Class I KNOX transcription factors promote differentiation of cambial derivatives into xylem fibers in the Arabidopsis hypocotyl

In higher plants, the bulk of cell divisions occur in tissues with a low degree of differentiation, the so-called meristems. Meristems are either placed at the tips of plant organs, i.e. apical meristems, or subapically, as lateral meristems. Derivatives of stem cells in apical meristems elongate along the main growth axes, whereas daughter cells of lateral meristems, e.g. of the vascular cambium, expand mainly radially and therefore contribute to the increment in diameter of plant organs. As both apical and lateral meristems fulfill the same basic meristematic functions of stem cell specification, maintenance, cell proliferation and differentiation, it has been suggested that these processes are regulated by the same or paralogous genes across different types of meristems (Groover et al., 2006; Aichinger et al., 2012). Stem cell maintenance in the shoot apical meristem (SAM) is partly directed by a negative-feedback loop between the homeodomain transcription factor WUSCHEL (WUS) and the peptide ligand-producing gene CLAVATA3 (CLV3) (Aichinger et al., 2012). WUS is expressed in the stem cell niche (organizing center) of the SAM from where the WUS protein can move to the stem cells to activate CLV3 expression in order to control the size of the stem cell population (Yadav et al., 2011). CLV3 peptide signaling in turn represses WUS transcription, resulting in negative-feedback regulation of WUS (Schoof et al., 2000; Brand et al., 2000). Similarly, WUS and CLV3 paralogs control stem cell behavior in the root apical meristem (WOX5 and CLE40) (Stahl et al., 2009) and the cambium (WOX4 and CLE41/44) (Hirakawa et al., 2010), suggesting that paralogous genes regulate stem cell maintenance across different classes of meristems.

Independently of the WUS-CLV3 regulatory circuit, the class I KNOX homeodomain transcription factor SHOOT MERISTEMLESS (STM) contributes to meristem maintenance in the SAM (Lenhard et al., 2002). STM is expressed throughout the meristem but not at the place of incipient leaves (Long and Barton, 1998). STM-null mutants in the Ler background fail to establish a SAM during late embryogenesis and weak mutant alleles lead to abortion of the SAM after the formation of only a few leaves (Endrizzi et al., 1996). KNAT1, another member of the class I KNOX family, is expressed in a similar pattern to STM (Lincoln et al., 1994) and acts redundantly with STM on meristem maintenance (Byrne et al., 2002). Furthermore, the stm meristem phenotype can be rescued by KNAT1 overexpression (Scofield et al., 2013). Together, these data have led to the interpretation of STM and KNAT1 as being required to prevent cell differentiation at the site of their expression and hence to keep cells meristematic. Conversely, absence of their transcripts from incipient leaves is thought to allow differentiation, as a consequence of which organogenesis from meristematic tissue can take place.

In wood-forming tissues, the cambium forms radial files of daughter cells that undergo rapid differentiation into xylem and phloem. In the xylem, a phase of radial cell expansion is followed by secondary cell wall (SCW) deposition, which represents a structural barrier preventing cell division and renders differentiation irreversible. In Arabidopsis, the establishment of the cambium in the hypocotyl is already completed early in development, approximately 2 weeks after germination (Busse and Evert, 1999). At this developmental stage, referred to as xylem I, xylary derivatives differentiate into vessel cells and chlorophyll-containing parenchymatic cells with unlignified walls (Chaffey et al., 2002). Later in development, upon flowering, xylem fibers with lignified SCWs are formed instead of parenchymatic cells (a phase that is referred to as xylem II) (Chaffey et al., 2002; Sibout et al., 2008), resulting in a similar anatomy to that of angiosperm wood. In contrast to the Arabidopsis inflorescence, the hypocotyl contains a continuous cylindrical cambium without disturbances of traces from lateral organs. Furthermore, an apical-basal developmental gradient with internodes of different developmental stages, as is characteristic for the Arabidopsis stem, is absent in the hypocotyl (Ragni and Hardtke, 2013; Schuetz et al., 2013).

