Pluripotency of Arabidopsis xylem pericycle underlies shoot regeneration from root and hypocotyl explants grown in vitro


(fax +33 1 44 27 45 82; e-mail


We have established a detailed framework for the process of shoot regeneration from Arabidopsis root and hypocotyl explants grown in vitro. Using transgenic plant lines in which the GUS or GFP genes were fused to promoters of developmental genes (WUS, CLV1, CLV3, STM, CUC1, PLT1, RCH1, QC25), or to promoters of genes encoding indicators of the auxin response (DR5) or transport (PIN1), cytokinin (CK) response (ARR5) or synthesis (IPT5), or mitotic activity (CYCB1), we showed that regenerated shoots originated directly or indirectly from the pericycle cells adjacent to xylem poles. In addition, shoot regeneration appeared to be partly similar to the formation of lateral root meristems (LRMs). During pre-culture on a 2, 4-dichlorophenoxyacetic acid (2, 4-D)-rich callus-inducing medium (CIM), xylem pericycle reactivation established outgrowths that were not true calli but had many characteristics of LRMs. Transfer to a CK-rich shoot-inducing medium (SIM) resulted in early LRM-like primordia changing to shoot meristems. Direct origin of shoots from the xylem pericycle occurred upon direct culture on CK-containing media without prior growth on CIM. Thus, it appeared that the xylem pericycle is more pluripotent than previously thought. This pluripotency was accompanied by the ability of pericycle derivatives to retain diploidy, even after several rounds of cell division. In contrast, the phloem pericycle did not display such developmental plasticity, and responded to CKs with only periclinal divisions. Such observations reinforce the view that the pericycle is an ‘extended meristem’ that comprises two types of cell populations. They also suggest that the founder cells for LRM initiation are not initially fully specified for this developmental pathway.


Totipotency is the ability of each cell in a multicellular organism to develop independently into a complete individual. It is a particular feature of plant cells, and allows spontaneous or experimental plant regeneration. Since the 1950s, in vitro cultures have been used to accumulate evidence of the intrinsic totipotency and regenerative capacity of plant cells. Full expression of totipotency occurs through somatic embryogenesis, while the regeneration of root and shoot meristems from differentiated cells reflects pluripotency, i.e. the possibility of more than one potential outcome (Verdeil et al., 2007). Such regeneration can originate directly without an intermediate unorganized callus, or indirectly following a callus stage (Gahan and George, 2008). Direct initiation occurs when explants contain cells that have retained the capacity to divide and to generate a root or shoot rapidly after they re-enter the cell cycle. When such pre-determined cells are not present, explants have to de-differentiate through cell division before they can form meristems. It is clear that not all types of cell have an equal capacity to retain pluripotency. However, there have been few studies of these differences in capacity at the molecular level. Much attention has been paid to the effect of phytohormones on regenerative processes, especially the antagonistic effects of auxins and cytokinins (CKs) on the regeneration of roots and shoots. A relatively high auxin/CK ratio induces root regeneration, a low ratio is expected to induce shoot regeneration, and an intermediate ratio leads to unorganized growth resulting in the formation of calli (Skoog and Miller, 1957).

The mechanism by which CKs induce the regeneration of shoot apical meristems (SAMs) is not fully understood, but recent tissue culture research has shown renewed interest in the process in Arabidopsis (Cary et al., 2002; Che et al., 2002, 2006, 2007; Gordon et al., 2007), while other studies have screened for genes that are involved in CK-independent regeneration of shoots (Banno et al., 2001; Daimon et al., 2003; Ikeda et al., 2006). Several key studies have taken advantage of the shoot-forming capacity of Arabidopsis root explants to assess the regenerative capacity of mutants or transgenic lines in which hormone levels or sensitivity have been altered (Andersen et al., 2008; Daimon et al., 2003; Endrizzi et al., 1996; Ikeda et al., 2006; Kakimoto, 1996; Mähönen et al., 2006; Ozawa et al., 1998). Most of these studies have used the two-step procedure developed by Valvekens et al. (1988) or modifications thereof. In the first step, cells are grown on a medium that contains a high auxin/CK ratio. The cells proliferate and give rise to outgrowths that are classed as calli. The medium in which the cells are grown is commonly named callus-inducing medium (CIM). It is thought that cells acquire the competence to form shoots during this phase (Cary et al., 2002; Christianson and Warnick, 1983) after 2–3 days (Che et al., 2007). After five days on CIM, explants are transferred onto a shoot-inducing medium (SIM) with a high CK/auxin ratio. In this step, the so-called calli form shoots.

In the study reported herein, we conducted a detailed examination of the culture of Arabidopsis root and hypocotyl explants on CIM and SIM, and elucidated the developmental pathways that lead to SAM regeneration. We also determined the sites of daily expression of developmentally important genes. The first set of marker lines indicated the sites of expression of genes that are involved in the development of SAMs, such as WUSCHEL (WUS), CLAVATA (CLV) and SHOOT MERISTEMLESS (STM). Expression of STM and WUS is required for the establishment and maintenance of embryonic SAMs (Clark et al., 1996; Laux et al., 1996; Long et al., 1996), and ectopic WUS expression is sufficient to stimulate SAM regeneration (Gallois et al., 2004). The domain of WUS expression is limited to a group of deep central cells that constitute the SAM organizing centre (Mayer et al., 1998). The CLV genes (CLV1, CLV2 and CLV3) are expressed in the superficial layers of those SAM central cells that are considered to be stem cells (Fletcher and Meyerowitz, 2000). The CLV complex regulates the size of the stem cell population by restricting WUS expression, while WUS induces CLV3 (Schoof et al., 2000). The function of the CLV genes is antagonized bySTM (Clark et al., 1996), a gene from the KNOTTED-like homeobox (KNOX) family, which is SAM-specific and is required to prevent the incorporation of central cells into lateral primordia (Endrizzi et al., 1996; Long et al., 1996).

