Polyamines are small polycationic amines that are widespread in living organisms. Thermospermine, synthesized by thermospermine synthase ACAULIS5 (ACL5), was recently shown to be an endogenous plant polyamine. Thermospermine is critical for proper vascular development and xylem cell specification, but it is not known how thermospermine homeostasis is controlled in the xylem. We present data in the Populus model system supporting the existence of a negative feedback control of thermospermine levels in stem xylem tissues, the main site of thermospermine biosynthesis. While over-expression of the ACL5 homologue in Populus, POPACAULIS5, resulted in strong up-regulation of ACL5 expression and thermospermine accumulation in leaves, the corresponding levels in the secondary xylem tissues of the stem were similar or lower than those in the wild-type. POPACAULIS5 over-expression had a negative effect on accumulation of indole-3-acetic acid, while exogenous auxin had a positive effect on POPACAULIS5 expression, thus promoting thermospermine accumulation. Further, over-expression of POPACAULIS5 negatively affected expression of the class III homeodomain leucine zipper (HD-Zip III) transcription factor gene PttHB8, a homologue of AtHB8, while up-regulation of PttHB8 positively affected POPACAULIS5 expression. These results indicate that excessive accumulation of thermospermine is prevented by a negative feedback control of POPACAULIS5 transcript levels through suppression of indole-3-acetic acid levels, and that PttHB8 is involved in the control of POPACAULIS5 expression. We propose that this negative feedback loop functions to maintain steady-state levels of thermospermine, which is required for proper xylem development, and that it is dependent on the presence of high concentrations of endogenous indole-3-acetic acid, such as those present in the secondary xylem tissues.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Polyamines are essential organic polycationic amines that have been implicated in several processes in plants, such as biotic and abiotic stress responses (Alcázar et al., 2006; Yamaguchi et al., 2006; Kusano et al., 2007; Naka et al., 2010; Gonzalez et al., 2011; Wang et al., 2011; Sagor et al., 2012), wound responses (Perez-Amador et al., 2002), nitric oxide signaling (Flores et al., 2008), fruit development (Nambeesan et al., 2010; Trénor et al., 2010) and stem growth and elongation (Hanzawa et al., 2000; Alcázar et al., 2005). The most common polyamines are the diamine putrescine, the triamine spermidine and the tetramines spermine and thermospermine. Putrescine is produced from ornithine by ornithine decarboxylase or from arginine by arginine decarboxylase. Spermidine and spermine production is catalysed by aminopropyltransferases, which transfer an aminopropyl residue from the decarboxylated S-adenosylmethionine to an amine acceptor on putrescine or spermidine to produce triamines and tetraamines, respectively. Thermospermine is a structural isomer of spermine that was only recently identified in plants (Knott et al., 2007; Naka et al., 2010; Rambla et al., 2010). It is synthesized by the thermospermine synthase ACAULIS5 (ACL5) (Knott et al., 2007) that is expressed specifically in early developing xylem vessel elements (Muñiz et al., 2008). Disruption of the function of ACL5 in Arabidopsis leads to plants with impaired stem elongation and thinner veins in leaves, as well as lack of secondary growth (Hanzawa et al., 1997, 2000; Clay and Nelson, 2005; Kakehi et al., 2008; Muñiz et al., 2008).
The events downstream of ACL5 expression have been subject of intensive study in recent years. At least two extragenic suppressors of the acl5 mutation have been described. One of them disrupts an upstream open reading frame and enhances translation of a basic helix-loop-helix (bHLH) transcription factor encoded by SUPPRESSOR OF ACAULIS51 (SAC51), and the second (sac52-1d) affects RPL10a, an important component of the large ribosomal subunit (Imai et al., 2006, 2008). However, the upstream events that regulate ACL5 expression are largely unknown. Class III homeodomain-leucine zipper transcription factors (HD-Zip III transcription factors) have been hypothesized to play a role in transcriptional control of ACL5, as both HD-Zip III transcription factors and ACL5 have been implicated in control of metaxylem development (Muñiz et al., 2008; Carlsbecker et al., 2010). AtHB8 is a particularly good candidate for control of ACL5 as it is the HD-Zip III family member that shows highest transcriptional alterations in the acl5 mutant background (Imai et al., 2006). Another important factor is auxin, which is well known for its role in xylem development (Uggla et al., 1996; Tuominen et al., 1997; Nilsson et al., 2008) as well as in transcriptional activation of both ACL5 (Hanzawa et al., 2000; Imai et al., 2006; Rambla et al., 2010) and AtHB8 (Baima et al., 1995). A model for thermospermine regulation of xylem differentiation involving auxin has been proposed, in which it was suggested that HD-Zip III transcription factors mark the procambial cells that are destined for xylem specification in an auxin-dependent manner, leading to up-regulation of ACL5 and concomitant differentiation of xylem vessel elements (Vera-Sirera et al., 2010; Takano et al., 2012). However, the role of the HD-Zip III transcription factors in control of ACL5 expression has not been studied in detail.
In the present study, we wished to elucidate the relationship between ACL5, auxin and HD-Zip III transcription factors. Populus trees were selected as the model system due to extensive development of xylem, which is the site of thermospermine production as well as auxin transport in plants. Tree stems therefore allow isolation of large amounts of tissues that are enriched in xylem elements transporting auxin and expressing ACL5. The ACL5 orthologue, POPACAULIS5, was cloned from hybrid aspen (Populus tremula × Populus tremuloides) and black cottonwood (Populus trichocarpa), and thermospermine levels were altered in trees by manipulating the expression levels of POPACAULIS5. In addition, PttHB8, a hybrid aspen homologue of AtHB8, was over-expressed in hybrid aspen to investigate the effect on POPACAULIS5 transcript levels. The results on expression levels of POPACAULIS5 and PttHB8, thermospermine accumulation and indole-3-acetic acid (IAA) measurements in the transgenic trees support a novel regulatory mechanism, mediated through auxin and PttHB8, for maintenance of thermospermine homeostasis in secondary xylem tissues of the stem.
