Spatial specificity of auxin responses coordinates wood formation

Spatial organization of signalling events of the phytohormone auxin is fundamental for maintaining a dynamic transition from plant stem cells to differentiated descendants. The cambium, the stem cell niche mediating wood formation, fundamentally depends on auxin signalling but its exact role and spatial organization is obscure. Here we show that, while auxin signalling levels increase in differentiating cambium descendants, a moderate level of signalling in cambial stem cells is essential for cambium activity. We identify the auxin-dependent transcription factor ARF5/MONOPTEROS to cell-autonomously restrict the number of stem cells by directly attenuating the activity of the stem cell-promoting WOX4 gene. In contrast, ARF3 and ARF4 function as cambium activators in a redundant fashion from outside of WOX4-expressing cells. Our results reveal an influence of auxin signalling on distinct cambium features by specific signalling components and allow the conceptual integration of plant stem cell systems with distinct anatomies.

I n multicellular organisms, communication between cells is essential for coordinated growth and determination of cell fate. In plants in particular, the flexible regulation of cellular properties by hormone signalling is important throughout the whole life cycle. This is because plants are sessile and continuously adapt their growth and development to their local environment. The basis of this plastic growth mode are local stem cell niches at the tips and along plant growth axes, called meristems 1 . The tip-localized shoot and root apical meristems (SAM and RAM, respectively) are essential for primary, or longitudinal, growth of shoots and roots, respectively. In turn, the vascular cambium is the predominant lateral meristem forming a cylinder of indeterminate stem cells at the periphery of growth axes and mediating radial growth by producing the vascular tissues phloem and xylem in a bidirectional manner 2,3 . This production is the basis of wood formation and is thus essential for the accumulation of a large proportion of terrestrial biomass.
The plant hormone auxin plays pivotal roles in local patterning and maintenance of stem cell niches in the SAM and RAM. In the SAM, auxin signalling is low in stem cells and increases during recruitment of cells for organ formation [4][5][6] . Cell wall modulation and the formation of vascular strands are two aspects promoted by auxin in this context 7,8 . In contrast, a maximum of auxin signalling is present in the quiescent centre, and the surrounding stem cells in the RAM and cell differentiation is, at least partly, driven by a decrease in signalling levels 9,10 . Therefore, the functions of auxin in both meristems are different and adapted to distinct niche requirements.
For the cambium, the role of differential auxin signalling along the radial sequence of tissues is still obscure. In Arabidopsis stems, apex-derived auxin is transported basipetally and distributed laterally across the cambial zone by the auxin exporters PIN-FORMED1 (PIN1), PIN3, PIN4 and PIN7 11,12 . Indeed, direct auxin measurements in Populus and Pinus trees showed that the concentration of the major endogenous auxin indole-3-acetic acid (IAA) peaks in the centre of the cambial zone and gradually declines towards differentiating xylem and phloem cells [13][14][15] . This observation prompted the idea that, in analogy to the situation in the RAM, radial auxin concentration gradients contribute to the transition of cambium stem cells to secondary vascular tissues 16,17 . Consistently, ubiquitous repression of auxin responses by expressing a stabilized version of the auxin response inhibitor PttIAA3 reduces the number of cell divisions in the cambium region of hybrid aspen trees 18 . In addition, however, the zone of anticlinal cell divisions characteristic for cambial stem cells is enlarged in PttIAA3-overexpressing trees. This suggests that auxin signalling not only promotes cambium proliferation but also spatially restricts stem cell characteristics within the cambium area 12,18,19 . Indeed, especially xylem formation is associated with a local increase of auxin signalling in other contexts 10,[20][21][22] that supports a role of auxin in the recruitment of cells for differentiation similarly as in the SAM. Therefore, it is currently unclear whether auxin signalling is predominantly associated with stem cell-like features or cell differentiation in the context of radial plant growth or how a positive effect on cambium proliferation and on the differentiation of vascular tissues is coordinated.
As a central cambium regulator, the WUSCHEL-RELATED HOMEOBOX4 (WOX4) transcription factor imparts auxin responsiveness to the cambium 23 . Equivalent to the role of WUSCHEL (WUS) and WOX5 in the SAM and RAM 24,25 , respectively, WOX4 activity maintains stem cell fate 23,26 . In turn, WOX4 transcription is stimulated by the leucine-rich repeat receptor-like kinase (LRR-RLK) PHLOEM INTERCALATED WITH XYLEM (PXY). Importantly, the expression domains of the WOX4 and PXY genes presumably overlap and are considered to mark cambium stem cells 23,[26][27][28] . However, a bipartite organization of the cambium zone was shown recently with PXY being expressed only in the proximal (xylem-facing) part 29 . Whether this organization reflects the existence of two distinct stem cell pools feeding xylem and phloem production, respectively, has still to be determined.
Here we identify functional domains of auxin signalling in the Arabidopsis cambium by local short-term modulation of auxin biosynthesis and signalling. We reveal that, while cambial stem cells do not appear to be a site of elevated auxin signalling, auxin signalling in these cells is required for cambium activity. By analysing transcriptional reporters and mutants of vasculatureassociated AUXIN RESPONSE FACTORs (ARFs), we identify ARF3, ARF4 and ARF5 as cambium regulators with different tissue-specificities as well as distinct roles in cambium regulation. Remarkably, whereas ARF3 and ARF4 act redundantly as more general cambium promoters, ARF5 acts specifically in cambium stem cells. In-depth analysis of the auxin-and ARF5-dependent transcriptome in those cells, together with protein-DNA binding assays and genetic analyses, demonstrates that the ARF5dependent attenuation of WOX4 is an essential aspect of auxin signalling during cambium regulation.

