Identification of new adventitious rooting mutants amongst suppressors of the Arabidopsis thaliana superroot2 mutation

Summary Auxin is a major regulator of adventitious rooting and, to better understand its role, we identified suppressor mutants of superroot2-1. This provides new resources for the discovery of genetic players involved in auxin signalling or auxin crosstalk with other hormones.


Introduction
Adventitious root (AR) formation defines the process whereby roots develop at any location other than from a root. In horticulture and forestry in which asexual plant propagation is widely used, AR formation from cuttings is a necessary step. Adventitious rooting is a complex organogenic process controlled by multiple endogenous and environmental factors (Geiss et al., 2009;da Costa et al., 2013), among which the plant hormone auxin plays a central role. Auxin has long been known as the rooting hormone (Haissig and Davis, 1994) and is used to promote development of AR in stem cuttings, but other phytohormones can also promote or accelerate rooting and are often used in association with auxin treatments (Kevers et al., 2009). In Arabidopsis, mutants overproducing auxin, such as the superroot mutants sur1 and sur2 or the yucca mutants, spontaneously develop AR in the hypocotyl when grown under light (Boerjan et al., 1995;Delarue et al., 1998;Zhao et al., 2001). Although auxin plays a central role in the determination of rooting capacity, its use is not always effective in promoting rooting of recalcitrant genotypes (Fogaça and Fett-Neto, 2005), highlighting the fact that in this developmental process, as in others, auxin interacts either with other endogenous factors or environmental stimuli such as light (Fett-Neto et al., 2001;Sorin et al., 2005).
Besides auxin, ethylene can also promote adventitious rooting. The ethylene-insensitive never ripe tomato mutant developed fewer AR than the wild type (Clark et al., 1999). More recently it was demonstrated that the positive role of ethylene in AR formation is likely due to a modulation of auxin transport, which is a central point of ethylene-auxin crosstalk (Negi et al., 2010). Ethylene could also promote rooting by stimulating cytokinin catabolism (Kuroha and Sakakibara, 2007). Cytokinin has been shown to act as a rooting inhibitor (Geiss et al., 2009;Ramirez-Carvajal et al., 2009). Therefore, downregulation of the endogenous cytokinin content via catabolism or repression of the cytokinin signalling pathway promotes adventitious rooting. Abscisic acid (ABA), like cytokinins, appears to be a negative regulator of adventitious rooting. The ABA-deficient tomato mutants flacca and notabilis produce an excess of AR on the stems (Tal, 1966). The AR phenotype of the notabilis mutant could be restored to wild-type phenotype by expressing SpNCED1, involved in ABA biosynthesis (Thompson et al., 2004). Niu et al. (2013) recently demonstrated that a high concentration of endogenous gibberellins in the stem of tobacco plants had an inhibitory effect on the early steps of adventitious root initiation in tobacco cuttings. In addition, it was shown that adventitious rooting is regulated by the stress-related hormone jasmonic acid (JA; Ahkami et al., 2009;Fattorini et al., 2009;Gutierrez et al., 2012). Strigolactones, a new class of phytohormones, were recently shown to have an inhibitory effect on adventitious rooting both in Arabidopsis and pea (Rasmussen et al., 2012). They might act by modulating the auxin level in the cells or tissues from which AR originate. Among endogenous compounds, it has also been reported that alkamides and nitric oxide (Pagnussat et al., 2003;Pagnussat et al., 2004;Campos-Cuevas et al., 2008;Mendez-Bravo et al., 2010), polyamines (Rey et al., 1994;Heloir et al., 1996), and flavonoids (Curir et al., 1990) are important players in the regulation of adventitious rooting.
Although these studies clearly indicate the importance of hormone homeostasis and hormone crosstalk, they do not reveal specific molecular mechanisms and additional investigation is needed to better understand the mechanism of auxin in the control of AR formation. In rice (Oryza sativa), CROWN ROOTLESS1 (CRL1)/ADVENTITIOUS ROOTLESS1 (ARL1) (Inukai et al., 2005;Liu et al., 2005) encodes a member of the plant-specific ASYMMETRIC LEAVES2/LATERAL ORGAN BOUNDARIES (LOB) protein family and has been shown to control AR development. Disruption of CRL1/ARL1 prevents initiation of adventitious crown root primordia in rice. CRL1/ARL1 is an auxin-responsive gene that contains two putative auxinresponse elements (AREs) in its promoter region. Its induction by auxin requires the degradation of Aux/IAA proteins and it was shown that the proximal AREs specifically interacted with a rice AUXIN RESPONSE FACTOR (ARF) and acted as a cis-motif for CRL1 expression (Inukai et al., 2005). CRL1/ARL1 can be considered as a positive regulator for crown root formation in rice. Similarly, the current study group has shown that AR initiation in Arabidopsis hypocotyls is controlled by a subtle balance between the negative regulator ARF17 and the positive regulators ARF6 and ARF8, which display overlapping expression domains, interact genetically, and regulate each other's expression at both transcriptional and post-transcriptional levels . More recently, this study group has shown that, in contrast to ARF17, ARF6 and ARF8 are positive regulators of the auxin-inducible genes GH3-3, GH3-5, and GH3-6 that fine-tune AR initiation in Arabidopsis hypocotyls by modulating JA homeostasis . These findings highlight a regulatory module at the crosstalk between jasmonate and auxin signalling pathways.
Genetic screens for suppressors have often been used to further investigate gene functions and to dissect signal transduction pathways. In Arabidopsis, suppressor screens have been used to identify genes/mutants that function in, for example, the auxin (Cernac et al., 1997), gibberellin (Jacobsen and Olszewski, 1993;Wilson and Somerville, 1995), and abscisic acid (Koorneef et al., 1982) pathways. The mutation sur2-1 causes a loss of function of the cytochrome P450 CYP83B1, and the mutant spontaneously produces an excess of AR (Delarue et al., 1998).
In order to identify more genes that might control the initiation or development of AR, the current work used the sur2-1 AR trait and performed a screen for suppressor mutants that produced fewer AR than sur2-1. This work isolated 46 such mutants representing 34 groups of complementation. This report describes mapping of 32 of these mutants and the phenotypic characterization of independent sur2-1 suppressors covering 26 complementation groups.