Despite the fact that STM homologs are expressed in the vascular cambium of Populus trees (Schrader et al., 2004; Groover et al., 2006) and although it was previously noted that STM in Arabidopsis is not exclusively expressed in the SAM but also subapically in tissues related to the vasculature, little attention has been paid to its function outside the SAM (Sanchez et al., 2012; Aichinger et al., 2012). Although phenotypes outside the SAM have been reported for knat1 mutants, e.g. downward-pointing siliques (Ragni et al., 2008), premature floral organ abscission (Shi et al., 2011) and aberrant lignification patterns in the inflorescence stem (Mele et al., 2003), the role of KNAT1 in cambium maintenance and differentiation of cambial derivatives has not yet been investigated. We made use of weak stm and knat1 mutant alleles and the simple anatomy of Arabidopsis hypocotyls in order to study whether their function as repressors of differentiation has been co-opted by the cambium during the evolution of secondary growth.

Arabidopsis hypocotyls grow radially from the cambium, the derivatives of which undergo gradual differentiation either into xylem fibers, parenchyma and vessels to the inner side or into phloem to the outer side. If STM and KNAT1 keep cambial cells undifferentiated and meristematic, similar to the SAM, their expression is expected to be highest in dividing cells of the cambium and absent in differentiating cambial daughter cells. In the hypocotyl, pKNAT1::GUS and pSTM::GUS activities were highest in the cambial zone, but unlike in the SAM, where STM and KNAT1 transcripts are not present at positions of organogenesis, they were also expressed in the phloem and in developing and mature xylem fibers (Fig. 1A-D). KNAT1 and STM transcripts were equally abundant in dissected hypocotyls of 6-week-old plants and in inflorescence meristems of the same age (Fig. 1E). Similar strength of expression in the two different types of meristems supports the hypothesis of KNAT1 and STM playing a role outside of apical meristems, but presence of their transcripts in differentiating and even mature cambial derivatives indicates that these two transcription factors fulfill additional or different tasks in the cambium than in the SAM.

We then isolated mutants of STM and KNAT1 in order to functionally address their role in the cambium of the hypocotyl. Because in plants homozygous for strong stm loss-of-function alleles, the SAM aborts early or is not established (Barton and Poethig, 1993; Endrizzi et al., 1996), we identified weak loss-of-function mutants in STM [stm-GK; NASC (N409575)] and KNAT1(knat1^bp-9). knat1^bp-9 and stm-GK carry a T-DNA insertion in the first and second intron, respectively, which caused a reduction of the wild-type transcript levels in both mutants by more than two orders of magnitude (supplementary material Fig. S1A). Both mutants accumulated the same amount of aerial biomass as wild type 6 weeks after germination (supplementary material Fig. S1B). Hypocotyls of knat1^bp-9 and stm-GK mutants were smaller in diameter than wild type (supplementary material Fig. S2) and transverse sections through the hypocotyl of 6-week-old plants revealed a reduction of the ratio between xylem II and total xylem, indicative of reduced xylem fiber formation (Fig. 2A,B). Hypocotyl diameter and xylem II formation were similarly reduced in stm-6, another weak mutant, as in stm-GK, whereas in the null mutant knat1^bp-11 (dela Paz et al., 2012), xylem fiber differentiation was almost entirely abolished (Fig. 2B; supplementary material Fig. S2). By contrast, single mutants for all the other members of the KNOX gene family did not display reduced xylem II (supplementary material Fig. S2). Similar to their hypocotyls, in the oldest internode of knat1^bp-9 and stm-GK inflorescence stems we frequently observed vascular bundles that did not form fibers (supplementary material Fig. S3).

