elongation between AtGA20ox1 and the two types of PttGID1 over-expressors might have the following explanation. Expression of GUS controlled by the 35S promoter in cross-sections of hybrid aspen stems has been shown to be heterogeneous: high in the cortex, phloem and cambial zone, but lower in the differentiating xylem (Nilsson et al., 1996; Chen et al., 2000). Therefore, the increased amount of GA resulting from the higher production of GA 20-oxidase in 35S:AtGA20ox1 plants could probably be transported to the xylem, leading to an increase in xylem GA levels and thus increasing fiber elongation. In contrast, GA receptors probably cannot move to the xylem, hence GA signaling and fiber elongation are likely to remain at wild-type levels in the xylem of our PttGID1 over-expressors. If this rationale is correct, xylogenesis and fiber elongation could be regulated by GAs in two distinct areas of the wood-forming tissues: within the cambial cells for xylogenesis, and within the developing xylem for fiber elongation.

A xylem-specific increase inPttGID1 expression has no impact on fiber elongation

To clarify the role of PttGID1 genes in fiber elongation during wood formation, we used the xylem-specific promoter (LMX5) of the gene AI164126, which induces high expression in developing xylem (Bjorklund, 2007), to drive the expression of PttGID1.1 and 1.3 genes in hybrid aspen. The plants transformed with the two transgenes showed PttGID1 expression patterns that were opposite to those found in the wild-type, producing up to four times more PttGID1 transcripts in the xylem than in the phloem, whereas wild-type transcript levels of PttGID1s were two-fold lower in the xylem than in the phloem (Figure 6a). Nevertheless, PttGID1 expression was still ten times higher in the xylem of the transgenic plants than in the wild-type.

Figure 6.

 Characterization of hybrid aspens over-expressing PttGID1.1, PttGID1.3 and AtGA20ox1 under the control of the LMX5 promoter.
(a) Expression levels of PttGID1 in the xylem and phloem of internode 55 in LMX5::PttGID1 over-expressors. The expression levels are relative to that of the At4g34270 ortholog, and to the value in the phloem of wild-type (set at 1). Values are means ± SE of two biological replicates.
(b) Mean cumulative shoot elongation for five lines of each construct at the day of sampling.
(c) Pith/xylem ratios in tangential sections of fully elongated internodes.
(d,e) Diameter and length of fibers in the same internodes.
(f) Expression levels of AtGA20ox1 in the xylem and phloem of internode 55 of AtGA20ox1 over-expressors. As AtGA20ox1 is an A. thaliana gene, its expression was not measured in wild-type. The expression levels are relative to the At4g34270 ortholog, and to the lower value (set at 1). Values are means ± SE of two biological replicates.

Considering the promoter used, it was no surprise that the phenotype of the LMX5::PttGID1 plants did not show any major modifications (Figure 6b). There were no significant differences in their secondary growth traits, including no significant reductions in the pith/xylem ratio (Figure 6c), and no significant changes in the diameter of their fibers (Figure 6d). Surprisingly, there was no difference in the length of their fibers either (Figure 6e), despite the higher expression of PttGID1 genes in the xylem, in contrast to what might be expected from an increase in GA signaling within this tissue.

Unidirectional transport of GAs from phloem to xylem in aspen

The unexpected lack of effects on fiber elongation in LMX5::PttGID1 plants prompted us to examine transgenic hybrid aspens expressing AtGA20ox1 under the control of the LMX5 promoter. As expected, because they strongly express AtGA20ox1 in the xylem (Figure 6f), these plants did not show significant differences in height (Figure 6b) or cross-sectional xylem areas (Figure 6c). However, the transgenic xylem fibers were significantly longer than in the wild-type plants (Figure 6d). These observations confirmed that increases in xylem GA levels are required to induce fiber elongation, supporting the hypothesis that GA is transported to the xylem, which could explain the GA-mediated increase in fiber length observed in the xylem of 35S::AtGA20ox1 plants. Moreover, although 35S::AtGA20ox1 and 35S::PttGID1 plants showed an increase in xylogenesis, due to strong expression of the respective transgenes within the cambium, LMX5::AtGA20ox1 plants did not show any increase in xylogenesis. These findings imply that the GA signaling enhancement is confined to the xylem in which AtGA20ox1 is expressed, suggesting that there is no transport of GAs from xylem to phloem, and thus corroborating the model suggested by Israelsson et al. (2005).

