Tissue-specific localization of gibberellins and expression of gibberellin-biosynthetic and signaling genes in wood-forming tissues in aspen


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


Bioactive gibberellins (GAs) are known regulators of shoot growth and development in plants. In an attempt to identify where GAs are formed, we have analyzed the expression patterns of six GA biosynthesis genes and two genes with predicted roles in GA signaling and responses in relation to measured levels of GAs. The analysis was based on tangential sections, giving tissue-specific resolution across the cambial region of aspen trees (Populus tremula). Gibberellin quantification by GC/MS-SRM showed that the bioactive GA1 and GA4 were predominantly located in the zone of expansion of xylem cells. Based on co-localization of the expression of the late GA biosynthesis gene GA 20-oxidase 1 and bioactive GAs, we suggest that de novo GA biosynthesis occurs in the expanding xylem. However, expression levels of the first committed GA biosynthesis enzyme, ent-copalyl diphosphate synthase, were high in the phloem, suggesting that a GA precursor(s) may be transported to the xylem. The expression of the GA signaling and response genes DELLA-like1 and GIP-like1 coincided well with sites of high bioactive GA levels. We therefore suggest that the main role of GA during wood formation is to regulate early stages of xylem differentiation, including cell elongation.


The formation of secondary xylem (wood) and phloem is initiated in the vascular cambium. The cambial initials give rise to xylem and phloem mother cells, which differentiate by expansion and maturation into the different cell types found in wood and bark. During wood development, the phases of division, expansion and maturation are separated in space and can therefore be recognized as distinct zones (Larson, 1994; Figure 1). Auxin is required for cambial growth and its concentration gradient across the cambial tissues has been suggested to provide positional information in wood development (Tuominen et al., 1997; Uggla et al., 1996, 1998, 2001). Gibberellins (GAs) act synergistically with auxin in stimulating cambial growth. Moreover, both analyses of transgenic trees overproducing GAs and application experiments have shown that GAs stimulate xylem fiber elongation (Digby and Wareing, 1966; Wareing, 1958; Little and Pharis 1995; Eriksson et al., 2000).

Figure 1.

Schematic overview of the cambial region including different types of phloem, cambium and xylem cells.
Not included in the figure are the ray cells.
(A–G) Approximate sample positions from where sections were pooled for expression analysis.

In recent years our understanding of GA biosynthesis and signaling has increased substantially through the identification of many genes involved in these processes. The biosynthesis of GAs can be divided into three stages: (i) the formation of ent-kaurene; (ii) the conversion of ent-kaurene to GA12; and (iii) the formation and deactivation of bioactive GAs (Figure 2; Hedden and Phillips, 2000; Olszewski et al., 2002; Yamaguchi and Kamiya, 2000). The enzyme that initiates GA formation is the proplastid-located ent-copalyl diphosphate synthase (CPS), which catalyzes the conversion of geranylgeranyl diphosphate (GGPP) to copalyl diphosphate (CDP). Copalyl diphosphate is subsequently converted to ent-kaurene by the action of ent-kaurene synthase (KS). Two membrane-associated cytochrome p450 monooxygenases then catalyze sequential oxidations of ent-kaurene, thereby producing GA12. Once GA12 has been formed it may or may not be 13-hydroxylated into GA53, giving rise to a branching point leading to two parallel pathways: the early non-hydroxylation pathway and the early 13-hydroxylation pathway. The enzymes responsible for catalyzing the last steps in active GA biosynthesis are soluble 2-oxoglutarate-dependent dioxygenases. The multifunctional enzyme GA 20-oxidase (GA20ox) converts the C20-GAs into C19-GA compounds by the successive oxidations of GA12 and GA53 to GA9 and GA19, respectively. The final conversion into the bioactive GAs (GA4 and GA1) is then catalyzed by GA 3-oxidase (GA3ox). Deactivation of GA4 and GA1 can in turn be catalyzed by GA 2-oxidases (GA2ox). High levels of bioactive GA can trigger a feed-back mechanism, which represses the expression of GA20ox and GA3ox, and upregulates GA2ox (Hedden and Phillips, 2000). Similarly, when GA levels are low, the opposite occurs, thus enabling the plant to maintain GA homeostasis. Studies in Arabidopsis, rice and barley have also identified several positive and negative regulators of GA signaling pathways, all involved in regulating GA responsiveness during development (reviewed in Gomi and Matsuoka 2003;Olszewski et al., 2002; Swain and Singh, 2005). Furthermore, a number of studies have provided evidence of interactions between GA signaling and biosynthesis (e.g. Peng et al., 1999).

