Red Chinese sand pear (cultivar “Meirensu”) fruits were obtained from a commercial orchard in Kunming, Yunnan, China. Ten mature trees that were similar in size and uniformly exposed to sunlight were selected. We bagged 80 fruitlets (the vertical diameter of fruit was around 2 cm) per tree 40 days after full bloom with double-layered yellow–black paper bags [Kobayashi (Qingdao) Co., Ltd., China]. Half of the fruits were re-exposed to sunlight after the bags were removed 10 days before harvest (the total bagging period lasted for about 3 months). The fruits that remained bagged served as control samples. Fruits were harvested 0, 1, 3, and 6 days after bag removal (DABR), stored on dry ice, and transported to the lab as quickly as possible. The skin from 10 individual fruits were pooled, instantly frozen in liquid nitrogen, and stored at −80°C for assays.

Total RNA from re-exposed and control fruit peels collected at 0, 1, 3, and 6 DABR was isolated using a CTAB method (Qian et al., 2014a). After ligation of 3′- and 5′- adapters, and RT-PCR amplification, small RNA (sRNA) library was size fractioned by PAGE gel. A mixed sRNA library was constructed and sequenced using the Illumina Genome Analyzer IIx (Illumina, Inc., Santa Clara, CA, USA). After filtering and adapter cutting, the clean tags were used by BLASTall to search the GenBank and Rfam 10.0 databases (Kozomara and Griffiths-Jones, 2014) to detect rRNAs, scRNAs, snoRNAs, snRNAs, and tRNAs, which were subsequently removed. The remaining sRNA tags were aligned with mRNA sequences to identify any degraded mRNA fragments, which were eliminated (http://peargenome.njau.edu.cn/default.asp?d=1&m=1) (Wu et al., 2013). The potential miRNAs were then mapped to the pear genome sequence (http://peargenome.njau.edu.cn:8004/default.asp?d=4&m=2) (Wu et al., 2013) using SOAP 2.20 (http://soap.genomics.org.cn/soapsplice.html), and their distribution on the genome was analyzed. Only sRNA tags that formed good stem-loop structures and had a miRNA/miRNA^* pair were considered as potential miRNAs. The secondary structure was predicted by RNAfold, and novel miRNAs were identified by MIREAP. The criteria used to identify candidate miRNAs were as previously described (Niu et al., 2013). Analyses of the high-throughput sequencing profiles were based on the number of reads. The miRNA expression levels were converted to transcripts per million normalized values as follows: normalized expression = actual miRNA count/total number of clean reads × 1,000,000. Raw data from sRNA sequencing was submitted to Sequence Read Archive (SRA ID: SRR5074534).

Equal amounts of RNA from seven independent pear fruit peel libraries (re-exposed and control fruit peels collected at 0, 1, 3, and 6 DABR) were pooled to construct a degradome library. After 5′ adapter contaminants were eliminated and genome mapping was completed as described for the sRNA data, the degradome sequencing data were analyzed using the CleaveLand (version 3.0) (Addo-Quaye et al., 2009) program. The alignment score threshold was set to 4.5 for conserved and moderately conserved miRNAs and to 5 for novel and candidate miRNAs (Xia et al., 2012). The apple consensus gene set from AppleGFDB and the annotation information for miRNA target genes were retrieved from the Genome Database for Rosaceae (Zhang et al., 2013). Degradome data were normalized to transcripts per million values. Based on the number of degradome sequences and their abundance values, the cleavage sites of miRNA targets were classified into 5 categories as previous study (Xia et al., 2012): Category 0: > 1 raw read at the position, read abundance at position is equal to the maximum read abundance on the transcript, and there is only one maximum on the transcript. Category 1: > 1 raw read at the position, read abundance at position is equal to the maximum read abundance on the transcript, and there is more than one maximum position on the transcript. Category 2: > 1 raw read at the position, read abundance at position is less than the maximum but higher than the median read abundance for the transcript. Category 3: > 1 raw read at the position, read abundance at position is equal to or less than the median read abundance for the transcript. Category 4: only 1 raw read at the position. Raw data from degradome sequencing was submitted to Sequence Read Archive (SRA ID: SRR5074535).