POPACAULIS5 is the Populus ACAULIS5 orthologue
In Arabidopsis, ACL5 encodes thermospermine synthase (Knott et al., 2007). Our search in the Populus genome retrieved a putative ACL5 sequence from P. trichocarpa, POPTR_0006s23880 (Phytozome Populus version 2; Goodstein et al., 2012) with 86.1% identity to ACL5. Two other sequences were found in the P. trichocarpa genome, POPTR_0008s15120 and POPTR_0010s09940, which show higher similarity amongst themselves than to ACL5 or to POPTR_0006s23880 (Figure 1a). BLAST searches of POPTR_0006s23880 against the Populus DB (http://www.populus.db.umu.se/; Sterky et al., 2004) retrieved a 496 bp EST (GI:3853671, GenBank EST database) from P. tremula × P. tremuloides, with 95% identity over 99% of the sequence (Figure 1b). Alignment of ACL5 amino acid sequences from Arabidopsis, Populus and other land plants showed that POPTR_0006s23880, hereafter called POPACAULIS5, is the sequence most similar to ACL5.
The relationship of POPACAULIS5 with other known and predicted ACL5 sequences was inferred from a phylogenetic analysis (Figure 1a), in accordance with previously reported relationships amongst plant aminopropyltransferases (Minguet et al., 2008; Rodriguez-Kessler et al., 2010). The tree shows two major clusters, one identified as predicted thermospermine synthase proteins and the other as thermospermine synthase-like proteins (Figures 1a and S1). The POPACAULIS5 putative sequence from P. trichocarpa and P. tremula × P. tremuloides PttACL5 group within the former cluster, in which the Arabidopsis ACL5 sequence is also included, suggesting that this is the most likely candidate to be thermospermine synthase. We confirmed that POPACAULIS5 is a true orthologue of ACL5 by demonstrating its thermospermine synthase activity in yeast cells (Figure S2).
35S::POPACAULIS5 trees show reduced overall growth and slight defects in xylem development
We isolated and cloned POPACAULIS5 cDNA from P. trichocarpa and P. tremula × P. tremuloides (PttACL5) for over-expression under the control of the 35S CaMV constitutive promoter. Transformation of hybrid aspen allowed recovery of 90 kanamycin-resistant transgenic lines for 35S::POPACAULIS5, and 39 kanamycin-resistant transgenic lines for 35S::PttACL5. Ectopic expression of POPACAULIS5 from both constructs resulted in the same dramatic changes in shoot and root development, and the transgenic lines obtained were grouped according to the severity of observed phenotypes (Figures 2a and S3a–f). Dwarf transgenic plants from group B (Figure 2a) were able to develop a rooting system and showed partial restoration of the wild-type phenotype once transferred to a plant growth regulator-free medium (Figure 2b). For this reason, we selected the transgenic lines 35S::POPACAULIS5-B2, B4, B13, B14 and B15 from group B for further analysis of the effects of POPACAULIS5 over-expression in trees (Figure 2c).
Tree height was significantly reduced in 35S::POPACAULIS5 trees grown in the greenhouse for a period of 2 months (Figure 2c,d). In addition, the diameter and internode length of the mature stem as well as the leaf size were smaller in the transgenic plants than in the wild-type (Figure 2e–g). Strong phenotypic variation was observed between independent transgenic lines (Figures 2c and S4a) and even in individual trees from the same transgenic line. Maturation of secondary xylem was screened along the stem by searching for the youngest internode with fully lignified xylem fibres, detected by the appearance of highly autofluorescing tissues in the secondary xylem. In wild-type, fully lignified secondary xylem fibres were found in the 25th internode. Xylem maturation was delayed in the transgenic lines as fully lignified secondary xylem fibres were only observed in the 32nd internode in B2 and the 39th internode in B14, and were absent from the stem in lines B4 and B13. We performed further analyses at internode 45 from the top (the so-called reference internode; see Experimental procedures). Xylem anatomy and cell morphology were analysed at the reference internode from transgenic lines B2 and B14, which showed a intermediate phenotype in xylem maturation as well as the various growth parameters (Figures 3a and S4a,b). The absence of ACL5 function in the acl5 mutants of Arabidopsis has a profound effect on xylem maturation, resulting in premature cell death of xylem elements and reduction in the complexity of secondary cell-wall patterning (Muñiz et al., 2008). By nitroblue tetrazolium staining of stem transverse sections, we observed that the width of the living xylem zone in the reference internode of the transgenic lines was similar to that of the wild-type (Figure 3b). The shorter distance from pith to cambium together with the equal width of the xylem living zone observed in the transgenic lines indicate that cell death in xylem fibres may be slightly delayed in the transgenic lines compared to the wild-type (Figure 3b). No major alterations were observed in the size (Figure S5a–f) or cell-wall patterning of xylem elements in the transgenic tree stem samples collected from the reference internode (Figures S6a and S7). However, a slight increase in the proportion of the primary xylem vessel types was observed (Figure S6b,c). Altogether, the lack of major defects in secondary xylem development as a consequence of manipulating POPACAULIS5 raised the question as to what the level of POPACAULIS5 over-expression was, and whether thermospermine was over-produced in the secondary xylem tissues of the 35S::POPACAULIS5 woody stems.
