Lateral root emergence in Arabidopsis is dependent on transcription factor LBD29 regulation of auxin influx carrier LAX3

Lateral root (LR) emergence represents a crucial developmental program enabling new primordia to breach the overlying endodermal, cortical and epidermal cell layers and enter the surrounding soil environment (Péret et al., 2009; Fig. 1A). This process is tightly controlled in order to limit damage to the parental root from which a new lateral root primordium (LRP) originates. The hormone signal auxin and several of its signaling and transport components have been demonstrated to play an essential role during LRP emergence (Swarup et al., 2008; reviewed in Lavenus et al., 2013). Auxin has a specialized transport machinery, and its signal triggers the degradation of AUXIN/INDOLE ACETIC ACID (AUX/IAA) proteins via the SCF^TIR1 complex (Dharmasiri et al., 2005; Kepinski and Leyser, 2005). The degradation of AUX/IAA proteins (that function as transcriptional repressors) allows interacting transcriptional proteins termed AUXIN RESPONSE FACTORs (ARFs) to change auxin responsive gene expression in order to elicit developmental responses (Dharmasiri and Estelle, 2004).

In the model plant Arabidopsis thaliana, LAX3, an auxin influx carrier belonging to the AUX/LAX gene family, plays an important role during LR development, by facilitating the emergence process (Péret et al., 2009; Swarup et al., 2008). Swarup et al. (2008) proposed a model in which an increased level of auxin in the cortical cells overlying LR primordia, induces LAX3 expression by targeting the degradation of the IAA14/SLR repressor protein, allowing ARF7 and ARF19 to activate auxin responsive genes. The arf7 arf19 double mutant and the iaa14/slr-1 gain-of-function mutant show impaired expression of LAX3, suggesting that LAX3 expression is mediated by the auxin signaling pathway module IAA14/SLR-ARF7-ARF19 (Swarup et al., 2008). In the cortical cells, the increase of LAX3 expression triggers a positive-feedback loop stimulating further auxin uptake from LRP. The consequence of auxin accumulation is the induction of expression of a set of cell wall remodeling genes, such as polygalacturonases and xyloglucan endotransglucosylase/hydrolase, which are involved in pectin polymer cleavage and cell wall loosening, respectively (Laskowski et al., 2006; Swarup et al., 2008). Consistent with the model that the LAX3 positive feedback loop is important for emergence, the expression of these cell wall remodeling enzymes at the emergence site is LAX3 dependent (Swarup et al., 2008). Hence, by inducing cell wall remodeling in overlying cells and triggering their separation, the growing primordium is able to pass through the outer root cell layers and emerge (Péret et al., 2009; Swarup et al., 2008).

LAX3 is expressed in just two cell files overlaying new LRP (Swarup et al., 2008; Fig. 1B). To understand how this striking pattern of LAX3 expression is regulated, we developed a mathematical model that captures the network regulating its expression and auxin transport within realistic three-dimensional cell and tissue geometries (Péret et al., 2013). Our model revealed that, for the LAX3 spatial expression to be robust to natural variations in root tissue geometry, the efflux carrier PIN3 is also required. To prevent LAX3 from being transiently expressed in multiple cell files, the model predicted that PIN3 and LAX3 genes must be induced consecutively. Consistent with this prediction, the translational inhibitor cycloheximide was observed to block auxin upregulation of transcript abundance of LAX3 (but not PIN3; Péret et al., 2013). Hence, LAX3 appears to function as a secondary (rather than primary) auxin responsive gene. However, the underlying molecular basis of LAX3 induction by auxin remained unresolved.

In this study, we initially describe that ARF7 is essential for auxin-dependent LAX3 induction during LR emergence. However, we demonstrate using several independent lines of experimental evidence that ARF7 regulates LAX3 indirectly. Instead, we report that the LATERAL ORGAN BOUNDARIES-DOMAIN/ASYMMETRIC LEAVES 2-LIKE (LBD/ASL) transcription factor (TF) family member LBD29 functions as a direct positive regulator of LAX3 auxin-dependent expression. Our results position LBD29 at a key node downstream of auxin and ARF7 in the LR emergence regulatory network.