Figure 2.

Schematic representation of the cellular localization of different steps in the gibberellin (GA) biosynthesis and deactivation pathway.
Bioactive GAs are depicted in a box. Abbreviations: GGPP, geranylgeranyl diphosphate; CPS, copalyl diphosphate synthase; CDP, copalyl diphosphate; KS, ent-kaurene synthase, ER, endoplasmic reticulum.

Numerous studies have tried to pinpoint the tissues and organs in which bioactive GAs are synthesized and perceived in plants. The presence of bioactive GAs in actively growing and elongating tissues, such as shoot apices, young, developing internodes and expanding leaves (Eriksson and Moritz, 2002; Jones and Phillips, 1966; Kobayashi et al., 1998; Potts et al., 1982) suggests that bioactive GAs are primarily synthesized at the site of action. The overlapping expression of genes involved in both GA biosynthesis and signaling in rice supports this hypothesis (Kaneko et al., 2003), although tissue-specific subdivision of different stages in GA biosynthesis cannot be excluded. Studies in germinating Arabidopsis seeds, for example, have revealed non-overlapping tissue-specific expression patterns of early and late GA-biosynthesis genes (Yamaguchi et al., 2001). Further support for the hypothesis that GA intermediates may be transported between organs comes from studies in which GA20ox was found to be expressed predominantly in young expanding leaves and GA3ox in elongating stems in pea (Ross et al., 2003), tobacco (Itoh et al., 1999; Tanaka-Ueguchi et al., 1998) and hybrid aspen (Israelsson et al., 2004). In addition, older studies have revealed the presence of GAs in both xylem and phloem sap (reviewed in Hoad, 1995; Lang, 1970), but the identity and importance of GAs in these compartments remain to be established.

Despite the importance of GAs in growth and development of secondary vascular tissue, knowledge about their origin and perception in these tissues is lacking. In this report, highly sensitive gas chromatography-mass spectrometry selected reaction monitoring (GC/MS-SRM) analyses were employed to visualize the distribution of various GAs across the cambial region tissues of aspen (Populus tremula) trees. Similarly, the expression patterns across the same tissues of six GA biosynthesis genes (PttCPS1, PttGA20ox1, PttGA20ox4, PttGA3ox1, PttGA2ox1 and PttGA2ox2) and two genes implicated in GA signaling (DELLA-like1) or responsiveness (GIP-like1) were analyzed. The results show a tissue-specific distribution pattern of GAs across the wood-forming tissues and suggest that GA biosynthesis and perception is co-localized to the zone of xylem fiber elongation.


Sampling strategy

The cambial region of a tree consists of dividing, expanding and maturing xylem and phloem cells. These developmental stages can be easily recognized anatomically and longitudinal tangential sectioning has already been used to obtain tissue-specific samples for biological analysis (Uggla and Sundberg, 2001). In the study presented here this approach was used to measure the distribution pattern of GAs and indole-3-acetic acid (IAA) as well as the expression pattern of key genes involved in GA biosynthesis and signaling across the cambial region during active growth. The tissues used for the hormone analyses were all obtained from individual tangential sections (Figure 1A–G). In contrast, some of the samples used in the gene expression analysis were pooled from several sections. In order to increase the general validity of the results, expression levels of the genes were analyzed in three different trees, in two of which hormone distributions were also analyzed.

IAA and gibberellins show distinct distribution patterns across the cambial region

The levels of IAA exhibited a characteristic peak in the cambial zone and decreased steeply towards the phloem and xylem, reaching low and relatively stable levels at the transition between expanding and secondary wall-forming cells (Figure 3a–b). This is in agreement with previous studies, in Scots pine (Uggla et al., 1996, 1998, 2001), hybrid aspen (Tuominen et al., 1997) and aspen (Hellgren et al., 2004), thereby validating our sampling procedure.

Figure 3.

Results of hormone analysis across the cambial region of two field-grown aspen trees.
(a, b) Concentrations and distribution patterns of IAA.
(c–h) Concentrations and distribution patterns of various gibberellins (GAs) .
Shaded areas mark the location of sections containing more than 50% of cambium tissue. The approximate tissue distribution within each individual section is given in the lowest part of the figure.