Thermospermine accumulation is suppressed in the 35S::POPACAULIS5 Populus woody stem, but not in leaves
The contents of the main polyamines putrescine, spermidine, spermine and thermospermine were measured in samples collected from leaves and scrapings of the living zone of the secondary xylem in the stem. Spermidine and spermine were highly abundant in all plant tissues, while putrescine and thermospermine were less abundant but still easily detectable (Figure 4a,b). As expected, the thermospermine content was higher in leaves of the transgenic lines compared to the wild-type. However, similar or even lower levels of thermospermine were observed in the secondary xylem tissues collected from the same tree stems (Figure 4a). Similar results with higher levels of thermospermine in the leaves but not in the stem were obtained in plants grown in vitro on auxin-free medium (Figure 4b). It was also interesting to note that the levels of the other polyamines were not increased in the stem or xylem samples of the transgenic lines (Figure 4a,b), making it unlikely that the lack of thermospermine over-accumulation in these samples is due to back-conversion of thermospermine to spermidine or putrescine. A decrease in spermidine and spermine levels was observed in several samples (Figure 4b), which may be related to increased overall polyamine catabolism, as recently proposed in Arabidopsis 35S::ACL5 plants (Marina et al., 2013).
Expression of POPACAULIS5 paralleled the changes in the thermospermine levels of the transgenic trees. Quantitative RT-PCR analysis of the same samples analysed for polyamine content showed approximately 40-fold and higher increases in POPACAULIS5 expression in leaves of the transgenic trees, but the expression in xylem tissues was unaltered in lines B2, B13 and B4 and even reduced in lines B14 and B15, compared to the wild-type (Figure 4a). To exclude the possibility that this result was due to inactivity of the 35S promoter in the secondary xylem, we analysed transgenic trees carrying a 35S::GUS:GFP construct, and demonstrated by histochemical GUS assay that the 35S promoter is active in the secondary xylem tissues (Figure S8). The lack of POPACAULIS5 over-expression in the secondary xylem explains the lack of major phenotypic changes in the xylem tissues of the transgenic trees, but also raises a question on what is the mechanism leading to suppression of POPACAULIS5 transgene expression in xylem tissues but not in leaves. Interestingly, increased expression of POPACAULIS5 was observed in the stem of plants grown in vitro on auxin-containing medium (Figures 2a and 4c), which led us to hypothesize that auxin levels somehow control accumulation of POPACAULIS5 transcripts and question whether the transgenic 35S::POPACAULIS5 trees had lower levels of auxin.
POPACAULIS5 over-expression suppresses endogenous auxin levels in the secondary xylem
IAA levels were measured in leaves and stems of in vitro grown plants and in leaves and secondary xylem tissues from greenhouse-grown trees. Plants grown in vitro showed similar IAA levels in leaves of wild-type and transgenic lines, but a decrease was found in the young stems compared to the wild-type (Figure 5a). In greenhouse-grown trees, IAA levels were strongly reduced in the leaves and in the secondary xylem tissues of the transgenic trees (Figure 5b). This was even more pronounced in lines with more severe reduction in growth. The low levels of auxin and the lack of POPACAULIS5 over-expression in xylem tissues of the transgenic 35S::POPACAULIS5 plants is suggestive of a negative feedback mechanism whereby increased POPACAULIS5 expression functions to reduce IAA levels, which in turn prevents further expression of POPACAULIS5. The reduction in the IAA levels in the 35S::POPACAULIS5 leaves (Figure 5b) most probably reflects functioning of the feedback mechanism in the leaf vasculature, which is the main site of auxin accumulation in this organ (Ljung et al., 2001; Teichmann et al., 2008; Petrásek and Friml, 2009).
We tested our hypothesis by studying whether exogenous auxin affects expression of POPACAULIS5. Auxin levels were modulated in young stem tissues of wild-type and two transgenic lines, B2 and B15, by exogenous application of indolebutyric acid (IBA) (Figure 6a–c). Stem pieces were first depleted of auxin for 16 h. As expected, exogenous supply of auxin resulted in a strong increase in POPACAULIS5 transcript levels after 4 h in both transgenic lines but not in the wild-type. Twenty-four hours after auxin treatment, the transcript levels decreased in the transgenic stems (40 h point; Figure 6b,c). To exclude the possibility that the auxin-induced increase in POPACAULIS5 expression was due to induction of the 35S promoter itself, transgenic Populus trees carrying a 35S::GUS:GFP construct were analysed and shown to be non-responsive to exogenous IBA on the basis of expression analyses of the GUS and GFP genes by quantitative PCR (Figure S9). Together, these findings suggest that auxin stimulates POPACAULIS5 expression at the post-transcriptional level.
