Proteomic Identification of Putative MicroRNA394 Target Genes in Arabidopsis thaliana Identifies Major Latex Protein Family Members Critical for Normal Development*

Expression of the F-Box protein Leaf Curling Responsiveness (LCR) is regulated by microRNA, miR394, and alterations to this interplay in Arabidopsis thaliana produce defects in leaf polarity and shoot apical meristem organization. Although the miR394-LCR node has been documented in Arabidopsis, the identification of proteins targeted by LCR F-box itself has proven problematic. Here, a proteomic analysis of shoot apices from plants with altered LCR levels identified a member of the Latex Protein (MLP) family gene as a potential LCR F-box target. Bioinformatic and molecular analyses also suggested that other MLP family members are likely to be targets for this post-translational regulation. Direct interaction between LCR F-Box and MLP423 was validated. Additional MLP members had reduction in protein accumulation, in varying degrees, mediated by LCR F-Box. Transgenic Arabidopsis lines, in which MLP28 expression was reduced through an artificial miRNA technology, displayed severe developmental defects, including changes in leaf patterning and morphology, shoot apex defects, and eventual premature death. These phenotypic characteristics resemble those of Arabidopsis plants modified to over-express LCR. Taken together, the results demonstrate that MLPs are driven to degradation by LCR, and indicate that MLP gene family is target of miR394-LCR regulatory node, representing potential targets for directly post-translational regulation mediated by LCR F-Box. In addition, MLP28 family member is associated with the LCR regulation that is critical for normal Arabidopsis development.

In eukaryotes, ubiquitination is a post-translational regulatory process that controls the level and/or activity of numerous proteins. Proteins destined for degradation via ubiquitination are covalently conjugated with ubiquitin, a small globular protein that serves as a tag for proteolysis in the 26S proteasome (1)(2)(3). Ubiquitination is essential for rapid physiological responses to both internal molecular and external environmental signals, in order for a specific cell or tissue to quantitatively and qualitatively modulate the pool of proteins that make up the proteome at any given time. In the model plant Arabidopsis thaliana (Arabidopsis), almost 6% of its known proteome, corresponding to the gene products encoded by 1600 Arabidopsis loci, is predicted to be involved in the ubiquitination-proteasome system (4). It is the multisubunit E3 ubiquitin ligase that directs the specificity of ubiquitination, and in Arabidopsis ϳ700 genes are predicted to encode the F-box subunit, revealing the biological significance of this component of the ubiquitin pathway for post-translational gene expression regulation in Arabidopsis (5,6).
F-box proteins are central components of a variety of protein complexes. The Suppressor of Kinetochore Protein 1 (SKP1)-Cullin (CUL)-F-Box (SCF) 1 complexes are the largest and best-characterized group of multisubunit RING domain E3 ubiquitin ligases (7,8). The F-box is bound to SKP1 via its highly conserved amino (N)-terminal 'F-box' motif, and to its target protein(s) or substrate(s) via its carboxyl (C)-terminal domains (9). The C-terminal region of each F-box protein consists of multiple domains, including WD40 domains, leucine-rich repeats (LRRs), and Kelch repeats (5) in an array of combinations, and it is the C-terminal structural variability that confers F-box target specificity.
In plants, the characterization of a small number of F-box proteins has revealed that they play functional roles in diverse cellular processes, including regulating responses to hormones and mediating pathogen defense as well as directing other essential physiological and developmental processes, including circadian rhythm and flowering time (10,11). The importance of F-box proteins to plant growth and development is further evidenced by the expression of several F-boxencoding genes under additional post-transcriptional regulation by a class of small regulatory nonprotein-coding RNAs, termed microRNAs (miRNAs). A number of molecular approaches have shown that the expression levels of F-box protein encoding genes, Transport Inhibitor Response 1 (TIR1), Leaf Curling Responsiveness (LCR) and More Axillary Growth 2 (MAX2) are regulated via miR393-, miR394-, and miR528directed mRNA-cleavage, respectively (12)(13)(14).
In Arabidopsis, miR394 is processed from two precursor transcripts, PRI-MIR394A and PRI-MIR394B (15). It has been shown to be involved in vasculature and leaf patterning formation (13), and crucial for shoot apical meristem (SAM) stem cell maintenance and competence, by post-transcriptional repression of its F-box target gene LCR (16), and more recently demonstrated to be involved in abiotic stress responses (17). The miR394-directed spatiotemporal regulation of LCR is crucial for normal SAM development as the inhibition of miR394 activity in the SAM has severe developmental consequences, including downward leaf curvature and eventual meristem termination (13). Although the miR394-LCR node has been well documented in Arabidopsis, the identification of proteins targeted by LCR F-box itself has proven problematic. It has previously been shown that LCR interacts with the well-known SAM stem cell feedback regulators Wuschel (WUS) and Clavata 3 (CLV3); however, neither WUS nor CLV3 is believed to be a direct target of LCR F-box (16).
Here, we used a liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS)-based approach to assess the proteome landscape of the SAM tissue of Arabidopsis plant lines that had been modified to have either elevated (LCR over-expression plants, LCR-OE) or reduced (LCR knockdown plants, LCR-KD) LCR levels. This quantitative MS approach identified a number of differentially accumulating proteins, including Major Latex Protein 28 (MLP28), in the modified plant lines. In silico, functional and molecular approaches suggested MLP28 as a LCR F-box target and identified other members of the MLP gene family as putative targets of LCR-mediated post-translational gene expression regulation. Transgenic plants were produced that expressed artificial miRNAs (amiRNAs) targeted against MLP28. These amiRNA-MLP28 lines had diminished MLP28 levels and displayed severe developmental defects, including elongated petioles, leaf morphology and shoot apex alterations, dwarfness and eventual premature death. These phenotypes resemble those of Arabidopsis plants modified to over-express LCR. Taken altogether, the data presented here show that the F-Box LCR mediates the degradation of MLP proteins and that the miR394-LCR node is associated with MLP28, to play a role for normal Arabidopsis development.

