StMYB44 negatively regulates phosphate transport by suppressing expression of PHOSPHATE1 in potato

Highlight Pi deficiency suppressed the transcription factor gene StMYB44, encoding the protein that interacts with Arabidopsis WRKY6 and StWRKY6 and negatively affects Pi distribution by suppressing the expression of PHOSPHATE1 in potato.


Introduction
Inorganic phosphates (Pi) are taken up by plants to meet the phosphorus (P) requirements for a variety of structural and physiological functions. Inadequate supply of P in soil negatively affects plant development and growth (Raghothama, 1999). Reshaping root architecture and development is one of the mechanisms to increase P i uptake, mobilization, and utilization upon P i deficiency (Devaiah et al., 2007b). Changing expression of Pi-responsive genes, and altering metabolic and developmental processes are molecular adaptations in this regard (Wu et al., 2003;Thibaud et al., 2010;Secco et al., 2013;Puga et al., 2014). Systematic transcriptional regulation of Pi-responsive genes is believed to be the major regulatory step in maintaining Pi homeostasis (Hammond et al., 2004). A number of transcription factors mediating plant responses to Pi starvation have been identified in Arabidopsis and rice, including MYB transcription factors, PHR1/OsPHR2, PSR1, AtMYB2, MYB62, and OsMYB2P-1 (Wykoff et al., 1999;Rubio et al., 2001;Zhou et al., 2008;Devaiah et al., 2009;Dai et al., 2012;Baek et al., 2013), WRKY transcription factors AtWRKY6, AtWRKY42, AtWRKY45, and AtWRKY75 (Devaiah et al., 2007a;Chen et al., 2009;H. Wang et al., 2014;Su et al., 2015), basic helix-loop-helix transcription factors OsPTF1 and bHLH32 (Yi et al., 2005;Chen et al., 2007, zinc-finger transcription factor ZAT6 (Devaiah et al., 2007b), and APETALA2/ETHYLENE RESPONSE FACTOR, AtREF070 (Ramaiah et al., 2014). Each transcription factor specifically activates or suppresses a single or multiple Pi-related genes in response to Pi starvation (Chen et al., 2009;H. Wang et al., 2014;Su et al., 2015). Nuclear proteins SPX1 and SPX2 carry an SPX domain, which exists in Pi sensors and other Pi starvation signaling proteins in yeast and plants. These proteins are found to inhibit the activity of PHR1 and OsPHR2 transcription factors by protein-protein interactions in response to Pi availability in Arabidopsis and rice (Puga et al., 2014;Z. Wang et al., 2014). It demonstrates the vital role of transcription factors in Pi signaling pathways by linking Pi perception and gene expression. Hence, identification of additional transcription factors will further broaden our understanding about the signaling process in plant responses to Pi deficiency.
Among various MYB families, R2R3-type transcription factors are the largest MYB family in plants (Stracke et al., 2001). Based on amino acid sequence similarities, 126 Arabidopsis R2R3-type MYB transcription factors are categorized into 22 subgroups, and the last subgroup of MYB transcription factors mainly mediates hormone signaling and abiotic stress responses (Jung et al., 2008). One of its members, AtMYB77, mediates auxin signaling by interacting with auxin response factors and regulating expression of auxin-inducible genes to control lateral root growth and development (Shin et al., 2007). Another member of this subgroup, AtMYB44, positively regulates drought tolerance by enhancing stomatal closure (Jung et al., 2008). In addition, AtMYB44 has also been shown to induce expression of ETHYLENE INSENSITIVE2 (EIN2), a central component in the ethylene signaling pathway . Interaction of MYBR1/AtMYB44 with ABA receptor PYR1-LIKE8 (PYL8) mediates leaf senescence and responds to stress and wounding (Jaradat et al., 2013), implying that members of this subgroup are involved in diverse physiological processes in plants.
Potato (Solanum tuberosum L.), the fourth largest food crop in the world, faces an array of abiotic stresses including drought, cold, and mineral deficiency (Leone et al., 1999). Unlike Arabidopsis and rice, little is known about the mechanisms to maintain mineral homeostasis in potato since relatively few genes involved in regulation of mineral uptake and distribution have been identified in this species.
The present study was designed to carry out RNA-Seqbased identification of genes, particularly those encoding transcription factors, whose expression is affected in potato roots by Pi starvation. The current study is to explore how StMYB44 (previously named tuber-specific and sucroseinducible element-binding factor), one of the transcription factors identified, is involved in regulation of Pi uptake and distribution in potato plant.