Differences in the regulation of GA signaling in the xylem ofLMX5::AtGA20ox1 andLMX5::PttGID1 plants

GA signaling is enhanced in the xylem of LMX5::PttGID1 and LMX5::AtGA20ox1 plants through increases in GA receptors and biosynthesis, respectively. However, the expected increase in fiber elongation is lacking in LMX5::PttGID1 plants. To determine the reasons for this, we analyzed GAs and gene expression in the wood-forming region of both types of transgenic plants to examine the effects of the transgenes on GA signaling in this region.

The xylem GA profiles in LMX5::AtGA20ox1 plants were quite similar to those of AtGA20ox1 over-expressors (Eriksson et al., 2000), with lower levels of C20 GAs and higher levels of C19 GAs, including bioactive GAs (but almost only GA4), than wild-type plants (Figure 7a). Surprisingly, a decrease in the amount of C20 GAs, and higher production of GA9 and GA20, was detected in the phloem of the transgenic plants (Figure 7b), and the levels of bioactive GA were increased much less than in the xylem. Interestingly, the level of only one bioactive GA (GA4) was enhanced in the LMX5::AtGA20ox1 transformants, which have elongated fibers. Hence, the gibberellin that plays an active role in fiber elongation is presumably GA4, in accordance with previous suggestions regarding the elongation growth of aspen (Israelsson et al., 2003).

Figure 7.

 GA content in LMX5::AtGA20ox1 and LMX5::PttGID1 plants, and PttGID1 feedback regulation.
(a,b) GA content in xylem (a) and phloem (b) of internode 55 of LMX5::AtGA20ox1 plants.
(c–f) GA content in xylem (c,e) and phloem (d,f) of internode 55 of LMX5::PttGID1 plants. (c,d) Early 13-hydroxylation pathway. (e,f) Non-early 13-hydroxylation pathway.
(g) Expression levels of PttGID1 genes in the xylem of internode 55 of LMX5::PttGID1. The expression levels of each gene are relative to the At4g34270 ortholog, and to the value of each gene in wild-type (set at 1). Values are means ± SE of two biological replicates.

LMX5::PttGID1.1 and 1.3 plants had very similar GA profiles, which differed from those of wild-type (Figure 7c–f). In the xylem, the GA content generated by the early 13-hydroxylation pathway (Figure 7c) showed similar changes to those observed in the PttGID1 over-expressors (Figure 4b), most notably reduced amounts of C19 GAs, including GA1. However, no changes were observed in the abundance of GAs produced via the other pathway (Figure 7e). This specificity suggests that the mechanisms regulating PttGID1 genes and GA biosynthesis in the wood-forming region of LMX5::PttGID1 plants differ from those in the internodes of 35S::PttGID1 plants. However, this difference in GA content resulting from only one biosynthesis pathway was surprising. A hypothesis that could explain this finding is that the affinity of the receptors for GA1 and GA4 may differ. However, this is not the likeliest explanation as it implies that each pathway is regulated by specific feedback mechanisms.

In addition to a regulatory effect on GA biosynthesis, increases in PttGID1.1 and 1.3 gene expression have a negative effect on the expression of PttGID1 counterpart genes (Figure 7i): PttGID1.1 is more weakly expressed in the plants expressing PttGID1.3 under the control of the LMX5 promoter than in wild-type plants (although the reduction is less severe than in the 35S::PttGID1 over-expressors; Figure 4a), and the expression of PttGID1.3 is slightly decreased in the LMX5::PttGID1.1 plants (Figure 7i). However, modifications of PttGID1.2 expression were less clear. Similar trends to those observed in the young internodes of the 35S::PttGID1 over-expressors (Figure 4a) were also observed in the xylem of mature internodes in plants that express the transgenes specifically in this tissue.

