Diversity of Stability, Localization, Interaction and Control of Downstream Gene Activity in the Maize Aux/IAA Protein Family

AUXIN/INDOLE-3-ACETIC ACID (Aux/IAA) proteins are central regulators of auxin signal transduction. They control many aspects of plant development, share a conserved domain structure and are localized in the nucleus. In the present study, five maize Aux/IAA proteins (ZmIAA2, ZmIAA11, ZmIAA15, ZmIAA20 and ZmIAA33) representing the evolutionary, phylogenetic and expression diversity of this gene family were characterized. Subcellular localization studies revealed that ZmIAA2, ZmIAA11 and ZmIAA15 are confined to the nucleus while ZmIAA20 and ZmIAA33 are localized in both the nucleus and the cytoplasm. Introduction of specific point mutations in the degron sequence (VGWPPV) of domain II by substituting the first proline by serine or the second proline by leucine stabilized the Aux/IAA proteins. While protein half-life times between ∼11 min (ZmIAA2) to ∼120 min (ZmIAA15) were observed in wild-type proteins, the mutated forms of all five proteins were almost as stable as GFP control proteins. Moreover, all five maize Aux/IAA proteins repressed downstream gene expression in luciferase assays to different degrees. In addition, bimolecular fluorescence complementation (BiFC) analyses demonstrated interaction of all five Aux/IAA proteins with RUM1 (ROOTLESS WITH UNDETECTABLE MERISTEM 1, ZmIAA10) while only ZmIAA15 and ZmIAA33 interacted with the RUM1 paralog RUL1 (RUM-LIKE 1, ZmIAA29). Moreover, ZmIAA11, ZmIAA15 ZmIAA33 displayed homotypic interaction. Hence, despite their conserved domain structure, maize Aux/IAA proteins display a significant variability in their molecular characteristics which is likely associated with the wide spectrum of their developmental functions.

In general, Aux/IAA proteins are transcriptional repressors that share a conserved domain structure. Moreover, they are confined to the nucleus [2,8]. Thus far, only AtIAA8 was localized in both the nucleus and the cytosol [9]. Domain I confers transcriptional repression of target genes, once the Aux/IAA and auxin response factor (ARF) protein complex interacts with promoters of downstream genes [10]. The transcriptional repressor function of Aux/IAA proteins arises from the conserved ERF-associated amphiphilic repression motif (EAR) [11], which permits the interaction with the co-repressor protein TOPLESS (TPL) and TPL-related proteins (TPR) [12]. Recent studies revealed that the former domain complex III/IV may form a type I/II Phox and Bem1p (PB1) protein-protein interaction domain [13,14].
Thus far only one Aux/IAA mutant is characterized in maize: the rootless with undetectable meristem 1 (rum1) mutant does not initiate seminal roots and lateral roots in the primary root. The mutation results in a 26 amino acid deletion, which includes domain II. The rum1 gene corresponds to ZmIAA10 [18]. Subcellular localization studies revealed that the wild-type protein RUM1 is localized in the nucleus, whereas the mutated form rum1-R is detected in the nucleus and cytosol due to the partial lack of the nuclear localization signal (NLS). The RUM1 protein has a half-life time of ,22 minutes compared to the very stable rum1-R mutant protein. It was demonstrated that RUM1 interacts with ZmARF25 and ZmARF34 [18]. In the present study, the maize Aux/IAA proteins ZmIAA2 (GRMZM2G159285), ZmIAA11 (GRMZM2G059544), ZmIAA15 (GRMZM2G128421), ZmIAA20 (GRMZM5G864847) and ZmIAA33 (GRMZM2G359924) were characterized to investigate their subcellular localization, repression of downstream gene expression, stability and interaction with other proteins and the variability and specificity of these functions.