Materials and methods
Identification of sur2-1 suppressors In order to avoid selecting wild-type seedlings due to potential contamination of the mutagenized population with wild-type seeds, the glabra1 mutation was introgressed in the sur2-1 mutant background. The glabra1 mutant was identified in the Versailles collection of T-DNA insertion lines and was therefore in the same genetic background as sur2-1. Homozygote seeds from the doublemutant superroot2-1gl1 (sur2-1gl1) (ecotype Wassilewskija, Ws-4) were mutagenized with ethyl methanesulphonate, as described in Santoni et al. (1994). The M2 progeny of 2345 independent lines was collected and screened as follows: seeds were surface sterilized, sown in vitro, stratified for 48 h at 4 °C, then transferred to a plant growth chamber but kept in the dark until the hypocotyls reached sizes of 5-6 mm. They were then transferred to the light for 7 d, as described previously (Sorin et al., 2005). The conditions in the controlledenvironment chambers were as follows: 130 μE m -2 s -1 irradiance on average, 16/8 light/dark cycle, 22/15 °C. 60% relative humidity. For dark growth conditions, Petri dishes were wrapped with three layers of aluminium foil and placed vertically. After 7 d in the light, the AR were counted on the hypocotyls. Seedlings that produced fewer AR than sur2-1gl1 were transferred to soil and grown to set seeds. The M3 was then rescreened to confirm the suppressor phenotype in the progeny. A schematic of the screening procedure is presented on Supplementary Fig. S1 available at JXB online.