Hypocotyls of stm-GK and knat1^bp-9 showed only small differences in total cambial activity, but instead of fibers (xylem II), they formed more xylem parenchyma cells (xylem I) per radial cambial cell file than did wild type (Fig. 2C). In contrast to xylem production, the number of phloem cells was not altered in hypocotyls of stm-GK and knat1^bp-9. In order to test whether diminished cell expansion contributed to reduced hypocotyl diameters of knat1 and stm mutants, we measured radial cell expansion of cambial derivatives, which is an early step in their differentiation into vessels or fibers. Both fibers and vessels of stm-GK and knat1^bp-9 did not expand to the same luminal areas as observed in mature vessels and fibers of wild-type hypocotyls (Fig. 3A). The degree of reduction of luminal area accounted for smaller diameters of mutant hypocotyls, indicating that KNAT1 and STM were required for early steps of xylem cell differentiation. We then examined expression patterns of ATHB8, an early marker of vascularization that, when overexpressed, promotes xylem fiber and vessel differentiation (Baima et al., 1995; Gardiner et al., 2011). In wild type, pAtHB8::GUS was expressed in the cambium and developing xylem, peaking in cell files, which give rise to xylem vessel cells, and to a lesser extent in cells at the origin of fiber cell files. By contrast, pATHB8::GUS expression was nearly absent in xylem fiber cell files in stm-GK and knat1^bp-9 (Fig. 3B-D). Taken together, reduction of STM and KNAT1 expression impaired early events of differentiation of cambial daughter cells into xylem fibers, whereas it only weakly affected overall cambial cell division activity. Promotion of xylem fiber differentiation by STM and KNAT1 stands in sharp contrast to their proposed function of preventing differentiation of meristematic cells in the SAM.

Final radial diameter of cambial derivatives is a result of the cell expansion rate and the residence time in the expansion zone (Skene, 1969; Cuny et al., 2014). Hence, large expansion zones, as observed for example during early wood formation, correlate with big luminal areas, whereas short expansion zones are associated with smaller lumina of cambial derivatives. Smaller cells in stm and knat1 mutants could therefore be a consequence of reduced residence time of cambial derivatives in the expansion zone rather than defective rate of cell expansion. We measured the radial dimension of the expansion zone, including the cambium, with the help of the JIM13 cell wall epitope (Hall et al., 2013), which marks mature phloem bundles and SCW of xylem fibers and vessels (supplementary material Fig. S4). At the time of transition to flowering, the cambium-expansion zones of knat1^bp-9 and stm-GK hypocotyls were of similar size to those in wild type. One week after transition to flowering, when the first mature xylem fibers had been formed, the cambium-expansion zone was reduced to about one-third of its previous size in wild type. By contrast, in both mutants, the cambium-expansion zones were significantly larger 1 week after transition to flowering than they were in wild type. Thus, the cell expansion defect observed in the mutants cannot be explained by a reduction in expansion zone sizes, suggesting that STM and KNAT1 have a direct positive impact on the rate of cell expansion.

Even though expression patterns support a local function of KNAT1 and STM in the hypocotyl, effects on fiber differentiation could be indirect, e.g. mediated through a mobile hormone-like factor derived from the SAM. Class I KNOX genes have previously been suggested to regulate the biosynthesis of the plant hormones cytokinin and GA in the SAM (Jasinski et al., 2005). Furthermore, graft-transmissible GA promotes fiber formation in the hypocotyl (Ragni et al., 2011). To address the issue of whether class I KNOX genes act locally in the hypocotyl/root to influence fiber differentiation, we used reciprocal grafting experiments. As we could not recover viable plants from graftings involving stm, we focused on reciprocal graftings with knat1. Five-day-old wild-type (Ler) hypocotyls were grafted on knat1^bp-1 stocks and hypocotyls of 6-week-old plants were sampled. The hypcotyl scion above the grafting junction was white (normal for the wild type), whereas the mutant stock was dark green (typical for knat1 hypocotyls) (supplementary material Fig. S5). Transverse sections of the grafted stocks, 3 mm below the grafting junction, were analyzed (Fig. 4). No phenotypic alterations were observed in wild-type hypocotyls carrying knat1^bp-1 scions, whereas in knat^bp-1 hypocotyls carrying a wild-type shoot, fiber formation was affected to a similar degree than in graftings between knat1^bp-1 stocks and scions. Reduced xylem fiber formation indicated that wild-type shoots were not sufficient to rescue the fiber differentiation phenotype in knat1^bp-1 hypocotyls. As expression of the typical knat1 phenotype in the hypocotyl did not depend on the presence of defective KNAT1 in the shoot/scion, KNAT1 probably acts locally, in the hypocotyl or root, on xylem fiber differentiation.