These maize Aux/IAA genes were selected because they represent different properties of members of this gene family (Table 1). First, these genes have different phylogenetic ancestries and relationships. Based on their relationship to the unduplicated sorghum genome, ZmIAA2 (subgenome 1) and ZmIAA11 (subgenome 2) emerged before the last whole genome duplication in maize ca. 5-12 million years ago [31], while the non-syntenic genes ZmIAA11, ZmIAA20 and ZmIAA33 emerged more recently by the duplication of individual ZmIAA genes. Moreover, these genes encode both classes of maize Aux/IAA proteins that are separated in the maize phylogenetic tree based on their sequence of the PB1 domain (class A: ZmIAA2, ZmIAA15; class B: ZmIAA11, ZmIAA20, ZmIAA33). Furthermore, these genes display distinct root-type specific gene expression profiles and gene expression kinetics in response to auxin (Table 1). While ZmIAA2, ZmIAA11, ZmIAA20 and ZmIAA33 display root-type specific expression patterns, ZmIAA15 is highly expressed in all root-types. Moreover, expression of ZmIAA2, ZmIAA11 and ZmIAA15 is significantly increased upon 1-NAA treatment within three hours (pattern A), while ZmIAA20 and ZmIAA33 are initially induced by auxin followed by a significant decrease of expression (pattern B, [5] and Table 1).
The goal of the present study was to functionally characterize five selected maize Aux/IAA proteins that displayed different attributes in earlier analyses with respect to stability, localization, protein-protein interaction and control of downstream gene activity and thus to unveil the diversity of these functions in maize.

Plant material and growth conditions
Seeds of the maize inbred line B73 were sterilized with 6% sodium hypochlorite for 10 min and afterwards rinsed in distilled water. Subsequently, the kernels were rolled up in germination paper (Anchor paper, www.anchorpaper.com) and placed in a 10 l bucket filled with ,2 l of distilled water. For RNA extraction, seedlings were germinated at 28 uC with a 16 h light (2700 lux) and 8 h dark cycle for six days in a plant growth chamber (Conviron Adaptis, www.conviron.com). The primary roots of these seedlings were harvested and immediately frozen in liquid nitrogen and stored at 280 uC.
For subcellular localization studies, seedlings were germinated in a plant growth chamber (Conviron Adaptis) at 28 uC with a 16 h light and 8 h dark cycle for three to four days until a 1 cm primary root was visible. Afterwards, the seedlings were transferred into an incubator (Binder, www.binder-world.com) at 26 uC in darkness for another eight to ten days. The etiolated leaves were used for protoplast isolation.

Maize protoplast isolation and transformation
Protoplast isolation of maize leaves was modified according to Sheen [32] and Nguyen et al. [33]. The middle part (6-8 cm) of etiolated 12 to 14 day-old-maize seedlings were cut in 0.5 -1 mm strips. Subsequently, leaf digestion was performed with an enzyme solution (1.5% cellulose RS, 0.3% macerozyme R10, 0.4 M sucrose, 10 mM MES, pH 5.7) in a petri dish as described by Sheen [32]. Washing and recovering of the maize protoplasts was achieved according to Nguyen et al. [33] in 14 ml conical falcon tubes. Afterwards, quality and viability of the protoplasts was determined via the microscope Axio Lab.A1 (Zeiss, www.zeiss.de) with a Fuchs-Rosenthal bright-line objective (Labor Optik, www. lo-laboroptik.de). Protoplasts were resuspended in MMG buffer

Subcellular localization
The C-terminal GFP fusion constructs established for the protein stability determination were also used for subcellular localization experiments in isolated maize leaf protoplasts. For ZmIAA11, ZmIAA11-P241S, ZmIAA11-P245L, ZmIAA20, ZmIAA20-P220S and ZmIAA20-P224L, additional N-terminal GFP fusion constructs were generated. To construct the pUC-35S-GFP-MCS the open reading frame of the GFP without stop codon was amplified with the oligonucleotides GFP-XbaI-F and GFP-SacI-R2 using the pUC-35S-MCS-GFP vector as template (Lab Ac #765). The GFP PCR-products were inserted in the XbaI/ SacI site of the backbone vector. The MCS was generated by PCR amplification (oligonucleotide primers: MCS4-SacI-F and MCS4-SacI-R) with the pUC-SYPNE vector (Lab Ac #519) as template. The MCS was inserted at the SacI site of the recipient vector (Lab Ac #949). The vector sequences were confirmed by sequencing.