Results
Auxin responses in stem cells stimulate cambium activity. In Arabidopsis stems, the activity of the common auxin response marker pDR5rev:GFP 30 was detected in vascular tissues (phloem and xylem) and cortical cells prior and during cambium initiation (Fig. 1a, Supplementary Fig. 1a) 23 . However, there was no overlap with pWOX4:YFP 23 or pPXY:CFP 28 reporter activities, the two canonical markers for cambium stem cells (Fig. 1a-c, Supplementary Fig. 1a-c) 23 . This suggested that auxin signalling in stem cells occurs at low levels or is even absent. To decide between both possibilities, we generated a plant line expressing an endoplasmatic reticulum (ER)-targeted Yellow Fluorescent Protein (YFP) under the control of the high-affinity DR5revV2 promoter, which recapitulated the pattern of DR5revV2 activity previously reported in roots ( Supplementary Fig. 1d-f) 31 . In the second internode of elongated shoots, pDR5rev:GFP and pDR5revV2:YFP activities were congruent but pDR5revV2:YFP activity also included the whole cortex as well as cambium cells marked by pPXY:CFP activity (Supplementary Fig. 1g-i). Immediately above the uppermost rosette leaf (denoted as stem base throughout the text), stem anatomy shows a secondary configuration, which is characterized by a continuous domain of cambium activity 23 . At this position, the expression domain of pDR5revV2:YFP was again broader than the domain of pDR5rev:GFP activity substantially overlapping with pPXY:CFP activity (Fig. 1d-f). Based on these observations, we concluded that the auxin signalling machinery is active in PXY-positive cambial stem cells.
To see whether auxin levels in PXY-positive cells were positively correlated with cambium activity, we used the WOX4 promoter, whose activity fully recapitulated the PXY promoter activity in the cambium (Supplementary Fig. 2a-i), for expressing a bacterial tryptophan monooxygenase (iaaM) in an inducible manner 32 . iaaM converts endogenous tryptophan to the IAA precursor indole-3-acetamide and was used before to boost endogenous IAA levels in Arabidopsis 33 . As a read out for cambium activity, we determined the amount of interfascicular cambium-derived (ICD) tissues ( Supplementary  Fig. 3a, b) 34 . Indeed, ethanol-based iaaM induction substantially stimulated the production of ICD tissues (Fig. 1g, h, m, Supplementary Fig. 3c-g), demonstrating that an increase of auxin biosynthesis in PXY-positive stem cells stimulates cambium activity.
To determine to which extent downstream components of the auxin signalling cascade are required in those cells, we blocked ARF activity by expressing a dexamethasone (Dex)-inducible variant of the stabilized AUX/IAA protein BODENLOS (Myc-GR-bdl) 35 under the control of the PXY promoter. Consistent with a role of ARF activity in cambium regulation, Dex treatments of pPXY:Myc-GR-bdl plants resulted in a strongly reduced amount of ICD tissues at the stem base (Fig. 1i, j, m) but not in an altered overall growth habit ( Supplementary Fig. 3h). Strikingly, Dex-treated pPXY:Myc-GR-bdl plants showed an even more pronounced repression of IC activity than the inhibition using the BDL promoter 35 (Fig. 1k, l, m, Supplementary Fig. 3i) whose activity was very broad including also PXY-positive cells ( Supplementary Fig. 4a-f). These observations indicated that local auxin signalling in PXY-positive stem cells stimulates cambium activity.
ARF genes are expressed in cambium-associated cells. To identify ARFs active in cambium stem cells, we mined public transcriptome datasets and found the ARF3/ETTIN, ARF4 and ARF5/MONOPTEROS genes to be co-induced with WOX4 and PXY during cambium initiation 23 . Indeed, pARF3:YFP, pARF4: YFP and pARF5:YFP promoter reporters were active in cambiumrelated cells at the stem base and the second internode (Fig. 2a, d, g, Supplementary Fig. 5a, d, g). However, while pARF3:YFP and pARF4:YFP reporters were active in rather broad domains including the phloem, the xylem and, partly, pPXY:CFP-positive cells ( Fig. 2a-f, Supplementary Fig. 5a-f), pARF5:YFP was exclusively active in cells marked by pPXY:CFP activity ( Fig. 2g-i, Supplementary Fig. 5g-i). Moreover, in second internodes, pARF3:YFP and pARF4:YFP activities were both detected in the starch sheath, the innermost cortical cell layer that is considered to serve as the origin of the IC ( Supplementary Fig. 5c, f arrows) 37,38 , while pARF5:YFP activity was restricted to vascular bundles ( Supplementary Fig. 5i). Indicating also a temporal difference between ARF3/4 and ARF5 activities, pARF3:YFP and pARF4:YFP reporters were active together with pPXY:CFP in interfascicular regions at positions approximately 5 mm above the stem base ( Supplementary Fig. 6a-f) where cortical cells start dividing to form the IC 34 . In contrast, no pARF5:YFP activity was detected in the same cortical cells (Fig. 2j-l). This observation suggested that ARF5 expression follows the expression of PXY during cambium initiation and is not active during early steps of cambium initiation. Consistently, in pxy-4 mutants where IC formation is largely absent in stems (see below 36 ) a pARF5:mCherry promoter reporter was only active in vascular bundles but not in interfascicular regions (Supplementary Fig. 6g-i). Taken together, these observations were in line with a role of ARF3 and ARF4 as The T-bars that extend from the boxes (whiskers) are expected to include 95% of the data. Significance is indicated by asterisks promoters of cambium activity and a role of ARF5 as a modulator of the established cambium.