Plant materials and growth conditions
Tetraploid potato (Solanum tuberosum L.), Désirée, plants were grown in a greenhouse under a 14 h light/10 h dark regime at 25 °C. Arabidopsis thaliana (ecotype Columbia) were grown in a growth chamber under a 14 h light/10 h dark cycle at 23 °C. Hoagland solution was used in hydroponic growth of potato plants, and was changed every other day. The Pi starvation was initiated by withdrawing Pi from the Hoagland solution when the potato plants were 1 month old. Roots were collected 5 d after the treatment and stored at -80 °C before RNA extraction.

Plasmid construction, and transformation of Arabidopsis and potato
The coding region of StMYB44 without the stop codon was amplified by PCR and cloned into the pAVA393 vector (von Arnim et al., 1998) to make the StMYB44:GFP fusion gene, which was then subcloned behind a double Cauliflower mosaic virus (CaMV) 35S promoter in the binary vector pCAMBIA1300S (Zhou et al., 2015). The complete vector was verified by sequencing and transformed into Agrobacterium tumefaciens GV3101 by electroporation. Arabidopsis transformation was performed by the floral-dip method (Clough and Bent, 1998).
For the CRISPR/Cas9 (clustered regularly interspaced short palindromic repeat/Cas9) vector, the sequence GAAGATGATACTATCATCAGG of the StMYB44 gene was used as the target sequence. Two primers were synthesized and annealed to form the dsDNA and cloned between two BsaI sites of the pKSE401 vector by Golden Gate cloning (Xing et al., 2014). The complete vector was verified by sequencing.
The 1.5 kb StMYB44 promoter upstream of the translation start codon was inserted between HindIII and BamHI sites of pBI101.2, and then transformed into A. tumefaciens GV3101.
The complete vectors were introduced into potato by Agrobacterium-mediated transformation as previously described (Chronis et al., 2013).
Protein structure analysis and phylogenetic tree analysis Predicted StMYB44 and homologs from Arabidopsis, tomato, tobacco, and cotton were aligned by using Clustal Omega (http:// www.ebi.ac.uk/Tools/msa/clustalo/). The phylogenetic tree was built with the Molecular Evolutionary Genetics Analysis (MEGA) software. Bootstrap analysis of the phylogenetic tree was performed using 100 replicates.
RNA extraction, library construction, RNA-Seq, and quantitative RT-PCR Total RNA was extracted from roots of potato plants by using an E.Z.N.A. ® Total RNA Kit I (Omega Bio-tek, Norcross, GA, USA). A 5 μg aliquot of total RNA was used for library preparation as previously described (Zhong et al., 2011). Sequencing was conducted on an Illumina HiSeq2500 at the Genomics Resources Core Facility of Weill Cornell Medical College.
Total RNA samples were treated with RQ1 DNase (Promega, Madison, WI, USA) for 30 min to remove genomic DNA, and then converted into cDNA using iScript™ Reverse Transcription Supermix (Bio-Rad, Hercules, CA, USA). Quantitative real-time PCR (qRT-PCR) was conducted in a CFX Connect Real-Time System with iTaq Universal SYBR Green Supermix (Bio-Rad). The thermal cycle involves 95 °C for 3 min, and 40 cycles of 95 °C 15 s and 60 °C for 60 s, followed by melt curve analysis to verify the specificity of amplification. The ΔΔCt method was used to calculate RT-PCR results with the potato Actin gene as an internal control.

RNA-Seq data processing and analysis
Libraries were sequenced on a HiSeq2500 (Illumina) using 101 base, single-end sequencing, and the quality of RNA-Seq data was determined by using FASTQC (v 0.10.1) (http://www.bioinformatics. babraham.ac.uk/projects/fastqc/). Reads were mapped to the reference S. tuberosum Group Phureja DM1-3 genome assembly PGSC v4.03 pseudomolecules (http://solanaceae.plantbiology.msu.edu/ pgsc_download.shtml) using TopHat2 , allowing up to two mismatches. Differentially expressed genes were identified using cuffdiff following normalization of transcript count information to RPKM (reads per kilobase of exon model per million mapped reads) (Mortazavi et al., 2008). Genes with a P-value <0.05 were considered to be differentially expressed.
GUS staining β-Glucuronidase (GUS) activity was assayed as previously described (Jefferson et al., 1987) in transgenic potato seedlings, leaves, flowers, and tubers expressing the Pro MYB44 :GUS chimeric gene using two independent transgenic lines for analysis.
To compare expression of the StMYB44 promoter upon Pi starvation, transgenic seedlings were transferred onto fresh medium with Pi (Hoagland solution) or medium without Pi (Hoagland solution without Pi) and grown for 5 d. GUS activity in the seedlings was examined as above.

