Bipartite promoter element required for auxin response.

Multiple mechanisms have been described for coordination of responses to the plant hormones auxin and brassinosteroids (Zhang et al., 2009). One unexplained phenomenon is the reliance of the auxin transcriptional response on a functional brassinosteroid pathway. In this study, we used luciferase reporters to interrogate the promoter of SMALL AUXIN-UP RNA15 (SAUR15), a well-characterized auxin and brassinosteroid early response gene in Arabidopsis (Arabidopsis thaliana). After identifying a minimal region sufficient for auxin response, we targeted predicted cis-regulatory elements contained within this sequence and found a critical subset required for hormone response. Specifically, reporter sensitivity to auxin treatment required two elements: a Hormone Up at Dawn (HUD)-type E-box and an AuxRE-related TGTCT element. Reporter response to brassinosteroid treatment relied on the same two elements. Consistent with these findings, the transcription factors BRASSINOSTEROID INSENSITIVE1-EMS SUPPESSOR1 and MONOPTEROS (MP)/ AUXIN RESPONSE FACTOR5 (ARF5) showed enhanced binding to the critical promoter region containing these elements. Treatment with auxin or brassinosteroids could enhance binding of either transcription factor, and brassinosteroid enhancement of MP/ARF5 binding required an intact HUD element. Conservation of clustered HUD elements and AuxRE-related sequences in promoters of putative SAUR15 orthologs in a number of flowering plant species, in combination with evidence for statistically significant clustering of these elements across all Arabidopsis promoters, provided further evidence of the functional importance of coordinated transcription factor binding.

Auxin and brassinosteroids are essential for photomorphogenesis in Arabidopsis (Arabidopsis thaliana) and promote growth in many plant species Stewart and Nemhauser, 2010). Microarray studies have identified a large number of early response genes shared by both hormones (Goda et al., 2004;Nemhauser et al., 2006). Many of these common targets are synergistically induced when both hormones are provided exogenously (Nemhauser et al., 2004;Vert et al., 2008). In addition, auxin and brassinosteroid transcriptional response pathways also exhibit interdependence; changes in gene expression in response to one hormone require the function of the other (Nemhauser et al., 2004(Nemhauser et al., , 2006Nakamura et al., 2006;Hardtke, 2007).
The close relationship between auxin and brassinosteroids likely reflects several levels of cross-regulation.
For example, brassinosteroids have been shown to promote polar auxin transport, leading to altered auxin distribution within and between organs (Bao et al., 2004;Li et al., 2005). In addition, auxin regulation of brassinosteroid biosynthesis is required to sustain a robust auxin response in diverse tissues (Mouchel et al., 2006;Beuchat et al., 2010). Bioinformatic analyses of target genes for each hormone suggest that auxin and brassinosteroid could directly affect one another's signaling pathways (Goda et al., 2004;Nemhauser et al., 2006). Recent studies have supported this model with evidence for cross-regulation of both auxin-and brassinosteroid-associated transcription factors (Nakamura et al., 2003b(Nakamura et al., , 2006Kim et al., 2006;Vert et al., 2008;Chandler et al., 2009). How these interactions manifest in transcriptional complexes on target promoters remains largely unknown.
The transcription factors regulated by auxin and brassinosteroids are well characterized, as are their cognate cis-regulatory elements. The AUXIN RE-SPONSE FACTOR (ARF) family of transcription factors binds to auxin-responsive elements with the sequence TGTCTC, called canonical AuxREs (Ulmasov et al., 1995) or ARFATs (Higo et al., 1999). A repression complex that can be removed in the presence of auxin modulates ARF activity (Gray et al., 1999(Gray et al., , 2001Tiwari et al., 2001;Szemenyei et al., 2008). Brassinosteroids change gene expression by activating the BRASSINOZOLE RE-SISTANT1 (BZR1; Wang et al., 2002)/BRASSINOSTE-ROID INSENSITIVE1 (BRI1)-EMS SUPPESSOR1 (BES1; Yin et al., 2002) family of transcription factors. BES1 and BZR1 have been shown to bind to E-boxes 1 This work was supported by the National Science Foundation (grant no. IOS-0919021) and the University of Washington. C.L.W. was supported by the Developmental Biology Predoctoral Training Grant (T32HD007183) from the National Institute of Child Health and Human Development.
* Corresponding author; e-mail jn7@uw.edu The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiology. org) is: Jennifer L. Nemhauser (jn7@uw.edu). [C] Some figures in this article are displayed in color online but in black and white in the print edition. [W] The online version of this article contains Web-only data. [OA] Open Access articles can be viewed online without a subscription.
In this study, we used the promoter of SMALL AUXIN-UP RNA15 (SAUR15) to dissect the mechanism for auxin reliance on brassinosteroids for induction of target genes. SAUR15 belongs to one of the first identified families of auxin early response genes (McClure et al., 1989) and has been studied extensively as a brassinosteroid early response gene (Yin et al., 2005;Vert et al., 2008;Sun et al., 2010;Yu et al., 2011). Here, we show that auxin and brassinosteroid responses require the same cis-regulatory elements: a Hormone Up at Dawn (HUD)-type E-box in combination with a nearby AuxRE variant. BES1 and MONOP-TEROS (MP)/ARF5 bind to the promoter region containing these sites, and binding of either transcrip-tion factor can be enhanced by treatment with either auxin or brassinosteroids. Analysis of putative orthologs from diverse species reveals conservation of clustered AuxRE variants and HUD elements, and a number of other key Arabidopsis genes in the auxin and brassinosteroid pathways share similar promoter architecture. This work points to an additional mechanism of coordination between auxin and brassinosteroid transcriptional responses.

