Functional analysis of the HD-Zip transcription factor genes Oshox12 and Oshox14 in rice

The homeodomain-leucine zipper (HD-Zip) transcription factor family plays vital roles in plant development and morphogenesis as well as responses to biotic and abiotic stresses. In barley, a recessive mutation in Vrs1 (HvHox1) changes two-rowed barley to six-rowed barley, which improves yield considerably. The Vrs1 gene encodes an HD-Zip subfamily I transcription factor. Phylogenetic analysis has shown that the rice HD-Zip I genes Oshox12 and Oshox14 are the closest homologues of Vrs1. Here, we show that Oshox12 and Oshox14 are ubiquitously expressed with higher levels in developing panicles. Trans-activation assays in yeast and rice protoplasts demonstrated that Oshox12 and Oshox14 can bind to a specific DNA sequence, AH1 (CAAT(A/T)ATTG), and activate reporter gene expression. Overexpression of Oshox12 and Oshox14 in rice resulted in reduced panicle length and a dwarf phenotype. In addition, Oshox14 overexpression lines showed a deficiency in panicle exsertion. Our findings suggest that Oshox12 and Oshox14 may be involved in the regulation of panicle development. This study provides a significant advancement in understanding the functions of HD-Zip transcription factors in rice.


Introduction
Plant genomes contain a large number of transcription factors (TFs) that regulate the expression of several downstream targets. In Arabidopsis, approximately 1,500 TFs have been identified and are divided into a number of classes, such as the MADS box, AP2/ERF, Dof, Myb, Hsp, bZIP, NAC and homeobox genes [1][2][3]. In addition, the rice genome contains more than a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Plant transformation and growth conditions
Oryza sativa (L.) Japonica cultivar Zhonghua 11 was used for stable rice transformation. Embryonic calli were induced on scutella from germinated seeds and rice transformation with the binary vector constructs was performed as described previously except that Agrobacterium tumefaciens strain LBA4404 was used [38]. Prior to growth in the greenhouse, transgenic seedlings were selected on a half-strength Murashige-Skoog medium supplied with 0.7% type I agarose (Sigma) and 25 mg/mL hygromycin B (Duchefa, Haarlem, The Netherlands). Regenerated transgenic plantlets were transferred to the greenhouse and grown in hydroponic culture with a regime of 16 h light, 28˚C and 85% relative humidity.
Transgenic and wild type rice seeds were first surface sterilized with 70% ethanol for 30 s and 2% sodium hypochlorite (v/v) for 30 min. The seeds were then rinsed five times in sterile water and immersed in water in the dark for two days at 28˚C to induce germination. Additionally, transgenic seeds were selected for one week on half-strength MS media containing hygromycin B to screen transgenic plants. Finally, the germinated seeds were transferred to the greenhouse in three L pots (diameter 19 cm, depth 14.5 cm) filled with soil. The conditions in the greenhouse were as follows: temperature, 28˚C day/25˚C night; photoperiod, 12 -h light/dark; 85% relative humidity, and 450 μM m −2 s −1 light intensity. Plants were watered twice a week using modified half-strength Hoagland nutrient solution [39].
To evaluate the agronomic traits of the transgenic rice plants, plant height, number of tillers per plant, panicle length and number of primary branches per panicle, were measured at maturity in ten plants from each of three independent lines. The data were analyzed by Student's ttest. The plant height was measured from the base of the stem to the top of the flag leaf.

Yeast one-hybrid system
To study the DNA binding properties of the Oshox12 and Oshox14 proteins, expression vectors for use in the yeast one-hybrid system were made. The full length cDNA of Oshox12 was amplified from λFLC-1-B-Oshox12 with the Oshox12 cDNAFW (5'-CGGAATTCCCATGG GCCGTGAGGAGGATGAGAAG-3') and Oshox12/14 cDNAREV 5'-(GCGTCGACCC CTCGACGGATCAGGCCCTTA-3') primers, then the EcoRI and SalI fragment of the Oshox12 full length open reading frame (ORF) was cloned into yeast expression vector pRED-ATGb cut with the same restriction enzyme, resulting in pRED-ATGb-Oshox12.
