AUXIN UP-REGULATED F-BOX PROTEIN 1 REGULATES THE CROSSTALK BETWEEN AUXIN TRANSPORT AND CYTOKININ SIGNALING DURING PLANT ROOT GROWTH 1

Plant root development is mediated by the concerted action of the auxin and cytokinin phytohormones, with cytokinin serving as an antagonist of auxin transport. Here, we identify the AUXIN UP-REGULATED F-BOX protein (AUF) 1, and its potential paralog AUF2, as important positive modifiers of root elongation that tether auxin movements to cytokinin signaling. The AUF1 mRNA level in roots is strongly up-regulated by auxin but not by other phytohormones. Whereas the auf1 single and auf1 auf2 double mutant roots grow normally without exogenous auxin and respond similar to wild type upon auxin application, their growth is hypersensitive to auxin transport inhibitors with the mutant roots also having reduced basipetal and acropetal auxin transport. The effects of auf1 on auxin movements may be mediated in part by mis-expression of several PIN-FORMED (PIN) auxin efflux proteins, which for PIN2 reduces its abundance on the plasma membrane of root cells. auf1 roots are also hypersensitive to cytokinin and have increased expression of several components of cytokinin signaling. Kinematic analyses of root growth and localization of the cyclin-B mitotic marker showed that AUF1 does not affect root cell division but promotes cytokinin-mediated cell expansion in the elongation/differentiation zone. Epistasis analyses implicate the cytokinin regulator ARR1 or its effector(s) as the target of the SCF ubiquitin ligases assembled with AUF1/2. Given the wide distribution of AUF1/2-type proteins among land plants, we propose that SCF AUF1/2 provides additional crosstalk between auxin and cytokinin, which modifies auxin distributions and ultimately root elongation. B , Immunoblot detection of ARR1. Crude extracts were prepared from roots harvested from 10-d-old seedlings and subjected to SDS-PAGE and immunoblot analysis with anti-ARR1 antibodies. Duplicate loads are shown to account for lane to lane variation. Equal protein loading was confirmed by immunoblot analysis with antibodies against the 26S proteasome subunit RPT1a. crosstalk between cytokinin and auxin transport.


INTRODUCTION
negatively regulate cytokinin signaling by binding to and interfering with type-B ARRs. The transient transcriptional induction of type-A ARRs by type-B ARRs serves to dampen cytokinin responses by negative feedback (To et al., 2007;To and Kieber, 2008).
To maintain the correct balance between meristem maintenance and cell differentiation, an auxin/cytokinin crosstalk is used to adjust the influence of these two hormones, especially with respect to root and shoot identity. Auxin regulates the size of the RM by promoting cell division, whereas cytokinin acts in the transitional region overlapping between the distil RM zone and proximal elongation/differentiation (EDZ) zone to promote root cell elongation/differentiation (Blilou et al., 2005;Dello Ioio et al., 2007). The connection between these two competing processes is primarily mediated by SHORT HYPOCOTYL (SHY)-2, an AUX/IAA transcriptional repressor whose expression is activated by cytokinin through the AHK3/ARR1 pathway (Dello Ioio et al., 2008). Increased SHY2 down-regulates the expression of multiple PIN proteins in the root, thus limiting the formation of auxin maxima and subsequent cell divisions. By contrast, auxin promotes SHY2 degradation through the SCF TIR1/AFB1-3 ubiquitylation machinery, thus relieving SHY2 repression on auxin redistributions. SHY2 also attenuates cytokinin synthesis to provide a second feedback loop (Dello Ioio et al., 2008).
Through these flexible interconnected circuits, auxin and cytokinin are delicately balanced to antagonistically regulate root cell development and organogenesis.
Here, we describe a second FBX type encoded by the AUXIN UP-REGULATED FBX (AUF)-1 and possibly the AUF2 loci within the Arabidopsis UPS that connects auxin and cytokinin during root development. As the name implies, AUF1 was first noticed by the strong increase in its mRNA level upon treating seedlings with auxin. auf1 loss-of-function mutants have normal responsiveness to exogenous auxin but are hypersensitive to the auxin transport inhibitors 1-naphthylphthalamic acid (NPA) and 2,3,5-triiodobenzoic acid (TIBA) with respect to root elongation, and have reduced rates of acropetal and basipetal auxin transport in roots. The www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. transcript abundance for several PIN genes is altered in homozygous auf1 plants, which, at least for PIN2, decreases the accumulation of this efflux facilitator on root cell plasma membranes. auf1 root elongation is also hypersensitive to exogenous cytokinin with auf1 roots expressing higher levels of the type-A response regulators ARR5 and ARR15 in response to the hormone. Kinematic analyses pinpointed the root elongation defect to the zone of rapid expansion in the EDZ. Cytokinin-treated auf1 cells exited this zone earlier than wild type.
Given the widespread distribution of AUF1/2 proteins among land plants, the SCF complexes assembled with these FBX proteins likely target a conserved positive effector in the cross-talk between auxin and cytokinin that regulates auxin movements and ultimately root elongation.