Quantification of GAs revealed a good correlation between the distribution of GAs in the early non-hydroxylation and the early 13-hydroxylation pathway across the cambial region (Figure 3c–h). Levels of the bioactive GA precursors GA20 and GA9 exhibited peaks in the phloem and a smaller peak in the expanding xylem cells (Figure 3c–d). In contrast to this pattern, the bioactive GA4 and GA1 peaked in expanding xylem cells (Figure 3e–f), with only trace amounts present in the phloem and cambial zone. GA4 was always present at a higher concentration than GA1. This is consistent with recent findings that the concentration of GA4 is higher than that of GA1 in young elongating internodes of hybrid aspen (Israelsson et al., 2004). The expanding xylem cells also exhibited the highest levels of deactivated GAs (Figure 3g–h). However, deactivated GAs were also present in the phloem, which may partly explain the low concentration of GA4 and GA1 in this tissue.

Expression of gibberellin biosynthesis genes across the cambial region

Real-time RT-PCR analysis of the ent-copalyl diphosphate synthase (CPS) gene that encodes the first enzyme committed to GA biosynthesis revealed a phloem-enriched transcript distribution (Figure 4a–c). The primers were designed based on a full-length PttCPS1 cDNA previously isolated from young elongating hybrid aspen shoots (data not shown). In most species examined, CPS is encoded by one gene (reviewed in Hedden and Phillips, 2000), but in pumpkin two functional CPS enzymes have been characterized with distinct expression patterns (Smith et al., 1998) According to BLAST searches of the Populus trichocarpa genome (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html), two CPS genes are present in Populus (data not shown). Therefore, several gene-specific primer pairs for the putative PtCPS2 were designed and evaluated, but no expression of corresponding transcripts was detected. It was thus concluded that only one CPS is expressed during wood formation in aspen.

Figure 4.

Relative expression levels of genes encoding gibberellin (GA) biosynthesis enzymes in three field-grown aspen trees.
Expression was detected by real-time RT-PCR by employing gene-specific primer pairs and 18S as a reference gene, see Experimental procedures for details. The shaded areas mark the location of pooled samples containing more than 50% cambium tissue. The approximate tissue distribution within each sample of pooled sections is given in the lowest part of the figure. The asterisk (*) indicates a sample that was not included in the analysis.

PttGA20ox1, which encodes the multifunctional enzyme GA20ox has previously been characterized in hybrid aspen (Populus tremula × P. tremuloides; Eriksson and Moritz, 2002). In contrast to the distribution of PttCPS1 transcript, the expression of PttGA20ox1 was low in the phloem and cambial zone but peaked in expanding xylem cells. This peak co-localized with the peak of bioactive GAs (Figures 3e–f and 4d–f). Because GA20ox belongs to a small multigene family (Hedden and Phillips, 2000), we searched the poplar genome (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html) for genes homologous to PttGA20ox1 and identified three additional putative GA20ox genes (data not shown). Analysis of these sequences revealed that the primers for PttGA20ox1 were also likely to amplify a putative GA20ox2. However, preliminary gene-specific analysis of the GA20ox2 showed that the two genes had very similar expression patterns to that of GA20ox1 during wood formation (data not shown). The least similar GA20ox gene to PttGA20ox1 of the three was named PtGA20ox4 (Joint Genome Institute (JGI) ID no. 86560) after confirming its putative gene identity using National Centre for Biotechnology Informationblastx (highest score GA20ox: 533; e-150, GenBank accession no. CAB96202). Interestingly, PtGA20ox4 was found to be expressed at a higher overall level than PttGA20ox1 and with a distribution flanking the expression of PttGA20ox1 (Figure 4g–i).

One GA3ox is encoded by PttGA3ox1 as described in Israelsson et al. (2004) and catalyzes the conversion of GA9 and GA20 into GAs with biological activity. PttGA3ox1 expression peaked in the phloem (Figure 4h–i), although, in one replicate, PttGA3ox1 levels also exhibited a second peak in the expanding xylem, where the concentrations of GA1 and GA4 were highest (Figures 3e and 4j). Therefore, the results do not definitively establish that PttGA3ox1 has a tissue-specific transcription pattern. Similarly to GA20ox, GA3ox belongs to a small multigene family (Hedden and Phillips, 2000) and according to a preliminary genomic search (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html) there are three additional GA3ox genes in Populus (data not shown). However, sequence analysis of these genes predicted that the primers used for PttGA3ox1 were gene specific.