Subcellular localization of StMYB44
Agrobacterium cells containing 35S:StMYB44-GFP and 35S:GFP plasmids, respectively, were infiltrated into 4-week-old Nicotiana benthamiana leaves. Three days after infiltration, the leaves were detached and green fluorescent protein (GFP) signals were examined under a Leica TCS-SP5 confocal microscope (Leica Microsystems Exton, PA, USA) with excitation wavelength at 488 nm and emission wavelength at 500-520 nm.
Six-day-old transgenic Arabidopsis seedlings expressing the 35S:StMYB44-GFP and 35S:GFP transgenes were used to study subcellular localization. Nuclei of root cells were stained with DAPI solution at 10 μg ml -1 (w/v) for 10 min, and then washed three times with water. Transgenic Arabidopsis seedlings expressing 35S:GFP were used as the control. GFP and DAPI signals were examined using a Leica TCS-SP5 confocal microscope with excitation wavelengths 488 nm for GFP and 405 nm for DAPI (Zhou et al., 2011).

Pi content determination
Pi content was determined as previously described (Jain et al., 2007) by grinding 6-20 mg of fresh shoot or root samples to a fine power in liquid nitrogen. The ground samples were suspended in 500 µl of 1% glacial acetic acid and immediately frozen in liquid nitrogen again and thawed. After centrifugation at 13 000 rpm for 1 min, 50 µl of supernatant were used in a phosphomolybdate colorimetric assay (Ames, 1966). To make Pi contents comparable, seedlings of wildtype and individual transgenic potato lines were grown in the same Magenta box containing 4.3 g l -1 Murashige and Skoog (MS) salt, 0.17 g l -1 NaH 2 PO 4 ·H 2 O, 0.1 g l -1 inositol, 0.4 mg l -1 thiamine HCl, 30 g l -1 sucrose, and 1.8 g l -1 gelrite. Two weeks after subculture, shoot and root samples were collected for Pi content determination.

Protein-protein interaction by BiFC
The coding sequences of StMYB44, AtWRKY6(At1g62300), and StWRKY6 (NM_001318697, initially named StWRKY31, but it is more similar to AtWRKY6), a homolog of WKRY6 from potato, without stop codons were amplified by PCR and cloned into the KpnI and XmaI sites of the bimolecular fluorescence complementation (BiFC) vectors pSPYCE and pSPYNE, respectively (Waadt et al., 2008). After confirmation by sequencing, the vectors were transferred into A. tumefaciens GV3101 and agroinfiltrated into 4-week-old N. benthamiana leaves. Three days after infiltration, the leaf discs were detached and examined by confocal microscopy for the yellow fluorescent protein (YFP) signal with excitation wavelength at 488 nm and emission filter at 520 nm.