A simple hypothesis that could explain the lack of phenotypic effects, despite the increase of PttGID1 in the xylem, is that there is no enhancement of GA signaling in the xylem of these lines. In accordance with this hypothesis, the PttGID1 genes driven by the LMX5 promoter were only slightly over-expressed relative to the increases when the 35S promoter was used (Figure 3a and 6a), and the cumulative effect of down-regulation of the PttGID1 paralogs (Figure 7i) and reductions in GA levels (Figure 7c), which together could lead to substantial reductions in GA signaling, may have compensated for their slight over-expression. Our study provides evidence of a complex regulatory mechanism that could maintain homeostatic GA signaling in the plant by modulating GA responsiveness at both the GA biosynthesis and GID1 receptor levels. Slight changes in GA signaling, such as the changes induced in the LMX5::PttGID1 plants, might activate this homeostatic mechanism, thereby maintaining a constant level of GA signaling and phenotypic stability. However, the effects of exogenous GA or the 35S promoter may be too strong for this subtle mechanism to correct, leading to changes in GA signaling and hence phenotypic alterations. Furthermore, the expression of AtGA20ox1 also affected the GA signaling and phenotype of aspen, even when driven by the LMX5 promoter, presumably because this gene affects GA biosynthesis directly, and increasing its activity has stronger effects than the feedback mechanisms regulating the endogenous PttGA20ox1 genes.

Conclusion

In the present study, we have shown that aspen has four homologs of OsGID1, and that the two we used in our transgenic studies (PttGID1.1 and PttGID1.3) function as GA receptors. Over-expression of PttGID1 and AtGA20ox1 in aspen using two promoters, the strong 35S promoter or the weak but specific xylem LMX5 promoter, resulted in transgenic plants with phenotypic divergences from the wild-type. 35S::PttGID1 plants had similar traits to 35S::AtGA20ox1 plants, with increased elongation and secondary growth (Eriksson et al., 2000). However, 35S::PttGID1 plants did not show any increase in xylem fiber length, in contrast to 35S::AtGA20ox1 plants. Furthermore, neither LMX5::PttGID1 nor LMX5::AtGA20ox1 plants showed any obvious phenotypic changes from wild-type, except that, like 35S::AtGA20ox1 plants, LMX5::AtGA20ox1 plants had longer xylem fibers. The results of these studies of transgenic plants clearly suggest that GAs play two tissue-specific roles in the wood-forming zone, as enhancement of GA signaling appears to stimulate increases in xylogenesis in the cambium zone and fiber elongation in the developing xylem.

Experimental procedures

Plant material

Transformed A. thaliana (Columbia) were selected by growing them in vitro on 1/2 × MS medium with 50 μl ml−1 of kanamycin. After potting, the transformants and controls were grown in climate chambers under 16/8 h photoperiods with day/night temperatures of 22°C/18°C.

After transformation and selection, transgenic and wild-type hybrid aspen (P. tremula L. × tremuloides Michx. clone T89), at approximately 10 cm tall and with adequate root systems, were transferred from in vitro culture to soil in a greenhouse with 20 h light photoperiods, 60% minimum relative humidity, and a temperature of approximately 22°C. Two approximately 12-year-old aspens (P. tremula), growing in northern Sweden (64° 21′N, 19° 46′E), were also used.

Sampling for GID1 expression study

Vegetative tissues of two 3-month-old hybrid aspens, representing eight developmental states (the leaves and internodes at three levels, shoot apices, and the last 3 cm of the roots) were harvested. Longitudinal tangential cryosections (3 × 15 mm) were made across the wood-forming tissues of 12-year-old aspen trees as described by Uggla and Sundberg (2001) to collect a set of eight samples representing characteristic cell types within the cambial region, as described by Israelsson et al. (2005). All tissues were immediately frozen in liquid nitrogen and stored at −80°C.