Causal links have been established between expression of these developmental genes and the presence of certain hormones. WUS represses the transcription of several ARABIDOPSIS RESPONSE REGULATOR genes (ARR5, ARR6, ARR7 and ARR15) that act in the negative feedback loop of CK signalling (Leibfried et al., 2005). In the presence of auxin, WUS over-expression induces somatic embryogenesis (Zuo et al., 2002). KNOX gene over-expression leads to the growth of ectopic buds on leaves, and increases the capacity of leaf explants to form shoots in tissue culture (Chuck et al., 1996; Hake et al., 2004; Sinha et al., 1993). Over-expression of the isopentenyltransferase (IPT) gene, which encodes a protein involved in CK biosynthesis, increases both shoot regeneration and the expression of KNOX genes (Rupp et al., 1999). Direct evidence that CKs are involved in mediating the control of SAM maintenance by KNOX genes has also been reported (Jasinski et al., 2005; Yanai et al., 2005). In addition, high auxin levels down-regulate the expression of KNOX genes (Shani et al., 2006), and it has been suggested that KNOX proteins may inhibit the transport of auxin (Soucek et al., 2007). Our collection of SAM-specific marker lines also included a line expressing AtML1 (MERISTEM LAYER 1), a gene that specifies the identify of the L1 epidermal layer of both SAMs and early lateral root meristems (LRMs) (Sessions et al., 1999).

The second set of marker lines used in the present study expressed root meristem-specific genes. In the root tips, stem cells for all cell types surround a small group of organizing cells, called the quiescent centre (QC), and together they form a niche of stem cells (Wildwater et al., 2005). QC25 is a QC-specific marker (Sabatini et al., 2003), while PLETHORA1 (PLT1) is expressed in both the QC and surrounding stem cells (Aida et al., 2004). Auxin accumulation is found in the QC and root apical meristems (RAMs) are sensitive to any perturbation in the endogenous level of auxin (Friml et al., 2003; Sabatini et al., 1999). PLT1 and normal auxin signalling are required for specification of the QC and maintenance of RAMs (Aida et al., 2004). ROOT CLAVATA-HOMOLOG1 (RCH1) is a RAM-specific gene that is expressed in the RAM cortical and external cells (Casamitjana-Martínez et al., 2003). The third set of marker lines used in the present study expressed markers for mitotic activity (AtCYCB1, which encodes the mitotic cyclin B1), auxin response (DR5; Ulmasov et al., 1997) and transport (PIN1; Galweiler et al., 1998), and CK response (ARR5; D’Agostino et al., 2000) and biosynthesis (IPT5; Miyawaki et al., 2004).

Using data obtained from these marker lines, we present a detailed view of the processes that occur during in vitro culture of Arabidopsis root and hypocotyl explants. We demonstrated that the outgrowths that were initiated on CIM, which were previously considered to be calli, were in fact LRM-like primordia that originated from the pericycle opposite the protoxylem poles. The regeneration of SAMs occurred as a result of re-determination of these LRM-like primordia, which changed into SAMs after transfer to SIM. We also showed that, when they were cultured directly on CK-containing media, xylem pericycle cells were able to regenerate SAMs directly in places where lateral roots should have been initiated, thus demonstrating their broad pluripotent character. This pluripotency was accompanied by the capacity to retain diploid status over several rounds of cell division, which allowed regeneration of true-to-type primary and secondary meristems.


Identification of target cells for de-differentiation on CIM

The plant lines used were from different ecotypes. Therefore, a preliminary study was performed to determine whether they all showed a similar morphological time course when cultured on CIM and SIM (Figure S1). The regenerative efficiency of explants from transgenic lines was demonstrated to be similar to that of their wild-type counterparts. The regenerated shoots developed roots easily and flowered normally after acclimatization.

Regardless of the line used, rapid internal modifications occurred when the explants were cultured on CIM. Although not visible macroscopically during the first two days (Figure 1a,b,j), there was active resumption of cell division, which started exclusively in the pericycle opposite the protoxylem poles (Figure 1d,e,m). Longitudinal sections showed that several contiguous xylem pericycle cells were involved (Figure 1g,p). After 3–5 days, the succession of periclinal and anticlinal divisions in the xylem pericycle gave rise to multiple protuberances positioned in two longitudinal rows (Figure 1e,f,k,l,n–r). This was in contrast to the phloem pericycle, in which only a few cells divided periclinally. After 4–5 days, tissues external to the pericycle derivatives were exfoliated (Figure 1f,l,o,r), such that explants became restricted to the initial central cylinder surrounded by protuberances that were derived from the xylem pericycle. These protuberances were covered with a continuous cell layer and were organized internally into cell files (Figure 1f,o,r). Some of the protuberances looked like LRM primordia and others were more rounded (Figure 1f,n,q,r). None of them had the unorganized appearance that is common to true calli. Depending on the ecotypes, the protuberances were more or less fused, with some of them looking like root fasciations.

Figure 1.

 Structural evolution of root and hypocotyl explants on CIM.
(a–c, j–l) Morphology.
(d–f, m–o) Transverse sections.
(g–i, p–r) Longitudinal sections showing the symmetric rows of dividing xylem pericycle cells together with more or less fused LRM-like structures.
Arrowheads on (d) and (e) indicate the protoxylem poles. Staining was with toluidine blue. Scale bars = 5 mm (a), 3 mm (b, c, j–l), 40 μm (d), 80 μm (e, f) or 130 μm (g–i, m–r).