Identification of Pi starvation-responsive genes in potato roots by RNA-Seq
To investigate the regulatory mechanism of potato in response to Pi deficiency, differentially expressed genes in roots under Pi-sufficient (Hoagland solution with 0.5 mM KH 2 PO 4 ) and Pi-deficient (Hoagland solution without Pi) conditions were examined by RNA-Seq. A previous study on rice subjected to Pi starvation elucidated a 2-to 3-fold change in Pi content in shoots and roots, but substantial numbers of differentially expressed genes were not observed until 3-7 d (Secco et al., 2013). Therefore, to obtain a relatively comprehensive list of genes involved in the responses triggered by Pi deficiency, plant materials examined in this study were collected 5 d after Pi withdrawal.
A total of 31.5 million reads were sequenced from six libraries generated from three biological repeats of Pi-deficient and sufficient samples. Statistical analysis indicated the differential expression of 359 genes upon Pi limitation, of which 221 genes were expressed at a minimum 1.6-fold higher level (see Supplementary Table S1 at JXB online). The rest of the genes were found to be reduced >1.6-fold upon Pi starvation as compared with control (Supplementary Table S2). Functional categorization of these genes revealed their involvement in diverse biological processes including cellular response to phosphate starvation and phosphate ion homeostasis (Fig. 1A). Further analysis indicated that several genes, including those encoding Inorganic Phosphate Transporter, four purple acid phosphatases, and three SPX domain-containing proteins were strongly up-regulated after Pi starvation treatment. In contrast, PHOSPHATE2 (PHO2), a gene encoding a ubiquitin-conjugating E2 enzyme mediating the degradation of Phosphate Transporter 1 (PHT1) and PHOSPHATE1 (PHO1), was observed to be dramatically suppressed in potato roots (Supplementary Table S2). These results indicated that a 5 d Pi starvation treatment had successfully triggered comprehensive molecular responses in potato.
The replacement of phospholipids in membranes with glycolipids and sulfolipids is one of the typical responses of plants to Pi starvation (Härtel et al., 2000). In this study, five genes, namely those encoding two glycosyltransferases, 1,2-diacylglycerol 3-beta-galactosyltransferase, digalactosyldiacylglycerol synthase 2 (DGD2), and riboflavin kinase/ FMN adenylyltransferase, involved in the glycolipid biosynthetic process were identified. The Sulfate Transporter 3.4-encoding gene was observed to have greater abundance (17.5-fold increase) upon Pi starvation (Supplementary  Table S1), suggesting an increase in S uptake or transport to meet the demand for the elevated biosynthesis of sulfolipids (Misson et al., 2005).
Previous studies have indicated that starch accumulates in response to Pi deprivation (Calderon-Vazquez et al., 2008;Hammond and White, 2008). The abundances of transcripts of starch synthase VI and two phosphofructokinase genes, involved in starch synthesis, were observed to be ~3-fold higher in Pi-depleted potato roots (Supplementary Table S1). The increased expression of these genes was also reported in Pi-deficient potato leaves (Hammond et al., 2011).

Verification of gene expression by quantitative RT-PCR
qRT-PCR was used to verify the expression of several genes potentially involved in Pi uptake and signaling, including those encoding IDS4 SPX Domain-containing Protein 2-Like, Xenotropic and Polytropic Murine Leukemia Virus Receptor IDS-4, Inorganic Phosphate Transporter, Purple Acid Phosphatase 1, and two purple acid phosphatases. Altered expression of these selected genes was consistent with that from the RNA-Seq approach although the scale of the fold changes differed between two approaches (Fig. 1B).
Among the Pi starvation-responsive genes, a number of targets, including seven up-regulated and nine down-regulated transcription factors, with potential signaling functions in response to Pi starvation were identified (Supplementary Table S3). StMYB44 (PGSC0003DMG400003316), a potato homolog to AtMYB44 and a member of the important MYB family subgroup 22, was down-regulated in roots by Pi starvation, as shown by both RNA-Seq and qRT-PCR (Fig. 1B). This gene was selected for a more comprehensive analysis of its involvement in regulation of Pi starvation responses.

Isolation and structure analysis of potato StMYB44
The ORF of StMYB44 was isolated from potato cultivar Désirée by PCR. Sequencing analysis showed that the 963 bp long ORF encoded a protein of 320 amino acid residues with a predicted molecular mass of 35.02 kDa and an isoelectric point of 9.24 by using Compute pI/Mw software online (http://web.expasy.org/compute_pi/). The deduced protein shared 49% sequence identity with MYB44 in Arabidopsis. AtMYB44 belongs to R2R3-MYB subgroup 22 carrying R2 and R3 MYB repeat domains ( Fig. 2A; Supplementary  Fig. S1). Most members of this group in Arabidopsis contained the conserved motifs 22.1 (TGLYMSPxSP) and 22.3 (GxFMxVVQEMIxxEVRSYM) (Stracke et al., 2001). Further analysis indicated that another conserved motif, 22.2, (D/EPP/MTxLSLP) is present between motifs 22.1 and 22.3 among the members of this group in Arabidopsis. StMYB44 carried the 22.2 and 22.3 motifs but lacked the 22.1 motif ( Fig. 2A, B), indicating that it could have different physiological roles from its homologs in Arabidopsis. In addition, phylogenetic analysis showed that StMYB44 and its orthologs from tomato and tobacco form one clade with high bootstrap numbers (Fig. 2C), indicating that the divergence of StMYB44 occurred after the split of Solanaceae and Brassicaceae.