EXPERIMENTAL PROCEDURES
Plant material and growth conditions-Arabidopsis thaliana (Arabidopsis) ecotype Columbia-0 (Col-0) was used the wild-type (WT) background. Identification of Arabidopsis T-DNA insertion mutant lines was carried out using the Arabidopsis gene mapping tool T-DNA express (http://signal.salk.edu/cgi-bin/tdnaexpress) (18). Transformant lines were generated by floral dipping (19) using Agrobacterium tumefaciens (Agrobacterium, strain GV3101)-mediated transformation. Seeds were placed on selective Murashige and Skoog (MS) media that contained the appropriate selective agent to identify primary transformants (T0). T1 lines that segregated at a ratio of ϳ3:1 on selective plates were transferred to soil for seed collection. Homozygous T2 lines were identified on selective agar plates and confirmed homozygous via standard PCR-based genotyping and used for further phenotypic and molecular analyses. For all Arabidopsis lines used in this study, seeds were stratified via a 48-hour incubation at 4°C and were subsequently cultivated at 21-23°C under a 16-hour light/8-hour dark day-night cycle.
Transient Agro-infiltration-Agrobacterium infiltration (Agro-infiltration) method was conducted according to previously described (20). Transient expression assays were performed with 4-week-old wildtype Nicotiana benthamiana plants that were cultivated under standard glasshouse conditions of at 22-23°C and 16-hours of light and 8-hours of dark. Equal volumes of Agrobacterium cultures, each containing the desired binary plasmid, were mixed prior to co-infiltration. Final dilution of cultures used in co-infiltration assays was 0.33OD, and each Agro-infiltration experiment was performed three times and infiltrated leaves were assessed after 3 days of transient expression.
Plasmid constructs-MiR394 overexpression constructs were generated by amplification of 200bp of the precursor of MIR394B, flanking mature miRNA sequences, and fused to Cauliflower mosaic virus (CaMV) 35S promoter (35Spro) presented in the pART7 vector (21). Subsequent cloning steps were performed by excision of the 35Spro/ miRNA overexpression sequence/terminator fragment using NotI restriction digestion, ligation into similarly digested vector pART27, and generation of plant expression vector (21). MiR394 sponge constructs were generated by GeneArt Gene Synthesis (Life Technologies, Carlsbad, CA/USA). The STTM format followed the design of Yan et al. (2012), and construct designs are illustrated in supplemental Fig. S1. Nucleotide sequences were designed to target miR394, and contained additional spacer nucleotides and 5Ј XhoI and 3Ј XbaI restriction sites, for subsequent cloning steps in pART7 and pBART, generating plant expression vector (22). MiRNA-resistant LCR transgene (LCR-OE) was generated by site-directed mutagenesis using the QuikChange Lightning Multisite-Directed Mutagenesis Kit according to the manufacturer's instructions (Agilent Technologies, Santa Clara, CA/USA). The modified sequence was subsequently cloned into the plant expression vector pBART.
Gateway-compatible plant transformation pEarleyGate201, obtained from TAIR was used to epitope tag (HA; hemagglutinin) the N terminus of each protein of interest (23). Amplicons of the LCR and TIR1 coding sequences were fused in frame to the Cauliflower mosaic virus 35S promoter (35Spro) housed in the shuttle vector pART7. The resulting 35S promoter/gene of interest fragments were subsequently cloned into the plant expression vector pBART. The pBART vector alone (no inserted sequences) used as the empty vector control and the GUS plant expression vector (35Spro::GUS; pBART housing a 35S promoter-GUS fragment) was used as Agrobacterium infiltration internal control.
Gateway-compatible plant transformation pSITE-YFP vectors, obtained from TAIR, were used to express fusions to monomeric YFP for the expression in plant cells (24,25). The N-terminal region of YFP was fused with MLP28 and MLP423 and the YFP C terminus was next fused with LCR F-box. Bimolecular fluorescence complementation (BiFC) were performed as previously described (26). Plants agroinfiltrated for the BiFC assay were assessed for YFP expression under the stereo fluorescence microscope with FITC and YFP filter sets (Ste-REO Lumar V12, SteREO Lumar V12, Zeiss, Oberkochen, Germany).
␤-Glucuronidase (GUS) reporter gene constructs were generated by PCR amplification of genomic fragments, immediately upstream of the start codons of MIR394A (4 Kb), MIR394B (4 Kb), LCR (1.6 Kb, plus additional 1.1 Kb downstream of ATG sequence), MLP28 (1.6 Kb) and MLP423 (1.8 Kb), which were subsequently cloned into the pRITA::GUS vector (21). Subsequently, the generated promoter::GUS fragment was cloned into the plant expression vector pBART. Staining for GUS activity determination was performed as previously described (27). Images of GUS stained plants were obtained using the stereo fluorescence microscope under white light.
Artificial miRNAs were generated to direct RNA silencing of MLP28 using the pBlueGreen vector system that includes the MIR159B primary miRNA coding sequence fused to the 35S promoter (28).
All generated plant expression vectors were used to stably transform wild-type Arabidopsis plants via Agrobacterium-mediated transformation. All oligonucleotides used in this study to generate plant expression vectors via a PCR-based cloning approach are all listed in supplemental Table S1.
