MicroRNA390-Directed TAS3 Cleavage Leads to the Production of tasiRNA-ARF3/4 During Somatic Embryogenesis in Dimocarpus longan Lour

Trans-acting short-interfering RNAs (tasiRNAs) originate from TAS3 families through microRNA (miRNA) 390-guided cleavage of primary transcripts and target auxin response factors (ARF3/-4), which are involved in the normal development of lateral roots and flowers in plants. However, their roles in embryo development are still unclear. Here, the pathway miR390-TAS3-ARF3/-4 was identified systematically for the first time during somatic embryo development in Dimocarpus longan. We identified the miR390 primary transcript and promoter. The promoter contained cis-acting elements responsive to stimuli such as light, salicylic acid, anaerobic induction, fungal elicitor, circadian control, and heat stress. The longan TAS3 transcript, containing two miR390-binding sites, was isolated; the miR390- guided cleavage site located near the 3′ end of the TAS3 transcript was verified. Eight TAS3-tasiRNAs with the 21-nucleotides phase were found among longan small RNA data, further confirming that miR390-directed TAS3 cleavage leads to the production of tasiRNA in longan. Among them, TAS3_5′D5+ and 5′D6+ tasiRNAs were highly abundant, and verified to target ARF3 and -4, implying that miR390-guided TAS3 cleavage with 21-nucleotides phase leading to the production of tasiRNA-ARF is conserved in plants. Pri-miR390 was highly expressed in friable-embryogenic callus (EC), and less expressed in incomplete compact pro-embryogenic cultures, while miR390 showed its lowest expression in EC and highest expression in torpedo-shaped embryos (TEs). DlTAS3 and DlARF4 both exhibited their lowest expressions in EC, and reached their peaks in the globular embryos stage, which were mainly inversely proportional to the expression of miR390, especially at the globular embryos to cotyledonary embryos (CEs) stages. While DlARF3 showed little variation from the EC to TEs stages, and exhibited its lowest expression in the CEs stage. There was a general lack of correlation between the expressions of DlARF3 and miR390. In addition, pri-miR390, DlTAS3, DlARF3 and -4 were up-regulated by 2,4-D in a concentration-dependent manner. They were also preferentially expressed in roots, pulp, and seeds of ‘Sijimi’ longan, implying their extended roles in the development of longan roots and fruit. This study provided insights into a possible role of miR390-tasiRNAs-ARF in plant somatic embryo development.