Class III HD-Zip PttHB8 over-expression stimulates POPACAULIS5 expression
HD-Zip III family member AtHB8 is a good candidate for control of ACL5 expression in Arabidopsis (Baima et al., 2001; Imai et al., 2006; Carlsbecker et al., 2010). We therefore tested whether the autoregulatory feedback mechanism of ACL5 expression proceeds through AtHB8. In Populus, the closest homologue to AtHB8 is PttHB8 (Ko et al., 2006a). First, an in silico analysis was performed using the PlantPAN database (Chang et al., 2008) to analyse the target promoter region of POPACAULIS5, which resulted in identification of several regulatory homeodomain cis-elements (Figure S10a). The search predicted that the transcription factor ATHB9/PHV, which has high affinity in vitro for the pseudo-palindromic sequence GTAAT(G/C)ATTAC (Sessa et al., 1998), targets POPACAULIS5. Next, the FootprintDB database (Contreras-Moreira, 2010) was interrogated to identify HD-Zip III AtHB8 and PttHB8 DNA-binding signatures. We found that PttHB8 had a similar DNA-binding signature to the one identified for AtHB9/PHV, which suggests that AtHB8/PttHB8 binds to the POPACAULIS5 promoter (Figure S10b,c). In addition, we found other putative regulatory elements, such as an auxin response factor cis-element, which is known to be enriched in the 5′ flanking region of genes up-regulated by IAA (Ulmasov et al., 1999; Goda et al., 2004).
In the in vitro transgenic 35S::POPACAULIS5 lines grown on auxin-containing medium, PttHB8 expression was significantly suppressed compared to the wild-type (Figure 7a). In the greenhouse-grown trees, PttHB8 expression was suppressed in the leaves from most of the lines but was unaltered or slightly up-regulated in xylem tissue in comparison to the wild-type (Figure 7b). These results demonstrate that the POPACAULIS5 over-expression observed in the plants grown in vitro and in the leaves of greenhouse-grown trees (Figure 4a,c) suppresses PttHB8 expression (Figure 7a,b). We also showed that expression of PttHB8 was rapidly induced by exogenous auxin (Figure 6d–f), and it is therefore possible that the suppression of PttHB8 expression by POPACAULIS5 over-expression is mediated through IAA.
To further understand PttHB8 involvement in POPACAULIS5 regulation, we expressed a miRNA165/166 mis-regulated form of PttHB8 under the control of the 35S promoter in hybrid aspen. Three 35S::PttHB8-miRNAd transgenic lines, L175, L176 and L179, were obtained. As previously observed for 35S::POPACAULIS5 trees, up-regulation of PttHB8 occurred only in leaves but not in the stem (Figure 8). This suggests that POPACAULIS5 and PttHB8 expression levels are controlled by the same mechanism. Most importantly, over-expression of PttHB8 in the leaves resulted in increased levels of POPACAULIS5, suggesting that PttHB8 activates expression of POPACAULIS5 either directly or indirectly.
This work focused on POPACAULIS5, which was shown to be the true orthologue of Arabidopsis thermospermine synthase ACL5. Our results provide evidence that thermospermine levels are strictly controlled by a negative feedback mechanism involving POPACAULIS5, auxin and the HD-Zip III transcription factor PttHB8. The evidence for the existence of the feedback mechanism is based on the surprising inability to increase the levels of POPACAULIS5 transcript or thermospermine in xylem tissues by ectopic expression of POPACAULIS5 under the control of the CaMV 35S promoter in transgenic Populus trees (Figure 4a). This inability correlated with reduced accumulation of auxin (Figure 5b), suggesting that over-production of thermospermine suppresses biosynthesis of auxin. This is in accordance with the opposite situation observed in the Arabidopsis acl5 mutant, in which increased auxin levels were present in young seedlings (Vera-Sirera et al., 2010). However, auxin is known to induce expression of ACL5 (Hanzawa et al., 2000; Rambla et al., 2010), and this was also shown for POPACAULIS5 in the Populus stem (Figures 4c and 6b,c). Therefore, thermospermine and auxin are part of a feedback loop that involves a negative effect of thermospermine on auxin and a positive effect of auxin on thermospermine. The presence of a negative feedback loop was first noticed when it was observed that acl5 mutants had increased expression levels of ACL5 (Hanzawa et al., 2000; Imai et al., 2006; Muñiz et al., 2008). Here, we have identified auxin as a mediator of this feedback control.
The negative feedback loop functions to suppress thermospermine levels specifically in secondary xylem tissues, as high expression levels of POPACAULIS5 from the CaMV 35S promoter were observed in the leaves of the transgenic Populus trees (Figure 4a,b). It is probable that the feedback mechanism operates specifically in the xylem vessel elements, due to the fact that both ACL5 and HB8 are specifically expressed in such elements (Baima et al., 2001; Muñiz et al., 2008; Zhang et al., 2011). The absence of the feedback loop in other cell types leads to high levels of thermospermine (Figure 4) that appear to be detrimental to overall growth of the plants (Figures 2 and S3). The decrease in the height of the 35S::POPACAULIS5 trees may be a direct effect on apical growth but also a secondary effect of the smaller leaf size, impaired photosynthetic capacity or impaired water transport capacity of the transgenic trees, for example. This decrease is somewhat surprising considering that suppression of ACL5 expression in Arabidopsis causes dwarfism of the inflorescence stem (Hanzawa et al., 2000; Clay and Nelson, 2005; Muñiz et al., 2008). Hence, one would expect an increase rather than decrease in height of the stem in an ACL5 over-expressor. We can only speculate about the reasons for this, but it is a general phenomenon that plant hormones function in a dose-dependent manner until a threshold level, after which increases in hormonal concentrations become inhibitory (Srivastava, 2002). The thermospermine levels resulting from 35S::POPACAULIS5 expression appear to exceed the threshold level for thermospermine action in control of height growth. It is quite probable that the threshold for optimal thermospermine levels is very low or maybe even very close to zero in cells other than xylem vessel elements as they normally do not synthesize any thermospermine. In conclusion, we propose that any increase in thermospermine levels of non-vessel cells is detrimental for growth, while a decrease in the thermospermine concentration in the xylem vessel elements, such as in the acl5 mutant, also leads to reduced growth of the inflorescence stem due to problems in xylem specification.