For the yeast expression vector of Oshox14, the full length cDNA of Oshox14 was first amplified from λFLC-1-B-Oshox14 with the Oshox14 cDNAFW (5'-CGGAATTCCCATGGA CCGATACGGCGAGAAGCA-3') and Oshox12/14 cDNAREV primers, and then the NcoI and XhoI fragment of Oshox14 ORF was cloned into pACTII (pACTII-Oshox14). After the sequence was confirmed, the NcoI and BglII fragment of Oshox14 derived from pACTII-Oshox14 was subcloned into pUC28 (pUC28-Oshox14) and cut with NcoI and BamHI. The EcoRI and SalI fragment of the Oshox14 full-length ORF from pUC28-Oshox14 was then inserted into pRED-ATGa with the same enzymes in frame, resulting in pRED-ATGa-Oshox14. Yeast transformations were performed by the LiAc method, essentially as described by Gietz [40]. Yeast transformants were grown on a selective medium without histidine and uracil but with 10 mM 3-AT (to suppress background growth on CM minimal medium lacking histidine) [41]. The yeast reporter strains 4AH1-HIS3 and 4AH2-HIS3 have been described previously [42,43]. These strains contain tetramers of the AH1 (CAAT(A/T)ATTG) and AH2 (CAAT(C/G)ATTG) sequences which are consensus binding sites for HD-Zip I and II proteins respectively. The 4AH1 and 4AH2 sequences are in front of the HIS3 reporter gene which is integrated via the pINT1 yeast one-hybrid system at the non-essential PDC6 locus [42]. All handlings of yeast were performed as described previously [43][44][45].

Protoplast isolation and transformations
Protoplast isolation was performed as described by Chen [46]. To isolate protoplasts from young seedling tissues, rice seeds were germinated on half-strength MS medium under light for three days. Seedlings were then cultured on the same medium in the dark at 28˚C for 10-12 days. Tissues of young seedlings (the stems including sheaths) were cut into approximately 0.5 mm strips and placed in a dish containing K3 medium [47] supplemented with 0.4 M sucrose, 1.5% cellulase R-10 (Yakult Honsa) and 0.3% macerozyme R-10 (Yakult Honsha). The chopped tissue was vacuum-infiltrated and digested at 28˚C with gentle shaking at 40 rpm. After incubation, the K3 enzyme medium was replaced by the same volume of W5 solution (154 mM NaCl, 125 mM CaCl 2 , 5 mM KCl and 2 mM MES, adjusted to pH 5.8 with KOH). Protoplasts were released by shaking at 40 rpm for 1 h, followed by filtering through a 35 μm nylon mesh. Protoplasts were collected by centrifuging at 1,300 g for 5 min at 4˚C. Pellets were resuspended in suspension solution (0.4 M mannitol, 20 mM CaCl 2 , 5 mM MES, adjusted to pH 5.8 with KOH). Transfection with effector/reporter constructs was performed as follows: 200 μL (usually 1.5~2.5×10 6 cells/mL) of suspended protoplasts was added to the tube with 10 μg plasmid DNA (including the effector and reporter); then, 220 μL of 40% (w/v) PEG 4,000 prepared with 0.1 M Ca(NO 3 ) 2 and 0.4 M mannitol solution, pH 7, was added, and the mixture was incubated at room temperature for 20 min. After incubation, 750 μL W5 medium was added slowly without mixing, and the protoplasts were transferred to a microtiter plate (12 wells) with 750 μL W5 medium, which was incubated overnight in a room at 25˚C in the dark [46]. After 16 h incubation, protoplasts were harvested and lysed in GUS extraction buffer. After centrifugation, the soluble protein concentration was determined using the Bradford assay [48].

Subcellular localization analysis
To prepare the GFP-tagged translational fusion constructs, the coding region of Oshox12 was amplified by PCR from construct Pro35S-Oshox12 using primers 35Sfor1 (5'-ATCCCA CTATCCTTCGCAAGACCC-3') and Oshox12GFPR (5'-CATGCCATGGCGCTGAAT TGGTCGTAGA-3'). Oshox14 was amplified by PCR from construct Pro35S-Oshox14 with primers 35Sfor1 and Oshox14GFPR (5'-CATGCCATGGCGATCAATCCATACAGG-3'). The resulting fragments were cut with SalI and NcoI and fused in frame to the N-terminus of the sGFP (S65T) coding sequence under the control of the CaMV 35S promoter in vector pTH-2 [50] and the sequences were confirmed (Baseclear, The Netherlands). Subcellular localization of the Oshox12-GFP and Oshox14-GFP fusion proteins and a GFP control in protoplast using transient transformation was performed as described above. The GFP signal was visualized with confocal laser scanning microscopy (Leica SP5) at 16 h after transformation.