ARF control of cambium proliferation. To find indications for these roles, we analysed cambium activity in mutants for the respective ARF genes ( Supplementary Fig. 7a-e). Consistent with a positive effect of ARF3, arf3 mutants 37,39 showed significantly reduced cambium activity (Fig. 3a, b, e, f, m). This reduction was further increased upon depletion of ARF4 activity by introducing the arf4-2 mutation 37 into the respective arf3 mutant backgrounds ( Fig. 3c, d, g, h, m) although the primary stem conformation was similar as in wild type ( Supplementary Fig. 7f-m).
Consequently, we concluded that cambium activity is positively regulated by ARF3 and ARF4, which, as in other contexts 37,38 , act in a concerted fashion. In contrast, cambium activity was enhanced in the hypomorphic arf5 mutant mp-S319 40 (Fig. 3i, j, n, Supplementary  Fig. 7a, c, n, o), suggesting that ARF5 counteracts cambium proliferation. Enhanced cambium activity was also found in mp-S319 hypocotyls ( Supplementary Fig. 7r-t) arguing against the possibility that increased cambium activity was observed in stems due to an earlier cambium initiation in interfascicular regions. To confirm a negative impact of ARF5 on cambium activity, we generated adult plants of the strong ARF5 loss-of-function mutant mp-B4149, which is usually seedling lethal 41 , and wildtype plants through tissue culture 42,43 (Supplementary Fig. 7d, e, p, q). Indeed, mp-B4149 plants showed enhanced ICD formation   comparable to mp-S319 mutants (Fig. 3k, l, n). Further confirming a negative effect of ARF5 on cambium activity, ubiquitous expression of a Dex-dependent GR-ARF5 protein fusion using the 35S promoter 44 led to significantly reduced tissue production under long-term induction ( Fig. 4a-c, j).
To test whether the identified ARFs function in PXY-positive stem cells, we first employed the PXY promoter to express GR-ARF5ΔIII/IV, a truncated variant of ARF5 lacking the domains III and IV releasing it from AUX/IAA-based repression 45 . Indeed, long-term Dex treatment of pPXY:GR-ARF5ΔIII/IV plants resulted in reduced cambium proliferation ( Fig. 4d-f, j) indicating a stem cell-specific role of ARF5. In contrast, the same treatment of a pPXY:GR-ARF3 line did not influence cambium activity ( Fig. 4g-i, j, see below) arguing against a rate-limiting role of the non-AUX/IAA-dependent 46 ARF3 protein in those cells. Collectively, we concluded that ARF3 and ARF4 on one side and ARF5 on the other side represent two subgroups of ARF transcription factors with differences in both their spatio-temporal expression and roles in cambium regulation.
ARF5 restricts the number of undifferentiated cambium cells.
To dissect the ARF5-dependent control of cambium stem cells, we took advantage of the DEX-inducibility of our pPXY:GR-ARF5ΔIII/IV and of a p35S:Myc-GR-bdl line. By determining transcript abundance of the direct ARF5 targets ATHB8 and PIN1 47,48 at different time points after Dex treatment, 3 h of treatment was identified as being optimal for observing shortterm effects on gene activity ( Supplementary Fig. 8a, b). After determining genome-wide transcript profiles at that time point, we identified a common group of 600 genes with altered transcript levels in both the pPXY:GR-ARF5ΔIII/IV and the p35S: Myc-GR-bdl line (p < 0.01 according to Fisher's Exact Test; Fig. 4k and Supplementary Data 1). The 600 genes represented various functional categories, including primary auxin response (IAAs, SAURs, GH3s), xylem and phloem formation (IAA20 and IAA30 22 Fig. 8e), indicating that we indeed revealed ARF5dependent genes in stems. Strikingly, while our expectation was that genes, which are induced by GR-ARF5ΔIII/IV induction, would be repressed by the auxin signalling repressor bdl and vice versa, we observed 144 genes (24%) that were either induced (73 genes) or repressed (71 genes) by both transgenes (Fig. 4k). This indicated that in PXY-positive cells ARF5 antagonizes the effect of overall auxin signalling on a substantial subset of target genes. Since we observed opposing effects of ARF5 and overall canonical auxin signalling on cambium activity, we suspected that genes integrating these effects are among the 144 genes behaving in an unexpected manner. Interestingly, 11 genes out of the 144 were also found in a set of genes that are differentially expressed during IC formation 36 , one of them being WOX4 that was repressed by both GR-ARF5ΔIII/IV and Myc-GR-bdl induction (Fig. 4l).