Expression and subcellular localization of StMYB44
To examine the tissue-specific expression patterns of StMYB44, a 1.5 kb fragment upstream of the start codon was fused to the GUS reporter gene, and transformed into potato. GUS assay showed StMYB44 expression in almost all potato tissues, including young seedlings, roots, mature leaves, flowers, and tubers, although the expression in young leaves, root tips, stigma, and anthers was stronger than that in other tissues ( Fig. 3A-G). Examination of GUS activity in Pro StMYB44 :GUS seedlings grown in either Pi-sufficient or deficient conditions showed reduced staining in roots responding to Pi deficiency, further confirming the results of RNA-Seq and qRT-PCR (Fig. 3H).
To determine the subcellular localization of StMYB44, a StMYB44:GFP fusion protein was expressed in tobacco  leaves by agroinfiltration. Confocal microscopic analysis showed the exclusive accumulation of StMYB44:GFP in the nucleus, whereas only GFP protein driven by the same CaMV 35S promoter was found in the cytosol and nucleus (Fig. 4A).
In addition, roots of 6-day-old seedlings of two stable transgenic Arabidopsis lines expressing 35S:StMYB44:GFP were stained with DAPI, a reagent specifically staining the nucleus. The overlap of the GFP and DAPI signals verified the nuclear localization of the StMYB44 protein, consistent with its function and the transient localization studied in tobacco leaf. As a control, GFP was detected in both the cytosol and nucleus (Fig. 4B).

Overexpression of StMYB44 results in low Pi accumulation in potato shoots
In order to reveal the physiological function of StMYB44, four transgenic potato lines with the highest expression level among 35 independent transgenic lines overexpressing StMYB44:GFP driven by a double 35S promoter were analyzed further (Fig. 5A). Under regular growth conditions in a greenhouse, the StMYB44 overexpression lines were dwarf with small and curly leaves, and produced smaller and fewer tubers, indicating that elevated expression of StMYB44 affected potato development and tuber yield (Fig. 5B-D).
Pi levels were measured in 14-day-old wild-type and transgenic potato seedlings grown on Pi-sufficient medium (MS medium containing 1.25 mM Pi KH 2 PO 4 ). The shoot Pi contents of transgenic plants ranged from 6.25 nmol mg -1 FW to 8.54 nmol mg -1 FW and that of the wild type was 11.02 nmol mg -1 FW (Fig. 6A), while no significant difference was detected between the roots of wild-type and transgenic potato plants (Fig. 6B), indicating the negative effect of StMYB44:GFP overexpression on translocation of Pi from roots to shoots.
In order to dissect the downstream genes controlled by StMYB44, we compared global gene expression profiles between two independent transgenic lines (OE22 and OE30) and wild-type seedlings by RNA-Seq analysis. A total of 80.2 million reads were obtained from nine libraries generated from three biological repeats of wild-type and two transgenic lines. Of the 174 differentially expressed genes, 52 and 122 were discovered to be up-regulated and down-regulated, respectively, over 4-fold (Tables 1, 2). The RNA-Seq analysis showed that the averaged expression of StMYB44 in the two chosen transgenic lines was increased by 44.3-fold, which is consistent with the qRT-PCR analysis of StMYB44overexpressing lines, demonstrating the authenticity of RNA-Seq in identification of the differentially expressed genes in this study. Among the up-regulated genes in the StMYB44overexpressing lines, Purple Acid Phosphatase 3 (PAP3; PGSC0003DMG403007838), a gene involved in the release of the phosphate from phosphate ester under phosphate starvation conditions (Bozzo et al., 2002;Y. Zhang et al., 2014), was identified. At this point, it remained unclear whether the enhanced expression of PAP3 was caused directly by the overexpression of StMYB44 or a feedback response due to reduced Pi accumulation in shoots. More interestingly, the transcript abundance of potato PHOSPHATE1 (StPHO1; PGSC0003DMG400017163) was observed to be reduced in the transgenic potato. In Arabidopsis, PHO1 is responsible for loading Pi into the xylem in roots and its translocation from root to shoot; accordingly, mutation of this gene results in reduced Pi accumulation in shoot tissues (Poirier et al., 1991;Hamburger et al., 2002). Potato StPHO1, sharing 67% amino acid identity with the Arabidopsis PHO1 ( Supplementary Fig. S2), is predominantly expressed in potato roots (Fig. 6C), similar to the expression pattern of PHO1 in Arabidopsis (Hamburger et al., 2002), and is expected to confer a similar Pi-translocating function in potato. qRT-PCR was used to confirm the decreased expression of StPHO1 detected by RNA-Seq, and the result showed that the expression of StPHO1 was significantly reduced in the StMYB44-overexpressing potato roots (Fig. 6D). Since PHO1 is the only identified gene with a known function related to Pi loading and translocation, the lowered Pi accumulation in the transgenic potato shoots could be attributed to the reduction of StPHO1 expression, caused by the overexpression of StMYB44, suggesting that StMYB44 negatively regulates Pi translocation from roots to shoots by specifically suppressing the expression of StPHO1.