We initially addressed the ARF-dependent regulatory mechanisms controlling auxin-inducible LAX3 expression. It was previously reported that the abundance of LAX3 mRNA is reduced in an arf7 mutant background, suggesting that LAX3 expression is dependent on this ARF transcription factor (Okushima et al., 2007; Swarup et al., 2008). To test this, a RT-qPCR-based assay was designed to transcript profile Arabidopsis wild-type (WT) and arf7 mutant roots exposed to external auxin (1 µM IAA) for varying lengths of time. Transcript profiling revealed that auxin induction of LAX3 mRNA was abolished in the arf7 mutant background (Fig. 1C). Similarly, when pLAX3:GUS and pLAX3:LAX3-YFP reporters (Swarup et al., 2008) were expressed in arf7, auxin induction of both reporters was abolished (Fig. S1AB). Hence, ARF7 appears to be crucial for auxin induction of LAX3 expression.

We next addressed whether ARF7 regulated LAX3 auxin induction directly or indirectly. Whilst the LAX3 promoter sequence contains a canonical ARF binding motif (−939 bp), transcript profiling of auxin-treated WT roots revealed that upregulation of LAX3 mRNA was first detected 6 h after hormone induction (Fig. 1C). Induction of primary auxin-responsive transcripts is typically detected within minutes rather than hours, suggesting that LAX3 may be a secondary auxin-responsive gene and hence not a direct target for ARF7 regulation (Péret et al., 2013; Mellor et al., 2015). This is consistent with bioinformatics analysis of transcriptomic data from the arf7 arf19 mutant complemented by a dexamethasone (DEX)-inducible ARF7-GLUCOCORTICOID RECEPTOR (GR) fusion protein under its native ARF7 promoter (as described in Lavenus et al., 2015), which categorized LAX3 as a positive indirect target of ARF7 (Fig. S1C).

To directly test the functional importance of the AuxRE motif within the LAX3 promoter for auxin induction and gene activity, we performed targeted in vitro mutagenesis studies. The LAX3 (−939 bp) AuxRE motif and two mutant variants (termed IVM1 and IVM2; Fig. S2A) were re-introduced (as part of a functional pLAX3:LAX3-YFP transgene) into an aux1 lax3 mutant background. Phenotypic analysis revealed that, like the wild-type pLAX3:LAX3-YFP reporter, both IVM1 and IVM2 promoter transgene variants retain their ability to rescue LR emergence in the aux1 lax3 mutant background (Fig. S2B). Confocal imaging confirmed that both IVM1 and IVM2 variants of the pLAX3:LAX3-YFP reporter also retained their ability to be induced by auxin in root cortical cells (Fig. S2C).

To independently assess the role of the AuxRE versus other regulatory regions, a LAX3 promoter deletion approach was also employed (Fig. 2). The 1.98 kb LAX3 promoter was truncated at four different positions (Fig. 2A; termed Δ1, Δ2, Δ3 and Δ4), to create increasingly shorter promoter fragments, then fused to the LAX3-YFP reporter and transformed into the aux1 lax3 double mutant. Multiple transgenic lines expressing each pLAX3:LAX3-YFP promoter deletion were initially scored for complementation of the aux1 lax3 LR phenotype. All Δ1 and Δ2 promoter deletion lines fully complemented the aux1 lax3 LR defect (Fig. 2B) and retained auxin-inducible expression (Fig. 2C,D). This result demonstrates that the 826 bp sequence upstream of the start of the LAX3 coding sequence (that does not contain the AuxRE) is sufficient to drive auxin-inducible expression and promote LR emergence. In contrast, all Δ3 promoter deletion lines only partially complemented the aux1 lax3 LR phenotype, whereas no complementation was observed for any of the Δ4 promoter deletion lines (Fig. 2B). Quantitative analysis of Δ3 and Δ4 reporter lines following auxin treatment gave similar results (Fig. 2C,D). Hence, multiple independent lines of evidence suggest that the AuxRE promoter element is not necessary for auxin-inducible LAX3 expression. Instead, other regulatory motifs encoded closer to the start of the LAX3 coding sequence appear to be functionally important. We conclude that ARF7 regulates auxin-inducible LAX3 expression indirectly, raising the question about which transcription factor(s) acts between ARF7 and LAX3 in the LR emergence gene regulatory network.