Genomic Analysis of FBX Gene Pair Potentially Regulated by Auxin
During our attempts to define the functions of the nearly 900 FBX loci in Arabidopsis by various "omic" approaches (Gagne et al., 2002;Hua et al., 2011), we noticed that the expression of one FBX gene designated AUF1 (At1g78100 in the C1 subclade) was shown in the Genevestigator (https://www.genevestigator.ethz.ch (Hruz et al., 2008)) and eFP DNA microarray databrowsers (www.Arabidopsis.org) to be strongly up-regulated by the natural auxin IAA but not by several other phytohormones or their precursors. The increase in AUF1 mRNA abundance ranged between 5 and 8 fold when whole seedlings were treated to 1 μ M IAA ( Figure 1A). Further analyses of the microarray datasets revealed that AUF1 is expressed in most tissues with the root-specific datasets in particular revealing high expression in the maturing cortical and epidermal cell files (Brady et al., 2007). To confirm the auxin upregulation, we exposed Arabidopsis seedlings to 0.1 μ M IAA for 5 hr and then subjected root total RNA to RNA gel-blot analysis with an AUF1-gene specific probe. A strong increase in the expected 1-kb AUF1 transcript was evident, suggesting that the corresponding protein regulates auxin-dependent processes ( Figure 1B).
Sequence searches of the Arabidopsis thaliana Col-0 genome revealed that AUF1 has an obvious paralog designated AUF2 (At1g22220). It is also on chromosome 1 but on the opposite side of the centromere relative to AUF1. Representative AUF2 transcripts are available in the Arabidopsis Expressed Sequence Tag (EST) database (www.Arabidopsis.org/), and we could generate a sequence-confirmed AUF2 cDNA by reverse transcribed (RT)-PCR, indicating that the locus is expressed. Whereas 105 ESTs have been reported for AUF1, only 9 have been reported for AUF2, suggesting that AUF2 is expressed at considerably lower levels than AUF1. Other details about AUF2 expression patterns are not yet known, mainly because www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. of its omission from the Affymetrics ATH1 DNA microarrays commonly used to analyze the Arabidopsis transcriptome (www.Arabidopsis.org/). However, quantitative real time (q) RT-PCR showed that AUF2 expression is not affected by IAA ( Figure 1C). This is in contrast to the strong and transient up-regulation of AUF1 expression which peaks between 2 and 5 hrs after IAA exposure and declines back to normal levels after 10 hr.
Transcripts from both AUF1 and AUF2 were predicted to be intronless, which was subsequently confirmed by DNA sequence analysis of cDNAs generated from seedling mRNA by reverse transcribed (RT)-PCR. The single contiguous reading frames encode polypeptides of 334 and 311 residues, respectively, which share 71/63% amino acid sequence similarity/identity ( Figure 2). Like of other members of the FBX superfamily (Gagne et al., 2002;Hua et al., 2011), a signature FBX domain was predicted with high probability (1.2e-04 (AUF1) and 1.2e-04 (AUF2) by HMMER analysis) near the N-terminus of both proteins. Typically, the region C-terminal to the FBX domain contains one or more motifs that help recognize ubiquitylation targets. No known protein-protein interaction domains were obvious in AUF1 or 2 by searching with PFAM, but several stretches enriched in leucine, isoleucine, methione and valine, and interspersed with large bulky hydrophobic residues were evident, suggesting that leucine-rich-type repeats (LRRs) are present, but in a non-canonical arrangement ( Figure 2). Subsequent searches detected AUF1/2-related genes in all available land plant genomes examined, including the moss Physcomitrella patens and the lycopod Selaginella moellendorffii, but not in any animal, fungal or algal species, implying that the corresponding proteins have land plant-specific functions (Suppl. Figure 1). In fact, AUF1/2 are members of a widely-distributed collection of land plant FBX genes undergoing strong purifying selection (Hua et al., 2011), suggesting that the corresponding AUF1/2 proteins direct a conserved and likely essential ubiquitylation event important to terrestrial plant life. Amino acid alignment of AUF1/2related proteins revealed strong conservation in the FBX domain as well as in the C-terminal region bearing the potential LRRs ( Figure 2). Multiple AUF1/2 genes are present in all the species with well annotated genomes, including two each in P. patens, S. moelendorffii and Pinus strobus, three is Oryza sativa, and four in Zea mays (Suppl. Figure 1), raising the possibility that the AUF family expanded early in land plant evolution. While the clustering of monocot paralogs with close relatives from other species rather than with each other supports this scenario, the eudicot paralogs often clustered together with their paralogs, suggesting that the eudicot AUF genes expanded by more lineage-specific events (Suppl. Figure 1).

Reverse Genetic Analysis of AUF1 and 2
To help define the functions of AUF1/2, we acquired several T-DNA insertion mutants affecting the coding regions that should strongly impair synthesis of the corresponding proteins ( Figure 1C). The three auf1 alleles prevent synthesis of the full-length AUF1 transcript as detected by RT-PCR and/or RNA gel blot analysis ( Figure 1B,E). However, RT-PCR detected partial transcripts emanating downstream from the insertion sites in auf1-1 and auf2-1 plants and upstream of the insertion site in auf1-3 plants. The auf1-1 and auf1-2 transcripts likely reflect cryptic promoter activity within the T-DNA insertion. Nonetheless, these two mutants should represent strong alleles based on their disruption of the coding region for the FBXdomain which is required for docking AUF1 with the rest of the SCF complex. The auf1-3 mutant protein should be missing much of its predicted LRR-like target recognition module, likely rendering this truncation inactive even if expressed and assembled into an SCF auf1-3 complex via its intact FBX domain.
Only a single T-DNA insertion line was available for AUF2 ( Figure 1D). RT-PCR analyses showed that the auf2-1 allele also disrupts synthesis of a full-length AUF2 mRNA but that partial transcripts before and after the T-DNA accumulate ( Figure 1E). Even if translated, the resulting truncated polypeptides should be missing either the FBX domain or much of the target recognition module, thus likely compromising both fragments if synthesized separately.