Knock-out of StMYB44 by using the CRISPR/ Cas9 system
The CRISPR/Cas9 system was employed to generate StMYB44 knock-out lines in potato (Xing et al., 2014) using the nucleotide sequence from 376 to 396 of StMYB44 mRNA as guide RNA. After Agrobacterium-mediated transformation, a total of 11 kanamycin-resistant potato lines were obtained, and PCR genotyping detected Cas9 in all transgenic lines ( Supplementary Fig. S3A). A fragment of ~300 bp of StMYB44 harboring the target region was amplified by PCR and sequenced ( Supplementary Fig. S3B), indicating that 9 out of 11 lines carried mutant StMYB44 alleles, with a 81.8% frequency of gene editing for this CRISPR/Cas9 system in potato. The targeted mutations ranged from 2 to 120 deleted nucleotides in all of these nine transgenic plants. In addition to deletions, insertions of nucleotide A or T were observed in four lines, C12, C17, C19, and C21, consistent with reports on other plant species (H. Ma et al., 2015). Désirée is a tetraploid potato cultivar, and single nucleotide polymorphism (SNP) analysis of the cloned StMYB44 fragments indicated that it had eight alleles of StMYB44 ( Supplementary Fig. S4). The sequencing results also indicated that wild-type alleles of StMYB44 were still present in C3, C12, C17, and C21, showing that not all the alleles in transgenic potato were modified. To better understand the degree of the reduced expression of StMYB44 in these lines, the expression level of wild-type StMYB44 in leaves was measured by RT-PCR. Expression of wild-type StMYB44 was hardly detected in the selected transgenic plants, while it remained high in wild-type plants ( Supplementary Fig. S3C).
All StMYB44 knock-out lines displayed no visible phenotype compared with the wild type under normal conditions, except one line, C14, which grew more slowly and carried smaller leaves than the wild type. No statistically significant difference was observed in Pi content in shoots and roots between three selected StMYB44 knock-out lines and wild-type plants ( Supplementary Fig. S5A, B). StPHO1 was expressed similarly in both the transgenic potato roots and wild-type roots ( Supplementary Fig. S5C), implying that knock-out of StMYB44 is not enough to increase transcription of StPHO1, probably due to the presence of other negative transcription factors. Expression of StPHO1 and Pi contents were similar in C14 and the wild type ( Supplementary  Fig. S5), suggesting that the abnormal phenotypic change of C14 was most probably caused by an insertion in a development-related gene instead of Pi metabolism.

StMYB44 interacts with AtWRKY6 and StWRKY6 in vivo
Transcription factor AtWRKY6 binds to the W-boxes in the AtPHO1 promoter and suppresses its expression in Arabidopsis (Chen et al., 2009). Three W-boxes [(T) TGAC(C/T)] were identified in the 1.5 kb StPHO1 promoter. In addition, a MYB-binding site I (MBSI) (CCGTTG), located 297 bp upstream of ATG, was identified in the potato PHO1 (Fig. 7A). EMSA showed that Arabidopsis AtMYB44 directly binds to the MBSI motif (Jung et al., 2012). It is highly possible that StMYB44 could regulate the transcription of StPHO1 by directly binding to the MBSI motif as StMYB44 shares a high amino acid identity with AtMYB44 in the DNA-binding domains ( Supplementary  Fig. S1). Furthermore, the juxtaposition of two kinds of cis-elements led to speculation that StMYB44 interacts with AtWRKY6 to form a protein complex. This hypothesis was tested by fusing StMYB44 to the C-terminal half of YFP (StMYB44-cYFP), and AtWRKY6 and StWRKY6 to the N-terminal half (AtWRKY6-nYFP and StWRKY6-nYFP), respectively, and introduced into tobacco leaves by agroinfiltration. Co-expression of 35S:StMYB44-cYFP and 35S:AtWRKY6-nYFP, 35S:StMYB44-cYFP, and 35S:StWRKY6-nYFP resulted in a YFP signal in the nucleus (Fig. 7B). In contrast, no YFP signal was detected when combinations of 35S:nYFP and 35S:cYFP, 35S:StMYB44-cYFP and 35S:nYFP, 35S:cYFP and 35S:AtWRKY6-nYFP, and 35S:cYFP and 35S:StWRKY6-nYFP were expressed (Fig.  7B). These results demonstrated that StMYB44 interacts with AtWRKY6 and StWRKY6 in planta.