RNA extraction and cDNA preparation

Internodes were ground in a mortar with liquid nitrogen, while the other samples (apices, leaves, roots and cryosections) were disrupted using a MM 301 vibration mill (Retsch GmbH at a frequency of 30 Hz for 2 min with tungsten carbide beads (Retsch GmbH, http://www.retsch.com/). RNA was extracted from the apices, leaves and internodes of both hybrid aspen and A. thaliana using an Aurum Total RNA Minikit (Bio-Rad, http://www.bio-rad.com/), and from cryosections and roots using a Plant RNA Isolation Aid and a RNAqueous-Micro Kit (Ambion, http://www.ambion.com/). A DNase treatment was applied using a DNA-free kit (Ambion). RNA (1 μg, or the maximum amount obtained from the cryosections) was used to prepare cDNA samples using an iScript cDNA synthesis kit (Bio-Rad).

Plasmid constructs

The open reading frames of PttGID1.1 and PttGID1.3 were amplified from cDNA of mature leaves using pfuTurbo Hotstart DNA polymerase (Stratagene, http://www.stratagene.com/) and primers 5′-CACCATGGCTGGAAGTAATGGAGTTAATCT-3′ and 5′-GTTAACCCTATTAACAGTTAGGACTCAC-3′ for the former, and 5′-CACCATGGCTGGTAGCAATGAAGTCA-3′ and 5′-GGTTAMGAAGTYTATTAACAGTTAGAATTCACRAAC-3′ for the latter. PCR products were cloned into pENTR/D-TOPO (Invitrogen, http://www.invitrogen.com/), and recombined using the Gateway system into pPCV812-LMX5-GW (Swe Tree Technologies, http://www.swetree.com/) and pK2GW7 (Karimi et al., 2002). The AtGA20ox1 gene in the vector pDONER (provided by Dr Sven Ericksson, Umeå Plant Science Centre, Sweden) was recombined into the same vectors. The constructs were transferred into Agrobacterium strain C58pMP90-RK.

Plant transformation

Arabidopsis thaliana was transformed by the floral dip method, as described by Clough and Bent (1998). Only eight of the PttGID1.1 transformants and seven of the PttGID1.3 transformants had no multiple insertions. Gene expression levels were measured in the T3 generation by quantitative RT-PCR using primers RT-PttGID1-F (5′-ACCCAATGTTTGGTGG-3′) and RT-PttGID1-R (5′-ATTACATGCTGGATGGTCCCT-3′). The ten plants of each line that expressed each construct most strongly were examined at the phenotypic level.

Hybrid aspen was transformed as described by Nilsson et al. (1992). We used the five lines for each of the constructs that most strongly expressed the respective transgenes as measured by quantitative RT-PCR using primers RT-PttGID1-F and RT-PttGID1-R for the PttGID1 genes, and RT-AtGA20ox1-F (5′-TGAGAGTGTTGGCTACGCAAGCAGTTTCAC-3′) and RT-AtGA20ox1-R (5′-GCTCATGTCGTCGCAAAACCGGAAAGAAAGG-3′) for AtGA20ox1. These lines were propagated in vitro, and three plantlets per line were retained for further study.

Growth measurements and sampling

The hybrid aspens were measured and their leaves counted 4 weeks after potting (time point 0), and every week for 6 weeks. The first internode was defined as the first occurring below a leaf at least 1 cm long.

After 10 and 11 weeks, for 35S::PttGID1 and LMX5::PttGID1 plants, respectively, internodes 8–10 and 55 were sampled and frozen in liquid nitrogen. The younger internodes were ground with liquid nitrogen, and RNA and GAs were extracted from them for analysis by quantitative RT-PCR and HPLC-GC/MS. The older internodes were used for anatomical characterization, fiber measurements, and separate measurements of GAs in phloem and xylem tissues (for the transgenic plants containing the LMX5 promoter).

