Main

Plant organs grow axially during primary development and then radially during secondary development. In many plant organs—including woody stems—secondary development is organized by two concentric, cylindrical meristems that undergo periclinal cell divisions (that is, the division plane is parallel to the surface of an organ or a tissue): the inner vascular cambium proliferates to produce conductive xylem and phloem, and the outer phellogen (also called cork cambium) produces protective tissue called phellem4. In the root of A. thaliana, secondary meristems originate from parenchymal cells located between primary xylem and phloem (a tissue known as procambium) and from cells surrounding the procambium, xylem and phloem, which form a tissue known as the pericycle5,6 (Fig. 1a, Extended Data Fig. 1a, b).

Fig. 1: Lineage tracing revealed the origin of the vascular cambium stem cells.
figure 1

a, The activation of secondary root growth is characterized by the initiation of periclinal cell divisions in procambial cells (black arrowheads), followed by divisions in pericycle (white arrowheads), leading to the formation of vascular cambium and phellogen. Cell divisions are marked only in top half of the cross-section. b, Lineage tracing initiated from the primary vasculature (top) showed the importance of an initial xylem connection for the ontogeny of the cambium (yellow dashed line). The bottom panel shows the four major types of clonal sectors obtained. Frames are coloured to match the corresponding cell lineages in the top panels. The numbers in the fate map (bottom right) refer to the observed number of the sectors. c, d, Lineage tracing in active cambium revealed a common stem cell for xylem and phloem. GUS sectors (black bars) plotted on the basis of the position of thinnest cell wall (arrow) in each radial file, which represents the most recent periclinal cell division in cambium (y = 0). Scale bars, 50 μm. All experiments were repeated four times.

Full size image

To understand, at cellular resolution, how the vascular cambium and phellogen develop from the procambium and pericycle, we designed a Cre–lox recombination system to carry out cell-lineage tracing in the Arabidopsis root (Extended Data Fig. 2a, b). Clones marked by expression of β-glucuronidase (GUS) were induced during the activation of root secondary growth (Fig. 1a, b), and the subsequent sectors were analysed after 11 days. GUS sectors showed highly repetitive patterns that indicated that the secondary development is a tightly controlled process (Fig. 1b, Extended Data Fig. 2c). This enabled us to develop a fate map of procambial and pericycle cells during the morphogenesis of root secondary tissue (Fig. 1b). The fate map showed that phellogen originates from the pericycle (dark and light purple, in Fig. 1b), confirming previous anatomical observations4,5,6. Pericycle cells adjacent to the primary xylem (xylem-pole pericycle (XPP) cells, dark purple in Fig. 1b) are also the founder cells of the vascular cambium. The proliferation of XPP cells thus leads to the separation of the two concentric meristems and production of all the secondary tissues from their circumferential position (Extended Data Fig. 2d). Our finding underscores the pluripotency of XPP cells, which are also the source of lateral roots and regenerated tissue (callus)7. The fate map also revealed that only those procambial cells that are in physical contact with primary xylem vessels (red, in Fig. 1b) give rise to vascular cambium, secondary xylem, and phloem. Thus, these xylem-adjacent procambial cells, together with xylem-adjacent daughters of XPP cells, act as the stem cells of vascular cambium during the activation of secondary growth. The fact that procambial and pericycle cells that are not in contact with xylem (pink and light purple, respectively, in Fig. 1b) do not form vascular cambium suggests that cellular position (that is, being adjacent to xylem), rather than identity during primary development, determines the origin of the vascular cambium. This conclusion is further supported by lineage tracing in roots that have transiently been treated with the phytohormone cytokinin, which imparts procambium identity to cells in the protoxylem position. These cells are located adjacent to the remaining xylem and formed only vascular cambium, whereas cells in the XPP position—no longer adjacent to xylem cells—formed only phellogen and not vascular cambium (Extended Data Fig. 2d–g). Together, these findings suggest that xylem or xylem precursor cells cause neighbouring cells to become vascular cambium stem cells.