SAUR15 Is Induced by Auxin and Brassinosteroid Treatment
Growth promotion by auxin or brassinosteroids requires the function of both signaling pathways (Nakamura et al., 2003a(Nakamura et al., , 2003b(Nakamura et al., , 2006Nemhauser et al., 2004). It is less clear if this phenomenon reflects a convergence at the level of transcriptional control . To determine whether auxin and brassinosteroids coordinately regulate activity on target gene promoters, we analyzed the regulatory region of SAUR15, one of the best characterized auxin-and brassinosteroid-responsive genes (Gil et al., 1994;Green, 1996, 1997;Nakamura et al., 2003a). SAUR15 is rapidly induced following hormone treatments (Gil et al., 1994;Nakamura et al., 2003a). In our conditions, SAUR15 mRNA levels more than quadrupled within 1 h of auxin or brassinosteroid treatment (Fig. 1A). As previously reported, response dynamics differed somewhat between the two hormones (Nakamura et al., 2003a). Seedlings treated with the natural auxin indole-3-acetic acid (IAA) sustained a similar level of mRNA induction after 1 or 3 h. For seedlings treated Figure 1. SAUR15 is an auxin-and brassinosteroid (BR)-responsive gene. A, SAUR15 is responsive to both auxin and brassinosteroids. Quantitative RT-PCR analysis was performed on three independent biological replicates of total RNA isolated from seedlings following treatment with mock, auxin, or brassinosteroids for 1 or 3 h. Expression at 1 h under mock treatment is set to 1. Error bars represent SE. B, The intergenic region between At4g38840 and At4g38850 (SAUR15) contains a number of cisregulatory elements important for response to auxin and brassinosteroids. Red bars located above the line represent predicted or known brassinosteroid cis-regulatory elements: E1-E5, E-boxes. Blue bars located below the line represent potential auxin cis- with the brassinosteroid brassinolide, mRNA levels were clearly elevated within 1 h and were consistently higher after 3 h (Fig. 1A).
SAUR15 (At4g38850) is an obvious target for analysis of regulatory regions. Similar to other SAUR genes, it contains no introns, and there is only a small upstream intergenic region shared by At4g38840. Within the putative SAUR15 promoter sequence, there are several predicted auxin and brassinosteroid cis-regulatory elements (Fig. 1B). For brassinosteroids, there are five E-boxes (Fig. 1B, red ticks numbered E1-E5) but no BRRE elements. E1, E2, E3, and E5 are HUD elements with the sequence CACATG/CATGTG (Michael et al., 2008). HUD elements were previously reported as overrepresented in the promoters of cycling genes related to diverse hormone pathways (Michael et al., 2008). Previous studies have shown that MYB30 can directly interact with BES1, bind to a site immediately adjacent to E4, and increase induction of SAUR15 expression (Li et al., 2009). For auxin, there are 10 TGTC/GACA core elements (Fig. 1B, blue ticks numbered A2-A11), including one canonical AuxRE (GA-GACA; A5). A1 overlaps with E1 and represents an AuxRE variant with a single base insertion in the middle of the element (TGTGCTC). A related element found in the promoter of the brassinosteroid biosynthetic gene DWF4 was recently shown to be important for auxin response (Chung et al., 2011).