In order to identify putative transcription factors that regulate LAX3, a yeast one-hybrid (Y1H) assay was performed. As bait, the LAX3 promoter (1374 bp) was fused to lacZ and HIS3 reporter genes. In the Y1H system, reporter gene expression is activated when a TF interacts with the DNA bait. A root TF collection (Gaudinier et al., 2011) containing >650 genes was fused to a transcription activation domain (AD) and used as a prey. A total of 17 root stele-expressed TFs were found to bind to the LAX3 promoter sequence (Table S1). These transcription factors included five homeodomain leucine zipper (HD-Zip) proteins (HAT22, ATHB52, ATHB6, PHV/ATHB9, ATHB40), two zinc finger-homeodomain (ZF-HD) proteins (HB21, HB30) and the LBD/ASL protein LBD29.

LBD29 represents the most promising candidate identified in the Y1H screen for an intermediary transcriptional regulator between ARF7 and LAX3 based on several criteria. First, LBD29 expression is induced minutes after auxin treatment (Okushima et al., 2007). Second, characteristic of many primary auxin-response genes, LBD29 expression is strongly induced following cycloheximide (CHX) treatment (Okushima et al., 2007). Third, LBD29 expression can be induced in pARF7:ARF7-GR/arf7 arf19 seedlings following treatment with DEX plus auxin (Okushima et al., 2007). Fourth, ARF7 has been shown to bind to LBD29 promoter fragments containing AuxRE motifs using EMSA and ChIP-PCR techniques, respectively (Okushima et al., 2007; Lavenus et al., 2015). Finally, overexpression of LBD29 can also partially restore LR formation in the arf7 arf19 mutant background (Okushima et al., 2007).

To directly test the functional importance of LBD29 for LAX3 auxin-inducible expression, we employed the lbd29-1 T-DNA insertion line (SALK_071133). In this SALK line (the only insertion line currently available for this gene), the T-DNA sequence is inserted in the LBD29 promoter region 3′ of the nearest AuxRE to the transcription start site (Fig. S3A). Whilst the position of the T-DNA insert does not disrupt the coding sequence, RT-qPCR analysis revealed that despite a higher basal level of expression, it significantly attenuated (>10-fold compared with WT) LBD29 auxin-inducible expression (Fig. S3B,C). Interestingly, LAX3 mRNA abundance was no longer auxin inducible in the lbd29-1 insertion line compared with the WT (Fig. 3A), suggesting that this T-DNA allele attenuates the ability of LBD29 to function as an auxin-inducible positive regulator of LAX3 expression. To verify this, the lbd29-1 insertion line was also crossed with the pLAX3:LAX3:YFP reporter (Swarup et al., 2008). In the WT, LAX3-YFP is constitutively expressed in central root stele tissues, plus a small number of cortical cells (and later in epidermal cells) overlying new LR primordia (Fig. 3C,D) (Swarup et al., 2008). However, in the lbd29-1 mutant background the pLAX3:LAX3-YFP reporter was no longer expressed in cortical cells overlying new primordia (Fig. 3E-F). IAA treatment also could not induce ectopic expression of LAX3 in the lbd29-1 mutant (Fig. 3I,J) compared with the WT (Fig. 3G,H).

We next examined the effect of the lbd29-1 allele on auxin-regulated genes controlled by LAX3. For example, the POLYGALACTURONASE (PG) gene is auxin and LAX3 dependent and is specifically expressed at sites of LR emergence (Swarup et al., 2008). Consistent with a role for LBD29 mediating upregulation of LAX3, lbd29-1 roots also failed to show induced PG expression following auxin treatment in contrast to the WT (Fig. 3B). Hence, LBD29 appears to not only control upregulation of LAX3 via auxin in outer root tissues but also (indirectly) impacts expression of LAX3-dependent downstream targets of the LR emergence gene regulatory pathway.

Transcript profiling and reporter studies have demonstrated that LBD29 is essential for auxin-inducible expression of LAX3 and other genes involved in cell separation in cells overlying new LRP (Fig. 3). Logically, as a regulator of LAX3 expression (and its downstream targets), disrupting the induction of LBD29 (in the case of the lbd29-1 allele) should cause a lax3-like mutant phenotype and delay LR emergence. To detect any change in the lbd29-1 LR emergence rate, we employed a gravistimulation-based bioassay (Péret et al., 2012), which after subjecting seedlings to a 90° gravity stimulus, leads to the highly synchronized temporal development of a new primordium on the outer side of the root bend (Lucas et al., 2008). For example, 18 h after a gravistimulation, WT root bends contain mainly stage II LRP, whereas by 42 h, many primordia were close to emergence (stage VI or VII) or already emerged (stage e; Fig. 4). In the case of lax3, 18 h after a gravity stimulus, mutant roots displayed proportionately more stage I primordia compared with the WT (Fig. 4A). However, the biggest phenotypic difference was observed 42 h after the gravistimulus, since no emerged LR were detected in lax3 compared with the WT, and most primordia were still at earlier stage IV or V (Fig. 4A). Phenotypic analysis of the lbd29-1 insertion line revealed an even greater delay in LR development compared with lax3, featuring a higher proportion of stage IV and V primordia (Fig. 4A), consistent with LBD29 acting upstream of LAX3 in the LR emergence regulatory pathway.