Discussion
The regulatory mechanism of Pi starvation responses in plants has been the subject of intensive investigation. A number of key genes controlling Pi homeostasis and regulation have been identified in plants, mainly in Arabidopsis and rice, where mutants or transgenic plants are easier to acquire (Rubio et al., 2001;Hamburger et al., 2002;Chen et al., 2009;Rouached et al., 2010;López-Arredondo et al., 2014). However, relatively limited information is available in species in which either a mutant collection does not exist or the generation of transgenics is more difficult. Here, the RNA-Seq approach was selected which has been widely used to study dynamic changes in gene expression in Pi-deficient plants, including Arabidopsis, white lupin, rice, and wheat (Lan et al., 2012;Oono et al., 2013a, b;O'Rourke et al., 2013;Secco et al., 2013). In the present study, a transcriptomic  analysis was conducted of potato root in response to low P i by RNA-Seq, and StMYB44, one of the transcription factor genes identified, was characterized in more detail.

Genetic responses to Pi starvation in potato roots
A total of 359 genes were identified to be P i deficiency responsive (Supplementary Tables S1, S2). The number of differentially expressed genes is higher than that (147) in Pi-deficient maize roots at day 3 after treatment, and lower than that (967) in maize roots at day 6 after treatment, as reported in previous studies (Calderon-Vazquez et al., 2008). These genes functioned in diverse biological processes as shown by Gene Ontology (GO) analysis, including Pi homeostasis and other related metabolic processes (Fig. 1A), suggesting that Pi deficiency causes profound changes in these processes in potato roots. Common genetic responses to Pi starvation in potato and other plant species regarding Pi uptake, distribution, and signaling, lipid metabolism, carbon assimilation, and other stress pathways were observed, supporting the notion that Pi-deficient responses are largely conserved among plants (Franco-Zorrilla et al., 2004;Calderon-Vazquez et al., 2008). Interestingly, differentially expressed genes involved in two biological processes, cellular responses to fungus and nitric oxide, and negative regulation of endopeptidase activity, were also identified. The connections between Pi starvation and these biological processes were not reported in previous studies and thus could be interesting to explore in future research.   Although the focus of this study was on molecular responses to Pi starvation in potato roots, a comparison of our results with one of the previous studies in which the potato leaf was analyzed led to the identification of similar or distinct metabolic pathways between the two tissues. A few pathways involved in starch accumulation, protein degradation, lipid metabolism, and S uptake were activated, and the expression of the associated genes encoding starch synthase, phosphofructokinase, E3 ubiquitin ligase, and ubiquitin-protein ligase, plus SUT3, was found to be increased in both leaf and root tissues (Supplementary Table S1) (Hammond et al., 2011). In contrast, different responses to Pi deficiency were also observed between shoot and root in potato. For example, the patatin-encoding gene and four Phospholipase A1 (PLA1) genes were down-regulated in roots, while two Phospholipase D (PLD) genes were up-regulated in potato leaves under Pi-limiting conditions (Hammond et al., 2011). As the main tuber storage proteins, patatins also possess phospholipase A2 (PLA2) activity (Senda et al., 1996). PLDs hydrolyze structural phospholipids, while PLAs hydrolyze galactolipids more efficiently than phospholipids and are involved in auxin signaling in roots (Rietz et al., 2010;Canonne et al., 2011). The up-regulation of PLD genes in shoot and down-regulation of PLA genes in root indicated that the breakdown of phospholipids mainly occurs in the shoot while an altered auxin signaling mediated by PLAs occurss in root during Pi deficiency.