At the onset of secondary growth (Fig. 1a), primary xylem cells have already undergone programmed cell death as the final step of their differentiation. We therefore hypothesized that xylem induces stem-cell identity in the neighbouring cells earlier, during primary development, when cells with xylem identity are alive and capable of signal transduction. A family of five redundantly acting HD-ZIP III transcription factors determine xylem identity during the primary development of the Arabidopsis root8,9,10. Additionally, an HD-ZIP III orthologue in Populus regulates vascular cambium morphogenesis11. HD-ZIP III transcription factors are thus likely candidates for regulating signals for stem-cell identity. HD-ZIP III transcription factors are expressed in the cambial zone; AtHB8 at least is expressed predominantly on the xylem side of the vascular cambium, adjacent to the cells undergoing cell division (highlighted with expression of AINTEGUMENTA (ANT)), which is consistent with the idea that genes for HD-ZIP III transcription factors mark cells with secondary xylem identity (Fig. 2a, Extended Data Fig. 3a, b). Only mutants that lack at least four of the HD-ZIP III family genes show patterning defects and reduced xylem formation in the primary root vasculature8. The same mutant combinations also show scattered cambial divisions and xylem patterning during secondary development (Extended Data Fig. 3c, d). To examine whether these severe developmental defects are a direct consequence of the loss of HD-ZIP III transcription factors or a secondary consequence of defects in primary development, we constructed a line with an inducible12 miR165a, a microRNA (miRNA) that targets the mRNAs of all five HD-ZIP III transcription factors13 (Extended Data Fig. 4a). Ubiquitous induction of miR165a from germination onwards led to severe inhibition of primary xylem differentiation (Extended Data Fig. 4b), similar to the phenotype of the quintuple HD-ZIP III mutant8. Induction of miR165a during the activation of secondary growth initially blocked vessel differentiation, which indicates that the HD-ZIP III transcription factors are also required for secondary xylem-vessel formation (Fig. 2b). Dispersed cell divisions and erratic xylem-vessel formation occurred in the remaining undifferentiated tissue, resembling the phenotypes of quadruple HD-ZIP III mutants (Extended Data Fig. 3d). Induction of miR165a during secondary growth forced xylem parenchyma cells and cells with xylem identity—which were previously quiescent—to re-enter mitosis, generating a population of dividing cells in the place of xylem (Fig. 2c, Extended Data Fig. 4c). These results indicate that the HD-ZIP III transcription factors are required for cellular quiescence in xylem cells. WUSCHEL-RELATED HOMEOBOX 4 (WOX4)14 and PHLOEM INTERCALATED WITH XYLEM (PXY, also known as TDIF RECEPTOR (TDR))14,15 are key cambial regulators. They show expression maxima on the xylem-side of the cambium and weak expression in the adjacent dividing cells (marked with ANT expression), which strongly resembles the expression pattern of AtHB8 (Fig. 2a). Induction of miR165a led to downregulation of WOX4, PXY and ANT (Fig. 2d, Extended Data Fig. 4d). These data imply that xylem-identity regulators—the HD-ZIP III transcription factors—define a stem-cell organizer that confers cambium stem-cell identity to the adjacent cells, in a non-cell-autonomous manner.

Fig. 2: HD-ZIP IIIs determine the organizer of the vascular cambium.
figure 2

a, Confocal microscopy of cross-sections of ANT double markers in 14-day-old plants. White arrows indicate pairs of nuclei in recently dividing cells. Insets, merged reporter channels. b, Bright-field microscopy of root after miR165a induction (Ind) for 6 days (in 5-day-old plants). Red dots, secondary xylem vessels (SXV); quantification presented on the right. Box-and-whisker plot shows median and interquartile range; whiskers show range, dots show outliers. Two-tailed Wilcoxon test. n, independent roots. c, Confocal microscopy of roots after 5-ethynyl-2′-deoxyuridine (EdU) staining to detect S-phase nuclei reveal a centripetal shift of the cell division domain after a 2-day miR165a induction (in 8-day-old plants). Asterisks mark cambium. Histograms showing the relative distances of EdU-positive nuclei in the radial direction from the root midpoint (xylem axis: x = 0; root surface, x = 1), as depicted in mock. nn, total number of nuclei. RD, relative distance from root midpoint. d, Confocal microscopy of fluorescent markers after 2-day miR165a induction (in 8-day-old plants). Dashed lines, borders of the secondary tissues. The numbers in the bottom left corner of subpanels represent the frequency of the observed phenotype. a, c, d, Cell wall staining in grey. p, phloem; x, xylem. All experiments were repeated at least three times. Scale bars, 50 μm.