Anatomical characterization

Tangential sections of internode 55 from three plants representing each of two lines carrying each transgenic construct and three wild-type plants were manually cut and observed under an Axioplan microscope (Zeiss, http://www.zeiss.com/). Images were taken, and the areas of pith and xylem were measured using axiovision 4.6 software (Zeiss).

Fiber measurements

Pieces of outer xylem (approximately 1 mm × 1 cm × 0.5 mm) from internode 55 of the plants used for the anatomical characterization were macerated to separate the wood cells as described by Franklin (1945). The cells were observed under an Axioplan 2 microscope (Zeiss), and length and diameter of 100 fibers per plant were measured using AxioVision 4.6 software.

Hormonal treatment in vitro

Sets of three hybrid aspen plants from a line over-expressing PttGID1.1 (82.10), a line over-expressing PttGID1.3 (87.4) and wild-type were placed in plastic jars (1 L) containing 100 ml 6 × MS medium, with 10 μm paclobutrazol or 10 μm GA4. Images of the plants were taken every week for 4 weeks using a Canon EOS D60 camera (Canon, http://www.canon.com/), and their lengths were measured using ImageJ software (http://rsb.info.nih.gov/ij/). This experiment was performed twice.

Quantitative GA analysis

GAs from tissue samples were purified and analyzed, essentially as described by Eriksson et al. (2000). Briefly, 50 mg samples(fresh weight) were extracted using 500 μl of MeOH:H2O:CH3COOH solvent (800:190:10 v/v/v) containing stable isotope internal standards (2H2 GAs purchased from Professor L. Mander, Australian National University, Canberra, Australia) using a MM 301 vibration mill at a frequency of 30 Hz for 3 min after adding 3 mm tungsten carbide beads (Retsch GmbH) to each tube to increase the extraction efficiency. After centrifugation for 10 min at 14 000 g, the supernatant was subjected to solid-phase extraction purification using 100 mg ISOLUTE C8(EC) cartridge (Sorbent AB, http://www.sorbent.se/). Then, after methylation with ethereal diazomethane, the samples were subjected to reverse-phase HPLC. The final analysis was performed by GC/MS-selected reaction monitoring using a JMS MStation 700 instrument (JEOL Inc., http://www.jeol.com/).

Real-time PCR analysis

Two quantitative RT-PCR techniques were applied: the SYBR Green fluorochrome technique, to quantify expression of the AtGA20ox1, PttGA20ox1 and PttGID1 family genes, and the TAQ-MAN technique using dual-labeled fluoregenic probes, to distinguish expression of the PttGID1 genes individually. For each case, the best reference gene was chosen using GeNorm (Vandesompele et al., 2002) as suggested previously by Gutierrez et al., (2008a,b). The methods are described in detail in Appendix S1. The expression data presented in the figures are the means of technical replicates (±SE) for the organs and cryosections of wood-forming tissue, and the means of two or three biological replicates for LMX5 and 35S transformants, respectively.

Phylogenetic analysis

The predicted protein sequences of the four GID1 sequences detected in aspen were aligned with known GID1 sequences from O. sativa and A. thaliana using the Clustal W program (Thompson et al., 1994). A phylogenetic tree was generated using the neighbor-joining method (Saitou and Nei, 1987) with 1000 bootstrap replicates.

Acknowledgements

We acknowledge Inga-Britt Carlsson for GA extraction and quantification, Kjell Olofsson for assistance in cryosectioning and anatomical characterization, Laurent Gutierrez for advice about quantitative RT-PCR and many comments on the manuscript, and SweTree Technologies for the LMX5 promoter. This work was financially supported by the Swedish Research Council, Formas, the Kempe foundation, Swedish University of Agricultural Sciences, FuncFiber (a FORMAS-funded center of excellence) and the Berzelii Center.

Ancillary