Genetic segregation and complementation analysis
The suppressors were backcrossed twice with the original sur2-1gl1 line used as the female parent, except for the male sterile suppressor 677/2191, where sur2-1gl1 was used as male parent. Segregation analysis was performed in the progeny of the second backcross (Supplementary Table S1). Suppressor mutants showing similar phenotypes that were identified early in the screening process were crossed to each other for complementation analysis. Once the coarse mapping information was available, only those mutants that were mapped in the same region were crossed for complementation analyses.
Mapping of sur2-1 suppressors To map the suppressor mutations (Ws-4 background), phenotyped mutant seedlings that yielded fewer AR than sur2-1gl1 were identified in an F2 population obtained by crossing the suppressor mutant with atr4-1, an allele of the sur2 mutant in a Columbia-0 (Col-0) background (Smolen and Bender, 2002). Using standard protocols, genomic DNA was extracted from entire mutant seedlings grown in vitro as previously described by (Sorin et al., 2005) and used as template for mapping as described by Pacurar et al. (2012).

Hypocotyl and root measurements
Hypocotyl and root measurements were performed as described by Gendreau et al. (1997). Seedlings were grown vertically in Petri dishes. The plates were placed directly in the light (without prior etiolation) for 7 d. Plates were photographed and hypocotyls and roots measured using ImageJ software (http://rsb.info.nih.gov/ij/ index.html). All measurements were performed on three independent biological replicates with a minimum of 40 seedlings each.

Counting of adventitious and lateral root number
ARs were counted on hypocotyls of etiolated seedlings, following the same protocol described above for the screening. Lateral roots were counted on light grown seedlings used for the hypocotyl and root measurements. All counting was performed on three independent biological replicates with a minimum of 40 seedlings each.

Determination of free IAA concentrations
The screen and characterization of the suppressor mutants was performed over a long period of time. Therefore, the endogenous free IAA content was measured in two periods (2008 and 2011) using different methods. Seedlings were grown in vitro, in controlled diurnal environment with a 16/8 light/dark cycle (average irradiance 130 μE m -2 s -1 ) and 60% relative humidity. The root was removed and the aerial parts were pooled to obtain an average of 15 mg fresh weight per sample. Samples were extracted, purified and analysed by liquid chromatography multiple-reaction monitoring mass spectrometry (LC-MRM-MS), as described by Kowalczyk and Sandberg (2001) or by gas chromatography selected-reaction monitoring mass spectrometry (GC-MRM-MS) as described by Andersen et al. (2008). In the two experiments, measurements were performed with three independent biological replicates.

RNA isolation and cDNA synthesis
RNAs from wild type and suppressor mutants were prepared as previously described . RNA (5 μg) were treated with DNaseI using a DNAfree Kit (Ambion) and cDNA was synthesized as described by Gutierrez et al. (2012). All cDNA samples were tested by PCR using specific primers flanking an intron sequence to confirm the absence of genomic DNA contamination.

Real-time PCR experiments and data analysis
Transcript abundance was assessed by quantitative real-time (RT) PCR according to this study group's previously described procedure . All quantifications were repeated with three independent biological replicates.
Steady-state levels of uncleaved ARF transcripts were quantified using primers spanning the miRNA target site. The following standard protocol was applied for the amplification of each mRNA: 10 min at 95 °C, followed by 40 cycles at 95 °C for 10 s, 60 °C for 15 s (except for GH3.5 for which the annealing temperature was 65 °C), and 72 °C for 15 s. The sequences of primers used for all target genes are presented in Supplementary Table S2. APT1 and TIP41 had previously been validated as the most stably expressed genes of the 11 tested  and were used to normalize the RT-PCR data. The normalized expression patterns obtained using both reference genes were similar, so only the data normalized with APT1 are shown. CT and E values were used to calculate expression using the formula E T where Cal and M are subscripts to the superscript CT. where T is the target gene and R the reference gene, CT is the crossing threshold value, M refers to cDNA from the mutant line, and Cal refers to cDNA from the calibrator. The data for the suppressor mutants are presented as relative to the calibrator (i.e. the wild type or sur2-1gl1).
All RT-PCR results presented are means from three independent biological replicates. For each independent biological replicate, the relative transcript amount was calculated as the mean of three technical replicates, using the method for calculation of standard errors in relative quantification recommended by Rieu and Powers (2009).