Source data

Full size image

The loss of the HD-ZIP III transcription factors also caused decreased production of differentiated phloem cells (Extended Data Fig. 4e) and lower expression of the early phloem marker PHLOEM-EARLY-DOF 1 (PEAR1)16,17 (Fig. 2a, d, Extended Data Figs 1a, 4d), which suggests that the organizer cells maintain a common stem cell for secondary xylem and phloem. To test this hypothesis, we traced cell lineages during secondary growth. GUS clones were generated in 15-day-old or 30-day-old Arabidopsis roots, which contain 2–4 (at 15 days old) or 7–9 (at 30 days old) layers of vascular cambial cells that could histologically be identified (Extended Data Fig. 1b). In each radial cell file, three different types of GUS sectors were observed (Fig. 1c, Extended Data Fig. 5a–c). GUS clones that originated from the xylem (green) or phloem (yellow) side of the cambium differentiated into the corresponding cell types or divided before differentiation. By contrast, cells located in the centre of the cambium (red) produced broad GUS sectors that spanned into the differentiated xylem and phloem (Fig. 1c, d, Extended Data Fig. 5c, d). Similarly, induction of clones in the ANT expression domain—marking a narrow layer of dividing cells—led to production of GUS sectors that spanned in both directions (Fig. 2a, Extended Data Fig. 5e–h). These data demonstrate that the vascular cambium comprises a single, bifacial stem cell in each radial cell file, and that stem cells are marked with ANT expression. These stem cells produce both xylem and phloem—thus confirming the 145-year-old theory of the uniseriate vascular cambium3,4, at least in the Arabidopsis root. This theory states that a single layer of ‘initials’ (stem cells) produces both xylem and phloem; and was later elaborated upon by the observation that the position of the initials varied between neighbouring radial files1,4. Our experimental data support these later modifications; the variable location of the stem cells can be observed as a considerable radial distribution of the GUS sectors that originate from stem cells (Fig. 1d, Extended Data Fig. 5d).

To test whether the HD-ZIP III transcription factors are sufficient to induce the formation of an ectopic organizer within the cambium, we inducibly overexpressed a version of AtHB8 (termed AtHB8d) that was resistant to all miR165 and miR166 miRNAs, in the stem-cell domain under the ANT promoter12. Induction of pANT:XVE>>AtHB8d-YFP rapidly inhibited cell division in the stem cells, thus confirming the ability of the HD-ZIP III transcription factors to promote cellular quiescence (Extended Data Fig. 6a–c). This local quiescence was associated with a centrifugal shift—that is, towards phloem—in the domain of high cell division and organizer marker expression (Extended Data Fig. 6b, d), followed by the formation of ectopic xylem vessels in the cambium (Extended Data Fig. 6e). These data demonstrate that, of the HD-ZIP III family genes, at least AtHB8 is sufficient to induce organizer function within the vascular cambium.

Auxin promotes expression of HD-ZIP III transcription factors in vascular tissues during primary development, and their expression often correlates with auxin signalling maxima18,19. Because the auxin maximum reaches a peak on the xylem-side of tree vascular cambia20,21, we next studied the pattern of the response to auxin signalling in the Arabidopsis root cambium. Auxin response markers showed a local maximum in the xylem domain, and tapered off towards the cambium stem cells, which resembles the expression pattern of AtHB8, WOX4 and PXY (Figs 2a, 3a, Extended Data Fig. 7a, b). Auxin signalling is mediated by AUXIN RESPONSE FACTOR (ARF) transcription factors22. We focused our expression analysis on class A ARFs, because their activity is preferentially auxin-regulated22; MONOPTEROS (MP) (also known as ARF5), a class A ARF, has previously been shown to regulate normal vascular formation23 and cambium-cell division by controlling the expression of WOX424. Of the five class A ARFs, three (MP, ARF7 and ARF19) showed strong and consistent expression in the vascular cambium (Extended Data Fig. 7c, d). Because the arf7 arf19 double mutant did not show clear patterning defects (Extended Data Fig. 8a), and strong mp mutants failed to initiate roots25, we developed an inducible artificial miRNA (amiRNA) line that targets MP—which we name amiMP—in the arf7 arf19 background. Ubiquitous induction of amiMP led to a reduction in transcript levels of MP, downregulation of WOX4 and a subset of genes for HD-ZIP III transcription factors, and a reduction in secondary vessel formation (Fig. 3b, c, Extended Data Fig. 8b, c), similar to the miR165a-overexpression phenotype (Fig. 2b). Similarly, ubiquitous induction of an inducible, stable ARF-signalling repressor axr3-1 (ref. 26) led to a strong downregulation of the organizer, stem-cell and phloem-identity genes, as well as the genes of HD-ZIP III family, followed by termination of vascular cambium activity (Fig. 3d, Extended Data Fig. 8d–f). These data show that auxin signalling through transcription factors encoded by MP, ARF7 and ARF19—and perhaps other ARF genes—promotes the expression of HD-ZIP III genes, PXY and WOX4 in the stem-cell organizer, and that this signalling module is required to maintain vascular cambium identity. Because manipulations of levels of HD-ZIP III transcription factors lead to changes in auxin response (Extended data Fig. 7e–g), our data further suggest that ARF and HD-ZIP III transcription factors, and auxin, are under positive feedback regulation in the vascular cambium.