LRM and non-LRM characteristics of the outgrowths initiated on CIM

Similarities between the outgrowths that initiated on CIM and early LRMs led us to examine their functional organization. Many common features (Figures 2–4) were found, including origination from the xylem pericycle (Figure 2a,l), reactivation of mitotic activity (Figure 2b,m), and expression of the DR5 auxin-responsive synthetic promoter (Figure 2d,e,o,p). As seen for LRMs occurring in planta, the LRM-like protuberances were characterized by early expression of PIN1 (Figure 2f,g,q,r), ARR5 (Figure 2h,i,s,t) and IPT (Figure 2j,k,u,v). However, some differences in these patterns were observed compared to LRMs in planta. (i) Initiation of LRM-like outgrowths involved 4–6 xylem pericycle cells instead of three (Figure 2a versus 2l). (ii) The DR5 signal was less concentrated in the tip and spread all over the protuberances (Figure 2e versus 2p). (iii) PIN1 accumulation was not so polarized nor centralized (Figure 2g versus 2r), indicating alterations in auxin polar transport. (iv) Sites of ARR5 expression were also less centralized in the internal cells (Figure 2i versus 2t). (v) IPT expression occurred in the whole surrounding cell layer instead of being restricted to the juvenile root cap (Figure 2k versus 2v). The external layer of the outgrowths had root cap characteristics, such as statoliths (Figure 3a, stars) and the AtML1 gene was expressed (Figure 3b,c) in the outermost layer (Figure 3c), as previously reported when LRMs are initiated (Sessions et al., 1999).

Figure 2.

 Cellular and functional LRM parameters of outgrowths initiated on CIM compared to true LRMs initiated in planta at early (a, b, d, f, h, j and m, o, q, s, u) and late (c, e, g, i, k and n, p, r, t, v) stages.
(a, l) Electron micrographs of transverse root sections at sites initiating LRM (a) or LRM-like structures (l). Stars indicate protoxylem poles; p, pericycle.
(b–k, m–v) Patterns of expression of reporter genes (GUS or GFP) fused to promoters of genes encoding markers of mitotic activity (CYCB1), auxin response (DR5) or transport (PIN1), and cytokinin response (ARR5) or biosynthesis (IPT5).
Scale bars = 20 μm (a, l) or 80 μm (b–k, m–v).

Figure 3.

 LRM parameters of the LRM-like structures initiated on CIM.
(a) Electron micrograph showing statoliths (stars) in the root cap-like layer.
(b, c) Expression of the AtML1 L1 marker gene after 5 days.
Scale bars = 2 μm (a), 80 μm (b) or 100 μm (c).

Figure 4.

 Patterns of expression of LRM-specific genes QC25, PLT1 and RCH1 in LRM-like structures initiated on CIM compared to LRMs initiated in planta.
Scale bars = 80 μm.

In addition, the RAM-specific marker genes QC25, PLT1 and RCH1 were expressed in the LRM-like protuberances (Figure 4). QC25 expression was restricted to QC cells in LRMs (Figure 4a). It was not expressed in the early pericycle derivatives on CIM (Figure 4b), but was expressed in a large central domain in 5-day-old LRM-like primordia (Figure 4c). As shown by Aida et al. (2004), PLT1 is expressed in the LRM stem-cell niche (Figure 4d); such expression was also found in the early stages of inititation of LRM-like protuberances (Figure 4e). The domain of PLT1 expression eventually covered a large area beneath the cap-like cells (Figure 4f). The RAM-specific RCH1 gene was also expressed in LRMs (Figure 4g) and LRM-like protuberances (Figure 4h,i). All these features indicated that the protuberances that were initiated on CIM were not true calluses but organized structures that resemble LRMs, in which PIN1 and QC25 expression was mislocalized, differentiation of correct radial organization did not occur, and a centralized stem-cell niche was absent. When such reactivated explants were maintained on CIM over a 5-day period, LRM-like primordia did not elongate but differentiated into superficial hairs that resembled root hairs (as in Figure 5a). Transfer to hormone-free medium resulted in some LRM-like protuberances converting into roots. When observed after 3 days on CIM (i.e. during the developmental step in which competence to form shoots was acquired; Che et al., 2007) and at the onset of their initiation, the LRM-like protuberances were found to comprise 2–4 cell layers (Figures 1e and 5d) (as for true LRMs during early initiation) and expressed the same genes as early LRMs.

Figure 5.

 Morphogenetic comparison of events induced by complete CIM and media containing only 2.2 μm 2,4-D, NAA or IAA, or 0.2 μm kinetin.
Large arrows, protoxylem poles; small arrows, periclinal divisions in phloem pericycle. Scale bars 1 mm (a–e), 80 μm (f–o, r–t) or 130 μm (p, q).

Surprisingly, LRM-like primordia that arose from root or hypocotyl explants that were grown on CIM expressed some SAM-specific genes (Figure S2), such as CLV3 and LFY. These genes are not usually expressed in LRMs. Occasionally, hypocotyls also expressed STM and CUC1 at a low level. Neither WUS nor CLV1 were expressed at any stage on CIM.

Differential effects of auxins and CKs on xylem and phloem pericycle cells

To determine the respective involvement of 2,4-D and kinetin in the earliest behaviour of xylem and phloem pericycle cells, root explants from the Col ecotype were placed on modified CIM that contained 2,4-D or kinetin alone. The presence of 2.2 μm 2,4-D produced LRM-like protuberances that originated from the xylem pericycle (Figure 5a,f,k,p), as did complete CIM (Figure 5b,g,l,q). When explants were left on CIM or in the presence of 2,4-D alone for more than 5 days, the protuberances increased in size (Figure 5a,b and Figure S3) and showed differentiation of superficial root hairs, particularly with 2,4-D alone. Such protuberances remained provided with a limiting layer for up to 15–20 days (Figure S3), before becoming progressively disorganized, and the increasing disorganization was accompanied by an increasing loss of capacity to regenerate shoots (Figure S3). In the early stages, xylem pericycle cells reacted by undergoing periclinal divisions, but no divisions occurred at the phloem poles (Figure 5g,l). After 5 days in the presence of 2,4-D alone, the protuberances were less fused than those grown on CIM (Figure 5p,q). In addition, spontaneous LRMs that were present in the initial explants lost the capacity to elongate. Reducing the amount of 2,4-D from 2.2 to 0.44 μm induced few protuberances, which were spaced far apart, and 0.22 μm induced only occasional true lateral LRMs. At a concentration of 4.4 μm 2,4-D, most xylem pericycle cells reactivated, which gave rise to continuous protuberances along both xylem poles. This indicated that the number of xylem pericycle cells that were recruited longitudinally increased with increasing 2,4-D concentrations, as has already been reported to occur with increasing concentrations of naphthaleneacetic acid (NAA; Benkovàet al., 2003) and indoleacetic acid (IAA; Laskowski et al., 1995). In all cases, 2,4-D acted only on xylem pericycle cells, and, depending on the concentration, initiated organized meristematic structures that ranged from true lateral LRMs to completely fused LRM-like protuberances.