INTRODUCTION
MicroRNAs and tasiRNAs are distinct classes of small RNAs, which control many aspects of development in plants by guiding silencing of target RNAs via cleavage or repression mechanisms (Chapman and Carrington, 2007). MiRNAs arise from endogenous transcripts that can form local fold-back structures, whereas tasiRNAs are generated by primary TAS transcripts that are initially targeted and sliced by miR173 (TAS1 and TAS2), miR828 (TAS4) or miR390 (TAS3) (Pullman et al., 2003;Wilson et al., 2005;Axtell et al., 2006;Kikuchi et al., 2006;Montgomery et al., 2008b). The TAS3 family is conserved in plants and has two sites that are complementary to miR390 (Krasnikova et al., 2013;Xia et al., 2013), while the TAS1, TAS2, and TAS4 families are found only in Arabidopsis thaliana or close relatives, and each of them contain only a single site complementary to miR173 or miR828 (Pullman et al., 2003;Wilson et al., 2005;Axtell et al., 2006;Kikuchi et al., 2006;Montgomery et al., 2008b).
The roles of the pathway miR390-TAS3-ARF3/-4 in root and flower development are conserved and clear in plants; however, their functions in plant embryo development are unclear. For example, miR390 peaked in mature embryos , and three detected tasiRNAs generated from TAS3, which were triggered by miR390, peaked in mature embryos in larch (Larix leptolepis; Zhang et al., 2013), suggesting that the miR390-TAS3-tasiRNA pathway might play regulatory roles during CE development. In Valencia sweet orange, miR390 is involved in globular-shaped embryo formation , and overexpression of Csi-MIR390 callus lines caused lost embryogenesis capacity . During Gossypium hirsutum SE, miR390 reached the highest expression level at the EC stage, remained at moderate levels during embryo development, and one TAS3 transcript was verified as the target of miR390 (Yang et al., 2013). As mentioned above, notably, the roles of the pathway miR390-TAS3-ARF3/-4 in embryo development are somewhat confused in different plants, and their functions in plant embryos remain unclear.
Dimocarpus longan (longan), a member of the Sapindaceae family, is an exotic subtropical fruit mainly planted in Northern Abbreviations: 2,4-D, 2,4-dichlorophenoxyacetic acid; ARF, auxin response factor; EC, friable-embryogenic callus; cDNA, complementary DNA; CDS, coding DNA sequence; CEs, cotyledonary embryos; GEs, globular embryos; ICpECs, incomplete compact pro-embryogenic cultures; miRNAs, microRNAs; PPRs, pentatricopeptide repeat proteins; pre-miRNA, precursor miRNA; pri-miRNAs, primary miRNA transcripts; RLM-RACE, RNA ligase-mediated rapid amplification of cDNA ends; RT-PCR, reverse transcription-polymerase chain reaction; RT-qPCR, real-time quantitative reverse transcription PCR; SA, salicylic acid; SE, somatic embryogenesis; Tail-PCR, thermal asymmetric interlaced PCR; tasiRNAs, trans-acting short-interfering RNAs; TEs, torpedo-shaped embryos; TSSs, transcription start sites; UTR, un-translated region. Burma, and Northeast and Southern China. Longan fruit are preferably eaten fresh because of their sweet flavor and beneficial health effects, and usually contain a relatively large black or brown seed at maturity, with a large quantity of polysaccharides (Tseng et al., 2014). The seeds have also been used as bioactive ingredients in many traditional Chinese medicines to improve human health and increase the immunomodulatory capacity (Rangkadilok et al., 2005). However, the molecular mechanism of longan seed development remains unclear because of the extreme genetic heterozygosity exhibited and the difficulty of sampling the early embryos (Liu et al., 2010;Lai and Lin, 2013;Lin and Lai, 2013a,b;Lin et al., 2015). To avoid these difficulties, longan SE, which resembles zygotic embryogenesis, has previously been used widely as a model system to investigate in vitro and in vivo regulation of embryogenesis in plants Lin and Lai, 2010). Our previous work indicated that dlo-miR390 a.1 and -a * .1 accumulated during heart-and TE stages (Lin and Lai, 2013b). However, how miR390 directs the formation of tasiRNAs, and down-regulates the expression of target ARF3 and -4 during longan SE, remains unclear.
In this study, the conserved pathway miR390-TAS3-ARF3/-4 was identified in longan using an integrated strategy including RT-PCR/RACE, computational, genome-wide expression profiling and experimental validation. First, the primary miR390 and TAS3 transcripts were cloned by RT-PCR and RACE. The promoter of primary miR390 was isolated by Tail-PCR and its cis-acting elements were predicted using bioinformatic methods. The TAS3 tasiRNAs triggered by miR390 were detected from a longan small RNA database (Lin and Lai, 2013b). TasiRNA-ARF targets were identified by a modified RLM-RACE. The expressions of primary miR390, DlTAS3, DlARF3 and -4 were analyzed in D. longan 'Honghezi' SE and in 'Sijimi's tissues. Their responses to 2,4-D stimuli were also determined. These results revealed a possible role for the conserved pathway miR390-TAS3-ARF3/-4 in longan somatic embryo development.

Plant Materials and Treatments
The synchronized embryogenic cultures from D. longan 'Honghezi' used in this study were friable-EC, ICpECs, GEs, TEs, and CEs; Lai and Chen, 1997). For RT-qPCR analysis, the EC were cultured on Murashige and Skoog medium (MS; Murashige and Skoog, 1962) supplemented with 1.0 mg/L 2,4-D and 2% sucrose (pH 5.8), for 20 days, then transferred to MS liquid medium containing 2,4-D (0, 0.5, 1.0, 1.5, and 2 mg/L) in a rotary shaker at 150 rpm for 24 h, respectively. All samples were collected and stored at −80 • C for subsequent analyses.
Samples of roots (R), leaves (L), floral buds (FB), flowers (F), young fruits (YF), mature fruits (MF), pericarp (P1), pulp (P2), and seeds (S) collected from at least six rootstock plants of the same cultivar, D. longan 'Sijimi, ' were provided by the experimental fields of the Fujian Academy of Agricultural Science in Putian, and used for RNA extraction.