We completed 5′-rapid amplification of cDNA ends (RACE) using the SMARTer RACE cDNA Amplification Kit (Clontech, Palo Alto, CA, USA) to verify the miR156-mediated cleavage events identified during degradome sequencing. A 2-μg sample of mixed total RNA isolated from seven independent pear fruit peels (re-exposed and control fruit peels collected at 0, 1, 3, and 6 DABR) was ligated with 5′ RNA adapters at room temperature. All polymerase chain reaction (PCR) products were inserted into the pMD18-T vector (Takara, Dalian, China) and sequenced. Specific primers were designed for nested PCR (Supplementary Table S1 in Supplementary File 1).

An additional analysis involving the Gene Ontology (GO) databases was completed to more thoroughly characterize the functions of miRNA targets. Blast2GO was used to store and manage information from the GO databases (https://www.blast2go.com). All targets were identified using BLASTx searches against the GO protein database. The GO analyses were completed using a query search of the GO databases.

The hidden Markov model (HMM) profile of the SBP domain (Accession no. PF03110) was downloaded from the Pfam database (http://www.sanger.ac.uk). This domain was then used as a query for BLASTP searches of the GenBank non-redundant protein database and the Pear Genome Database (http://peargenome.njau.edu.cn:8004/default.asp?d=4&m=2) (Wu et al., 2013). All hits with an E < 0.01 were collected. The non-redundant protein sequences encoded by putative PpSPL genes were manually checked for the presence of the SBP domain, and genes without a complete SBP domain were eliminated. All putative PpSPL genes were confirmed again by cloning and sequencing, using a mixed cDNA template obtained from different tissues of “Meirensu” plants, including young fruits, young leaves, petals, young stems, and the skin of mature fruit. Primer details are provided in Supplementary Table S1 in Supplementary File 1. The PpSPL genes were named according to their gene ID order in the pear genome.

Multiple sequences were aligned using DNAMAN software (version 5.2). The sequence logo was obtained using the online WebLogo platform (http://weblogo.berkeley.edu). Phylogenetic trees were constructed using the neighbor-joining method of the MEGA 4.0 program (Tamura et al., 2007), and the bootstrap test was replicated 1,000 times. The miR156 target sites were identified using sequence alignments and manual analyses.

Total anthocyanin content was measured using a pH differential method and was calculated as mg cyanidin-3-galactoside per 100 g fresh tissue (mg Cy-3-gal/100 g FW) (Dussi et al., 1995). Fruit tissue (1 g) was mixed with methanol and 0.01% HCl, and then centrifuged at 18,514 g for 20 min at 4°C. The absorbance (at 510 nm and 700 nm) of a 100-μL extract in buffers (pH 1.0 and 4.5) was measured using a DU800 spectrophotometer (Beckman Coulter, Fullerton, CA, USA). Total anthocyanin content was calculated as follows: A = [(A₅₁₀− A₇₀₀) pH_1.0− (A₅₁₀− A₇₀₀) pH_4.5], with a cyanidin-3-galactoside molar extinction coefficient of 3.02 × 10⁴.

Total RNA was extracted using a modified CTAB method (Qian et al., 2014a). The total RNA sample was treated with DNase I to remove any contaminating genomic DNA, and then quantified. First-strand cDNA was synthesized from 4 μg DNA-free RNA using a Revert Aid™ First-Strand cDNA Synthesis Kit (Fermentas, Glen Burnie, MD, USA). A 2-μL sample of 10-fold diluted cDNA was used as the template for gene cloning and quantitative real-time polymerase chain reaction (qPCR) analyses.

The qPCR mixture contained 10.0 μL SYBR Premix Ex Taq™ (Takara, Ohtsu, Japan), 0.4 μL each primer (10 μM), 2 μL cDNA, and 7.2 μL RNase-free water in a total volume of 20 μL. The PCR was completed using a LightCycler 1.5 instrument (Roche, Germany) with the following program: 95°C for 30 s; 40 cycles of 95°C for 5 s and 60°C for 20 s. A template-free control was included for each primer pair. The qPCR primers (Supplementary Tables S2, S3 in Supplementary File 1) were designed using Primer 3 software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). All qPCR data were normalized using the threshold cycle value for the Pyrus species actin gene (PyActin, JN684184). Each sample was analyzed three times.