Mass spectrometry analysis-Proteins were extracted from shoot apices dissected from Arabidopsis seedlings [ϳ8 plants per biological replicate; wild-type Arabidopsis and LCR-overexpression (OE) and LCR-kockdown (KD) transformant lines] as described previously (29). Briefly, apices were tip-probe sonicated (3 ϫ 30 s) in 8 M urea, 1% triton X-100 in 100 mM triethylammonium bicarbonate (TEAB) (pH 7.5) and centrifuged at 13,000 ϫ g for 10 min. The supernatant was precipitated with 20% trichloroacetic acid overnight at Ϫ20°C and the protein pellets washed briefly with ice-cold acetone. Proteins were resuspended in 8 M urea in 100 mM TEAB (pH7.5), reduced with 10 mM dithiothreitol for 60 min at room temperature and alkylated with 25 mM iodoacetamide for 60 min at room temperature in the dark. The reaction was diluted 5-fold with 100 mM TEAB and digested with trypsin (1:50 trypsin/protein) overnight at 37°C. Peptide preparation, stable isotope labeling with isobaric tags for relative and absolute quantitation (iTRAQ; AB Sciex, Framingham, MA), peptide fractionation, and nano-reverse phase liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-MS/MS) was performed as described previously (30). Briefly, 100 g of peptide was labeled with iTRAQ according to the manufacturer's instructions and desalted with hydrophilic-lipophilic balance solid phase extraction (Waters; Milford, MA). The peptides were resuspended in 90% acetonitrile, 0.1% TFA and fractionated into 10 -11 fractions on an in-house packed amide-HILIC column (320 m x 17 cm with m particles; Tosho, Tokyo, Japan) using an Agilent 1200. The gradient was 90 -60% acetonitrile containing 0.1% TFA over 30 min at 6 l/min. Each fraction was resuspended in 0.1% formic acid and separated on an in-house packed C18AQ column (75 m x 17 cm with 3 m particles; Dr Maisch, Ammerbuch,Germany) using an Easy nLC-II. The gradient was 0 -30% acetonitrile containing 0.1% formic acid over 120 min at 250 nl/min. The separation was coupled to either an LTQ-Orbitrap XL or LTQ-Orbitrap velos. For LTQ-Orbitrap XL analysis, an MS1 precursor scan was measured at 400 -1600 m/z (30,000 resolution and 1e 6 AGC) followed by data-dependent MS2 analysis by both LTQ-CID (35 NCE and 3e 4 AGC) and Orbitrap-HCD (55 NCE, 7500 resolution and 4e 5 AGC) of the top three most intense ions. For LTQ-Orbitrap velos analysis, a similar data-dependent acquisition was performed except the top seven most intense ions were analyzed by Orbitrap-HCD only (48 NCE,7,500 resolution and 4e 4 AGC). An additional biological replicate was performed with 3-plex dimethyl labeling essentially as described previously (31). These peptides were fractionated by amide-HILIC and analyzed on an LTQ-Orbitrap XL essentially as described above excepted data-dependent acquisition was performed on the top ten most intense ions with LTQ-CID. All resulting raw data were processed using Proteome Discoverer v1.4beta (Thermo Fisher Scientific, Waltham, MA/USA) and searched against the Arabidopsis TAIR-10 database (35,386 entries) with SequestHT. The parameters included a tolerance of 20 ppm for MS1 and 0.02 Da for HCD-MS/MS or 0.6 Da for CID-MS/MS. The data was searched with Met oxidation as a variable modification and Cys carbamidomethylation as a fixed modification with either fixed modification of iTRAQ peptide N terminus and Lys or, fixed modification of 3-plex demethylation of peptide N terminus and Lys (in three separate searches). The enzyme was full trypsin cleavage and all results were filtered to 1% FDR with Percolator (32). Proteins were normalized to the median of all peptide spectral matches and statistical analysis was conducted in Perseus. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (33) with the data set identifier PXD002390.
Experimental Design and Statistical Rationale-For the proteomic analysis of plant tissue, a total of three biological replicates were performed where each replicate consisted of eight pooled dissected shoot apices from Arabidopsis transformant seedlings. We investigated differentially regulated proteins between both over-expression and knockdown of LCR relative to wild-type. We hypothesized that protein(s) regulated by LCR would show both decreased expression with LCR overexpression and increased expression with LCR knockdown and, these proteins would be prioritized for further validation. Because of available starting material and resources, two biological replicates were performed with iTRAQ and one additional biological replicate was performed with 3-plex dimethylation. Identification of significantly regulated proteins was performed by analyzing the two iTRAQ replicates with a one-sample t test corrected for multiple testing with a Benjamini Hochberg FDR of 1%. Significantly regulated proteins were further filtered to display only very high confident hits to validate which displayed Ͼ50% fold-change in both biological replicates and displayed Ͼ50% regulation in the third dimethylation biological replicate. Although we hypothesized LCR-dependent regulation would result in differential abundance between overexpression and knockdown and this would enable efficient stratification of targets for validation, the observed reciprocal response was not robust and there were very little proteome changes. There are numerous reasons why this reciprocal regulation was not observed. For example, overexpression of an E3 ligase may not necessarily result in robust down-regulation of target proteins because other cofactors or the proteasome itself may be rate-limiting. However, because the candidate list of proteins was small, we attempted validation for a number of targets.
