Rice osa-miR171c Mediates Phase Change from Vegetative to Reproductive Development and Shoot Apical Meristem Maintenance by Repressing Four OsHAM Transcription Factors

Phase change from vegetative to reproductive development is one of the critical developmental steps in plants, and it is regulated by both environmental and endogenous factors. The maintenance of shoot apical meristem (SAM) identity, miRNAs and flowering integrators are involved in this phase change process. Here, we report that the miRNA osa-miR171c targets four GRAS (GAI-RGA-SCR) plant-specific transcription factors (OsHAM1, OsHAM2, OsHAM3, and OsHAM4) to control the floral transition and maintenance of SAM indeterminacy in rice (Oryza sativa). We characterized a rice T-DNA insertion delayed heading (dh) mutant, where the expression of OsMIR171c gene is up-regulated. This mutant showed pleiotropic phenotypic defects, including especially prolonged vegetative phase, delayed heading date, and bigger shoot apex. Parallel expression analysis showed that osa-miR171c controlled the expression change of four OsHAMs in the shoot apex during floral transition, and responded to light. In the dh mutant, the expression of the juvenile-adult phase change negative regulator osa-miR156 was up-regulated, expression of the flowering integrators Hd3a and RFT1 was inhibited, and expression of FON4 negative regulators involved in the maintenance of SAM indeterminacy was also inhibited. From these data, we propose that the inhibition of osa-miR171c-mediated OsHAM transcription factors regulates the phase transition from vegetative to reproductive development by maintaining SAM indeterminacy and inhibiting flowering integrators.

Introduction microRNAs (miRNAs) regulate gene expression by sequence-specific cleavage or translational repression of cognate mRNAs in plants and animals [1,2]. They are involved in most of the essential physiological processes in plants, including organ development, hormone signalling, and stress response [3][4][5][6]. It is particularly worth mentioning that they have diverse roles in plant development, such as phase transition, flowering, leaf morphogenesis, meristem identity, and other aspects of plant development [7][8][9].
Higher plants pass through a series of developmental states to complete their life cycles. During post-embryonic life, juvenile-adult transition (also known as the vegetative phase change) and vegetative-reproductive transition represent the two main developmental transitions [10]. These transitions are important, firstly, plants can enter a reproductive stage under appropriate environmental conditions only during the adult phase [10]; therefore, the juvenile -adult phase change plays a critical role in plant development. In rice, this phase transition is usually associated with a series of changes in a range of species-specific traits, including the presence of a mid-rib, size and shape of the leaf blades, shoot apical meristem (SAM) size, and photosynthetic rate, etc [11]. Although the mechanisms underlying the vegetative phase change remain largely unknown, recent studies have revealed that miRNAs are involved in this vegetative phase change across species. Of them, miR156 and miR172 are well known for playing critical roles in the phase change of several species, including Arabidopsis [12,13], maize [14] and rice [15]. The expression of these two miRNAs is negatively correlated; thus, miR156 is intensively expressed during the juvenile phase to control shoot development, while miR172 is strongly expressed during the adult phase. miR156 targets the SQUAMOSA PROMOTER BINDING PROTEIN LIKE (SPL) transcriptional factors, which control the transition from juvenile to flowering stage by regulating the expression of a class of MADS box genes [13,16,17]. Over-expression of miR156 prolongs the juvenile-phase, produces more tillers, delays flowering, and reduces the number of spikelet [14,16,18,19]. On the other hand, miR172 targets AP2like transcription factors, promoting both vegetative phase change and floral induction [20]. Over-expression of miR172 leads to earlier flowering and produces abnormal floral organs [21]. Together with the juvenile-adult transition, the vegetative-reproductive phase transition, known as the floral transition, is the most dramatic phase change in plant development. This transition is regulated by a complex genetic network that monitors the developmental state of the plant, as well as the environmental conditions such as photoperiod and level of phytohormones [16,22]. Recent molecular biological and genetic advances have revealed that FLOWERING LOCUS T (FT) in Arabidopsis, a long-day plant, and Heading date 3a (Hd3a) and Rice FT-Like 1 (RFT1) in rice, a short-day plant, encode florigen as a mobile leaf-derived signal aimed to trigger floral transition [23][24][25]. In addition, following the transition from the vegetative to the reproductive phases, the fate of the vegetative SAM changed, transforming into an inflorescence meristem (IM). Throughout this process, the stem cell must balance the maintenance of totipotent, undifferentiated stem cells and generation of differentiation cells [26]. In Arabidopsis, the WUSCHEL-CLAVATA (WUS-CLV) feedback loop pathway is one of the best-characterised signalling mechanisms involved in the regulation of meristem identity maintenance [27]. WUS is known to promote the expression of CLV3 and SAM maintenance, while CLV3 negatively regulates SAM maintenance by restricting the expression of WUS [27]. At the same time, recent studies in both maize and rice have suggested that the WUS-CLV feedback loop pathway is conserved in grasses [28][29][30]. microRNA171 (miR171) family is one of the most ancient and well conserved miRNA families known to date, which has been isolated from a large number of plant species, from mosses to flowering plants [31][32][33]. This family is known to target the HAM (hairy meristem) genes (HAMs, also known as SCARECROW-LIKE6 or LOST MERISTEMS), which encode a member of the GRAS (GAI-RGA-SCR) plant-specific transcription factor family [34,35]. The GRAS genes play important roles in diverse cellular processes, including light and hormone signalling pathways and meristem maintenance [36]. Previous studies divided the GRAS family into 13 subfamilies based on phylogenetic data -AtSHR, AtPAT1, AtSCR, AtSCL4/7, AtLAS, Os19, HAM, Os4, Pt20, DLT, AtSCl3, DELLA, and LISCL, which have distinctly conserved domains and functions [37]. To date, the functions of some of the GRAS genes have been identified in rice. The DELLA protein OsSLR1 plays a role inhibiting gibberellin (GA) signalling [38]. MOC1 encodes a GRAS protein that controls the formation of auxiliary meristems in rice [39]. OsCIGR1 and OsCIGR2 can be induced in the presence of the elicitor N-acetylchitooligosaccharide and exogenous gibberellins [40]. In Arabidopsis, miR171 highly expressed in the inflorescence, and it is known to regulate HAM expression through mRNA cleavage [41]. Overexpression of miR171 or loss of AtHAM1,2,3 function produces pleiotropic phenotypes, including fewer cauline and rosette leaves, reduced shoot branching, increased chlorophyll content, shorter primary roots, and abnormal flower patterning [42,43]. In addition, atham1,2,3 mutant exhibits loss of indeterminacy in both the shoots and roots, aberrant shoot phyllotaxis and lateral organs, and altered meristem morphology [43]. Further, a more detailed analysis of atham1,2,3 mutant demonstrated that HAM1 and HAM2 are important to promote cell differentiation at the periphery of the shoot meristems and to help maintain their polar organization [44]. In barley, over-expression of miR171 has been also associated with pleiotropic phenotypes, including an extended vegetative phase, an increased number of short vegetative phytomers, and a delay in the differentiation of spikelet meristems into floral organs [45]. These results suggest that miR171 plays a conservative role in regulating meristem identity, but the regulation of the phase transition may be monocot-specific functions.
Set against this background, little is known about the function of miR171 and their targets in rice and how miR171 mediates the phase transition from vegetative to reproductive development. In this study, we identified a rice mutant resulting from T-DNA insertion at the promoter of OsMIR171c gene, where OsMIR171c was up-regulated. This mutant produced severely delayed-heading (flowering) phenotype. We found that the rice osa-miR171c-OsHAMs module is involved in the maintenance of shoot apical meristem indeterminacy and vegetative to reproductive phase change.