Southern and northern blot hybridization
Southern and northern blot analyses were performed as described by Memelink et al. [51]. For Southern blot analysis, rice genomic DNA was isolated from young leaves in 96 tube-tacks (Qiagen) by dry-grinding using a Mixer Mill MM300 (Retch, Germany) with 4 mm stainless steel beads followed by DNA extraction according to Pereira and Aarts [52]. Ten μg of DNA per sample was digested with HindIII (only one cut site in the T-DNA region), fractionated on a 0.8% agarose gel run in TAE and transferred onto Hybond N + membranes (Amersham) under alkaline conditions. The hygromycin phosphotransferase II (hptII) gene (1 Kb) was excised from vector pC1300intB-35SnosBK (GenBank accession AY560326) as XhoI fragment. Hybridizations were performed with 32 P-labelled hptII-probe at 65˚C in hybridization mixture (10% dextran sulphate, 1 M NaCl, 1% SDS, 100 μg/mL of denatured salmon sperm DNA). The membranes were washed once in 2X SSC and 1% SDS at 65˚C for 30 minutes and once in 2X SSC and 0.1% SDS at 65˚C for 30 minutes. For northern blot analysis, 20 μg of total RNA per sample was electrophoresed in formaldehyde agarose gel and transferred to Hybond-N + membrane. Baked blots were (pre)-hybridized in 1 M NaCl, 1% SDS, 10% dextrane sulfate and 50 μg/mL denatured herring sperm DNA at 65˚C, washed with 0.1 XSSPE and 0.5% SDS at 42˚C and autoradiographed. Probes were labeled by random priming with 32 P-dCTP. Equal loading of RNA samples was verified on the basis of ethidium bromide staining of ribosomal RNA bands.

GA treatment of plants
To evaluate the response of sheathed panicle to exogenous GA, transformed lines overexpressing Oshox14 lines with strong phenotype were sprayed with 20 μM GA 3 (Gibberellic acid) at the end of panicle differentiation. For each independent line, five transformed plants were treated.

Histochemical localization of GUS activity
Plant materials were vacuum infiltrated for 20 min in GUS staining solution containing 100 mM phosphate buffer pH 7.7, 2 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid (X-Gluc; Biosynth AG), 0.5 mM potassium ferricyanide, 10 mM EDTA and 0.1% Triton X-100 and incubated at 37˚C for 1 h to overnight, depending on staining intensity. The samples were cleared by several changes of 70% (v/v) ethanol and stored at 4˚C.
For sectioning, the samples were dehydrated in a graded ethanol series from 70% to 100% and embedded in Technovit 7100 resin (Kurzer, Wehrheim, Germany), polymerized at 37˚C, and cut into 3-5 μm sections that were stained with toluidine blue. The samples were viewed using a Leica MZ12 stereo microscope or a Leitz Diaplan microscope with bright-field optics settings, and images were acquired with a Sony 3CCD Digital Photo Camera DKC-5000. Greenhouse-grown plants were photographed with a Canon EOS 350D camera.

Phylogenetic analysis of Oshox12 and Oshox14
Previous work has shown that the barley Vrs1 gene suppresses the development of lateral spikelets and that loss of function in Vrs1 lines results in complete conversion of two-rowed barley into six-rowed barley [34,36]. BLAST searches of the rice (http://rice.plantbiology.msu. edu/analyses_search_blast.shtml) with Vrs1, found that the rice HD-Zip I family Oshox14 and Oshox12 proteins had the highest similarity. Oshox14 is the closest homologue to Vrs1 based on the sequence comparison, but Oshox14 is not highly expressed in panicles compared to Oshox12 [12]. Thus, Oshox14 may be closer to Vrs1 in function than Oshox12 is. In Arabidopsis, the HD-Zip I members Athb-53, Athb-21 and Athb-40 are closest to Vrs1 and HVhox2 [36]. Furthermore, previous studies have shown that rice Oshox12, Oshox14 and Arabidopsis Athb-53, Athb-21, Athb-40 are all in the so-called δ clade, which is characterized by a unique intron between the fourth and fifth leucine of the zipper region (the so-called L4-L5 group) whereas all other family I HD-Zips in rice have the intron between L5 and L6 [12].