A Short Promoter Region Is Sufficient for Normal Auxin Response and Localization of SAUR15
To identify a short promoter sufficient for auxin response, we assayed a series of deletions of the intergenic region upstream of SAUR15. These putative promoter regions were used to drive expression of a luciferase reporter (LUC). First generation transformed (T1) seedlings were presprayed with luciferin to remove accumulated LUC enzyme and subsequently exposed to auxin for 3 h. Auxin response for each construct was determined by averaging fold-change for a large number of T1 seedlings (24 independent transformants for pS15-5E and 30 independent transformants for all other constructs). When the reporter was driven by the full intergenic region upstream of SAUR15 (pS15-5E), seedlings showed similar auxin response to previous studies with the well-characterized SAUR-AC1 reporter (Gil and Green, 1997;Fig. 2A). A reduction of the putative promoter to 290 bp [pS15-2E(ii)::LUC] resulted in a modest increase in auxin sensitivity, suggesting there may be negative regulators that can bind these upstream regions. The slightly longer pS15-2E(i)::LUC reporters showed similar auxin response to the full-length constructs. The additional sequence included in this reporter construct does not contain any TGTC/GACA cores, so any negative regulation is unlikely to be ARF mediated. Auxin response of reporter constructs with only 200 bp of the upstream regulatory region resembled full-length constructs, suggesting they retained core sites required for hormone sensitivity [ Fig To test whether our deletions changed the pattern of SAUR15 expression, we engineered additional seedlings with the pS15-5E, pS15-2E(ii), and pS15-1E(i) regulatory regions driving expression of the uidA gene encoding GUS. The original SAUR-AC1::GUS reporter included the intergenic region between SAUR15 and At4g38840, the entire coding region of At4g38840, and part of the coding region of At4g38830. All of the shorter Figure 2. A short region is required for auxin sensitivity and localization of SAUR15. A, Auxin response requires a region containing three TGTC/GACA cores (A2-4), but not the A5 element. Relative luciferase activity is defined as activity after 3 h divided by baseline. Median response levels are indicated by horizontal green lines. Linear regression was used to test for significant differences in auxin response between constructs. Asterisks indicate lines with significant differences in auxin response from pS15-5E. pS15-2E(ii) showed significantly higher auxin response, while pS15-1E(ii) and pS15-0E showed a significantly lower auxin response. Twenty-four independent transformants were evaluated for pS15-5E. At least 30 T1 seedlings were analyzed for all other lines. Error bars represent SE. B, Expression pattern of SAUR15 is retained in short auxin-responsive reporter lines. Deletion constructs with auxin response [pS15-5E::GUS, pS15-2E(ii)::GUS, or pS15-1E(i)::GUS] show similar expression pattern as the original SAUR-AC1::GUS reporter. For all seedlings, staining is highest in the petioles of young leaves and the hypocotyl region closest to the shoot apical meristem. Staining can also be detected in the vasculature of cotyledons. No staining is detected in the roots. Scale is the same in all panels; bar represents 1 mm.
constructs we analyzed showed similar staining patterns to seedlings containing the original SAUR-AC1 construct (Fig. 2B). While there may be functionally important sequences outside of the minimal region defined here, the 200-bp region in the pS15-1E(i) construct was sufficient for both auxin response and proper localization of reporter expression.