Isolation of the miR390 Gene from D. longan
To gain a more complete understanding of the miR390 gene and its evolution in plants, based on the previously published small RNA data (BioSample accession SAMN04120614, Bio-Project ID PRJNA297248; Lin and Lai, 2013b) and the longan EC transcriptome data (SRA050205; Lai and Lin, 2013), the longan primary miR390 gene was obtained from longan EC DNA using PCR with sense (390F) and anti-sense (390R) primers. To determine the 5 -end, a GeneRacer Kit (Invitrogen, Carlsbad, CA, USA) was used for full-length cDNAs synthesis from a mixture of total RNAs (EC, ICpEC, GE, and HE), following the manufacturer's instructions. 5 RACE PCR was carried out using two nested PCR reactions and a combination of GeneRacer forward primers (5 Primer and 5 Nested Primer) and specific reverse primers (390-5RACE1 and 390-5RACE2). To obtain the gene regulatory sequence for the longan miR390 gene, genome walking was performed to obtain flanking genomic DNA using TAIL-PCR (Takara, Japan). Its promoter was cloned by three nested PCR amplifications from a longan EC DNA template with the same forward primer, AP2, and specific reverse primers (390-SP1-SP5). The list of primers for PCR amplification is shown in Table 1.

MiR390-Targeted TAS3 Gene Cloning
In Arabidopsis, miR390 has been shown to target TAS3 (Montgomery et al., 2008a), but there is no evidence for the existence of TAS3 in D. longan. To verify its existence and determine the structure of the TAS3 transcript, the conserved region of the TAS3 transcript was first obtained using RT-PCR with sense (TAS3-F) and anti-sense (TAS3-R) primers designed from known TAS3 sequences. To determine the 5 -and 3 -ends, 5 and 3 RACE PCR were carried out by nested PCR reactions using a combination of GeneRacer primers and the specific primer pairs TAS3-5P1/5P2 and TAS3-3P1/TAS3-3P2. The list of primers for PCR amplification is shown in Table 1.

Bioinformatics Analysis
The miR390 and TAS3 genes were compared against the miRBase server (Release 21: June 2014) and NCBI using BLAST. Multiple alignment analysis and secondary structure analysis were performed using DNAMAN ver. 6.0 (Lynnon Biosoft, Pointe-Claire, QC, Canada). Cis-acting elements analysis was carried out using the PlantCARE database (Lescot et al., 2002). For phylogenetic analysis, MEGA 5.02 (Kumar et al., 2004) was performed according to the neighbor-joining method (Saitou and Nei, 1987) with 1000 bootstrap replicates. The psRNATarget analysis (Dai and Zhao, 2011) was carried out for miRNA and tasiRNA target prediction.
Cleaved Target mRNA Identification with 5 RLM-RACE TasiRNA produced from the TAS3 transcript has been shown to target ARF3 and -4 (Marin et al., 2010). To discover the target transcripts of TAS3 tasiRNA in longan, a modified 5 RLM-RACE experiment was set up. Here, the transcripts of ARF3 (GenBank accession No. KJ200347.1) and -4 (GenBank accession No. KJ200347.1) were used to designed RACE primers. The cDNA was synthesized using a GeneRacer Kit using RNA extracted from friable-EC, ICpECs, GEs, and HEs of longan embryogenic cultures. PCR amplification of cDNA fragments using 5 RACE outer primers and gene-specific reverse primers was performed. PCR fragments were cloned and sequenced to identify the 5 -end of the amplified target genes.

RNA Extraction and RT-qPCR
For RT-qPCR analysis, total RNAs were extracted from the above-mentioned samples using a TRIzol Reagent kit (Life Technologies, Grand Island, NY, USA). cDNA for mRNA quantification was subsequently synthesized using PrimeScript TM RT Reagent Kit Perfect Real Time (Takara Code, DR037A, Japan). For RT-qPCR amplification, cDNA was diluted 15 times before use. RT-qPCR was performed using LightCycler 480 equipment (Roche Applied Science, Switzerland) using the manufacturer's instructions. Three reference genes DlFSD1a, EF-1a, and eIF-4a  were used to normalize our signal. All reactions were performed in triplicate. Statistical analysis was performed using SPSS 19. The gene names and primer sequences are provided in Table 2.

