Proper gibberellin localization in vascular tissue is required to control auxin-dependent leaf development and bud outgrowth in hybrid aspen


(fax +46 90 786 8120; e-mail


Bioactive gibberellins (GAs) are involved in many developmental aspects in the life cycle of plants, acting either directly or through interaction with other hormones. One way to study the role of GA in specific mechanisms is to modify the levels of bioactive GA in specific tissues. We increased GA catabolism in different parts of the vascular tissue by overexpressing two different GA 2-oxidase genes that encode oxidases with affinity for C20- or C19-GA. We show that, irrespective of their localization in the vascular tissue, the expression of different members of this gene family leads to similar modifications in the primary and secondary growth of the stem of hybrid aspen. We also show that the precise localization of bioactive GA downregulation is important for the proper control of other developmental aspects, namely leaf shape and bud dormancy. Expression under the control of one of the studied promoters significantly affected both the shape of the leaves and the number of sylleptic branches. These phenotypic defects were correlated with alterations in the levels and repartitioning of auxins. We conclude that a precise localization of bioactive GA in the vasculature of the apex is necessary for the normal development of the plant through the effect of GAs on auxin transport.


Gibberellin biosynthesis

The gibberellins (GAs) are plant hormones that play important roles in regulating cell differentiation and elongation throughout the life cycles of plants, from germination to seed production, through primary and secondary growth, leaf expansion, flowering and fruit formation (Olszewski et al., 2002; Sun and Gubler, 2004). They are a large group of tetracyclic diterpenes, produced from geranylgeranyl diphosphate by a complex pathway involving three cell compartments and three classes of enzymes (Hedden and Phillips, 2000). The late biosynthesis pathway, located in the cytosol, requires two families of 2-oxoglutarate-dependent dioxygenases (2ODDs) to convert C20-GAs into C19-GAs, the immediate precursors and bioactive forms of this hormone: GA 20-oxidase (GA20ox) and GA 3-oxidase (GA3ox). The two main bioactive GAs are GA1 and GA4, and the overall GA response is a function of the accumulation of these molecules.

There are a number of different methods for functionally inactivating GAs: conjugation (Schneider and Schliemann, 1994), epoxidation (Zhu et al., 2006), methylation (Varbanova et al., 2007) and 2-oxidation. 2-oxidation is the most significant GA inactivation mechanism. The enzymes that catalyze this activity are encoded by a family of 2ODDs, the GA 2-oxidases (GA2ox), which are divided into two groups on the basis of their substrate specificity: one group is selective for C20-GAs (Schomburg et al., 2003), whereas the other is selective for C19-GAs (Hedden and Phillips, 2000; Thomas et al., 1999). C19-GA 2-oxidation is considered to be the primary inactivation pathway (Rieu et al., 2008). Single GA2ox mutants do not exhibit an obvious phenotype because of functional redundancy. Regulation of bioactive GA levels also involves feedback and feed-forward regulation of the expression of 2ODDs (Yamaguchi, 2008).

Role of GAs in primary and secondary growth

The most obvious phenotypic effect of GA relates to stem elongation (Koornneef and Van der Veen, 1980). Indeed, many GA mutants were originally isolated during screens for dwarf plants, with the height defect occurring as a result of fewer and shorter internodes. GAs are therefore involved in regulating the primary growth of plants. Hybrid aspen (Populus tremula×tremuloïdes) and tobacco (Nicotiana tabacum) strains that overexpress GA20ox or the GA receptor (GID1) are taller and have longer internodes than their wild-type counterparts (Biemelt et al., 2004; Eriksson et al., 2000; Mauriat and Moritz, 2009), but they are also wider (with increased xylogenesis), suggesting that GA also exerts control over secondary growth. The presence of GA signaling in a specific area of the wood-forming region of hybrid aspen appears to be important for inducing xylogenesis and elongation of the fibers (Mauriat and Moritz, 2009). These observations are in agreement with the hypothesis that there is a GA concentration gradient present across this region, with levels of bioactive GA reaching a peak in elongating xylem (Israelsson et al., 2005).

Role of GAs in apical dominance

Another characteristic of the GA biosynthesis mutants such as ga1.3 in Arabidopsis is their reduced apical dominance (Sun and Kamiya, 1994). Analysis of a perennial grass overexpressing GA catabolism genes indicated that GAs are also involved in the control of axillary bud outgrowth (Agharkar et al., 2007). However, the control of apical dominance by auxins and the way in which auxin control interacts with that exerted by cytokinins (CKs), and the newly characterized strigolactone (SL), have been studied in much more detail (Gomez-Roldan et al., 2008). These three hormones have been found to move systemically through the plant to control bud outgrowth (Leyser, 2009). The polar transport of auxin down the shoot from the apex plays a central role in the establishment and maintenance of apical dominance (Leyser, 2009; Ljung et al., 2001). This inhibits the release of auxin from the lateral buds (Prusinkiewicz et al., 2009). CK is produced in the roots and transported acropetally; its effects are known to counteract those of auxin (Bangerth, 1994; Sachs and Thimann, 1967). Strigolactone is also synthesized in the roots and transported acropetally, and has a negative effect on bud activation (Gomez-Roldan et al., 2008; Umehara et al., 2008). The relationship between auxin and CK in this process is well characterized: auxin downregulates CK biosynthesis, inhibiting the normally positive effect of CK on bud outgrowth (Li and Bangerth, 2003; Tanaka et al., 2006). However, the relationship between auxin and SL is not completely understood. It has been speculated that auxin stimulates SL biosynthesis, and that SL in turn modulates polar auxin transport to avoid bud outgrowth, either systemically or locally (Brewer et al., 2009; Crawford et al., 2010). The role of GAs in this network is not well characterized. It has been shown that GAs can improve auxin transport in the stem of the Alaska pea (Pisum sativum; Jacobs and Case, 1965; Scott et al., 1967). However, there have been no subsequent reports confirming the modality of GA action during apical dominance.

