Strigolactone-Based Node-to-Bud Signaling May Restrain Shoot Branching in Hybrid Aspen.

The biosynthesis and roles of strigolactones (SLs) have been investigated in herbaceous plants, but so far, their role in trees has received little attention. In this study, we analyzed the presence, spatial/temporal expression and role of SL pathway genes in Populus tremula � Populus tremuloides. In this proleptic species, axillary buds (AXBs) become para-dormant at the bud maturation point, providing an unambiguous starting point to study AXB activation. We identified previously undescribed Populus homologs of DWARF27 (D27), LATERAL BRANCHING OXIDOREDUCTASE (LBO) and DWARF53-like (D53-like) and analyzed the relative expression of all SL pathway genes in root tips and shoot tissues. We found that, although AXBs expressed MORE AXILLARY GROWTH1 (MAX1) and LBO, they did not express MAX3 and MAX4, whereas nodal bark expressed high levels of all SL biosynthesis genes. By contrast, expression of the SL perception and signaling genes MAX2, D14 and D53 was high in AXBs relative to nodal bark and roots. This suggests that AXBs are reliant on the associated nodes for the import of SLs and SL precursors. Activation of AXBs was initiated by decapitation and single-node isolation. This rapidly downregulated SL pathway genes downstream of MAX4, although later these genes were upregulated coincidently with primordia formation. GR24-feeding counteracted all activation-related changes in SL gene expression but did not prevent AXB outgrowth showing that SL is ineffective once AXBs are activated. The results indicate that nodes rather than roots supply SLs and its precursors to AXBs, and that SLs may restrain embryonic shoot elongation during AXB formation and para-dormancy in intact plants. � The Author(s) 2019. Published by Oxford University Press on behalf of Japanese Society of Plant Physiologists.

Introduction

In deciduous trees, crown architecture arises through the coordinated action of terminal and axillary meristems (AXMs). In contrast to annuals, like Arabidopsis (Grbić and Bleecker 2000, Long and Barton 2000, Greb et�al. 2003), the AXMs of deciduous trees arise in the axils of emerging leaves and produce axillary buds (AXBs) with bud scales (Garrison 1955). The timing and pattern of branch formation reflect different branching styles. In sylleptic branching, newly formed AXBs produce branches in the same season, whereas in proleptic branching they may produce them only in the following seasons (Hall� et�al. 1978, Ceulemans et�al. 1990, Wu and Stettler 1998, Barth�l�my and Caraglio 2007). The sylleptic branching pattern is strongly influenced by the prevailing environmental conditions, revealing considerable plasticity in architectural design. On the other hand, in proleptic species AXB outgrowth is postponed to the next growing season, resulting in a more robust branching pattern (Cline 1997).

Hybrid aspen (Populus tremula � Populus tremuloides, clone T89) is a typical proleptic species. AXBs develop during the growing season until the dwarfed side shoot, enclosed by five scales, has produced about 10 embryonic leaves. This point is referred to as the bud maturation point (BMP; Rinne et�al. 2015). These mature AXBs partially dehydrate and remain para-dormant, at least until the next growing season, but they can be activated expeditiously by decapitation. This allows the investigation of processes that exclusively relate to AXB activation, and not to AXB formation. In sylleptic tree species, where branches are initiated in the same season, such unambiguous starting point is lacking.

In woody perennials, very little is known about the molecular processes that control branching. By contrast, these processes are under intensive investigation in herbaceous annuals, like Arabidopsis and pea, as well as in the monocot rice (Sorefan et�al. 2003, Domagalska and Leyser 2011, Wang and Li 2011). They show that the AXB activation is regulated by a network of interacting hormones. Although auxin and cytokinins are the classic branching hormones (King and Van Staden 1988, M�ller and Leyser 2011), recent work with woody species shows that gibberellins (GA) also play a role (Ni et�al. 2015, Rinne et�al. 2016). Crucial newcomers in this network are carotenoid-derived terpenoid lactones, referred to as strigolactones (SLs) that suppress branching (Gomez-Roldan et�al. 2008, Umehara et�al. 2008, Ferguson and Beveridge 2009).

So far, all natural SLs have been isolated from root exudates and identified based on their capacity to stimulate germination of parasitic plant seeds (Kobae et�al. 2018). The first SL, identified in root exudates of cotton, was named strigol because it stimulated the germination of witchweed (Striga lutea Lour) seeds (Cook et�al. 1966, Cook et�al. 1972). When Striga seeds are in close proximity of the roots of a strigol exuding host plant, they will germinate and parasitize the plant. The finding that not only host plants but also non-hosts like cotton exuded SL-like compounds to the rhizosphere, indicated that SLs had some distinct function unrelated to parasitic seed germination (Wang and Bouwmeester 2018). Such non-host SL exudation was found to attract arbuscular mycorrhizal (AM) fungi to colonize plant roots, particularly under conditions of phosphate starvation (Yoneyama et�al. 2007, L�pez‐R�ez et�al. 2008, Carbonnel and Gutjahr 2014). In an established symbiotic relationship, the AM fungi deliver phosphate to the plant, while in return the plant provides sugars (Akiyama et�al. 2005, Besserer et�al. 2006).

In addition to inhibiting shoot branching and attracting AM fungi, SLs have crucial roles in secondary growth, root development and leaf senescence (Gomez-Roldan et�al. 2008, Umehara et�al. 2008, Kapulnik et�al. 2011, Rasmussen et�al. 2012, Yamada et�al. 2014). The role of SL in shoot branching has been analyzed in branching mutants of Arabidopsis (Sorefan et�al. 2003, Booker et�al. 2004, Booker et�al. 2005, Domagalska and Leyser 2011, Seto and Yamaguchi 2014), pea (Beveridge et�al. 1997, Hamiaux et�al. 2012), petunia (Drummond et�al. 2009, Hamiaux et�al. 2012) and rice (Wang and Li 2011, Zhang et�al. 2014).

A generalized SL pathway can be subdivided into three distinct parts, which are spatially separate: the plastid, the cytoplasm/symplasm and the nuclei of cells in the target areas (Fig.�1). In the plastid carotenoid pathway (Matusova et�al. 2005), all-trans-β-carotene is converted to the biosynthetic intermediate carlactone (CL), which is exported to the cytoplasm (Alder et�al. 2012, Kobae et�al. 2018, Yoneyama et�al. 2018). CL biosynthesis involves three important classes of plastid enzymes. In Arabidopsis, these include the isomerase DWARF27 (D27), and two carotenoid cleavage dioxygenases (CCD7 and CCD8), encoded by MORE AXILLARY GROWTH3 (MAX3) and MORE AXILLARY GROWTH4 (MAX4), respectively.