To provide additional independent genetic evidence to probe the function of LBD29 during LR emergence, we produced a transgenic line expressing the LBD29 genomic sequence fused to the SRDX transcriptional repressor domain in the WT (Col-0) background. The resulting gLBD29-SRDX fusion protein is designed to repress LBD29 target genes by blocking their transcription, thereby mimicking an lbd29 loss of function allele. Phenotypic analysis of two independent gLBD29-SRDX lines revealed that both exhibited delayed LR development and featured a higher proportion of stage V primordia versus the WT, as in lbd29-1 (Fig. 4B), consistent with LBD29 acting as a positive regulator of the LR emergence regulatory pathway. Further independent confirmation of the role of LBD29 during LR emergence was generated by expressing a translational fusion of the LBD29 protein to the vYFP marker in the lbd29 mutant background. The lbd29-1 line transformed with the pLBD29:LBD29-vYFP construct exhibited full restoration of the wild-type LR phenotype using the LR bending assay (Fig. 4C).

Further characterization of WT versus the lbd29-1 T-DNA insertion line was performed to investigate any additional alterations of its LR phenotype in 10-day-old seedlings. This revealed that the number of emerged LRs was reduced in the lbd29-1 insertion line (Fig. S4A-C) but that neither the stages of LR primordia distribution (Fig. S4D) nor the total number of LR primordia (Fig. S4E) was affected in lbd29-1 compared with the WT. Taken together, these results suggest that the reduced number of emerged LR in the lbd29-1 insertion line is not related to a defect in LR initiation or primordia development, but is due to slower organ emergence.

Several independent lines of evidence reported in this paper suggest that LBD29 regulates LR emergence by controlling LAX3 expression. To test whether the LBD29 transcription factor binds directly to the LAX3 promoter in planta, we performed chromatin immunoprecipitation (ChIP) qPCR analysis using an anti-GFP antibody on nuclear samples prepared from WT (Col-0) (as a negative control) and 35S:LBD29-GFP lines. Four regions of the LAX3 promoter were tested by ChIP-qPCR for LBD29-GFP enrichment (Fig. 5A). All four regions exhibited enrichment (∼5-fold) but in a fragment spanning −387 to −192 bp from the start codon of the LAX3 promoter, LBD29-GFP enrichment was enriched ∼25-fold compared with the WT control (Fig. 5B). In silico sequence analysis identified 10 LBD/AS2 family binding motifs (Husbands et al., 2007) in the LAX3 promoter (Table S2). Three of these motifs occurred within the fragment that exhibited the highest enrichment for LBD29-GFP using ChIP-qPCR (Fig. 5B). Our ChIP-based results suggest that LBD29 directly binds to the LAX3 promoter in planta, presumably activating its expression.

To validate that LBD29 functions as a transcriptional activator for LAX3 expression, we tested the ability of LBD29 to transactivate a LUC-Trap reporter (Lau et al., 2011) fused to the LAX3 promoter. The pLAX3:LUC plasmid was co-expressed in protoplasts derived from tobacco suspension cells with other test plasmids and imaged using the dual luciferase transient expression assay. A two-fold increase in relative luciferase activity was detected when the pLAX3:LUC reporter was co-transfected with the LBD29 effector plasmid (Fig. 5C) but not with the ARF16 control (Fig. 5C). Our transient expression data suggest that LBD29 can operate as a positive transcriptional regulator of LAX3 expression.

We reasoned that if LBD29 positively regulates LAX3 expression, an LBD29 overexpression line is likely to cause overexpression of a LAX3 reporter even in the absence of exogenous auxin. To test this, we crossed the pLAX3:LAX3-YFP reporter (Swarup et al., 2008) with the overexpression line of LBD29 (35S:LBD29-GFP) (Okushima et al., 2007). Lines homozygous for both transgenes displayed ectopic expression of the LAX3-YFP reporter in both root cortical and epidermal cells in the absence of exogenous auxin application (Fig. 5D). Taken together, these results suggest that LBD29 positively regulates LAX3 transcription.