In wild type, transition to xylem II development took place after the formation of ∼11 parenchymatic cells per cambial cell file (Fig. 2C). By contrast, onset of the xylem II phase was delayed in knat1^bp-9 (after 19 cells) and stm-GK (after 15 cells). Initiation of SCW deposition in xylem fibers of the hypocotyl correlates with induction of flowering; in recombinant inbred lines of a Uk-1×Sav-0 cross, the major QTL for fiber differentiation localizes to FLC, a negative regulator of flowering time (Sibout et al., 2008). We therefore tested whether delayed flowering accounts for delayed xylem II development in stm and knat1 mutants. Although weak stm mutants rarely form flowers (Endrizzi et al., 1996; Felix et al., 1996), both knat1 and wild type flowered after the formation of ∼11 rosette and three cauline leaves, and also bolted at the same time (Fig. 5). Reduced xylem fiber formation in knat1 was therefore not a consequence of delayed flowering, and KNAT1 seems to act downstream of or in parallel to regulators of flowering time. Notably, both mutants produced the first xylem fibers 1-2 weeks later than the wild type, where the first mature fibers were observed 4 weeks after germination, shortly after initiation of flowering took place (supplementary material Fig. S6). In wild-type hypocotyls, xylem fiber production stopped between 6 and 7 weeks after germination, corresponding to the time point when the first silique was completely senescent. Similarly, fiber production ceased after 6 and 7 weeks in knat1^bp-9 and stm-GK, respectively. Importantly, neither of the mutants reached the same level of xylem II formation at the end of their life cycle (10 weeks after germination). Hence, the mutants could not compensate later initiation of fiber differentiation by a later cessation of fiber production.

KNAT1 expression levels and patterns were unaltered in hypocotyls of stm-GK, whereas in knat1^bp-9, STM transcript abundance was mildly but significantly reduced to 66% of the expression level in wild type (P<0.05; Fig. 6F; supplementary material Fig. S7). In order to test whether KNAT1 function is required for STM expression, we introgressed knat1^bp-1, a knat1-null mutation (dela Paz et al., 2012), into the Col-0 background (referred to as knat1^bp-1C). In knat1^bp-1C hypocotyls, STM expression was more strongly reduced to 30% of the wild-type level than in knat1^bp-9, but not entirely abolished. This indicates that STM expression depends partially on KNAT1 activity. However, because in stm-GK, which showed a less pronounced xylem phenotype than both knat1^bp-1C and knat1^bp-9, STM expression is reduced by more than two orders of magnitude (Fig. 6F), the effects of KNAT1 on fiber differentiation are unlikely to be mediated solely through STM.

In a severely dwarfed double mutant of the weak stm-GK and knat1^bp-9 alleles (supplementary material Fig. S1B, Fig. S2A and Fig. S8), xylem fiber differentiation was completely abolished (Fig. 6A-D), indicating that STM and KNAT1 are redundantly required for xylem fiber differentiation. In comparison with the single mutants, radial xylem increment and transverse luminal area of vessel cells were synergistically reduced in hypocotyls of stm-GK;knat1^bp-9 (Fig. 6E). However, residual cambial activity and radial organization of cambial cell files were still retained in the double mutant (Fig. 6B), suggesting that meristem activity and maintenance are less sensitive to reduced KNAT1 and STM function than fiber differentiation. In accordance with this, WOX4 and PXY/TDR, which are associated with cambial activity, showed either increased or unaltered expression in 6-week-old hypocotyls of the double mutant, whereas ATHB8 was less strongly expressed than in wild type (Fig. 7A). Importantly, transcript levels of xylem fiber identity genes SND1 and NST1, and their downstream target SND2 (Wang and Dixon, 2012), were all very strongly reduced in hypocotyls of stm-GK;knat1^bp-9. By contrast, the expression level of the phloem pole marker APL (Bonke et al., 2003) was not altered, and the expression of the vessel identity gene VND7 (Wang and Dixon, 2012) was increased in the double mutant.