Western Blot Analysis-Proteins were extracted in extraction buffer [100 mM Tris HCl pH 8.0, 0.04% DTT, 10% glycerol, and protease inhibitor tablet cOmplete ULTRA Tablets (Roche, Basel, Switzerland)] and concentration was determined using the Protein Assay kit (Bio-Rad, Hercules, CA/USA). SDS-PAGE polyacrylamide midi gels Nu-PAGE Bis-Tris (Life Technologies) were used to separate 10 g of total protein under denaturing conditions. Western blotting was conducted by probing electroblotted nitrocellulose membranes with monoclonal anti-HA antibody (1:10,000; Sigma-Aldrich, St. Louis, MO/USA) or anti-␤-glucuronidase (N-Terminal) antibody (1:5000; Sigma-Aldrich) and a shared host-specific secondary antibody. Bands were visualized with the Western Lightning Plus ECL (Perkin-Elmer, Waltham, MA/USA). Assays were repeated at least 3 times per each analyzed MLP-HA tagged protein.
Quantitative RT-PCR (RT-qPCR) Analysis-Total RNA was extracted from frozen plant material using TRIzol Reagent according to the manufacturer's instructions (Life Technologies). For RT-qPCR, 5 g of DNase (Promega, Fitchburg, WI/USA)-treated total RNA was used for first-strand cDNA synthesis with oligo (dT) 23 primer and reverse transcription with the Superscript III reverse transcriptase (Life Technologies) according to the manufacturer's protocol. RT-qPCR was carried out on an Mx3000P instrument (Agilent Technologies), and Brilliant II SYBR Green QPCR Master Mix (Agilent Technologies) was used for the three-step cycling reactions, following the manufacturer's instructions. All RT-qPCR reactions (for both reference and specific genes) were carried out in biological and technical triplicate. Cyclophilin (Cyclophilin 5; AT2G29960) was used to normalize gene expression using the comparative quantification program and data was analyzed with MxPro QPCR Software (Agilent Technologies).

RESULTS
In Silico Analysis of the Structure of LCR F-box-like Proteins-Extensive BLASTP searches of publically available plant proteome data sets (http://blast.ncbi.nlm.nih.gov/Blast. cgi) (34), using the 467-amino acid (aa) sequence of the Arabidopsis LCR F-box as the query, generated an extensive list of putative LCR F-box-like proteins. This initial list was reduced to 11 unique full-length LCR-like proteins (Table I), and each contained an N-terminal F-box motif with similarity to residues 112 to 158, of the Arabidopsis LCR query sequence (Table I and Fig. 1A). These F-Box motif sequences showed a high level of conservation even though they come from evolutionary diverse plant species (Fig. 1A). Bioinformatic analysis of the secondary structure of the Arabidopsis LCR F-box C terminus revealed that this region contains a number of putative protein-protein interaction domains, termed Kelch repeats. Four Kelch repeats were detected, at LCR aa residues 163-211, 252-305, 307-354, and 406 -456 (Fig. 1A). In addition, a high level of sequence conservation was also detected in the C-terminal region of the 11 assessed LCR F-box-like proteins, which at nucleotide level contain the miR394 target site also conserved, to further indicate that each is a true representative of a LCR F-box-like protein encoded by each respective plant species (data not shown).
The secondary and tertiary structure of Arabidopsis LCR was predicted and the three-dimensional model shows two distinct structural domains: (1) an F-box motif in the N terminus forming a stem-like structure, and; (2) a C-terminal domain, consisting of four Kelch repeats that make a solenoidlike structure (Fig. 1B). Together, the four Kelch repeats form a four-stranded ␤-sheet, which, via hydrophobic interactions, generate a conserved ␤-propeller tertiary structure (Fig. 1C). In addition to this in silico evidence for Arabidopsis LCR to function as a canonical F-box protein, previous protein-protein interaction studies (40) have demonstrated that Arabidopsis LCR interacts with multiple members of the Arabidopsis SKP1-like (ASK) family (Fig. 1D). Generation of Arabidopsis Plants with Altered Micro-RNA394 and LCR Levels-Several approaches were used to generate transgenic plants with altered miR394-LCR regulatory pathway, in which either miR394 or LCR levels were directly targeted for deregulation. To specifically alter miR394 accumulation, miR394 overexpressing transgenic plant lines (miR394-OE) were generated by transforming plants with a construct containing 200bp of the precursor MIR394B, flanking the mature miR394 sequence. MIR394-OE plant lines highly accumulated miR394 as assessed by small RNA northern blotting and displayed upward leaf curvature when compared with WT plants ( Fig. 2A and 2B). We next attempted to generate miR394A/B knockdown Arabidopsis lines using two different "sponge" approaches (41). This technology uses complementary nucleotide sequences to the mature miRNA including mismatches between bases 10 and 11, preventing slicer activity of AGO proteins, and hence acting as a "sponge" of mature miRNAs sequences. In the first format, the miR394-SPO sponge construct was made from an artificially synthesized DNA sequence encoding 10 non-cleavable repeated regions complementary to miR394 sRNA but harboring mismatched bases across the cleavage site (supplemental Fig. S1A). However, using this approach, miR394 levels matched WT lines as assessed by small RNA northern blotting, and the miR394-SPO transformed lines closely resembled WT plants ( Fig. 2A and 2B). The second format followed the design of short tandem target mimic (STTM) constructs (42) (supplemental Fig. S1B). Successful knockdown was achieved below the limit of detection as assessed by small RNA northern blotting and the miR394-STTM plant lines displayed a range of altered morphology, including dramatically down-curled leaves ( Fig. 2A and 2B).