Fig. 1

Generalized scheme of SL biosynthesis and signaling. The schema envisions three compartments: the biosynthetic compartment of the plastid (green) where CL is produced, the cytoplasmic and the symplasmic compartment (light blue) where excreted CL is converted to CLA/4DO by MAX1, and the nucleus (grey) where perception occurs. CLA is converted to MeCLA, and further by LBO. The SL-like compounds downstream of MeCLA and 4DO (stippled line) are imported into the nuclei of target cells, where AtD14/D14 interacts with the F-box protein MAX2/D3 in an SL-dependent manner to ubiquitinate and degrade the transcription repressor SMXL/D53, resulting in expression of BRC1.

CL is a chemically stable and graft-transmissible intermediate that must be converted by the ER-anchored enzyme MORE AXILLARY GROWTH1 (MAX1; Cytochrome P450) to carlactonoic acid (CLA; Abe et�al. 2014) or 4-deoxyorobanchol (4DO; Alder et�al. 2012). CL and CLA are non-canonical SLs that possess the essential enol ether-D-ring moiety required for biological activity (Zwanenburg et�al. 2009) but not the complete ABCD ring system found in canonical SLs (Yoneyama et�al. 2018). CLA, the universal precursor of a variety of species-dependent SLs (Iseki et�al. 2018), is methylated to methyl carlactonoate (MeCLA) in roots and shoots by an unidentified enzyme (Abe et�al. 2014, Iseki et�al. 2018, Yoneyama et�al. 2018). In Arabidopsis, MeCLA is a substrate of the 2-oxoglutarate-dependent dioxygenase LATERAL BRANCHING OXIDOREDUCTASE (LBO), which oxidizes MeCLA to a compound referred to as MeCLA+16D (Brewer et�al. 2016).

SL perception requires the F-box protein MAX2 and the unconventional hormone receptor DWARF14 (D14), a protein of the α/β-fold hydrolase superfamily. In Arabidopsis, as well as rice, SL triggers interactions among D14, MAX2 and SMXL/D53 in the nuclei of target cells (Zhou et�al. 2013, Zhao et�al. 2015, Liang et�al. 2016, Yao et�al. 2016). In rice, the D53 protein was identified as a repressor of the SL signaling pathway, which is targeted for degradation after SL treatment (Jiang et�al. 2013). The rice F-box protein DWARF3 (D3, ortholog of Arabidopsis MAX2) plays a crucial role in mediating this degradation. It requires D14 to ubiquitinate D53 for degradation by the D14-SCFD3 ubiquitin ligase, to promote SL signaling and responses (Jiang et�al. 2013, Zhou et�al. 2013). A downstream target of SL signaling is the gene BRANCHED1 (BRC1)/TEOSINTE BRANCHED1 (TB1), which encodes a transcription factor that suppresses shoot branching (Doebley et�al. 1997, Aguilar-Mart�nez et�al. 2007, Finlayson 2007, Finlayson et�al. 2010, Seale et�al. 2017).

Although in annuals SL biosynthesis and signaling genes are largely conserved (Yao et�al. 2018), in woody perennials their presence and function remain mostly unexplored. Given the distinct initiation, development and composition of AXBs in hybrid aspen (Rinne et�al. 2015), it is uncertain if the complete pathway is present and functionally conserved in Populus. So far, few SL pathway genes have been identified in perennials (Wang and Li 2006, Czarnecki et�al. 2014, Zheng et�al. 2016). However, it has been reported that Populus root exudate contains 4DO, a canonical SL, and the non-canonical SLs CLA and MeCLA (Xie 2016). Tentative evidence indicates that the inhibition of shoot branching is mediated by non-canonical SLs (Yoneyama et�al. 2018). Indeed, grafting experiments with pea, Arabidopsis and petunia showed that root-produced CL can be imported by the shoot (Beveridge et�al. 2000, Morris et�al. 2001, Turnbull et�al. 2002). However, to inhibit branching in Arabidopsis it must be converted by MAX1 to CLA, as CL is ineffective in max1 mutants (Scaffidi et�al. 2013). In addition, these studies showed that SL biosynthesis genes can also be expressed in shoots, as a wild type scion on an SL-deficient mutant stock does not display a branching phenotype. However, so far SL-like compounds have not been isolated from shoots, indicating that their levels may be very low (Kobae et�al. 2018).

In AXBs of hybrid aspen, two MAX1 orthologs and two orthologs of the SL target gene BRC1 are expressed (Rinne et�al. 2015). All four genes were upregulated during AXB development, reaching their highest levels in mature AXBs, whereas decapitation at the BMP downregulated them in the proximal AXBs (Rinne et�al. 2015). In agreement with this, in Populus � canescens, knockdown of SL biosynthesis genes reduced BRC1 expression and induced branching, like knockdown of BRC1 and BRC2 (Muhr et�al. 2016, Muhr et�al. 2018). Together, these findings suggest that at least part of the SL biosynthesis and signaling genes as well as downstream targets are functional in Populus.

Our first aim was to investigate whether close homologs of the Arabidopsis and rice SL pathway genes (Fig.�1) were present in the Populus trichocarpa genome (Tuskan et�al. 2006), and if and where they were expressed in hybrid aspen. In addition, we aimed to assess whether the unique lifestyle of woody perennials and their complicated bud structure would put different demands on the spatial layout of the SL biosynthesis and signaling paths. Here, we, identified all SL pathway genes in the P. trichocarpa genome, and analyzed their expression in roots and shoot tissues of the non-branching hybrid aspen seedlings (Fig.�2). Nodal bark, rather than root tips, appeared to be major hubs for SL biosynthesis, whereas the AXBs were dominant centers of SL perception. Decapitation-activated AXBs rapidly downregulated SL pathway genes coincident with the start of embryonic shoot (ES) elongation, suggesting that SL inhibits this elongation in intact plants.

Fig. 2

Cartoon depicting the position of young and mature AXBs. The developing young AXBs become para-dormant at the BMP. Sink node and source node denote bark tissue, isolated from the nodes of young and mature AXBs, respectively (hatch pattern). Root material was isolated exclusively from root tips.