Auxin and Brassinosteroid Induction of SAUR15 Requires Two Crucial Elements
To directly assess the contribution of specific cisregulatory elements to auxin response, we systematically mutated TGTC/GACA cores (A1-A5) and E-boxes (E1-E2) in the pS15-2E(ii)::LUC construct. The pS15-2E(ii)::LUC construct was selected for further analysis because it showed the lowest levels of LUC activity in the absence of hormones (Supplemental Fig. S1) and a robust increase in auxin-induced promoter activity ( Fig. 2A). The low basal level of reporter expression reduced noise in the luciferase assays and allowed detection of smaller effects on reporter activity. We began our analysis with the canonical AuxRE (TGTCTC, A5) and its close variant (TGTGCTC, A1) because these seemed the most likely to confer auxin responsiveness (Chung et al., 2011). Contrary to this expectation, neither A5 nor A1 were essential for auxin response, although A5 did contribute to the magnitude of the induction by auxin (Fig.  3A). In contrast, most T1 seedlings transformed with reporters lacking the A2, A3, and A4 TGTC/GACA cores [pS15-2E(DA4A3A2)::LUC] showed a complete loss of auxin response. Loss of A2 alone [pS15-2E (DA2)::LUC] drastically reduced auxin induction of LUC activity, demonstrating that A2 is acting as an essential AuxRE in the SAUR15 promoter. Recently, another element with the same sequence as A2 (CA-GACA) was found to act as a bona fide AuxRE in the unrelated ATHB8 promoter (Donner et al., 2009). Consistent with this finding, we observed that loss of A2 alone [pS15-2E(DA2)::LUC] drastically reduced auxin induction of the reporter, demonstrating that A2 (CAGACA) but not A5 (GAGACA) is the AuxRE for SAUR15.
We also tested whether E-boxes, which have been linked to brassinosteroid responses, had any role in auxin effect on transcription. Surprisingly, mutations in both E-boxes closest to the transcriptional start site [pS15-2E(DE2E1)::LUC] eliminated auxin response. Loss of E1 alone [pS15-2E(DE1)::LUC] but not E2 [pS15-2E(DE2)::LUC] was sufficient for this effect. Previous work on the soybean (Glycine max) SAUR15A gene (Xu et al., 1997) identified an auxin-responsive domain containing an E-box but no AuxRE-related sequence. Our results, in combination with this previous work, lend functional significance to computational studies showing a significant overrepresentation of E-boxes in shared auxin and brassinosteroid targets (Nemhauser et al., 2004). It is also worth noting that constructs with a reduced response to auxin treatment showed somewhat lower basal levels of luciferase activity (Supplemental Fig. S1), perhaps reflecting a reduced response to endogenous hormone levels. Another possibility is that fluctuations in basal levels among reporter lines could reflect the impact of additional regulatory pathways acting on SAUR15 or additional modulators of auxin response.
Because previous studies had demonstrated auxin dependence on brassinosteroids, we wanted to check whether this relationship extended to specific ciselements. Two T1 plants representing median auxin responses were selected for each reporter construct. By using isogenic lines rather than individual T1 plants, we could expose presprayed seedlings to mock or hormone treatments and measure luciferase activity at the identical time point. In this way, we could eliminate any potential effect of the circadian clock or extended duration of submergence, while also directly comparing reporter responses for each hormone. Auxin response in all T2 seedlings was similar to what was observed for the same constructs in the T1 experiments, suggesting that neither the circadian clock nor extended duration of submergence interfered with measurements of auxin response (Fig. 3B).
Seedlings from T2 families carrying pS15-2E(ii):: LUC, as well as those containing mutations in predicted cis elements, were exposed to 3 h of mock, auxin, or brassinosteroid treatments (Fig. 3B). Unlike in the T1 analysis, average relative luciferase activity for each T2 family was determined by dividing the average luciferase activity of all hormone-treated seedlings by the average luciferase activity of all mock-treated seedlings. Similar to the results in the T1 plants, mutations in A2 or E1 substantially reduced auxin response (Fig. 3B). Brassinosteroids had a more subtle effect on reporter induction than auxin, reducing the power to detect statistically significant hormone effects. Despite this limitation, brassinosteroid response did follow the same pattern as auxin response, requiring both A2 and E1 sites (Fig. 3B).