To investigate whether the observed effects of altered auxin signalling levels on gene transcription were based on local changes of gene activity in PXY-positive stem cells, we introgressed fluorescent promoter reporters into our pPXY:GR-ARF5ΔIII/IV line and a line expressing Myc-GR-bdl under the control of the PXY promoter (pPXY:Myc-GR-bdl) and investigated reporter activity 1 day after Dex induction. As predicted by our transcriptional profiling (Supplementary Data 1), a YFP reporter driven by the cambium-associated REDUCED IN LATERAL GROWTH1 (RUL1) 36 promoter (pRUL1:YFP) was more active in Dex-treated than in mock-treated pPXY:GR-ARF5ΔIII/IV lines (Fig. 5a, b, e, f) but repressed in Dex-treated p35S:Myc-GR-bdl plants (Fig. 5c, d, g, h). Instead, the pWOX4: YFP reporter displayed a reduced activity in pPXY:GR-ARF5ΔIII/ IV and pPXY:Myc-GR-bdl plants upon Dex treatment (Fig. 5i-p) confirming our expectations. Of note, a reduction was only observed in cambium-borne stem cells but not in xylem parenchyma cells where pWOX4:YFP reporter activity was found as well, demonstrating that the effect was cambium-specific ( Fig. 5i-p). In contrast to RUL1 and WOX4 promoters, PXY promoter activity visualized by the pPXY:CFP reporter did neither respond to Dex treatments of pPXY:GR-ARF5ΔIII/IV nor of pPXY:Myc-GR-bdl plants ( Fig. 5q-x). These results demonstrated that our transcriptome analyses faithfully reported on gene activity in PXY-positive stem cells in response to different members of the auxin signalling machinery. Moreover, we concluded that there are distinct auxin response signatures of individual stem cell-related genes.
Because ARF5 induction resulted in both, WOX4 repression and the induction of xylem-and phloem-related genes (Supplementary Fig. 8b), we reasoned that the repressive effect of ARF5 on cambium proliferation was due to an influence on the transition of cambial stem cells to vascular cells. To test this, we analysed the stem cell marker pWOX4:YFP in mp-S319 mutants. Indeed, the radial extension of the pWOX4:YFP domain was increased in mp-S319 plants (Fig. 6a-c), suggesting that the number of undifferentiated cambium cells was higher when ARF5 activity was reduced. Consistently, when analysing the anatomy of the cambium zone predominantly the size of the domain of undifferentiated cells was increased ( Fig. 6d-f), which was caused by an increase in cell number (Fig. 6g). This alteration resulted specifically in an increased ratio of undifferentiated cells to xylem cells (Fig. 6h). As in stems, increased cambium activity in mp-S319 hypocotyls resulted mostly in an increase in phloem production but less in an increase in xylem production ( Supplementary Fig. 7r-t). Likewise, Dex treatment of pPXY: GR-ARF5ΔIII/IV lines resulted in a decrease of the width of the cambium domain as well as the number of cambium cells. Moreover, predominantly the width of the phloem domain and the number of phloem cells was reduced (Fig. 7a, b). As in mp-S319 mutants, this mostly affected the ratio of cambium/xylem cell numbers but to a lesser extent the cambium/phloem cell numbers (Fig. 7c, d). Taken together, these observations indicated that ARF5 predominantly fulfils its function by promoting the transition of undifferentiated stem cells to differentiated xylem cells.
WOX4 mediates ARF5 activity. Owing to the fact that WOX4 was the only stem cell-associated gene being repressed by both GR-ARF5ΔIII/IV and Myc-GR-bdl induction and was identified as a mediator of auxin responses before 23 , we hypothesized that ARF5 acts on cambium activity by directly regulating WOX4.
Indeed, the expression domain of the transcriptional pWOX4:YFP reporter 23 almost completely overlapped with pARF5:mCherry at the stem base (Fig. 8a-c). Moreover, WOX4 transcript levels were increased in mp-S319 mutant stems (Fig. 8d), indicating that the endogenous ARF5 gene is required for the regulation of WOX4. Importantly, the negative effect of GR-ARF5ΔIII/IV induction on WOX4 activity was also observed in the presence of the protein biosynthesis inhibitor cycloheximide (CHX) (Fig. 8e), which was in line with a direct regulation of WOX4 by ARF5. Consistently, transient expression of ARF5ΔIII/IV in cultured cells had, similar as on other genes directly repressed by ARF5 50 , a strong effect on the activity of a pWOX4:LUC promoter reporter, while this effect was only minor when ARF3 was expressed (Fig. 8f). This suggested that, in comparison to ARF3, ARF5 substantially influenced the activity of the WOX4 promoter. Confirming this conclusion, neither cambium specific nor global induction of GR-ARF3 activity led to a significant change in WOX4 expression in wild-type or arf3;arf4 double mutants although IPT3, a putative downstream target of ARF3 51 , was induced ( Supplementary  Fig. 9a-c). Furthermore, WOX4 expression was not significantly altered in the arf3;arf4 double mutant ( Supplementary Fig. 9d), making it rather implausible that ARF3 and ARF4 act on cambium activity by regulating WOX4.
To confirm direct binding of the ARF5 protein to the WOX4 promoter, we first predicted ARF-binding sites using a conventional weight matrix approach 52 . In total, we found two topscored potential sites, located 335 bp (CAGACA) and 2175 bp (TGTCATtaCCGACA) upstream of the transcriptional start site ( Supplementary Fig. 10). Performing electrophoretic mobility shift assays (EMSAs) using DNA oligomers covering sequences from those sites and sites shown to be bound by ARF5 before 50,53 , binding was indeed observed for the first but not the second predicted site (Fig. 8g, h). These observations demonstrated a strong potential of the ARF5 protein to repress WOX4 transcription via direct promoter binding.
To analyse the relevance of the observed effect of ARF5 on WOX4 activity, we determined ICD extension in mp-S319 and wox4-1 single and double mutants. While mp-S319 showed enhanced cambium activity (Fig. 9a, b, g), activity was similar in wox4-1 single and in mp-S319;wox4-1 double mutants (Fig. 9c, d, g), suggesting that WOX4 is required for an ARF5-dependent repression of cambium activity. In comparison, depletion of ARF5 activity in mp-S319;pxy-4 double mutants lead to a mild suppression of cambium defects observed in pxy-4 single mutants (Fig. 9e, f, g) 36 , suggesting that the epistatic relationship between WOX4 and ARF5 is specific.