Total RNA was extracted using the pBiozol Total RNA Extraction Reagent (BioFlux, China). According to the instructions provided for the miRNA cDNA Synthesis Kit (Takara), 1 μg total RNA was polyadenylated by poly(A) polymerase. The poly(A)-tailed total RNA was reverse-transcribed by PrimeScript® RTase with a universal adapter primer (containing oligo-dT). The qPCR analysis was completed using SYBR® Premix Ex Taq II (Perfect Real Time) (Takara) and a LightCycler 1.5 instrument. Each sample was analyzed three times, and 5S rRNA was used as the internal control gene (Wu et al., 2014; Niu et al., 2016). The qPCR primers are listed in Supplementary Table S4 in Supplementary File 1.

A yeast two-hybrid assay was conducted using the Matchmaker™ Gold Yeast Two-Hybrid System (Clontech). We used the pGADT7 vector, which contains the GAL4 activation domain, and the pGBKT7 vector, which carries the GAL4 DNA binding domain. The gene open reading frames were inserted into the multiple cloning sites of pGBKT7 and pGADT7 using the In-Fusion® HD Cloning Kit (Clontech) to obtain the gene-BD and gene-AD plasmids, respectively. Details of the primers used during the yeast two-hybrid assay to obtain the open reading frames are listed in Supplementary Table S5 in Supplementary File 1. The gene-BD plasmid was inserted alone or together with the gene-AD plasmid into the Y2HGold yeast strain according to a polyethylene glycol/lithium acetate method. The autoactivation of the colonies transformed with the gene-BD plasmid was tested on synthetically defined (SD) medium lacking tryptophan, but supplemented with 5-bromo-4-chloro-3-indolyl α-D-galactopyranoside (X-α-Gal) and aureobasidin A (AbA). The co-transformed colonies were selected on SD medium lacking leucine and tryptophan (i.e., double dropout medium), and screened for growth on quadruple dropout SD medium (i.e., lacking adenine, histidine, leucine, and tryptophan) supplemented with X-α-Gal and AbA to verify positive interactions.

After running ANOVA, LSDs (α = 0.05) were calculated for mean separations of anthocyanin concentration using the Data Processing System (DPS, version 3.01, Zhejiang University, Hangzhou, China). Differences of gene expression were statistically evaluated by the analysis of variance and Tukey's test using SPSS 19.0 (SPSS, Chicago, IL, USA). Probability values of <0.05 were considered statistically significant.

A total of 13,954,538 raw reads were acquired through high-throughput sequencing of sRNA from pear fruit peels of bagged plants. Corrupted adapter sequences (e.g., missing 3′ adapter), sequences shorter than 17 bases or 25 bases after removing the 3′ adapter, and junk reads were eliminated. The remaining 9,915,560 clean reads were used for predicting miRNAs (Supplementary Table S6 in Supplementary File 1). A total of 167 conserved, 34 moderately conserved, and 21 novel miRNAs were identified in the sRNA library (Supplementary Tables S7, S8 in Supplementary File 1). Among the detected miRNAs, 26 miR156 members were identified, including some that were highly expressed in fruit peels such as miR156c (424 reads). Some members were detected at very low levels, including miR156sd (1 read) and miR156bd-3p (1 read). In addition, all the novel miRNAs were highly expressed in fruit peel, with the raw reads from 125 to 1,866 (Supplementary Tables S8 in Supplementary File 1).

For the degradome sequence analysis, ~17.71 million raw reads were obtained, and 9,052,407 unique reads were ultimately generated. These reads were subsequently screened and analyzed with CleaveLand 2.0 (Addo-Quaye et al., 2009). A total of 130 targets with cleavage sites grouped into five categories (i.e., 0–4) were identified in fruit peels (Supplementary Table S9 in Supplementary File 1), with 86 targets having GO numbers (Supplementary Table S10 in Supplementary File 1). Many of the identified targets for the conserved pear miRNAs were transcription factors, such as the auxin response factor targeted by miR160, ethylene-responsive transcription factor targeted by miR172, NAC domain protein targeted by miR164, and a GRAS family transcription factor targeted by miR171 (Supplementary Table S10 in Supplementary File 1). Among 21 novel miRNAs, miRC9 and miRC10 targeted 4 (Pbr008410.2, Pbr008785.1, Pbr037485.1, and Pbr019133.1) and 2 (Pbr001375.1 and Pbr002864.1) genes, respectively, and two of these target genes had GO numbers, with Pbr008410.2 and Pbr001375.1 encoding a homeobox-leucine zipper protein and glutamate dehydrogenase, respectively (Supplementary Tables S9, S10 in Supplementary File 1). Notably, four pear SPL genes were targeted by miR156. Pbr019232.1, which is homologous to A. thaliana SPL9, was targeted by miR156ba. Pbr020183.1 (SPL6 homolog), Pbr037448.1, and Pbr038099.1 (SPL13 homolog) were targeted by miR156a/g/s/sa (Supplementary Table S10 in Supplementary File 1).