The reference plasmid encoding for renilla LUC (Luciferase) and the reporter plasmid encoding for firefly LUC, both driven by a 35S Cauliflower mosaic virus (CaMV) promoter, and the effector plasmid were co-transformed into Arabidopsis protoplast. For the transformation 1 mg plasmid DNA of the reference plasmid, 3 mg of reporter and 5 mg of effector plasmid were used, which were determined by a dose-response assay as described in the Dual-Luciferase reporter assay system manual (Promega, www. promega.de).
To determine the luciferase activities, the Dual-Luciferase reporter assay system (Promega) was used and quantified in a plate reader (CLARIOstar, www.bmglabtech.com). Three biological replicates were measured per Aux/IAA gene and luciferase activity of each transformant was determined three times independently. The values were normalized with the corresponding values of the internal control renilla LUC, which minimized the experimental variability introduced by the variation of the transfection efficiency and the protoplast viability. Differences of the relative luciferase activities were determined by a one-sided Student's t-test. Interaction studies were conducted by bimolecular fluorescence complementation (BiFC) as described by Walter et al. [34]. Fusion plasmids of ZmIAA2, ZmIAA11, ZmIAA15, ZmIAA20 and ZmIAA33 were generated with the C or N-terminal fragment of YFP. Donor plasmids encoding the open reading frame without stop codon of the genes of interest (ZmIAA2: Lab Ac #880, ZmIAA11: Lab Ac #883, ZmIAA15: Lab Ac #887, ZmIAA20: Lab Ac #913, ZmIAA33: Lab Ac #891) were digested with BamHI/XbaI (ZmIAA33), BamHI/XhoI (ZmIAA2, ZmIAA11, ZmIAA15) or SpeI/KpnI (ZmIAA20) to release the Aux/IAA open reading frames. These open reading frames were introduced into the corresponding restriction sites of the recipient vectors pUC-SPYNE-152 ( [34] and Lab Ac #518) and pUC-SPYCE ( [35] and Lab Ac #519). Likewise, interactions were studied with RUM1 (ZmIAA10: GRMZM2G037368) and RUL1 (ZmIAA29: GRMZM2G163848). The split-YFP constructs of RUM1 (Lab Ac #528 and 531) were generated as described in von Behrens et al. [18]. The full-length open reading frame of rul1 was amplified with the oligonucleotide primer rul1-BamHI-fw and rum1-KpnI-rv. The donor vectors pUC-SYPCE and pUC-SYPNE-152 were digested and the PCR product ligated into the restriction site BamHI/KpnI. The sequences of the C-or Nterminal fragment YFP -RUL1 constructs (Lab Ac #529 and 530) were validated via sequencing. As a negative control an uncharacterized protein of the barwin gene family (BARW, GRMZM2G117942) was used. The corresponding C and Nterminal YFP fusion constructs were co-transformed into Arabidopsis protoplasts, according to Berendzen et al. [36] and analyzed by flow cytometry. In total 76 different samples were measured in three replicates. Per 96 well plate one biological replicate of each sample was measured.
A linear model was fitted with an additive effect for treatment, corresponding to the protein interaction partners. In order to normalize the data between different plates, an additive effect for plates was fitted. Inspection of the studentized residuals suggested that a logarithmic transformation stabilized the variance and led to an approximate normal distribution. Thus, all analyses were performed on the logarithmic scale. Note that an additive model on the transformed scale corresponds to a multiplicative model on the original scale. According to this model, responses were expected on the original scale of the different treatments on one plate to be a multiple of the responses on another plate. This expectation also met with the observation.
Each interacting protein pair was compared to the mean of its two controls by specifying the corresponding linear contrast. In this contrast, the protein interacting pair was assigned a coefficient of 2, whereas the two controls were each assigned coefficients of -1. T-tests were performed for all contrasts of interest. The p-values were adjusted for multiplicity using the simulation method of Edwards and Berry [37]. The family-wise type I error rate was set at a = 5%.
To confirm the interaction results obtained from the quantitative BiFC assays, split-YFP experiments were performed in maize protoplasts. To confirm the expression of the proteins, Western blot experiments were performed as previously described [18].