Plant materials and growth conditions
Rice (Oryza sativa L.) were grown in a controlled paddy of the South China Botanical Gardenduring natural growing seasons or in a greenhouse at 28°C for 14-h (day) and 10-h (night) circadian cycle in winter. This planting was permitted by South China Botanical Garden. The seedlings were grown in Hoagland's Solution [46] under normal growth conditions, and samples were collected at different stages for expression pattern analyses. Rice japonica cultivar Zhonghua11 (ZH11, wild type) and mutant dh were used in this study.

Quantitative RT-PCR analysis of gene expression
Small RNA and total RNA were extracted from different organs of rice using RNAiso for small RNA (Takara, Code No. 9753A) and RNAiso Plus for total RNA (Takara, Code No. 9108) and digested with DNase I (Takara, Code No. 2212) according to the manufacturer protocols. Reverse transcription for small RNA was performed using M-MLV Reverse Transcriptase (Promega, Cat#M1701) in a stem-loop RT-PCR reaction [47]. The specific reaction component and protocol were conducted as described previously [48]. Quantitative real-time RT-PCR (qRT-PCR) reactions for miRNA and mRNA were also performed as previously described [48]. The relative expression of the genes was normalized to the expression level of U6 and e-EF-1a with biological repeats in triplicate. U6 and e-EF-1a were used as internal controls for miRNA and mRNAs, respectively. All primers used are listed in S1 Table. RNA ligase-mediated 5 0 -rapid amplification of cDNA ends (5 0 -RLM-RACE) To identify the cleavage sites of the target mRNAs, a modified 5 0 -RLM RACE was carried out following a protocol previously described [49]. Total RNA was directly ligated to a synthesised RNA adaptor (GCUGAUGGCGAUGAAUGAACACUGCGUUUGCUGGCUUUGAU¬ GAAA), which has a 5 0 -hydroxyl group at both ends and thus can only ligate to 5 0 -phosphorylated RNA, including truncated products of miRNA-guided mRNA cleavage. The ligated product was directly reverse transcribed using an oligo-(dT) primer. The cDNA was amplified by nested PCR, and the final PCR product was gel purified and sub-cloned into the pGEM-T Easy Vector (Promega, Cat#A1360) for sequencing.