Due to the results from sRNA and degradome sequencing, miR156a and miR156ba were chosen as two examples of miR156s, and the genomic region 1,500 bp upstream of pre-miR156a and pre-miR156ba was considered the promoter region and was analyzed using the PlantCARE online program (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/). We identified 16 light-responsive elements in the pre-miR156a promoter (i.e., two Box 4 elements, six G-boxes, one GA-motif, one GAG-motif, two GT1-motifs, one I-box, one MRE, and two sp1 elements) (Table 1). The pre-miR156ba promoter contained one AE-box, two Box 4 elements, seven G-boxes, one GAG-motif, three GT1-motifs, one TCCC, and two sp1 elements (Table 1).

We isolated 19 pear PpSPL genes, which were named based on the gene ID order in the pear genome (i.e., PpSPL1–19) (Supplementary Table S11 in Supplementary File 1). All PpSPL family members shared a conserved SBP domain, with two zinc finger-like structures (i.e., Zn-1 and Zn-2) and a highly conserved bipartite nuclear localization signal (NLS; Figure 1). To further study the evolutionary relationships among PpSPL genes and SPL genes from other plant species, we prepared a dataset of 95 SPL sequences from apple, grape, rice, tomato, and A. thaliana. The amino acid sequences of the SBP domains (75 amino acids) encoded by the SPL genes were used to construct a phylogenetic tree with the neighbor-joining algorithm (Figure 2). The plant SBP domains were clustered into seven distinct groups (i.e., G1–G7). The PpSPL family members were distributed over all seven cluster groups, but were most closely associated with apple genes. Additionally, PpSPL9 and PpSPL13 were homologs of A. thaliana AtSPL9, which negatively regulates anthocyanin biosynthesis.

Based on analyses of the degradome, PpSPL9 (Pbr019232.1), PpSPL10 (Pbr020183.1), PpSPL15 (Pbr037448.1), and PpSPL18 (Pbr038099.1) were targeted by miR156 in fruit peels (Figure 3A; Supplementary Table S10 in Supplementary File 1). Additionally, the cleavage site of the miR156ba-targeted PpSPL9 gene was also confirmed by 5′-RACE nested PCR (Figure 3B). Some miR156-targeted PpSPL genes may have been missed during the analysis of the degradome because of inactive genes or low levels of target gene expression in mature pear fruit peels. To assess this possibility, we completed target prediction analyses for all identified PpSPL genes based on sequence complementarity with miR156. We determined that an additional seven PpSPL genes (i.e., PpSPL1, PpSPL4, PpSPL5, PpSPL6, PpSPL12, PpSPL13, and PpSPL14) were putative miR156 targets (Figure 3A). A total of 11 PpSPL genes were potentially regulated by miR156. Interestingly, all 11 miR156-targeted PpSPL genes were clustered into G1–G4 in the phylogenetic tree (Figure 2).

Anthocyanin started to accumulate in pear fruit peels at 3 DABR, and reached about 4.5 mg Cy-3-gal/100 g FW at 6 DABR, while no anthocyanin was detected in control fruit peels (Figures 4A,B). Additionally, after bags were removed, the expression of most anthocyanin biosynthetic genes was upregulated, with peak levels at 1 DABR for PpPAL, 3 DABR for PpCHS, PpCHI, PpF3H, PpANS, and PpUFGT, and 6 DABR for PpDFR (Figure 5). Notably, the PpCHS, PpANS, and PpUFGT transcript levels at 3 DABR were about 200-, 40-, and 600-fold higher than the transcript levels at 0 DABR, respectively (Figure 5). Among the transcription factors, the expression levels of PpMYB10 and PpWD40 increased significantly, and peaked at 1 DABR (i.e., about a 3-fold increase; Figure 5).