The results for kinetin contrasted markedly with those for 2,4-D. Regardless of the concentration, kinetin induced greening of the explants but never activated pericycle cell division at the xylem poles; neither did it produce LRM-like structures. Kinetin at 0.2 μm and higher concentrations (0.4, 4 and 20 μm) induced only periclinal divisions in phloem pericycle cells (Figure 5h,m, small arrows). This produced 2–4 cambial-like layers that were devoid of organogenetic properties (Figure 5r). At sites where ordinary lateral LRMs were formed, 4–20 μm kinetin induced vacuolation of meristematic cells and loss of LRM determination by arresting PLT1 expression. Consequently, we conclude that the 2,4-D contained in CIM was responsible for the reactivation of xylem pericycle cells and the initiation of the LRM-like structures, while kinetin induced only re-entry of some phloem pericycle cells into the cell cycle, and these did not participate in initiation of the protuberances.

Comparing the effects of 2.2 μm 2,4-D with those of NAA and IAA at similar molar concentrations, we observed that both auxins induced true LRMs from xylem pericycle cells, as usually reported. The few LRMs observed after 2.2 μm IAA treatment were narrow and widely spaced (Figure 5e,j,o,t), as seen on hormone-free medium, but were larger, more numerous, and grouped in clusters of two or three in response to NAA treatment (Figure 5d,i,n,s). In both cases, the LRMs displayed classical organization and gave rise to elongated roots.

Indirect origin of SAMs regenerated from LRM-like protuberances after transfer to SIM

Four days after transfer onto SIM, the LRM-like protuberances had increased in size and turned green (Figure 6c,i). The first leafy shoots became visible after 5–7 days for root explants and 8–9 days for hypocotyl explants. They developed at the surface of the LRM-like protuberances (Figure 6c,e,k,m) in a polar manner, mainly in the basal region of the root explants (Figure 6o) and in the apical region of the hypocotyl explants (Figure 6p). True roots developed at the opposite ends, despite the high CK concentration. The extent to which the LRM-like protuberances developed depended on the ecotype and the polarity, with the result that regenerated SAMs arose further from (Figure 6f,n) or closer to (Figure 6d,l) the initial central cylinder. The surrounding root cap-like layer persisted for up to 7–9 days (Figure 6h,j,l) unless SAMs formed. Cells located in the region of the expected root-like stem-cell niche underwent re-organization after transfer to SIM (Figure 6j). They reformed into roundish domains of small, isodiametric meristematic cells (Figure 6l,n) from which regenerated SAMs emerged (Figure 6k,n), and the external root cap-like layer became disrupted (Figure 6d,n). Thus, indirect SAM regeneration occurred through reconversion of early LRM-like protuberances into SAMs, which suggested a change of stem-cell determination after transfer to CK-rich medium. Note that the differential capacity of Arabidopsis ecotypes to regenerate shoots was not correlated with their initial capacity to regenerate LRM-like structures, but rather with the capacity of these structures to switch into SAMs.

Figure 6.

 Morphogenetic events following the transfer onto SIM of root explants (a–f) and hypocotyl explants (g–n) that had previously been grown for 5 days on CIM (a–n).
Regenerated shoots are recognizable by the presence of anthocyanidins in leaf primordia. Regeneration occurred in a polar manner in both root (o) and hypocotyl (p) explants. Arrowheads, protoxylem poles.
Scale bars = 1.5 mm (a, c, e, g, i, k, m, o, p) or 160 μm (b, d, f, j, h, l, n).

In situ gene expression during re-determination of early LRM-like protuberances into SAMs

Although SAM regeneration was asynchronous, careful microscopic observations of events that occurred after transfer onto SIM allowed us to distinguish gene expression at sites in which SAMs were presumed to form from that at sites that were not involved in the process. In both root and hypocotyl explants, expression of a number of SAM-specific genes occurred in the LRM-like structures prior to any visible sign of de novo SAM organization. Correspondingly, LRM-specific genes (QC25, RCH1 and PLT1) and PIN1 were down-regulated after 2–3 days on SIM, i.e. before formation of the round-shaped domain.

The WUS gene (which was not expressed during growth on CIM) was expressed at the same time as greening of the explants occurred after 3–4 days on SIM. Expression started in the innermost pericycle derivatives (Figure 7a); then expression stopped in the inner derivatives and new expression occurred in the round-shaped domain (Figure 7b,c). Expression in this domain became progressively restricted to the few cells that corresponded to the organizing centre of the incipient SAM (Figure 7d,e). CLV3 was initially expressed on CIM and continued to be strongly expressed on SIM in deep domains during the first 3 days (Figure 7f). It was then expressed in both the deep cells and the round-shaped domain of LRM-like protuberances, with the exception of the external root cap-like layer (Figure 7g,h). Expression of CLV3 continued in deeper cell layers in which no SAM formation was expected and in large areas of the nascent SAMs (Figure 7i,j). The CLV1 gene was not expressed on CIM and only weakly expressed in internal cells after 2–3 days on SIM (Figure 7k). The location of CLV1 expression changed after 5 days to a well-delimited area within the round-shaped domain (Figure 7l,m). At 7–9 days on SIM, CLV1 expression was confined exclusively to the newly formed SAMs in an area that surrounded the previously described WUS-expressing domain (Figure 7n,o). The STM gene, which was rarely expressed on CIM, showed significant expression after 3 days on SIM, in the innermost cells of the explants (Figure 7p). After that, internal expression was no longer observed, and STM expression was only observed in LRM-like protuberances with properties that were indicators of shoot formation (Figure 7q,r). At 7–9 days, STM expression was restricted to the newly formed SAMs (Figure 7s,t). Lastly, the AtML1 gene, which was already expressed superficially in the LRM-like protuberances on CIM, continued to be expressed after transfer to SIM (Figure 7u). Thereafter, AtML1 gene expression became restricted to the tip of the protuberances (Figure 7v). More precisely, expression of AtML1 in the superficial layer stopped before exfoliation. Expression of AtML1 then started in the underlying layer which became the external layer of the round-shaped domain (Figure 7w) and of the new shoots (Figure 7x,y).