To alter LCR levels, a miR394-resistant LCR construct, with an altered miR394-binding site containing four silent point mutations was transformed into WT plants (supplemental Fig.  2A). Referred herein as LCR-OE plants (LCR overexpression), these transformants displayed ϳ2to 3-fold increase in LCR mRNA and a pronounced change in leaf morphology ( Fig. 2A  and 2C). Some of the lines showed severe developmental defects, characteristic of shoot apical meristem (SAM) termination, leading to premature death and failure to flower (supplemental Fig. S2B-S2E). Conversely, Arabidopsis plant lines with greatly reduced LCR expression were obtained from the SALK collection of T-DNA insertion mutant lines (18).
The T-DNA insertion is in the first exon of the LCR gene (SALK_136833). These plants, hereafter named LCR-KD (LCR knock down) had 4 -5-fold reduced expression of LCR mRNA and showed a subtle leaf polarity phenotype with the leaves curling in a slightly upwards direction, similar to that observed in the miR394-OE plants ( Fig. 2A and 2C).
The transgenic plants with altered miR394 expression were also assessed for LCR mRNA levels by RT-qPCR (Fig. 2C). This demonstrated an inverse relationship between the LCR mRNA levels and the observed phenotype. Specifically, overexpression of miR394 reduced LCR mRNA levels and resulted in leaves curling in an upwards direction. This phenotype was recapitulated by directly knocking down LCR. Alternatively, knocking down miR394 increased LCR mRNA levels resulting in leaves curling downwards combined with developmental defects. This phenotype was also recapitulated by directly overexpressing LCR.
Identification of Differential Proteins in Arabidopsis Plants with Altered LCR Levels-To identify LCR F-box regulated proteins, proteomic analysis was performed from the shoot apex of WT and the transformant lines, LCR-OE and LCR-KD. LCR-OE and LCR-KD lines were selected as they had been previously determined to have the most elevated and reduced LCR expression, respectively, among the transgenic lines (Fig. 2C). Proteins from the three groups were digested with trypsin and labeled with isobaric tags for relative and absolute quantification (iTRAQ) in biological duplicate prior to analysis by LC-MS/MS. An additional biological replicate was performed using 3-plex dimethyl labeling and LC-MS/MS which served as validation. A total of 4676 proteins were identified in the iTRAQ replicates with 3093 quantified with Ն2 peptides in both biological replicates ( Fig. 3A and supplemental Table  S2-S3). These data showed very low global variation and only two proteins were up-regulated with LCR-KD and one protein down-regulated with LCR-OE (Ͼ1.5-fold and adjusted p Ͻ 0.05). The single protein down-regulated with LCR-OE was the major latex protein 28 (MLP28; AT1G70830.1) whereas the two proteins up-regulated with LCR-KD were SOUL-like heme containing protein (AT1G17100.1) and the zinc-binding ribosomal-like protein (RPL37A; AT1G15250.1). Of the 3093 proteins quantified in both iTRAQ replicates, a total of 2922 proteins were quantified in the 3-plex dimethylation experiment. The three proteins regulated in both iTRAQ experiments were similarly regulated in the dimethylation experiment (Fig. 3B).
Transient co-expression of LCR and MLP28 results in reduced MLP28 accumulation-To experimentally test whether MLP28 and SOUL are targeted by LCR for post-translational regulation, plasmids (35Spro::MLP28::HA and 35Spro::SOUL:: HA) encoding hemaglutinin (HA)-tagged versions of both putative targets were generated. Young leaves of Nicotiana benthamiana plants were independently Agro-infiltrated with these constructs and the accumulation of MLP28 and SOUL assessed, by Western blotting. In all experiments, MLP28-HA accumulated to high levels but the levels of SOUL-HA were undetectable; further analysis of SOUL was therefore not pursued.
To verify that MLP28 is a specific target of LCR F-box, the 35Spro::MLP28::HA vector was co-infiltrated into young N. benthamiana leaves along with four F-box vectors: (1) 35Spro::LCR, encoding a full-length wild-type version of the Arabidopsis LCR gene; (2) 35Spro::mLCR, encoding a modified miR394-resistant full-length version of the Arabidopsis LCR gene (supplemental Fig. S2A); (3) LCRpro::mLCR, encoding the modified LCR transgene under the control of the endogenous LCR promoter, and; (4) 35Spro::P0, which produces the unrelated, but well-characterized, Polerovirus P0 F-box protein. Co-infiltration of the 35Spro::MLP28::HA with the P0 or with an "empty" 35Spro::HA construct were measures taken to monitor for nonspecific effects.
Western blotting showed that MLP28-HA expressed from the 35Spro::MLP28::HA construct accumulated to high levels when co-infiltrated with empty vector (Fig. 4A). Co-expression of MLP28-HA with LCR from either the 35Spro::LCR or 35Spro::mLCR constructs, resulted in a marked reduction in MLP28 levels as assessed by anti-HA Western blotting. A slight reduction in MLP28 levels was observed with co-expression of MLP28-HA with the LCRpro::mLCR construct. This is consistent with LCR targeting MLP28 for ubiquitination and hence degradation. The undiminished MLP28-HA levels from co-infiltration with 35Spro::P0 indicates that the reduction of MLP28-HA accumulation in these experiments was specifically caused by LCR.