Finally, the transcription of two expressed genes encoding putative GA deactivating enzymes, PttGA2ox1 and PttGA2ox2 (described in Israelsson et al., 2004), were analyzed. No amplification of PttGA2ox1 transcript was detected even after 45 cycles of PCR cycling. The lack of transcript amplification was not due to improper primer design as the Polymerase chain reaction (PCR) efficiency of PttGA2ox1 primers using plasmid template was highly comparable to that of PttGA2ox2 (data not shown). However, PttGA2ox2 was relatively highly expressed in the tissues flanking the cambium (Figure 4j–l). This correlates to the high levels of deactivated GAs detected in both phloem and expanding xylem (Figure 3g–h).

Expression of gibberellin signaling and response genes

To determine whether the distribution pattern of biologically active GAs correlated with the expression of genes involved in GA signaling and GA responsiveness, we analyzed the expression profiles of two genes (designated DELLA-like1 and GIP-like1, for reasons described below) encoding proteins that act downstream of GA perception. DELLA proteins are growth repressors that are degraded in the presence of GA and act as key components in the GA signaling pathway (reviewed in Olszewski et al., 2002). In Arabidopsis and rice, the transcription of the DELLA genes repressor of gal-3 (RGA) and slender rice (SLR1) are both upregulated in response to bioactive GA treatment (Itoh et al., 2002; Ogawa et al., 2000; Silverstone et al., 1998). A DELLA homologue was identified in the Populus expressed sequence tag database (EST DB) and selected for real-time Reverse Transcription (RT)-PCR primer design and expression analysis (accession no. BU815161). To confirm the gene identity, we first obtained the full-length sequence by blast searches in the P. trichocarpa genome (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html, data not shown). The deduced amino acid sequence contained the characteristic DELLA domain and the gene was subsequently named DELLA-like1 (60% identity at amino acid level to AtRGA1). As seen in Figure 5(a–c), the DELLA-like1 gene expression profile exhibited a specific peak in expanding xylem where bioactive GA levels are highest (Figure 3e–f) in accordance with a GA-induced DELLA expression pattern.

Figure 5.

Relative expression levels of genes encoding gibberellin (GA) signaling and response proteins in three field-grown aspen trees. Expression was detected by real-time RT-PCR by employing gene-specific primer pairs and 18S as a reference gene, see Experimental procedures for details. The shaded areas mark the location of pooled samples containing more than 50% cambium tissue. The approximate tissue distribution within each sample of pooled sections is given in the lowest part of the figure. The asterisk (*) indicates a sample that was not included in the analysis.

The other gene selected for expression analysis was a Populus GA responsive gene similar to the Arabidopsis GASA4 (Aubert et al., 1998; accession no. AI166057). This Populus gene has previously been reported to be upregulated in GA overproducing hybrid aspen trees (Israelsson et al., 2003). An annotation update on the deduced amino acid sequence revealed it was most similar (72% identity) to a GA-induced Petunia hybrida GIP-like protein (accession no. AJ417391). Hence, the Populus gene was subsequently renamed from GASA4-like to GIP-like1. Among all assayed genes, GIP-like1 exhibited the highest transcript levels, and it was specifically expressed in tissues with high concentrations of GA1 and GA4 (Figure 5d–f). Compared with its lowest expression levels (in phloem; sample A), GIP-like1 expression was several hundredfold higher in expanding xylem in all three trees. Although the cambial tissue contained relatively low concentrations of bioactive GAs, abundant GIP-like1 expression was detected in the cambium-enriched sample C. This may be due to the small amounts of expanding xylem that were unavoidably present in this fraction. Overall, the expression of DELLA-like1 and GIP-like1 was co-localized with the peak levels of GA1 and GA4 in the cambial region of the aspen trees. Furthermore, the high abundance of GIP-like1 transcripts suggest that it may be used as a sensitive marker of sites expected to actively perceive the GA signal during wood formation.


In order to understand the role of GAs in wood development, it is important to determine the tissues and developmental stages in which GA is synthesized and perceived. In the present investigation, we employed a tangential cryosectioning technique that permitted nearly tissue-specific sampling from the cambial region of actively growing aspen trees. We determined the levels of IAA and various GA metabolites across these different tissues and subsequently analyzed their distributions in relation to expression levels of six GA biosynthesis genes and two GA signaling and response genes.

Can the different distribution patterns of bioactive gibberellins and IAA help explain the roles of the two plant growth regulators?