Results

Expression of SL biosynthesis genes in hybrid aspen

The first enzyme in the SL biosynthesis pathway is isomerase D27 (Fig.�1), but so far it has not been reported for woody perennials. We identified three close homologs of the Oryza sativa D27 gene (Lin et�al. 2009) in the P. trichocarpa genome (Supplementary Fig. S1), and named them D27a, D27b and D27c. Transcripts of D27a and D27c were expressed in most plant parts, whereas D27b was undetectable. D27a transcript levels were higher than those of D27c, in the apex and particularly in the sink and source leaves (Fig.�3A). Remarkably, in roots, thought to be the major source of SL, D27a transcripts were undetectable, and D27c expression was also very low (Fig.�3A). However, roots expressed MAX3 and MAX4, the two downstream SL biosynthesis genes that mediate CL production (Figs.�1, 3B). By contrast, the expression of MAX3 and MAX4 was virtually absent in developing and mature AXBs, but surprisingly the associated nodes expressed both genes at high levels (Fig.�3B). The expression of MAX3 was higher than that of MAX4 in both sink and source nodes. The sink nodes, which support the young developing AXBs, expressed both genes at very high levels (Fig.�3B, inset). MAX1 genes were expressed in all plant parts, including the AXBs (Fig.�3C). However, because the AXBs themselves did not express MAX3 and MAX4, MAX1 must serve to convert imported CL. As the expression of MAX3 in both sink and source nodes, and MAX4 in sink nodes, were at exceptionally high levels compared to roots (Fig.�3B), the AXBs of hybrid aspen are likely to import CL from the nodes rather than from the roots.

Fig. 3

Expression of SL biosynthesis genes in different plant parts. Expression (fold change) was analyzed in mature AXBs, young AXBs, apex, root, sink node, source node, sink leaves and source leaves. (A) D27a and D27c. (B) MAX3 and MAX4. (C) MAX1.1 and MAX1.2. (D) LBO. (E) D14a and D14b. (F) MAX2a and MAX2b. (G) D53-like1 and D53-like3. (H) BRC1 and BRC2. Values represent the means of three biological replicates �SE (n = 6 plants). nd, not detected. The expression value of the mature AXBs or roots was set at 1.

In Arabidopsis, a downstream product of CLA is the methyl ester MeCLA (Fig.�1), which can directly interact with the SL signaling component D14 (Abe et�al. 2014). However, MeCLA is also substrate for LBO (Fig.�1), and conversion into other SL-like compounds might be required for at least some of its bioactivity (Brewer et�al. 2016). To date, no information is available about its precise role, and whether it is conserved in woody species.

To identify the LBO gene, we searched the P. trichocarpa genome for a putative ortholog of AtLBO (encoded by locus At3g21420) and identified a protein encoded by Potri.010G023600 as PtLBO. The number of amino acids in PtLBO (364 aa) is identical to that in AtLBO (Supplementary Fig. S2) and exhibits 84% similarity and 66% identity at the amino acid level. LBO was expressed throughout the plant, including AXBs, but the highest relative expression was found in source nodes, followed by sink nodes and roots (Fig.�3D).

Expression of SL signaling genes in hybrid aspen

In Arabidopsis, the α/β-hydrolase D14 and the F-box protein MAX2 are essential components in the SL-dependent suppression of AXB outgrowth. D14, thereby, functions as an SL receptor with catalytic activity. Although D14 is localized in the cytoplasm and nucleus (Chevalier et�al. 2014), the nuclear pool is responsible for D14 function (Liang et�al. 2016). SL triggers the physical interaction among nuclear-localized D14, MAX2 and SMXL7/D53 in the nuclei of target cells, resulting in degradation of SMXL7/D53 (Liang et�al. 2016). D14 as well as MAX2 homologs have been identified previously in a Populus species (Czarnecki et�al. 2014, Zheng et�al. 2016), but their tissue-specific expression and role in AXBs have not been investigated. The present data show that in hybrid aspen all plant parts expressed D14a and D14b. Transcript levels in AXBs were two to three times higher than in roots, while levels in the corresponding nodes were somewhat lower than in roots (Fig.�3E). A similar trend was found for MAX2a and MAX2b transcripts, although here the lowest expression level was in roots instead of nodes (Fig.�3F). The relative expression of both signaling genes, MAX2 and D14, was highest in AXBs. However, MAX2b was also well expressed in source tissues (Fig.�3E, F). Thus, although the production of SL-like compounds downstream of CL occurs predominantly in both sink and source nodes, SL perception appears particularly dominant in AXBs (Fig.�3E–G).

Downstream targets of SL signaling in hybrid aspen

In rice, enhanced SL signaling results in the proteasomal degradation of the OsD53 (Fig.�1), a suppressor of SL signaling, resulting in inhibition of AXB activation and outgrowth (Jiang et�al. 2013, Zhou et�al. 2013). Using phylogenetic analysis, we identified three P. trichocarpa homologs of OsD53, which we named D53-like1, D53-like2 and D53-like3 (Supplementary Fig. S6). All three genes were expressed throughout the plant, with the possible exception of roots in the case of D53-like1 and D53-like2 (Fig.�3G;Supplementary Fig. S3A). As D53-like2 was unresponsive to decapitation, we considered it not relevant for branching (Supplementary Fig. S3B). Although hardly expressed in roots, AXBs and their associated nodes expressed D53-like1 and D53-like3 at appreciable levels (Fig.�3G). Expression in the apex was about half of that in AXBs and nodes.

Among the downstream targets of SL in Arabidopsis is the branch-inhibitor gene BRC1 (Fig.�1), which encodes a class II TB1 CYCLOIDEA PCF (TCP) type transcription factor (Aguilar-Mart�nez et�al. 2007, Finlayson 2007) that represses cell proliferation (Schommer et�al. 2014). As we showed previously, hybrid aspen has two BRC genes, BRC1 and BRC2, which are upregulated in developing AXBs (Rinne et�al. 2015). Here, we confirm that BRC1 and BRC2 are highly expressed in AXBs, but that their relative expression elsewhere in the plant was very low or undetectable, except for BRC2 in the shoot apex (Fig.�3H). This suggests that SL signaling targets BRC1 and BRC2 in the dwarfed side shoots of the AXBs to inhibit outgrowth.