Cloning and Analysis of the D. longan miR390 Gene Sequence
To determine miR390s structure and the length of the gene in the D. longan genome, the full-length cDNA sequence of the primary miR390 transcript was obtained by RT-PCR and 5 -RACE. According to previous studies, the longan mature miR390 (Lin and Lai, 2013b) was predicted to derive from the transcript Contig648826, which was retrieved from longan EC transcriptome data (SRA050205; Lai and Lin, 2013). To verify the existence of the transcript, it was cloned from longan EC using RT-PCR and 5 RACE. Sequence analysis showed that the transcript was 845 bp in length and had no introns (GenBank accession No. KJ372216), and the first nucleotide (transcription initiation site, TTS) of the full-length cDNA was A ( Figure 1A); the mature sequences dlo-miR390a-5p and dlo-miR390-3p were found to exactly match the 5 arms of the transcript ( Figure 1A). A sequence of 117 bp extracted from between the positions of dlo-miR390-5p and -390-3p in the transcript was further used to predict the secondary structure using DNAMan6.0 software. The analysis showed that the sequence could form the classic stemloop structure and the minimum free energy of the structure was −44.7 kcal/mole ( Figure 1B). The sequence has similarities with existing precursor miR390s in Malus domestica, G. hirsutum, Glycine max, and Cucumis melo using BLAST analysis, and is phylogenetically closely related to pre-miR390 transcripts of Arabidopsis lyrata, Populus trichocarpa, and Vitis vinifera, but was distant from pre-miR390s of C. melo, M. domestica, and Prunus persica ( Figure 1C).

Isolation of the Longan miR390 Promoter and Analysis of its Structural Features
Promoter analysis is an essential step in the identification of regulatory networks. Here, the partial 5 -flanking region of miR390 (927 bp) upstream of the TTS was isolated from longan friable-EC genomic DNA (GenBank accession No. KJ372219), and a promoter motif search was performed by PlantCARE (Lescot et al., 2002). The analysis indicated that the classical promoter elements, such as a TATA box and a number of CAAT boxes, were located in the expected position of the miR390 promoter (i.e., within −30 to −90 nucleotides upstream of the TTS). Potential cis-regulatory elements associated with light, including an ACE motif, an AE-box, an AT1-motif, three Box 4 sites, a CATT-motif, an Sp1 motif and a chs-CMA2b motif, and the cis-acting elements related to stress-related responses containing a HSE motif (involved in heat stress responses) and TC-rich repeats (involved in defense and stress responses), were also detected in the promoter region. In addition, a TCA-element involved in SA responsiveness; an ARE cis-acting regulatory element that is essential for anaerobic induction; two AT-rich sequence elements for maximal elicitor-mediated activation (two copies); a Box-W1 element involved in fungal elicitor responsiveness; a Skn-1 motif and a GCN4 motif, which are involved in endosperm expression; a circadian element involved in circadian control; a cis-acting element that confers high transcription levels (5 UTR Py-rich stretch); and four elements of unknown function were also present in the promoter ( Figure 1A). These results indicated that the miR390 gene could be regulated by environmental (light and stress) factors and hormones (SA).
Cloning and Analysis of miR390-Targeted TAS3 Gene in Longan In this study, based on the sequences of TAS3 in other plants, the full-length cDNA of TAS3 was obtained through RT-PCR and 5 /3 RACE using longan EC cDNA as the template. Sequence analysis showed that the TAS3 transcript, named DlTAS3, comprised 579 bp, with a 23-bp poly A tail (GenBank accession No. KJ372220). Sequence alignment analysis indicated that DlTAS3 has high similarities with Arabidopsis TAS3 (AT3G17185.1). The mature miRNA sequence, dlo-miR390a-5p, was predicted to bind exactly to the DlTAS3 transcript at two sites, located at positions 292-312 and 472-491 bp, respectively (Figure 2A). The phylogenetic analysis ( Figure 2B) suggested that DlTAS3 in D. longan is closely related to TAS3 in Citrus sinensis, Lotus japonicus, and Solanum lycopersicum, but was distant from TAS3 in A. thaliana and Oryza barthii.
MiR390 Guides In-phase Processing of DlTAS3 Transcripts TAS3 tasiRNAs originate from sequences between the two miR390 target sites (Kikuchi et al., 2006;Montgomery et al., 2008a). Here, the sequence of DlTAS3 between the two binding sites was 158 bp, and the dlo-miR390-guided cleavage site was predicted to be located near the 3 end of the DlTAS3 transcript, while dlo-miR390 interacts in a non-cleavage mode with a second site near the 5 end, which was found to have a conserved mismatch at the center of the binding region (Figure 2A). MicroRNAs direct tasiRNA biogenesis in plants, and miR390guided cleavage was shown to set the 21-nucleotides phase for tasiRNA precursor processing (Kikuchi et al., 2006). Here, according to the dlo-miR390-guided cleavage site, eight potential tasiRNAs with the 21-nucleotides phase were predicted to be produced from miR390-guided DlTAS3 cleavage (Figure 2A; Table 3). In-phase, the 21-nucleotides positions on the 5 side of the miR390 cleavage site were named 5 D1+, 5 D2+, and so on. Among these tasiRNAs, the TAS3_ 5 D5+ and 5 D6+ have a high similarity, with only four nucleotides differences at the 3 end of the tasiRNAs (Figure 2C), and they are also very similar to the 5 D7+ and 5 D8+ of the TAS3 in A. thaliana (Kikuchi et al., 2006), G. max (Hu et al., 2013), M. domestica, C. sinensis, and P. trichocarpa (Figure 3). These results suggested that miR390directed TAS3 cleavage leading to the production of tasiRNA is conserved in plants.