To further determine the evolutionary distances among these HD-Zip I proteins and Vrs1, a systematic phylogenetic analysis of the HD-Zip I proteins isolated from Arabidopsis, barley and rice was performed. This phylogenetic analysis confirmed that Oshox12, Oshox14 and Vrs1 were in the same clade (S1A Fig). Alignment of the entire amino acid sequence showed that rice Oshox14 shared the maximum amino acid sequence similarity with Vrs1, and the degree of full length protein sequence identity to Vrs1 reached 63.27% (S1B Fig 53]. To confirm affinities of Oshox12 and Oshox14, we studied the binding of Oshox12 and Oshox14 using yeast one-hybrid system. For this experiment, yeast strains containing a chromosomally integrated HIS3 reporter gene preceded by upstream AH1 (construct 4AH1-HIS3) or AH2 (construct 4AH2-HIS3) tetramers were used. The results showed that the 4AH1-HIS3 yeast strain transformed with either pRED-ATGb-Oshox12 or pRED-ATGa-Oshox14 grew well on a medium lacking histidine but containing up to 10 mM 3-AT (Fig 1A  and 1B), whereas no growth was observed in yeast strains with the 4AH2-HIS3 reporter or with the empty pRED-ATGb expression vector. Our results indicate that both Oshox12 and Oshox14 are able to bind the AH1 sequence, but not AH2 in yeast.

Interaction of Oshox12 and Oshox14 with the AH1 (CAAT(A/T)ATTG) sequence in rice.
To further confirm binding of the Oshox12 and Oshox14 proteins to the AH1 sequence, transient expression assays were carried out with effector and reporter plasmids in rice protoplasts. Two reporter plasmids, 4AH1-90-GUS and 4AH2-90-GUS, in which the AH1 and AH2 tetramers were fused to a CaMV-90 CaMV 35S minimal promoter were used [54]. The constructs Pro35S-Oshox12 and Pro35S-Oshox14, which contained Oshox12 and Oshox14 expressed under control of the CaMV 35S promoter, were used as effectors (Fig 2A). The GUS expression in protoplasts indicates that Oshox12 and Oshox14 are capable of activating transcription of the reporter gene when the upstream HD-Zip binding site AH1 is present, but cannot activate transcription of the reporter gene when upstream HD-Zip binding site AH2 is present (Fig 2B and 2C). These results show that Oshox12 and Oshox14 can bind specifically to the AH1 (CAAT(A/T)ATTG) DNA sequence and activate reporter gene expression in rice protoplasts, which is consistent with the result obtained in the yeast experiments and is also in line with results obtained for other HD-Zip I and II proteins in gel shifts and yeast experiments with AH1 and AH2 [42,43].
Subcellular localization of Oshox12 and Oshox14 and expression pattern of ProOshox12-GUS and ProOshox14-GUS in rice. To study the subcellular localization of Oshox12 and Oshox14, full length Oshox12 and Oshox14 clones were fused in frame to GFP, expressed from the CaMV 35S promoter and observed in transiently transformed rice protoplasts. As shown in Fig 3A, in the control vector, GFP signals were observed in both the cytosol and the nucleus. In contrast, we observed that the Oshox12-GFP and Oshox14-GFP signals were located exclusively in the nucleus, suggesting that both Oshox12 and Oshox14 are nuclear-localized proteins (Fig 3A).
The expression patterns of Oshox12 and Oshox14 were further studied using a promoter-GUS fusion construct. In total 20 and 33 independent transgenic Nipponbare lines were made that expressed the constructs ProOshox12-GUS and ProOshox14-GUS, respectively. GUS reporter gene activity was detected in seedlings and in tissues of mature plants. The X-Gluc staining showed that Oshox12 is expressed in nodes and young leaves and in the vegetative growth stage, highly expressed in glume, anther, palea and lodicules in mature plants (Fig 3B). Previously, RT-PCR results showed that Oshox12 was predominantly expressed in panicles at 10 and 15 DAF [12]. The GUS staining result is consistent with the Oshox12 expression profile deducted from the Rice Genome Annotation Project (RGAP, http://rice.plantbiology.msu. edu/index.shtml) Database (S2A Fig) and a recent work by Gao et al. [55]. ProOshox14-GUS was mainly expressed in the reproductive organs, such as anther and pistil (Fig 3C), which is also consistent with the results from the RGAP Database (S2A Fig). Although the RT-PCR results implicated that the highest level of Oshox14 expression was found in the stem rather than in the other detected organs, no expression of Oshox14 in leaf sheath [12]. In addition, the available microarray-based expression profile for different development stages suggest that both Oshox12 and Oshox14 are highly expressed in calli, hull and panicle, with low expression in the radicle and root (S2B Fig) [56].