Fig. 3: Auxin signalling is required for xylem differentiation and organizer identity.
figure 3

a, Confocal microscopy of DR5:nYFP and cAtHB8-YFP (heat map) root cross-sections. Nuclear fluorescence signal (circles indicate examples of regions of interest) was quantified in three cell layers from the most recent cell division (arrowheads in micrographs and on x axis) towards xylem (positions −1, −2 and −3) and phloem (positions 1, 2 and 3). nn = total number of nuclei. Cell wall staining in grey. p, phloem; x, xylem. b, Quantitative PCR with reverse transcription (qRT–PCR) analysis of marker-gene expression in amiMP T1 transformants in arf7 arf19 background with the lowest level of MP transcripts. Each point represents an individual amiMP T1 transformant (n = 10) or individual arf7 arf19 plant (n = 5) of the same age. The P values of a one-sided t-test comparing amiMP to arf7 arf19 controls are provided under each gene name, and significant values are in bold. c, Bright-field microscopy of pUBQ10:XVE >> amiMP transformants in the arf7 arf19 background induced for 7 days (in 5-day-old plants). Yellow dots, secondary xylem vessels. Experiment was repeated three times with similar results. d, qRT–PCR analysis of marker-gene expression in inducible axr3-1 after 24h induction (in 8-day old plants). Data are from three experiments (n = 16 roots per treatment). In a and d, box-and-whisker plots show median and interquartile range; whiskers show range. One-tailed t-test. In a, dots represents independent data points. Scale bars, 10 μm (a), 50 μm (c).

Source data

Full size image

Finally, we tested whether the components that determine the stem-cell organizer are sufficient to induce vascular cambium formation when ectopically expressed. We generated activation clones in the phloem pole pericycle and surrounding cell files using an inducible S17 promoter12,16 (Extended Data Fig. 9a, b). Even though AtHB8d was sufficient to re-specify the organizer when overexpressed in stem cells of the vascular cambium, it was unable to induce xylem vessel formation or expression of cambial markers in the S17 domain (Extended Data Figs. 6, 9c, d). However, a proportion of clones expressing MPΔ (a dominant active version of MP27) formed ectopic vessel elements and had periclinal cell divisions in the adjacent cells (Fig. 4a, Extended Data Fig. 9e, f). Time-course experiments revealed that MPΔ cell-autonomously induced WOX4, PXY and AtHB8 expression, followed by differentiation of the clones into vessel elements (Fig. 4a, b). During differentiation into vessels, MPΔ clones induced ectopic expression of WOX4, PXY and AtHB8 in the adjacent cells, which indicates that MP can also promote organizer-cell identity in a cell-non-autonomous manner (Fig. 4b). Additionally, the early phloem marker PEAR1 appeared ectopically at a distance of a few cells from the clone, which suggests that the presence of an MPΔ clone is sufficient to promote the formation of a vascular cambium (Fig. 4c). MPΔ induces expression of the genes that define the stem-cell organizer before xylem vessel formation; this suggests that a cellular identity induced by high levels of auxin signalling, rather than the presence of xylem vessels, defines the organizer. Supporting this, clonal activation of VASCULAR-RELATED NAC-DOMAIN 6 (VND6)—which directly induces terminal differentiation of xylem vessels28—failed to promote organizer-gene expression within the clone (Extended Data Figs. 1a, 9g). In the longer term, a proportion of VND6-induced vessels were surrounded by cells that express a subset of cambial markers; however, these cells rarely divided (Extended Data Fig. 9e, f). This indicates that the chemical or physical properties of terminally differentiated xylem vessels are not sufficient to induce organizer formation. Together, these data demonstrate that a locally high level of auxin signalling within root vascular tissue is sufficient to establish an organizer cell that promotes stem-cell-like divisions and identity in the adjacent cells (Fig. 4d). Our data reveal the dynamic nature of the organizer; differentiation of the organizer into a xylem vessel leads to formation of a new organizer in the adjacent cambial stem cell, thus ensuring the maintenance of the vascular cambium (Extended Data Fig. 10).

Fig. 4: MP is sufficient for the formation of an ectopic organizer.
figure 4

a, Bright-field micrograph of an MPΔ-GUS clone with periclinal cell divisions induced cell-non-autonomously. Red dots, GUS-negative cells dividing periclinally around the MPΔ-GUS clone. b, Confocal analysis of marker expression within MPΔ-YFP clones (top) and around MPΔ-YFP-induced ectopic vessels (bottom). Cell wall staining in grey. Insets show YFP channel, except for insets of PXY and AtHB8 in bottom panels, which show RFP channel. c, PEAR1 is induced a few cell layers away from MPΔ-induced ectopic vessels (asterisk). An arrow indicates a gap between ectopic vessel and PEAR1 expression. d, A model illustrating how a local auxin maximum and downstream components define an organizer that positions the stem cell in the adjacent cell. Curved arrows represent bidirectional cell proliferation from the bifacial stem cell. The numbers in the left corner of subpanels represent the frequency of the observed phenotype. HD-ZIP IIIs, HD-ZIP III genes; IAA, indole-3-acetic acid. Scale bars, 10 μm (c), 50 μm (a, b).