We then tested whether KNAT1 and STM are co-expressed with SCW genes. Among the 100 most highly co-expressed genes of KNAT1 and STM across a collection of more than 300 microarrays (Expression Angler) (Toufighi et al., 2005), 52 genes were co-expressed with both STM and KNAT1. Strikingly, among these 52 co-regulated genes we found 20 members of the 44 genes that make up the SCW regulon described by Persson et al. (2005,) (supplementary material Table S1). We tested whether KNAT1 and STM act upstream of the SCW regulon by performing qRT-PCR on 6-week-old hypocotyls of stm-GK and knat1^bp-9, as well as on the respective double mutant. Cellulose synthase A (CESA) 4, CESA7 and CESA8, which are specific for SCW synthesis (Taylor et al., 2003), were significantly less strongly expressed in knat1^bp-9 than in wild type (Fig. 7B). Partial removal of STM function in the knat1^bp-9 background (i.e. in stm-GK;knat1^bp-9) reduced their transcript levels by more than two orders of magnitude. By contrast, five of the six CESA genes, which are specifically required for cellulose synthesis during primary wall formation (Persson et al., 2007; Sullivan et al., 2011), were not differentially expressed in knat1^bp-9;stm-GK. Together, these results suggest that STM and KNAT1 are redundantly required for the expression of secondary wall genes but are not involved in cellulose deposition during primary growth.

Opposing functions of STM and KNAT1 in the cambium, when compared with their functions in the SAM, could be due to differential, meristem-specific expression of genetic interactors. BLADE-ON-PETIOLE1 (BOP1) and BOP2, which encode BTB/POZ domain and ankyrin repeat-containing proteins (Norberg et al., 2005; Hepworth et al., 2005), antagonize the effect of KNAT1 on pedicel angle in the Arabidopsis inflorescence stem and overexpression of BOP1 and BOP2 phenocopies the premature lignification phenotype of knat1 mutant stems (Khan et al., 2012). We tested whether BOP1 and BOP2, which are not expressed in the summit of the SAM, have a function in the cambium. Shortly after the onset of cambial activity, in 2-week-old hypocotyls, pBOP1::GUS was expressed in the cortex but was absent from the inner tissues (supplementary material Fig. S9A). At this stage, pBOP2::GUS was expressed in a reciprocal pattern, namely in the periderm, phloem, cambial zone and xylem (supplementary material Fig. S9B). In 3-week-old hypocotyls, pBOP1::GUS activity was no longer detectable in transverse sections of the hypocotyl (Fig. 8A), whereas pBOP2::GUS was strongest in the cambial zone and in the developing phloem, similar to pKNAT1::GUS, but notably absent from the developing xylem and mature xylem fibers (Fig. 8C). In the knat1^bp-1C-null mutant background, both pBOP1::GUS and pBOP2::GUS were overexpressed and, interestingly, BOP2 expression occurred ectopically in xylem parenchyma cells (Fig. 8A-D). qRT-PCR confirmed that BOP1 and BOP2 were strongly de-repressed in knat1^bp-1C and to a lesser extent in stm-1^C (stm-1 introgressed into Col-0) hypocotyls (Fig. 8G), suggesting that KNAT1 and STM act upstream of BOP1 and BOP2. By contrast, we did not observe any change in gene expression levels of KNAT1 and STM in a bop1;bop2 double mutant (Fig. 8E-G), indicating that negative regulation of KNAT1 expression by BOP1 and BOP2, as occurs in the embryonic hypocotyl (Jun et al., 2010), is no longer active during later stages of hypocotyl development.

In order to test whether BOP genes are of functional importance during secondary xylem formation and whether ectopic BOP expression in xylem parenchyma could explain the knat1 fiber phenotype, we analyzed a bop1;bop2 double mutant. bop1;bop2 hypocotyls displayed increased xylem fiber formation, whereas ectopic expression of BOP1 led to an opposite phenotype and was sufficient to prevent xylem fiber differentiation to a similar degree as in strong knat1 mutants (Fig. 8H-L). Furthermore, the knat1^bp-1C fiber differentiation phenotype, but not the reduced cambial activity, was completely suppressed by bop1;bop2 (Fig. 8I-M; supplementary material Fig. S10), indicating that KNAT1 acts on fiber differentiation independently of cambial activity through the repression of BOP genes. Ectopic BOP expression in xylary parenchyma is therefore likely to account for repression of xylem fiber differentiation in knat1 mutants.