Figure 7.

 SAM-specific gene expression in root explants after transfer onto SIM during the time course of SAM regeneration from LRM-like protuberances.
Scale bars = 1 mm (a, d, g, i, k, q, x), 250 μm (b, f, l, n, p, s, u, v) or 80 μm (c, e, h, j, m, o, r, t, w, y).

This detailed analysis revealed that SAM regeneration occurs by reconversion of the LRM-like protuberances that was achieved through orchestrated patterns of gene expression in the round-shaped domain that formed in the LRM-like structures. Not all the LRM-like protuberances were at similar stages of development. This precluded elucidation of more precise kinetics of the successive involvement of SAM-specific genes. However, WUS expression occurred before that of CLV1 and CLV3 at sites of SAM formation. Furthermore, STM expression was observed either before or afterWUS expression at these sites, which suggested that both genes are involved in the process independently. The L1 layer of the incipient SAM was identified throughAtML1 expression prior to WUS expression, which suggests that the L1 cells play a role in subsequent SAM organization. It should also be noted that global analysis of gene expression could not account for the kinetics that are related specifically to SAM initiation, because gene expression occurred in sites that were not involved in the process.

Direct SAM regeneration from root xylem pericycle upon direct culture on SIM or CK-containing media

To determine whether it was possible to bypass the first step on CIM that led to development of the shoot-forming LRM-like protuberances, root and hypocotyl explants were grown directly on either classical SIM, or modified SIM that contained only 24.6 μm 2-isopentenyladenine (2-iP) or kinetin. Shoots regenerated within 7 days at various points along the root explants (Figure 8a–i) in addition to the crown, which is where the first lateral roots normally form. These shoots appeared acropetally in two rows, in an alternating manner. The mean spaces between two successive shoots were similar to those between lateral roots. This indicates that direct shoot formation occurred in sites where LRM initiation would have occurred. Such replacement of LRMs by SAMs was observed only with high CK concentrations (24.6 or 49.2 μm). In contrast, regardless of ecotype, no shoots formed from hypocotyls that had no pre-determined sites for LRM initiation. All the SAMs that regenerated directly formed initially as lateral roots from xylem pericycle cells (Figure 8j, arrows). The emergence of SAMs was accompanied by exfoliation of external tissues of the initial explant (Figure 8k,l). These results indicate that the shoot-forming effect of CK was seen only at sites that were already involved in early stages of spontaneous lateral rooting, and thus were competent to respond to the CK signal. At these sites, expression of CLV3 in the xylem pericycle started within the first day (Figure 9a,e) and WUS during the second day (Figure 9i,l,m). STM expression started later, after 3–5 days (Figure 9v). As in the case of indirect SAM regeneration, CLV3 was not expressed exclusively at SAM-forming sites (Figure 9d,h). At these sites, CLV3 expression occurred in a zone similar to the previously described round-shaped domain (Figure 9d), before spreading over the entire regenerated SAM (Figure 9h). WUS expression began in the central cells of the developing SAM (Figure 9l) and was subsequently restricted to the expected location for the organizing centre (Figure 9p).

Figure 8.

 Direct shoot regeneration on root explants grown for 3–12 days on media containing 24.6 μm 2-iP and 0.9 μm IAA (SIM), or only 24.6 μm 2-iP or kinetin.
Regenerated SAMs arose exclusively from xylem pericycle cells (j–l) at sites normally involved in spontaneous LRM initiation. Conversion of early LRMs into SAMs occurred, as in the two-step procedure, through establishment of a round-shaped domain of cells dividing in all planes (k) at the surface of which SAMs were organized (l). Arrowheads, protoxylem poles.
Scale bars = 10 mm (a–i) or 80 μm (j–l).

Figure 9.

 Sites of expression of SAM-specific genes during direct SAM formation from xylem pericycle of root explants grown on 24.6 μm 2-iP-containing medium.
Scale bars = 150 μm (a–c, e–g, i–k, m–o, q–s, u–w) or 80 μm (d, h, l, p, t, x).

Expression of WUS and STM started earlier on 2-iP-containing medium than on SIM, which suggests that both genes are regulated positively by CKs and negatively by auxins. This negative regulation might explain the absence of WUS and STM expression on auxin-rich CIM that was reported previously. Further experiments on conditions for direct SAM regeneration were performed using solitaryroot1 (slr1) explants. The slr1 mutant, a gain-of-function mutation in the IAA14 gene that is involved in auxin signalling, is deficient in LRM initiation (Fukaki et al., 2002). However, it reacts to CYCD3 over-expression by triggering pericycle cell division that does not result in LRM formation (Vanneste et al., 2005). When grown for 5 days on CIM, and then on SIM, slr1 root explants were able to regenerate some SAMs from the dividing xylem pericycle cells (data not shown). This suggests that the shoot-forming capacity of CKs requires dividing cells that are not necessarily committed to LRM determination.