StMYB44 is a negative regulator of Pi transport from roots to shoots
The major purpose of this study was to identify the novel signaling transducers in potato in response to Pi deficiency. MYBs are among the well-characterized transcription factors regulating Pi deficiency responses. According to the phenotypic effect of either overexpression or knock-out of these MYB genes on Pi homeostasis, PHR1/OsPHR2, PSR1, AtMYB2, and OsMYB2P-1 had positive effects on Pi uptake or transport (Wykoff et al., 1999;Rubio et al., 2001;Zhou et al., 2008;Dai et al., 2012;Baek et al., 2013), whereas MYB62 negatively regulates Pi content in the shoot by reducing Pi uptake and acid phosphatase activity (Devaiah et al., 2009). Our study demonstrated that StMYB44 plays a negative role in Pi transport from root to shoot by regulating the transcription of PHO1. Genetic analysis has already demonstrated that the transcription of PHO1 is negatively regulated by the transcription factor AtWRKY6 in Arabidopsis (Chen et al., 2009). Regulation of PHO1 by StMYB44, a transcription factor from a different family from AtWRKY6, in plant roots indicated an additional regulatory mechanism of Pi transport, expanding our knowledge of the physiological functions of this gene family.
It is important to realize that the strong shoot morphological alterations in the StMYB44 overexpression lines are less likely to be caused by the reduced allocation of Pi from root to shoot. How StMYB44 mediates the growth and development of potato is worth future exploration, although current interest is focused on its involvement in Pi metabolism.

Control of PHO1 expression by multiple transcription factors
PHO1 is responsible for Pi transport from roots to shoots by loading Pi to the xylem (Hamburger et al., 2002). Transcription of PHO1 is under tight control in response to Pi availability since PHO1 was induced by Pi starvation and quickly recovered by Pi resupply in rice (Secco et al., 2013). A number of cis-elements, which can be recognized by several regulatory proteins including MYB transcription factors, in the promoter of Arabidopsis PHO1 were predicted. Similarly, the promoter region of StPHO1 was predicted to harbor several regulatory cis-elements, including binding sites for both WRKY transcription factors (W-box) and MYB transcription factors, suggesting that MYB transcription factors, such as StMYB44, could be involved in the regulation of StPHO1 expression by binding directly to its cis-elements. Moreover, it is known that not only can WRKY transcription factors physically interact with other members in the same family but they can also interact with transcription factors or regulatory proteins in other families. For example, AtWRKY6 and AtWRKY42 interacted with each other in Arabidopsis (Chen et al., 2009). HvWRKY38 interacted with Barley Prolamin-Box Binding Factor (BPBF), a non-WRKY transcription factor, to repress the expression of Amy32b in barley aleurone cells (Zou et al., 2008). These results demonstrated that these interactions could play an important role in the regulation of genes controlled by WRKY proteins, as documented previously (Chi et al., 2013). This study showed that StMYB44 physically interacts with AtWRKY6 and StWRKY6 in vivo. To our knowledge, this is the first time that these two classes of transcription factors, WRKY and R2R3 MYB, were demonstrated to interact in the nucleus. These interactions allow us to propose that StMYB44 forms a complex with StWRKY6 in potato to regulate StPHO1 expression synergistically. Under normal conditions, expression of PHO1 is tightly controlled by the transcriptional complex to avoid overaccumulation of Pi in shoots, while upon Pi deficiency, removal of repressors StWRKY6 and/or StMYB44 leads to a lowered abundance of the transcription factor complex, facilitating the transcription of PHO1 and associated Pi transport from root to shoot. Further studies, such as functional analysis of other Pi deficiency-responsive transcription factors or identification of StMYB44-interacting proteins, would not only advance our knowledge on the regulatory mechanism of potato in response to Pi starvation, but also shed light on the selection of candidate genes that could be used for genetic enhancement of Pi deficiency tolerance in potato and other crops.
RNA-Seq data in this study have been deposited in GenBank with accession no. SRP083083.

Supplementary data
Supplementary data are available at JXB online. Fig. S1. Comparison of StMYB44 and other MYBs from Arabidopsis. Fig. S2. Alignment of Arabidopsis PHO1 and StPHO1. Fig. S3. Knock-out of StMYB44 by the CRISPR/Cas9 system. Fig. S4. Eight SNPs in the StMYB44 fragment in Désirée. Fig. S5. Pi levels and expression of StPHO1 remain unchanged in StMYB44 knock-out lines. Table S1. Up-regulated genes by Pi starvation in potato roots. Table S2. Down-regulated genes by Pi starvation in potato roots. Table S3. Transcription factors responding to phosphate starvation. Table S4. Sequences of primers used in this study.

Author contributions
XZ and CZ conceived the project and designed the experiments; XZ and MZ performed the experiments; JH and MI provided technical assistance; XZ, LL, and CZ analyzed the data and wrote the article.