Several previous reports have demonstrated the importance of polyamines for cambial development and xylem differentiation (Vera-Sirera et al., 2010; Tisi et al., 2011; Waduwara-Jayabahu et al., 2012). Disruption of ACL5 in Arabidopsis results in over-proliferation of xylem vessel elements with spiral or reticulate secondary wall thickenings, a smaller size of the vessel elements and lack of xylem fibres (Muñiz et al., 2008). However, only modest defects in xylem development were observed in the transgenic 35S::POPACAULIS5 Populus trees. Expansion of the secondary xylem was reduced in the 35S::POPACAULIS5 trees, but this was not surprising considering the severe reduction in the overall height growth of these trees. Despite this, the morphology of xylem elements was scarcely altered, as the size and abundance of the various xylem cell types as well as the type of secondary cell-wall thickenings of vessel elements were similar in wild-type and transgenic trees. The unaltered xylem morphology correlates well with the fact that thermospermine levels were not increased in the secondary xylem tissues of the transgenic trees. It is also possible that the slight defects observed in xylem development are due to increased thermospermine levels in the early stages of plant growth in the auxin-rich environment in vitro (Figure S3). Alternatively, thermospermine levels may fluctuate slightly during growth of the trees, and measurement of the thermospermine levels in the pool of xylem tissues perhaps did not reveal these fluctuations. Fluctuations are expected to occur as a result of the presumable strong over-expression of POPACAULIS5 from the 35S promoter, which must be counteracted by the negative feedback loop. We observed high variation in growth within transgenic lines that may reflect these fluctuations. Similar variation was previously observed within transgenic lines in Populus trees, in which expression levels of the HD-Zip III transcription factor family member POPREVOLUTA were increased (Robischon et al., 2011).
An interesting question is whether the decrease in secondary growth of the stem of the transgenic lines is due to the low IAA levels found there. IAA is known to be a central regulator of cambial growth and xylem specification (Ohashi-Ito and Fukuda, 2010; Ursache et al., 2013), and alterations in levels of auxin have long been known to have severe effects on xylem development (Gälweiler et al., 1998; Hardtke and Berleth, 1998). It was also recently reported that 2,4-dichlorophenoxy acid and other auxin synthetic analogues induce ectopic xylem vessel differentiation in acl5 but not in wild-type Arabidopsis (Yoshimoto et al., 2012a,b). The authors suggested that xylem differentiation is controlled by auxin, and that thermospermine acts to suppress IAA synthesis and/or sensitivity. Our data provide the evidence for suppression of IAA levels by thermospermine. However, we also showed that inclusion of auxin in the growth medium in vitro did not alleviate growth defects but instead further reduced xylem differentiation in the transgenic 35S::POPACAULIS5 trees (Figure S3c,e), supporting the function of thermospermine rather than auxin as the central regulator of xylem differentiation during secondary growth of the stem.
On the basis of our data, we propose operation of a mechanism in secondary xylem tissues to maintain thermospermine at safe levels in order to facilitate its fundamental role in xylem differentiation. In this proposed mechanism, IAA mediates POPACAULIS5 expression through PttHB8, and thermospermine levels feedback control PttHB8 and consequently POPACAULIS5 transcript levels through repression of IAA in a loop (Figure 9). How POPACAULIS5 or thermospermine functions to suppress IAA biosynthesis is not clear currently. Another open question is how the interaction between PttHB8 and POPACAULIS5 takes place. Our results on up-regulation of POPACAULIS5 expression as a result of over-expression of PttHB8, as well as identification of several regulatory homeodomain cis-elements in the promoter of POPACAULIS5 suggests that POPACAULIS5 expression may be under direct transcriptional regulation by PttHB8. Transcriptional control of POPACAULIS5 levels by auxin and PttHB8 does not explain the lower levels of POPACAULIS5 transcript and thermospermine in the secondary xylem tissues of the 35S::POPACAULIS5 trees. We propose therefore that POPACAULIS5 is regulated by auxin in the transgenic 35S::POPACAULIS5 trees through post-transcriptional control of mRNA stability. Hence, POPACAULIS5 over-expression in the xylem elements of the 35S::POPACAULIS5 trees reduces auxin levels, which in turn results in destabilization of the POPACAULIS5 transcripts. The feedback mechanism cannot cope with excessive amounts of auxin, as inclusion of auxin in the in vitro medium resulted in high levels of POPACAULIS5 expression and damaging levels of thermospermine in the 35S::POPACAULIS5 stems. Whether post-transcriptional regulation of POPACAULIS5 also occurs for the endogenous transcript in the wild-type plants is not known, but is feasible as it would allow rapid alterations in thermospermine homeostasis. It is therefore possible that auxin mediates transcriptional activation of ACL5 through HB8, and that the post-transcriptional control allows rapid alterations in thermospermine signalling, especially in situations where IAA concentrations are excessive. Similar kind of mechanisms involving both transcriptional and post-transcriptional control of gene expression by IAA have been shown in the context of other genes as well, such as the AUX/IAA genes (Benjamins and Scheres, 2008). In any case, it is clear from our data that thermospermine levels must be tightly controlled in the secondary xylem tissues of the stem to ensure proper xylem differentiation, and that auxin is a central component in this control.