AUF1 Mutants Display Defects in Auxin Transport
Under normal laboratory growth conditions, homozygous seedlings for each of the three auf1 alleles, auf2-1, and the auf1-2 auf2-1 double mutant were phenotypically indistinguishable from the wild-type Col-0 parent and showed normal fertility and genetic segregation. Thus, we conclude that the AUF1 and AUF2 proteins separately or together are not essential for most, if not all, aspects of Arabidopsis development, and reproduction. Given the strong increase in AUF1 mRNA by IAA, we predicted that auf1-2 and auf1-2 auf2-1 plants would show defects in auxin perception and signaling. In contrast, the mutant plants responded normally to exogenous IAA and the synthetic auxin 1-naphthaleneacetic acid (1-NAA) as measured by auxin-induced inhibition of root elongation and promotion of lateral root emergence, and by the rate of root curvature induced by gravity, a well described response that depends upon local changes in the distribution of auxin ( Figure 3A and Suppl. Figure 2D,E). auf1-2 root elongation was similarly unaffected by several other growth regulators, including jasmonic acid, abscisic acid, and the ethylene precursor 1 aminocyclopropane-1-carboxylic acid (data not shown). However, subtle differences in auxin signaling were detected for auf1 plants using several molecular markers of auxin signaling. For example, the expression of the AUX/IAA gene IAA1, which is up-regulated by 1 μ M IAA (Abel et al., 1994), was poorly responsive in the auf1-2 background (Suppl. Figure 2C). The well-characterized auxin-responsive reporter DR5pro: GUS (Ulmasov et al., 1997) also showed a dampened response to exogenous IAA in auf1-2 plants. In the absence of IAA, both wild-type and auf1-2 RMs harboring DR5pro:GUS expressed β-glucuronidase (GUS), as observed by histochemical staining, in a small collection of cells comprising the quiescent center (QC) and the root stem cell niche (Suppl. Figure 2A).
IAA treatment (1 μ M) greatly expanded the zone of expression into the EDZ and mature zone (with root hairs) with the expression less robust in auf1-2 and auf1-2 auf2-1 roots as compared to wild-type and auf2-1 roots (Suppl. Figure 2A). This dampened auxin response for auf1 plants could also be seen by quantitative 4-methyllumbelliferyl-β-D-glucuronide (MUG) assays measuring GUS activity in whole root tips treated with 1 μ M IAA (Suppl. Figure 2B).
To examine the ability of auf1/2 plants to maintain appropriate auxin maxima when auxin transport is compromised, we tested the response of the mutants to NPA and TIBA, two drugs that inhibit polar auxin transport (Lomax et al., 1995). NPA works by blocking the action of the ABCB and potentially the PIN families (Noh et al., 2001;Bouchard et al., 2006;Petrasek et al., 2006), whereas TIBA appears to impair cycling of the PIN family between the plasma membrane and endosomes (Geldner et al., 2001). Strikingly, root elongation of seedlings homozygous for each of the three auf1 alleles and the auf1-2 auf2-1 combination, but not the auf2-1 allele, was hypersensitive to both inhibitors ( Figure 3B). The most significant effects were seen for 5-10 μ M of either TIBA or NPA where a >2-fold difference in root length was observed after a 7-day exposure. The response appeared specific for roots as no obvious differences in shoot growth were evident between NPA-treated wild-type and auf1 plants.
We further confirmed the NPA-hypersensitive phenotype by rescuing auf1-2 plants with transgenes expressing full-length AUF1 under its native promoter, either as an N-terminal fusion to the Flag epitope tag or to green fluorescent protein (GFP). Root growth in the presence of NPA for multiple independent AUF1pro:AUF1-Flag auf1-2 and AUF1pro:AUF1-GFP auf1-2 lines was restored to near the wild-type rate and significantly better than the auf1-2 parent (Suppl. The hypersensitivity of auf1 plants to NPA and TIBA strongly suggested that AUF1 promotes auxin transport. To test this hypothesis, we measured the movement of 3 H-IAA after localized application (Lewis and Muday, 2009). As can be seen in Figure 4A, IAA transport was significantly depressed in the basipetal (shootward) direction in auf1-2 and auf1-3 roots but not auf2-1 roots, and significantly depressed in the acropetal (rootward) direction in auf1-2 roots. Transport in both directions was further dampened by 1 μ M NPA, with the NPA-treated auf1-2 roots having only ~30% of the auxin transport rate found in non-treated wild type roots.
To test the likelihood that NPA-treated auf1-2 plants have altered auxin distributions, we indirectly measured auxin maxima using the DR5pro: GUS reporter (Ulmasov et al., 1997). As with auxin (Suppl. Figure 2A), NPA treatment of wild-type roots expanded the zone of GUS expression basally from the QC and stem cell niche into the RM ( Figure 4B). Notably, the NPAinduced expansion of auxin maxima was substantially more robust in auf1-2 roots and extended further basally into the region of root hair emergence that likely includes the EDZ ( Figure 4B).
The edge of high DR5pro:GUS expression was remarkably sharp, with little or no staining evident in mature cells, suggesting that NPA traps basipetal auxin flow at a defined boundary in auf1-2 roots. Quantitatve measure of DR5pro:GUS activity showed that auf1-2 root tips had overall 3-4 fold more GUS upon NPA treatment as compared to wild-type or auf2-1 roots ( Figure   4C). Taken together, the data suggest that AUF1 regulates auxin distribution primarily in the small region surrounding the RM/EDZ junction, a location consistent with its reported expression patterns (Brady et al., 2007).