Treatment with Either Auxin or Brassinosteroids Can Enhance DNA Binding by Key Transcription Factors
E-boxes and AuxRE elements have been linked with brassinosteroid and auxin responses, respectively. Chromatin immunoprecipitation (ChIP) assays have shown that BES1 can bind the promoter of SAUR15 in the region around E5 and E4 (Yin et al., 2005). This area is approximately 500 bp upstream of the transcriptional start site and outside of the minimum region defined here. We performed additional ChIP assays on pBES1::BES1:GFP seedlings to test for binding near the A2 and E1 sites. Brassinosteroid treatment caused a substantial enrichment in BES1 binding in this region (Fig. 3C). This finding is consistent with recent global ChIP data showing BES1 binding enriched near transcriptional start sites (Yu et al., 2011). BES1 binding further upstream, perhaps in conjunction with MYB30, is not essential but may enhance brassinosteroid re- Figure 3. cis-Regulatory elements are required for auxin and brassinosteroid response. A, A2 and an E1 are required for auxin response. Relative luciferase activity is defined as activity after 3 h divided by baseline. Median response levels are indicated by horizontal green lines. Linear regression was used to test for significant differences in auxin response between constructs. Asterisks indicate lines with significant differences in auxin response from pS15-2E(ii). Loss of the canonical AuxRE [pS15-2E (DA5)] led to a modest albeit statistically significant reduction in auxin response. Loss of A2, A3, and A4 [pS15-2E(DA4A3A2)] or A2 alone [pS15-2E(DA2)] essentially eliminated auxin response. Loss of E1 alone [pS15-2E(DE1)] or in combination with a loss in E2 [pS15-2E(DE2E1)] led to a similarly severe loss of reporter sensitivity to auxin treatment. At least 30 T1 seedlings were analyzed for all lines. Error bars represent SE. B, Brassinosteroid response follows a similar pattern as auxin response. Two representative T1 individuals carrying each construct were selected for propagation. Luciferase activity of 24 seedlings from each T2 line was measured following 3 h of mock, auxin, or brassinosteroid (BR) treatment. As seen in the T1 screen, loss of A2, A3, and A4 [pS15-2E(DA4A3A2)] or A2 alone [pS15-2E(DA2)] eliminated auxin and brassinosteroid response. Loss of E1 alone [pS15-2E(DE1)] or in combination with a loss in E2 [pS15-2E(DE2E1)] led to a similarly severe loss of reporter sensitivity to auxin and brassinosteroid treatment. Asterisks indicate significant differences in hormone response from pS15-2E(ii) identified using a two-way ANOVA followed by a Tukey's honestly significant differences test. Two biological replicates were assayed for each T2 line under each condition. Arrowheads indicate location of mutated elements. Error bars represent SE. C, Brassinosteroids and auxin can regulate MONOPTEROS/ARF5 and BES1 binding to DNA. ChIP assays were used to test candidate transcription factor binding to the region of the SAUR15 promoter containing the essential A2 and E1 elements. Protein binding was assayed following 3 h of mock, auxin, or BR treatments. Binding of BES1:GFP to the SAUR15 promoter could be detected in mock treated seedlings but was clearly enhanced by both hormone treatments. Similarly, MP:GFP binding was enhanced by treatment with either auxin or brassinosteroids. Results are expressed as a ratio of experimental condition to no antibody control, normalized to binding within the SAUR15 open reading frame. Each treatment is the average of at least two independent biological replicates. Position of the PCR amplicons is indicated on schematic. Error bars represent SE. D, E1 is required for brassinosteroid enhancement of MP/ARF5 binding to DNA. ChIP assays were used to test MP:GFP binding to transgene promoters containing intact or mutated E1 sites. Results from two independent biological replicates are expressed as in C. Error bars represent SE. [See online article for color version of this figure.] sponse. Auxin treatment was also able to increase BES1 binding, raising the possibility of a larger transcriptional complex. Such a model is complementary to the recent finding that auxin induces expression of a brassinosteroid biosynthetic enzyme through inhibition of BZR1 binding to its promoter (Chung et al., 2011).
Analysis of MP/ARF5 provided further evidence for a larger transcriptional complex representing factors regulated by multiple pathways. Previous studies suggested that MP/ARF5 regulates SAUR15 expression (Schlereth et al., 2010) and can bind to A2-like elements (Donner et al., 2009). We performed ChIP assays on pMP::MP:GFP seedlings using primers flanking the critical minimal region where we observed BES1 binding. Although we found little evidence of MP/ ARF5 binding in mock-treated samples, binding was clearly enriched following treatment with either auxin or brassinosteroids (Fig. 3C). Although no direct interaction has been shown between BES1 and any of the ARFs, other proteins may be acting as bridges between the two factors. If BES1 was required for brassinosteroid enhancement of MP/ARF5 binding, then we reasoned that loss of the E1 site should reduce the effects of brassinosteroids on MP/ARF5. To test this, we crossed pMP::MP:GFP plants to both pS15-2E(ii)::LUC and pS15-2E(DE1)::LUC plants. ChIP assays with transgenespecific primers showed brassinosteroid-induced increase in MP/ARF5 binding did largely depend on an intact E1 site (Fig. 3D).

Conservation of Critical cis-Regulatory Elements
SAURs were originally identified in soybean (McClure et al., 1989) and can be identified in the genomes of many flowering plants. If the elements identified in this study are important for gene function, they would likely be preserved during species divergence. To test this prediction, we compared upstream regulatory regions of putative SAUR15 orthologs in a number of species at increasing evolutionary distance from Arabidopsis. These included the close relative Arabidopsis lyrata, poplar (Populus spp.), soybean (Glycine max), and rice (Oryza sativa). This group spans more than 100 million years of evolutionary time (Reineke et al., 2011). Although blocks of extended shared sequence were still readily detectable in A. lyrata, this was not the case with any of the other promoters.
Within 250 bases upstream of the predicted transcriptional start sites of every putative ortholog, we were able to identify HUD elements (CACATG/ CATGTG; Fig. 4). Of all possible 5-mers contained within the 250 upstream bases, only four sequences were exactly conserved across all six genes examined here (AGACA, TCATG, TTCTC, TTTCT). This is consistent with a recent bioinformatic analysis showing rare conservation of plant promoter sequences across the monocot-dicot divide (Reineke et al., 2011). One of the 5-mers that was absolutely conserved was the A2type of AuxRE variant (AGACA). In fact, at least one AGACA element was found within 50 bp of a HUD element in every promoter examined here.
Given the small number of elements found to be required for hormone response in the SAUR15 promoter, it is not clear how many of the elements identified in this bioinformatic analysis are functionally relevant. One way to enrich for functionality is to focus on genes induced by auxin or brassinosteroids. Among stringently identified auxin-induced genes, nearly 14% (27 of 198) have both a HUD and an A2-like element within 250 bp of their predicted transcriptional start site; a similar percentage of brassinosteroid-induced genes exhibit this architecture (four of 28). In contrast, only one-half as many cytokinin-induced genes meet this criterion (four of 60). Of the 31 auxin-or brassinosteroid-induced genes with putative bipartite promoter elements (Supplemental Table S2), .60% (19) were identified as BES1 or BZR1 targets in global ChIP studies as compared with ,17% of the whole genome. The results are similarly striking when the list of genes is compared with stringently identified MP targets: 20% (six) are likely bound by MP, compared with 0.4% of the genome as a whole.