Longan tasiRNA Abundance Analysis and their Target Prediction
To verify the existence of DlTAS3 tasiRNAs in longan, the tasiRNAs predicted in Section "MiR390 Guides In-phase  Processing of DlTAS3 Transcripts" were searched for in a longan small RNA dataset using local BLAST (Lin and Lai, 2013b). Among the eight tasiRNAs, the reads of TAS3_5 D5+ and 5 D6+ siRNAs were more abundant, with 284 and 478 reads. This result was similar to previous findings in which the reads of TAS3 5 D7+/5 D8+ tasiRNAs were also high in G. max (Hu et al., 2013). The signatures of 5 D2-and 5 D4+ siRNAs were 14 and 44 reads, respectively, while the expressions of other tasiRNAs were very low, and the 5 D3+ siRNA was not detected in longan SE (Figure 2A). These results further proved that dlo-miR390 was sufficient to trigger secondary tasiRNA biogenesis, and miR390-guided TAS3 cleavage with the 21-nucleotides phase was conserved in plants.
These results suggested that tasiRNAs are broadly involved in plant development by guiding the cleavage of different targets, similar to plant miRNAs.

Validation of tasiRNA-Guided Cleavage of Target Gene DlARF3 and -4 mRNAs in Longan
Four ARF genes have been validated as the target of TAS3 5_D7+/5_D8+ siRNAs in Arabidopsis (Kumria et al., 2003). In our study, only DlARF3 and -4 were predicted to be targets of TAS3_5 D5+/5 D6+ tasiRNAs, which are similar to TAS3 5_D7+/5_D8+ siRNAs in Arabidopsis. Therefore, to further verify the cleavage of DlARF3 and -4 by the TAS3_5 D5+/5 D6+ tasiRNAs, a modified RLM-RACE analysis was performed, which resulted in the detection of fragments of DlARF3 and -4 mRNAs from longan embryogenic tissues (Figure 4). Fragment sequence analysis showed that the cleavage sites in DlARF3 and -4 mRNAs are located at positions corresponding to the amino acid sequences KVLQGQE; in addition, DlARF3 was cleaved between the 10th and 11th nucleotides complementary to miR390, while DlARF4 was cleaved between the 9th and 10th nucleotides of miR390 (Figure 4). These results clearly indicated that TAS3_5 D5+/5 D6+ tasiRNAs cleaved the DlARF3 and -4 mRNAs during longan SE, which further proved that the tasiRNA-directed ARF mRNA cleavage was conserved in plants.