Auxin is an important regulator in the growth and development of primary and secondary vascular tissues. It also induces differentiation of vascular strands in callus and explants (Jacobs, 1984; Mattsson et al., 1999). We confirmed the characteristic peak of IAA in the cambial zone of P. tremula (Figure 3a,b) as previously described in Scots pine (Uggla et al., 1996, 1998, 2001), hybrid aspen (Tuominen et al., 1997) and P. tremula (Hellgren et al., 2004). In contrast to IAA, the distribution of the bioactive GA4 and GA1 exhibited distinct peaks in the expanding xylem (Figure 3e,f) suggesting that GA plays a role in the processes of expansion and/or elongation of xylem elements. Application of a GA biosynthesis inhibitor to wood-forming tissues of Eucalyptus globulus resulted in decreased GA levels and shorter xylem fibers, whereas the radial width of the fibers were not affected in a study by Ridoutt et al. (1996). Together with the documented effect of GA on longitudinal cell expansion (e.g. Shibaoka, 1991), these observations suggest that the main effect of GA may be on xylem fiber elongation rather than radial expansion. However, it should be emphasized that fibers elongate by tip growth and not by diffuse expansion (Lev-Yadun, 2001) and thus the effects of GAs on cell elongation in other systems are not always applicable to fiber growth.

Application of GA to decapitated poplars has been found not only to stimulate fiber elongation, but also to have a synergistic effect with auxin on xylem production, and when low IAA/GA ratios were applied phloem production was stimulated (Wareing, 1958;Digby and Wareing, 1966). However, the lack of detectable levels of bioactive GA in the cambial zone and phloem suggest that any role that GA may play in planta on cambial cell division is either of limited importance, or that the level of GA is very precisely controlled to keep the phloem to xylem ratio low. Too high levels of GA might change the balance between IAA and GA and thus the xylem to phloem ratio. However, it should be noted that, in transgenic GA-overproducing hybrid aspen trees, an increase was found in both the number and length of xylem fibers as compared with wild type, while the amount of phloem tissue remained the same (Eriksson et al., 2000; Israelsson, 2004). Although these transgenes again demonstrate that GA has the potential to affect cambial cell division and xylem production, it must also be mentioned that the transgenes did also contain increased levels of IAA (Israelsson et al., 2003), indicating that the increased number of xylem fibers might be due to altered IAA levels. Taken together, the exogenous GA stimulation on fiber length demonstrated by Digby and Wareing (1966) and Ridoutt et al. (1996) together with the localization of endogenous GA to the expansion zone demonstrated by us supports the idea that a major function of GA during wood formation is to stimulate fiber elongation.

Subdivision of early and late gibberellin biosynthesis in the cambial region

Expression analysis of PttCPS1, which encodes the first enzyme committed to GA biosynthesis, revealed a phloem-enriched transcript distribution (Figure 4a–c), while expression of PttGA20ox1, which acts late in GA biosynthesis, was detected mainly in expanding xylem tissue, as were the bioactive GA1 and GA4 (Figures 3e,f and 4d–f). Transport of an intermediate GA between different tissues has been suggested to occur during Arabidopsis seed germination, based on the expression pattern of AtCPS1 in embryo provasculature compared to the expression of AtKO1, AtGA3ox1 and AtGA3ox2 in the cortex and endodermis (Yamaguchi et al., 2001). As suggested by Yamaguchi and co-workers, the highly hydrophobic ent-kaurene could be transported, possibly by association with a carrier protein. In wood-forming tissues, ray cells symplastically interconnect the phloem with the developing xylem, thus providing a feasible translocation route for ent-kaurene, or other precursors of C19-GAs in trees (Figure 6). Once present in the developing xylem, ent-kaurene could be further metabolized, enabling de novo production of C19-GAs.

Figure 6.

A working model for the biosynthesis of gibberellin (GA) during wood formation.
The concentrations of C19-GAs (e.g. GA9) are abundant in the phloem and elongating xylem of the cambial region of aspen. The bulk of the GAs in the phloem may originate from long-distance transport from young leaves and thus may be limited to the phloem sap. However, the C19-GA precursors such as ent-kaurene or C20-GAs may be synthesized in the phloem, transported symplastically through the ray cells and then used in de novo C19-GA biosynthesis within the developing xylem. Once the limiting C19-GAs have been produced on site they are subsequently converted into bioactive GAs (e.g. GA4). These last stages of GA formation may occur mainly in the elongating xylem fibers where bioactive GAs are suggested to act and, to a lesser degree (if at all), in the xylem ray cells or the conductive vessel elements. Based on results from another woody angiosperm (Ridoutt et al., 1996), GA may act on xylem fiber elongation (increase in length through intrusive tip growth) rather than xylem fiber radial expansion.