AUF1 Modifies PIN Genes Expression
Unlike many other mutations that effect root auxin transport in either the acropetal or basipetal directions (e.g., (Okada et al., 1991;Lewis et al., 2009)), the auf1 mutants significantly reduced auxin transport in both directions, suggesting that AUF1 participates in a global auxin transport-regulating network, genetically upstream of tissue-specific auxin transport facilitators. Upon comparison of wild-type and auf1-2 lines introgressed with the same PIN2pro:PIN2-GFP insertion event, considerably less GFP fluorescence was detected in the root tip in the absence of AUF1 ( Figure 5A). Lower PIN2-GFP signal in auf1-2 roots could have been generated by defects related to the asymmetric plasma-membrane partitioning of PIN2  (Blilou et al., 2005), or its continuous auxin-dependent cycling between the plasma membrane and endosomes (Geldner et al., 2001;Pan et al., 2009). However, the intracellular distribution of PIN2-GFP in root epidermal cells appeared indistinguishable between wild type and auf1-2, with a majority of the fluorescence in both lines localized to what appears to be the basal plasma membrane ( Figure 5A). Moreover, the cycling of PIN2-GFP between the plasma membrane and endosomes was unaffected. Upon treating roots with the drug brefeldin A (BFA), which inhibits endosome cycling of PIN proteins (Geldner et al., 2001), PIN2-GFP accumulated in the endosome-like "BFA bodies" of auf1-2 root epidermal cells at a rate comparable to that of wild type (Suppl. Figure 4). suggesting that AUF1 has an isoform-specific control on PIN protein accumulation ( Figure 6B).
It was also possible that AUF1 affects expression of other key auxin transporters, including AUX1, ABCB4 and ABCB19 (Yang et al., 2006;Lewis et al., 2007;Wu et al., 2007). However, qRT-PCR analyses of auf1-2 roots detected no significant change in their transcript levels as compared to wild type (Suppl. Figure 5).

auf1 Mutants are Hypersensitive to Cytokinin
The aforementioned antagonistic connection between cytokinin and auxin transport through SHY2 (Dello Ioio et al., 2008), and our observations that AUF1 inactivation alters the expression of some PIN genes similar to that observed upon cytokinin treatment ( To test this scenario, we performed a dose-response analysis on auf1 root growth using two natural cytokinins, kinetin and zeatin. Like the response to auxin-transport inhibitors, all three mutant alleles of auf1 and the auf1-2 auf2-1 double, but not auf2-1, were significantly more hypersensitive to these cytokinins ( Figure 6A,B). Moreover, this hypersensitivity could be reversed by introducing the AUF1pro:AUF1-Flag and AUF1pro:AUF1-GFP transgenes into the auf1-2 background (Suppl. Figure 3B). Like NPA, exogenous cytokinin also dampened IAA transport, which was further decreased in the auf1-2 background ( Figure 4A). The inhibition of www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. root growth by cytokinin could be mediated in part by the ability of cytokinin to stimulate ethylene production, which in turn blocks cell elongation (Chae et al., 2003). In support, we found that simultaneous treatment with kinetin and the ethylene biosynthesis inhibitor 2aminoethoxyvinyl glycine (AVG) abolished the kinetin hypersensitivity of auf1-2 roots (Suppl. Figure 6), further highlighting the important hormonal crosstalk that regulates root growth.
One initial response to exogenous cytokinin is the rapid and transient increase in transcripts encoding the type-A negative effectors ARR5 and ARR15 (D'Agostino et al., 2000;Kiba et al., 2002). The increases occur between 10-15 min after cytokinin treatment with the mRNA levels quickly returning back to pre-treatment levels after 30 min. This rise is abolished in cre1-12 ahk3-3 double mutants, thus implicating the two affected cytokinin receptors ( Figure   6C, D and (Higuchi et al., 2004)). When ARR5/15 transcript accumulation was similarly tested in the auf backgrounds by semi-quantitative RT-PCR and qRT-PCR, we found that the levels of both were substantially up-regulated by 0.5 μ M kinetin in auf1-1 and auf1-2 plants as compared to wild-type and auf2-1 plants ( Figure 6C,D). The rise and fall of mRNA levels were similar among the time courses, indicating that the magnitude but not the duration of the responses was accentuated by AUF1 inactivation. By comparison, the transcript levels for neither AUF1 nor AUF2 were increased by kinetin ( Figure 1C), consistent with a scenario in which the decrease of cytokinin responsiveness through AUF1 is mediated indirectly by its auxin upregulation.
Cytokinin affects root development by increasing the rate of cell differentiation, which in turn reduces the size of the RM (Dello Ioio et al., 2008;Moubayidin et al., 2010). To determine if the cytokinin effect on auf1 plants was through an effect on RM size, we compared the number of cortical cells in the file that span the distance from the root quiescent center (QC) to the first noticeably elongated cell marking the beginning of the EDZ. Similar to previous reports (Dello Ioio et al., 2008), we found that exogenous kinetin (0.5 μ M) substantially reduced the number of cells in the wild-type RM. However, the RM of auf1-2 roots was similar in size to wild type without kinetin and was reduced to an equivalent amount upon kinetin treatment ( Figure   7E,F). Collectively, the data imply that AUF1 does not control RM size but instead may affect elongation/differentiation of the downstream EDZ in response to cytokinin.

AUF1 is Required for the Cytokinin-Mediated Promotion of Root Cell Differentiation
To help localize where AUF1 controls root growth in response to cytokinin, we used computer-assisted kinematic analyses of individual roots to describe their growth dynamics in detail (Beemster andBaskin, 1998, 2000;Miller et al., 2007). Growth velocity profiles along the length of wild-type roots revealed that cell growth begins to accelerate ~200  Figure 7A). Cytokinin suppresses growth velocity to produce shorter roots at any point in time. Whereas inactivation of AUF1 did little to affect growth velocity in the absence of kinetin (~230 versus ~250 μ m/hr), growth velocity of the mutant was significantly reduced in its presence (~130 to ~90 μ m/hr) ( Figure 7A), thus explaining why auf1 mutant roots are shorter than wild type after cytokinin treatment.
The first derivative of the velocity profiles was used to produce axial Relative Elemental Growth Rate (REGR) profiles (Beemster and Baskin, 2000). As shown in Figure 7B Differences in the REGR profiles of auf1 versus wild-type roots exposed to cytokinin could reflect differences in the rates of cell production by the RM and/or differences in cell expansion. To observe an effect on cell division, we introgressed the mitotic reporter CYCLINB1:1pro:GUS-DBox into auf1-2 plants, which transiently accumulates GUS at the G2 stage of the cell cycle (Colon-Carmona et al., 1999). From counts of GUS-stained cells, we found that both wild-type and auf1-2 RMs had similar mitotic numbers without treatment and that the numbers decreased equally upon 0.5 μ M kinetin treatment, indicating that AUF1 does not modify cell division rates in the RM. To test for changes in cell expansion rates, we measured the lengths of cortical cells along the entire root axis. Here, a significant effect on cytokinin-treated auf1-2 roots was observed. From initial microscopic analysis of the EDZ region, it was clear that the auf1-2 cortical cells were considerably smaller than similarly treated wild-type cells ( Figure 7D