Role of GA in leaf development

Plants treated with GAs or overexpressing GA biosynthesis genes produce longer leaves than their wild-type counterparts; treatment with a GA inhibitor or overexpression of GA catabolism genes reduces the size of leaves (Ross et al., 1993). The growth capacity of the leaves is proportional to the level of GAs. In Arabidopsis transformed with a MEA::PsGA2ox2 construct (but not in 35S::GA2ox plants), the leaves were smaller and exhibited an altered phenotype: they were misshapen and had abnormal vasculature (Singh et al., 2009), which is characteristic of plants overexpressing the class-I KNOX genes or having altered auxin transport. The role of auxin and the KNOX genes in determining leaf vasculature patterning and shape has been discussed in a number of recent reviews (Braybrook and Kuhlemeier, 2010; Moon and Hake, 2011; Scarpella et al., 2010).

Although plant hormones are known to be responsible for controlling leaf developmental processes, such as shape definition and vasculature patterning, no role has been demonstrated previously for GAs. Hormone-induced regulation is part of a complex response network involving a number of factors (Weiss and Ori, 2007). The existence of crosstalk between auxin and GAs has been demonstrated in processes such as root elongation (Fu and Harberd, 2003), stem formation (Nemhauser et al., 2006) and fruit formation (de Jong et al., 2011). Auxin was shown to regulate the GA response by promoting degradation of DELLA in the roots (Fu and Harberd, 2003), and to stimulate the biosynthesis of GAs by upregulating the expression of GA20ox or GA3ox, or downregulating the expression of GA2ox (de Jong et al., 2011; Frigerio et al., 2006; Nemhauser et al., 2006; O’Neill and Ross, 2002). However, as with apical dominance, observations of the action of GAs on auxin transport (Bjorklund et al., 2007) during wood formation in hybrid aspen suggest the existence of another type of crosstalk between these two hormones.

This article describes an investigation into the effects of decreasing the levels of bioactive GAs in specific vasculature stages of hybrid aspen. Effects were found in different developmental mechanisms involving vascular tissue formation, from primary and secondary growth to apical dominance and leaf development. Moreover, evidence supporting the occurrence of crosstalk between GAs and auxin was found: GAs appear to control auxin transport during the apical dominance and leaf development processes.


Downregulation of GA level in the vasculature of hybrid aspen

To investigate the effects of reducing the levels of early GA precursors or bioactive GAs on different developmental stages of hybrid aspen, two genes involved in the 2-oxidation of GA at different steps in the GA biosynthesis pathway were expressed under the control of two promoters, leading to expression in different parts of the vasculature. The two GA2ox genes –AtGA2ox8 and AtGA2ox2– were chosen because the enzymes that they encode catabolize early precursors of GA (C20-GA) or late precursors and bioactive GAs (C19-GA), respectively (Schomburg et al., 2003; Thomas et al.,1999) (Figure S1). The promoters used induce gene expression in the vasculature (Bjorklund, 2007): LMP1 induces higher expression on the phloem side of the cambial zone in the stem, and LMX5 induces expression on the xylem side of the cambial zone.

Not all of the attempted combinations of the two GA2ox genes with the two promoters resulted in viable transgenic plants. No plants were regenerated from the LMP1::AtGA2ox8 transformation, and those containing the LMX5::AtGA2ox2 construct failed to take root in vitro. However, a number of plants were regenerated using the LMP1::AtGA2ox2 and LMX5::AtGA2ox8 combinations. It is possible that the localized decrease in GA levels and the blocking of the biosynthetic pathway are both important for full plant development.

Quantitative PCR (qPCR) analyses were performed on different organs from the LMP1::AtGA2ox2 and LMX5::AtGA2ox8 plants to determine where the two promoters induced the expression of the transgenes (Figure 1). As expected, the transgene was most highly expressed in the stems of the LMP1::AtGA2ox2 plants, especially in the phloem of old internodes. However, it was also present in the apex and young leaves, possibly in their vasculature (Figure 1a). AtGA2ox8 induced by LMX5 was most highly expressed in the stem, especially in the xylem of older internodes. AtGA2ox8 was also found to be expressed in old leaves, possibly in their mature vasculature (Figure 1b). Interestingly, both transformations induced expression of the transgenes in the apices: LMP1::AtGA2ox2 was almost as highly expressed in the apex as in the phloem, whereas LMX5::AtGA2ox8 exhibited a very low level of expression in the apex compared with that in the xylem (almost 1000-fold lower) (Figure 1b).

Figure 1.