An earlier study investigating the spatial and temporal expression patterns of selected members of the LOB/AS2 gene family during root development reported (using GUS-based reporters) that LBD29 is expressed in LRP (but not overlying) cells (Okushima et al., 2007). Given the large body of evidence we have assembled showing that LBD29 is required to bind directly to the LAX3 promoter to activate its expression, we generated new fluorescence-based LBD29 transcriptional and translational reporter lines to address its binding ability.

We initially fused an ER-localized CFP reporter to the LBD29 promoter sequence. Transgenic lines expressing the pLBD29:erCFP transcriptional reporter clearly exhibited a CFP signal in new LRP plus a small number of cells directly overlying new primordia (Fig. 6A,B). Roots were clearly observed to express the LBD29 driven erCFP reporter in cortical cells overlying Stage I/II LRP, coincident with pLAX3:LAX3-YFP expression first being detected (Fig. 1; Swarup et al., 2008). The pLBD29:erCFP transcriptional reporter was also clearly expressed in endodermal cells overlying LRP (Fig. 6A,B). However, no LAX3 expression was detected in this tissue (Fig. 1; Swarup et al., 2008), suggesting that additional transcriptional repressor proteins may be required to impose the observed pattern of LAX3 spatial expression.

A similar spatial expression pattern was observed after fusing the LBD29 genomic sequence to a single copy of YFP (VENUS) and creating the pLBD29:gLBD29:Venus translational reporter line (Fig. 6C,D). The pLBD29:gLBD29:Venus reporter clearly exhibited a nuclear-localized YFP signal in a subset of cells directly overlying new LRP (Fig. 6C,D), consistent with LBD29 encoding a transcription factor. In addition, temporal analysis of the pLBD29:gLBD29:Venus reporter line (Fig. 6) revealed that its induction preceded LAX3 expression (Fig. 1B). Finally, RT-qPCR assays revealed that the pLBD29:gLBD29:Venus reporter line when used to complement the lbd29-1 LR emergence defect (Fig. 4C) was also able to restore auxin-inducible LAX3 expression (Fig. S5). In summary, our results are consistent with the proposed role of LBD29 as a transcriptional regulator of LAX3.

LBD/ASL genes encode a plant-specific family of transcription factors (Husbands et al., 2007) that have been implicated in a variety of developmental processes during leaf, flower and root morphogenesis (Iwakawa et al., 2002; Majer and Hochholdinger, 2011; Okushima et al., 2007, 2005; Soyano et al., 2008; Xu et al., 2008). In Arabidopsis thaliana, the LBD gene family contains 43 members, of which the LOB gene is the founding member (Shuai et al., 2002). LOB plays a role in organ separation and lateral organ development and encodes a transcription factor, expressed at boundaries within shoot tissues (Shuai et al., 2002). In roots, family members LBD16, LBD18 and LBD29 play important roles during LR development (Okushima et al., 2005, 2007; Lee et al., 2009; Lavenus et al., 2015; Fig. 7A). During LR initiation, LBD16 has been shown to play an important role promoting asymmetric cell division of LR founder cells, controlling polarized nuclear migration to the common cell pole between pairs of founder cells (Goh et al., 2012). LBD18 and LBD33 positively regulate the cell cycle via the transcriptional regulation of E2Fa (Berckmans et al., 2011). LBD29 has also been reported to be involved in the regulation of the cell cycle progression during LR formation (Feng et al., 2012).

In this current study, we demonstrate that LBD29 plays a key role during LR emergence. Several other LBD genes have been linked with this developmental program. Like lbd29-1, the lbd16-1 and lbd18-1 single mutants also displayed a reduction in LR emergence, which was enhanced in the lbd16-1 lbd18-1 double mutant (Lee et al., 2009; Okushima et al., 2007). LBD18 has been demonstrated to function downstream of the auxin influx carrier LAX3 during LR emergence (Lee et al., 2014). In addition, the triple mutant lbd16-1 lbd18-1 lbd33-1 displayed a further reduction in the number of LRs emerged compared with any of the single or double mutants mentioned above (Goh et al., 2012). These observations may indicate the presence of functional redundancy between selected members of the LBD gene family. Alternately, it could reflect that these LBD genes play roles during distinct (but interacting) LR developmental processes such as organ initiation, patterning and emergence, which have additive phenotypic effects when disrupted in a multiple mutant background.