Target prediction of miRNAs
The target genes of the miRNA were predicted using psRNA Target (http://bioinfo3.noble.org/ psRNATarget/) with default parameters. The information on the target genes was obtained from Gene Bank and listed in S2 Table. Southern blot analysis Southern blot analysis was performed following a previously described protocol (Roche, Cat#11745832910). For this analysis, ten μg of genomic DNA were digested using Xho I or Xba I. The probe was prepared from a PCR-amplified fragment of the hygromycin B phosphotransferase gene.

Subcellular localization of OsHAMs
A green fluorescent protein (GFP) fusion protein was constructed using full-length OsHAM cDNA with a C-terminal fusion of the GFP clone under the control of a CaMV35S promoter. Rice protoplast preparation and transformation were conducted as previously described [50]. Subcellular distribution of the GFP fusion protein was examined using a confocal laser scanning microscope (ZEISS-510 Meta). Excitation was achieved using an argon laser at 488 nm (GFP), and emission of GFP was detected from 492 to 550 nm. Auto-fluorescence of chlorophyll was simultaneously detected between 650 and 730 nm. The images presented are average projections of 8-20 optical sections.

Statistical analysis
Data were represented as mean ± standard deviation (SD). Student's t-test was used to analyse all the data presented as the mean ± SE. A p value <0.01 or <0.05 was considered as statistically significant. The statistical analyses were performed in SPSS 12.0.

Identification and molecular analyses of a late-heading rice mutant
In our rice T-DNA insertion collection, one line displayed a serious delayed-heading (flowering) phenotype in the primary transgenic plant (T 0 ). Heading date is one of the main agronomic traits in rice [51]. Thus, we aimed to characterise this trait because of its importance in agriculture. The late-heading phenotype was observed from the T 1 to the T 3 generations when the plants were grown in a paddy field under normal growth conditions (Fig 1A). For this mutant, the time required between sowing and heading was at least twice as long as the timing exhibited by the wild-type Zhonghua11 (ZH11) (Fig 1B). This mutant, therefore, was named as dh (delayed heading) mutant. Because dh mutant continues growing and tillering during the long vegetative stage, it eventually produces more tillers (Fig 1D), higher plants (Fig 1C), and more nodes (Fig 1E and 1F) than that by the wild-type ZH11.
We performed molecular analyses to characterize the dh mutant (S1 and S2 Figs). Southern blot analysis showed that the dh mutant presented one T-DNA insertion locus (S1B Fig). Thermal asymmetric interlaced (Tail) PCR [52] was used to identify the T-DNA insertion site. The sequence of the genomic fragment flanking the T-DNA insertion site showed that the T-DNA was inserted in the promoter region of the gene Os04g0623901; therefore, the gene is downstream of Ubiquitin and the 35S promoter in the dh mutant, and there is a double-enhancer in the 35S promoter (S1A Fig), which has observed being able to activate the downstream gene [53]. The corresponding mRNA of Os04g0623901 (OsMIR171c) is AK242153, which is predicted to encode a precursor of the osa-miR171c. The results obtained by qRT-PCR showed that the expression of OsMIR171c was up-regulated at least seventy-fold in the leaves of dh mutant ( Fig 1G and S1C Fig), compared with that of the ZH11. The late-heading phenotype of the dh mutant was observed in the T 0 transgenic rice, suggesting that the mutation may be dominant. To verify this, we performed a co-segregation analysis between the late heading and T-DNA insertion heterozygous plants (S2 Fig). The late-heading phenotype co-segregated with T-DNA, and the segregation ratio obtained for heterozygous plants in a three generation family was well within the expected 3:1 ratio (χ 2 , p = 0.05; S3 Table). These data suggested that dh mutant present a single T-DNA insertion site, and the late-heading phenotype is caused by a single sporophytically controlled Mendelian locus. Similarly, over-expression of miR171 has been also associated with an extended vegetative phase in barley [45]. Therefore, we deduced that T-DNA insertion enhances OsMIR171c expression, which consequently leads to the appearance of the late-heading phenotype in dh mutant.