Phenotypes of transgenic rice plants overexpressing Oshox12 and Oshox14. Oshox12 was further investigated by gain-of-function studies. For this, thirty-one independent T 0 lines were obtained and overexpression of Oshox12 was confirmed by northern blot analysis (S3A Fig). Further Southern blot analysis showed that 16 lines were single-copy (S3B Fig). Three independent transgenic lines (OX12-23, OX12-29 and OX12-33), with high expression levels and obvious phenotypic differences were selected for further phenotyping. We observed that overexpression of Oshox12 induces a semi-dwarf phenotype accompanied by low fertility ( Fig  4A). Although both wild type and Oshox12-OX lines had five nodes at maturity, plant height of Oshox12 transgenic plants was reduced because of the shortened uppermost internode (data not shown). The average plant heights of the three Oshox12-OX lines were 63.7 cm, 66.34 cm and 64.7 cm, respectively, whereas that of wild type plants reached 85.1 cm on average ( Fig  5A). Thus, the average plant height in the three Oshox12-OX lines was decreased by 23% (P<0.01). Furthermore, the Oshox12-OX lines displayed a decrease in tiller number (though this effect was not significant) ( Fig 5B).

Fig 2. Interactions of Oshox12 and Oshox14 with the HD-Zip binding site AH1 (CAAT(A/T)ATTG) in a transient expression system using rice protoplasts. (A) Schematic overview of the effector and reporter constructs
Oshox12 is predominantly expressed in the panicle suggesting that it has a function in the development of this tissue [12]. An examination of the panicle architecture in the Oshox12-OX lines revealed significant decreases in panicle axis length and primary branch number (Fig 5C  and 5D). In the transgenic plants, the main panicle length was reduced by 20% from an average of 19.6 to 15.6 cm (P < 0.01, n = 10) (Fig 5C). The number of primary branches per main panicle (Fig 5D) was also determined. On average, panicles from lines OX12-23, OX12-29 and OX12-33 had 6.4, 7.2 and 6.8 primary branches, respectively, while the wild type panicles had 9.6 primary branches, representing a significant reduction in the Oshox12-OX lines (Fig 5D). In addition, we found the grain number to be reduced in the Oshox12-OX lines (Fig 4A panels  b-c). On average, the grain numbers from lines OX12-23, OX12-29 and OX12-33 were 324.6, 379.6 and 357, respectively, while the wild type had 803.8 grains per plant; this reduction was also significant. Taken together, these data indicate that Oshox12 might be involved in panicle development.
Our next step was to examine whether Oshox14 functions as a developmental regulator and to determine whether it shows functional similarities to Vrs1, for which purpose we made transgenic rice plants overexpressing Oshox14. Thirty-four independent T 0 lines were obtained and over-expression of Oshox14 was confirmed by northern blot analysis (S4A Fig). Southern blot analysis showed that four lines were single-copy (S4B Fig). Three primary transformants with high Oshox14 expression levels of Oshox14, containing the sense gene construct (lines OX14-9, OX14-10, OX14-45) were found to be severely retarded in growth at the seeding stage (Fig 4B panel a) and showed difficulties with panicle exsertion through stem and leaf sheath at the mature stage (Fig 4B panels c, e). The plants with the strongest phenotypes showed fully sheathed panicles. To clarify whether this defect was accompanied by abnormalities in leaf sheath development or internode elongation, we performed an anatomical study of sections from the first internodes and leaf sheath. The results showed that no difference between the first internodes at anatomical level (Fig 4C panels a, b); however, the Lugol staining experiment showed decreased starch content in stems of the Oshox14-OX plants (Fig 4C panels c,  d). Further histological sectioning showed that additional differences in the structures of the leaf sheath. In general, the Oshox14 overexpressing lines have more turns of the flag leaf sheath than that of the wild type ( Fig 4D). The severity of the phenotype in these transgenic plants was correlated with the expression levels found in northern blotting (S4A Fig). It is known that the leaf sheath from rice elongates rapidly in response to treatment with GA [57]. Thus, we treated Oshox14 overexpressing lines with the strong phenotypes with 20 μM GA 3 at the end of panicle differentiation, which led to the panicle being exserted from the culm and the flag leaf (Fig 4B panel d).