 Expression levels of the transgenes in LMP1::AtGA2ox2 and LMX5::AtGA2ox8 plants.
(a) AtGA2ox2 expression in different organs of LMP1::AtGA2ox2 plants. (b) AtGA2ox8 expression in different organs of LMX5::AtGA2ox8 plants. YL, leaves smaller than 1 cm in length; L10 and L30, leaves 10 and 30, respectively, counted from the apex; IN10, internode below L10; P and X, phloem and xylem of the internode below L30. qPCR normalized against the At4g34270 ortholog of hybrid aspen (identified as the most appropriate reference gene for these conditions using GeNorm), and relative to the value obtained in the organ where the gene is the most highly expressed (set at 1). Values are means ± SEs of three biological replicates.

LMP1::AtGA2ox2 and LMX5::AtGA2ox8 transformations have similar effects on primary growth

The LMP1::AtGA2ox2 and LMX5::AtGA2ox8 plants exhibited a lower growth rate than the wild-type plants (Figure 2a–d). As the presence of GAs is necessary to induce elongation of the stem, this may result from the expression of the transgenes in the apex, which could potentially lead to a decrease in bioactive GAs in this tissue. GA levels were quantified in the apices of the two different GA2ox overexpressors, and in the wild-type plants (Figure 2g,h): the C19-GAs from the early and non-early hydroxylated pathways (Figure S1) were between two- and threefold less abundant in the two transgenics. This is probably because of increased GA catabolism, which is induced at an early stage in the biosynthetic pathway by AtGA2ox8 expression, or at a later stage by AtGA2ox2. In either case, elevated catabolism causes reductions in the levels of GAs, so both constructs have similar effects on the growth of the plants. The semi-dwarf phenotypes of the LMP1::AtGA2ox2 and the LMX5::AtGA2ox8 plants were both correlated with the presence of fewer and shorter internodes (Figure 2e,f).

Figure 2.

 Characterization of the phenotypes and growth of the hybrid aspens LMP1::AtGA2ox2 and LMX5::AtGA2ox8.
Phenotypes of 3-month-old plants grown in the glasshouse: (a) three LMP1::AtGA2ox2 plants (line 72.8) with a wild-type plant to the left; (c) three LMX5::AtGA2ox8 plants (line 68.17) with a wild-type plant to the left. Cumulative shoot elongation of plants grown in the glasshouse: (b) wild-type hybrid aspen (T89) and three LMP1::AtGA2ox2 lines; (d) three LMX5::AtGA2ox8 lines. Time point 0 corresponds to 4-week-old plants. The average number (e) and length (f) of the internodes of the transgenic hybrid aspens are also shown; values are means ± SD of three lines. GA levels in the apex: (g) non-early 13-hydroxylation pathway; (h) early hydroxylation pathway. GA values are means ± SD of two wild-type and six transgenic replicates. ND, not determined.

The secondary growth of trees expressing LMP1::AtGA2ox2 or LMX5::AtGA2ox8 is similarly modified

The global stem shape was modified in both types of transgenic trees when compared with the wild-type plants. The stems of the transgenics were less tapered than those of the wild-type plants: both groups had similar stem diameters in the apical region (at internode 10), but the stems of the transgenics were appreciably narrower at lower levels (internodes 30 and 50; see Figure 3a). This indicates that the secondary growth of the plants was affected by the transgenic modifications.

Figure 3.

 Characterization of the secondary growth of the hybrid aspens LMP1::AtGA2ox2 and LMX5::AtGA2ox8.
(a) Stem diameter at internodes 10, 30 and 50 counted from the top of the plant. (b) Ratio of phloem bark area/xylem area of internode 50. GA content in phloem (c, e) and xylem (d, f) of internode 50. (c, d) Non-early 13-hydroxylation pathway. (e, f) Early 13-hydroxylation pathway. GA values are means ± SDs for two wild types and five transgenic replicates. ND, not determined.

To investigate this modification of secondary growth, cross sections were taken from each plant at internode 50, and the areas of the xylem and bark phloem were estimated (Figure 3b). The ratio of bark phloem to xylem was approximately one-third higher in the LMP1::AtGA2ox2 and LMX5::AtGA2ox8 sections: proportionally, the transgenics had a wider phloem and less xylem than the wild-type plants.

To determine whether there was any connection between GA levels and xylogenesis, the GA levels were measured in the phloem and xylem of the samples taken at internode 50 of the LMP1::AtGA2ox2, LMX5::AtGA2ox8 and wild-type plants (Figure 3c–f). In the wild-type samples, the results were reasonably consistent with the published data on GA levels in cryosections from wood-forming regions (Israelsson et al., 2005). C19 precursors (GA9 and GA20) (Figure S1) were more abundant in the phloem than in the xylem, whereas the levels of deactivated GAs (GA8 and GA34) were similar in both samples, and bioactive GAs were more abundant in the xylem than in the phloem (significantly for GA1 but not for GA4).

The LMP1::AtGA2ox2 and LMX5::AtGA2ox8 samples contained lower quantities of all GAs (C20 and C19) – late precursor, bioactive and deactivated – than the wild-type samples. In the phloem sample from the LMP1::AtGA2ox2 plants, the levels of the late GA precursors (C19 and C20) and bioactive GAs were very low, but levels of deactivated GAs (GA8 and GA34), although lower than in the wild-type sample, were not as greatly reduced as the bioactive GAs. These findings are consistent with an increase in AtGA2ox2 activity in these plants. In the xylem samples, GAs were present in only trace amounts, explaining the low rate of xylogenesis in these transgenic plants.