Discussion
Similar to apical meristems, the regulation of the vascular cambium has been tightly associated with the plant hormone auxin for several decades 15,17,54 . However, spatial organization of functional signalling domains and the role of auxin signalling in controlling different aspects of cambium activity remained unknown. Here we show that auxin signalling takes place in cambium stem cells and that this signalling is crucial for cambium activity. We also show that not only stem cell activity in general but also the balance between undifferentiated and stem cells depends on the auxin signalling machinery with ARF5 fulfilling a rather specific and WOX4-dependent role in this respect. Thus auxin-related signalling controls distinct aspects of cambium activity important for a dynamic tissue production and a complex growth process.
The concentration of IAA peaks in the centre of the cambial zone in Populus and Pinus 13-15 and transcriptional profilings indicated a spatial correlation of this peak with the expression of auxin signalling components 17,55 . However, genes responding to auxin were rather expressed in developing xylem cells arguing that sites of intense auxin signalling and of downstream responses do not necessarily overlap 18 . Consistently, our analysis of the highly sensitive auxin response marker pDR5revV2:YFP revealed a moderate auxin response in PXY-positive stem cells and a higher response in differentiated vascular tissues. Importantly, the auxin response in the PXY-positive region is overall pivotal for cambium activity since its local repression resulted in reduced tissue production similar as found in wox4 or pxy mutants defective for canonical regulators of stem cell activity 23,28,36 . This demonstrates that, in the cambium, auxin signalling promotes stem cell activity in a cell-autonomous manner. Interestingly, ARF5 and auxin signalling acts upstream of WOX5 in the context of RAM organization 24,56 but differentiation of distal root stem cells is promoted by ARF10 and ARF16 56,57 . In comparison, ARF5 restricts the stem cell domain in the SAM by repressing stem cellrelated features 5 . In the RAM and the SAM, ARF5 expression is found next to the expression domains of the central regulators WOX5 and WUS, respectively 5,24,58 , whereas it overlaps completely with the domain of WOX4 expression in the cambium. Thus a division of labour of different auxin signalling components is found in various plant meristems and recruitment of distinct factors and adaptation of expression domains seem to have happened during the evolution of those systems. ARF5 plays a major role in translating auxin accumulation into the establishment of procambium identity in embryos and leaf primordia (reviewed in ref. 54 ). However, ARF5 is also tightly associated with xylem formation via its direct targets TMO5 and ATHB8 47,59,60 . In fact, we identified those and other xylemassociated genes like ACAULIS5 (ACL5), SUPPRESSOR OF ACAULIS5 LIKE3 (SACL3) and BUSHY AND DWARF2 (BUD2) 21,60,61 to be induced upon GR-ARF5ΔIII/IV induction. Together with the observation that the domain of WOX4-positive cells is enlarged and the cambium/xylem ratio but not the cambium/ phloem ratio is increased in arf5 mutants, this suggests that ARF5 promotes the transition of cambium stem cells to xylem cells (Fig. 9h). Because we also revealed a negative effect of ARF5ΔIII/ IV induction on WOX4 activity, a responsiveness of the WOX4 promoter in transient expression systems, binding of the ARF5 protein to a distinct canonical AuxRE element in the WOX4 promoter and an epistatic genetic relationship between WOX4 and ARF5, we propose that ARF5 fulfils its function partly by directly attenuating WOX4 activity. Therefore, we propose that ARF5 acts as one hub modulating the activity of a multitude of genes in PXY-positive cells to foster the transition from stem cells to differentiated vascular cells. Error bars represent ±standard deviation. Significance is indicated by the asterisk those cells counteract cambium activity overall (Fig. 9h). A scenario in which a concerted action of various ARFs regulate different aspects of cambium activity provides an explanation for why ARF5 acts as a transcriptional repressor of WOX4 in a cambium context but WOX4 depends positively on auxin signalling overall. Competition for promoter binding between ARF5 and ARFs with a stronger positive effect on WOX4 transcription would classify ARF5 as a transcriptional repressor in this specific case. Alternatively, ARF5-dependent recruitment of other transcriptional repressors, whether directly or indirectly, to the WOX4 promoter would have the same effect. Especially our finding that ARF5 binds only to one out of the two predicted ARF-binding sites in in vitro assays gives room for a more complex occupancy of the WOX4 promoter by ARF transcription factors in vivo. Consistent with the possibility that ARF5 does not necessarily act as a transcriptional activator, it represses AUXIN RESPONSE REGULATOR15 (ARR15) and STOMAGEN in the SAM and in leaf mesophyll cells, respectively 5,50 . As observed for WOX4, both promoters are yet induced by ARF5 in transient expression systems 50 . The role of ARF5 in transcriptional regulation does therefore depend on the target promoter and the respective cellular environment, a phenomenon which has also been described for other ARF transcription factors 62,63 .