Mature osa-miR171c is the main up-regulated product in the dh mutant
To understand the biological function of osa-miR171c, we first searched the miRNA database (miRBase) [54], and found that eleven osa-miR171 have been deposited in the database. Sequence alignment indicated that miR171c is conserved across a variety of plant species, and that there are fourteen identical conserved nucleotide positions in two blocks of conserved regions (S3A Fig). Of them, seven OsMIR171 genes generate one identical type of mature osa-miR171 (b/c/d/e/f/m/n). The OsMIR171c gene is transcribed as a long transcript, the osa- The osa-miR171c precursor can be processed into two different mature forms, namely, osa-miR171c and osa-miR171c Ã ; therefore, we examined the expression level of the OsMIR171c transcript ( Fig 1G) and the two mature osa-miR171c forms ( Fig 1H) in the dh mutant by qRT-PCR. The expression level of mature osa-miR171c was up-regulated by 0.5-to 3-fold in different developing organs, compared to that of ZH11; while the level of mature osa-miR171c Ã did not differ between the ZH11 and the dh mutant ( Fig 1H). In addition, the expression level of osa-miR171c was generally much higher than that of osa-miR171c Ã in all the tissues examined. However, the elevated levels of mature osa-miR171c (0.5-to 3-fold) were not directly related to the up-regulated level of OsMIR171c (> 70 fold; Fig 1G and 1H and S1C Fig), it might be explained by that the mature osa-miR171c also is the product of the OsMIR171b/d/e/f/m/n genes (S3A Fig). These results suggested that mature osa-miR171c is the predominant form between the two mature osa-miR171c forms, and it is clearly up-regulated in the dh mutant.

Four OsHAMs transcription factors are the targets of osa-miR171c in rice
The psRNA target server (http://plantgrn.noble.org/psRNATarget) predicted the existence of nine genes (S2 Table) as the targets of osa-miR171c. Eight of them were predicted to be cleaved under osa-miR171c direction, except for AK101142, which may be translationally repressed by osa-miR171c. Therefore, for another eight putative targets except for AK101142, we could check the expression changes in the WT and dh mutant by qRT-PCR. Semi-qRT-PCR analysis showed that these eight genes were expressed in the leaves of ZH11 (Fig 2A); subsequently, rice leaves were used to detect the mRNA expression level of the eight putative targets by using qRT-PCR with primer pairs spanning the miR171 cleavage site. The expression of five genes (Os02g0662700, Os02g0663100, Os04g0555000, Os06g01052000, and Os10g0551200; Fig 2C) were down-regulated in the leaves of the dh mutant; however, the expression of the other three genes ( Fig 2D) showed no obvious change, when compared with those of ZH11.
To further determine whether the 5 target genes are cleaved in the osa-miR171c direction, a 5 0 -RLM-RACE assay was performed using RNA isolated from mixed samples of the shoots and the panicles of ZH11 (Fig 2B). Sequencing of the 5 0 -RLM-RACE clones indicated that the cleavage sites of four genes (Os04g0555000, Os02g0662700, Os02g0663100, and Os06g0105200) were within the osa-miR171c-HAM complementary region ( Fig 2B). The cleavage site of OsHAM3 was different from the others. No cleavage was detected for Os10g0551200 in any of the tested samples. Taken together, we concluded that at least four OsHAM genes could be the targets of osa-miR171c in rice.
The four targets belong to the HAM subfamily of the extensive GRAS family of plant-specific transcription factors (S4 Fig), and they contain all the highly conserved domains of the GRAS family, except for the RVER domain in their C-termini SAW motif (S4A Fig) [35]. A phylogenetic analysis based on HAMs amino acid sequences from HAMII (containing the miR171-binding sequence) of 36 representative species showed that the HAMs formed two distinct groups of monocot and eudicot [43] (S4B Fig). The four osa-miR171c targets were all grouped in the monocot clade; we named these four genes as OsHAM1 for Os02g0662700, OsHAM2 for Os02g0663100, OsHAM3 for Os04g0555000 and OsHAM4 for Os06g0105200. To investigate how osa-miR171c controls OsHAM expression in developing organs in rice, we analysed the parallel expression of osa-miR171c and its targets in ZH11 using qRT-PCR  (Fig 3). First, we tested the expression pattern of osa-miR171c in developing organs using stem-loop qRT-PCR (Fig 3,). The results showed that osa-miR171c was expressed in all the tested organs. The highest expression levels were detected in the booting panicle and the weakest expression levels were in the shoot apex of different developmental stages. In general, osa-miR171c was mainly expressed in reproductive organs, which was consistent with a role regulating the timing of floret initiation and development in rice and consistent with previous data published for Arabidopsis and barley [45,55].
Second, primer pairs spanning the cleavage sites of the target mRNAs were used in a qRT-PCR reaction to detect intact transcripts of the target genes (Fig 3). The four OsHAM genes shared a similar expression pattern. They were expressed in all organs, but their transcript abundance varies considerably among them. The highest expression of the four genes was detected in the shoot apex of the 10-leaf stage seedlings at the onset of floret transition. On the other hand, the lowest expression levels were detected in the booting panicle. In the shoot apex, the expression of the four OsHAM gradually increased with shoot development. When SAM transform into the floral meristem, the expression levels decrease sharply. These expression profiles suggested that OsHAMs plays a role in controlling phase transition.
In conclusion, osa-miR171c expression level was very low at the 10-leaf stage, when plants were during the late adult vegetative phase and immediately prior to flowering initiation, especially in the shoot apex; however, expression levels increased steadily during panicle development (Fig 3). In contrast, OsHAMs were expressed intensively during the adult vegetative phase, although their expression gradually declined along with vegetative to reproductive phase transition (Fig 3). These data strongly support the hypothesis that the gradual decrease of OsHAM expression in the shoot apex during phase transition is inversely proportional to osa-miR171c expression level. However, in other organs, osa-miR171c level was not always inversely correlated with the expression of the four targets, indicating that osa-miR171c and its target genes interact in a complex manner in rice.