Plant material, growth conditions and sampling
Hybrid aspen (Populus tremula L× P. tremuloides Michx.; clone T89) was sub-cultured on MS basal salt medium at half-strength (Murashige and Skoog, 1962) (termed auxin-depleted medium). The Populus trichocarpa Nisqually-1 clone was maintained in the greenhouse. Plants were grown in growth chambers at 21°C under a 16 h light/8 h dark photoperiod. Transgenic and wild-type plants were transferred to soil, and trees were grown for 2 months in the greenhouse at 21°C under a 18 h light/6 h dark photoperiod. The greenhouse growth experiment was performed twice.
Sampled tissues were directly frozen in liquid nitrogen when collected and stored at −80°C. Leaves, the first internode (apical stem) and the internode closest to the base (basal stem) of plants grown on auxin-containing medium were collected and pooled in groups of ten from each line for gene expression analysis. Leaves, the stem between the third and the seventh internode from the top (young stem), the stem between the seventh and the basal internode (older stem), and root apices from plants grown in vitro on auxin-depleted medium were ground to powder and portioned for gene expression, polyamine and IAA quantification analyses. For greenhouse-grown trees, the five youngest fully expanded leaves were collected. Secondary xylem tissues were obtained between stem internodes 40 and 45 (from the top) by peeling off the bark and scraping the surface of the frozen woody core until emergence of fully mature wood. As no fully mature wood was present in lines B4 and B13, the whole xylem part of the stem was used. The tissues were ground to powder and portioned for gene expression, polyamine and IAA quantification analyses.
To identify the P. trichocarpa and P. tremula × P. tremuloides putative ACL5 coding regions, POPACAULIS5 and PttACL5, respectively, we performed BLAST/browse searches in various databases [JGI Populus trichocarpa version 1.1 (http://genome.jgi-psf.org/Poptr1_1/Poptr1_1.home.html; Tuskan et al., 2006)], [Phytozome Populus version 2 (http://www.phytozome.net/poplar.php; Goodstein et al., 2012), and Populus DB (http://www.populus.db.umu.se/; Sterky et al., 2004)] using the Arabidopsis ACL5 sequence (GI:145358223; AT5G19530) as the query. Predicted amino acid sequence alignments were performed using ClustalX (http://www.clustal.org/clustal2/) or MUSCLE (http://www.ebi.ac.uk/Tools/msa/muscle/). For the phylogenetic analysis, putative POPACAULIS5 (POPTR_0006s23880, Phytozome Populus version 2; Goodstein et al., 2012; GI:224088768, Genbank), ACL5-like (POPTR_0008s15120, GI:224102051; POPTR_0010s09940, GI:224108055) and P. tremula × P. tremuloides (PttACL5; GenBank accession number JX444689) sequences were used together with predicted sequences from genomes within the rosid clade, comprising representative orders of angiosperms: evolutionary history was inferred using the neighbour-joining method (Saitou and Nei, 1987). Phylogenetic analyses were performed using MUSCLE and MEGA4 (Edgar, 2004; Tamura et al., 2007). The alignment is provided in Figure S1.
Identification of transcription factor-binding sites in the POPACAULIS5 gene promoter
The PlantPAN database (http://plantpan.mbc.nctu.edu.tw/; Chang et al., 2008) was used to identify cis-elements related to HD-Zip III transcription factors in the POPACAULIS5 putative promoter region, ranging from −1 to −3487 bp upstream of the translation starting site. The FootprintDB (http://floresta.eead.csic.es/; Contreras-Moreira, 2010) was scanned using the PtrHB8 protein sequence (POPTR_0006s25390) to identify DNA-binding proteins that bind to similar DNA motifs and to identify the amino acids residues that interact with DNA.
Isolation of POPACAULIS5 and PttHB8 coding regions
Aliquots (1 μg) of total RNA, extracted using an RNeasy plant mini kit (Qiagen, http://www.qiagen.com/) from shoot apices of P. tremula × P. tremuloides and P. trichocarpa, were used for cDNA synthesis using a 1st Strand cDNA synthesis kit for RT-PCR (Roche, http://www.roche-applied-science.com/) and oligo(dT), according to the manufacturer's instructions. A 1028 bp sequence downstream of the start codon of PttACL5 and POPACAULIS5 was amplified from the cDNA of P. tremula × P. tremuloides and P. trichocarpa, respectively, and cloned into the pCR2.1 vector (Invitrogen, http://www.invitrogen.com/). The primers used were POPACAULIS5 forward, 5′-ATGGGTACTGAGGCAGTTGAG-3′ and reverse 5′-CCCACCAGCAAAGGTATGAG-3′. Similarly, a 2817 bp sequence downstream of the start codon of PttHB8 was isolated from hybrid aspen using the primers PttHB8 forward, 5′-ATCTCTAATCCGATCTACGCCAGG-3′ and reverse 5′-GCTCCCAAAGGTTTTTAGGC-3′, by amplification from cDNA and cloned into pCR2.1 vector. Sequence identity was confirmed by sequencing.
Site-directed mutagenesis of the PttHB8 miRNA 165/166 binding site
A site-directed mutagenesis approach was used to prevent miRNA165/166 from cleaving the PttHB8 transcript (Emery et al., 2003; Zhong and Ye, 2004; Kim et al., 2005). In the miRNA binding site of the isolated PttHB8 cDNA sequence, two nucleotides (T and G) were replaced by A nucleotides by PCR amplification. The complete pCR2.1 target plasmid bearing the HD-Zip III isolated sequence was amplified using Phusion Hot-Start DNA polymerase (Finnzymes, http://www.thermoscientificbio.com/finnzymes/). The mutations were introduced using the mutated forward primer 5′-CTGGGATGAAGCCTGGACCAGATTCCATTGG-3′ (point mutations are underlined) and the reverse primer 5′-GCATTTGGACCCACTCCACAGCAGTTCCAGT-3′. The mutated PCR product (named PttHB8-miRNAd) was then re-circularized by ligation using T4 Quick DNA ligase (New England Biolabs, https://www.neb.com/). The synonymous point mutations were confirmed by sequencing, and the PttHB8-miRNAd cDNA sequence was used to construct the vector for over-expression under the control of the constitutive CaMV 35S promoter as described below.