LMP1::AtGA2ox2, but not LMX5::AtGA2ox8, causes altered branching

Although the two constructs reduced the rate of growth, their semi-dwarfism was not associated with identical phenotypes. Whereas LMX5::AtGA2ox8 plants resembled short, wild-type plants (Figure 2c), LMP1::AtGA2ox2 plants differed in the size and shape of their leaves, and in the presence of sylleptic branches (branches growing outwards during the first year of plant growth) (Figure 2a). In order to highlight the differences in the forms of the branches, the apical part of a wild-type stem and a LMP1::AtGA2ox2 stem are shown after defoliation (Figure 4a). The lateral buds were never dormant and began to develop in the apex, in contrast to the wild-type and LMX5::AtGA2ox8 plants, in which the buds were very small (Figure 4b,c). Subsequently these LMP1::AtGA2ox2 plants showed a much higher percentage of lateral buds producing a branch than did the wild-type and the other transgenic plants (Figure 4d).

Figure 4.

 Characterization of the shoot branching.
Characteristic stems of a wild-type plant (a1) and a LMP1::AtGA2ox2 plant (a2) after defoliation. Dissected apices of a wild type (b) and a LMP1::AtGA2ox2 plant (c). (d) Percentage of lateral buds producing a branch. Values are means ± SDs of three lines. (e, f) Measurement of IAA transport in the transgenic hybrid aspens. (e) Endogenous IAA and (f) [13C]IAA contents in wild-type hybrid aspen, LMP1::AtGA2ox2 and LMX5::AtGA2ox8 plants after decapitation and treatment with [13C]IAA for 3 days. Values are means ± SDs of a minimum of three plants per condition.

Changes in apical dominance are usually correlated with modified indole-3-acetic acid (IAA) levels or transport in the plant. To determine if this was the case here, the level of endogenous IAA and the movement of exogenous IAA were measured in the stems (Figure 4e,f). IAA levels and transport in the LMP1::AtGA2ox2 plants differed substantially from those in the wild type and LMX5::AtGA2ox8. Specifically, the IAA levels at internode 10 of the LMP1::AtGA2ox2 plants was half that at the corresponding internode of the wild-type plants and the other dwarf plants (Figure 4e). To evaluate the IAA transport capacity of the three genotypes, the plants were decapitated and lanolin blocks with or without 13C-labeled IAA were attached to the apex. In the semi-dwarf branching plants, the levels of 13C-labeled IAA were half those in the wild-type and LMX5::AtGA2ox8 plants (Figure 4f). The endogenous IAA levels and IAA transport capacity seemed to be affected significantly in LMP1::AtGA2ox2 plants, but not in the other transgenic hybrid aspen.

Leaf development is affected in hybrid aspen expressing LMP1::AtGA2ox2, but not in plants expressing LMX5::AtGA2ox8

Both transgenic trees exhibited significantly different leaf phenotypes. The leaves of the LMP1::AtGA2ox2 plants were half the size of those of the wild type and the other transgenic trees (Figure 5a,b). GAs are known to be involved in leaf growth, and correlations between the levels of bioactive GA and leaf size have been noted in different species (Eriksson et al., 2000; Ross et al., 1993). In this study, although both transgenics expressed a GA2ox gene, only the leaves of LMP1::AtGA2ox2 plants were smaller. However, AtGA2ox8 under the control of the LMX5 promoter was found to be expressed at a low level in the young leaves (Figure 1b), indicating that the level of bioactive GA in the developing leaves of this transgenic were probably not affected significantly. Conversely, AtGA2ox2 under the control of the LMP1 promoter was highly expressed in young leaves. It therefore appears that the size of the leaves is influenced by the quantity of bioactive GA present in the leaves.

Figure 5.

 Characterization of leaves of LMP1::AtGA2ox2 and LMX5::AtGA2ox8 hybrid aspen.
(a) Area of leaves 6 and 10 (counted from the apex). Values are means ± SDs of three lines. (b) Pictures of characteristic leaves 6 and 10 of a wild-type hybrid aspen, a LMP1::AtGA2ox2 plant and a LMX5::AtGA2ox8 plant. Scale bar: 1 cm.
(c) Tip of young leaves (smaller than 1 cm in length) of a wild-type hybrid aspen, a LMP1::AtGA2ox2 plant and a LMX5::AtGA2ox8 plant after clearing in chloral hydrate for 1 week.
(d) Ratio of auxin level in the leaves: tip/base (next to the petiole). (Student’s t-test between T89 and LMP1::AtGA2ox2: P = 0.054.)
(e, f) Relative expression levels of PtPIN1 and ARK2 during the development of the leaves (YL, leaves smaller than 1 cm in length; L10, leaves number 10 counted from the apex). qPCR normalized against the At4g34270 ortholog (selected with GeNorm as the most appropriate reference gene in these conditions), and relative to the value of the wild type in the apex (set at 1). Values are means ± SEs of three biological replicates. (Student’s t-tests performed on YL samples: T89 versus LMP1::AtGA2ox2, P = 0.037; T89 versus LMX5::AtGA2ox8, P = 0.0515.)