DISCUSSION
Genome-wide surveys and computational analyses of auxin and brassinosteroid targets have provided clues into the regulatory landscape required for hormone response (Nemhauser et al., 2004;Priest et al., 2009;Keilwagen et al., 2011;Yamamoto et al., 2011). However, these approaches have a limited ability to define the function of specific cis-elements. Using the promoter region of SAUR15 as a test case, we identified a short region sufficient for both auxin and brassinosteroid response [ Fig. 3B; pS15-1E(i)::LUC]. A detailed analysis of specific cis-elements revealed two critical elements coregulated by both hormones: an AuxRE variant (TGTCTG/CAGACA) and a HUDtype E-box (CACATG/CATGTG). Disruption of either of these cis-regulatory elements resulted in a loss of hormone response.
In this work, we found that despite apparent complexity in quantitative responses to hormone treatment, there is an absolute need for two elements: one predicted to be regulated by auxin and the other by brassinosteroids. This binary requirement for each pathway to be present to allow response to the other can be compared to an AND logic gate. The use of AND logic can lead to more robust pathway function because cells need to receive sustained signals from both pathways to activate gene expression. A recent global transcriptome analysis of the effect of the brassinosteroid biosynthetic inhibitor brassinazole on auxin response showed a striking reduction of auxin response when brassinosteroid levels were compromised (Chung et al., 2011). Enhancement of BES1 and MP/ARF5 binding with application of either hormone (Fig. 3C), combined with evidence that brassinosteroid enhancement of MP/ARF5 requires an intact E-box (Fig. 3D), make a strong case that interdependence of auxin and brassinosteroid signaling reflects joint regulation of promoter activation. Full activation of BES1 and MP/ARF5 require additional pathway-specific events, which could be facilitated by preformation of a poised complex bound to DNA.
Recent evidence of BES1 interaction partners may provide a clue to why auxin response requires the E1 site. BES1 has been shown to interact with proteins that promote transcriptional elongation (Li et al., 2010). In this way, BES1 may be needed to globally prime growth-associated targets by promoting more efficient transcription. Interestingly, the critical mammalian bHLH transcription factor c-Myc may have a parallel function. A recent study of c-Myc targets showed enrichment of E-boxes within 200 bp of the transcriptional start site specifically among genes associated with paused RNA Pol II (Rahl et al., 2010). Interaction with RNA Pol II may also explain why even in plant species with larger average intergenic regions, HUD-type E-boxes and A2-like elements were still found close to transcriptional start sites in putative SAUR15 orthologs (Fig. 4). A model where BES1 acts to boost response to various stimuli is consistent with the role of brassinosteroids in auxin response analyzed here and the observed role of brassinosteroids in enhancing differential cell fate determination during root epidermal development (Kuppusamy et al., 2009).
Our analysis of putative SAUR15 orthologs promoter regions revealed conservation specifically of HUD type E-boxes. HUD elements are overrepresented in phytohormone gene promoters and are believed to play a critical role in the temporal synching of growth-related processes (Michael et al., 2008). Expression of a number of key brassinosteroid biosynthesis and signaling genes show diel cycling (Bancos et al., 2006;Michael et al., 2008). In addition, many genes related to auxin metabolism and response are regulated by the clock (Covington and Harmer, 2007). Clustering of HUD and A2-like elements could facilitate coordination of auxin and brassinosteroid responses to specific times of day.
Auxin promotion of BES1 binding was an unexpected result (Fig. 4), particularly given previous evidence that auxin treatment does not affect the phosphorylation status of BES1 (Yin et al., 2002), and BES1 needs to be dephosphorylated to bind DNA (Vert and Chory, 2006). Phosphorylation of ARF2 by the brassinosteroid-regulated kinase BIN2 was shown to reduce ARF2 binding to DNA (Vert et al., 2008), but little is known about whether posttranslational mod-ifications play a major role in regulating ARF-DNA associations. Each hormone could influence binding of the other pathway's transcription factor through regulation of cofactors. For example, MYB77 has been shown to interact with ARF7 (Shin et al., 2007) and MYB30 has been shown to directly interact with BES1 (Li et al., 2009). An alternative explanation is that each hormone is affecting the level of the other. An auxininduced rise in brassinosteroid levels might be sufficient to trigger the modest increase in binding we observed, yet not sufficiently change phosphorylation status to be detected by western blots. BREVIS RADIX is known to act in just such a secondary cross-regulation loop (Mouchel et al., 2006), linking auxin perception with increased expression of a key enzyme in brassinosteroid biosynthesis. Brassinosteroids could also cause a rise in auxin levels, though for the timelines used here that is less likely. Studies of the auxin-labile Aux/IAAs (Zenser, et al., 2003) suggest that auxin levels are not significantly affected by brassinosteroid treatment.
Another question raised by our results is why the widely used DR5 promoter, which consists only of multimerized TGTCTCs, is responsive to both auxin and brassinosteroids (Nakamura et al., 2003b;Nemhauser et al., 2004). One possible explanation is that the number of AuxRE repeats in the synthetic promoter may bypass the need for whatever function is provided here by the E-box. For example, there may be a change in chromatin structure or strength of induction when multiple ARFs can bind to the same promoter. It is worth noting that few natural promoters show extensive AuxRE arrays. Another interesting possibility is that different ARFs have different requirements for full activation. The SAUR15 promoter examined here was analyzed only in seedlings. There have been reports of different relationships between auxin and brassinosteroids in other cell types, such as mesophyll protoplasts . These differences could be explained by which ARFs or other potential partner proteins are available in specific environments and whether brassinosteroids have any effect on those partners. Brassinosteroids have been shown to induce expression of some ARFs and a number of Aux/IAA corepressors (Goda et al., 2004;Nemhauser et al., 2004;Kim et al., 2006). They have also been implicated in inhibition of DNA binding by repressor ARFs (Vert et al., 2008). These mechanisms could contribute to the brassinosteroid effect on isolated AuxREs, as well as those found in native promoters.
Global transcriptome experiments suggest that there may be multiple classes of auxin-and brassinosteroidresponsive genes (Goda et al., 2004;Nemhauser et al., 2004Nemhauser et al., , 2006, a conclusion supported by this study. The bipartite element identified here likely identifies one class of genes with a distinct mode of auxin:brassinsteroid cross-regulation. By analyzing the functional requirements of elements found in other promoters, further groupings may be found. Investigation of other promoters with predicted bipartite elements, across diverse tissues and under different environmental conditions, is needed to test whether shared architecture leads to a functional module of coregulation. In particular, using genes with a stronger hormone induction than what was observed here for SAUR15 could provide important additional information. This type of directed functional analysis of response elements is essential for "ground-truthing" global data mining efforts and for ultimately understanding gene network dynamics.

Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) ecotype Columbia was used as wild type throughout this study. Several constructs were previously characterized and generously provided by others (SAUR-AC1::GUS, Dr. Pamela Green [University of Delaware]; pMP::MP:GFP seeds, Dr. Dolf Weijers [Wageningen University]; and pBES1::BES1:GFP seeds, Dr. Zhi-Yong Wang [Standard University]). Seeds were surface sterilized for 15 min in 70% ethanol with 0.05% Triton X-100 followed by 5 min in 95% ethanol. After sterilization, seeds were spread onto plates containing 0.53 Linsmaier and Skoog (LS) media (Caisson Laboratories, Inc.) with 0.8% agar and stratified in the dark at 4°C for 2 d. All seedlings were 9 d old at the time of the assays. Seedlings were grown at 20°C in 12-h-light/12-h-dark conditions in 35 mmol m¯2 s¯1 white light, except for T1 seedlings, which were grown in 80 mmol m¯2 s¯1 white light. Plants were transformed by the floral dip method (Clough and Bent, 1998), and transgenic lines were identified by selection on kanamycin (Fisher Scientific).

Quantitative RT-PCR
Light-grown seedlings were immersed in 0.53 LS media (Caisson Laboratories, Inc.) containing mock (0.1% dimethyl sulfoxide; Sigma), 1 mM brassinolide (Chemiclones), or 1 mM auxin (IAA; Bioworld) treatments for 1 and 3 h. Total RNA was extracted by using an RNeasy kit (Qiagen), and first strand cDNA was synthesized by using an iScript cDNA Synthesis kit (Bio-Rad). cDNAs were diluted 20-fold and combined with SYBR master mix (Bio-Rad). Three independent biological replicates were analyzed with a Chromo4 Real-Time PCR machine (MJ Research). Expression for each gene was calculated using the formula (E target ) 2DCPtarget(control 2 sample) /(E ref ) 2DCPref(control 2 sample) (Pfaffl, 2001) and normalized to a reference gene, At1g13320.

Plasmid Construction
All deletion constructs were generated by PCR amplification within the intergenic region between At4g38840 and At4g38850 (SAUR15) from wild-type genomic DNA. Amplified regions were inserted into a pCR-BLUNT vector (Invitrogen), then digested and ligated into pGreen I 0029 (Hellens et al., 2000) containing the LUC+ reporter. Phusion site-directed mutagenesis kit (New England BioLabs) was used to mutate transcriptional elements in pS15-2E(ii). Primers for deletion constructs and site-directed mutagenesis are listed in Supplemental Table S3.