Light regulates the expression of osa-miR171c and the four OsHAMs
Nine cis-elements in OsMIR171c promoter were predicted to be involved in the plant response to light conditions (S4 Table). For this reason, we investigated the relation between osa-miR171c and OsHAMs expression under different photoperiodic conditions (Fig 4). During a day and night cycle, osa-miR171c expression increased rapidly with sunlight and reached a peak in the early morning (6:00 to 10:00), it then decreased rapidly and was kept at a relative low levels throughout the day (Fig 4B). In addition, osa-miR171c expression level was higher under long days than under short day conditions (Fig 4A). In contrast, the expression of the four OsHAMs increased during the evening to reach a peak at midnight and then decreased at dawn (Fig 4B). The expression levels of osa-miR171c were inverse correlated with those of OsHAMs following a natural day/night photoperiod.
The expression data obtained using RiceXpro (http://ricexpro.dna.affrc.go.jp/) showed that the expression of the four OsHAMs is diurnal changed (data not shown); therefore, we analysed the circadian pattern in the expression of osa-miR171c and OsHAMs. Under natural photoperiodic conditions, osa-miR171c expression showed a weak diurnal pattern with the expression level peaking just immediately after dawn (Fig 4C). On the other hand, OsHAMs expression exhibited an obvious diurnal expression pattern with the expression levels peaking at night ( Fig  4C). To distinguish whether the expression of osa-miR171c and OsHAMs was mediated by a circadian rhythm control or only by light, we analysed their expression under continuous light and under continuous darkness, by using the rice diurnal gene OsGI as a control [56]. The expression of osa-miR171c was reduced to a relative low level under continuous darkness; Spatial and temporal parallel expression analyses of osa-miR171c and OsHAMs in different ZH11 organs at vegetative and reproductive stages. Data were calculated from three replicates. ZH11 plants were grown in a paddy for microRNA and RNA extraction. The shadow showed inverse correlation of expression change between osa-miR171c and 4 OsHAM genes during transit from vegetative to reproductive stage. e-EF-1a and U6 were used as OsHAMs and osa-miR171c internal controls, respectively. Expression level of osa-miR171c and 4 OsHAMs was normalized with their expression at 2L-R stage, respectively. 2L-R, however, such reduction was not observed when the plants were grown under continuous light (Fig 4D). Similarly, the expression of OsHAMs was inhibited under continuous light while during continues darkness it exhibited the similar diurnal pattern than that observed under natural light conditions (Fig 4D). These results suggested that osa-miR171c and OsHAM expression was primarily regulated by light, rather than indirectly through changes in the circadian rhythm.