Statistical analysis
Statistical analysis was performed using the GraphPad Prism version 6.0d for Mac (www.graphpad.com). One-way ANOVA combined with the Tukey's multiple comparison post-test was used for multiple means' comparisons. Linear regression and Pearson correlation coefficient r were calculated using the GraphPad package.

Suppressor mutant isolation and genetic characterization
A total of 2345 individually harvested M2 seed stocks derived from ethyl methanesulphonate-mutagenized sur2-1gl1 double-mutant plants were screened for a phenotype showing fewer adventitious roots than sur2-1gl1. In order to limit finding mutants altered in the expression of genes already known to be involved in the control of root development and/ or auxin signalling, suppressor mutants that had a shorter primary root or fewer or no lateral roots were eliminated as much as possible.
A total of 46 mutants showing fewer AR than sur2-1gl1 double mutants were identified, and the suppressor phenotype was confirmed in the M3 generation and after at least two backcrosses with sur2-1gl1, as described in Supplementary Fig. S1. Segregation analysis was performed for 33 of these mutants (Supplementary Table  S1). Of the analysed mutants, all but one (no. 1626) were recessive. In some cases, the proportion of plants exhibiting the suppressor phenotype was lower (P<0.05) than the expected 3:1 sur2-1gl1:mutant segregation (Supplementary  Table S1). No embryo lethality or germination problems were observed, suggesting that these suppressors could be double mutants. In two cases (nos. 266 and 2037), several alleles were identified ruling out this hypothesis, at least for these two mutants.
Considering the number of mutants, in the first instance only those that looked alike were crossed to determine complementation groups. In parallel, the mutations were mapped on the chromosomes using the strategy and the InDel markers reported previously (Pacurar et al., 2012). Mutants that were located in the same region were crossed to check for potential allelism. The 46 mutants represented 34 groups of complementation (Supplementary Table S1). Most mutants were single-allele mutants. One particular gene (no. 266) showed a very high mutation rate with eight alleles, while others had two (nos. 420, 677, and 1583) or three (no. 2037) alleles (Fig. 1). These results indicate that, although a rather small number of mutagenized lines were analysed, saturation of the genome was likely reached. Surprisingly, no linkage with any of the markers tested could be found for two mutations (nos. 1738 and 1788). This cannot be explained by a segregation defect since they both clearly segregated as single-locus recessive mutations (Supplementary Table S1) and the phenotype (although mild for no. 1788; Fig. 2A) was easily detectable in the mapping population. The difficulty in positioning these mutations suggests that they might be located in regions with hot spots of recombination.
Besides mutations in genes likely to be required for auxin biosynthesis, this work identified mutants in signalling and/ or regulatory genes. Two allelic mutants (nos. 677, 2191) of ABERRANT LATERAL ROOT FORMATION 4/ALF4 were identified (Fig. 1, Supplementary Fig. S2). ALF4 encodes a nuclear-localized protein of unknown function that is necessary for lateral root formation (DiDonato et al., 2004;Dubrovsky et al., 2008). The two alleles identified had a phenotype very similar to that reported for alf4-1 (i.e. sterile, bushy plantlets that are unable to form lateral or adventitious roots; Celenza et al., 1995). Both alleles were mutated at the exact same nucleotide (Supplementary Fig. S2). This is a rare event but not impossible since it had already been shown for other mutants' alleles (Busch et al., 1996).
This study identified mutations in three genes involved in auxin signalling (Fig. 1, Supplementary Fig. S2, Supplementary Table S1). These were no. 2244, an allele of auxin response 1 (axr1), no. 1626, a gain-of-function allele of short hypocotyl 2 / iaa3 (shy2), and two alleles of rub-conjugating enzyme 1 (rce1) (nos. 420 and 1375.2). AXR1 encodes a subunit of a heterodimeric RUB-activating enzyme (del Pozo et al., 1998), SHY2/IAA3 encodes an Aux/IAA protein (Tian and Reed, 1999), and RCE1 is involved in the regulation of auxin signalling and ethylene biosynthesis (Dharmasiri et al., 2003;Larsen and Cancel, 2004;Pacurar et al., 2012). For the remaining suppressors, either no other candidate gene was clearly identified or the mapped region was still too large to choose a candidate for sequencing. With the optimization of next-generation sequencing, it will become easier to obtain a full genome sequence in order to identify the mutation.