Due to the phenotypic abnormalities in the lines with weaker phenotypes, we could obtain only a small number of T 2 seeds for further study, which included phenotyping for plant height, tiller number, main panicle length and numbers of primary branches per panicle ( Fig  6). Though line OX14-27 displayed only a non-significant decrease in plant height (Fig 6A), the tiller number, main panicle length and numbers of primary branches per panicle, were significantly different than those of the wild type (Fig 6B-6D). Examination of the panicle architecture in the zero expression line OX14-30 showed no difference from that of the wild type (Fig 6). This result may be explained by the weak overexpression of the Oshox14 construct in OX14-27.

Discussion
The HD-Zip TF family is one of the largest super-families of homeobox genes in plants [6][7][8][9]12] and an increasing amount of knowledge is being acquired about their functions in rice [7,27,28,54,58]. In barley, the HD-Zip I member Vrs1 is important in determining grain yield [34, 36] via an effect on inflorescence architecture. The architecture of the inflorescence plays a key role in the determination of grain yield, but our understanding of the genetic control of this complex trait is still limited [59]. In rice, the HD-Zip I genes, Oshox12 and Oshox14 are close homologues of Vrs1 [12,36]. Based on an overexpression analysis of these two genes, we propose a function of Oshox12 and Oshox14 in panicle and sheath development.
The model plants Arabidopsis and rice have 17 and 14 HD-Zip I genes, respectively [11,12]. Rice Oshox12 and Oshox14 and Arabidopsis Athb-21, Athb-40 and Athb-53, belong to a relative small subfamily characterized by an intron between the fourth and fifth leucine of the zipper region (originally called the δ clade or L4-L5 group) whereas many other HD-Zip I genes have an intron between the fifth and sixth leucine of the zipper region [11,12]. The transcript levels of Arabidopsis Athb-21, Athb-40 and Athb-53 are upregulated upon exposure to ABA and salinity stress [14] and these genes are thought to be involved in ovule development [60]. During root development, Athb-53 also plays an important role in auxin/cytokinin signaling [61]. Based on phylogeny, a set of 178 HD-Zip I proteins from different plant species was divided into six groups (I to VI) [62]. Based on this analysis, the δ clade members Oshox12 and Oshox14 [12] were included in group VI [62]. Furthermore, this analysis revealed a set of 20 conserved motifs in the amino-terminal (NTR) and carboxy-terminal regions (CTR). Group VI proteins, including Oshox12 and Oshox14 share in common that they have a so-called motif 10 in the NTR which is also unique to this group but for the moment a precise function is yet unclear although some group VI proteins have a nuclear localization signal in motif 10 [62]. In addition, Oshox12 and Oshox14 have three and six putative phosphorylation sites (Ser, Thr, Tyr), respectively in the CTR, but no sumoylation site was found. Both Oshox12 and Oshox14 possess the so-called AHA (Aromatic, large Hydrophobic, Acidic context) motif in the CTR, which is responsible for transcriptional activation. In addition, both TFs contained a high frequency of aromatic amino acid phenylalanine (Phe) in the CTR but a precise function for this phenomenon is unclear yet [62].
Consistent with the known function of TFs, the GFP-tagged fusion constructs indicated that Oshox12 and Oshox14 are both nuclear-localized proteins. A similar result for Oshox12 was also reported elsewhere [55]. In general, HD-Zip I family members bind in vitro and in vivo to the 9-bp pseudopalindromic cis-element, AH1 (CAAT(A/T)ATTG) and AH2 (CAAT (C/G)ATTG [42,43,52,54]. Our yeast one-hybrid experiment suggests that Oshox12 and Oshox14 specifically bind to the proposed AH1 sequence. It is obvious that these proteins can activate reporter gene expression by an intrinsic activation domain which was also observed for other family I proteins from rice [27,43]. Oshox12 and Oshox14 as transcriptional activators were further confirmed by transient assays in rice protoplasts using the GUS reporter gene. HD-Zip TFs generally form homodimers or heterodimers to regulate downstream gene expression [17,42]. Oshox12 was shown that it can form homodimers as well as heterodimers  with Oshox14 in a bimolecular fluorescence complementation (BiFC) system [55]. Interestingly, both Oshox12 and Oshox14 can also interact with ELONGATED UPPERMOST INTER-NODE1 (EUI1) in yeast one-hybrid and electrophoretic mobility shift (EMSA) assays, and the EUI1 gene contains a similar AH1 (CAAT(A/T)ATTG) sequence element in its promoter region [55]. Taken together, the results of the yeast and protoplast experiments support the functions of Oshox12 and Oshox14 as transcriptional activators, which is characteristic of HD-Zip I family TFs [43,44,63].