Up-regulation of osa-miR171c drastically delays rice heading date
The most conspicuous feature of the dh mutant was the occurrence of delayed heading date ( Fig 1A). To test whether the late-flowering phenotype of the dh mutant can be changed under different photoperiods, we grew the dh mutant under different photoperiodic conditions. The results showed that the late-flowering phenotype of the dh mutant did not change with the photoperiodic conditions (data not shown). We subsequently compared the expression of nine flowering-related genes of the rice flowering regulation model [57] between the dh mutant and ZH11 using qRT-PCR (Fig 5). Leaf samples were collected during the floral transition stage following direction [58], from 21-days-old plants (growing under short and natural day conditions) and 30-days-old plants (growing under long day conditions). The leaf samples were collected 2 h after dawn, when the transcription of these genes was at the highest level [58]. Among the 12 genes investigated, 3 genes (Hda3, RFT1, and Ehd1) showed significantly lower transcription levels in the dh mutant than in the ZH11 for the three photoperiodic conditions studied (Fig 5A-5C), while the other six genes (OsHd1, OsEhd2, OsGhd7, OsMADS56, OsG1, and OsMADS50) did not show a significant difference in expression between dh mutant and ZH11 (Fig 5D-5I). These results also suggested that osa-miR171c and its targets may regulate a flowering pathway different from the pathway that studied before.
We further observed the morphological change of shoot apical meristem (SAM) in dh mutant (Fig 6). SAM of ZH11 stopped producing leaves and converted into an inflorescence meristem (IM) (Fig 6A), with the initial SAM elongation observed at 40 days after germination (DAG); however, the dh mutant continued to produce new leaves at this stage (Fig 6B), indicating that the juvenile-adult transition was delayed in dh mutant. SAM in dh mutant converted into IM after 150 DAG, at a stage when the ZH11 seeds had been already harvested for a long time. To further understand the mechanism underlying the delay in the juvenile-adult phase transition, we used qRT-PCR to examine osa-miR156 and osa-miR172 expression (Fig 6C). In leaves of dh mutant 5-leaf stage seedlings, osa-miR156 expression level was higher than that in ZH11; however, osa-miR172 expression did not differ between them. miR156 has been demonstrated to be related with a delay in the juvenile-adult phase change [19]. Therefore, we deduced that osa-miR171c and its targets played a role in the juvenile-adult phase change.
Finally, scanning electron microscope (SEM) was used to observe SAM morphological changes during vegetative to reproductive phase transition. SAM did not differ morphologically between dh mutant and ZH11 during the seedling stage (Fig 6D and 6G). However, when SAM of ZH11 underwent transition from vegetative to reproductive phase to generate rachis branches (Fig 6E and 6F), SAM in dh mutant still maintained the vegetative identity and continued to produce more leaves than those in ZH11 (Fig 6H and 6I). At the same time, SAM enlarged gradually during development in dh mutant. At 80 DAG, SAM showed a flat shape in roots of 2-leaf stage seedlings; 2L-S, 2-leaf stage seedlings; 10L-L, the 10th leaf; 2L-SA, shoot apex of 2-leaf stage seedling;4L-SA, shoot apex of 4-leaf stage seedling;10L-SA, shoot apex of 10-leaf stage seedling; 1P, developing panicles with a length of 1 cm; BP, booting panicle.  dh mutant (Fig 6J), and at 100 DAG, it became even bigger and flatter (Fig 6K), reaching a three times bigger size than of those in the ZH11 (Fig 6L).
To understand the causes leading to the delayed heading observed in dh mutant, we compared the transcription pattern of four SAM identity genes expressed in shoot apex (before the period which convert to the IM) by qRT-PCR between dh mutant and ZH11 (Fig 6M). The results indicated that the genes OsFON1, OsFON4 (orthologs of Arabidopsis CVL1 and CLV3, respectively), and WOX4 were significantly down-regulated in the dh mutant, whereas OSH1 remained unchanged. As a result, similar to the function of miR171 in Arabidopsis, osa-miR171c affects the maintenance of SAM indeterminacy by regulating WUS-CLV feedback loop. Therefore, we deduced that osa-miR171c and its targets can delay the heading date by influencing the maintenance of SAM identity.

Up-regulation of osa-miR171c results in abnormal reproductive organs
We also analysed the effect of osa-miR171c in the development of reproductive organs (Fig 7). The results showed the diameter of dh mutant stems was significantly wider than that of ZH11 (S6 Fig) and the nodes number of dh mutant were more (Fig 1E and 1F). The panicle of dh  Function of osa-miR171c in Phase Change mutant showed pleiotropic morphological abnormalities, such as shorter and more primary branches, dense spikelet and longer awn (Fig 7C and 7D). The number of primary branches increased almost two-fold from 7 ± 1 in ZH11 (Fig 7B) to 12 ± 2 in dh mutant (Fig 7C and 7D). In dh mutant, the spikelet agglomerated, forming an extremely condensed pattern in the majority of the primary branches; on the other hand, at the top of the lemma in the mutant, a long awn was observed (Fig 7C and 7D). SEM of inflorescence and spikelet development showed that formation of the primary branches, secondary branches, and spikelet did not differ between ZH11 and dh mutant (Fig 7D and 7E).
The transcription of genes (APO1, APO2, and ASP1) related to primary branch development [59,60], were compared by qRT-PCR (Fig 6M) in shoot apex undergoing transition to inflorescence meristem between dh mutant and ZH11. The results showed that the expression of APO1, APO2, and ASP1 did not differ between ZH11 and dh mutant. However, SAM in dh mutant was very large and produced a larger inflorescence meristem than ZH11 (Figs 6E, 6K, 7E  and 7F). In addition, it produced more primary branches, which agrees with a previous report showing that the primary branch number is determined by the initial size of the reproductive apex [61]. These results suggested that osa-miR171c and its target genes are involved in different aspects of meristem function, including initiation and maintenance.