Full size image

We uncovered a mechanistic framework that defines the stem-cell organizer of the vascular cambium (Fig. 4d). Similarly to the cells in the stem-cell organizer (quiescent centre) of the root apical meristem29, the cells in the cambium organizer are quiescent and are defined by an auxin response maximum. However, unlike the root quiescent centre, the stem-cell organizer of the vascular cambium is dynamic: as the root grows thicker, organizer cells differentiate and stem cells acquire organizer identity (Extended Data Fig. 10a). Cells of the organizing centre of the shoot apical meristem also are replaced as they differentiate, although in this meristem the location of the organizer is stable. All three organizers express a member of the WOX gene family: WUSCHEL in the shoot apical meristem, WOX5 in the root meristem2 and WOX4 in the vascular cambium, which underscores the related identity of cells in the three locations. Future challenges are to identify the cell-non-autonomous stem-cell factors that originate from the organizer cells (Fig. 4d), and to link other known cambial regulators30 to the signalling module that defines the organizer (Extended Data Fig. 10).

Methods

No statistical methods were used to predetermine sample size. The experiments were not randomized and investigators were not blinded to allocation during experiments and outcome assessment.

Gene accession numbers

Gene accession numbers are as follows: CYCB1;1, AT4G37490; VND6, AT5G62380; WOX4, AT1G46480; PXY, AT5G61480; PEAR1 (also known as EPM, S29 or Dof2.4), AT2G37590; ANT, AT4G37750; AtHB8, AT4G32880; CNA (also known as AtHB15), AT1G52150; PHB, AT2G34710; PHV, AT1G30490; REV, AT5G60690; MIR165A, AT1G01183; MP, AT1G19850; ARF6, AT1G30330; ARF7, AT5G20730; ARF8, AT5G37020; ARF19, AT1G19220; APL, AT1G79430; and S17, AT2G22850.

Plant material and cloning

Most of the entry vectors were generated by PCR amplification of the desired DNA sequence with the primers indicated in Supplementary Table 1, and recombination of the purified PCR product into MultiSite Gateway compatible pDONR entry vectors (Supplementary Table 2). The PCR products that were obtained were amplified in two steps (marked with asterisks in Supplementary Table 2), first with gene-specific primers with partial attB overhangs and then with adaptor primers to generate PCR products with full length attB recombination sequences. The dominant version of cAtHB8 (AtHB8d) has a mutation (CCG to CTG) in the last codon of the miR165a-recognition sequence, similar to the dominant REVOLUTA mutant rev-10d31. The entry vector p221z-c AtHB8d(−)stop was generated by site-directed mutagenesis using p221z-c AtHB8(−)stop as a template and primers AtHB8dF1 and AtHB8dR1, using the QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies). The entry vector p221a-CYCB1;1-Cre was made by PCR amplification of Cre recombinase32 from p221z-CRE1 using the CreMfeF and CreMfeR primers, and the resulting PCR product was ligated into the unique MfeI restriction site of p221a-CycB1-dbox26. MicroRNA against MONOPTEROS (amiMP) was obtained by PCR-mutagenizing the MIR167a precursor sequences (At3g22886)33 using the following primers: aMIR-MP_1, aMIR-MP_2, aMIR-MP_3 and aMIR-MP_4 (sequences in Supplementary Table 3). amiMP was then cloned into pENTR/D_TOPO to create pENT-AtMIR167a-MP. To construct gANT-GFP, genomic fragment of ANT was amplified by using primers ANT(−)5172 and ANT(+)2885R, and cloned into pAN19 (a modified pUC18). GFP was inserted into the C terminus of ANT coding sequence by using inverse PCR. The resultant gANT-GFP fragment was re-cloned into pBIN30 (a modified pBIN19) binary vector. pSm43GW was generated from pHm43GW in two steps. First, hygromycin resistance was replaced with a -SphI-YFP-Kpn1- PCR fragment using the YFPfrg_F1 and YFPfrg_R2 primers and the SphI and KpnI restriction sites to obtain pYm43GW. Next, the YFP fragment was replaced with the sulfadiazine-resistance gene amplified from GABI_874H08 line gDNA with primers sulf-KpnIF and sulf-SnaBIR, and the resulting PCR product was cloned into pYm43GW using the KpnI and StuI restriction sites.

MultiSite Gateway technology (Invitrogen) was used to combine promoter (1st box), gene of interest (2nd box) and tag or terminator (3rd box) Gateway entry clones with Gateway-compatible binary destination vectors12,34 in a MultiSite LR Clonase reaction. All expression vectors generated in this study are listed in Supplementary Table 4.