Any distinct regulatory functions that LBD family members play during LR development cannot be explained simply on the basis of each gene exhibiting contrasting spatio-temporal expression patterns (Fig. 7A). For example, whilst LBD16 and LBD29 are both expressed in pericycle cells prior to LR initiation (Goh et al., 2012; Fig. 6C), only transgenic lines expressing a LBD16-SRDX transcriptional repressor protein block the initial asymmetric cell division in LR founder cells (Goh et al., 2012), whereas gLBD29-SRDX lines are defective in organ emergence (Fig. 4). This suggests that LBD16 and LBD29 proteins target distinct sets of genes during LR development. Currently, the molecular basis of LBD16 and LBD29 target specificity is unclear. Several LBD proteins have been reported to interact with other family members or distinct classes of transcription factors that may help confer target specificity. For example, LBD18 and LBD33 dimerize to regulate cell cycle by activating E2FA transcription during LR initiation (Berckmans et al., 2011) and LBD6 (AtAS2) interacts with a MYB transcription factor AtAS1 during leaf development (Xu et al., 2003). In summary, LBD family members play key regulatory roles throughout plant development, including LBD16, LBD18 and LBD29, which are critical for LR development.

Auxin functions as a key regulatory signal during LR development (reviewed by Lavenus et al., 2013). In Arabidopsis, auxin alters the expression of large numbers (>2000) of genes during LR development via ARF transcription factors such as ARF7 and ARF19 (Okushima et al., 2005; Vanneste et al., 2005). ARF7 is particularly important as it functions to activate gene regulatory networks that control LR initiation, patterning and emergence (Lavenus et al., 2015; Swarup et al., 2008). ARF7 (and ARF19) regulates LR development (in part) by activating the expression of several auxin-inducible LBD family members (Okushima et al., 2005), as ectopic expression of LBD16 and LBD29 can partially rescue the arf7 arf19 LR-less phenotype (Okushima et al., 2007). LBD16 and LBD29 represent key nodes within the LR gene regulatory network (GRN; Lavenus et al., 2015). ChIP-qPCR experiments have recently demonstrated that LBD16 and LBD29 represent direct targets for ARF7 binding and regulation (Lavenus et al., 2015).

Whilst LBD16 function was closely associated with the GRN controlling early stages of LR initiation (Goh et al., 2012; Lavenus et al., 2015), the role for LBD29 had been less clear until our genetic studies employing lbd29-1, pLBD29:gLBD29:Venus rescue of lbd29-1 and gLBD29-SRDX lines revealed that LBD29 mediates auxin-regulated LR emergence (Fig. 4 and Fig. 7A). The current study has also revealed that LBD29 is essential for the auxin-inducible expression of the auxin influx carrier LAX3 (Fig. 3 and Fig. 7B) by directly binding the LAX3 promoter (Table S1; Fig. 5). Promoter deletion experiments show that a region between −570 and −363 is essential auxin induction by LAX3 (Fig. 2, between Δ3 and Δ4). In silico analysis of the LAX3 promoter sequence allowed the identification of a high number of LBD binding sites immediately downstream of this region (Fig. 5 and Table S2). Consistently, ChIP q-PCR experiments confirmed that LBD29 binds this region of the promoter, which triggers LAX3 induction in response to auxin.

During LR emergence, cell separation in tissues overlying new organs requires auxin induction of cell wall-remodeling genes, such as PG that degrades pectin, leading to cell wall breakdown (Laskowski et al., 2006). Auxin-inducible PG expression in cells overlying LRP is dependent on the auxin influx carrier LAX3 (Swarup et al., 2008). The absence of PG expression following auxin treatment in the lbd29-1 line (Fig. 3B) suggests that LBD29 controls PG expression via its regulation of LAX3. LBD18 has been reported to induce the expression of another cell wall-remodeling gene, EXPANSIN 14 (EXP14), during LR emergence (Lee et al., 2012). Hence, both LBD18 and LBD29 regulate the induction of cell wall-remodeling enzymes that promote LR emergence. However, LBD29 and LBD18 function at distinct positions in the LR emergence GRN, upstream and downstream of LAX3, respectively (Fig. 7B; Lee et al., 2014).