ARR1 is Epistatic to AUF1
Prior studies connected cytokinin to auxin transport and root growth via the action of the type-B response regulator ARR1 (and possibly ARR12) whose expression is concentrated in the transition zone encompassing the distal RM and the proximal EDZ (Dello Ioio et al., 2007). To examine whether AUF1 and ARR1 might interact genetically, we compared the response of auf1-2 and auf1-3 roots to cytokinin and NPA with that of the arr1-5 null mutant of ARR1 and an arr1-5 auf1-2 double mutant. As shown in Figure 8A, untreated roots from all the genetic backgrounds grew similar to wild type. However, whereas the two auf1 mutants were significantly shorter than wild type in the presence of 5 μ M NPA or 0.5 μ M kinetin, the arr1-5 and arr1-5 auf1-2 plants resembled wild-type.
This rescue demonstrated that the hypersensitivity of auf1 roots to both cytokinin and NPA can be reversed by removing ARR1 ( Figure 8A), The apparent rescue of auf1 roots by the arr1-5 mutation, suggested that ARR1 is in the same pathway and epistatic to AUF1. An obvious possibility is that the ARR1 protein is the direct target of SCF AUF1 , with the corollary expectation that ARR1 is stabilized and thus more abundant in auf1 plants. However, immunoblot analyses of ARR1 in whole root extracts revealed that its abundance with or without kinetin or IAA pretreatment was not consistently increased in the auf1-2 background ( Figure 8B and data not shown). This lack of effect could imply a more complex relationship between ARR1 and AUF1, or the possibility that localized ARR1 turnover is masked when whole roots are examined.