Phenotypic characterization of the suppressor mutants
After two backcrosses with sur2-1gl1, a more detailed phenotypic characterization of 26 independent suppressor mutants was performed. They were characterized at the seedling level (Figs 2 and 3) and in soil (Fig. 4). sur2-1gl1 had on average 14 AR and the wild type an average of 1.5 AR. The average number of AR for most of the suppressors lay between these two values ( Fig. 2A). For a few of the mutants (nos. 1626/shy2, 1844, 2035, 2244/axr1, and 2249) the average number of AR was not significantly different from that of the wild type ( Fig. 2A), and in one case (no. 2125) the average number of AR was significantly lower than in the wild type. Suppressor 2125 also had a short primary root and almost no emerged lateral roots (Figs 2B-D and 3), suggesting that the mutation alters general root development. Nevertheless, it retained some sur2-1gl1 features such as a longer hypocotyl than the Ws-4 wild type (Figs 2E and 3B) and very small rosette with leaves bearing long petioles (Fig. 4). This suppressor is representative of suppressors that were not studied further because of their strongly altered root development.
Another type of suppressor is represented by no. 2249, which has an average number of AR closer to that of Ws-4 ( Fig. 2A), but a root system closer to that of sur2-1gl1 in terms of primary root length and lateral root density (Fig. 2B-D).
No. 2249 also had a sur2gl1-like hypocotyl length, petioles, and epinastic cotyledons (Fig. 3A, B). Although AR and lateral roots emerged, they grew very slowly or stopped elongating (Fig. 3A, B), suggesting that no. 2249 also impacted the general root development. The mutants that are of most interest for the future are those which retain a reasonably strong root system but for which a significant reduction in the average number of AR is observed, such as nos. 1738, 1745, 1788, 1848, 1932, 1977, 2101, 2146.4, 2310, and 2344. These mutants had a hypocotyl length in between that of sur2-1gl1 and the wild type (Fig. 2E), and their phenotype in soil varied between that of wild type and sur2-1gl1 (Fig. 4).