Of the miR156 members, miR156a/g/s/sa and miR156ba shared highly similar sequences. Therefore, a conserved primer was designed for analyzing the total expression of miR156a/g/s/sa and miR156ba. After bags were removed from plants, miR156a/ba/g/s/sa transcript abundance increased, peaking at 1–3 DABR (i.e., >3-fold increase) followed by a gradual decrease, but the expression level was still relatively high at 6 DABR (Figure 6). In contrast, miR156a/ba/g/s/sa transcript abundance in the control samples was low at all time points.

We did not detect the expression of PpSPL6, PpSPL8, PpSPL11, PpSPL12, PpSPL14, and PpSPL19 in mature “Meirensu” pear fruit peels. The expression of PpSPL3, PpSPL4, and PpSPL15 was upregulated after bag removal, and peaked at 6, 3, and 1 DABR, respectively (Figure 7). The transcription of PpSPL13 was decreased at 1DABR but upregulated at 3 DABR (Figure 7). The expression of PpSPL2, PpSPL5, PpSPL7, PpSPL9, PpSPL10, PpSPL16, PpSPL17, and PpSPL18 was downregulated at 1 DABR, with the greatest decrease observed for PpSPL9 and PpSPL10 (Figure 7). Among these nine downregulated SPL genes, five were targeted by miR156, namely PpSPL5, PpSPL9, PpSPL10, PpSPL13, and PpSPL18. PpSPL9 and PpSPL13 were homologs of A. thaliana SPL9.

Because anthocyanin biosynthesis involves three transcription factors (i.e., MYB, bHLH, and WD40), the interaction among these proteins was tested with a yeast two-hybrid assay. The PpMYB10, PpbHLH3, PpbHLH33, and PpWD40 open reading frames were inserted into the pGBKT7 and pGADT7 vectors to obtain BD-bait and AD-prey protein complexes. We used SD medium lacking tryptophan, but supplemented with X-α-Gal and AbA, to verify the autoactivation of BD-bait protein complexes. PpMYB10-BD, PpbHLH3-BD, and PpWD40-BD were not autoactivated on medium containing 125 ng/ml AbA. However, PpbHLH33-BD was autoactivated on medium with 1,000 ng/ml AbA (data not shown), so it could not be analyzed by the yeast two-hybrid assay. Our results indicated that PpMYB10 and PpWD40 did not interact with each other, while both transcription factors interacted with PpbHLH3 and PpbHLH33 (Figure 8). These findings suggested that a MYB10/bHLH/WD40 protein complex exists in pear.

In this study, DFR expression was slightly upregulated (<2-fold increase) during anthocyanin accumulation (Figure 5), and NAC genes were not considered. Thus, proteins encoded by genes that were downregulated after bags were removed from plants (e.g., PpSPL2, PpSPL5, PpSPL7, PpSPL9, PpSPL10, PpSPL13, PpSPL16, and PpSPL17) were fused to BD, while PpMYB10, PpbHLH3, and PpbHLH33 were fused to AD to test potential interactions in a yeast two-hybrid assay. Unfortunately, PpSPL2-BD, PpSPL9-BD, and PpSPL18-BD were autoactivated (data not shown). Our results indicated that PpSPL10 and PpSPL13 interacted with PpMYB10 (Figure 9), while there was no interaction between PpSPL and PpbHLH (Supplementary Figure 1).

Different family members of miR156 and SPL genes have been isolated in plants (Cardon et al., 1999; Yu et al., 2015). There are eight miR156 genes in the A. thaliana genome (i.e., miR156A, miR156B, miR156C, miR156D, miR156E, miR156F, miR156G, and miR156H). The expression patterns of these genes were highly tissue-specific. For example, miR156A was highly expressed in the first two pairs of leaves in seedlings, miR156B was specifically expressed in the central cylinder of the primary root primordial cells, and miR156D was highly expressed in seedlings (Yu et al., 2015). This variability in expression patterns suggests that the miR156 genes have diverse roles influencing plant growth and development.

Because of their crucial contribution to the regulation of a broad range of developmental processes, SPL genes have been subjected to genome-wide identification and analysis in diverse plant species. Consequently, 17, 27, 19, 19, and 15 SPL gene family members have been identified in A. thaliana (Cardon et al., 1999), apple (Li et al., 2013), grapevine (Wang et al., 2010), rice (Xie et al., 2006), and tomato (Salinas et al., 2012), respectively. Of the known SPL genes, only SPL9 and SPL10 in A. thaliana and SPL1 in peach have been shown to participate in regulating anthocyanin biosynthesis (Gou et al., 2011; Zhou et al., 2015). There is currently no information regarding if specific miR156 and SPL gene family members are involved in regulating anthocyanin accumulation in pear.