Interestingly, a second putative GA20ox, PtGA20ox4, was expressed at low levels in expanding xylem, but was highly expressed in flanking tissues (Figure 4d–i). A possibility that cannot be excluded is that the PtGA20ox4 transcript profile is governed, at least in part, by feedback loops imposed by the high and low bioactive GA levels detected in expanding xylem and flanking tissues, respectively. Conversely, the expression of PttGA20ox1 could be mainly driven by developmental and tissue-specific cues that overcome any feedback regulation imposed by the relatively high bioactive GA levels in expanding xylem. It remains to be established whether the lack of bioactive GAs in the cambium and phloem can be explained by factors such as cell-specific compartmentalization of CPS1, GA20ox4 and GA3ox1 or a high turn-over rate provided by GA2ox2. There was no clear xylem-enriched transcription of the gene encoding PttGA3ox1, the enzyme that catalyzes the conversion of inactive GAs into GAs with biological activity (Figure 4j–l). However, based on the phenotype of transgenic 35S-GA20ox and 35S-GA3ox hybrid aspen plants, where only the former show substantially increased GA levels and increased xylem growth, the activity of GA20ox relative to GA3ox has been suggested to be the limiting factor in GA biosynthesis and, thus, to play a major role in the regulation of wood formation by GAs (Eriksson et al., 2000; Israelsson et al., 2004). Therefore, the phenotype of transgenic 35S-GA20ox shows that there is in vivo evidence for a GA 3-hydroxylation in xylem, although specific expression profiles of the remaining GA3ox genes will help resolve which GA3ox is essential for a GA biosynthesis in the xylem.

Based on our data, we hypothesize that de novo production of the C19-GAs GA9 and GA20 occurs in expanding xylem (Figure 6). In addition, this is dependent on transport of a GA precursor(s) from adjoining tissues for bioactive GA formation. In contrast, the high levels of C19-GAs in the phloem may originate from young expanding leaves where GA20ox transcripts are abundant (Israelsson et al., 2004; Itoh et al., 1999; Ross et al., 2003; Tanaka-Ueguchi et al., 1998) and would thus be restricted to the phloem assimilate stream. Future studies using inducible systems that can block or increase the transcription of relevant genes in specific tissues, for instance PttCPS in the phloem, PttGA20ox in the leaves or PttGA20ox in the xylem, may finally help to resolve how the GA4 and GA1 pools in the xylem are derived.

Overlap of bioactive gibberellins and the expression of gibberellin signaling and response genes

The DELLA proteins are known to act as growth repressors of GA-mediated processes (reviewed in Olszewski et al., 2002). Upon receiving a GA signal, at least some of these proteins are degraded through targeted polyubiqutination and subsequent degradation by the 26S proteasome (Fu et al., 2002; McGinnis et al., 2003; Sasaki et al., 2003). Following GA treatment in Arabidopsis, the DELLA protein RGA becomes destabilized (Silverstone et al., 2001), whereas RGA transcription is upregulated (Silverstone et al., 1998). Similarly, in Arabidopsis and rice, the transcription of the DELLA genes RGL2 and SLR1 also become upregulated in response to bioactive GA treatment (Itoh et al., 2002; Ogawa et al., 2000; Tyler et al., 2004). It is feasible that the presence of GA triggers increased expression of DELLA genes, thereby compensating for the increased protein degradation. In agreement with the cited studies, we detected higher amounts of DELLA-like1 gene expression in tissues with high concentrations of bioactive GA (Figures 3c,d and 5a–c). Furthermore, in agreement with results from Arabidopsis showing a high background expression of RGA throughout the plant (Silverstone et al., 1998), the real-time RT-PCR analysis also revealed relatively high levels of DELLA-like1 expression in the cambium and phloem, where levels of bioactive GA were low. In addition, we assayed a gene similar to the GA responsive GIP genes isolated in P. hybrida (Ben-Nissan and Weiss, 1996; Ben-Nissan et al., 2004). The exact functions of the GIP proteins remain to be established, although overexpression studies have implicated a role in shoot elongation and transition to flowering for one of them (Ben-Nissan et al., 2004). In our study, the expression of GIP-like1 dramatically increased (several hundredfold) in tissues where the concentration of bioactive GA was highest (Figures 3c,d and 5d–f). This suggests that GIP-like1 expression may be used as a responsiveness marker gene for tissues actively perceiving GA signals during wood formation in trees.