DISCUSSION
Previous studies identified a set of transcriptional and proteolytic feedback loops that antagonistically connect auxin and cytokinin signaling during root growth and differentiation (Blilou et al., 2005;Dello Ioio et al., 2007;2008 reduced its protein abundance. A role for AUF1 in cytokinin perception was supported by a hypersensitivity of auf1 root growth to exogenous cytokinin and by the transiently enhanced upregulation of several type-A ARRs that negatively regulate cytokinin signaling. Part of the cytokinin effect may involve enhanced synthesis of ethylene, which can also alter auxin transport (Vandenbussche et al., 2003;Negi et al., 2007). While the collective data indicate that AUF1 plays a role in modulating auxin transport in roots, it is notable that AUF1 is not essential for either root or shoot growth under normal conditions. Consequently, AUF1 must play a more subtle role in fine tuning the process(es) that dictate auxin movements, auxin maxima, and hormonal crosstalk, which seem to remain somewhat robust in its absence.
Our results with AUF1 provide further support for a role of cytokinin in controlling auxin transport and, in particular, we identify a role for AUF1 in affecting PIN gene expression and ultimately PIN protein accumulation in roots. Presumably, reduced PIN levels dampen auxin distributions sufficiently to make auf1 seedlings more sensitive to auxin transport inhibitors such as NPA and TIBA. One striking feature of auf1-2 plants treated with NPA is the dramatic accumulation of auxin (as observed by DR5pro:GUS expression) in the region encompassing the distal RM and proximal EDZ. Together with kinematic analysis of root growth, the data imply that AUF1 has its greatest effect on auxin concentrations in this small region, which importantly coincides with the region where auxin transport is predicted to be most affected by cytokinin (Dello Ioio et al., 2008). One interesting point from the study of Ruzicka et al. (2009) and observed here is that cytokinin or the elimination of AUF1 do not dampen the expression of all PIN genes but have contrasting roles in decreasing PIN2, 3 and possibly 4 transcript levels while simultaneously increasing PIN7 transcript levels. These differential effects could trap auxin in particular cells/tissues within the RM and EDZ by encouraging the acropetal transport responsible for delivering shoot-derived auxin into this region while simultaneously discouraging the basipetal transport responsible for moving auxin away from the apical maximum.
Interestingly, auf1 mutants share similar phenotypes with several described mdr/abcb mutants, thus supporting further a connection between AUF1 and auxin transport. For instance, the mdr1-1 mutant responds like wild type to exogenous auxin and has normal gravitropism, yet exhibits an 80% reduction of acropetal auxin transport in the root (Lewis et al., 2007). Mutants in PIS1, which encodes the ABCG37 transporter, were shown previously to be hypersensitive to TIBA and NPA as measured by an increased inhibition of root growth, but respond normally to exogenous IAA (Fujita and Syono, 1997 defective in both acropetal and basipetal auxin transport. However, at least for major auxin transporters in roots AUX1, ABCB4 and ABCB19 (Yang et al., 2006;Lewis et al., 2007;Wu et al., 2007), AUF1 appears to have no role in controlling their expression.
The exact role(s) of Arabidopsis AUF1 (and possibly AUF2) in the crosstalk between auxin transport and cytokinin signaling remains unclear. Based on the strong homology within the predicted FBX domains, we expect that both assemble into SCF E3 complexes that recognize the same or similar target(s). Unfortunately, attempts to confirm this assembly have been unsuccessful, primarily due to: (i) our failure to express sufficient quantities of tagged AUF1 variants (Flag or GFP) in planta that could be used to isolate the entire SCF complex, and (ii) the propensity of AUF1 to auto-activate Y2H assays, thus precluding its use in paired interaction studies with Arabidopsis SKP1 proteins. The poor expression of AUF1 is not without precedent as a number of other FBX proteins have been shown to express poorly and/or be inherently unstable as a result of an intrinsic auto-ubiquitylation activity of SCF E3s, which triggers turnover of the FBX subunit by the 26S proteasome (Bosu and Kipreos, 2008;An et al., 2010). In our case, even pretreating AUF1-Flag or AUF1-GFP plants with MG132 failed to permit transgenic AUF1 protein detection with anti-Flag or anti-GFP antibodies. In the absence of direct data, we note that AUF1 and 2 phylogenetically cluster based on their FBX domains close to several well characterized Arabidopsis FBX proteins (Gagne et al., 2002;Hua et al., 2011), including SLEEPY1 which has been shown previously to incorporate into an SCF E3 complex (McGinnis et al., 2003).
While we presume based on sequence homology that AUF2 functions like AUF1, we failed to associate these potential paralogs either genetically or by expression studies. Unlike AUF1, expression of AUF2 is not upregulated by auxin. Furthermore, auf2 mutants were indistinguishable to wild type under all conditions tested, and the auf1-2 auf2-1 double mutant did not display any exaggerated phenotypes compared to the auf1-2 single mutant. Consistent with the low expression of AUF2 relative to AUF1, we propose that AUF2 if active plays a minor, more constitutive role in whatever process(es) these two FBX proteins control.
The central unresolved question pertains to the identity of the AUF1/2 substrate(s). The widespread distribution of AUF-type genes within the plant kingdom implies that these substrate(s) direct a key conserved step in root/rhizoid development that appeared early in land plant evolution. Based on the auf1 phenotypes, we propose that SCF AUF1/2 targets for ubiquitylation a positive regulator in the crosstalk between cytokinin signaling and auxin transport. Over-accumulation of this regulator in auf1 roots accentuates responsiveness to cytokinin, which through the SHY2 feedback loop (and possibly AVG-sensitive ethylene synthesis) dampens the expression of specific PIN genes (e.g., PIN2, 3 and 4) Ruzicka et al., 2009). Concomitant reductions in protein levels (e.g., PIN2) reduce auxin transport below a critical threshold which, in turn, makes auf1 roots more sensitive to auxin transport inhibitors. Coincidently, co-expression correlations derived from nearly all available Arabidopsis microarray datasets found that the expression patterns of AUF1 most closely matched that of SHY2 (Pearson Correlation coefficient = 0.53 (http//atted.jp)) proposed to be at the core of the cytokinin/auxin transport connection.
Previous studies showed that SHY2 expression is up-regulated by cytokinin via the type-B ARR1 transcription factor and that its increased protein abundance then down regulates the expression of multiple PIN genes, thus attenuating auxin transport and altering auxin maxima (Dello Ioio et al., 2007). The end result is shorter roots caused by increased differentiation rates in the RM and EDZ. Auxin conversely, targets SHY2 for breakdown via the SCF TRI1/ABF1-3 E3s, thus relieving this repression. Taken together, we propose a model that explains the auf1 phenotypes and their restoration in the arr1-5 auf2-1 combination. The model states that SCF AUF1 targets ARR1 for ubiquitylation and subsequent turnover in the absence of cytokinin.
Low ARR1 levels would dampen SHY2 expression, thus attenuating its repressive effects on the expression of some, but not all, PIN genes. Increased PIN levels driving robust auxin transport would then promote maintenance of the RM and delay elongation/differentiation, thus increasing root growth. The auxin-induced expression of AUF1 in the RM/EDZ could reinforce ARR1 turnover and subsequent SHY2 downregulation by increasing the concentration of the SCF AUF1/2 complex. Cytokinin in contrast could antagonize this promotional effect on growth by blocking ARR1 breakdown by SCF AUF1/2 . ARR1 stabilization by cytokinin or AUF1 inactivation increases SHY2 transcription, with the increased SHY2 protein levels then repressing both cytokinin synthesis and PIN expression, resulting in reduced auxin transport. As observed here for auf1 mutants, increased ARR1 would also up-regulate transcription of the ARR5 and ARR15 type-A negative effectors of cytokinin signaling (Taniguchi et al., 2007).
While the data are consistent with ARR1 being the SCF AUF1/2 substrate, we failed to detect a marked increase of ARR1 protein in auf1 backgrounds from the immunoblot analysis of whole seedlings or even just roots with or without auxin and cytokinin pretreatment.
Consequently, other modes of ARR1 downregulation are possible, including AUF1 affecting the abundance and/or activity of other positive/negative regulators associated with cytokinin signaling, including a secondary effect on auxin transport by ethylene. However, given that the SCF AUF1/2 complex may have a highly restricted effect on auxin transport (i.e., distal RM and proximal EDZ), the ability to detect of ARR1 stabilization in auf1 plants may require focused analysis of just the small zone of root tissue where AUF1 and ARR1 overlap (i.e., distal RM and proximal EDZ (Brady et al., 2007;Dello Ioio et al., 2007)). Whatever the SCF AUF1/2 substrate(s), www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. their identities will likely reveal an additional regulatory mechanism within the root tip that dynamically responds to auxin/cytokinin crosstalk.