Stability of maize Aux/IAA proteins is very variable
The rapid degradation of Aux/IAA proteins in response to high auxin levels is crucial for auxin signal transduction and thus controls the activity of downstream gene activity. To compare the stability kinetics of the five selected maize Aux/IAA proteins, Cterminal green fluorescent protein (GFP) fusions were generated. Two different point mutations were introduced into the degron sequence (VGWPPV) of the Aux/IAA proteins. These mutations resulted in a substitution of the first proline (P) by serine (S), or the exchange of the second proline (P) by leucine (L). In Arabidopsis, it was demonstrated that these specific substitutions resulted in more stable proteins compared to the wild-type [19,38]. The relative GFP fluorescence intensity was determined over a time period of two hours in Arabidopsis protoplasts by flow cytometry, after treatment with the synthetic auxin 1-NAA and cycloheximide. As a control, the GFP fluorescence intensity of free GFP was measured. For each of the five Aux/IAA proteins, different average half-life times were determined: ZmIAA2-GFP had an average half-life time of ,11 minutes, ZmIAA20-GFP and ZmIAA33-GFP displayed a half-life time of ,40 minutes and ZmIAA11-GFP and ZmIAA15-GFP displayed the longest average half-life times of ,60 and ,120 minutes, respectively ( Figure 1). As a consequence of the substitution of the amino acid P to S or L in the degron sequence, all modified Aux/IAA-GFP proteins were significantly more stable than their respective wild-type protein during the whole time course of 120 minutes. Most mutated Aux/ IAA-GFP proteins appeared slightly less stable than the GFP control, as indicated by a minimal decrease of the measured GFP fluorescence intensity over time. However, these values were not significantly different from the GFP control. The degradation of the Aux/IAA-GFP proteins was independently confirmed by Western blot experiments ( Figure S2).

Subcellular localization of Aux/IAA proteins
Aux/IAA proteins are transcriptional repressors and therefore act in the nucleus. To survey the subcellular localization of the selected Aux/IAA proteins, C-terminal GFP-fusion proteins of wild-type and the two mutated forms were generated for all five Aux/IAA genes and transiently expressed in maize mesophyll protoplasts. The wild-type GFP-fusion proteins of ZmIAA2-GFP, ZmIAA11-GFP and ZmIAA15-GFP were exclusively accumulated in the nucleus. In contrast, ZmIAA20-GFP and ZmIAA33-GFP were localized in both the nucleus and the cytosol in maize protoplasts ( Figure 2). Both, ZmIAA20 and ZmIAA33 have an incomplete NLS lacking the amino acids KR in the first part of the bipartite NLS ( Figure S1) which might explain their subcellular localization pattern. ZmIAA11-GFP and ZmIAA20-GFP displayed a compartmentalized expression in the nucleus (Figure 2). To investigate if the compartmentalization in the nucleus is a consequence of the C-terminal GFP fusion, N-terminal GFPfusion proteins of wild-type and their mutated forms were generated. Identical results, i.e. compartmentalized GFP signals, were observed for GFP-ZmIAA11 and GFP-ZmIAA20. Moreover, the N-terminal GFP-fusion GFP-ZmIAA20 displayed the same localization in both nucleus and cytoplasm as the C-terminal GFP-fusion ZmIAA20-GFP. All mutated proteins displayed the same subcellular localization as the five wild-type proteins ( Figure  S3).