Based on the analysis of the distribution of GAs in wood-forming tissues, the present study suggests that the role of GA is restricted mainly to fiber cell elongation during wood formation. Any stimulation by GA of cambial growth is probably of limited importance based on the very low bioactive GA concentrations found in the cambium. At least the last stages of GA biosynthesis in expanding xylem are suggested to occur in situ, although the possibility that a precursor(s) is (are) transported to the tissue cannot be excluded. Finally, GA induced signaling and responsiveness, as measured by the transcription of two selected genes, coincided well with sites of high bioactive GA levels.

Experimental procedures

Plant material and sample preparation

The plant material consisted of approximately 40-year-old, 15-m-tall, vigorously growing aspen (P. tremula) trees growing in northern Sweden (64°21′N, 19°46′E). Three trees were sampled on June 30, 2002, during the most active period of cambial growth. The samples, consisting of extraxylary tissues and a few annual rings of wood were collected around the stem. The blocks were immediately frozen in liquid nitrogen, transported to the laboratory on dry ice and stored at −80°C.

Analysis was performed on 30-μm-thick, longitudinal tangential sections obtained across the wood-forming tissues by cryosectioning approximately 3 × 15 mm specimens. The sampling procedure and anatomical characterization was performed as described in Uggla and Sundberg (2001).

Real-time RT-PCR

Frozen sections from the cambial region were disrupted using an MM 301 Vibration Mill (Retsch GmbH & Co. KG, Haan, Germany) at a frequency of 30 Hz for 2 min with 3-mm tungsten carbide beads (Retsch GmbH & Co. KG,) added to each tube to increase the homogenization efficiency. During the last 1.5 min of disruption, a lysis/binding solution including Plant RNA Isolation Aid (Ambion, Austin, TX, USA) was included and total RNA was subsequently isolated using an RNAqueous®-Micro kit (Ambion). Genomic DNA was removed by treatment with the DNase I included in the kit. All steps were performed according to the manufacturer's instructions. Synthesis of cDNA was carried out with the obtained total RNA and random poly-d(N)6 oligonucleotide primers using an Invitrogen SuperscriptTM II Reverse Transcriptase Kit with the addition of RNAseOUTTM (Invitrogen, Carlsbad, CA, USA). The amount of RNA added to each cDNA synthesis reaction was not known due to the scarce amount of starting plant material (approximately 1–2 mg). Primers for real-time RT-PCR were designed using Beacon Designer software (v2.1; Premier Biosoft International, Palo Alto, CA, USA) for amplicon lengths of between 100 and 150 bp and are shown below (Table 1). The majority of the primers constructed for real-time PCR analysis were based on hybrid aspen (P. tremula × P. tremuloides) sequences that had been obtained either by manual cloning in our group, or by blast searches among approximately 100 000 EST sequences in the Populus DB (Sterky et al., 2004). A primer pair for the constitutively expressed control gene 18S was also included.

Table 1.  Oligonucleotide primers used for real-time RT-PCR expression analysis and amplicon sizes
Gene nameGenBank acc. no./ additional info.Amplicon size (bp)Real-time RT-PCR primers forward and reverse
PttCPS1Cloned in our lab1485′-CCTCCATCCATCTCTTGTCTAC-3′
PttGA2ox1See PatGA2ox1, AY392094965′-TGGTAGTGGATCCCCAACTC-3′