sur2-1 suppressor mutants are not systematically reduced in free IAA content
Because sur2-1 is an auxin overproducer, this work checked the level of endogenous free IAA in the aerial part of 9-d-old seedlings of several representative suppressors grown directly in light, as described by Sorin et al. (2005). The measurements were performed at two different periods using Ws-4  Fig. 1. Chromosomal localization of sur2-1 suppressor mutations. From left to right: chromosomes 1, 2, 3, 4, and 5. Black indicates chromosomes; horizontal white bars indicate the positions of identified suppressor genes, whose names are indicated in boxes with arrows; numbers in brackets indicate alleles. Light and dark grey boxes indicate regions containing the mutations (numbers on the right of the chromosomes) that still need to be identified, each flanked by two indel markers (identified on the left of the chromosomes). and sur2-1gl1 as controls (Fig. 5A, B). In the second experiment, the difference in IAA content between these two genotypes (Fig. 5B) was not as high as in the one observed in the first experiment (Fig. 5A), probably because seedlings used for Fig. 5B were not exactly at the same development stage as those used for Fig. 5A. This study group has previously shown that the endogenous content of IAA in the aerial part of sur2-1 has a tendency to decrease after several days in the  light (Barlier et al., 2000). Although minimized, the difference between sur2-1gl1 and wild type was still significant and allowed discrimination of the suppressor mutants. In the majority of the suppressors analysed, the endogenous level of auxin was not significantly different from that of the original sur2-1gl1 line (Fig. 5). In two cases (nos. 266/asa1/wei2 and 1745), the free IAA content was similar to wild-type level (Fig. 5A, B) suggesting that the corresponding mutations could alter expression of genes involved in or controlling IAA biosynthesis. As stated above, no. 266 is allelic to asa1/wei2 and is likely to be affected in tryptophan, and thus IAA, biosynthesis. Mutant 1745 remains to be identified, and since no obvious candidate gene is present in the mapped region it is expected that a new player in the regulation of auxin metabolism will be identified. For no. 1626/shy2, the auxin content was higher than in sur2-1gl1 (Fig. 5B). No. 1626 is a dominant mutation affecting expression of the SHY2/IAA3 gene (Tian and Reed, 1999), which acts as a negative regulator of auxin signalling (Tian et al., 2002); therefore, the increased endogenous auxin level is likely to be explained by a feedback regulation due to the fact that the mutant cannot sense auxin. Previous work has shown that AR development in Arabidopsis hypocotyl does not necessarily correlate with the endogenous auxin content (Sorin et al., 2006;Gutierrez et al., 2012). The current work shows that sur2-1 suppressor mutants retaining wild-type level of IAA (nos. 266/asa1/wei2 and 1745) still produce more AR than the wild type, whereas others with an IAA level similar to sur2-gl1 (e.g. nos. 420/rce1, 494, 1977, and 2035) produce  significantly fewer AR than sur2-1gl1. These results corroborate the fact that the number of AR developed in the Arabidopsis hypocotyl is not necessarily correlated to the endogenous content of auxin (Fig. 5C).