Up-regulation of osa-miR171c leads to abnormalities in leaf morphology
To investigate whether osa-miR171c affects the development of vegetative organs, we analysed leaf morphology using SEM (S8 Fig). Rice leaves typically show a parallel venation pattern, and Function of osa-miR171c in Phase Change the three major longitudinal veins-mid vein (MV), large vein (LV), and small vein (SV) usually lie parallel along the distal axis of the leaf [62,63]. The number of SVs between LVs was quite irregular in leaf blades of dh mutant. ZH11 leaves showed five SVs (S8A Fig); however, the SV number ranged from three to six in dh mutant (S8B Fig). The diameter of the mastoideus was larger in dh mutant (S8D Fig) than in ZH11 (S8C, S8E Fig) on the adaxial side of the leaves. These data showed that osa-miR171c also affects the development of vegetative organs.

Discussion
The miR171 family are highly conserved across land plants, from moss to flowering plants of both monocots and dicots [33]. Based on the characterization of miR171 over-expressing transgenic plants, the miR171 family has been shown to be not only involved in the negative regulation of shoot branching in Arabidopsis [42], but also in the mediation of phase transitions and floral meristem determinacy in barley [45]. Here, we identified that four OsHAMs transcripts are cleaved under direction of osa-miR171c in rice (Fig 2B and 2C), and the up-regulation of osa-miR171c resulted in prolonged vegetative phase and serious delayed rice heading date. We found that the osa-miR171c-OsHAMs module may be involved in three different pathways controlling phase transition (Fig 8).