All five Aux/IAA genes are active repressors of downstream gene expression
Transcriptional repression of Aux/IAA proteins is mediated by domain I. To investigate the regulation of downstream genes by the five maize Aux/IAA proteins, the Dual-Luciferase reporter assay was applied in Arabidopsis protoplasts. This system allows to test if proteins containing a GAL4-DBD (DNA Binding Domain), Fusion proteins containing a N-terminal yeast GAL4 DNAbinding Domain and full-length wild-type Aux/IAA proteins at the C-terminus were used as effectors (Figure 3). The reporter plasmid encoded firefly luciferase (LUC) proteins driven by a minimal TATA box of the CaMV 35S promoter combined with an upstream GAL4-UAS binding site. Arabidopsis protoplasts were co-transformed with three plasmids including the reference plasmid encoding for renilla LUC, a reporter plasmid encoding for firefly LUC and either the control GAL4-DBD effector plasmid or effector plasmids encoding for one of the five GAL4-DBD-ZmIAA proteins. Relative firefly luciferase activity of the Aux/IAA proteins was measured and quantified relative to the control GAL4-DBD protein. All GAL4-DBD-ZmIAA proteins exerted a significant reduction of relative luciferase activity on the downstream reporter gene compared to the GAL4-DBD control (Figure 3). GAL4-DBD-ZmIAA2 and GAL4-DBD-ZmIAA15 displayed the strongest reduction, with only ,30% residual relative luciferase activities. Moreover, the relative luciferase activity was ,36% for GAL4-DBD-ZmIAA11 and ,48% for GAL4-DBD-ZmIAA33, whereas GAL4-DBD-ZmIAA20 displayed ,77% activity compared to GAL4-DBD control (Figure 3). These results suggest a significant transcriptional repression of downstream genes by the tested Aux/IAA proteins but also a considerable variability between the repressor function of different Aux/IAA proteins.

Homo and heterointeraction of maize Aux/IAA proteins including RUM1 and its paralog RUL1
The PB1 domain complex of Aux/IAA proteins allows interaction with other Aux/IAA proteins or auxin response factors (ARFs). To examine homotypic interactions of the five Aux/IAA proteins surveyed in the present study, quantitative bimolecular fluorescence complementation (BiFC) experiments were performed by flow cytometric analysis of transformed protoplasts. Subsequently, the results obtained in flow cytometric analyses were independently confirmed by analyzing BiFC interactions in individual maize protoplasts under a Zeiss LSM 780 (attached with Axio Observer Z1, Zeiss) confocal laser scanning microscope. Homotypic interaction of the Aux/IAA proteins ZmIAA11, ZmIAA15 and ZmIAA33 was detected in quantitative FACS experiments ( Table 2) and confirmed in subsequent confocal laser scanning microscopic studies of individual protoplasts ( Figure 4A). Homotypic interactions of ZmIAA2 and ZmIAA20 were neither detected by flow cytometry nor when analyzing BiFC interactions in individual maize protoplasts in a confocal laser scanning microscope (Table 2). Moreover, interaction of the five maize Aux/IAA proteins studied here was tested with the previously characterized paralogous RUM1 (ZmIAA10) and RUL1 (ZmIAA29) proteins involved in maize lateral and seminal root formation. Interactions were detected for all five Aux/IAA proteins with RUM1, whereas only ZmIAA15 and ZmIAA33 interacted with RUL1 ( Figure 4B and Table 2). These interactions were validated quantitatively by flow cytometry and confirmed in individual protoplasts by confocal imaging. None of the negative controls using the BARW protein, which does not interact with Aux/IAA proteins, revealed an interaction with any of the five Aux/IAA proteins ( Figure S4). Expression of the proteins in the interaction studies was confirmed by Western blot analyses ( Figure  S5). A summary of all observed Aux/IAA interactions is provided in Figure 4C).

Discussion
The maize genome contains 34 Aux/IAA genes [5], which encode plant-specific transcriptional regulators of auxin signaling [39]. Canonical Aux/IAA proteins consist of domains I, II and PB1 and contain a bipartite NLS [40]. The five Aux/IAA proteins characterized in the present study (ZmIAA2, ZmIAA11, ZmIAA15, ZmIAA20, and ZmIAA33) display the canonical domain structure.