A role for ethylene in adventitious root initiation in Arabidopsis hypocotyl
Three of the genes identified are involved in auxin biosynthesis (nos. 266/ASA1/WEI2 and 1583/ASB1/WEI7; Stepanova et al., 2005) or auxin signalling (no. 420/RCE1; del Pozo and Estelle, 1999;Dharmasiri et al., 2003). These are also linked to ethylene signalling or biosynthesis since the wei2 and wei7 mutants have been described as weak ethylene-insensitive mutants (Stepanova et al., 2005) and it was demonstrated that the RCE1 gene is required for a proper regulation of ethylene biosynthesis (Larsen and Cancel, 2004). The rce1-2 allele was shown to overproduce ethylene and displayed the characteristic ethylene-related tripleresponse phenotype when grown in the dark (Larsen and Cancel, 2004). Because ethylene was shown to play a role in the regulation of AR formation in other species (da Costa et al., 2013), these suppressors were characterized further. When grown in the dark, seedlings of nos. 266/asa1/wei2, 1583/asb1/wei7, and 420/rce1 showed a shorter hypocotyl than the wild type (Fig. 6A, B) and a reduced apical hook for nos. 266/asa1/wei2 and 1583/asb1/wei7 or an accentuated apical hook in the case of no. 420/rce1 ( Fig. 6A; Pacurar et al., 2012). Indeed, these are the characteristic phenotypes of asa1/wei2, asb1/wei7, and rce1 mutants. The suppressor mutant 494 had a phenotype similar to nos. 266/asa1/wei2 or 1583/asb1/wei7 when grown in the dark but had no apical hook 5 d after germination (Fig. 6), suggesting that it could also be affected in ethylene response or biosynthesis.
To discriminate between potential ethylene insensitivity or modification of ethylene biosynthesis, this work quantified the expression of the ACO gene, which encodes the 1-aminoacyclopropane-1-carboxylic acid (ACC) oxidase. It has previously been shown that ethylene biosynthesis was correlated with ACO activity (Abeles et al., 1992;Larsen and Cancel, 2004) and it was anticipated that overexpression or downregulation of ACO1 would result in an up-or downregulation of ethylene production. The relative transcript abundance of ACO1 was unchanged in the sur2-1gl1 mutant compared to the wild-type level (Fig. 7A), whereas it was significantly increased in no. 420/rce1 and significantly reduced in no. 494. This coincides with the wild-type-like apical hook phenotype of sur2-1gl1 and the increased or reduced apical hook of nos. 420/rce1 and 494, respectively. Because ethylene was shown to promote AR development in rice and tomato (Clark et al., 1999)m a decrease in ethylene biosynthesis could explain the reduced number of AR in no. 494. However, no. 420/rce1, which in contrast is likely to overproduce ethylene, when compared to no. 494, has a similar reduction in average AR number. In addition, it is interesting to note that the relative abundance of the RCE1 transcript is unchanged in sur2-1gl1 and nos. 266/asa1/wei2 and 1583/asb1/wei7 mutants compared to wild-type level, but is significantly reduced in no. 494 (Fig. 7B). This was unexpected, considering that no. 494 had an opposite ethylene-related apical hook phenotype compared to no. 420/rce1. Nevertheless, it has been reported that ethylene could have differential effect on adventitious rooting, depending on the stage of the induction phase (da Costa et al., 2013). Therefore, although it seems unlikely that the phenotypes of nos. 420/rce1 and 494 mutants are due to modification of endogenous ethylene level, the current work cannot exclude that, in no. 494, a reduced ethylene production has a negative effect on the early events of AR induction phase, whereas an excess of ethylene in no. 420/rce1 could inhibit a later stage during the induction phase. Future identification of the no. 494 mutation and further characterization of these suppressor mutations both in the sur2-1gl1 and in the wild-type background could help understanding of how ethylene and auxin interact in the control of adventitious rooting. sur2-1 suppressor mutations differentially affect the ARF/GH3 regulatory module In the past few years, studies have shown that a regulatory module, composed of three AUXIN RESPONSE FACTOR (ARF) genes (ARF6, ARF8, and ARF17) and three auxinresponsive genes (GH3.3, GH3.5, and GH3.6), controls adventitious rooting in Arabidopsis hypocotyls . ARF proteins bind AREs present in the promoters of auxin-inducible genes such as Aux/IAA, SAUR, and GH3 and either repress or activate their transcription (Takase et al., 2003;Tiwari et al., 2003). Whereas ARF6 and ARF8 positively regulate GH3.3, 3.5, and 3.6 expression and thereby adventitious rooting, ARF17 is a negative regulator. The average number of AR positively correlates with the abundance of GH3.3, 3.5, and 3.6 transcripts and proteins (Sorin et al., 2006;Gutierrez et al., 2012). Thus, the current work checked whether nos. 266/ASA1/WEI2, 420/RCE1, 1583/ASB1/WEI7, and 494 mutations had an impact on the ARF/GH3 regulatory module. The relative transcript amount of the three ARF and three GH3 genes were quantified in these suppressor