The simple anatomy of the cambium in Arabidopsis hypocotyls and the use of weak loss-of-function mutant alleles permitted us to uncover a promotive function of KNAT1 and STM on xylem differentiation, a yet unknown role that is in sharp contrast with their suggested inhibitory mode-of-action on cell differentiation in the SAM. Although there is a large body of evidence to indicate that STM and KNAT1 repress differentiation in the SAM, conclusions about the functions of these genes were often based on spectacular phenotypes resulting from ectopic overexpression of class I KNOX genes, typically leading to adventitious meristems and highly lobed leaf margins (Hay and Tsiantis, 2010). Little attention has been given to evidence challenging the view of STM and KNAT1 being simply repressors of differentiation. For example, the recessive waldmeister (stm^wam) mutant shows fasciated inflorescences and delayed senescence (Felix et al., 1996) characteristics, which are not consistent with the proposed repressive function of STM on differentiation. Additionally, stm^gorgon, a partial loss-of-function mutant, displays fasciated shoot apical and inflorescence meristems (Takano et al., 2010), resulting in a shoot architecture reminiscent of clavata mutants. CLV3 transcript levels, which are reduced in the nonsense mutant stm-1, and WUS are increased in stm^gorgon (Takano et al., 2010). As stm^gorgon is a recessive hypomorphic mutation it seems likely that different levels of STM activity can have different, even opposite, effects. In the hypocotyl, however, strong stm mutants, which did not establish a SAM, had a similar, although more pronounced, xylem phenotype compared with weak stm mutants. Notably, in both weak and strong mutants of stm and knat1, cambial growth was retained. Thus, rather than being the result of different STM or KNAT1 levels in the SAM versus the cambium, the apparently opposite consequences of reduced levels of STM and KNAT1 between the two types of meristems are likely caused by a different mechanism. Similar STM and KNAT1 expression levels in inflorescence meristems and the cambium (Fig. 1) are in keeping with this conclusion.

Spicer and Groover (2010) proposed that regulators of meristem maintenance and activity from the SAM have been co-opted during the evolution of cambia. As different WUS and CLV3 paralogs govern meristem regulation in apical and lateral meristems, neo-functionalization of WUS and CLV paralogs may have only required divergence in their promoter but not protein-coding sequences. Recruitment of class I KNOX genes from apical to lateral meristem regulators would have required an expansion of their expression domain to the cambium but did not involve gene duplication. Divergent functions of STM and KNAT1 between different types of meristems could be the result of meristem type-specific transactivation targets, due to different chromatin structures, to genetic or protein-protein interactions, or to concentration-dependent binding to a different suite of promoters.

As class I KNOX genes are present in the moss Physcomitrella patens (Sakakibara et al., 2008), a non-vascular plant that lacks a functional xylem, lignified xylem fibers and a cambium, their ancestral function cannot be specific to cambia and their role might be of a more general nature, e.g. promoting cell division. In seedlings of strong stm mutants, meristem-like structures occur in the axils of cotyledons, which are much smaller than the wild-type SAM and comprise only slightly enlarged cells (Endrizzi et al., 1996; Barton and Poethig, 1993). This suggests that in the SAM, STM is required for cell division and has only limited effects on the repression of differentiation. By contrast, our results show that, in the cambium, cell proliferation is less sensitive to reduced KNAT1 and STM levels than cell differentiation. We suggest that in the cambium at sites of high class I KNOX activity, cell proliferation overrides the simultaneously induced differentiation mediated by repression of BOP expression. Conversely, in the expansion zone, STM and KNAT1 activity may drop below a threshold required for cell proliferation that is, however, still sufficient to repress BOP expression and as a consequence allow differentiation. By contrast, due to the absence of BOP1 and BOP2 expression from the center of the SAM (Norberg et al., 2005), rather than due to an effect on cell differentiation, reduced STM and KNAT1 function would only affect cell division activity in the SAM. In line with this hypothesis, no SAM phenotype has been reported for recessive BOP mutants to date. Rather than meristem type-specific access to a different suite of promoters, the discrepancy between the SAM and cambium phenotype of stm and knat1 mutants could therefore be a consequence of differential expression of BOP genes.

Plants were grown under standard greenhouse conditions on soil with 16 h light per day and at 23°C. Unless otherwise stated, hypocotyls were dissected 6 weeks after germination, which corresponded to the onset of senescence of the first silique. Hypocotyls were cut basal to the petioles of the cotyledons and 3 mm long segments (rootward) were employed for anatomical analyses and RNA extractions. KNOX mutants were purchased from the Arabidopsis Stock Centre, Nottingham, UK. Accession numbers and primers used for genotyping can be found in supplementary material Tables S2 and S3.