Capacity of xylem pericycle derivatives to maintain diploidy over long-term cell proliferation

We used nuclear DNA flow cytometry to determine whether regenerated SAMs had the same ploidy levels as SAMs in planta, and whether indirect regeneration through re-specification of early LRM-like protuberances allowed clonal fidelity of the regenerants. Both roots and hypocotyls from 7-day-old plantlets from various genetic backgrounds were initially largely polyploidy, with DNA contents ranging from 2C to 16C (Figure 10a, labelled To). Culture on CIM for 5 days resulted in enrichment of the explants in diploid cells during establishment of LRM-like protuberances (Figure 10a). Nine days after transfer to SIM, we separately analysed non-dividing parts of the initial explants, LRM-like protuberances devoid of SAMs, and regenerated SAMs. This revealed that the regenerated SAMs and LRM-like protuberances contained a majority of 2C and 4C nuclei (Figure 10b). Root explants that were maintained for 40 days on CIM, and that consequently provided a higher number of proliferative pericycle cell derivatives, were also analysed and found to be mainly diploid (Figure 10c).

Figure 10.

 Frequencies (F) of nuclear DNA contents (qDNA; semi-log scale) in initial explants (To), after 5 or 40 days on CIM, and in several parts of shoot-forming root explants after transfer onto SIM.


We studied in detail the process of SAM regeneration from Arabidopsis root and hypocotyl explants grown in vitro, using either a two-step procedure (Valvekens et al., 1988) or direct culture in the presence of CKs. We show that the xylem pericycle had a higher pluripotency than expected, and present a comprehensive and new framework for users of Arabidopsis tissue culture, with insights into competence and determination.

Xylem pericycle cells are more pluripotent than usually thought

The root pericycle is the first layer of the central stele to be derived from the procambium that is already recognizable above the QC. In most spermatophytes, pericycle cells in contact with the xylem poles are direct target sites for auxin-induced LRMs that are formed through periclinal and anticlinal cell divisions. The process of LRM formation in Arabidopsis roots has been extensively studied in recent years (Beeckman et al., 2001; Blakely et al., 1988; Casimiro et al., 2003; Dubrovski et al., 2000; Himanen et al., 2002, 2004; Malamy and Benfey, 1997; Parizot et al., 2008; de Smet et al., 2006). LRMs initiate along two files that correspond to both xylem pericycle cell files, with variable numbers and spacing. All pericycle cells remain more meristematic than the surrounding cells (Dolan et al., 1993). They are also able to retain expression of cell-cycle genes such as CDC2a (Hemerly et al., 1993; Martinez et al., 1992) and CYCA2 (Burssens et al., 2000). These properties are probably responsible for the capacity of pericycle cells to rapidly re-enter the cell cycle without undergoing real de-differentiation. In addition to root-forming ability, pericycle cells are involved in secondary root growth and divide periclinally to give rise to the phellogen and parts of the vascular cambium (Esau, 1965). We showed that when root explants were cultured directly in the presence of CKs, the xylem pericycle regenerated SAMs indirectly through LRM-like structures, or directly at sites involved in LRM formation. Although the shoot-forming capacity of the pericycle has, as far as we are aware, never been reported in Arabidopsis, it has been described in some woody and herbaceous species (Bonnett and Torrey, 1966; Projetti and Chriqui, 1986). The pericyclic origin of the outgrowths that are initiated on CIM has previously been reported by Cary et al. (2002) and Gordon et al. (2007). However, it had not been assigned precisely to the xylem poles, nor has the direct generation of SAMs from the xylem pericycle been reported. In addition to their root-forming capacity and their involvement in cambium formation, the shoot-forming capacity of xylem pericycle cells described here led us to conclude that they are more pluripotent than has previously been thought. Such observations reinforce the concept of the pericycle as an ‘extended meristem’ (de Smet et al., 2006). Determination of the ability of the xylem pericycle to express genes that are involved in the specification of LRM or SAM stem cells provides a foundation for further research to establish whether this tissue is a true stem-cell layer. Calli that are established in tissue culture from polysomatic species display frequent variations in their ploidy levels (Lee and Phillips, 1988). Although most organs of Arabidopsis undergo endo-duplication during differentiation, pericycle cells are diploid (Beeckman et al., 2001). We found that the xylem pericycle derivatives were able to retain diploidy over several rounds of cell division, which ensured the establishment of true-to-type primary and secondary meristems. This suggests that xylem pericycle cells have particular mechanisms for maintaining clonal fidelity that could also be part of their pluripotency.

Xylem and phloem pericycle cells react differently to CKs

In addition to their distinct involvement in primary and secondary meristem establishment as discussed above, xylem and phloem pericycle cells have a number of differences. These include size (Casero et al., 1989; Dubrovsky et al., 2000; Laskowski et al., 1995), ultrastructure (Himanen et al., 2004; Parizot et al., 2008) and distribution of plasmodesmata (Complainville et al., 2003). There are also differences in the levels of extensins and arabinogalactans (Casero et al., 1998; Knox et al., 1989), and in their reactions to inhibition of auxin transport and treatment with NAA (Himanen et al., 2004; Parizot et al., 2008). Following auxin treatment, xylem pericycle cells undergo periclinal and anticlinal mitoses, which lead to LRM formation. By contrast, phloem pericycle cells do not divide, which is most likely a result of expression of the KRP2 cyclin-dependent kinase inhibitor gene that blocks transition from G1 to S phase (Himanen et al., 2002). In our study, we found that the phloem pericycle did not react to auxins but divided periclinally in response to CK treatment. This raises questions about the involvement of hormones in controlling planes of cell division and the involvement of CK in establishing secondary meristems. We found that the shoot-forming effects of CK depended on the developmental stage of the xylem pericycle and its derivatives.

In whole plants, exogenous CKs repress LRM formation (Li et al., 2006; Wightman et al., 1980), and transgenic lines that over-express genes for CK degradation display enhanced root branching (Werner et al., 2003). Increasing CK levels, through by IPT expression, targeted to either the xylem pericycle or young lateral root primordia, revealed that only xylem pericycle cells were CK-sensitive. This indicates that CKs have direct effects on root founder cells that might be related to delayed cell division (Laplaze et al., 2007) and blockage of transition from G2 to M phase (Li et al., 2006). Upon culture of root explants directly in the presence of CK, we found that LRM-forming sites present at time 0, and LRM-like structures initiated on CIM, responded to CK by switching their end state to SAM initiation. These responses are both stage- and concentration-dependent. Shoot-forming responses occur only in dividing cells that are already engaged in LRM initiation but not yet committed to late stages of LRM patterning. This suggests that re-specification of LRM or LRM-like primordia into SAMs occurs only during a particular developmental window, and that LRM founder cells are not initially fully specified for this developmental pathway.