Decapitation-induced developmental changes in AXBs

To assess the role of SL biosynthesis and signaling in the activation of mature, developmentally inactive AXBs, plants were decapitated at the BMP. Changes in gene expression were analyzed in the AXB proximal to the decapitation point. To provide context to these gene expression changes, we investigated the time-frame of decapitation-induced developmental changes in the proximal AXB (Fig.�4). The lengths of the AXBs and ESs were measured, and the number of embryonic leaves counted at regular intervals post decapitation (Fig.�4B). The length of the proximal AXB increased gradually after decapitation, and the increase was statistically significant after 48 h. The elongation of the ES shoot followed a similar pattern, albeit a statically significant increase occurred 1 d earlier (Fig.�4B). The number of embryonic leaves was constant over the entire 96-h period, showing that no neo-formed leaves were produced (Fig.�4B). Together, the data show that decapitation-induced changes in gene expression during the first 48 h clearly relate to elongation of the ES stem, and not to the formation of new leaves at the shoot apical meristem (SAM) of the ES.

Fig. 4

AXB size, ES length and the number of embryonic leaves (EL) in the AXB proximal to the plant decapitation point. (A) The size of the AXBs was measured as indicated (L, upper panel). ES length was measured from the surface of the SAM to the basal line connecting the outer bud scales (middle panel). EL number, as depicted, was counted in fixed AXBs (lower panel). (B) Upper row photographs show AXB sizes, measured at 0, 2, 6, 12, 24, 48, 72 and 96 h post decapitation. The lower row shows ES lengths at the same time points. The line graphs show AXB size (left), ES length (middle) and EL number (right) at the indicated times after plant decapitation. Different letters indicate statistical differences between the time points (one-way ANOVA with post hoc Fisher’s LSD test; P-value at least <0.05; NS, not significant). Scale bars, 1.0 mm.

Post-decapitation expression of SL biosynthesis and signaling genes

To pinpoint the role of SL biosynthesis and signaling in the early activation events of the proximal, mature AXBs, we restricted our analyses to the genes that were expressed in the AXBs themselves (Fig.�3). The expression of D27, MAX1, LBO, D14, MAX2 and D53-like genes, as well as the downstream target genes BRC1 and BRC2, was analyzed in AXBs during the critical 0–48 h post-decapitation period (Fig.�5).

Fig. 5

Expression of SL pathway genes in AXBs proximal to the plant decapitation point. Gene expression (fold change) was analyzed at 0, 2, 6, 12, 24 and 48 h post decapitation. (A) D27a and D27c. (B) MAX1.1 and MAX1.2. (C) LBO. (D) D14a and D14b. (E) MAX2a and MAX2b. (F) D53-like1 and D53-like3. (G) BRC1 and BRC2. Values represent the means of three biological replicates �SE (n = 6 plants). Values were calculated relative to the AXBs at t = 0, set at 1. One-way ANOVA (P-value; NS, not significant). Asterisks indicate the first significant change in gene expression (Fisher’s LSD test; P-value at least <0.05).

The expression of D27a and D27c was somewhat reduced between 6 and 12 h after decapitation, and thereafter gradually recovered (Fig.�5A), although these changes were not statistically significant. As MAX3 and MAX4 were not expressed in AXBs of intact plants (Fig.�3B), the modest decapitation-induced alterations in the two D27 genes might not relate SL-mediated events in the AXBs. MAX1.1 and MAX1.2 expressions, and putative CLA production, were significantly reduced by decapitation between 2 and 6 h (Fig.�5B). LBO expression showed a statistically significant increase that started between 12 and 24 h (Fig.�5C).

Although D14a and D14b were specifically expressed at high levels in all AXBs of intact plants (Fig.�3E), decapitation significantly reduced transcript levels in the AXB proximal to the decapitation point (Fig.�5D). The transcript levels of both D14 genes declined significantly between 2 and 6 h, and onward. Although D14b expression diminished more gradually, both D14 genes had the same low level at the 48 h time point (Fig.�5D). The expression of F-box genes MAX2a and MAX2b also decreased relatively early, between 2 and 6 h post decapitation, although MAX2a expression tended to recover (Fig.�5E). Of the two D53-like genes, D53-like1 expression was significantly reduced by decapitation between 0 and 2 h. By contrast, the decrease in D53-like3 expression was only transient, and it increased significantly between 12 and 24 h (Fig.�5F).

The downstream target of SL signaling, BRC1, which was highly expressed in mature AXBs of intact plants (Fig.�3H), was rapidly and strongly downregulated after decapitation between 0 and 2 h in the proximal AXB. BRC2 expression was more gradually and modestly reduced (Fig.�5G).

Taken together, the downregulation of MAX1.1, MAX1.2, D14a, D14b, MAX2b, D53-like1, BRC1 and BRC2 as well as the upregulation of LBO (Fig.�5) preceded the initial phase of ES stem elongation, and the subsequent neo-formation of leaves (Fig.�4).

Developmental changes in AXBs of GR24-treated single-node systems

As we found that SL pathway dynamics within the AXB-node complex reflected the transition from inactivity to activation, we hypothesized that an increase in SL content will prevent AXB activation. To investigate the effect of SL application, we used single-node systems to xylem-feed the SL analog GR24 into AXBs (Fig.�6). These systems, commonly used to study bud burst (Rinne et�al. 2011, Brewer et�al. 2015, Rinne et�al. 2016, Seale et�al. 2017, Xie et�al. 2017), are particularly useful in woody perennials where direct application to the buds is ineffective. Xylem-feeding also enables the investigation of AXB activation independent from the constraints of apical dominance and leaf- or root-derived signals.

Fig. 6

AXB size, ES length and the number of embryonic leaves (EL) in young and mature AXBs (YB, MB) of single-node systems in water and GR24. Lengths of YB, MB and ES, as well as EL counts, were obtained as indicated in Fig. 4A. YBs and MBs: upper row photographs show example AXBs, and lower row photographs show longitudinally cut example AXBs at given days post isolation. The three line graphs show AXB size (left), ES length (middle) and EL number (right) at the indicated times after treatment with water and GR24. Insets: repeats of day 7. Different letters and asterisk indicate statistical differences (one-way ANOVA with post hoc Fischer’s LSD test; P-value at least <0.05). Scale bars, 1.0 mm.