Several sets of transcriptome data have shown that Oshox12 and Oshox14 are highly expressed in the panicle [12,55]. Like Oshox12, Oshox14 is mainly expressed in the panicle, even though its expression level in the panicle is less than that of Oshox12 [12]. In barley, the Oshox12 and Oshox14 homologue, Vrs1, is involved in determining the number of rows of spikelets by suppressing the development of lateral rows [34]. Base on the microarray data, the expression patterns and our phylogenetic analysis, we suggest that Oshox12 and Oshox14 might be involved in panicle development, which is further supported by our promoter-GUS expression analysis. This experiment revealed that Oshox12 displayed a tissue specific pattern with the highest expression in glume, anther, palea and lodicules. Our data suggest that Oshox12 function is necessary in different tissues and that this gene may be involved in panicle development. According to the GUS analysis, the promoter of Oshox14 was also mainly expressed in reproductive organs, such as anther and pistil.
Defects in the elongation of the uppermost internode lead to panicle enclosure and thus greatly reduce seed production by blocking normal pollination [64]. Our previous work has shown that Oshox12 and Oshox14 are highly expressed in panicles, suggesting a role in panicle development [12]. Consistently, our findings here demonstrate that overexpression of Oshox12 results in reduced length of the panicle axis, reduction of primary branch number and a consequent decrease in grain yield (Fig 4A and Fig 5). At the heading stage, the Oshox12 overexpression lines exhibited a shortened uppermost internode thereby reducing plant height (Fig 5A). Independently of our work, it was recently demonstrated that Oshox12 regulates panicle exsertion in rice by directly modulating the expression of EUI1, which encodes a cytochrome P450 monooxygenase CYP714D1 that deactivates bioactive GAs and plays a crucial role in panicle exsertion in rice. Panicle exsertion principally depends on the elongation of the uppermost internode [55]. In rice, there are six groups of internode elongation mutants, which are classified based on the elongation pattern of the upper internodes [65]. In the'sh' type, the uppermost internode shows no elongation with the panicle enveloped in the leaf sheath, which results in a sheathed panicle. The rice leaf sheath is an important part of the plant where considerable critical metabolic and regulatory activities occur, and these processes eventually control rice height and robustness. Several mutants with sheathed panicle phenotypes have been identified, including shp1-5, dsp1, sui1-1 and sui1-2 [63]. However, the mechanism underlying sheathed panicles remains unclear. In this study, through the overexpression of Oshox14, we found that transgenic plants overexpressing Oshox14 display sheathed panicles, showing that the overproduction of Oshox14 also alters panicle development. Microscopic analysis indicates that the cells in the uppermost internode appear the same in the wild type and the Oshox14 overexpression line, but that the starch content of the transgenic plant stems was decreased. Our experiments with GA treatment showed that the function of Oshox14 in panicle exsertion may relate to GA signaling. It was reported that Oshox12 is also involved in regulating panicle exsertion and response to endogenous GA [55]. Thus, in summary, our results strongly suggest that both Oshox12 and Oshox14 play important roles in regulating the length of the uppermost internode, probably via GA signaling.
In this study, we demonstrate the roles of Oshox12 and Oshox14 in panicle and sheath development. Improving crop productivity by selection for the components of grain yield and for optimal plant architecture has been the key focus of national and international rice breeding programs. However, the detailed molecular mechanisms by which Oshox12 and Oshox14 regulate panicle development remain largely unknown, and further genetic analyses of downstream target genes need to be undertaken, including the use of mutant alleles. Elucidation of these downstream events will be one of the keys in understanding the roles of these HD-Zip I TFs and their potential in rice yields improvement.