Outgrowths initiated on CIM are not calli but LRM-like structures

It is commonly accepted that outgrowths that are initiated from root explants during pre-incubation on 2,4-D-rich CIM are calli. However, we demonstrated that, regardless of ecotype, the outgrowths were not true calli but structures that had a range of LRM characteristics. These LRM-like structures originated as LRMs, had a root cap-like layer, and expressed root meristem-specific genes such as RCH1, PLT1 and QC25 in enlarged domains of expression that are comparable to those of LRMs. Other differences from true LRMs existed with respect to the number of pericycle cells that participate in LRM initiation, reduced spacing up to fasciation, and absence of elongation. LRM-like primordia had an altered PIN1 polarity and DR5 expression gradient. They also lacked apical organization, with a correctly positioned QC provided with local auxin accumulation. In contrast to results obtained with IAA and NAA, 2,4-D promotes division but not elongation of tobacco cells (Campanoni and Nick, 2005). Instead of inducing normal lateral roots in chicory, 2,4-D stimulates pericycle reactivation but inhibits the conversion of primordia to lateral roots (Vuylsteker et al., 1998). During LRM formation, PIN1 localization is detected in the founder cells, followed by polarization that points towards the primordium tip, a relocation that correlates with DR5 gradients (Benkovàet al., 2003). Defects in LRM development, very similar to those in LRM-like primordia, have been reported to occur in NAA-treated pin3 pin7 and pin1 pin3 pin4 mutants that are deficient in auxin efflux (Benkovàet al., 2003). gnom mutants that are impaired in polar auxin transport also react to 2,4-D by pericycle cell proliferation. This gives rise to multi-layered structures that are not organized into LRMs, but display depolarized PIN1 localization and absence of DR5::GUS activity (Geldner et al., 2004). 2,4-D also has effects on the levels of auxin, ethylene and abscisic acid (Raghavan et al., 2006). High 2,4-D concentrations modify roots through impaired auxin flow and root patterning, which could explain the LRM-like phenotype that is induced by CIM. Finally, other genes that are expressed during early LRM formation, such as IPT5 (Miyawaki et al., 2004) and ARR5 (Lohar et al., 2004), are also expressed in LRM-like protuberances. The finding that CIM induces LRM-like primordia rather than calli should be kept in mind when using Arabidopsis root cultures to investigate hormone signalling or responses.

The competence to regenerate shoots is associated with particular steps of LRM or LRM-like development

As reported by Che et al. (2002, 2007), 2–3 days growth on CIM are required to reach the developmental stages required for shoot formation upon transfer to SIM. We found that the capacity to regenerate shoots in response to CK was present in root explants. This occurs at sites that are engaged in spontaneous LRM formation, at which the cells are dividing but are not yet organized into mature LRMs. More than 12–15 days growth on CIM leads to large LRM-like protuberances that progressively lose their capacity to form shoots. Old LRM-like structures were partly disorganized, which suggests that there was no clear boundary between LRM-like outgrowths and those that could be considered to be true calli. This indicates that the capacity to respond to SIM or CKs requires a dividing cell state that is provided with early parameters of LRM or LRM-like determination. Pre-incubation on CIM seems only to be useful for increasing the number of sites that enter the first stages of LRM initiation, which correspondingly increases the potential number of regenerated shoots.

Shoot regeneration proceeds from re-determination of early LRM-like structures into SAMs

We have shown that the regeneration of shoots by both direct and indirect means occurred through CK-induced transdifferentiation of early LRMs into SAMs, rather than through real cell de-differentiation. Re-determination of Arabidopsis early RAMs into SAMs has been reported to occur via ectopic expression of WUS (Gallois et al., 2004; Zuo et al., 2002) but never via CK treatment. As previously shown by Che et al. (2002, 2007), we found that CK-response regulators (A-type ARR genes) were expressed during growth on CIM. More precisely, ARR5 was expressed in the central domain of the LRM-like primordia (Figure 2i,t), which reconverts into the round-shaped domain from which SAMs regenerate. Despite the lack of lineage studies, observation by microscope suggests that this domain is established in place of the nascent root stem-cell niche. CKs prevent formation of the auxin gradient that is required to pattern root primordia (Laplaze et al., 2007). It may be hypothesized that the early stages of LRM initiation are pluripotent with potential alternative end states, and are able to produce either RAMs or SAMs depending on the auxin/CK ratio in subsequent culture. The capacity to form shoots is acquired when the LRM-like protuberances or LRMs are still at the early stages of initiation, when root stem-cell niches have just been established, as indicated by PLT1 expression. This may be related to the recent finding that CK transiently antagonizes the ability of auxins to specify the root stem-cell niche during zygotic embryogenesis (Müller and Sheen, 2008). CKs also inhibit cell proliferation in root meristems (Dello Ioio et al., 2007; Li et al., 2006; Werner et al., 2003). Mis-expression of CLV3, as observed on CIM, induces consumption of root meristems (Fiers et al., 2005). All these factors may contribute to the loss of LRM identity after transfer onto SIM. The results from our study provide a significant foundation for further studies on the process of meristem transdifferentiation, and may be helpful in enabling recalcitrant species to regenerate.