In this study, we identified 26 miR156 members (Supplementary Tables S7 in Supplementary File 1), and 19 pear PpSPL genes (Supplementary Table S11 in Supplementary File 1). According to the phylogenetic tree, the SBP domains were clustered into seven distinct groups (i.e., G1–G7), which was similar to the results of a previous study on apple (Li et al., 2013). The PpSPL family members were more closely related to the SPL genes from eudicots, including apple, grape, tomato, and A. thaliana, than the rice SPL genes (i.e., monocotyledonous species). These findings imply that SPL genes in eudicots diverged more recently from a common ancestor than the SPL genes of monocots. In addition, 11 PpSPL genes are potentially regulated by miR156 (Supplementary Tables S10 in Supplementary File 1; Figure 3), which was similar to apple, where more than half of the MdSBP genes (15 out of 27) are targeted by miR156 (Li et al., 2013).

During the accumulation of anthocyanin after bags were removed from plants, the expression of most anthocyanin biosynthetic genes were upregulated. PpCHS and PpUFGT expression was especially upregulated at 3 DABR, with about 200- and 600-fold increases over expression levels at 0 DABR (Figure 5). Chalcone synthase catalyzes the first committed step of flavonoid and anthocyanin biosynthesis, and the CHS gene is considered to be a UV-B-responsive gene. Additionally, it is targeted by HY5, which is a bZIP transcription factor that mediates UV-B-induced gene expression changes downstream of UVR8 (Binkert et al., 2014). In our previous studies, CHS expression was induced by light (Qian et al., 2013) and methyl jasmonate (Qian et al., 2014b). Similar results were reported for other species, including apple (Ubi et al., 2006), Chinese bay berry (Niu et al., 2010), and Petunia hybrida (Koes et al., 1989). The UFGT enzyme catalyzes the production of stable anthocyanin by adding sugar moieties to the anthocyanin aglycone. The transcription of UFGT is one of the most rate-limiting steps in the anthocyanin biosynthesis pathway (Montefiori et al., 2011). In white grapes, the loss of fruit skin coloration is due to the absence of UFGT transcription (Boss et al., 1996). In litchi (Litchi chinensis), of the six genes encoding proteins associated with anthocyanin biosynthesis, the UFGT genes were the only ones whose transcript levels were significantly correlated with the pericarp anthocyanin content (Wei et al., 2011). These observations suggest that UFGT is required for anthocyanin accumulation.

Among transcription factors, the PpMYB10 and PpWD40 expression levels increased significantly, with peaks at 1 DABR (Figure 5). Additionally, the R2R3 MYB class of transcription factors is the dominant regulator of anthocyanin biosynthesis in several plant species, including pear (Feng et al., 2010), apple (Espley et al., 2009), grape (Kobayashi et al., 2004), and Chinese bayberry (Niu et al., 2010). Specific R2R3 MYBs were reported to activate the promoter regions of anthocyanin biosynthetic genes, such as DFR (Niu et al., 2010) and UFGT (Wang et al., 2014), thereby promoting anthocyanin accumulation. In Chinese sand pear, PpMYB10 expression was considerably upregulated during anthocyanin biosynthesis in different developmental stages (Feng et al., 2010), following methyl jasmonate treatments (Qian et al., 2014b), and after postharvest irradiation (Qian et al., 2013). These findings, together with the results of the present study, indicate that upregulated PpMYB10 expression is related to light-induced anthocyanin accumulation in pear. Interestingly, almost no anthocyanin concentration is detected in European pear under bagging treatment, and the expression of PpMYB10 is not up-regulated as well (Qian et al., 2013), indicating that bagging treatment doesn't activate the light transduction pathway in European pear, or probably the regulation of PpMYB10 expression from the down-stream transcription factors of light transduction pathway (such as HY5) is blocked due to certain factors. Thus, the transcription of PpMYB10 is not triggered and the anthocyanin biosynthesis pathway is not activated.