Phloroglucinol staining was performed on hand sections or 30 µm vibratome sections. For vibratome sections, hypocotyls were embedded in 5% agarose. The sections were incubated in 70% ethanol for 1 min and then directly transferred to phloroglucinol (in 20% HCl) for 3 min. Sections were mounted on glass slides in chloral hydrate:glycerol:H₂O (8:3:1) clearing solution. For imaging, an Axiocam mounted on an Axioplan 2 stereomicroscope (Zeiss) was used. The Axiovision 4.8 software was employed to measure sizes and areas.

GUS stainings were performed on freshly harvested hypocotyl segments of 3 mm length. The segments were immediately immersed in GUS staining solution (Fischer et al., 2006) and placed on ice. After vacuum infiltration of the staining solution, hypocotyls were incubated at 37°C on a shaker for 3 h. Subsequently, the tissue was fixed in FAE (2% formaldehyde, 5% acetic acid, 63% ethanol) for 1 h and stored in 70% ethanol. For microscopy, GUS-stained hypocotyls were hand sectioned, mounted and observed as the phloroglucinol stained samples.

For Toluidine Blue staining, hypocotyls were fixed in FAE for 1 h and then dehydrated in an ethanol, isopropanol, Roti-Histol (Roth Laborbedarf, Germany) concentration series and embedded in paraffin (Rotiplast, Roth Laborbedarf) (Escalante-Pérez et al., 2009). Blocks were sectioned into 30 µm sections using a sliding microtome (Reichert, Austria). After removal of the paraffin with xylene, sections were stained in 1% Toluidine Blue O (TBO, Sigma) for 1 min.

Total RNA was isolated from freshly harvested hypocotyls or other tissues using the RNeasy Plant Mini Kit (Qiagen) following the manufacturer's guidelines. Prior to first-strand cDNA synthesis (QuantiTect Reverse Transcription, Qiagen) the Ambion Turbo DNase kit (Invitrogen) was employed to remove residual gDNA. Total RNA (1 µg) was used for first-strand cDNA synthesis. qRT-PCR was carried out using the ROCHE LightCycler480 SYBR Green I Master kit (Roche) and reactions were run on a LightCycler 480 (Roche) with preincubation at 95°C for 5 min; and amplification cycles of 95°C for 10 s, 61°C for 10 s and 72°C for 10 s. Data were analyzed using LightCycler 480 Software Release1.5.0 (Roche). Amplification efficiencies were calculated with the help of cDNA dilution series and relative expression values were calculated according to Pfaffl (2001). ACTIN2 expression was used as an internal standard. Primers are listed in supplementary material Table S3.

For immunolocalization with the monoclonal rat JIM13 antibody (Complex Carbohydrate Research Center, CCRC, University of Georgia, Athens, GA, USA), hypocotyls were fixed in 4% formaldehyde in 1× PME (100 mM PIPES, 1 mM MgSO₄, 2 mM EGTA) and 30 µm vibratome sections were cut. BSA (5%) was used for blocking. Primary and secondary [Alexa Fluor 568 goat anti-rat IgG (H+L), Life Technologies] antibodies were diluted 1:40 and 1:300, respectively. After immunolabeling, sections were counterstained with calcofluor white (0.001%). Images were acquired with a Zeiss LSM780 confocal microscope with 405 nm excitation and 415-445 nm emission for calcofluor white, and 561 nm excitation and 575-600 nm emission for the secondary antibody.

Grafting was carried out between scions and stocks of hypocotyls as described previously (Turnbull et al., 2002) using 5-day-old seedlings grown on 1/2 MS plates at 24°C with 16 h light per day. Pieces of silicon tubes (3-5 mm) were used during grafting. Successful grafts were transferred into soil 5 days after grafting and grown under 16 h light conditions at 23°C. Hypocotyls were sampled when plants were 6 weeks old.

We thank Angela Hay (University of Oxford, UK) and Wolfgang Werr (University of Cologne, Germany) for providing us with knat1^bp-9 and pSTM::GUS seed stocks, respectively. We also thank Björn Sundberg for constructive discussions and advice.