Interaction between osa-miR171c and OsHAMs
HAM family has been predicted to be the target of miR171 in Arabidopsis, Nicotiana benthamiana, barley, and Larix kaempferi (Lamb.) Carr [41,42,45,64]. In rice, there are seven putative HAM genes [37], five of them contain osa-miR171c recognition sites (S2 Table), while the other two HAMs genes are the putative rice homologs of NSP2 [65]. In our study, we showed that osa-miR171c can cleave OsHAM1, OsHAM2, OsHAM3, and OsHAM4 mRNAs (Fig 2B  and 2C). Several lines of evidence support this conclusion: first, the over-expression of osa-miR171c resulted in a reduction in OsHAMs transcripts; second, 5 0 -RLM-RACE data showed that OsHAMs were the target of osa-miR171c; and third, the expression of OsHAMs was inversely correlated with osa-miR171c expression in shoot apex during floret transition. These results showed that at least four OsHAMs could be the target genes of osa-miR171c in rice.
Positive and negative feedback loops of miRNA-target modules have been described for the miRNA regulation in plants [13]. In Arabidopsis, miR171a, miR172b, and miR156a are found to be positively regulated by their target genes [13,66]. We found that the inverse expression correlation between four OsHAMs and osa-miR171c was mainly observed in the shoot apex during floret transition (Fig 3). In other organs, the expression patterns of osa-miR171c and OsHAMs nearly coincided in the temporal scale (Fig 3), similar to that reported in barley and Arabidopsis [42,45]. Therefore, we proposed that this expression pattern is related to the existence of a regulatory feedback loop between HAMs and MIR171 genes.
The osa-miR171c-OsHAMs module negatively regulates phase change in rice Our findings suggested that osa-miR171c represses the juvenile-adult phase change in rice by regulating the expression of miR156, in addition, up-regulation of osa-miR171c caused the delayed vegetative phase transition traits, such as continuously produce of the leaf primordium (Fig 6B). miR156 and miR172 have been shown to play a critical role in the vegetative phase change in several plant species [13,18]. The maize Cg1 mutant has been shown to extend juvenile phase through the over-expression of miR156 [14]. Down-regulation of the target gene Glossy15 of miR172 exhibited a shortened juvenile phase [67]. In rice, the phenotype associated with the over-expression of miR156 is similar to that of dh mutant, delaying the heading date and increasing tiller numbers [68,69]. A phenotype over-expressing miR172 was not observed in dh mutant [21]. In the dh mutant, miR156 expression was up-regulated and miR172 expression remained unchanged (Fig 6C), which is similar to the expression patterns of miR156 and miR172 found in the miR171 over-expressing barley [45]. Recently, observations by Xue et al. [66] showed that the targets of miR156 and miR171 HAM-SPL can interact with each other to affect a series of developmental events, including flowering, by repressing SPL activity in Arabidopsis. Thus, the interplay between the two timing miRNAs expression played an important role in phase change.
Despite showing a prolonged juvenile phase, dh mutant also delayed the heading date ( Fig  1A and 1B). Photoperiod is the most important environmental cue for rice flowering [70]. Rice presents two independent photoperiod pathways, one involving Hd1 and the other involving Ehd1, which control heading date by regulating Hd3a and RFT1 [57]. In the dh mutant, Ehd1 and Hd3a/RFT1 expression was almost completely repressed (Fig 5A-5C). According to a flowering network proposed in rice [57], we found that the expression levels of six genes (RID1/OsID1/Ehd2, OsMADS50, OsGI, Hd1, OsMADS56, and Ghd7), located upstream of Ehd1 and Hd3a, did not differ between ZH11 and the dh mutant (Fig 5D-5I). RFT1 and Hd3a proteins were produced in leaves and moved from the leaves to SAM to promote floral transition [57]. However, the expression of RFT1 and Hd3a were completely down-regulated in leaves in dh mutant (Fig 5A, 5C). Through this process, the floral meristem identity determined genes located in the SAM cannot accept the flowering signal leading to late-flowering. We also found that the expression of OsPHYC was much higher in dh mutant, compared to that in ZH11 (S7 Fig). Mutation of either phyB or phyC causes moderate early flowering under longday photoperiod in rice [71]. PhyC is thought to play an important role for flowering in barley [72]. Therefore, we believe that OsPHYC may be influencing flowering in rice. Thus, osa-miR171c regulates flowering time by promoting the expression of OsPHYC independently of the Hd1/Ehd1-Hd3a/RFT1 pathway (Fig 5), that is, osa-miR171c works in distinct way in leaves to affect the reproductive transition. These data suggested that osa-miR171c plays important roles in two independent transitions, by repressing the juvenile-adult phase change and the vegetative-reproductive stage transition. By regulating several different genetic pathways, osa-miR171c integrates the phase transitions and may be acts as a master switch regulator of phase transition.
The osa-miR171c-OsHAMs module is required for the maintenance of SAM indeterminacy Maintenance of an indeterminate state and regulation of stem cell population are fundamental aspects of the SAM function. In Arabidopsis, stem cell maintenance is regulated by WUSCHEL-CLAVATA (WUS-CLV) feedback loop genes [27]. In rice, FON1 (similar to CLV1) and FON4 (similar to CLV3) negatively regulate stem cell maintenance [30,[73][74][75], while WOX4 and OSH1 promote stem cell identity [76,77]. In dh mutant, the shoot apex maintained indeterminacy for a longer period. Three genes (FON1, FON4 and WOX4), related with stem cell maintenance, were significantly down-regulated in dh mutant, whereas OSH1 remained unchanged (Fig 6M). This suggests that the WUS-CLV feedback loop pathway was mainly affected in dh mutant. The fon4 mutant had enlarged shoot apex and inflorescence meristem, which causes an increase in culm thickness and primary rachis branch number [74,75]. dh mutant also present these phenotypes, which may result from the down-regulation of the FON4 gene. However, it is unclear what the role of OsHAMs is in the WUS-CLV feedback loop pathway. Mutation of Arabidopsis AtHAM1,2,3 genes altered the relative expression positions of WUS and CLV3 in SAM [44], and generated supernumerary cell layers between the L1 layer and the organizing centre in SAM [43]. On the other hand, the normal expression of CLV3 in WT not only relied on the basipetal transport of an L1-derived signal, but also required the acropetal transport of WUS signal [78]. In dh mutant, SAM also produced so many supernumerary meristem cell layers, blocking the transport of the L1-derived and WUS signals, and causing the alteration of FON4 expression pattern. Therefore, we hypothesise that the function of OsHAMs is required for normal OsFON4 expression, similar to Arabidopsis. These data suggested that osa-miR171c-OsHAMs was involved in the maintenance of SAM indeterminacy by affecting the WUS-CLV3 feedback loop.
Here, we also compared the expression of several genes (ASP1, APO1 and APO2) known to be related to primary branches number [59,60,79], because dh mutants showed an increase number of primary branches. However, the expression of these genes did not differ between dh mutant and ZH11 (Fig 6M). Previous studies have shown that the number of primary branches in rice is determined by the initial size of the reproductive shoot apex [61]. In dh mutant, the initial size of the reproductive shoot apex is obviously bigger than that in ZH11 (Fig 6E, 6J and  6K). This is consistent with the increased number of primary branches found in dh mutant (Fig 7A-7C).
These results suggested that maintenance of SAM indeterminacy by miR171 among Arabidopsis, barley and rice is conserved through the regulation of the WUS-CLV feedback loop [42,45].
The osa-miR171c-OsHAMs module role in the phase transition regulatory pathway Based on data from Arabidopsis and barley, miR171c has been confirmed to play a role in controlling the phase change [42][43][44][45]. Here, we proposed that the osa-miR171c-OsHAMs module regulates phase change in rice through three regulatory pathways (Fig 8). Firstly, up-regulation of osa-miR171c prolongs vegetative growth and delays the juvenile-adult phase transition by increasing the expression of miR156, similar to barley [45]. Secondly, osa-miR171c and its targets could affect the expression of florigens RFT1 and Hd3a by altering the expression of OsPHYC. Thirdly, osa-miR171c and its targets affect SAM maintenance by regulating the homologs of the WUS/CLV3 feedback loop, similar to Arabidopsis [44].