Delimiting LAX3 expression to two adjacent cortical and epidermal cell files overlaying new LRP is crucial for ensuring auxin-regulated cell separation occurs solely along their shared walls (Swarup et al., 2008; Fig. 1B). To understand how LAX3 spatio-temporal expression is regulated, Péret et al. (2013) developed a mathematical model that captures the network regulating its expression and auxin transport within realistic three-dimensional cell and tissue geometries. To prevent LAX3 from being transiently expressed in multiple cell files overlying LRP, the model predicted that this auxin influx carrier must be induced by auxin after an auxin efflux carrier (later demonstrated to be encoded by PIN3; Péret et al., 2013). Hence, PIN3 and LAX3 expression would need to be sequentially induced by auxin. Given that PIN3 is regulated as a primary auxin-responsive gene, LAX3 would be required to be induced as a secondary auxin-responsive gene. ARF7 regulating LAX3 auxin-inducible expression via an intermediary transcriptional factor (LBD29) would fulfill this temporal regulatory requirement. Hence, the sequential induction of LBD29 and LAX3 by auxin (Fig. 7B) represents an important regulatory motif within the LR GRN that is required to coordinate cell separation and organ emergence.

The Arabidopsis ecotype Columbia (Col-0) was used as the wild type in all experiments. The 35S:LBD29-GFP in the arf7 arf19 mutant background, arf7 (nph4-1), arf19-1, lbd29 and lax3 mutants have been previously described (Okushima et al., 2007, 2005; Feng et al., 2012; Swarup et al., 2008). The genomic LBD29-SRDX lines in Col-0 background (gLBD29-SRDX 1.3 and 2.1) were produced as described in Goh et al. (2012). Seeds were surface sterilized and plated on 0.5× MS medium (Sigma), 1% bacto-agar (Appleton Woods). Seeds were stratified at 4°C overnight and grown vertically (22°C, continuous light). Medium was supplemented either with 1 µM IAA (indole-3-acetic acid) (Sigma) or 10 µM cycloheximide (CHX) (VWR International).

RNA extractions, reverse transcription and quantitative PCR were performed as previously described (Péret et al., 2013). Primer sequences for LAX3 and PG have also been described (Péret et al., 2013).

Three-day-old seedlings grown on vertical plates were subjected to 90° gravistimulations for 18 and 42 h (Voß et al., 2015). In addition, 10-day-old seedlings grown vertically were harvested to analyze the developmental stages of LR primordia. In this phenotypic study, the total number and stages of LR primordia were counted and determined as described by Malamy and Benfey (1997). Root length was measured using ImageJ (ImageJ 1.40 g).

GUS staining and clearing was done as previously described (Péret et al., 2013). Confocal microscopy was performed using a Leica SP5 confocal laser-scanning microscope (Leica Microsystems). Cell walls were stained using propidium iodide (Sigma) (10 µg/ml) for 2 min.

The experiments were performed as described in Gaudinier et al. (2011). Interactions were called for TFs that activated at least one reporter assay.

Chromatin immunoprecipitation (ChIP) and subsequent quantitative PCR (input DNA dilution 1000×) were performed as previously described (Lavenus et al., 2015). Primers were designed to amplify 150-200 bp fragments and are listed in Table S3. Relative enrichment of the target region was normalized against TUB3 (TUBULIN BETA CHAIN 3, AT5g62700). Relative enrichments of LBD29-GFP proteins were analyzed at four regions of the LAX3 promoter. Transgenic roots of the LBD29-O^ex line were analyzed by ChIP using anti-GFP antibodies. Values were normalized to internal controls (relative to input and to TUB2). Data represent the mean±s.e. of four technical replicates, and two biological replicates were performed.

Transient expression assay was performed on protoplasts as previously described (Bielach et al., 2012). Protoplasts were co-transfected with 1 µg reporter plasmid containing the luciferase (LUC) reporter gene, 1 µg plasmid effector and 2 µg normalization construct expressing Renilla LUC gene (De Sutter et al., 2005). Firefly luciferase (fLUC) activity values were normalized with the luciferase activity derived from the internal standard plasmid coding for the Renilla luciferase (rLUC) gene under the control of 35S CaMV promoter. Both luciferase activity were measured subsequently on a Synergy H1 with double injector (Biotek). The mean value (±s.e.) was calculated from six measurements on three independent experiments.