All constructs were dipped into the Col-0 ecotype, and single insertion lines were screened based on Mendelian segregation of a selection marker. Several lines for each newly made construct were used for phenotype observation. Plant material produced in this study as well as published lines provided by others are indicated in Supplementary Table 5. The following transgenic and mutant lines have been described elsewhere: cna-2 phb-13 phv-11, athb8-11 cna-2 phb-13 phv-118,35, arf7-1 arf19-136, DR5rev::YFPnuc37, DR5rev:erGFP38, DR5v2:n3GFP39, DR5rev:GUS40, pWOX4:erYFP41, pMP:MP-GUS42, pARF7:GUS and pARF19:GUS43, gARF8-GUS and gARF6-GUS33, pG1090:XVE >> axr3-1-RFP26, pAPL:GUS44, gVND6-GUS45 and pAtHB8:GUS46.

Plant growth and chemical treatments

For almost all experiments, seeds were surface-sterilized in 70% ethanol, then in 20% chlorine, washed twice in H2O and plated on half-strength GM medium (1/2 GM) containing 0.5 × MS salt mixture with vitamins (Duchefa), 1% sucrose, 0,5g/l MES pH 5.8 and 0.8% agar and vernalized at 4 °C for 2 days. The age of the seedlings was measured from the moment they were placed in a vertical position in the growth cabinet, in which they were grown at 22 °C under long-day conditions. For experiments with soil-grown plants, seeds were directly placed on soil and grown under the same conditions.

17-β-ooestradiol (Sigma), a synthetic derivative of oestrogen, and 6-benzylaminopurine (BAP; Sigma)—a cytokinin—were prepared as 20 mM stock solutions in DMSO and stored at −20 °C. EdU, a thymidine analogue (Thermo Fisher), was prepared as a 10 mM stock solution in DMSO and stored at 4 °C. XVE-based gene induction (‘Ind’ in figure panels) was carried out by transferring plants to plates supplemented with 5 μM 17-β-ooestradiol or an equal volume of DMSO (‘mock’ in figure panels) unless otherwise indicated. The working concentration of BAP was 0.1 μM. For EdU treatment, plants were transferred into 1/2 GM liquid medium supplemented with 10 μM EdU and, eventually, with inducer or mock for 16 h. Unless stated otherwise, the age of the seedlings at the start of the treatment is indicated in parentheses in the figure legends.

GUS staining, microtome sectioning and histology

GUS detection and histology analysis were performed as previously described47, with slight modifications. First, samples were vacuum-infiltrated with the GUS staining solution for 1 h and then incubated at 37 °C until the desired level of GUS staining was obtained (from 2 h to overnight). Samples were cut on a microtome in 10-μm sections from 5 mm below the root–hypocotyl junction. For histological analysis of clones, the upper 1–1.5-cm root segment was cut every 200 μm to avoid repeated analysis of the same cells. Finally, root sections were stained for 5 s in a 0.05% (w/v) ruthenium red solution (Fluka Biochemika) or a 0.05% (w/v) toluidine blue solution in deionized water. To stain phloem companion cells, a 0.5% (w/v) safranin solution in 50% ethanol and a 1% (w/v) Alcian blue aqueous solution were used. All histological stainings used in each figure are indicated in Supplementary Table 6.

Fluorescent marker analysis, vibratome sectioning and EdU detection

To analyse fluorescent reporters from a lateral view, roots were fixed in 4% paraformaldehyde (PFA, Sigma) in 1× PBS pH 7.2 overnight, washed twice in 1× PBS and cleared in ClearSee solution48 supplemented with 1 μg/ml calcofluor white (Sigma) for at least three days.

Vibratome sectioning was used for fluorescent marker analysis and EdU detection. Roots were fixed in 4% PFA in 1× PBS pH 7.2 for 1 h under vacuum, washed in ice-cold 1× PBS and embedded in 4% agarose in 1× PBS. Samples were cut on a vibratome in 200-μm sections from 5 mm below the root–hypocotyl junction. Agarose slices that were obtained that contained root cross-sections were washed twice in 1× PBS, cell walls were stained with 1 μg/ml calcofluor white (Sigma) in 1× PBS for between 3 h and overnight and finally cleared with ClearSee solution. To visualize EdU-positive nuclei with the Click-iT EdU Alexa Fluor 488 Imaging Kit (Thermo Fisher), fixed cross-sections were washed twice in 1× PBS and then incubated in a modified EdU detection cocktail49 at room temperature for 1 h, washed twice in 1× PBS and stained with 1 μg/ml calcofluor or DAPI solution in 1× PBS.