Expression Profiling of miR390-DlTAS3-DlARF3/-4 during Longan SE
To systematically analyze the function of miR390-TAS3-ARF3/-4 during longan SE, expression profiling was performed using five embryogenic cultures. RT-qPCR analysis showed that the pri-miR390, miR390, DlTAS3, DlARF3 and -4 exhibited different temporal and spatial expressions ( Figure 5A). Pri-miR390 was highly expressed in EC, and less expressed in ICpECs, while it had a stable expression level with no fluctuations from GEs to CEs, suggesting that the accumulated pri-miR390 in the EC stage may be necessary for the maintenance of embryonic callus in an undifferentiated state in longan. miR390 showed its lowest expression in EC and highest expression in TEs, with a general lack of correlation of pri-miRNAs at the transcription level. It is worth noting that DlTAS3 and -4 showed similar expression patterns. They both exhibited their lowest expressions in EC, and reached their peaks in the GEs stage, which were mainly inversely proportional to the expression of miR390, especially at the GEs to CEs stages. DlARF3 showed little variation from the EC to TEs stages, and exhibited its lowest expression in the CEs stage. There was a general lack of correlation between the expressions of DlARF3 and TAS3. We concluded that miR390 down-regulation of TAS3 leading to the production of tasiRNAs via cleavage of DlARF3/-4 mRNAs during somatic embryo development in longan.

MiR390-DlTAS3-DlARF3/-4 Expressions in Response to Auxin in Longan EC
A previous study showed that miR390 and tasiRNA ARF are regulated by auxin concentration (IAA; Yoon et al., 2010). Embryogenesis is inhibited by exogenously supplied 2,4-D (>10 −9 M) or indoleacetic acid (IAA; >10 −10 M) in plants    (Ruduś et al., 2002). Here, to examine how the pathway of miR390-TAS3-ARF3/-4 expression is regulated by 2,4-D, their expression levels were monitored in longan EC exposed to different 2,4-D concentrations ( Figure 5B). The results showed that their levels all accumulated at 1.0 mg/L 2,4-D compared with auxin-free medium under the same conditions, but decreased at higher concentrations, except for DlTAS3. In contrast to miR390-ARF3/-4, DlTAS3 was expressed at its lowest level at 0.5 mg/L 2,4-D compared with auxin-free, increased at 1.0 mg/L, but decreased at 1.5 mg/L, and reached its peak at 2.0 mg/L ( Figure 5B). This result demonstrated that the presence of auxin in the medium controlled the level of miR390-TAS3-ARF3/-4, which is similar to the results of a previous study (Yoon et al., 2010).

MiR390-DlTAS3-DlARF3/-4 Expressions in 'Sijimi' Longan Tissue Types
To more precisely pinpoint the locations of miR390-TAS3-ARF3/-4 expressions, the tissues of the cultivar 'Sijimi' longan at nine stages of development, roots (R), leaves (L), floral buds (FB), flowers (F), young fruits (YF), mature fruits (MF), pericarp (P1), pulp (P2), and seeds (S) were used to assay RNA accumulation. The results showed that primary transcripts of miR390, DlTAS3, DlARF3 and -4 exhibited more or less expression in vegetative and reproductive organs, and were preferentially expressed in roots (Figure 5C), suggesting they might affect 'Sijimi' root development. In addition, DlTAS3, DlARF3 and -4 were also expressed at relatively high levels in pulp (P2) and seeds (S), while the primary miR390 transcript expression was an exception, implying that TAS3-ARF3/-4 may also be involved in the fruit development of longan. Furthermore, the expressions of pri-miR390 in young fruits (YF), mature fruits (MF), and pericarp (P1), DlTAS3 in pericarp (P1), and DlARF4 in leaves (L) and flower buds (FB) in mature fruits were low. These data revealed extensive regulation roles of miR390-TAS3-ARF3/-4 signal in the developmental stages of vegetative and reproductive growth in 'Sijimi' longan.

DISCUSSION
MiRNAs and tasiRNAs are small RNAs of ∼21 nucleotides in length that have various roles as negative regulators of mRNA targets in plant physiology and development (Chuck et al., 2009). TasiRNAs originate from TAS3 families through miR390-guided initiation-cleavage of primary transcripts and target ARF2/-3/-4, which are involved in the normal development of lateral roots (Marin et al., 2010;Yoon et al., 2010) and flowers in plants (Matsui et al., 2014); however, their roles in embryo development are still unclear. Here, we cloned, identified and examined the expressions of primary miR390 and its promoter, DlTAS3, tasiRNA, and its targets DlARF3/-4 in longan. This study was the first systematic investigation of the conserved pathway miR390-TAS3-ARF3/-4 in longan SE, thereby providing insights into a possible role in plant somatic embryo development.