Degradation of Additional MLP Gene Family Members by Arabidopsis LCR F-box-
The regulation of MLP28 in LCR transgenic Arabidopsis lines identified by quantitative proteomics and, the co-expression analysis in Agro-infiltrated N. benthamiana leaves suggested that MLP28 is targeted by the Arabidopsis LCR for post-translational regulation. We therefore performed a bioinformatics analysis of 12 additional members of the highly conserved Arabidopsis MLP gene family. Comparison of their amino acid (aa) sequences showed that MLP28 has its highest sequence identity with MLP31, MLP34, and MLP43 (supplemental Fig. S3A) and together they form a phylogenetic clade that is distinct from MLP165, MLP168, MLP328, MLP329, and MLP423 (supplemental Fig.  S3B). The tertiary structures of each of Arabidopsis MLP gene family members was predicted (supplemental Fig. S4) and revealed that they all adopt highly similar structures, despite only regions of similar sequences. This suggested that multiple MLP gene family members could be potential targets of LCR F-box-mediated post-translational gene expression regulation.
We next investigated the potential regulation of additional MLP family members by LCR. The full-length sequences of four MLP gene family members were cloned into a HA-tagged plant gene expression vector. Three of them, MLP31, MLP34, and MLP43, are highly similar to MLP28 whereas MLP423, has less sequence similarity. Each of the four MLP::HA expression vectors was co-infiltrated into young N. benthamiana leaves along with either; (1) an empty control vector; (2) the 35Spro::LCR vector, or; (3) the 35Spro::TIR1 vector, which is a well-characterized plant F-box protein and was included as a negative control. Western blotting revealed that co-expression of the 35Spro:LCR vector with each of the four MLP::HA vectors negatively affected the abundance of each HA-tagged MLP (Fig. 4B-4E). The MLP31 and MLP34 HA-tagged proteins showed dramatically decreased accumulation when coexpressed with the LCR transgene, but not when co-infiltrated with the TIR1 vector ( Fig. 4B and 4D). When co-expressed with LCR transgene, a modest reduction in MLP43 and MLP423 HA-tagged protein levels was detected by Western blotting (Fig. 4C and 4E). Together, these results show that MLP proteins are degraded by LCR and strongly suggested that other MLP gene family members in addition to MLP28 are targeted by LCR F-box for post-translational gene expression regulation.
In Vivo Assessment of LCR F-box Target Interaction-The Bimolecular Fluorescence Complementation (BiFC) assay was used to study protein-protein interactions in vivo (25). BiFC assay relies on the fluorescent signal that only forms when two fragments of a fluorescent protein is brought together by physical interaction of the proteins under investigation, which is fused to these fragments. In order to determine if the observed degradation of MLP proteins in the presence of LCR F-Box is directly driven by the physical interaction of the targeted MLP with LCR, two MLP gene family representatives with lower sequence similarity, namely MLP28 and MLP423, were selected for inclusion in this analysis.
BiFC analysis demonstrated in vivo interaction between LCR F-Box and MLP423 via the clear visualization of Yellow Fluorescent Protein (YFP) in N. benthamiana leaves co-expressing LCR:cYFP and MLP423:nYFP vectors (Fig. 4F). In addition, no fluorescence was observed when the TIR1::cYFP negative control vector was co-infiltrated with MLP423::nYFP vector (Fig. 4G). Furthermore, confocal microscopy revealed that LCR F-Box and MLP423 interaction occurred in the nuclear envelope and in the cytoplasm of YFP-expressing cells (Fig. 4H). No YFP fluorescence was observed in N. benthamiana leaves co-expressing LCR:cYFP and MLP28:nYFP vectors (data not shown). Together, these results indicate that the detection of physical interaction between LCR and MLP423, and not between LCR and MLP28, may be because of a more rapid degradation rate of MLP28, brought about by LCR F-box-directed ubiquitination, than MLP423.
MLP28 and MLP423 Gene Expression in Arabidopsis-To determine if the expression domains of MLP28 and MLP423 overlap with those previously determined for MIR394A, MIR394B, and LCR (supplemental Fig. S6), promoter-GUS expression vectors MLP28pro::GUS and MLP423pro::GUS were generated.
In Arabidopsis plants stably transformed with the MLP28pro::GUS, the GUS staining was clearly visible in the vasculature of rosette leaves, petioles, lateral root meristems, anther filaments and developing siliques, as well as in the embryonic tissues of developing seeds (Fig. 5A-5F). In MLP423pro::GUS lines, GUS activity was observed at low levels in all vegetative tissues of young plants, and more intensely in the petioles and at the base of newly emerged and emerging leaves (Fig. 5G). In more mature MLP423pro::GUS plants, the GUS staining was restricted to the base of trichomes, in the leaf (Fig. 5H), and to sepals, petals, anther filaments and the style, in floral tissues (Fig. 5I). MLP423 promoter activity was visible throughout mature silique tissues but had a more restricted pattern in developing siliques (Fig. 5J). It was also evident in the seeds of MLP423pro::GUS plants (Fig. 5K-5L).
These observed expression domains partially correlate with those of MIR394A, MIR394B and LCR (supplemental Fig. S5). This shows that MLP28 and MLP423, are transcribed in vegetative and/or reproductive tissues where LCR F-box can regulate the stability of their encoded proteins.
Characterization of Arabidopsis plants with repressed MLP expression-To assess whether disrupted MLP28 and MLP423 expression would have a negative effect on Arabidopsis development, putative T-DNA knockout insertion mutant lines were obtained from the publically available collection (18). Homozygous mutant plant lines mlp28 (CS366498), mlp423-1 (SALK_042869) and mlp423-2 (SALK_022306C) were characterized at both the phenotypic and molecular level.
Molecular analyses confirmed down-regulation of the MLP423 gene in the two lines, with both mlp423-1 and mlp423-2 displaying mild alterations in the leaf curvature (supplemental Fig. S6). At the phenotypic level, mlp28 plants were indistinguishable from WT (Fig. 6A). At the molecular level, RT-qPCR showed that MLP28 expression was highly elevated (Fig. 6E). The T-DNA insertion in mlp28 is immediately 5Ј of the MLP28 transcription start site which may explain its elevated, rather than repressed, transcription.