We also found that in addition to being smaller than the wild type, the leaves of LMP1::AtGA2ox2 had a modified shape (Figure 5b): the leaves were shorter with a more rounded extremity (being less pointed than the wild-type leaves), and differed in the extent of their vasculature (Figure 5c). The shape and vasculature of leaves results from the coordinated regulation of many factors, including transcription factors (such as the KNOX genes) and hormones (Moon and Hake, 2011). To better understand the cause of the changed leaf phenotypes, we analyzed auxin levels and the expression of the auxin transporter (PttPIN1) and KNOX genes.

The level of auxin in the tip and the base (next to the petiole) of the leaves was measured (Figure 5d). The ratio of the auxin concentrations at the tips and bases of the leaves was similar in wild-type and LMX5::AtGA2ox8 plants, but was lower in the LMP1::AtGA2ox2 plant, suggesting that the spatial partitioning of auxin was altered in these leaves. PttPIN1 is an auxin transporter known to be involved in leaf development. Its expression in the apex, young leaves and elongated leaves was monitored by qPCR analysis (Figure 5e): its expression pattern was found to be unaltered (relative to the wild type) in the LMX5::AtGA2ox8 plants, but was decreased in the apex sample (which contains the leaf primordia) of the LMP1::AtGA2ox2 plants (Figure 5e).

Studies in other species have established that polar auxin transport also negatively regulates the KNOX genes, so we monitored the expression level of ARBORKNOX2 (ARK2, the hybrid aspen BP ortholog; Du et al., 2009). ARK2 was found to be upregulated in the young leaves of both of the GA2ox transgenic plants (Figure 5f).


To gain new insights into the role of GAs, we investigated hybrid aspen plants with increased rates of 2-oxidation of GA in specific parts of the vascular tissue. We demonstrated that changes in the expression patterns of specific GA2ox genes can have a significant effect on the phenotype of the transgenic trees, and that this, to some extent, is the result of differences in their auxin response.

Increased GA catabolism in the vascular tissues reduces primary and secondary growth

We planned to investigate transgenic plants transformed with different combinations of four constructs, consisting of two promoters, LMP1 and LMX5 (which are known to promote expression specifically in different parts of the vascular tissue), paired with the AtGA2ox2 or AtGA2ox8 genes. However, we were unable to obtain viable plants from two of these combinations, and therefore focused on the effects of the two viable combinations. Although the two promoters were both shown to be expressed in the vasculature of young tissues, the levels of expression differed: LMP1 induced high expression in the apex, whereas LMX5 induced a much lower level of expression (Figure 1). It is possible that this variable expression is the reason why the LMP1::AtGA2ox8 and LMX5::AtGA2ox2 constructs did not produce viable plants. LMP1::AtGA2ox8 expression may have significantly decreased GA levels by promoting the degradation of early intermediates in GA biosynthesis; the resulting GA deficiency in their young tissues may have prevented the plants from regenerating. In contrast, the LMX5 promoter induced lower levels of AtGA2ox2 expression in the young tissue, allowing the growth of some plants. However, these plants did not survive further because of their inability to produce roots. The presence of bioactive GAs in the xylem seems to be crucial for adventitious rooting. Although a connection between GA levels and KNOX genes has been established (Bolduc and Hake, 2009), Ikezaki and co-authors have shown that KNOX genes are involved in the regulation of adventitious rooting in Arabidopsis, and that in this species, this role is independent of the GA pathway (Ikezaki et al., 2009).

The plants obtained with the two other constructs, LMP1::AtGA2ox2 and LMX5::AtGA2ox8, exhibited a decreased growth rate relative to the wild-type plants (Figure 2a–d). The dwarfism obtained with GA2ox of different subgroups (AtGA2ox2 uses the C19-GAs, and AtGA2ox8 modifies the C20-GAs; Figure S1) is consistent with that observed in other species. Indeed, overexpression of AtGA2ox8 or AtGA2ox7 in Arabidopsis also causes dwarfism (Schomburg et al., 2003); in hybrid aspen (Populus tremula × Populus alba), dwarf and semi-dwarf trees were obtained by overexpression (by activation tagging) of PtaGA2ox1, a GA2ox from the same subfamily as AtGA2ox2 (Busov et al., 2003).

Hybrid aspens with activation-tagged PtaGA2ox1 have shorter, but equivalent numbers of internodes when compared with their wild-type counterparts (Busov et al., 2003). The LMP1::AtGA2ox2 and LMX5::AtGA2ox8 plants had shorter and fewer internodes (Figure 2e,f). Compared with wild-type trees, hybrid aspens overexpressing genes involved in GA biosynthesis or a GA receptor have similar numbers of internodes, but their internodes are longer than in the wild type (Eriksson et al., 2000; Mauriat and Moritz, 2009). By using specific promoters, one obtains a decreased concentration of GA in the apex (Figure 2g,h) rather than a more widespread effect. This leads not only to a change in elongation but also in cell division, causing slower development. GA is also known to be involved in cell division, for example in the control of root cell proliferation (Achard et al., 2009; Ubeda-Tomas et al., 2009).