Characteristics of Longan Primary miR390 and its Promoter
MicroRNAs and tasiRNAs are both produced from larger capped and polyadenylated precursor non-coding transcripts of a few hundred base pairs in size (Krasnikova et al., 2009). However, the precursors of miRNAs or tasiRNAs are little studied and cloned, and have only been reported in some genome-sequenced or model plants, such as Arabidopsis (Xie et al., 2005) and G. max (Han et al., 2014). In non-genome-sequenced or nonmodel plants, it is hard to clone the precursors of small RNA. Here, based on the transcriptome data and RT-PCR/RACE, we successfully cloned the primary transcripts of miR390 and TAS3, and identified their TTS without genome data. These results laid the foundation for further experiments to identify the functions of longan miR390 and TAS3.
MiR390 has two genomic loci, on chromosome 2 (MIR390a) and 5 (MIR390b) in Arabidopsis (Montgomery et al., 2008a); the mature miR390 mostly originates from the MIR390a and not from the MIR390b locus in roots (Marin et al., 2010). MIR390a exhibited higher processing accuracy and efficiency of miR390 than the MIR390b (Cuperus et al., 2010). In longan, only one primary miR390 was cloned and its precursor was closest to pre-miR390s identified from C. sinensis and O. sativa, but closer to pre-miR390a and close to pre-miR390b in Arabidopsis, suggesting that the longan pre-miR390 cloned in our study perhaps has a similar function to the Arabidopsis MIR390a. Previously, miR390, which has been reported to respond to UV-B in Populus tremula (Jia et al., 2009), Cd (Ding et al., 2011) and As (Srivastava et al., 2013) stress in rice, was down-regulated in response to Al (Chen et al., 2012) and Hg-toxicity (Zhou et al., 2012) in Medicago truncatula, and up-regulated under drought stress in Vigna unguiculata (Barrera-Figueroa et al., 2011) and Brachypodium distachyon (Budak and Akpinar, 2011). In our study, the cis-acting elements of miR390 promoter related to  TC-rich repeats (defense and stress responses), and a Box-W1 element, which is a binding site for the WRKY transcriptional regulators that control pathogen defense, wound response, and senescence (Eulgem et al., 2000), were also identified, indicating the potential role of miR390 in stress responses in longan. In addition, miR390 was differentially regulated under heat conditions in switchgrass (Hivrale et al., 2015) and Brassica rapa (Yu et al., 2012). In longan, a cis-acting element related to heat stress responses (HSE motif) was identified, which was also present in miR390 promoter in strawberry (Li et al., 2014), indicating a potential role for heat in the control of miR390 accumulation in plants.
Previous studies showed that miR390 and tasiRNA ARF were positively regulated by auxin (IAA) concentration, but were not sensitive to abscisic acid, gibberellins, and cytokinin (Yoon et al., 2010;Li et al., 2013). Here, although no auxin-related cis-elements were identified in the longan miR390 promoter, pri-miR390 was up-regulated by 2,4-D treatment, suggesting that
ARF elements may be present outside of the 927 bp promoter region; in addition, DlTAS3, DlARF3 and -4 expressions were also up-regulated by 2,4-D in a concentration-dependent manner, implying that 2,4-D may influence the activity of the tasiRNA-ARF pathway through its concentration-dependent regulation of miR390 expression. However, the negative regulatory relationship was not observed among miR390, DlTAS3, DlARF3 and -4 under 2,4-D treatment, suggesting that the molecular mechanism of these genes response to 2,4-D is more complicated than expected, this needs further experimental verify. It is worth noting that the primary miR390, whose promoter contains a SA-responsive element (TCA element), was not sensitive to SA treatment (50, 75, and 100 mg/L) compared with hormone free (data not shown). A previous study showed that the mature miR390s response to auxin was significantly broader than that of the miR390 promoter GUS expression (Yoon et al., 2010), leading to a possible interpretation that SA affects the expression of mature miR390 but does not affect the expression of primary miR390 in longan.