Microscopy and image processing

Histological cross-sections were imaged using a Leica 2500 microscope (20× and 40× objectives). Confocal analysis was carried out with Leica SP5 and SP8 microscopes. Confocal images were acquired with the Leica LAS AF Software using ClearSee, 1× PBS or water as an imaging medium. All combinatorial fluorescence analyses were run in sequential scan mode. Confocal settings may have varied between experiments but always remained the same for an experimental sample and the corresponding control. However, to better visualize calcofluor-stained cell walls, the signal was sometimes unevenly adjusted between the sample and the control. EdU images represent maximal projections of five consecutive optical sections captured each 5 μm in the z-direction.

Microsoft PowerPoint with image compression mode off was used to compile all the figures and to modify brightness and contrast, which was always identically applied to sample and control pictures. Images in Fig. 1a were cropped and rotated using Corel Photo Paint X6, which occasionally resulted in empty corners; for aesthetic reasons, these were filled with a triangle of a similar colour as the image background.

Marker expression quantification, EdU and xylem vessel histograms

Changes in marker expression were analysed from the lateral view; optical sections were captured in the middle of the root, which was determined by the root edges being in focus in the transmitted light channel. The data represent the relative width of the fluorescent domain compared with the root diameter. DR5:nYFP, DR5v2:nGFP and cAtHB8-YFP fluorescence were captured as one stack of 16-bit confocal images from 15 cross-sections. The signal intensities of fluorescent nuclei were measured in ImageJ using multi-point tool with mark width set to 15 pixels. Relative intensities were determined by comparing the signal with the brightest value in the whole dataset. For EdU and vessel distribution histograms, relative distances of EdU nuclei and vessels were determined by comparing their distance from the centre of the xylem axis with the radius of the root. If not indicated otherwise, 25 cross-sections were analysed. Histograms represent frequencies of EdU nuclei or vessels along the radial axes divided into 10 bins in percentage. Distance measurements were done using Leica Lite or ImageJ.

amiMP analysis

Owing to persistent silencing of pUBQ10:XVE >> amiMP (arf7-1 arf19-1) in the T2 and T3 generations, individual plants were analysed in the T1 generation. Transformants in the arf7 arf19 or wild-type backgrounds were grown on plates with 20 μg/ml phosphinothricin (basta, Duchefa) for 5 days without noticeable phenotypes caused by the selection, and then induced together with the controls for 7 days. Segments of the upper 2 cm of the root were collected for histological analysis, and the remaining (lower) part was used for RNA isolation. Because arf7-1 arf19-1 does not produce lateral roots, the lower segment contained RNA exclusively from the primary root. Cell divisions were counted as the number of cells in one radial cell file between the primary phloem pole and the xylem axis.

qRT–PCR

RNA was extracted from the top 2 cm of the main root with lateral roots removed using the GeneJET Plant RNA purification kit (Thermo Fisher) and treated with DNase (Thermo Fisher). cDNA was synthetized from 100 ng (AtHB8d- and miR165a-overexpressing plants), 500 ng (axr3-1-overexpressing plants and miR165a-overexpressing plants for PXY) or 50 ng (amiMP plants) of RNA using MAXIMA H Minus reverse transcriptase (Thermo Fisher) with Oligo(dT) primers (Thermo Fisher). The PCR reaction was done on a Bio-Rad CFX384 cycler using EvaGreen qPCR mix (Solis Biodyne) and the following program: 10 min at 95 °C, 50 cycles (10 s at 95 °C, 10 s at 60 °C, 30 s in 72 °C). All primers used in qRT–PCR are listed in Supplementary Table 6: primers for ACT2, UBQ10 and ANT50; PHB, PHV, AtHB8 and CNA8; PXY51; and TIP4152 have previously been described. Results were normalized, following previously published methods53,54, to the reference genes ACT2, UBQ10 and TIP41. Three biological replicates were used for each line and time point, as well as three technical replicates, except in the analysis of amiMP lines (see above).

RNA in situ hybridization

Primers used for RNA probe synthesis have previously been published 8. Sample preparation, sectioning, probe synthesis using T7 RNA polymerase (Thermo Fisher) and hybridization were performed as described55 with slight modifications. The uppermost 1-cm segment of the main root was cut from 21-day-old soil-grown plants, and the tissues were vacuum-infiltrated in FAA (50% ethanol; 5% acetic acid; 3.7% formaldehyde) for 30 min and then fixed overnight at 4 °C. For paraffin-embedding, a Leica ASP200 vacuum tissue processor was used, following the protocol indicated in Supplementary Table 7.