For the luciferase assays, a genomic DNA sequence corresponding to 1374 bp upstream of the start codon of the LAX3 gene was isolated and then amplified using the forward and reverse primers 5′-ATAAATCTGCAGAGTCATGATCCTTTT-3′ and 5′-TCTTTAAATAGACCATGGAAAAGCTTTTTC-3′ containing PstI and NcoI sites, respectively, and ligated into a LucTrap vector (Lau et al., 2011) to generate luciferase fusion. The coding sequence of LBD29 was amplified to introduce HindIII and BamHI restriction sites at each extremity respectively (5′-ATCAAGCTTATGACTAGTTCCAGCT CTAGCTCT-3′ and 5′-GATGGATCCATATCACGAGAAGGAGATGTAGCC-3′) and subsequently cloned into pJIT60 vector to generate the plasmid effector (Schwechheimer et al., 1998). Full-length ARF16 cDNA was introduced to the pJIT60 vector, using BamHI and EcoRI restriction sites (5′-AAAACGGGATCCAAAAATATGATAAATGTGATGAATCCA-3′ and 5′-AAAGAATTCGCCAAGTTATACTACAACGCTCTCACT-3′). pJIT60 vector contained a double cauliflower mosaic virus 35S promoter.

Chimeric constructs were also created for the Y1H system. Genomic DNA from Col-0 was used to amplify 1422 bp of the LAX3 promoter, just upstream of the translational start codon (5′-TTCTGCTTTTTGAATATTACACCATT-3′ and 5′-TTTTCTCTTCTTCTCTCAGTTTCTTTAGC-3′) and was cloned into pENTR 5′-TOPO TA vector (Invitrogen). The correct clone was recombined with pMW2 (HIS3 reporter vector) and pMW3 (lacZ reporter vector) (Brady et al., 2011) using LR clonase II. Because the stele-expressed transcription factor collection (Gaudinier et al., 2011) did not contain LBD29 transcription factor, a construct was generated to clone the LBD29 coding sequence into the pDest-AD-2 µ plasmid. The cDNA sequence of LBD29 was amplified (5′-CACCATGACTAGTTCCAGCTC-3′ and 5′-CGAGAAGGAGATGTAGCCAAAATT-3′) and cloned into the pENTR-D-Topo entry vector (Invitrogen). The entry vector was used in a gateway LR cloning reaction (LR clonase II; Invitrogen) with pDest-AD-2 µ (Gaudinier et al., 2011) to create a GAL4-activation domain fusion Y1H prey vector.

The LAX3 promoter was cloned from pENTR11-LAX3-YFP (Swarup et al., 2008) into pBluescript KM+ (Invitrogen) using unique KpnI and SpeI restriction sites. The plasmid was PCR amplified using primers Lx3-25, 5′-TTTCTAAGAAATTAGTGGGTTAAATAAAGC-3′ and Lx3-26, 5′-AGTCTCCTTTTTAGCCCCATGCTTTTACAATGG-3′, which were designed to modify a single nucleotide within the auxin-response element (GAGACA to GAGACT). PCR amplification was carried out using Pfx proofreading DNA polymerase. Purified PCR products were digested with DpnI, treated with T4 polynucleotide kinase (NEB) and ligated with T4 ligase (NEB). Point-mutated promoters were cloned back into pENTR11-LAX3-YFP and sequenced to check no other mutations were created during the PCR.

For PCR-generated promoter deletions, a combination of primers Lx3-R2, 5′-TTCTAAGTAATTCCCTGCGACC-3′ and (KpnI)-Lx3-22, 5′-CCGGTACCTTTCTAAGAAATTAGTGGGTTA-3′ for Δ2 and (KpnI)-Lx3-23, 5′-CCGGTACCAATATGTTTTATTCATTGTTTC-3′ for Δ4 were used. PCR amplification was carried out using Pfx proofreading DNA polymerase. Purified PCR products were digested with DpnI and cloned into pENTR11-LAX3-YFP using KpnI and SpeI restriction sites.

For deletions generated using restriction enzymes, KpnI and MunI for Δ1 or only BamHI for Δ3 were used. The correct band was gel purified, if necessary 3′ overhang was filled using T4 DNA polymerase (NEB) and both fragments ligated using T4 ligase. Constructs were then cloned in the binary pGWB7 vector using the Gateway LR reaction (Invitrogen).

We acknowledge the support of glasshouse technicians at the University of Nottingham for help with plant growth and the Nottingham Arabidopsis Stock Centre (NASC) for providing Arabidopsis lines.