Experimental procedures

Plant materials and growth conditions

We used a range of homozygous transgenic lines of Arabidopsis thaliana (L. Heynh) that carried either a GUS or GFP expression cassette fused to various promoters, and the corresponding wild-type ecotypes. The CLV1::GUS (Schoof et al., 2000) and CLV3::GUS (Groß-Hardt et al., 2002) lines were in the Ler background. The WUS::GUS (Groß-Hardt et al., 2002) line was in the Col-0 × Ler background. The STM::GUS (Long et al., 1996), QC25::GUS (Sabatini et al., 2003), ARR5::GUS (D’Agostino et al., 2000), IPT5::GUS (Miyawaki et al., 2004) and LFY::GUS (Weigel et al., 1992) lines were in the Ws background. The AtML1::GUS2 (Sessions et al., 1999), CYCB1;1::GUS (which included a labile fusion protein; Colón-Carmona et al., 1999), DR5::GUS (Ulmasov et al., 1997), PIN1::PIN1:GFP (Galweiler et al., 1998), RCH1::GFP (Casamitjana-Martínez et al., 2003) and PLT1::GFP (Aida et al., 2004) lines were in the Col-0 background. The CUC1::GFP (Mo 223::GFP; Cary et al., 2002) line was in the C24 background.

Seeds were surface-sterilized with 70% ethanol for 2 min and 5% calcium hydrochloride for 5 min, then washed in sterile water before being sown on Petri dishes that contained Murashige and Skoog medium (1962) (Duchefa, supplemented with 10 g l−1 sucrose and 0.7% agar (Duchefa). The pH was adjusted to 5.7 with 1 m KOH. The prepared dishes were cold-treated for 24 h at 4°C before transfer to the growth chamber under a 16 h light/8 h dark cycle at 25 ± 1°C with 110 μmol m−2 sec−1 photon flux intensity delivered by Biolux tubes L58W/840 (Osram,

Root and hypocotyl explants were excised from 7-day-old sterile seedlings. For the two-step procedure (Valvekens et al., 1988), explants were first placed on CIM that consisted of B5 medium (Gamborg et al., 1968) supplemented with 0.5 g l−1 2-(N-morpholino) ethane sulphonic acid, 2.2 μm 2,4-D, 0.2 μm kinetin, 2% glucose and 0.7% agar. After 5 days, they were transferred onto SIM, prepared as CIM with respect to basal composition, but containing 24.6 μm 2-iP and 0.9 μm IAA as hormones. The hormones were dissolved in DMSO and added to the medium after autoclaving. In order to examine the effects of the auxins in CIM, modified media that contained 2.2 μm NAA or IAA, or other 2,4-D concentrations, were utilized. Other modified CIM or SIM that contained only 2-iP or kinetin were also used. The conditions under which the explants were grown were as described above.

Macroscopic and microscopic GUS assays

Five to ten samples per condition were fixed at 4°C in 90% acetone for 10 min, rinsed three times with 100 mm sodium phosphate buffer (pH 7), and then incubated for 30 min at room temperature in the same buffer containing 2 mm potassium ferricyanide and 0.5 mm potassium ferrocyanide (Jefferson et al., 1987). The samples were then incubated overnight at 37°C in the same medium supplemented with 2 mm X-Gluc (Duchefa) after vacuum infiltration. The reaction was stopped by washing with 100 mm phosphate buffer. Whole samples were examined using a Zeiss stereomicroscope (, and images were captured using a Zeiss AxioCam camera. For in situ studies, representative GUS-positive samples were fixed at 4°C for 1 h in 1.5% glutaraldehyde/4% paraformaldehyde in 50 mm sodium phosphate buffer, pH 7. They were then washed in phosphate buffer, dehydrated, and embedded in Araldite 502/Embed 812/DDSA/BDMA (15/25/55/2.4 ml; Electron Microscopy Sciences, Sections 3 μm thick were cut using a Microm HM 355 (, counterstained with 1% aqueous basic fucshin, and photographed using a Zeiss Axioscop microscope equipped with a Zeiss AxioCam MR camera.

GFP and CFP detection

GFP and CFP fluorescence were visualized with an AxioImager Z1 fluorescence microscope equipped with a Zeiss ApoTome structured illumination system. The filters used were 38 HE-GFP (excitation 470 nm; emission 525 nm) for GFP and 47 HE-CFP (excitation 436 nm; emission 480 nm) for CFP. Some samples were counterstained with FM4-64 (Invitrogen;

Electron microscopy

The samples were fixed with 2% glutaraldehyde/1.5% paraformaldehyde in 0.1 m sodium cacodylate buffer, pH 7.5, for 4 h at room temperature. After washing, they were post-fixed in 1% osmium tetroxide in 0.1 m sodium cacodylate buffer for 6 h at room temperature. They were then rinsed, dehydrated in an ethanol/propylene oxide series, and embedded as described above. Ultrathin sections (80–90 nm) were stained with saturated uranyl acetate in 50% ethanol and lead citrate (Reynolds, 1963) for 15 min each. Grids were observed at 80 kV under a Philips EM 201 transmission electron microscope ( at the Electron Microscopy Service of IFR 83 (Université Pierre et Marie Curie).

DNA flow cytometry

Nuclei were isolated by mechanical chopping with a razor blade in Galbraith’s buffer (Galbraith et al., 1983) supplemented with 5 μg ml−1 Hoechst 33342 (Sigma-Aldrich, The extracted nuclei were filtered through a 40 μm nylon mesh. Analysis was performed on an EPICS Elite ESP flow cytometer (Beckman Coulter; with laser excitation (40 mV) at 351–364 nm, at IFR 87 (Gif-sur-Yvette, France). Five to ten thousand nuclei were analysed per sample, and 5–10 samples were pooled for each analysis.


The authors are grateful to T. Laux and M. Lenhard (University of Friburg, Germany), M. Yanovski (University of California, San Diego, U.S.A.), E. Benkovà (University of Tübingen, Germany), T. Schmülling (University of Berlin, Germany), P. Laufs (INRA, Versailles, France), P. Benfey (Duke University, Durham, U.S.A.) and B. Scheres (University of Wageningen, The Netherlands) for providing the lines used in this study. They also thank R. Sablowski (John Innes Centre, Norwich, U.K.), T. Beeckman (VIB, Gent, Belgium) and G. Vigil for fruitful discussions, P. Dumont for greenhouse and seed collection management and O. Catrice (CNRS, Gif-sur-Yvette, France) for his assistance in flow cytometry. R. Atta was a PhD student with a financial support from the Egyptian government.