Domain II of Aux/IAA proteins, contains a degron sequence, which confers the rapid degradation of Aux/IAA proteins by interacting with the SCF TIR1 complex [41]. As a consequence, Aux/IAA proteins are short-lived proteins with a half-life time of ,6 to 8 min in pea [42] and ,10 to ,80 min in Arabidopsis thaliana [38,43]. Specific point mutations in the short degron sequence VGWPPV stabilizes these Aux/IAA proteins [19,28,38,44]. The half-life time of maize RUM1 (ZmIAA10) has been previously determined as ,22 min [18]. In the present study, half-life times of maize Aux/IAA proteins ranged from ,11 min (ZmIAA2), to ,120 min (ZmIAA15) in Arabidopsis protoplasts. ZmIAA20 and ZmIAA33 displayed a half-life time of ,40 min. Both genes are closely related to each other and display 77% identity in their amino acid sequence. Similarly, Arabidopsis Aux/IAA proteins, that share a high similarity outside the conserved domain II, displayed similar half-life times [44], suggesting that the degron sequence is required for the degradation of Aux/IAA proteins but that the half-life time of these proteins is also determined by sequences outside the degron motif [44]. The half-life time of ZmIAA15 (,120 min) exceeded the most stable Arabidopsis Aux/IAA proteins AXR3/IAA17 [28] and IAA28 at ,80 min [44]. The difference in stability of the different maize Aux/IAA proteins tested here might be a result of tissue-or organ-specific regulation by yet unidentified proteins or unidentified stabilons of degrons outside of domain II [44]. However, these significant differences in maize Aux/IAA protein half-life times, should rather be considered as relative than absolute because the data was obtained in heterologous Arabidopsis protoplasts. In line with the predicted molecular function of the degron motif, point mutations, substituting the first P by S or the second P by L, in the sequences of the five Aux/IAA proteins analyzed here resulted in proteins that displayed almost no degradation (,10%) over a time period of 120 min. Similar results were described for the mutated maize ZmIAA10 protein rum1-R in which the degron sequence was completely deleted [18].
Commonly, Aux/IAA proteins hold a bipartite nuclear localization signal (NLS), which directs them into the nucleus. Nuclear localization was observed for the Aux/IAA proteins ZmIAA2, ZmIAA11 and ZmIAA15 and their mutated forms. Previously we demonstrated that the RUM1 protein (ZmIAA10) is also localized in the nucleus [18]. Similar results were described for the Aux/IAA proteins AXR2-GUS and AXR3NT-GUS in Arabidopsis, which were localized in the nucleus [28,38]. Moreover, nuclear localization was observed for the Arabidopsis Aux/IAA proteins IAA26 [45], AXR2, MSG2, SLR [46], BDL [47] and IAA17 [28,48]. In contrast, ZmIAA20 and ZmIAA33 were not exclusively expressed in the nucleus but also in the cytosol. This might be a result of the two missing conserved amino acids KR of the first part of the NLS. In Arabidopsis, only one Aux/IAA protein, AtIAA8, was thus far localized in the cytosol and nucleus [9]. A function of Aux/IAA proteins in the cytosol has been described recently by the interaction of LSD1 (LESIONS SIMULATING DISEASE RESISTANCE 1) with AtIAA8 [49]. LSD1 is a cytosolic protein which negatively controls cell death and disease resistance. Furthermore, it sequesters the transcriptional factor AtbZIP10 [50]. Hence, ZmIAA20 and ZmIAA33 might not exclusively function in the transcriptional regulation of auxin signaling but may also play a role in cytosolic processes.
All Aux/IAA proteins analyzed in the present study repressed downstream gene expression as demonstrated by a dual luciferaseassay. Relative luciferase activity was repressed between ,23% (GAL4-DBD-ZmIAA20) and ,70% (GAL4-DBD-ZmIAA2 and GAL4-DBD-ZmIAA15) by the Aux/IAA effectors in comparison to the GAL4-DBD control. Previously we demonstrated that maize RUM1 (ZmIAA10) exerted a transcriptional repression of ,59% on downstream target genes [18]. These values should rather be considered relative than absolute because these experiments have been performed in heterologous Arabidopsis protoplasts. In Arabidopsis, the repressor function of domain I of sixteen Aux/IAA proteins was determined by GUS reporter assays. These Aux/IAA proteins repressed the GUS reporter gene between 51% to 89% relative to the control [51]. In Arabidopsis, a correlation was observed between Aux/IAA protein stability in wild-type versus mutant proteins and their repression of early auxin response genes. Stabilizing mutations in domain II, resulted in enhanced repression of downstream targets by these proteins. In contrast, mutations in domain I and III were able to partially reverse repressor function [51].