Remarkably, not only xylem-related but also phloem-related genes are activated in stems when inducing GR-ARF5ΔIII/IV plants. This would argue for a general role of ARF5 in promoting vascular differentiation and for the existence of one pool of stem cells marked by PXY promoter activity and feeding both xylem and phloem production. Alternatively, promotion of xylem differentiation is translated rapidly into the promotion of phloem differentiation by cell-to-cell signalling. The fact that the stem cell-to-phloem ratio is not altered in arf5 mutants would argue for the latter option. Consistent with a crucial role of cell-autonomous auxin signalling in cambium stem cells, ARF3, ARF4 and ARF5 expression was found in PXY-positive cells with ARF5 being exclusively active in those. ARF3 and ARF4 have previously been shown to act in part redundantly in the establishment of abaxial identity in lateral organs 37 . In line with this function, we found both genes being mostly expressed distally of the cambium in phloem-related cell types. In fact, the lack of any effect on cambium or WOX4 activity when modulating ARF3 activity exclusively in PXYpositive cells suggests that at least ARF3 functions outside of this domain when regulating cambium activity and that ARF transcription factors positively regulating WOX4 transcription still have to be discovered. Whether the phloem-related expression of ARF3 and ARF4 modulates the activity of cambium regulators expressed in areas distally to the PXY expression domain like MORE LATERAL GROWTH1 (MOL1) 29 or CLAVATA/ESR41/ 42/44 (CLE41/42/44) 26,27,64 remains to be determined. Transcriptionally, at least, our modulation of auxin signalling had no effect on MOL1 or CLE41/42/44 mRNA abundance.
Collectively, we found a role of auxin signalling in the cambium sharing features with both the situation in the RAM where auxin regulates cell divisions 65 and the SAM where auxin, and particular ARF5, is strongly correlated with cell differentiation 4 . Thereby, we enlighten a long-observed role of auxin signalling in radial plant growth and reveal that its function is partly specific in different stem cell niches. The involvement of different auxin signalling components regulating individual aspects of meristem activity may provide a set-up required for regulating a complex developmental process by one signalling molecule.  Sterile culture. To induce adventitious root formation in the strong arf5 mutant allele mp-B4149, seeds were liquid sterilized by 70% ethanol and incubation in 5% sodiumhyperchloride followed by three washes with ddH 2 O. After stratification at 4°C in the dark for 3 days, seeds were sown on 1/2 Murashige-Skoog (MS) medium plates (including B5 vitamins) in rows and grown vertically. After 7 days of growth under SD (8 h light, 16 h dark) conditions, rootless mutant as well as wild-type-looking seedlings from the segregating population were bisected with a scalpel and transferred to adventitious root-inducing medium (1/2 MS (including B5 Vitamins) + 1.5 % sucrose + 3 µg/ml indole butyric acid + 0.7 % agar + 50 µg/ ml ampicillin) 43 . After additional 2 weeks of growth under SD conditions, successfully rooted seedlings were transferred to soil and grown under SD conditions for 1 additional week before they were transferred to LD conditions. Plants that survived the transfer were genotyped for mp-B4149 and only wild-type and Plasmid construction. To avoid diffusion, all fluorescent proteins were targeted to the ER by fusing them to the corresponding sequence motif (ER+HDEL motif 67 ). For generating pDR5revV2:YFP (pKB46), we initially inserted the ADAPTOR PROTEIN-4 MU-ADAPTIN (AP4M, At4g24550) terminator, amplified from genomic DNA using the At4g24550_for1/At4g24550_rev1 primer pair, into pLC075 containing the DR5revV2 promoter fragment 31 using BamHI/XhoI restriction sites. A fragment carrying the ER-EYFP-HDEL coding sequence (CDS) was inserted in the resulting pLC075:AP4Mterm using the BamHI restriction site, to obtain pLC075:YFP:AP4Mterm. The complete reporter fragment was inserted in the binary vector pGreenII0179 68 using KpnI/XhoI restriction sites. For generating p35S:Myc-GR-bdl (pKB9), the Myc-GR-bdl fragment was amplified from genomic DNA of pRPS5a:Myc-GR-bdl 58 using the Myc_for1/BDL_rev3 primer pair. The resulting fragment was inserted into the pGreen0229 vector 68 containing the 35S promoter (pGreen0229-35S) using XbaI/BamHI restriction sites. To produce pAlcA: iaaM (pKB2), the iaaM CDS was amplified from piaaM (pIND:IND-iaaM) 69 using the IAAMfor1/IAAMrev1 primer pair and introduced into pGreen0229-AlcA 32 using AatII/EcoRI restriction sites. pWOX4:AlcR (pTOM55) was produced by amplifying the WOX4 promoter using the primers WOX4for11/WOX4ref9 and inserting the resulting fragment into pAlcR-GUS 32 using SpeI/NotI sites. The pPXY: Myc-GR-bdl (pKB45) construct was generated by cloning the Myc-GR-bdl fragment, amplified from pKB9 using the Myc_for5/BDL_rev7 primer pair, in pGreen0229 containing the PXY promoter (pTOM50 28 ) using NcoI/Cfr9I restriction sites. The promoter regions of BDL 35 were amplified from genomic DNA using the BDL_for2/BDL_rev4 and BDL_for3/BDLrev5 primer pairs. Both fragments were cloned into pGreenII0179 using