The slower primary growth is also followed by an obvious reduction in the extent of secondary growth. The global shape of the stems provides information about the nature of the reduction. As observed in the activation-tagged PtaGA2ox1 hybrid aspen (Busov et al., 2003), the LMP1::AtGA2ox2 and LMX5::AtGA2ox8 plants presented stems that were less tapered than in the wild type (Figure 3a). The young internodes were of the same width as the corresponding wild-type internodes, whereas the older internodes were narrower. The narrow internodes appeared to result from a lower level of xylogenesis in these plants (Figure 3b). Similar observations were made in studies on tobacco overexpressing one AtGA2ox and rapeseed overexpressing AtGA2ox8, where the dwarf plants were shown to have decreased xylogenesis and lignification (Biemelt et al., 2004; Zhao et al., 2010). However, hybrid aspen and tobacco transformed with 35S::AtGA20ox1 was observed to have more extensive xylogenesis (Biemelt et al., 2004; Eriksson et al., 2000). In hybrid aspen, an increase in the production of bioactive GAs appears to be the cause of the increased xylogenesis (Eriksson et al., 2000).

In conclusion, our results indicate that the levels of bioactive GAs are important for controlling xylogenesis in hybrid aspen. Decreases in the levels of GA precursors, or of bioactive GAs in the xylem (Figure 3c,d), reduce secondary growth.

Hybrid aspen defective in GA in early vasculature produces sylleptic branches

In many tropical species, sylleptic branching appears to be correlated with high growth rates in the main shoot. However, in a study of three hybrid poplars with different growth rates and sylleptic branching aptitudes, Cline and Dong (2002) found that the smallest hybrid exhibited the highest degree of sylleptism. In our study, we also observed that the semi-dwarf LMP1::AtGA2ox2 plants exhibited a higher degree of sylleptism than the wild-type hybrid aspen (Figure 4d). Other semi-dwarf plants (LMX5::AtGA2ox8 plants) did not exhibit any branching (Figure 4d). The presence of sylleptic branches corresponds to a loss of apical dominance earlier than might normally be expected (after the first winter). Among the hybrid poplars studied by Cline and Dong (2002), the shortest tree with the highest loss of apical dominance also exhibited the weakest response to exogenous auxin applied to the apex to inhibit bud outgrowth. Similarly, we observed lower levels of endogenous IAA and decreased IAA transport in the semi-dwarf branching trees than in the semi-dwarf non-branching trees (Figure 4e,f).

A number of early studies on apical dominance in non-woody species showed that GA amplifies auxin action (Jacobs and Case, 1965; Scott et al., 1967). For example, in studies involving the application of lanolin containing auxins or a mixture of auxins and GAs to decapitated pea seedlings, auxin was transported further down the stem in the presence of GA (Scott et al., 1967), suggesting that GA enhances auxin transport. Such an increase of auxin transport resulting from the presence of GA has also previously been hypothesized by Bjorklund et al. (2007). In our study, the decreased GA levels in the LMP1::AtGA2ox2 plants could explain the loss of apical dominance. However, the LMX5::AtGA2ox8 plants also exhibited a significant decrease in bioactive GA, but did not lose their apical dominance (Figure 4). The basis of this phenotype may relate to the specific type of GA2ox or the precise localization of the expression of the transgene. The development of lateral buds in the apex (Figure 4b,c) suggests that the altered GA levels affected IAA biosynthesis or transport in this young tissue. The levels of bioactive GA were low in the apices of the two types of transgenic plants (Figure 4e,f). However, even allowing for low GA levels in a specific region of the apex of the LMX5::AtGA2ox8 plants, this does not mean that low levels of GAs exist in all parts of the apex. As the measurements were performed on complete apices, rather than on the different tissues of a dissected apex, the concentration of GA in specific apex tissue layers could not be assessed. It was possible to monitor the patterns of expression from the promoters used in this study: the LMX5 promoter promotes gene expression in mature vasculature, and at a much lower level in young tissue, whereas the LMP1 promoter primarily induces expression in the apex and young tissue (almost at the same level than in the phloem) (Figure 1). Thus, the localization of the transgene expression directed by the promoters had a greater influence on the effects on auxin than the type of GA2ox being expressed.

Precisely localized downregulation of GA levels also affects leaf development

Leaf size was reduced in the LMP1::AtGA2ox2 plants. The LMP1 promoter induces a high level of gene expression in young leaves, but the LMX5 promoter induces only a very low level. Therefore, GA levels would be expected to be lower in the young leaves of the LMP1::AtGA2ox2 plant. Similar findings were obtained in studies on hybrid aspen overexpressing genes involved in GA biosynthesis, which had larger leaves that exhibited higher levels of bioactive GA (Eriksson et al., 2000). A correlation between the level of GA and the size of the leaves has also been shown in Arabidopsis (Ross et al., 1993).