Luciferase Analysis
Seedlings growing on plates were presprayed 16 h prior to assay with 2.5 mM D-luciferin (Biosynth) and 0.01% Triton X-100. At the beginning of the hormone analysis, whole seedlings were transferred from agar plates to 96well Optiplates (PerkinElmer) containing 0.53 LS media. Luciferin and either hormone or mock treatments were subsequently added to each well. Luciferase activity was measured using a Victor V 3 Plate Reader (PerkinElmer). Baseline measurements of luciferase activity for T1 seedlings were taken after 20 min to allow for autoluminescence to decay (measured by examining seedlings lacking reporter constructs). Hormone responses were measured at 3 h. For experiments with T1s, ratios of luciferase activity at 3 h versus baseline were calculated for each individual line. Linear regression was used to test for significant differences in auxin response between constructs. For T2 experiments, significant differences in auxin or brassinosteroid response between lines were identified using a two-way ANOVA followed by a Tukey's honestly significant difference test.

ChIP Assays
Nine-day-old seedlings (pBES1::BES1:GFP and pMP::MP:GFP) were crosslinked in 1% formaldehyde under vacuum. Seedlings were subsequently frozen in liquid nitrogen and ground to a fine powder with mortar and pestle. Samples were resuspended in nuclei extraction buffer [0.25 M Suc, 100 mM MOPS, pH 7.6, 10 mM MgCl 2 , 5% Dextran T-40, 2.5% Ficoll, 20 mM b-mercaptoethanol, and mini-Complete Proteinase Inhibitor tablet (Roche Applied Science)], filtered through Miracloth (Fisher Scientific), and centrifuged to collect nuclei. Nuclei were lysed with Nuclei lysis buffer (50 mM Tris-HCl, pH 8, 10 mM EDTA, and 1% SDS). ChIP dilution buffer was added (1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris, pH 8.0, 167 mM NaCl, and 0.01% SDS), and chromatin was fragmented using a Fisher Dismembrator (Fisher Scientific). An aliquot of fragmented chromatin served as an input control for qPCR analysis, and the remainder was subjected to immunoprecipitation. Dynabeads Protein A (Invitrogen) coupled with anti-GFP (Ab290, Abcam) antibody were used to enrich for pMP::MP:GFP-or pBES1::BES1:GFPcontaining chromatin fragments. Samples were washed and eluted off of Dynabeads using nuclei lysis buffer, and cross-links were reversed by incubating with 300 mM NaCl. DNA was purified using a PCR clean-up kit (Qiagen). ChIP assays were normalized to a region in the ACT2 3#untranslated region, and results are expressed as ratios of qPCR signal to the no antibody IP. ChIP results represent the average of at least two independent biological replicates normalized to the region designated ORF in SAUR15 open reading frame (Fig. 3C). Primers for ChIP qPCR analysis are listed in Supplemental Table S4.

Phylogenetic Tree and Promoter Analysis
Putative orthologs for SAUR15 were identified using the "Peptide Homologs" function of Phytozome v.5.0 (http://www.phytozome.net/) and the "orthologs of other genes" function of PLAZA database (Proost et al., 2009). Five hundred bp of sequence 5# of the coding region was analyzed for E-boxes and TGTCT-like elements. NJplot software (http://pbil.univ-lyon1.fr/software/njplot.html) was used to construct an unrooted phylogenetic tree for aligned coding sequences of putative SAUR15 orthologs using the neighbor-joining method. Multiple alignments of coding sequences were performed using ClustalX (http://www.ebi.ac. uk/Tools/msa/clustalw2/). Conservation of possible 5-mers were calculated using a custom script in MATLAB (MathWorks). Whole-genome analysis was based on TAIR10 annotations (http://arabidopsis.org). Custom scripts (perl, http://www.perl.org/) were used to extract the first 250 bp of upstream regulatory regions for all protein-coding genes and count occurrences of HUD and A2-type elements. Element distributions were compared in R (http://www.r-project.org/) using a general linear model given a Poisson distribution. Promoters were divided into those containing a specific element or not (HUD-containing, non-HUD; A2-containing, non-A2). This categorization was used to determine if the occurrence of one element was a significant factor in determining the likelihood of finding the other element. In both cases (frequency of HUD elements in A2-containing promoters and frequency of A2 elements in HUD-containing promoters), there was a positive, highly significant coefficient (P , 0.0001). All scripts are available upon request.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Baseline expression levels of reporter constructs. Table S1. Arabidopsis genes with both HUD and A2-type elements.

Supplemental
Supplemental Table S2. Auxin-and/or brassinosteroid-induced genes with HUD and A2-type elements.
Supplemental Table S3. Primers for deletion and mutation constructs.
Supplemental Table S4. Primers for ChIP assays.