NotI/XbaI and Cfr9I/KpnI restriction sites, respectively. The resulting plasmid (pKB27) was used to produce the pBDL: YFP (pKB28 using ER-EYFP-HDEL) and pBDL:Myc-GR-bdl (pKB29) constructs by inserting fragments carrying the respective CDSs using NcoI/Cfr9I restriction sites. For generating ARF3, ARF4 and ARF5 reporter constructs, promoter regions of the three genes were amplified from genomic DNA using the ARF3for1/ARF3rev1 and ARF3for2/ARF3rev2, ARF4for1/ARF4rev1 and ARF4for2/ARF4rev2 and MP_for7/ MP_rev5 and MP_for5/MP_rev6 primer pairs. Both fragments were cloned for each gene into pGreen0229 using NotI/BamHI and BamHI/KpnI (ARF3), NotI/ SpeI and Cfr9I/KpnI (ARF4) and NotI/BamHI and BamHI/ApaI (ARF5) restriction sites. The resulting plasmids (pKG40 (ARF3), pKG41 (ARF4) and pKB3 (ARF5)) were used to produce the pARF3:YFP (pKB30), pARF4:YFP (pKB31), pARF5:YFP (pKB24) and pARF5:mCherry (pKB4) constructs by inserting fragments carrying the respective CDSs using BamHI, SpeI and NdeI/XhoI restriction sites, respectively. To produce p35S:GR-ARF3 (pKB42) and p35S:GR-ARF5 (pKB17), we amplified the GR open reading frame from pKB9 using the GR_for1/GR_rev3 primer pair and inserted the resulting fragment in pGreen0229-35S using XbaI/ Cfr9I restriction sites. An additional unannotated SalI restriction site in the pGreen0229 backbone was removed by PCR-based silent mutagenesis using the NOS-mut_for1/NOS_mut_rev1 primer pair. In the resulting pGreen0229-35S:GR (pKB41) vector, we inserted the ARF3 and ARF5 CDS, amplified from cDNA using the ARF3_for4/ARF3_rev4 and MP_for16/MP_rev14 primer pair, using SalI/Cfr9I and SalI/EcoRI restriction sites, respectively. pPXY:GR-ARF3 (pKB43) was generated by cloning the GR-ARF3 fragment, amplified from pKB42 using the GR_for5/ ARF3_rev4 primer pair, in pTOM50 using NcoI/Cfr9I restriction sites. For generating pPXY:GR-ARF5ΔIII/IV (pKB25), the GR-ARF5ΔIII/IV fragment with a stop codon was amplified from pKB17 using the MP_for18/MP_rev16 primer pair and inserted in pTOM50 using NcoI/Cfr9I restriction sites. For generating the pWOX4: LUC (firefly);p35S:LUC (Renilla) (pKB55) reporter, the pZm3918:LUC (firefly) fragment in pZm3918:LUC (firefly);p35S:LUC (Renilla) (pGreen-LUC-REN) was excised by digest with KpnI/XbaI and replaced by a pWOX4:LUC (firefly) fragment previously excised from pWOX4:LUC (pMS80) using KpnI/NheI restriction sites. To produce p35S:ARF3 (pKB44), the ARF3 CDS was amplified from pKB42 using the ARF3_for6/ARF3_rev4 primer pair and introduced it in pGreen0229-35S using XbaI/Cfr9I restriction sites. For generating p35S:ARF5ΔIII/IV (pKB40), the ARF5ΔIII/IV CDS was amplified from pKB25 using MP_for17/MPrev15 and introduced in pGreen0229-35S using XbaI/EcoRI restriction sites. All constructs were sequenced, and after plant transformation by floral dip 70 , single copy transgenic lines were identified by Southern blot analyses and representative lines were used for crosses and further analyses. The promoter regions of RUL1 36 were amplified from genomic DNA using the RULfor2/RULrev2 and RULfor3/RULrev3 primer pairs. Both fragments were cloned into pGreen0229 68 using XhoI/EcoRI and SpeI/NotI restriction sites, respectively. The resulting plasmid (pJA16) was used to produce the pRUL1:YFP (pJA19 using ER-EYFP-HDEL) construct by inserting fragments carrying the respective CDS using NdeI/XmaI restriction sites. For generation of the bacterial expression vector pMG210, a 6xHis tagged N-terminal fragment of the maltose binding protein (MBP) 71 was amplified from pMAL-c2X (New England Biosciences) using the malE_fw2/malE_rv1 primer pair and inserted into pMAL-c5X (New England Biosciences) using MunI/BglII restriction sites. The plasmid pMG210 was further modified by replacing the C-terminal fragment of MBP, which contains the Factor Xa protease cleavage site, with the respective fragment from pETMBP_1a 72 , which contains a TEV protease cleavage site, by using BglII/NcoI restriction sites to generate pJH036. To produce the final expression vector pMG307, the ARF5 DNA-binding domain (ARF5-DBD, aa 120-274) was amplified from cDNA using the ARF5-DBD_fw1/ARF5-DBD_rv1 primer pair and cloned into pJH036 using NcoI/XhoI and NcoI/SalI restriction sites, respectively. All primers mentioned in this section are listed in Supplementary  Table 1.  Transient reporter activity assays. For transient reporter activity assays, protoplasts derived from an Arabidopsis (Col-0) dark-grown root cell suspension culture (kindly provided by Claudia Jonak, GMI, Vienna) were isolated and transfected. For transfection, we used 10 µg of reporter construct (pKB55) containing p35S:LUC (Renilla) as an internal control and 10 µg of each effector construct. The transfected protoplasts were diluted with 240 mM CaCl 2 (1:3) followed by cell lysis and dual-luciferase assay using the Dual-Luciferase Reporter Assay System (Promega) and following the manufacturer's instructions. Luminescence was measured using a Synergy H4 Hybrid Multiplate Reader (BioTek). For each reporter/effector combination, 3-5 technical replicates were done and the experiments were repeated at least three times. For experimental analysis, Firefly Luciferase activity was normalized to Renilla Luciferase activity.