The two types of transgenic hybrid aspen also exhibited different leaf shapes. The leaf shape and its vasculature are defined very early in leaf development, in the leaf primordia, before the leaves attain 10% of their final size (Efroni et al., 2010). Thus, the spatial repartitioning of GA in the leaf primordia is important in the regulation of factors that determine the shape of the leaves (Hay et al., 2002). To our knowledge, there have been no reports of modifications of leaf morphology as a result of changes in the GA content of a whole plant. However, altered leaf shapes were observed in Arabidopsis transformed with the tissue-specific promoter construct MEA::PsGA2ox2 (Singh et al., 2009). Gene expression induced by this promoter is thought to be localized in the shoot apical meristem (SAM) and its general vicinity: it is not homogenously distributed throughout the apex. In our study, the LMP1 promoter induced gene expression in the very young vasculature (probably in the protophloem), resulting in a modified leaf form, whereas expression of a GA2ox gene under the control of the LMX5 promoter (in the xylem) did not affect leaf shape (Figure 5b). Xylem is formed at a later stage in the development of the vasculature than the (proto)phloem, and it is only partially developed in the leaf primordia. In the leaf primordia, when the main vein starts to be established, a localized decrease of bioactive GA (in the LMP1::AtGA2ox2 plants) in the young vasculature influences the future shape and vasculature of the leaf. Interestingly, however, this effect has not been noted in hybrid aspen overexpressing the same gene under the control of a 35S promoter (Busov et al., 2003). It is possible that the hormonal balance between the different parts of the primordia (as in the SAM) controls their development. In the LMX5::AtGA2ox8 plants, the decrease of GA levels occurs at a late stage and is unlikely to have a significant influence on those parts of leaf development that might affect the global form of the leaf vasculature.

An important element in leaf shape and vasculature patterning is the KNOX gene BP. The BP gene is expressed in the SAM but not in the leaves of simple leaf species such as Arabidopsis (Lincoln et al., 1994). However, expression has been noted in the leaves of a compound leaf species (Tomato) (Hareven et al., 1996). It has been shown that overexpression of this gene induces altered leaf shape in many plants, and that this only needs to occur in a specific region at a very early stage of leaf development in order to produce the effect (Kano-Murakami et al., 1993). BP and its orthologs are known to repress GA biosynthesis genes and activate GA 2-oxidation genes (Hay et al., 2002; Kusaba et al., 1998a,b). GAs are also known to regulate BP accumulation. For example in the MEA::PsGA2ox2 plants, a significant increase in expression of the class I-KNOX genes was reported (Singh et al., 2009). We observed an increase in ARK2 (hybrid aspen BP ortholog) expression of about 10-fold in the young leaves of LMP1:AtGA2ox2 compared with wild-type plants with normal leaves. However, a general increase in ARK2 expression in young leaves does not appear to directly influence leaf shape, as the gene is also upregulated in the normal-shaped leaves of LMX5::AtGA2ox8 trees (Figure 5f). It is therefore likely that upregulation of ARK2 only results in a modified leaf shape if it occurs very early in leaf development, as is the case in the LMP1:AtGA2ox2 plants. Increased expression resulting from altered GA levels occurring later on in leaf development, as observed in LMX5::AtGA2ox8 plants, does not seem to have any effect on the simple leaves of hybrid aspen.

Because auxin and its polar transport have also been reported to regulate vasculature patterning and leaf shape (Scarpella et al., 2010), we investigated whether IAA levels or transport were altered in the plants in our study. In the apex samples (containing the leaf primordia) of the LMP1::AtGA2ox2 plants the expression of PttPIN1 was significantly downregulated in comparison with the wild-type plants, and the repartitioning of auxin was found to be altered in the young leaves (Figure 5e,d). In the other transgenic hybrid aspen being tested, no significant modification of PttPIN1 expression or of auxin repartition was observed. The auxin transport therefore seems to be affected by the decreased GA levels in the early vasculature of the leaf primordia. An effect of GA on the PIN expression has also been suggested by Gou et al. (2010), but in a different developmental process: lateral root formation.

In conclusion, our results have improved our understanding of the effects of GA expression in hybrid aspen in two key areas. First, we have provided evidence that both primary and secondary growth are affected by an increase in GA catabolism, using a GA2ox gene encoding a member of one or the other GA2ox subfamily. Secondly, we showed that the precise localization of bioactive GA in the vasculature of the apex is necessary for normal plant development because of its effects on auxin transport.

Experimental Procedures

Plant material

The transgenic and wild-type hybrid aspens (P. tremula L. tremuloïdes Michx. clone T89) were transferred to soil after rooting in vitro. They were grown in a glasshouse with 18-h light photoperiods, 60% relative humidity and a temperature of 20°C.

Plasmid constructs

The open reading frames of AtGA2ox2 and AtGA2ox8 were amplified from cDNA of Arabidopsis thaliana Col-0 rosette leaves using pfuTurbo Hotstart DNA polymerase (Stratagene, and the primers described in Table S1. The products were cloned into pENTR/D-TOPO (Invitrogen,, and then recombined into pPCV812-LMP1-GW and pPCV812-LMX5-GW (Swe Tree Technologies, with the Gateway system. The four constructs were transformed into Agrobacterium strain C58PMP90-RK.

Plant transformation

Hybrid aspen were transformed with the four constructs following the protocol described in Nilsson et al. (1992). The three lines showing the highest expression of the respective transgene were selected by qPCR with the primers AtGA2ox2q and AtGA2ox8q from Rieu et al. (2008). Six plants per line were amplified in vitro and transferred to soil. A second independent growth trial containing five biological replicates was performed the following year.

Growth measurements and sampling

The plants in the first independent growth trial were measured every week from the fourth week after potting (time 0) for a period of 6 weeks. The number of internodes was counted, with the first internode (1) being defined as that below the first leaf of at least 1 cm length in the apex. The plants were sampled after 10 weeks in the glasshouse. For the gene expression analysis, the shoot apices, leaves and cross-sectional samples from internodes 1, 10 and 30 from one plant per line were harvested, frozen immediately in liquid nitrogen and kept at −80°C. These lines were considered to be biological replicates. For the cross-sectional samples from internode 30, the bark peel and the developing xylem scraped from the exposed wood surface were used as phloem and xylem samples, respectively. The same phloem and xylem samples were used for the analysis of gibberellins. Apices from two plants of each line were harvested and frozen immediately for GA analysis. Cross-sectional samples from internode 50 of each line were harvested and frozen for sectioning and measurement of phloem and xylem areas.

Trees grown in the second year of the study were sampled after 6 weeks in the glasshouse. For anatomical characterization, leaves 6 and 10 were sampled and scanned directly. Small disc areas were cut from the tip and the top of leaves 1, 2 and 3, frozen immediately in liquid nitrogen and stored at −80°C prior to IAA measurement.

RNA extraction and cDNA synthesis

All of the samples bar those from internode 30 were disrupted for 2 min using a MM301 vibration mill with tungsten carbide beads (Retsch GmbH, at a frequency of 25 Hz. The phloem and xylem of the internode-30 samples were separated and ground with a mortar after freezing with liquid nitrogen. RNA was extracted using the Aurum Total RNA Minikit (Bio-Rad,, and treated with a DNA-free kit (Ambion, A 1-μg sample of the extracted RNA was used for reverse transcription with an iScript cDNA synthesis kit (Bio-Rad). The cDNA was diluted 20-fold prior to qPCR analysis.

Quantitative PCR analysis

Quantitative PCR analysis (qPCR) was performed with the SYBR-GREEN fluorochrome. Primers specific to AtGA2ox2, AtGA2ox8 and PttPIN1 were based on those described previously (Rieu et al., 2008; Bjorklund et al., 2007). The primers used for ARK2 were RT-PttBP-F (5′-ATTGATCCCCGTGCTGAAGACC-3′) and RT-PttBP-R (5′-GCCACCTTCTCCGTCTCTGAA-3′), which amplified an amplicon of 196 bp with an annealing temperature of 60°C. The primers used for the reference genes were as described by Mauriat and Moritz (2009). The most appropriate reference gene for the study was selected using GeNorm (Vandesompele et al., 2002). The MIQE précis is accessible in Appendix S1.

Measurements of IAA transport

In the first growth trial, the IAA transport measurements were performed as described by Bjorklund et al. (2007). The plants were decapitated at internode 9. Three days after decapitation, a 1.5-cm length of stem was harvested 1 cm below the cut. For each line, the experiment was performed on one plant control, one untreated decapitated plant and two decapitated plants treated with 13C6-IAA. The samples were frozen immediately in liquid nitrogen and kept at −80°C.

Quantitative IAA measurement

The IAA measurements were performed on stem samples from the IAA transport experiment, and on the leaf discs. The analysis involved extraction of plant material in 80% MeOH (1% HOAc), with [13C6]IAA and [2H3]IAA as internal standards for the leaf disc and IAA transport experiments respectively, followed by SPE-purification (100 mg C8 cartridge; IsoElute; Sorbent AB, The carboxylic acid group of IAA was chemically modified with bromocholine according to the method described by Kojima et al. (2009), and the modified IAA was analyzed on an Agilent 6460 triple quadrupole mass spectrometer in multiple-reaction-monitoring (MRM) mode (see Appendix S2).

Quantitative GA measurement

The GAs were extracted, purified and analyzed from the apex, and from 50 mg of the ground phloem and xylem samples. The samples were extracted using 1 ml of 80% MeOH containing stable isotope internal standards (2H2 GAs purchased from Professor L. Mander, Australian National University, Canberra, Australia) using a MM301 vibration mill (Retsch GmbH) at a frequency of 30 Hz for 3 min with tungsten carbide beads. After centrifugation for 10 min at 14 000 g, the supernatant was subjected to two successive solid-phase extraction purifications: 100-mg ISOLUTE C18 (EC) cartridge (Sorbent AB) and 150-mg Oasis MCX cartridge (Waters, For the phloem and xylem samples an additional purification was performed using a 100-mg ISOLUTE SI cartridge (Sorbent AB). The carboxylic acid group of the GAs was modified with bromocholine according to Kojima et al. (2009) and analyzed on an Agilent 6460 triple quadrupole mass spectrometer in MRM mode (see Appendix S2).

Anatomical characterization

Anatomical characterization (measurement of the bark phloem and xylem area) was performed on internode 50 as described previously (Mauriat and Moritz, 2009).

The vasculature of young leaves was observed on small leaves (<1 cm in length) after 1 week in a clearing solution of chloral hydrate.

Statistical analysis

All the quantitative data have been statistically analyzed by a Bonferroni-corrected Student’s t-test: *P ≤ 0.05.


We acknowledge Inga-Britt Carlsson for GA and auxin extraction and quantification, Dr Yohann Boutté for comments on the article, and SweTree Technologies for the LMP1 and LMX5 promoters. This work was financially supported by the Swedish Research Council, Formas, the Kempe Foundation, Knut & Alice Wallenberg Foundation, Swedish University of Agricultural Sciences, FuncFiber (a FORMAS-funded center of excellence) and the UPSC Berzelii Center of Forest Biotechnology (funded by VR and Vinnova).