Protein-protein interaction studies revealed homo-and heterotypic interaction of maize Aux/IAA proteins. Homotypic interaction of ZmIAA11, ZmIAA15 and ZmIAA33 was determined by BiFC experiments in Arabidopsis protoplasts which were quantified by flow cytometry and subsequent independent verification by confocal microscopy in maize protoplasts ( Figure 4A and Table 2). In yeast-two hybrid experiments, homotypic interactions of Arabidopsis IAA1 and IAA2 and IAA4 of pea were described [52]. Protein-protein interaction of five Aux/IAA proteins with RUM1 (ZmIAA10) and RUL1 (ZmIAA29) were quantified in Table 2. Statistical evaluation of BiFC data using a lineal model for logarithmized data and simulation-based multiple comparisons by Edwards and Berry [37].   Table 2) and the verification of these results by qualitative BiFC experiments surveying protoplasts under a confocal microscope ( Figure 4A and B BiFC experiments by flow cytometry in Arabidopsis protoplasts. Independent confirmation was performed by confocal microscopy in maize protoplasts ( Figure 4B and Table 2). Hetero-interactions were observed for all Aux/IAA proteins tested with RUM1, while only two of five ZmIAA proteins (ZmIAA15 and ZmIAA33) interacted with RUL1. The Aux/IAA gene rul1 is expressed ,84 fold higher than rum1. Likewise ZmIAA15 is one of the highest expressed maize Aux/IAA gene in contrast to ZmIAA2, ZmIAA20 and ZmIAA33 which display relatively low expression levels [5]. However, the expression levels of the different Aux/IAA genes were not correlated with the interaction between them suggesting specificity of Aux/IAA interactions irrespective of their expression levels. In yeast-two-hybrid experiments, Arabidopsis IAA1 interacted with 15 different Aux/IAA proteins [52]. It was suggested that several hundred interactions might be responsible for the diversity of physiological and morphological responses of these transcription factors based on the observation of differential spatial and temporal expression patterns of Aux/IAA genes in plants [52].
The functional characterization of the Aux/IAA proteins revealed a wide variability and specificity of different molecular functions of Aux/IAA proteins. In combination with specific expression patterns in root and shoot tissues, these results support the notion of a wide variety of Aux/IAA protein functions in plant growth and development.

Supporting Information
Figure S1 Alignment of the maize Aux/IAA protein sequences relevant for this study, including RUM1 (ZmIAA10) and RUL1 (ZmIAA29). The protein sequences were compared by multiple alignments with ClustalOmega. The canonical domains are highlighted by grey shading. The bipartite nuclear localization signal is encircled in red. (PDF) Figure S2 Confirmation of Aux/IAA degradation by Western blot experiments. Western blot analysis of Aux/IAA-GFP constructs after 1-NAA (10 mM) and cycloheximide (100 mg/ml) treatment using an anti-HA antibody. Protein abundance was quantified at three different time points (0, 30 and 120 min). Lane order for ZmIAA15 was manually rearranged to correct for the initial omission of the 30 min sample which was loaded on the same gel in a lane to the right. (PDF) Figure S3 Subcellular localization studies of maize Aux/IAA proteins and their mutated forms in maize protoplasts. Detailed summary of the subcellular localization studies of maize Aux/IAA wild-type proteins and two mutated protein forms of each protein in maize protoplasts. As a control, the empty GFP construct constitutively expressing GFP, was localized in the cytosol and nucleus. ZmIAA2, ZmIAA11 and ZmIAA15 were confined to the nucleus. The specific point-mutations in the degron-sequence did not affect localization. For ZmIAA20 and ZmIAA33, wild-type and mutated proteins were localized in the cytosol and the nucleus. To investigate if the compartmentalization of the GFP signal is a result of C-terminal GFP, N-terminal GFP fusion constructs of ZmIAA11 and ZmIAA20 with their mutated forms were tested. The localization study displayed an accumulation in a compartmental manner in the nucleus as described before. (PDF) Figure S4 Negative controls for protein-protein interaction studies in maize protoplasts. In maize protoplasts split-YFP experiments were conducted to demonstrate that Aux/IAA proteins do not interact with the control protein BARW. Red: auto fluorescence, blue: DAPI counterstaining. (PDF)