Methods
RNA preparation and quantitative reverse transcriptase-PCR (qRT-PCR ARF-binding site prediction. Three thousand bp upstream of the start codon of the WOX4 gene were used for the ARF-binding site prediction. In this analysis, one frequency matrix from CIS-BP 77 (M0147_1.02) and two from a Dap-seq analysis 78 (ARF_ARF2_col_v31, ARF_ecoli_MP_col) were applied. We computed position weight matrices as log-odds weights 52 . The significance of potential sites was estimated as a p-value derived from a matrix score 79 .
ARF5-DBD purification from bacteria. For the Escherichia coli expression culture 100 ml LB medium containing ampicillin (100 µg/ml) and chloramphenicol (25 µg/ ml) was inoculated with 5 ml of an overnight culture of the Rosetta (DE3) pLysS strain (Novagen) previously transformed with pMG307 and grown at 37°C until an OD 600 of 0.6 was reached. Expression of the (6xHis)MBP-ARF5-DBD fusion protein was induced by adding IPTG to a final concentration of 1 mM, and the culture was incubated for additional 5 h. Cells were harvested by centrifugation at 3000 × g for 20 min and the pellet stored at −20°C. All subsequent steps were performed on ice or at 4°C. The recombinant fusion protein was released from thawed cells by resuspension of the pellet in 2 ml lysis buffer (20 mM Tris-HCl pH 8, 150 mM NaCl, 2 mM dithiothreitol (DTT), 0.1% Lysozyme (Roth), 7.5 U/ml Benzonase (Novagen), 1× complete EDTA-free Protease Inhibitor (Roche), 0.2% Nonidet P40) and pulsing three times for 10 s at 5% amplitude with the S-4000 sonicator (Misonix) using a microtip. The lysate was cleared by centrifugation at 10,000 × g for 20 min and the supernatant loaded on a pre-equilibrated Ni-NTA purification column. To prepare the Ni-NTA purification column, 400 µl of Ni-NTA Superflow resin (Qiagen) was applied to a Micro Bio-Spin Column (Bio-Rad), washed with 2 ml Milli-Q water (Micropore) and equilibrated with 2 ml wash buffer 1 (20 mM Tris-HCl pH 8, 150 mM NaCl, 2 mM DTT, 0.2% Nonidet P40, 10 mM Imidazole pH 8) using gravity flow. The loading step was repeated twice to enhance binding of the (6xHis)MBP-ARF5-DBD fusion protein. The resin was sequentially washed with 2 ml of each of the following buffers: wash buffer 1, wash buffer 2 (20 mM Tris-HCl pH 8, 150 mM NaCl, 2 mM DTT, 10 mM Imidazole pH 8), wash buffer 3 (20 mM Tris-HCl pH 8, 1 M NaCl, 2 mM DTT, 10 mM Imidazole pH 8), and wash buffer 4 (20 mM Tris-HCl pH 8, 1 M NaCl, 2 mM DTT, 20 mM Imidazole pH 8). The fusion protein was released from the resin by adding 1 ml elution buffer (20 mM Tris-HCl pH 8, 1 M NaCl, 2 mM DTT, 330 mM Imidazole pH 8), incubated with 10 µg TEV protease overnight and the reaction was subsequently loaded on a pre-equilibrated SP purification column. Preparation of the SP purification column was done by applying 400 µl of SP Sepharose Fast Flow (GE Healthcare) to a Micro Bio-Spin Column (Bio-Rad), washing with 2 ml Milli-Q water (Micropore) and equilibrating with 2 ml SP wash buffer (20 mM Tris-HCl pH 8, 150 mM NaCl, 2 mM DTT) using gravity flow. The loading step was repeated twice to enhance binding of the ARF5-DBD. The resin was washed by adding 2 ml of SP wash buffer, and the purified protein was released upon addition of 250 µl SP elution buffer (20 mM Tris-HCl pH 8, 500 mM NaCl, 2 mM DTT). Concentration of the purified ARF5-DBD was determined by absorption at 280 nm considering molecular weight and molar extinction coefficient and purity verified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by Coomassie Brilliant Blue staining.
Fluorescent EMSAs. EMSAs were performed using fluorescent probes produced by annealing complementary pairs of CY5 5ʹ-labelled oligonucleotides (Eurofins) in annealing buffer (10 mM Tris-HCl pH 8, 1 mM EDTA, 50 mM NaCl). Oligonucleotide sequences are listed in Supplementary Table 1. For the binding reactions, 12.5 pmol of the ARF5-DBD was incubated with 200 fmol CY5 probe in binding buffer (final reaction conditions: 10 mM Tris-HCl pH 7.4, 75 mM NaCl, 50 ng/µl Poly(dI-dC) (Thermo Scientific), 1 mM DTT) for 20 min at room temperature. Samples were mixed with 6× Orange G loading dye and applied to a 6% native polyacrylamide gel in 0.5× TBE buffer. Electrophoresis was conducted at 300 V (~20 mA per gel) for 20 min and the gel shifts were subsequently recorded by using the Advanced Fluorescence Imager (Intas) and the ChemoStar Professional software with red dye filter settings (628/32-716/40). Competitor assays were performed as described above but protein amount was doubled and binding reactions were pre-incubated for 10 min with the non-labelled competitor before adding the fluorescent probe.
Data availability. Raw sequencing data produced in this study have been uploaded to NCBI's Gene Expression Omnibus (GEO) database 74 and are accessible through GEO Series accession number GSE98193. The authors declare that all other data supporting the findings of this study are available within the paper and its supplementary information files.