In addition to assessing MLP28 expression in mlp28 and the amiR-MLP28 lines, the transcript levels of LCR, CLV3 and WUS were also analyzed. As expected, LCR expression remained at approximately wild-type levels in mlp28 plants and in amiR-MLP28 1.1, 1.2 and 1.3 lines (Fig. 6F). The amiR-MLP28 1.2 and 1.3 lines had decreased CLV3 expression (Fig. 6G), but surprisingly the repression was greater in amiR-MLP28 1.2 plants than in the 1.3 line (Fig. 6G), and the WUS levels were elevated to the greatest degree in mlp28 and amiR-MLP28 1.1 plants, the two plant lines displaying wild-type like phenotypes and no change to CLV3 expression (Fig. 6A-6B and 6G-6H). Together, the phenotypic data suggest that reduced MLP28 expression has severe consequences on Arabidopsis development, and furthermore, the molecular analyses indicate that WUS and CLV3 are deregulated. DISCUSSION In plants, LCR F-box is highly conserved and is currently the only gene known to be targeted by miR394-directed posttranscriptional gene expression regulation (15,13). The role of most functionally characterized F-box proteins is to target a specific protein or multiple proteins for ubiquitination and subsequent degradation via proteolysis in the 26S proteasome (1,2). This suggests that the overall biological role of miR394 in Arabidopsis and other miR394 encoding plant species is to add an additional layer of regulation to the LCR F-box pathway.
Members of the F-box protein superfamily are characterized by a highly conserved 40 -60 aa F-box motif in their N terminus (5,45,46). Structure predictions revealed that LCR F-box contains a conserved 47 aa F-box motif, and that this motif is highly conserved across a diverse range of plant species (Table I and Fig. 1). Structural analysis also suggested that the F-box motif of LCR is likely to form a stem-like structure that protrudes from the body of LCR protein (Fig.  1B). Generally, the N-terminal F-box motif mediates F-box protein binding to SKP1 in the SCF complex (40,47,48,8), and accordingly, the crystal structure of the previously characterized F-box TIR1 in association with ASK1, revealed an overall mushroom-shaped structure (49).
Previous protein-protein interaction studies have demonstrated that LCR F-box interacts with several ASK-like proteins, including ASK1 (40), and the structural predictions performed in this study (Fig. 1) strongly suggest that these protein-protein interactions are most likely mediated by the highly conserved F-box motif in the LCR N terminus. The LCR C terminus is predicted to form a ␤-propeller structure (Fig.  1C) and most likely confers the specificity of LCR F-box for its targeted protein(s), as previous research has demonstrated that the C-terminal region of individual F-box proteins encodes a variety of protein-binding domains that are usually responsible for substrate recognition by the SCF complex (5,50,51). However, the target proteins of most of the many hundreds of currently identified plant-specific F-box proteins remain to be experimentally determined.
Here, molecular approaches were taken to generate and study Arabidopsis lines with altered LCR expression. In LCR-OE line, a LCR-targeted protein would be expected to be rapidly degraded and therefore only accumulate to very low or even undetectable levels. Conversely, LCR-targeted proteins would be expected to have enhanced accumulation in Arabidopsis lines with reduced LCR levels, namely in LCR-KD plants. Comparative MS-based analysis identified members of two protein families, the MLP and SOUL protein families as putative candidates for LCR-mediated post-translational gene expression regulation, because of their differential accumulation in LCR-OE and LCR-KD plant lines (Fig. 3). SOUL proteins are ubiquitous in nature, functioning as tetrapyrrole carrier proteins in animals (52,53). In plants, tetrapyrroles include compounds such as chlorophyll and heme; however, the functional role that SOUL proteins mediate in plant cells remains to be determined (54 -56). Although SOUL was identified as a differentially expressed candidate protein by MS, subsequent molecular-based approaches to assess whether LCR directly post-translationally regulates SOUL protein stability were unsuccessful and therefore require further experimental investigation.
Members of the second group of MS-identified putative LCR target proteins, the MLP protein family, were first identified in opium poppy (Papaver somniferum) as latex-specific polypeptides (57), and have because been found to be highly conserved in plants (58,59). Although the Arabidopsis MLP protein family consists of twenty-four members (58,59), only peptides derived from family member MLP28 showed differential accumulation in LCR-OE and LCR-KD plants by LC-MS/MS. Additional analyses revealed that MLP family members are closely structurally related to one another (supplemental Fig. S3 and S4), and furthermore, that in addition to MLP28, Arabidopsis MLP family members MLP31, MLP34, MLP43, and MLP423 were also regulated but to differing degrees, by LCR F-box (Fig. 4). Together, the structural and molecular-based results reported here strongly indicated that MLP protein family members might be targeted by LCR F-box for post-translational gene expression regulation.
The structure of Arabidopsis MLP28 has previously been experimentally determined and demonstrated to be structurally related to members of the Bet v1 (from the Betula verrucosa) protein superfamily (58). Bet v1 proteins are characterized by the presence of hydrophobic pockets in their tertiary structure and compounds that are structurally similar to brassinosteroids and cytokinins have been shown to 'occupy' these hydrophobic pockets (60 -63). Indeed, the predicted structure of all analyzed members of the Arabidopsis MLP protein family showed high modeling confidence and aa identity with the Bet v1 superfamily proteins, Cytokinin-Specific Binding Protein (CSBP) and Pyrabactin Resistance [PYR]/ PYR-Llike 3 (PYL3). CSBP and PYL3 encode a cytokinin binding protein and an abscisic acid (ABA) receptor respectively (63,64), and together these analyses indicate that Arabidopsis MLP proteins could also potentially bind compounds with structural similarity to plant hormones in their hydrophobic pockets.
The post-transcriptional regulation of the TIR1 F-box by miR393 is well established (12,65). TIR1 post-translationally targets several Aux/IAA family members for ubiquitin-mediated degradation (12,65). Aux/IAAs are transcription regulators that repress the expression of several auxin-responsive genes and all 29 Aux/IAAs encoded by Arabidopsis are regulated to different degrees by TIR1 (66 -68). TIR1 is a hormone receptor and a hydrophobic pocket on the upper surface of the C terminus of TIR1 presents a binding site for auxin to allow TIR1 to act as the 'molecular glue' for Aux/IAA target protein interaction (49). If LCR F-box targets MLP gene family members for ubiquitin-mediated degradation, a similar F-box/ hormone-pocket/F-box-target mechanism can be envisaged for LCR F-box interaction with its MLP targets.
The detection of physical interaction between LCR and MLP423, and not between LCR and MLP28 was perplexing. The LC-MS/MS data clearly showed that of all the Arabidopsis MLP gene family members, only the accumulation of MLP28 was affected in the shoot apex tissue of LCR-OE and LCR-KD lines. In addition, Western blot analysis clearly demonstrated that LCR co-infiltration had a greater impact on MLP28 levels than on MLP423 accumulation. Together, these results indicate that LCR preferentially interacts with MLP28 over MLP423, but this preferential interaction could not be confirmed by the transient BiFC approach. One possible scenario is that LCR can interact with either MLP, but triggers ubiquitination of MLP28 at a much greater rate than its me-diated ubiquitination of MLP423. This would lead to the rapid degradation of the N-terminal fused YFP version of MLP28, thus preventing the generation of visible fluorescent signals in LCR:cYFP/MLP28:nYFP co-expressing cells. Conversely, a slower rate of LCR-mediated ubiquitination of the less preferred target, MLP423, would allow in planta detection of physical LCR/MLP423 interaction and YFP visualization.
Although a physical interaction between MLP28 and LCR F-box was not detected, our promoter::GUS transgene results showed that the expression patterns of MLP28, MLP423, LCR and LCR-targeting miRNA, MIR394, overlap in specific Arabidopsis tissues and stages of development ( Fig. 5 and supplemental Fig. S5). The vascular-restricted expression of MLP28 is consistent with the expression profile of latex-specific MLP genes in opium poppy (57). Additionally, MLP328 and MLP329 have been detected in Arabidopsis phloem sap, leading to the suggestion that they are involved in longdistance signaling and lipid transport (69). Interestingly, our analyses revealed that MIR394A is also expressed in the vascular tissue of young rosette leaves (supplemental Fig.  S5), and aberrant vascular pattern formation has been described in an Arabidopsis plant transformed with a miR394resistant LCR transgene (13). The promoter::GUS transgene approach also clearly demonstrated that all assessed loci, including MLP28, MLP423, LCR, MIR394A, and MIR394B, has their promoter active in the shoot apices, and the tissues surrounding this region, including the petioles and bases of emerging, or newly emerged leaves ( Fig. 5 and supplemental  Fig. S5). These expression data not only support previous microarray-based expression data for each of the analyzed loci (70), but further suggest that the miR394/LCR/MLP relationship forms a crucial gene expression regulation module in tissues that are central to Arabidopsis development.
The major developmental defects observed in Arabidopsis plants, when miR394-directed post-transcriptional repression of LCR is lost, are concentrated in the shoot apex region during the early stages of vegetative development. The leaves of LCR-OE plants are strongly downwardly curled and, in severe phenotypes, meristem aberration impairs further plant development ( Fig. 2 and supplemental Fig. S2). If the developmental defects displayed by LCR-OE plants were the result of enhanced LCR F-box-mediated post-translational gene expression regulation of MLP28, then plants with reduced MLP28 levels would be expected to express similar developmental phenotypes. Plants of two independent transgenic lines, expressing amiRNAs that reduce the expression of MLP28, displayed a range of phenotype abnormalities, from mild changes in leaf curvature through to the development of tiny plants with dramatic changes in leaf patterning and morphology and shoot apex defects ( Fig. 6C and 6D). These phenotypes that are highly suggestive of deficiencies in SAM development and resemble those displayed by Arabidopsis plants in which LCR expression is no longer correctly regulated by miR394 ( Fig. 2 and supplemental Fig. S2). The target protein(s) of LCR F-box has previously been suggested to enable the correct coordination of stem cell fate in the SAM, and that this is orchestrated by LCR F-box target protein(s) mediating the expression of SAM regulators, WUS and CLV3 (16). Indeed, slight perturbations in CLV3/WUS gene expression is presented herein, where CLV3 levels were altered in both LCR-OE and amiR-MLP28 lines that displayed developmental abnormalities. Curiously, a corresponding increase in WUS expression in these lines was not detectable although this may be because of difficulties in measuring genes that are expressed in only a few cells within the SAM.
Here we show that the miR394-regulated F-Box LCR degrades MLP proteins, and provides extensive evidences to suggest that MLP gene family, specifically MLP28 and MLP423, are target of miR394-LCR regulatory node, representing potential targets for post-translational gene expression regulation by LCR. In addition, and that is critical for normal Arabidopsis development. We propose that MLP28 and MLP423 family members are associated with the LCR regulation and a miR394/LCR/MLP regulatory module might exist in the shoot apical meristem, being critical for normal plant development.