Plant Growth Conditions
Arabidopsis thaliana ecotype Columbia (Col)-0 plants were used in all analyses. Prior to germination, seeds were vapor-phase sterilized and incubated in sterile water at 4°C for 2 d in the dark. For phenotypic assays, seedlings were grown vertically on half-strength Murashige and Skoog (MS) medium with 1% sucrose and 1% agar at 21°C in continuous white light. For liquid-grown seedlings, seeds were germinated and cultured in half-strength MS liquid medium under continuous white light. For various hormone and inhibitor treatments, the compounds were added to the germination medium. Effects on the compounds on root elongation were quantified using the ImageJ software package (http://rsb.info.nih.gov/ij/).

AUF Protein Sequence Alignments and Phylogenetic Analysis
The full-length AUF1 protein sequence was used as a query to search NCBI non-
The genotype of each line was determined by PCR of total genomic DNA using gene-specific primers in combination with T-DNA left border-specific primers. Primers used in this study are listed in Table 1  Complementation was examined using an AUF1 fragment generated by genomic PCR, which contained 2,500 bps of 5' sequence upstream of the ATG translation initiation codon and the region encompassing the full coding region. The product was cloned into the pDONR221 entry vector (Invitrogen), sequence confirmed, and then recombined into the destination vectors pEARLYGATE302 containing the Flag epitope (DYKDDDDK) or pMDC107 containing the fullcoding region of GFP, which appended the tags in-frame to the 3-end of the AUF1 coding region.
The constructions were transformed into homozygous auf1-2 plants by the Agrobacterium-mediated floral-dip method. Homozygous plants for all loci were confirmed by Basta resistance and by genomic PCR of T3 plants.

RNA-Gel Blot and qRT-PCR Analyses
For RNA-gel blot analysis, total RNA from 7-d-old roots was isolated according to (Smalle et al., 2002). 32 P-labeled riboprobes were synthesized with T7 or SP6 polymerases using the Riboprobe system (Promega) and the linearized pGEMT (Promega) cDNA constructions of AUF1 and 18S rRNA. Membranes were hybridized overnight at 65°C and washed as described (Smalle et al., 2002) prior to autoradiography.
RT-PCR analysis was conducted with total RNA isolated from liquid-grown plants using was isolated as above and reverse transcribed into cDNA using the Superscript II reverse transcriptase kit (Invitrogen). qRT-PCR amplification was performed with the MyiQ5 two-color real-time PCR detection system using SYBR Premix EX Taq (Takara). Relative expression was calculated by the comparative threshold cycle method, using reactions with the ACT2 or ACT4 transcripts as the internal control.

Auxin Transport Assays
Acropetal or basipetal transport of auxin was measured by applying agar droplets containing 3 H-IAA to the root/shoot junction zone or the root apex of 5-d-old seedlings grown under constant light. Measurement of IAA movement was as described by (Lewis and Muday, 2009). Each measurement represented the average of three independent assays, each of which was performed with 15-20 seedlings.

Histochemical Analysis and MUG Assay
Histochemical staining for GUS activity was conducted as published (Malamy and Benfey, 1997), using the substrate 5-bromo-4-chloro-3-indolyl β-D-glucuronic acid (X-gluc). For analysis of CYCLINB1;1 pro:GUS-DBox expression, roots from 7-d-old seedlings were stained overnight prior to light microscopy. For quantitative MUG assays of DR5pro:GUS expression after IAA treatment, 40 seedlings were grown for 7 d on solid MS medium containing each concentration of IAA. Crude extracts were incubated with a reaction mix containing 0.3 mM MUG, 50 mM Na 2 HPO 4 (pH 7.0), 10 mM 2-mercaptoethanol, 10 mM Na 2 EDTA and 1 mM phenylmethylsulfonyl fluoride for 10 min. MUG product fluorescence was determined with a Wallac microtitre plate fluorometer. Values were normalized against total protein concentrations, which were determined by absorption at 280 nm with a Nanodrop spectrophotometer (Thermo Scientific). For quantitative MUG assays of DR5pro:GUS expression in the roots treated with NPA, the seedlings were grown for 4 d on solid MS medium with or without NPA. The 2-3 mm apical portion from 50 roots for each genotype was collected, homogenized, and then assayed in triplicate as above.

Laser Scanning Confocal Microscopy
GFP fluorescence was imaged with a Zeiss 510-Meta scanning laser confocal microscope using 488-nm light excitation and 500 to 530-nm light emission. For imaging propidium iodide stained roots, the 543-nm line of the helium/neon laser was used for excitation, and emission was detected at 590 to 620 nm. Laser, pinhole, and gain settings of the confocal www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. microscope were kept identical among treatments for direct comparison. For analysis of PIN2 membrane cycling, 4-d-old liquid-grown PIN2pro:PIN2-GFP seedlings were pretreated with 50 μ M cycloheximide for 30 min and then exposed for additional times to 10 μ M BFA as described (Pan et al., 2009). To measure root meristem size, roots were stained with 10 μ g/ml propidium iodide for 10 sec prior to microscopy. Cells in the cortical cell file from the QC to the first noticeably more elongated cell were counted manually (Dello Ioio et al., 2007). Images were assembled using Photoshop version 4.0 (Adobe Systems).