GR24 (10 �M) was fed into the internode base of single-node systems for 3, 5 or 7 d. At day 3 (72 h), the young as well as the mature AXBs that were kept on water (controls) were already enlarging, while GR24-treated young and mature AXBs were slightly less elongated. However, at days 5 and 7 the effect was reversed, particularly in the case of mature AXBs (Fig.�6). The 7-d time point was repeated in a separate experiment, with a similar result. However, in both experiments, the promoting effect of GR24 on AXB size was not statistically significant (Fig.�6, inset). The sturdy outer scale of the mature AXBs did not elongate much, and AXB enlargement at day 7 was mostly due to the protrusion of the inner scales from the tip of the buds (Supplementary Fig. S4). GR24-feeding had a similar but more pronounced effect on the elongation of the ES. At the 7-d time point, the ESs of GR24-fed mature AXBs were significantly longer than the controls. The 7-d time point was repeated in a separate experiment, confirming that GR24 could enhance ES elongation once AXBs were activated (Fig.�6, inset).

Notably, in mature AXBs the number of embryonic leaves had increased from 10 to 12 by day 5 (120 h), and at day 7 (168 h) several additional primordia had emerged, with or without GR24. Young AXBs possessed fewer embryonic leaves at the time of single-node isolation, but also here the number rose steadily without any visible interruption. Although the GR24-fed young AXBs appeared to slightly delay leaf initiation, the differences in leaf numbers were not statistically significant (Fig.�6, inset).

Effects of GR24 on gene expression in AXBs of single-node systems

As GR24-feeding only affected the elongation of the ES in a statistically significant way, the early changes in gene expression must relate to ES elongation. Here, we investigated, how GR24-feeding would affect the SL pathway genes (Fig.�1) in mature and young AXBs. In the decapitation experiments, we probed the early changes in gene expression in the period preceding primordia formation (0–48 h). As in single-node systems, primordia formation started after day 3, we analyzed gene expression during an extended time-frame, including day 0, day 3 (72 h) and day 5 (120 h) (Fig.�7).

Fig. 7

Expression of SL pathway genes, downstream targets BRC1 and BRC2, and the modulating effect of GR24. Young AXBs (YB, left column) and mature AXBs (MB, right column) on single-node systems in water control (–▪–) and 10 �M GR24 (‐‐▪‐‐). Gene expression was analyzed at 0, 3 and 5 d post isolation. (A) MAX1.1 and MAX1.2. (B) D14a and D14b. (C) MAX2a and MAX2b. (D) D53-like1 and D53-like3. (E) BRC1 and BRC2. Values represent the means of three biological replicates �SE (n = 6 plants). Values were calculated relative to the AXBs at t = 0, set at 1. Two-way ANOVA (P-value shows statistical significance between treatments; NS, not significant). Asterisks indicate significant differences with day 0 within each treatment (Fischer’s LSD test; P-value at least <0.05).

Whereas the MAX1.1 and MAX1.2 were highly expressed in young and mature AXBs (Fig.�3C), both genes were downregulated in AXBs in water, like in decapitation, except for MAX1.2 in young AXBs. In all cases, GR24-feeding counteracted the change in expression (Fig.�7A), suggesting feedback on MAX1 gene expression. Both, the downregulation and the counter-effect of GR24, were statistically significant for MAX1.1. In both young and mature AXBs, D14a and D14b were significantly upregulated without GR24, unlike in decapitation, while GR24-feeding repressed this completely (Fig.�7B). In young AXBs, MAX2a and MAX2b showed a similar response as the two D14 genes in that both were significantly upregulated in water, while GR24 prevented this increase (Fig.�7C). Although in mature AXBs expression of MAX2a and MAX2b only slightly increased in the controls, GR24 had a statistically significant reducing effect on MAX2a (Fig.�7C).

D53-like1, encoding a putative repressor of SL signaling, was significantly downregulated in mature as well as in young AXBs, but GR24 prevented this decrease (Fig.�7D). This is in line with the decapitation experiments, where D53-like1 was significantly downregulated already at day 1 and continued to decline up to 48 h (Fig.�5). Conversely, D53-like3, which is more closely related to AtD53 than to OsD53 (Supplementary Fig. S6), was upregulated in both young and mature AXBs in water, but GR24 prevented it in both cases (Fig.�7D). The increased expression of D14a, D14b, MAX2a and MAX2b in AXBs without GR24 could indicate that SL perception increased in response to diminished signal supply, reflecting homeostasis because GR24-feeding prevented upregulation of these genes. By contrast, the downstream target genes BRC1 and BRC2 were not significantly affected, except for BRC1 in young AXBs.

In summary, the GR24-induced changes in expression of SL pathway genes in young and mature AXBs were quite similar, suggesting that the developmental stage is less important for the activation response. Although MAX1.1, MAX1.2 and D53-like1 were downregulated in the controls, D14a, D14b, MAX2a, MAX2b and D53-like3 were upregulated. The only exception appeared to be MAX1.2 in young AXBs, as it was not downregulated in controls. GR24-feeding counteracted these changes in all cases.

Discussion

The role of SL in shoot branching has been explored mainly in herbaceous plants (Gomez-Roldan et�al. 2008, Umehara et�al. 2008, Bennett and Leyser 2014, Marzec 2016, Waters et�al. 2017, Barbier et�al. 2019). This has yielded a wealth of data, showing that the studied species share the SL biosynthesis pathway that produces the universal precursor CLA, which is further converted to canonical and non-canonical SLs in a species-dependent fashion (Xie 2016, Iseki et�al. 2018). The physiological relevance of this diversity has remained unclear (Zwanenburg and Blanco-Ania 2018). Although both canonical and non-canonical SLs are found in root exudates (Xie 2016, Iseki et�al. 2018), tentative evidence shows that the SLs that regulate shoot branching are non-canonical (Yoneyama et�al. 2018). How plants spatially and temporarily control the biosynthesis of the SLs that are involved in shoot branching has not been adequately addressed and remains an important research target (Kameoka and Kyozuka 2018). Especially, there is a need to better understand branching in trees, because their architecture, coupled to their superior CO2 capture, is a critical element in mitigating climate change. To obtain more insight into the role of SLs in tree branching, we addressed the following questions. Are SL pathway and signaling genes conserved in the Populus genome? Is their expression spatially and functionally differentiated? Is the entire SL pathway operational in AXBs, independent of roots and shoot? Are SL biosynthesis and homeostasis affected by decapitation and GR24-feeding?

In addition to previously identified Populus homologs of SL biosynthesis and signaling genes (Czarnecki et�al. 2014, Rinne et�al. 2015, Muhr et�al. 2016), we identified three homologs of D27 (Supplementary Fig. S1), one of LBO (Supplementary Fig. S2) and three of D53 (Supplementary Fig. S6). The existence of multiple copies in the Populus genome is a likely result of genome duplication (Tuskan et�al. 2006). We found that the complete SL pathway of Arabidopsis and rice is conserved in Populus species and that in hybrid aspen the SL pathway genes show unique expression patterns (Figs.�1, 3), which might relate to distinct features of tree branching. Firstly, the perennial lifestyle and the expansive shoot systems of trees (Tomlinson 1983, Millet et�al. 1999, Barth�l�my and Caraglio 2007, Ni et�al. 2015, Rinne et�al. 2015) require a modified branching strategy with a strong emphasis on mechanisms that act locally to control AXB outgrowth (Fig.�8A). Secondly, the AXBs, targets of SL signaling, are distinct in trees. In Arabidopsis, AXMs arise in axils of mature rosette leaves and produce simple scale-less buds (Grbić and Bleecker 2000, Long and Barton 2000). By contrast, in most trees, AXMs arise at a very early stage in the axils of emerging leaves and produce complex AXBs with an enclosed ES and sturdy bud scales (Garrison 1955, Paul et�al. 2014, Rinne et�al. 2015). The outer scale presents a physical barrier that only gradually gives way (Supplementary Fig. S4).

Fig. 8

Working models for mature AXBs (MBs) of proleptic hybrid aspen. (A) Node-to-bud signaling in intact plants. Nodes are dominant production centers of SLs, which are supplied to the AXBs. AXBs only express MAX1 and LBO, which can convert imported CL, CLA, MeCLA (red arrow) into SL-like compounds that are hydrolyzed in nuclei by D14. Subsequent interaction with MAX2 results in ubiquitination and degradation of the transcription repressor D53. Downstream targets BRC1 and BRC2 inhibit AXB outgrowth. Roots hardly contribute SLs (purple arrow). Stippled lines delineate the node. (B) Proximal AXBs. All genes are initially downregulated (dark arrows) except for LBO, which is strongly upregulated (light blue arrow). (C) Single-node systems. MAX1 genes are downregulated (dark arrows), but GR24-feeding (red arrow) prevents it through feedback (red arrows). D14 and MAX2 genes are upregulated (blue arrows), which is prevented by GR24-feeding (red arrows). D53-like1 and D53-like3 are, respectively, downregulated and upregulated (dark and blue arrows), but GR24-feeding prevents this (red arrows). BRC1 and BRC2 are modestly downregulated at day 3 (dark arrowheads), and GR24-feeding does not affect this (red arrowheads).

Grafting experiments with herbaceous plants have shown that roots can act as the primary source for branch-inhibiting SLs (Beveridge et�al. 2000, Morris et�al. 2001, Turnbull et�al. 2002, Simons et�al. 2007). However, this does not necessarily reflect the situation in intact plants. Indeed, despite initial findings (Kohlen et�al. 2011, Kohlen et�al. 2012), xylem-transport of SL from roots to AXBs in intact plants has remained unconfirmed (Xie et�al. 2015, Yoneyama et�al. 2018). Nonetheless, our data indicate that AXBs in all likelihood receive CL and downstream products from elsewhere as, contrary to our initial assumption, AXBs themselves did not express MAX3 and MAX4, but they did express the downstream biosynthetic gene MAX1 (Fig.�3B). The few available studies on woody species did not detect MAX3 and MAX4 transcripts in AXBs and nodal bark, while MAX4 was expressed only in wood tissue (Djennane et�al. 2014, Muhr et�al. 2016). Our data show that in hybrid aspen both genes are highly expressed in nodal bark tissues, and in addition in roots. Although root tips and bark tissues of source nodes expressed MAX4 at similar levels, expression of MAX3 was 65 times higher in the nodes. Moreover, in young nodes (‘sink nodes’), MAX3 and MAX4 expressions were about 200 and 500 times higher, respectively, than in root tips (Fig.�3B). Based on our data, the bark of the AXB-associated nodes appears to be the main source of CL and downstream SL products. That in trees the nodes rather than the roots supply SLs to AXBs is a plausible conjecture, as it would allow for a more precise local control over branching of the expanding shoot system.

Young AXBs are active sinks that might import node-produced SLs along with sugars and other phloem-delivered compounds. As sugars can promote AXB outgrowth (Mason et�al. 2014), a steady inflow of CL and SL-like compounds might be required to keep BRC1 expression high in the maturing AXBs to prevent their outgrowth. Indeed, the expression of MAX1 and BRC1 steadily increases during AXB formation (Rinne et�al. 2015). As BRC1 and BRC2 are class II TCPs, which repress cell cycling (Schommer et�al. 2014), this suggests that during AXB formation SLs target BRC1 to constrain the developing ES. Although the nodal bark expressed all SL biosynthesis genes, AXBs appeared to express MAX1 and LBO (Fig.�3C, D), implying that they might convert imported CL and CLA, as well as MeCLA and other downstream products that require local conversion (Figs.�1, 8A). That LBO is also expressed in the AXBs themselves is biologically meaningful, considering that its bioactive product, like MeCLA, is chemically unstable, providing only weak inhibition of branching in a heterograft in Arabidopsis (Brewer et�al. 2016).

Root tips also expressed MAX3, MAX4, MAX1 as well as LBO, albeit at much lower levels than the nodes (Fig.�3B–D). Considering that e.g. MAX2b and D53-like1 are hardly expressed in roots (Fig.�3F, G), root-produced SLs might serve specific root functions, including attraction of AM fungi in the rhizosphere. It is known that Populus roots can exude CLA, MeCLA as well as the canonical SL 4DO (Xie 2016). Although nodes of hybrid aspen expressed all SL pathway genes, including D14 genes (Fig.�3E), MAX2 genes (Fig.�3F) and D53-like genes (Fig.�3G), particularly D14a and D14b were expressed at much higher levels in the AXBs. The expression of D14 in leaves (Fig.�3E) may take place in vascular tissues, as found for other species (Shen et�al. 2007, Stirnberg et�al. 2007, Zhou et�al. 2013, Soundappan et�al. 2015). As D14 is also present in the sieve tubes (Kameoka et�al. 2016), by default it could move out of source leaves through mass flow toward sinks. Expression of SL pathway genes in vascular tissues may facilitate systemic bud–bud competition by modulation of auxin transport (Shinohara et�al. 2013). Such systemic control by SL could also play a role in natural bud burst of hybrid aspen, for which AXBs require a pre-exposure to winter-chill that further upregulates the SL biosynthesis gene MAX1 in AXBs (Rinne et�al. 2018).

The role of LBO, which in Arabidopsis catalyzes the hydroxylation of MeCLA to the unidentified compound MeCLA+16D (Brewer et�al. 2016), remains enigmatic in our study. Although expressed in all plant parts (Fig.�3D), its expression was highest in the nodes of the mature AXBs, which are poised for outgrowth, supporting its presumed role in inhibiting the outgrowth of para-dormant buds. Notably, expression of LBO in decapitation-activated AXBs was significantly increased after 1 d (Fig.�5C). As BRC1 expression was already diminishing within 2 h, the increase in LBO expression might serve some as yet unidentified function. With the exception of LBO, the SL pathway genes were downregulated during the first 24 h in the bud activation process and followed by the start of ES elongation in the next 24 h (Fig.�4B).

The SL pathway has been shown to be subjected to homoeostatic control (Mashiguchi et�al. 2009), like the GA pathway (Hedden and Thomas 2012). In single-node systems, the upregulation of receptor complex genes between days 3 and 5 could represent a response to diminished signal supply, considering the preceding downregulation of MAX1 genes (Fig.�7A). Indeed, in support of this hypothesis, when signal supply was compensated by feeding GR24, the changes in the expression of the signaling genes were abolished (Figs.�7B–D, 8C). However, as most SL pathway genes may be subject to post-transcriptional and post-translational regulation (Zhou et�al. 2013, Marzec and Muszynska 2015, Hu et�al. 2017), this remains to be investigated.

It is unlikely that the initial triggers in decapitated plants and single-node systems are identical to those in natural branching because the constraints are different in all cases. For example, in decapitation experiments, the removal of the auxin-producing top part of the plant is crucial and could be the cause of AXB activation. However, in experiments with pea, auxin supply to the stump could not repress AXB outgrowth (Brewer et�al. 2015). In hybrid aspen, the high expression of SL biosynthesis genes in nodes of intact plants might prevent AXB activation in the intact plant, resulting in a proleptic branching style. Nonetheless, these high expression levels cannot prevent AXB outgrowth following decapitation. Sugar diversion (Mason et�al. 2014, Kebrom 2017) might play a role also in decapitated hybrid aspen but is unlikely to be a factor in single-node systems that lack leaves. Although root-produced cytokinins (CK) are missing in this system, nodes might produce some CK as a result of the absence of a polar auxin transport stream (Nordstr�m et�al. 2004, Tanaka et�al. 2006, Ferguson and Beveridge 2009).

Our data show that GR24-feeding cannot prevent outgrowth once AXBs are activated. As CK as well as GA can be locally produced in nodes and AXBs, it seems possible that they synergistically promote AXB activation, overriding SL effects by repressing MAX2 and the downstream effects on BRC1 (Ni et�al. 2015). Indeed, we showed that decapitation upregulates GA biosynthesis genes in AXBs (Rinne et�al. 2016), whereas SL pathway genes are downregulated (Fig.�5). GA not only represses SL perception, but it can also downregulate SL biosynthesis (Ni et�al. 2015, Ito et�al. 2017, Marzec 2017). Moreover, GA also reinvigorates symplasmic stem–bud connections by upregulating 1,3-β-glucanase genes (Rinne et�al. 2011, Rinne et�al. 2016), thereby potentially facilitating import of sugars and other nutrients to drive AXB outgrowth.

Conclusions

Nodes rather than distant roots may supply SL precursors and SLs to AXBs, whereas AXBs are sites of SL perception and BRC1 action (Fig.�8). Mature AXBs can also synthesize SL-like compounds downstream of CL, but probably not CL itself as MAX3 and MAX4 are not expressed in AXBs, while MAX1 and LBO are (Figs.�3B–D, 8A). As most SL pathway genes are downregulated by decapitation within hours, and ahead of ES elongation, SL might function in intact plants to inhibit AXB activation. Once activated, elongation of the ES might even be promoted by SL, as suggested by GR24-feeding of single-node systems. GR24-feeding data also support the notion that SL pathway genes are under homeostatic control. When apically produced auxin, root-produced cytokinins and leaf produced sugars are lacking, AXB still grow out despite high initial levels of SL gene expression in nodes, even after GR24-feeding. Although the initial triggers of AXB activation differ between intact plants, decapitated plants and single-node systems, the ensuing growth processes rapidly converge. SLs may restrain outgrowth only during AXB formation and para-dormancy in intact plants but cannot override the interacting factors that facilitate outgrowth of activated AXBs.

Materials and Methods

Plant material and sampling

Hybrid aspen (P. tremula � P. tremuloides) clone T89 was micro-propagated in vitro for 5 weeks in 20�C, planted in a mixture of soil/peat and perlite [4:1 (v/v)], fertilized with 4 g�l−1 Osmocote, grown in a greenhouse under long days (18 h light) at 20�C and 60% relative humidity, and watered twice a day. Natural daylight was supplemented by mercury-halide lamps with the lighting of 200–250 �mol�m−2�s−1 (Osram) to maintain an 18 h photoperiod. The plants were replanted in 13 cm pots when they were ca. 60 cm high. Experiments were started when the plants had reached a height of 80–100 cm, and leaf production rates and elongation were stable. The plants were subdivided into three groups. Group one was kept in long-day (LD) conditions and decapitated at the BMP (Fig.�2), at around 40 cm below the apex, to eliminate apical dominance. The position of the BMP was as described by Rinne et�al. (2015). Group two was kept in LD to collect various types of tissues and organs from intact plants. Group three plants were used for xylem-feeding experiments with single-node systems.

Measurements of AXB and embryonic shoot length, and embryonic leaf number

To record the developmental changes in AXBs proximal to the decapitation point AXB length was measured at 0, 2, 6, 12, 24, 48, 72 and 96 h post decapitation. At the same time points, the AXBs were cut longitudinally under a dissection microscope, and the length of the enclosed ES was measured from the top of the SAM to the middle of a line connecting the base of the outer scale (Fig.�4A). Lastly, comparable AXBs were fixed in 70% alcohol for assessing the neo-formation of leaves. Under a dissection microscope, the bud scales were peeled away, and the number of embryonic leaves was counted for each time point. Commonly the SAM contained one leaf buttress, which was included in the count.

AXB burst tests and GR24-feeding

To investigate the role of SL in AXB inhibition, we performed xylem-feeding experiments in combination with AXB burst tests under forcing conditions. As hybrid aspen is proleptic, the forced activation of AXBs represents a form of bud burst which, in contrast to sylleptic species and herbaceous plants, includes two processes, activation and outgrowth. For xylem-feeding, single-node systems without leaves were isolated from 6-week-old LD plants. The internode base was punched through pores in a Styrofoam sheet that was floated on water (control) or water supplemented with the synthetic SL analog (rac-GR24; Chiralix BV, The Netherlands) at a concentration of 10 �M. In preliminary experiment GR24 in concentrations of 1, 5 and 10 �M were tested, and 10 �M was chosen for the current experiments. In each treatment and time point, three replicates of young and three replicates of mature AXBs were used. AXB length, ES length and embryonic leaf number were recorded at days 0, 3, 5 and 7. The young buds and the mature AXBs below the BMP were harvested at days 0, 3 and 5 to analyze changes in the relative expression of SL pathway genes induced by decapitation, and by the combination of decapitation and GR24-feeding. The experiments were repeated at least twice.

Gene selection and identification

To examine the expression patterns of SL biosynthesis and signaling genes in 6-week-old intact plants, total RNA was extracted from different plant parts. These included the apex, young AXBs, the bark of the corresponding node of young AXBs (‘sink node’), sink leaves, mature para-dormant AXBs, the corresponding node of mature AXBs (‘source node’), source leaves and root tips (Fig.�2). In total, three AXBs above, and three below the BMP, as well as other tissues like indicated above, were collected from each of the six plants. Samples of two plants were pooled to obtain three biological replicates. Gene expression analyzes included Populus homologs of the Arabidopsis SL biosynthesis and signaling genes D27a, D27b, D27c, MAX1.1, MAX1.2, LBO, MAX3, MAX4, D14a, D14b, MAX2a, MAX2b, D53-like1, D53-like2, D53-like3, as well as the downstream target genes BRC1 and BRC2.

To assess decapitation-induced changes in gene expression, mature AXBs proximal to the decapitation point at the BMP (Fig.�2) were collected at days 0, 2, 6, 12, 24 and 48 h post decapitation. For each time point, RNA was extracted from three biological replicates, pooled as described above. Sampling after day 1 (24 h) and day 2 (48 h) were carried out at the same time of the day to avoid potential diurnal effects on gene expression.

To assess the role of exogenous SL on gene expression in AXBs of single-node systems, they were incubated in water with or without GR24. AXBs were collected after 0, 3 and 5 d of treatment. Gene expression was assessed for the SL biosynthesis genes D27a, D27c, MAX1.1 and MAX1.2, as well as the SL signaling genes D14a, D14b, MAX2a, MAX2b, D53-like1, D53-like2, D53-like3 and downstream targets BRC1 and BRC2.

RNA extraction, cDNA preparation and quantitative RT-PCR analysis

Total RNA was extracted from 0.2 to 0.3 g of frozen tissue and grinded in a mortar with 500 �l extraction buffer (Qiagen RLT buffer containing 1% PVP-40), followed by an addition of a 0.4 volume KoAC (pH 6.5) and further homogenization. Subsequently, the solution was transferred to a 2-ml tube, incubated on ice for 15 min, and centrifuged at 12,000 rpm at 4�C for 15 min. The supernatant was transferred to a new 1.5-ml tube, and a 0.5 ml volume of 100% ethanol was added. The mix was transferred to RNeasy spin columns and further processed in accordance with instructions of the Qiagen Plant RNA isolation kit. Genomic DNA was eliminated using TURBO™ DNase kit (Invitrogen) treatment according to the manufacturer’s instructions and cleaned using the total RNA purification system ‘Purelink RNA mini kit’ (Invitrogen). RNA was quantified with NanoDrop 1000, and the RNA quality was assessed with the Agilent 2100 Bioanalyzer system. One microgram of total RNA was reversely transcribed to cDNA with SuperScript� VILO™ reverse transcriptase (Invitrogen). Quantitative RT-PCR (qRT-PCR) was used to analyze transcript levels of all SL pathway genes. The reaction setup (20 �l total volume) was prepared using SYBR� select PCR master mix (Applied Biosystems). As a template, 2 �l of the cDNA (200 ng) were added. All the qPCR reactions were run with three biological replicates and analyzed in three technical repeats. Real-time qRT-PCR analyses were performed with the Applied Biosystems 7500 Fast Real-Time PCR system according to the manufacturer’s instruction. Thermocycling conditions were set to 50�C for 2 min, 95�C for 2 min, 45 cycles of 15 s at 95�C and 60 s at 60�C. In addition, each PCR reaction included a negative control to check for potential genomic DNA contamination. PCR amplification of Populus actin served as a reference gene for normalizing the relative transcript level. For a complete list of primers and genes used for qRT-PCR see Supplementary Table S1.

Statistical analysis

Statistical analyses were carried out using analysis of variance (anova) in combination with a post hoc test to determine significant differences between the subgroups. One-way anova in combination with Fisher’s LSD test was computed to monitor the decapitation-induced changes in transcript levels and to pinpoint the time within the 48 h trajectory when a significant change took place. To analyze the effect of GR24 on gene expression during the 7-d feeding experiment, two-way anova (time and treatment as factors) was used in combination with Fisher’s LSD multiple comparison test. The developmental changes during AXB activation (AXB length, ES elongation, leaf numbers) induced either by decapitation or isolation of the single-node systems and treated with or without GR24 were analyzed with one- or two-way anova and combined with Fisher’s LSD test. Computation was performed using Microsoft Excel data analysis (www.microsoft.com) and Minitab Statistical Software version 18.1 (www.minitab.com).

Bioinformatics

BLAST searches in GenBank, P. trichocarpa genome v3.0 and P. tremula � P. tremuloides (T89) v0.1 databases (http://www.ncbi.nlm.nih.gov/BLAST; http://www.phytozome.net; http://popgenie.org/) were used to identify SL biosynthesis and signaling genes. Gene-specific primer sequences for qPCR analysis were designed using Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/). Phylogenetic trees were created using the MEGA6 program (www.megasoftware.net) with the Neighbor–Joining method. Bootstrap support values are based on 1,000 replicates.

Funding

The Norwegian University of Life Sciences to NUK, and a FRIPRO grant of the Norwegian Research Council to CvdS and PLHR [nr. 263117].

Click here for additional data file.