During the biosynthesis of anthocyanin after bags were removed from plants, miR156 a/ba/g/s/sa abundance increased and peaked at 1–3 DABR (Figure 6). In contrast, we observed a decrease in the transcription levels of PpSPL2, PpSPL5, PpSPL7, PpSPL9, PpSPL10, PpSPL13, PpSPL16, PpSPL17, and PpSPL18 (Figure 7). Additionally, PpSPL9 and PpSPL13 were determined to be homologs of A. thaliana AtSPL9, which regulates anthocyanin biosynthesis (Gou et al., 2011). In A. thaliana, increased miR156 activity promotes anthocyanin accumulation, whereas SPL activity is negatively correlated with anthocyanin content in wide-type plants and transgenic lines expressing Pro35S:MIR156 and Pro35S:MIM156 (Gou et al., 2011). Furthermore, during anthocyanin accumulation in A. thaliana plants treated with salt and mannitol, miR156 expression levels increased, while SPL expression was downregulated (Cui et al., 2014). In the red-fleshed peach cultivar “Dahongpao”, SPL1 was highly expressed in white-fleshed young fruits, but expressed at low levels in red-fleshed ripening fruits (Zhou et al., 2015). These results indicate that miR156 expression is closely related to anthocyanin accumulation, while the expression of the target SPL genes is negatively correlated with anthocyanin biosynthesis. Concerned about the patterns anthocyanin accumulation and gene expression, we hypothesized the procedure of light-induced anthocyanin biosynthesis from up-stream to down-stream: Exposure to light increases the abundance of miR156 (peaked at 1 DABR), which degrades SPL genes (down-regulated at 1 DABR), resulting in the up-regulation of PpMYB10 (peaked at 1 DABR) and formation of the MYB/bHLH/WD40 complex, subsequently activates the expression of most structural genes including PpCHS, PpCHI, PpF3H, PpANS, and PpUFGT (peaked at 3 DABR). More proteins encoded by these genes are translated, catalyze the reactions of anthocyanin biosynthesis pathway, and finally lead to the accumulation of anthocyanin from 3 DABR to 6 DABR.

The regulation of anthocyanin biosynthesis involves numerous transcription factors, including MYB, bHLH, and WD40, which form the MYB/bHLH/WD40 complex that mediates the expression of anthocyanin biosynthetic genes in diverse species, including A. thaliana (Broun, 2005), Chinese bayberry (Liu et al., 2013), and strawberry (Schaart et al., 2013). Among these transcription factors, bHLH acts as a bridge interacting with MYB and WD40 to form the complex. An earlier study involving apple revealed that there is no direct interaction between MYB and WD40 proteins (An et al., 2012). To the best of our knowledge, the current study is the first to examine the MYB/bHLH/WD40 complex in pear. Our yeast two-hybrid results indicated that PpMYB10 and PpWD40 interacted with PpbHLH3 and PpbHLH33, but not with each other. PpbHLH3 and PpbHLH33 were observed to interact with each other (Figure 8). Our findings imply that the MYB10/bHLH/WD40 transcription factor complex is conserved among plant species.

Our results revealed that PpSPL10 and PpSPL13 (orthologs of A. thaliana AtSPL9) interacted with PpMYB10 (Figure 9). We also observed that DFR expression was slightly upregulated (<2-fold increase) during anthocyanin accumulation (Figure 5). Therefore, the miR156–SPL9–DFR pathway, which responds to abiotic stresses in A. thaliana (Cui et al., 2014), may not be responsible for the light-induced accumulation of anthocyanin in pear. According to the previous results in A. thaliana (Gou et al., 2011), we hypothesize the involvement of miR156-SPL module in light-induced pear peel coloration and anthocyanin accumulation probably through the interaction between SPL and MYB10, which is the crucial part of MYB/bHLH/WD40 complex.

In conclusion, after bag removal, during the accumulation of anthocyanin in pear and up-regulation of genes related to anthocyanin biosynthesis, the expression of miR156 and its target PpSPL genes, including PpSPL2, PpSPL5, PpSPL7, PpSPL9, PpSPL10, PpSPL13, PpSPL16, PpSPL17, and PpSPL18 were increased and decreased, respectively. In addition, three transcription factors MYB10, bHLH, and WD40 could form a protein complex, and PpSPL10 and PpSPL13 could also interact with PpMYB10. Our results provide some clues about the miR156-SPL module probably participating in light-induced red peel coloration in pear.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer JL declared a shared affiliation, though no other collaboration, with the authors LB and DZ to the handling Editor, who ensured that the process nevertheless met the standards of a fair and objective review.