Kinematic and Cell Length Measurements
For kinematic analyses of root growth, a thin layer of half-strength MS medium plus agar, supplemented with or without 0.5 μ M kinetin, was poured over a glass cover slide and cooled.
Seeds were sown on top of the agar, covered with a glass cover slip, and then allowed to germinate and grow vertically between the agar-glass interface. After 6.5 days, the sandwich containing the seedlings was mounted in a small growth chamber, placed in a vertical position, and then allowed to equilibrate for 1 hr prior to data collection (Miller et al., 2007). Images of the apical 2 mm of root were acquired using a horizontal Nikon light microscope, a 10x objective, and an AVT Pike camera every 30 sec for 20 min at a resolution of 1.77 μ m per pixel. The resulting Nomarski image stacks were analyzed by customized software, using an optical flow technique (Lucas and Kanade, 1981;Beemster and Baskin, 1998) to obtain tissue growth velocity along the axis.
To measure the Relative Elemental Growth Rate (REGR) profiles, we selected ~30 points along the midline of the organ on the first frame of the image stack and these points were tracked through the time series. Circular image patches surrounding the selected points were deformed and translated from the i th frame to best match corresponding image patches in the (i+1) th frame. The rate at which these image patches traversed the axis of the organ generated the tissue velocity. A flexible logistic function was fit to the velocity data, which resulted in a smooth representation of the velocity profile (Morris and Silk, 1992). The velocity profile was differentiated with respect to arc-length to obtain the REGR profile. A detailed description of the technique and software will be published elsewhere (N.D. Miller and E.P. Spalding, unpublished).
The average cell length in the cortical cell file was subsequently determined from the same roots used in the kinematic analysis. The roots were stained with propidium iodine, and imaged with a confocal microscope (Zeiss; LSM500) at 20X magnification. A series of overlapping root images were tiled to generate complete root cell pictures. The lengths of cells for a continuous single cortical cell file were measured along both sides of the root using the www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. ImageJ software package (http://rsb.info.nih.gov/ij/). The position of each cell was calculated from the cumulative length of all cells between it and the root quiescent center. Subsequently, a logistic function was fit to the data.

Supplemental Data
The following materials are available in the online version of the article.
Supplemental Table 1    For primary root length measurements, seedlings were germinated and exposed to the       The numbers reflect the ratio of expression between control and treated plants. ABA, abscisic acid; ACC, 1-aminocyclopropane-2-carboxylic acid; GA3, gibberellic acid-3; SA, salicylic acid; Me-JA, methyl-jasmonate. B, RNA gel blot analysis of AUF1 mRNA after IAA treatment. Ten µg of total RNA was extracted from 7-day-old roots from auf1-2, auf2-1 and auf1-2 auf2-1 seedlings treated with (+) or without (-) 0.1 µM IAA for 5 hrs. 18S rRNA was used as a loading control. C, qRT-PCR of the AUF1 and AUF2 mRNAs in response to kinetin (Kin) and IAA.
Seven-d-old seedlings were treated with 0.5 µM Kin or 0.1 µM IAA for the indicated times. Root total RNA was then subjected to qRT-PCR using gene-specific primers. Transcript levels were normalized using the ACTIN (ACT)-2 transcript as a control. D, Organization of the AUF1 and  Table 1. E, RT-PCR analyses of auf mutants. RNA was subjected to first-strand cDNA synthesis with genespecific primers 2 and 6 for AUF1 and AUF2, respectively, and then subjected to PCR using the indicated primers. RT-PCR of PAE2 mRNA was included as a control.    For primary root length measurements, seedlings were germinated and exposed to the indicated concentrations of IAA and the inhibitors TIBA and NPA for 7 d in constant light before measurement. For lateral root densities, the plants were first grown without IAA for 4 days and then transferred to medium containing various concentrations of IAA and grown for an additional 6 d. Each point represents the average (±SD) of >20 seedlings. A, Effect of IAA on root elongation and lateral root emergence. B, Effect of TIBA and NPA on root elongation.
Asterisks identify points with significant differences between wild type and the mutants containing the auf1 alleles (Student t-test, p-value <0.0047). Pictures of representative plants exposed to 10 µM TIBA or NPA are shown above each graph. Scale bars represent 5 mm. The white lines highlight root tips.    Seedlings were grown for 7 d before measurement. Asterisks identify points with significant differences between wild type and the mutants containing the auf1 alleles (Student t-test, pvalue <1.5e-13). C, Accumulation of the cytokinin-regulated ARR5 transcript in 10-d-old seedlings treated for various times with 0.5 µM kinetin as determined by semi-quantitative RT-PCR. RT-PCR of the ACTIN (ACT)-4 transcript was used to confirm the analysis of equal amounts of RNA. The cytokinin-insensitive cre1-12 ahk3-3 double mutant was used as a control. D, Accumulation of the cytokinin-regulated ARR5 and ARR15 transcripts in 10-d-old seedlings treated for various times with 0.5 µM kinetin as determined by qRT-PCR using the ACT4 transcript as the control. Each bar represents the mean (±SE) of three independent biological replicates in which each RNA sample was analyzed in triplicate. E and F, Root meristem size is unaffected by the auf1-2 mutation. E, Representative root tips from 7-d-old wild-type and auf1-2 plants treated with 0.5 µM kinetin. The cell walls were stained with propidium iodine and visualized by confocal fluorescence microscopy. The lines highlight the approximate position of the RM boundaries as defined by the zone between the stem cell niche and the first noticeably more elongated cortical cell in the proximal EDZ (Dello Ioio et al., 2007).   A, Removal of ARR1 reverses the hypersensitivity of auf1-2 roots to cytokinin and the auxin transport inhibitor NPA. Primary root length was measured on 7-d-old wild-type, auf1-2, auf1-3, arr1-5 and the arr1-5 auf1-2 double mutant seedlings grown without or with 0.5 µM kinetin or 5 µM NPA. Error bars indicate ±SD (n>20). Asterisks identify treatments with significant differences between wild type and the mutants containing the auf1 alleles (Student t-test, pvalue ≤0.0012). B, Immunoblot detection of ARR1. Crude extracts were prepared from roots harvested from 10-d-old seedlings and subjected to SDS-PAGE and immunoblot analysis with anti-ARR1 antibodies. Duplicate loads are shown to account for lane to lane variation. Equal protein loading was confirmed by immunoblot analysis with antibodies against the 26S proteasome subunit RPT1a. C, Possible model describing the role of SCF AUF1 E3 in the crosstalk between cytokinin and auxin transport.
www.plantphysiol.org on August 28, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved.