Arabidopsis MLK3, a Plant-specific Casein Kinase 1, Negatively Regulated Flowering and Phosphorylated Histone H3 in vivo

Background: Casein kinase 1 (CK1) family members are highly conserved serine/threonine kinase present in most eukaryotes with multiple biological functions. Arabidopsis MUT9-like kinases ( MLKs ) belong to a clade CK1 specific to the plant kingdom and have been implicated collectively in modulating flowering related processes. Three of the four MLKs ( MLK1/2/4 ) have been characterized, however, little is known about MLK3 , the most divergent MLKs. Results: We demonstrated that compared with wild type, mlk3 , a truncated MLK3 , flowered slightly early under long day conditions and ectopic expression of MLK3 rescued the morphological defects of mlk3 , indicating that MLK3 negatively regulates flowering. GA 3 application accelerated flowering of both wild type and mlk3 , suggesting that mlk3 had normal GA response. The recombinant MLK3-GFP was localized in the nucleus exclusively. In vitro kinase assay revealed that the nuclear protein MLK3 phosphorylated histone 3 at threonine 3 (H3T3ph). Mutation of a conserved catalytic residue (Lysine 175) abolished the kinase activity and resulted in failure to complement the early flowering phenotype of mlk3 . Interestingly, the global level of H3T3 phosphorylation in mlk3 did not differ significantly from wild type, suggesting the redundant roles of MLKs in flowering regulation. The transcriptomic analysis demonstrated that 425 genes significantly altered expression level in mlk3 relative to wild type. The mlk3 mlk4 double mutant generated by crossing mlk3 with mlk4 , a loss-of-function mutant of MLK4 showing late flowering, flowered between the two parental lines, suggesting that MLK3 played an antagonistic role to MLK4 in plant transition to flowering. Conclusions: A serine/threonine kinase encoding gene MLK3 is a casein kinase 1 specific to the plant species and represses flowering slightly. MLK3 located in nucleus catalyzes the phosphorylation of histone H3 at threonine 3 in vitro and an intact lysine residue (K175) is CK1s in regulating flowering time. This study investigated the

3 indispensible for the kinase activity. This study sheds new light on the delicate control of flowering by the plant-specific CK1 in Arabidopsis.

Background
Casein kinase 1 (CK1) family members were first characterized in the 1970s for their catalytic activity to phosphorylate casein in vitro [ 1]. These highly conserved serine/threonine-selective enzymes are ubiquitously expressed and play diverse cellular roles in eukaryotic organisms ranging from yeast to humans [2,3]. During evolution, the CK1 genes evolved an ancient duplication event, resulting in the split of the family into two subgroups, of which one contains members from plant species alone [2,4]. Mut9 from green algae Chlamydomonas was the founding member of the plant-specific CK1 and the kinase predominantly phosphorylated histone H3 at threonine 3 [4].
In the last decade, lots of progress has been made on researches of CK1s from plants, especially Arabidopsis. The four MUT9-LIKE KINASEs (MLK1-4) have revealed to collectively impact Arabidopsis growth and development by involving in multiple physiological processes [5][6][7][8][9]. Previously, we found that the mlk1 mlk2 double mutant with pleiotropic defects including dwarfism, altered leaf shape and low fertility, was hypersensitive to osmotic stresses [10], suggesting the essentiality for plant growth and stress response.
MLKs, also designated Photoregulatory Protein Kinases (PPKs), have been reported to implicate in the fine-tuning of plant response to solar radiation by coordinating the coaction of photoreceptors: phytochrome and cryptochrome [11]. For MLKs' involvement in phytochrome, the hint was from the observation that individual MLKs was co-purified with the circadian clock component ELF3 by affinity purification at the presence of phytochrome B (phyB) [12]. Ni and coauthors' finding that MLKs interacted and phosphorylated both phyB and phytochrom-Interacting Factor3 (PIF3) provided direct molecular evidence [13]. For the blue-light signaling, it was proved that phosphorylation of the blue-light receptor cryptochrome 2 (CRY2) was catalyzed by MLKs and the kinases functioned as interacting partners of CRY2 [11]. The physical structure analysis of MLKs revealed that both the conserved N-terminal domain and the non-kinase domain at Cterminal facilitated the interaction with PIF3 and CRY2 in response to red and blue light, respectively [11,13].
Recently, MLKs, also described as Arabidopsis EL1-like (AEL) proteins, were documented to phosphorylate ABA signal receptors PYRABACTIN RESISTANCE/PYR-LIKE (PYR/PYL) proteins to promote their degradation by ubiquitination, and triple mutants of mlks were hypersensitive to ABA treatment due to eliminated ABA responses [8]. Interestingly, one of the triple mutants, mlk1,3,4, showed elevated level of salicylic acid (SA)-induced defense marker genes, and MLKs were identified as interacting proteins with RADICAL-INDUCED CELL DEATH1 (RCD1), a positive regulator of SA signaling [7]. Taken together, Arabidopsis MLKs collectively play critical roles in diverse biological processes including stress response, light signaling and immunity.
Besides the common roles shared by CK1s, several CK1s have been shown to involve in flowering regulation in both monocot and dicot. In rice, for example, mutation of Early Flowering 1 (EL1)/Heading Date 16 ( Hd16), a plant-specific CK1, caused plants flowered 5-6 days earlier than wild type (Dai and Xue 2010). EL1/Hd16 has been proven to act as a multifunctional kinase to phosphorylate hierarchical flowering-regulating proteins, such as DELLA protein SLENDER RICE1 (SLR1), circadian clock component Pseudo-Response Regulator 37 (OsPRR37) and floral repressor GRAIN NUMBER, PLANT HEIGHT, AND HEADING DATE7 (Ghd7) [14][15][16]. On the other hand, Arabidopsis mlk4 flowered late relative to wild type [12]. Recent study showed that MLK4 interplaying with circadian clock component CCA1 to accelerate flowering [17]. The two cases suggested the opposite roles of the plant-specific CK1s in regulating flowering time. This study investigated the 5 function of MLK3, the most divergent MLKs , in flowering by focusing on a truncated MLK3 mutant allele with slight early flowering phenotype. Our molecular, histochemical and genetic findings revealed MLK3 function in flowering repression and the in vitro activity of the nuclear protein in catalyzing phosphorylation of H3 at threonine 3 Results MLK3 with conserved kinase domains belongs to a separate subgroup of CK1s specific to plant kingdom Previously, we have identified four plant-specific casein kinase 1 encoding genes, namely MLK1-4 from Arabidopsis [10]. The four MLKs shared sequence identity of 67.9% -91.1%.
Homology analysis of MLKs and the homologs from several crops demonstrated that the plant-specific CK1s were classified into two main branches (I and II). MLK3, distinct from its paralogs, was grouped into a separate branch (Additional file 1, Figure S1).
Consistently, the phylogenetic analysis showed that MLK3 was relatively distant from the other three MLKs (Fig. 1a). Sequence alignment showed that MLK3 shared the conserved CK1 functional domains including substrate recognition region, kinase catalytic loop, ATP binding site and a predicted nuclear localization signal [4] (Additional file 2: Figure S2).
The predicted isoelectric point of MLKs ranges from 9.09 to 9.66 (Additional file 3: Table   S1), suggesting the preference to acidic substrates, such as serine and threonine residues. Thus, albeit divergent from other MLKs in Arabidopsis, MLK3 possessed the common features of CK1, implying its enzymatic activity as a kinase in phosphorylating target protein(s).

MLK3 was ubiquitously expressed and the MLK3-GFP recombinant protein was localized in the nucleus
Three MLKs (MLK1, 2 and 4) have been functionally identified [17][18][19]. To investigate the biological role of MLK3, the most divergent MLK, we first examined its spatial and temporal expression patterns using semi-quantitative RT-PCR. As shown in Figure 1b, MLK3 was expressed in roots, stems, leaves and flowers, which is in agreement with the results from eFP Browser (http://bbc.botany.utoronto.ca/efp) [20]. Comparison of the absolute transcription level demonstrated that MLK3 transcript is the lowest, while MLK4 is the highest, which is about 2.8-fold of MLK3 (Additional file 4 Figure S3).
Considering that MLK3 has a predicted nuclear localization signal (Additional file 2: Figure S2) [4,10], the subcellular localization of MLK3-GFP recombinant protein was examined. As expected, when transiently expressed in tobacco leaves by infiltration, the green signal of the MLK3-GFP fusion protein was observed exclusively in the nucleus of the leaf epidermal cells as indicated by DAPI staining, while the signal of 35S::GFP control displayed a universal distribution in the epidermal cells of the tobacco leaves (Fig. 1c).
Hence, consistent with the previous findings in tobacco and Arabidopsis protoplasts [8,11,12], MLK3, like its paralogs, is a nuclear protein, implying its potential role in histone modification.

MLK3 phosphorylated histone H3 at threonine 3 in vitro
Given that nuclear protein MLK3 shares the canonical features of CK1, we tested whether MLK3 functions as protein kinase. First, MLK3 was fused with maltose binding protein (MBP) and expressed in E. coli BL21 (DE3) strain. The purified MLK3-MBP recombinant protein was then incubated with the phosphoryl donor ATP and substrate before being dotted on membrane. Finally, immuno-blotting was performed using an antibody specifically against phosphorylated H3T3 [4]. Our results showed that strong immunesignal (bottom row) was detected with substrate H3T3ph peptide (a phosphor-histone H3 (Thr3) peptide) (Fig. 2a), which was used as positive control, confirming the specificity of the antibody. For substrate of unmodified histone H3 peptide (H3), anti-H3T3ph signal was detected when both ATP and the recombinant protein MLK3-MBP were present (middle row), while anti-H3T3ph signal was undetectable in the absence of MLK3-MBP fusion protein (upper row) (Fig. 2a). The detection of the phosphorylated histone H3 at threonine 3 indicated that with ATP as phosphoryl donor, the recombinant protein MLK3-MBP catalyzed in vitro phosphorylation of unmodified histone H3 peptide substrate. Hence, MLK3 phosphorylated histone H3 at threonine 3 (H3T3ph), a predominant target of the plant-specific kinase Mut9 and MLK1 in Chlamydomonas and Arabidopsis, respectively [4,10].

Conserved lysine (K) 146 of MLK3 is essential for MLK3-modulated in vitro phosphorylation of histone H3T3
It has been reported that the conserved lysine residue (K174 for Mut9p, K175 for MLK4/PPK1) was essential for the catalytic activity of the kinases [4,11,21]. To test whether the counterpart lysine (K146) of MLK3 is critical for phosphorylation of histone H3T3, the conserved K146 was point mutated to arginine (R). The MBP-fused MLK3 (K146R) was purified from E. coli., then the kinase activity was examined by dot blotting as mentioned above and the reaction without enzyme was served as negative control.
Expectedly, H3T3ph peptide, the positive control, showed anti-H3T3ph signal in the immuno-analysis (bottom row) (Fig. 2b). No signal was detected for unmodified histone H3 peptide substrate with ATP served as phosphoryl donor, no matter the mutated recombinant protein MLK3 (K146R)-MBP was supplied or not (Fig. 2b). These results indicated that unlike MLK3, the point-mutated MLK3 (K146R) was catalytically inactive.
Therefore, an intact lysine at the conserved position is crucial for the substrate phosphorylation mediated by the plant-specific CK1.

MLK3 affected leaf growth and flowering time
To address the biological function of MLK3, a T-DNA insertion line (SALK_017102) was obtained [10]. PCR analysis revealed that the T-DNA was integrated into the 12 th exon of 8 MLK3, resulting in a truncated peptide of 498 amino acid residues (Fig. 3a, b). In homozygous mlk3 mutant, MLK3 transcript flanking the insertion site was undetectable by RT-PCR, while a transcript upstream of the insertion site was detected (Fig. 3c), suggesting the partial expression of MLK3. Hereafter, the primers flanking the T-DNA insertion site were used to analyze the expression of MLK3. Morphologically, during vegetative stage mlk3 was slightly smaller than wild type under LD (Fig. 3d). Measurement of the rosette leaf numbers showed that mlk3 had 1.8, 2.4 and 3.2 fewer leaves on average than wild type in Week-2, 3 and 4, respectively (P<0.05) (Fig. 3e), suggesting the progressive retardance of leaf growth in mlk3. Our calculation of leaf area (the 5 th leaf) demonstrated that the fifth leaf of mlk3 was about 2.4-5.0 mm 2 smaller than that of wild type during the three weeks (P<0.05) (Fig. 3f).
For flowering time, Huang et al., [ 12] revealed that in terms of the days to inflorescence at one centimeter, statistically mlk3 had a minor fewer number of days than wild type, demonstrated that similar to mlk3, DAG of the both lines was significantly fewer than wild type (P<0.05) (Fig. 5b), suggesting that the transgenic lines flowered earlier. Consistently, the leaf number of the transgenic mlk3 ectopically expressing MLK3 (K146R) did not significantly differ from that of mlk3 (Additional file 6: Figure S4). These results indicated that unlike MLK3, which successfully restored the early-flowering phenotype of mlk3, the catalytically inactive MLK3 (K146R) expressed constitutively did not alter either flowering time or leaf growth. Therefore, the conserved lysine K146, which is essential for phosphorylation of H3T3, is indispensable for MLK3-mediated flowering repression.

Truncation of MLK3 did not significantly affect the transcriptional level of the major flowering regulators nor the global intensity of phosphorylated H3T3
To profile the transcriptome of mlk3, RNA-sequencing was carried out. Consistent with the truncation of MLK3, the unique reads of mlk3 matching the exons downstream of the T-DNA insertion site was clearly depleted compared with wild type, while no significant difference was monitored upstream of the T-DNA interruption (Additional file 7: Figure S5).
By the criteria of │Log 2 FC│≥1 and P<0.01, a total of 425 genes were differentially expressed with 133 up-regulated and 292 down-regulated in mlk3 relative to wild type (Additional file 8: Figure S6a). None of the transcript of MLK3 paralogs was significantly changed in mlk3, implying no clear compensation of other MLKs. Based on gene annotation, no flowering regulator was differentially expressed in mlk3 (Additional file 9: Table S3), suggesting that truncation of MLK3 may not significantly affect the known flowering signaling components at the transcriptional level. Functional categorization of the DEGs based on gene ontology (GO) annotations revealed that the genes categorized to "negative regulation process", such as "developmental process" and "multicellular organismal process", were up-regulated, while the genes categorized to "positive regulation process" of the above two biological processes were down-regulated (Additional file 8: Figure S6b).
To compare the global level of H3T3ph between mlk3 and wild type, western blot was performed with anti-H3T3ph antibody. The intensity of H3T3ph in mlk3 was not notably different from that of wild type (Additional file 10: Figure S7), suggesting the functional redundancy of other MLKs, especially MLK1 and MLK2 [ 10]. Therefore, truncation of MLK3 caused no significant alteration on either the transcriptional level of the main components of flowering pathway or the intensity of H3T3ph globally.

MLK3 acts antagonistically to MLK4 in regulating flowering time
It has been documented that the loss-of-function mutant of MLK4, the closest paralog of MLK3, flowered late [12,17]. To investigate the genetic relationship between MLK3 and MLK4 in flowering regulation, we generated the mlk3 mlk4 double mutant by crossing the two single mutants (Fig. 6a). In homozygous mlk3 mlk4, the expression level of both MLK3 and MLK4 was eliminated to that of in the corresponding single mutants (Fig. 6b). Given that FT transcript was reduced significantly in mlk4 mutant [17], the expression level of FT in mlk3 mlk4 was tested by RT-qPCR. As shown in Fig. 6b, FT in mlk3 mlk4 was about 60% of wild type, while in mlk3 and mlk4, it was about 135% and 52% of wild type, respectively ( Fig. 6b), indicating a compromised level of FT in mlk3 mlk4 relative to its parental lines.
Statistical analysis of the average DAG showed that mlk3 mlk4 flowered at 30.2 DAG, while the two parental lines flowered at 19.7 DAG for mlk3 and 32.8 DAG for mlk4, respectively (Fig. 6c), indicating that mlk3 mlk4 flowered later than mlk3 but earlier than mlk4. These results suggested that MLK3 acted antagonistically to MLK4 in regulating flowering time.

Discussion
Casein kinase 1 (CK1) is a universal and conserved ser/thr protein kinase family in eukaryotes [3]. In mammals, six CK1 isoforms (alpha, beta, gamma, delta and epsilon) have been reported to involve in a variety of cellular processes such as circadian rhythm, chromosome segregation and cellular differentiation by phosphorylating a wide range of key regulatory proteins [3]. Plant kingdom possesses an independent clade kinase evolutionarily related but phylogenetically distinct from mammalian CK1. The plantspecific CK1 family in higher plants expanded due to the lineage duplication event [4,22].
Based on sequence analysis, green algae Chlamydomonas reinhardtii has one, namely Mut9, and Arabidopsis genome encodes four, rice six and soybean nine of the plantspecific CK1 members, respectively [4,22].
Several plant-specific CK1s have been documented to implicate in the transition from the vegetative to the reproductive stage through flowering pathways, such as phytohormone GA signaling, circadian clock and light signaling [5,16,23]. Consequently, flowering time was affected by mutation of the CK1 encoding genes in model plant rice and Arabidopsis.
On one hand, the rice mutant early flowering1 ( el1), which is deficient in a plant-specific casein kinase I, exhibited early flowering with slow leaf emergence rate [15,16]. On the other hand, mutation of Arabidopsis MUT9-LIKE KINASE4 ( MLK4) flowered late [12,17].
These findings indicate the two plant-specific CK1s from rice and Arabidopsis oppositely regulate flowering time. Consistent with the former case, we found that truncation of  [11]. Unsurprisingly, the latter one was co-purified with two of the evening complex components, ELF3 and ELF4, while the former one interplayed with ELF3 alone [11,12]. Hence, the plant-specific CK1s act non-redundantly in regulating flowering time.
It appears that the intact kinase activity of MLK3 is indispensible for Arabidopsis flowering regulation. Albeit relatively divergent from the other three MLKs, MLK3 shared the common functional domains with a sequence identity ranging from 67.9% to 72.7% [4,10]. As expected, the nuclear protein MLK3, like its homolog MUT9 and MLK1, in vitro phosphorylated histone H3 at threonine 3 (H3T3), a predominant phosphosite of Mut9 in Chlamydomonas [ 4,10]. In contrast, the enzymatic activity of MLK3(K146R) was abolished by mutating a conserved lysine residue (K146). This is consistent with the elimination of the catalytic activity of Mut9 or MLK4/PPK1 caused by mutating the counterpart lysine, i.e. K174 for Mut9 in Chlamydomonas and K175 for Arabidopsis MLK4/PPK1, respectively [4,11]. These findings suggested that the plant-specific CK1s shared the similar catalytic structure and mechanism in phosphorylating target proteins. In agreement with the in vitro result, constitutive expression of MLK3 but not the point-mutated MLK3 (K146R) complemented the defects of mlk3 in flowering. Thus, the intact enzymatic activity of MLK3 was required for proper flowering repression.
Diverse proteins including histones and flowering signaling components have been documented to be phosphorylated by the plant-specific CK1s. For example, MLK4 in vitro phosphorylated histone H2A on serine 95 although no notable difference of phosphorylated H2A S95 was detected between the late flowering mlk4 mutant and wild type [17]. Our results that MLK3 phosphorylated histone H3 at T3 in vivo and mutation of the kinase activity on H3T3 could not rescue the early flowering phenotype of mlk3 suggest that MLK3-mediated phosphorylation of H3T3ph plays a critical role in flowering.
In addition to histone proteins, recent studies on mlk3/ppk4-combined MLKs/PPKs triple mutants proved that MLK3/PPK4, together with its paralogs, targeted light signaling receptors or coordinators. For instance, the level of phosphorylated PIF3, a negatively regulator of flowering, was reduced in red-light hypersensitive triple mutant ppk124 and ppk134 [ 13,24], suggesting that normal light-induced PIF3 phosphorylation required the MLKs/PPKs collectively. For rice CK1 protein EL1, DELLA protein SLR1 was found to be phosphorylated specifically and EL1 was required for the negative effect of SLR1 on GA signaling including flowering regulation [16]. These findings support the notion that MLK3,

Plant materials and growth conditions
Arabidopsis seeds of ecotype Col-0, the T-DNA insertion lines of GK-756G08 and SALK_017102 were obtained from Dr. Cerutti's lab at University of Nebraska-Lincoln [10].
Imbibed seeds were treated at 4 °C for three days before germination in soil. Plants were grown under normal conditions (21°C) with long-day (LD) (16 h light/8 h dark) with light supplied at 100 µmol/m 2 /s.

Plasmid constructs and plant transformation
For pMLK3::GUS, a fragment of 1.5 kb upstream of the transcription start site of MLK3 was amplified from genomic DNA, and was cloned into pCAMBIA1301 vector using Pst I and Bgl  Table S4.

In vitro protein kinase activity assay
The open reading frame (ORF) of MLK3 without the stop code (TGA) was re-amplified. The 15 sequence confirmed amplicon was cloned into Not I and BamH I sites of vector pMAL-c5X (NEB). For generation of MLK3 (K146R)-MBP, the point-mutated MLK3 plasmid was used as template with the same primers for MLK3-MBP. Sequence verified PCR product was digested by Not I and BamH I, then ligated into pMAL-c5X. The MBP fused protein was expressed in E. coli BL21 (DE3) strain. The recombinant protein was purified with Amylose Resin (NEB) after induction for 16 hours by IPTG (0.1 mM). Protein kinase activity was assayed by dot blotting with antibody against H3T3 phosphorylation (Upstate, 07-424) as described [10]. An unmodified histone H3 peptide (residues 1-21) biotin conjugate (Upstate 12-403) was used as substrate.

mRNA sequencing and data analysis
For mRNA-sequencing, rosette leaves from 11-day-old plants grown under LD were used to isolate total RNA with TRIzol reagent (Invitrogen). Three independent replicates were used for RNA extraction and library preparation according to the Illumina's protocol. The libraries were sequenced using a Genome Analyzer (Illumina) and the transcription analysis was processed with a regular RNA-sequencing workflow on bioconductor. Briefly, the reference genome of Arabidopsis (ftp://ftp.ensemblgenomes.org/pub/plants/release-42/fasta/arabidopsis_thaliana/dna/) was used to map the clean reads after trimming the raw read data by TopHat 2.1.1 [25]. Gene expression values were calculated and differentially expressed genes were determined by DESeq2 package [26]. Gene Ontology plotting was performed using Bioconductor packages ggplot2, clusterProfiler [27] and org.At.tair.db.

Immunoblot analysis
For dot blotting, reactions containing substrate, ATP and phosphatase inhibitor (Roche Applied Science) were carried out in liquid medium in the presence or absence of MLK3-MBP/MLK3 (K146R)-MBP at 30 °C for 60 min. One ul of the reactions was spotted on a 16 nitrocellulose filter and air dried. Ponceau S stained membrane was pictured to indicate protein loading and was applied to regular immune blotting. For western blot, total protein isolated from two-week-old plants was separated on 15% SDS-PAGE. Antibody against H3T3 phosphorylation (Upstate, 07-424) was used with a modification-insensitive anti-H3 antibody (Abcam, ab1791) as an internal loading control.

Subcellular localization and nuclear staining
To generate MLK3-GFP fusion protein, the CDS of MLK3 was cloned into pENTR/D-TOPO

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.