Hybrid aspen transformation
Cloned sequences were sub-cloned into pDONOR221, recombined with the Gateway vector pK7GW2.0 (Invitrogen, http://www.invitrogen.com) for POPACAULIS5, PttACL5 and PttHB8 over-expression (Karimi et al., 2002), and introduced into Agrobacterium tumefaciens strain GV3101 pMP90 (Koncz and Schell, 1986). Hybrid aspen was transformed as previously described (Nilsson et al., 1992). Transformant selection and shoot elongation were performed using MS medium, with 20 g l−1 sucrose, 0.1 μg ml−1 IBA, 0.2 μg ml−1 6-benzylaminopurine, 500 μg ml−1 cefotaxime and 80 μg ml−1 kanamycin monosulfate (auxin-containing medium). To confirm insertions, PCR was performed using the primers POPACAULIS5 forward 5′-ATGGGTACTGAGGCAGTTGAG-3′ and reverse 5′-TCAATTTTTGTTAGCCACCCCATG-3′, PttHB8 forward 5′-ATCTCTAATCCGATCTACGCCAGG-3′ and reverse 5′-GAAAGACAGTGTAAGGAG-3′, 35S forward 5′-CTCATCAAGACGATCTACCCGAG-3′ and reverse, 5′-TGGGCAATGGAATCCGAGGAGGT-3′, NPTII forward 5′-GAATCGGGAGCGGCGATACCGTAAA-3′ and reverse 5′-CAAGATGGATTACACGCAGGTTCTC-3′, and for false-positive screening the virBG forward 5′-GCGGTGAGACAATAGGCG-3′ and reverse 5′-GAACTGCTTGCTGTCGGC-3′ primers. After shoot elongation, plants were transferred to half-strength MS medium for rooting.
Anatomical and ultrastructural analysis
Tree height, internode length, leaf dimensions and stem diameter at internode 35 and stem base (15 cm above soil) were measured. The maturation internode was determined as the youngest internode where the xylem showed signs of complete maturation based on the presence of fully lignified, highly autofluorescing xylem fibres as described by Bollhöner et al. (2012). The 45th internode (from the apex) found below the maturation internode in wild-type, B2 and B14 lines was selected as the reference internode. Viability of the xylary cells was determined by staining 0.5–1 mm stem sections with 10 mg ml−1 nitroblue tetrazolium in buffered succinate (Berlyn and Miksche, 1976; Gahan, 1984). Lignin staining was performed using a saturated solution of phloroglucinol (Sigma, http://www.sigmaaldrich.com/) in 20% HCl (Jensen, 1962). Sections were observed using a Zeiss Axioplan light microscope, and images were captured using a Axioplan digital camera and Axiovison 4.8 software (Zeiss, http://www.zeiss.com/). Measurements were taken at four positions around the circumference of the stem using Axiovision 4.8 software. Stem segments were fixed in FAA [50% (v/v) ethanol, 5% (v/v) acetic acid, 5% (v/v) formaldehyde] overnight, dehydrated in an ethanol series and gradually infiltrated into LR White (TAAB, http://www.taab.co.uk/). A Leica RM2155 microtome (Leica Microsystems, http://www.leica-microsystems.com) was used for sectioning. Sections were heat-fixed to slides, stained with toluidine blue O (Sigma-Aldrich, http://www.sigmaaldrich.com/), and mounted in mounting medium for observations. Tree growth, growth parameter and microscopy analyses were performed twice.
Electron microscopy images of fibre and vessel elements were taken from stem segments from the reference internode, fixed in 2.5% glutaraldehyde in 0.2 m sodium cacodylate buffer, embedded in Spurr resin (Sigma) according to Rensing (2002), and examined using a Hitachi H-7000 transmission electron microscope (Hitachi, http://www.hitachi.com/).
Fibre and vessel element measurements
Stem segments 1 cm long were collected below the reference internode. Pieces of wood were cut to exclude inner pith, outer bark and vascular cambium. The wood samples were immersed in a maceration alkaline solution as described by Berlyn and Miksche (1976). The wooden blocks were mechanically disaggregated. Xylem cell suspensions were observed with a light microscope as described above. Measurements of length and width of at least 200 fibre and 50 vessel elements were performed manually for at least four individual trees, and classified according to the secondary wall thickening patterns (Esau, 1977).
Histochemical GUS staining
Hand sections of stem segments were placed in 90% acetone for 30 min at −20°C, washed twice in distilled water, and incubated in X-Gluc staining solution (1 mm X-Gluc, 1% Triton X-100, 10 mm EDTA, in phosphate buffer) at 37°C in darkness until staining was visible. Stem segments were washed with distilled water, dehydrated in an ethanol series to 50%, fixed for 10 min in 5% formaldehyde/5% acetic acid/50% ethanol, washed with 50% ethanol for 2 min, cleared in 100% ethanol, incubated overnight in 70% ethanol at 4°C, mounted in 50% glycerol, and observed with a Zeiss Axioplan microscope.