PCR reactions were performed in a 96-well plate with a Bio-Rad iCycler iQTM Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). PCR products were detected by measuring signals from the fluorescent DNA dye SYBR Green (QuantitectTMSYBR® Green PCR Kit; Qiagen, Hilden, Germany) bound to them. All assays were carried out in triplicate and a set of template-free controls was included per gene amplification. Each PCR reaction mixture contained 1 μl of 1.5-fold diluted cDNA sample, 12.5 μl SYBR Green Master Mix, 7.5 pmol of forward and reverse primers (Invitrogen) and 0.25 pmol Fluorescein Calibration Dye (Bio-Rad) in a total volume of 25 μl. Assays were repeated on three biological replicates with independently isolated RNA and independently synthesized cDNA. Thermal cycling conditions were: a 15-min initial activation step at 95°C, followed by 45 cycles of 15 sec at 95°C, 30 sec at 57°C and 30 sec at 72°C, then 1 min at 95°C, 1 min at 55°C, a melting curve programme (80 cycles, 10 sec each, of 0.5°C elevations starting at 54°C) and a cooling step to 4°C. The presence of one product per assay was confirmed by analysis of the dissociation curves. iCycler iQTM software 3.0 (Bio-Rad) was used to calculate the first significant fluorescence signal above noise, the threshold cycle (CT). When necessary, the back-ground level was adjusted manually to cross an exponential portion of the amplification curves of all samples being compared on the 96-well plate. The average CT value for 18S in diluted cDNA samples was 10.8 ± 1.3 (SD). The PCR efficiencies (E) of each amplicon were calculated, using data from pooled cDNA originating from cambial region samples in fourfold serial dilutions, by the iCycler iQTM software 3.0 package. The relative transcript levels (RTLs) were calculated as follows: 100 000 × ECT Control/ECT Target, thus normalizing target gene expression to the control gene expression.

Quantification of IAA and gibberellin

Frozen samples were individually placed, with 500 μl of extraction medium (MeOH:H2O:CH3COOH; 800:190:10) including stable isotope internal standards [1 ng 13C6-IAA (Cambridge Isotopes Laboratories, Andover, MA, USA); 50 pg 2H2-GAs (purchased from Prof. L. Mander, Australia University, Canberra, Australia], into 1.5 ml Eppendorf tubes. They were then extracted using an MM 301 Vibration Mill (Retsch GmbH & Co. KG) at a frequency of 30 Hz for 3 min after adding 3-mm tungsten carbide beads (Retsch GmbH & Co. KG) to each tube to increase the extraction efficiency. After centrifugation in an Eppendorf centrifuge for 10 min at 112.5 g, the supernatant was evaporated to dryness in a Speed-vac concentrator (Savant Instruments Inc., Farmingdale, NY, USA). The residue was dissolved in 500 μl H2O and applied to a pre-equilibrated 100-mg C8EC ISOLUTE cartridge (Sorbent AB, Västra Frölunda, Sweden), after adjusting the pH to 3.0 with 1 m HCl. The column was washed with 2 ml of 1% aqueous acetic acid, and then GAs and IAA were eluted with 2 ml 80% MeOH.

After evaporation to dryness, the samples were methylated with ethereal diazomethane, dried and trimethylsilylated in 20 μl dry pyridine/N,O-Bistrimethylsilyltrifluoroacetamide/trimethylchlorosilane (50:50:1, volume in volume, v/v) at 70°C for 30 min. The derivatization mixture was then reduced to dryness and dissolved in 6 μl heptane. Samples (1 μl for IAA analysis, 3 μl for GA analysis) were injected in the splitless mode into an HP 5890 gas chromatograph (Hewlett Packard, Palo Alto, CA, USA) fitted with a fused silica glass capillary column (15 m long, 0.25 mm inner diameter) with a chemically bonded 0.25 μm DB-5MS stationary phase (J&W Scientific, Folsom, CA, USA). The injector temperature was 270°C. The column temperature programme varied depending on whether GAs or IAA were being analyzed. The column effluent was introduced into the ion source of a JMS–SX/SX102A mass spectrometer (JEOL, Tokyo, Japan). The interface temperature was 270°C and the ion source temperature 250°C. The acceleration voltage was 10 kV and ions were generated with 70 eV at an emission current of 300–800 μA. For quantification, samples were analyzed in SRM mode (Edlund et al., 1995; Moritz and Olsen, 1995). For the GA analyses, calibration curves were recorded from 0.1 to 20 pg GA, with 5 pg 2H2-GA as an internal standard. For IAA, the calibration ranged from 10 to 5000 pg, with 100 pg of 13C6-IAA as an internal standard.


We would like to thank Kjell Olofsson for excellent technical assistance in cryosectioning and anatomical characterization and (together with Inga-Britt Carlsson) sample preparation for GA analysis. We also thank Gabriella Nilsson for initial expression analysis of PttGA20ox1 in hybrid aspen stems and Dr Makiko Chono for cloning the PttCPS1 gene. The DoE Joint Genome Institute and Poplar Genome Consortium are gratefully acknowledged for providing genomic sequences from the ongoing Populus genome sequencing project (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html). This work was supported by FORMAS, the Swedish Research Council, EU-strategic funding and the Kempe foundation.