Clone induction

To induce clones before the onset of secondary growth, plates with 4-day- old seedlings grown on 1/2 GM plates with or without BAP were heat-shocked at 37 °C for 14 or 17 min, immediately cooled at 4 °C for 15 min and further grown on new 1/2 GM plates under normal growth conditions until GUS staining and histological analysis. For lineage tracing in active cambium, plants were grown for 15 days before heat-shock. Clonal activation of pUAS:MPΔ-GUS (Fig. 4a) was carried out in the same way and the clones were analysed 6 days later. For lineage tracing in one-month-old plants, whole 5-cm pots with individual plants were transferred into a 37-°C water bath for 15 min followed by a cooling step under cold tap water for 1 min. Clone activation in the phloem pole pericycle and surrounding cells (S17 domain) was performed by transferring 5-day-old seedlings into liquid 1/2 GM medium with 17-β-ooestradiol for 8 or 16 min followed by four 15-min washes in medium without an inducer. After that, plants were placed back on agar plates and grown under normal conditions. ANT:XVE >> cre transformants in the 35S:lox-GUS+/+ background were selected on plates with hygromycin and claforan. After four days, hygromycin-resistant plants were transferred on plates with claforan only, and grown further without a noticeable phenotype caused by antibiotics. Clones in the ANT domain were induced in a similar way as S17 clones, with slight modifications: 17-day-old plants were incubated with an inducer for 5 or 20 min and, after washing, they were placed on agar plates with claforan. To increase the proportion of clones originating from single cells, roots were pre-screened and only those with a low number of clones were analysed further.

Fate map and lineage-tracing quantification

Because the position of major cell types in the primary root vasculature is invariant, the fate map was generated by counting the frequencies of different GUS sectors and grouping them into three categories: (i) the most frequent basic sectors, (ii) large sectors, which partially share clonal boundaries with two or more basic sector types at once and (iii) half sectors, which consist of a subdomain of one of the basic sector types. Only basic sectors are presented in the fate map (Fig. 1b). BAP-mediated promotion of procambium formation in the place of protoxylem is not a fully penetrant phenotype56, and therefore the presence or absence of protoxylem ((+) protoxylem or (−) protoxylem, respectively, in Extended Data Fig. 2e, f) was observed before analysing the sectors.

Quantification of GUS sectors in active cambium was carried out by counting the cellular distance of sector boundaries (xylem side (y < 0); phloem side (y > 0)) from periclinally dividing cambial cells (y = 0) in the radial direction. Dividing cambial cells were identified histologically on the basis of the presence of a thin cell wall. In rare cases in which two dividing cambial cells were in a given cell file, the distance was counted from the cell wall between those cells. Only atypically large GUS sectors, sectors with unclear borders or sectors that were located far from the vascular cambium (Extended Data Fig. 5a, b, f) were excluded from quantification. Each bar in graphs (Fig. 1d, Extended Data Fig. 5d, h) represents the size and position of an individual sector; bars with extensions denote sectors beginning or ending with a xylem vessel.

Quantification of MPΔ and VND6 clones

The cell-autonomous effects of MPΔ and VND6 clones on cambial marker-gene induction were measured as the number of independent YFP-positive nuclei that showed marker expression within the clones two days after induction. The cell-non-autonomous effects were quantified as the number of ectopic vessels that showed marker expression in the surrounding cells seven days after induction. At least ten roots per genotype were quantified. The cell-non-autonomous effect of MP on periclinal cell division was analysed by counting the frequency of periclinal cell divisions induced 16 days after germination, in the GUS-negative cells surrounding the MPΔ-GUS clones or ectopic vessel. In these experiments (Fig. 4a, Extended Data Fig. 9e), periclinal cell divisions refer to cell division planes parallel to the surface of the neighbouring MPΔ- or VND6-induced vessels or clones. Only clearly ectopic vessels or clones (that is, outside the vascular cambium) were analysed further.

General methodology and statistical analysis

Unless otherwise stated, the number of individual plants, root sections or clones analysed is indicated as the n value in figure panels or legends. The fraction given in the corner of some figure panels denotes the frequency of the presented phenotype. Statistical analyses were all carried out in R version 3.5.0 (http://www.r-project.org/). For the mean comparison analysis, first the Shapiro–Wilk test for normality was performed and—upon failing to reject or rejecting this hypothesis—the appropriate one- or two-sided Student t-test, or the nonparametric Wilcoxon rank sum test, was performed, respectively. Regression analyses were performed following the affirmation of the normality assumption of the residuals. In general, we considered 0.05 as the significance threshold for P values. The Kullback–Leibler divergent distance was used to quantify the level of dissimilarity between histograms, and finally, the mean between the two directional values is reported. For all box plots, the centre line represents the median and the lower and upper limits of boxes indicates the 25th and 75th percentiles, respectively. Whiskers show the maximum or minimum values, and empty and filled circles represent the individual samples and outliers, respectively.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.