Quantitative reverse transcriptase RT-PCR
Total RNA was extracted from 100 mg of frozen powdered tissues from in vitro grown plants using an RNeasy plant mini kit (Qiagen) as described above, and extracted from the tree tissues as described by Chang et al. (1993). cDNA synthesis was performed on 1 μg DNase-treated total RNA using a Transcriptor HF cDNA synthesis kit (Roche) with oligo(dT) primers. Quantitative PCR was performed using a LightCycler 480 PCR system with LightCycler480 SYBR Green I Master Mix (Roche) to monitor double-stranded DNA products. Specific primer pairs were designed to generate amplicons of POPACAULIS5 and PttHB8 (POPTR_0006s25390). Pt1 (POPTR_0002s12910) or CYP2 (POPTR_0009s13270) were used as reference genes (Czechowski et al., 2005; Gutierrez et al., 2008). The primers for the NAC gene family member PNAC058 (POPTR_0013s11740) were as described by Hu et al. (2010). The amount of target transcripts was normalized using the ΔΔCT method (Livak and Schmittgen, 2001). For all experiments, the mean of triplicate quantitative PCR reactions was determined, and at least three biological replicates or pooled biological samples were used. The experiments were performed at least twice. The primers used were: POPACAULIS5 forward 5′-AAGATGCAGAGTGCCGAAGT-3′ and reverse 5′- GACTTGTGCTTGAGGGCTTC-3′, PttHB8 forward 5′-ATCTCTAATCCGATCTACGCCAGG-3′ and reverse 5′-CGCATAGAGCTTGGCTTAGG-3′, Pt1 forward 5′-GCGGAAAGAAAAACTGCAAG-3′ and reverse 5′-TGACAGCACAGCCCAATAAG-3′, and CYP2 forward 5′- TAAGACCGAATGGCTTGACG-3′ and reverse 5′- AGAACGCACCCCAAAACTACTA-3′.
Quantification of polyamines
Polyamines were extracted from approximately 100 mg of frozen tissues collected as described above, purified as described previously (Rambla et al., 2010), derivatized as described previously (Fernandes and Ferreira, 2000), and identified and quantified as described by Rambla et al. (2010). Representative mass spectra for the heptafluorobutyric derivatives of thermospermine, spermine, spermidine and putrescine are shown in Figure S11.
POPACAULIS5 thermospermine activity assays in yeast
The POPACAULIS coding sequence was extracted from the pCR2.1-POPACAULIS5 plasmid as an EagI fragment, and cloned into yeast expression vector pCM190 (Gari et al., 1997). The pCM190-POPACAULIS5 vector and control empty vector were introduced into yeast using the Yeastmaker yeast transformation system 2 (Clontech, http://www.clontech.com/). After lysis by intense vortexing with 100 μl of 0.5 mm diameter glass beads, polyamine levels in yeast extracts were determined by gas chromatography-mass spectrometry as described above.
Quantification of IAA
Tissues from trees and from in vitro grown plants were collected as described above, and 10–20 mg were used for quantification of free IAA content. Sample extraction and purification was performed as described by Andersen et al. (2008), with 500 pg 13C6-IAA internal standard being added to each sample before extraction. After derivatization, the samples were analysed by gas chromatography-selected reaction monitoring-mass spectrometry as described previously (Edlund et al., 1995).
Auxin treatments for expression analysis
Stem segments 3 cm long from 5-week-old in vitro grown wild-type and 35S::POPACAULIS5 transgenic lines were cut between internodes. Six segments from six individual plants were immediately frozen after cutting, representing the pooled control (0 h). All remaining stem segments were placed on auxin-free half-strength MS medium to deplete them from auxin for 16 h, after which pools of six segments derived from six individual plants were grouped and sampled (16 h). Half of the remaining segments were then placed in fresh half-strength MS medium (mock) and the other half were placed in the same medium containing 20 μm IBA, from which further sets of six stem segments, derived initially from six individual plants, were pooled after mock/IBA treatment for 4 h (20 h point), 24 h (40 h point) and 32 h (48 h point). As negative controls for the auxin treatment experiments, stems from two 35S::GUS:GFP Populus lines were used to monitor the response to 2 and 20 μm IBA. Total RNA, cDNA synthesis and quantitative RT-PCR were performed as described above. The experiment was performed twice.
The non-parametric Mann–Whitney U test was used to assess significant differences in gene expression, polyamine and IAA contents and tree growth parameters. An α value of 0.05 was considered significant. Statistical analyses were performed using Statistica software (Statsoft Inc., http://www.statsoft.com/) as described by Zar (1998).
We thank Brian Jones (University of Sydney, Australia/Umeå Plant Science Centre, Sweden) for the T89 clone, Max Cheng (University of Tennessee, Department of Plant Sciences, USA) for the P. trichocarpa Nisqually-1 clone, Veronica Bourquin and Lenore Johansson (Umeå Plant Science Centre, Sweden) for assistance with microscopy, and Alexander Makoveychuk (Umeå Plant Science Centre, Dept. Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, Sweden) for providing the 35S::GUS:GFP lines. This research was supported by the Fundação para a Ciência e a Tecnologia through projects PEst-OE/EQB/LA0004/2011 and PTDC/AGR-GPL/098369/2008, and grants SFRH/BD/30074/2006 (to A. Milhinhos) and SFRH/BD/78927/2011 (to A. Matos), the Swedish Research Council Formas (to H. Tuominen), the Swedish research council VR and the Swedish Governmental Agency for Innovation Systems Vinnova (to H. Tuominen), and Spanish Ministry of Economy and Innovation grant BIO2011-23828 (to J. Carbonell).