mutants and compared to sur2-1gl1 (Fig. 7C) and wild-type levels (Fig. 7D). In sur2-1gl1, the relative transcript amounts of the three GH3 genes were significantly higher than in the wild type, although those of ARF6, 8, and 17 were unchanged (Fig. 7D). This suggests that the induction by auxin of the expression of the three GH3 genes is not due to a modification of the ARF transcript level. ARFs proteins have a conserved N-terminal DNA-binding domain that binds to consensus AREs present in the promoter of auxin-inducible genes. In most cases, ARFs also contain a conserved C-terminal dimerization domain required for the heterodimerization with Aux/IAA proteins that repress the transcriptional activity of ARF proteins. The presence of auxin triggers the degradation, via the 26S ubiquitine-proteasome pathway, of Aux/IAA transcriptional repressors (reviewed in Chapman and Estelle, 2009). This facilitates the expression of auxin-inducible genes such as GH3 genes. Since ARF6, ARF8, and ARF17 transcript levels are not affected in sur2-1gl1, the overexpression of GH3 genes is likely due to the auxin-induced degradation of the Aux/IAA partners of ARF6 and ARF8.
In the four suppressors analysed, the reduced number of AR coincided with a significant decrease of the three GH3 genes compared to sur2-1gl1 (Fig. 7C). In nos. 266/asa1/wei2 and 1583/asb1/wei7, this could be explained by the reduction of the endogenous auxin content in the hypocotyl (Fig. 5A), but this is unlikely to be the only explanation. Indeed the relative transcript amount of the GH3 genes in nos. 266/asa1/wei2 and 1583, although significantly lower than in sur2-gl1, was still higher than in the wild type (Fig. 7D), which may explain the fact that these two mutants still have more AR than the wild type. In addition, in nos. 420/rce1 and 494 mutants, the relative expression of the GH3 genes is much lower than in sur2-1gl1, despite the endogenous auxin level remaining unchanged compared to sur2-1gl1 (Fig. 5A). In fact, the GH3.3, GH3.5, and GH3.6 relative transcript amounts in the suppressor mutants analysed did not correlate with the endogenous auxin content (Fig. 8A) but positively correlated with the number of adventitious roots (Fig. 8B). Since the endogenous IAA content cannot explain the reduction of the relative amount of GH3 transcripts in the suppressor mutants, this reduction is likely due to the downregulation of the regulatory ARF expression. In nos. 266/asa1/wei2 and 1583/asb1/wei7, both positive regulators (ARF6, ARF8), and the negative regulator (ARF17) are downregulated, whereas in nos. 420/rce1 and 494, the relative transcript amount of ARF17 remains unchanged (Fig. 7C, D). In these latter cases, the balance of transcriptional regulators is strongly in favour of the repressor, which according to previous data  is likely repressing GH3 gene expression despite the elevated level of auxin. The initial phenotypic characterization of these four suppressors suggests that the decreased number of AR could be due to perturbation of ethylene-auxin crosstalk, interpretation also supported by the recently shown coregulation of both ethylene and auxin biosynthesis by the enzyme VAS1 encoding pyridoxal-phosphate-dependent aminotransferase (Zheng et al., 2013). However, characterization of the expression of genes involved the core ARF/GH3 regulatory module shows that modification of the endogenous levels of auxin or ethylene is not directly responsible for the suppressor phenotypes. Besides the AR phenotype, the common feature between these four suppressors is the modification of the balance between the three ARF and three GH3 genes. The future identification of the gene mutated in no. 494 is likely to provide more clues on the molecular interactions involved in this pathway.

Concluding remarks
From numerous studies addressing different aspects of AR formation in a variety of species, it is obvious that adventitious rooting is controlled by many genes and exhibits considerable phenotypic plasticity. Genomics, transcriptomics, and proteomics studies performed with several species have started to unveil its complex genomic control and interactions (Kohler et al., 2003;Brinker et al., 2004;Sorin et al., 2006;Ramirez-Carvajal et al., 2009;Muthreich et al., 2013;Thirunavukkarasu et al., 2013;Wei et al., 2013). The current contribution performed a screen to identify new players involved in AR formation in the model plant A. thaliana. The findings confirm the multitude and variety of genetic controls of AR formation by identifying to date suppressor genes involved in auxin biosynthesis (ASA1/WEI2, ASB1/WEI7, TSB1), auxin signalling (SHY2, AXR1, RCE1), and lateral root formation (ALF4), which can also regulate AR formation. Numerous suppressor mutants remain to be identified and since no obvious candidate gene is present in the mapped regions, this study group expects to identify new regulators of the intricate network controlling AR formation. These are likely to act at the crosstalk of auxin and other signalling pathways.