TAS3 and tasiRNA Expressed in Longan SE
MiR390 targets TAS3. Here, a 579-bp cDNA was cloned from longan EC cDNA, which was homologous to TAS3 in plants. The data presented here provide further evidence for the existence of TAS3 in D. longan. Three TAS3 loci have been identified in Arabidopsis: TAS3a (At3g17185), TAS3b (At5g49615), and TAS3c (At5g57735; Howell et al., 2007). In our study, the DlTAS3 is phylogenetically closer to AtTAS3a than AtTAS3b/c, suggesting that DlTAS3 may have a similar function to AtTAS3a. TAS3 tasiRNAs arise from the TAS3 family, all members of which have conserved two miR390-guided target sites; for each member, the 3 miR390 target site, but not the 5 target site, was cleaved (Howell et al., 2007). Recently, however, the 5 miR390 binding site in TAS transcripts was demonstrated to undergo cleavage for the initiation of processing in dicots (Krasnikova et al., 2009). In this study, two miR390 target sites also existed in DlTAS3, and the cleaved site was predicted to be located near the 3 end of the DlTAS3. Moreover, the longan small RNA data analysis showed that tasiRNAs of 21 nucleotides formed by miR390guided cleavage on the 3 side of DlTAS3 were also found in longan SE, which further proved that the roles of miR390-cleaved TAS3 in the production of tasiRNAs are also functional and conserved in longan.

MiR390, DlTAS3, DlARF3 and -4 Define an Auto-regulatory Network in Longan Somatic Embryo Development
We established that the pathway miR390-TAS3-ARF3/-4 operates in longan, like other plants, but its roles during plant SE remained unclear. In our study, primary miR390 reached a peak in EC, which is consistent with the mature miR390 expression in G. hirsutum (Yang et al., 2013). The mature miR390 showed its lowest expression in EC and highest expression in TEs in longan, and reached a peak in globular-shaped embryo in Valencia sweet orange . This implied that miR390 has different expression patterns during SE, suggesting that miR390 has different roles during SE in different plants. In addition, there was a general lack of correlation between the expression of pri-miRNAs and the corresponding mature miR390s in longan, which is a common phenomenon in eukaryotes (Lee et al., 2008), suggesting that unknown mechanisms that control processing play a critical role in regulating the mature miRNA expression.
Previous studies showed that the miR390-TAS3 tasiRNA pathway might play regulatory roles in the development of mature embryos in larch . Here, DlTAS3 exhibited the lowest expressions in EC, and reached its peaks in the GEs stage, which was mostly the reverse of the expression of miR390, further confirming that miR390 cleaved DlTAS3, leading to the production of tasiRNAs in longan. A previous study observed that the expression pattern of TAS3_5 D7+ was identical to that of miR390 during SE in B. napus , suggesting that the expression of TAS3 tasiRNA was induced by miR390. In addition, TAS3 mRNA reached its peak at EC after 1-5 days of sub-culture in larch , which was distinct from that of longan SE, implying that the roles of the miR390-TAS3 signal are not conserved between angiosperm and gymnosperm species. Moreover, parallel expression of DlTAS3 and DlARF4 during longan SE was observed, while DlARF3 showed little variation among the tissues. Obviously, the expression levels between DlTAS3 and DlARF3 were largely not correlated at the transcript levels, which was also found between TAS3a-5 D6+ and its target ARFs in Triticum aestivum (Tang et al., 2012). Taken together, we propose that the pathway of miR390-TAS3-tasiRNA-ARF3/-4 operates in the development of longan embryo. MiR390-TAS3 tasiRNAs and ARFs define an autoregulatory network quantitatively by regulating lateral root growth (Marin et al., 2010). Here, the primary transcripts of miR390, DlTAS3, DlARF3 and -4 were all preferentially expressed in 'Sijimi' longan roots; therefore, we propose that their roles in roots are also conserved in longan. Besides, DlTAS3, DlARF3 and -4 were also expressed at relatively high levels in the pulp and seeds, implying extended roles in the development of longan fruit.
In summary, this study was the first to identify the longan miR390-TAS3 tasiRNA-ARF3/-4 axis, using a RACE strategy and the expression profiles of small RNA and their targets in longan SE. We demonstrated that MiR390, DlTAS3, DlARF3 and -4 define an autoregulatory network duing longan somatic embryo development.

AUTHOR CONTRIBUTIONS
YL participated in the study design, carried out the experimental work and wrote the manuscript. LL carried out the experimental work. ZL conceived of the study, and participated in its design and coordination and helped to draft the manuscript. LL, RL, WL, ZZ, YC, and XX prepared the materials. All authors read and approved the final version of the manuscript. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.