PbrmiR397a regulates lignification during stone cell development in pear fruit

Stone cell formation and lignin deposition mainly occur during early stages of pear fruit development

To examine lignin deposition in the flesh of pear fruit, hand-cut sections were stained with phloroglucinol-HCl (Wiesner reagent). As shown in Figure 1a, lignified stone cells were stained red violet. The staining pattern indicates that stone cells formed rapidly from 21 to 49 days after full bloom (DAF), after which their distribution became gradually diluted from 63 to 150 DAF (Figure 1a). Stone cell content accumulated rapidly in fruit from 21 to 49 days, remaining stable after 49 DAF (Figure 1b). In addition, lignin content in stone cells remained a stable level at each development stages (Figure 1c). In general, stone cell formation and lignin deposition in pear fruit occur during the early stages of fruit development. Stone cells and lignin will not be degraded after their formation but be maintained in fruit flesh.

Genomewide identification of LAC genes and correlation of LACs with lignin biosynthesis during pear fruit development

Based on the results of Basic Local Alignment Search Tool (BLAST) queries, 38, 48, 31, 37 and 34 LAC genes were identified from pear, apple, peach, strawberry and tobacco genomes, respectively. These homologues were named as PbrLACs, MdLACs, PpLACs, FvLACs and NbLACs (Table S1). Based on the results of phylogenetic analyses, LACs were grouped into six clades, clades 1–6 (Figure 2a). Eleven PbrLACs clustered with AtLAC4, AtLAC10, AtLAC11 and AtLAC16 in clade 1 and nine clustered with AtLAC17 in clade 2 (Figure 2b). It was previously reported that mutations of AtLAC4, AtLAC11 and AtLAC17 led to a lignin content reduction, while the mutations of AtLAC10 and AtLAC16 caused no altered phenotypes in Arabidopsis (Berthet et al., 2011; Cai et al., 2006; Zhao et al., 2013). We predicted that some of PbrLAC genes might be involved in lignin biosynthesis in pear fruit as they were clustered together with AtLAC4, AtLAC11 and AtLAC17.

RNA-Seq data of six pear cultivars (‘Dangshansuli’, ‘Hosui’, ‘Yali’, ‘Kuerlexiangli’, ‘Nanguoli’ and ‘Starkrimson’) were generated at six fruit development stages (beginning of fruit set, physiological fruit drop, fruit rapid enlargement, a month after fruit enlargement, pre-mature and mature stage) (Wu et al., 2013; Zhang et al., 2016). From these data, the transcripts were detected for 29 of the 38 PbrLAC genes. Six of these genes (PbrLAC1, 2, 3, 15, 18 and 20) were abundantly expressed during the early stage of fruit development while other genes were very weakly expressed and excluded from further studies (Table S2). In apple, none of the MdLACs were abundantly expressed during fruit flesh development (Table S3). In peach and strawberry, only PpLAC11 and FvLAC1, which belong to clade 6 in the phylogenetic tree, were expressed abundantly during fruit development (Tables S4 and S5) (Gu et al., 2016; Kang et al., 2013). However, in that clade, no genes were reported to be directly involved in lignin biosynthesis. The six PbrLAC genes were selected to verify their expression at eight different stages of fruit development for ‘Dangshansuli’ by q-PCR. All six genes were expressed abundantly at 21 and 35 DAF, prior to lignin content peaking at 49 DAF, after which expression decreased rapidly (Figure 2c). The expression levels of PbrLAC genes were consistent with lignin contents during fruit development. Based on the phylogenetic tree and expression analysis, these PbrLAC genes are candidates for catalysing lignin biosynthesis in the formation of stone cells (the specific trait that differs from the other three Rosaceae species) in pear fruit.

To examine the subcellular localization of these pear LAC proteins, PbrLAC-GFP expression vectors of PbrLAC1, 2 and 18, which had higher expression during fruit development according to the RNA-Seq data (Table S2), were transferred into onion epidermal cells. In plasmolysed cells, green fluorescence signals from the fusion constructs PbrLAC1-GFP, PbrLAC2-GFP and PbrLAC18-GFP were detected specifically in cell walls, and signals from the 35S-GFP control construct were detected throughout the cell (Figure S1). This result suggests that these LAC proteins are specifically located in cell wall and are colocated at the site of lignin biosynthesis.

Verification of PbrLACs and NbLACs as targets of PbrmiR397a

We have previously shown that 27 LAC genes identified from genomic sequences are potential target genes of PbrmiR397a (Wu et al., 2014). In this study, 20 NbLACs were identified as potential targets of miR397a in tobacco. The target sites of PbrmiR397a are located in sequences encoding a conserved Cu-oxidase domain in both pear and tobacco genes (Figure S2a). We performed a 5′-rapid amplification of cDNA ends (5′-RACE) analysis (Leng et al., 2016) using six PbrLACs (1, 2, 3, 15, 18 and 20) selected based on their strong expression in pear fruit and six tobacco NbLACs (1, 3, 10, 11, 12 and 17). The analysis verified that all 12 LACs were targets of PbrmiR397a (Figure 3a). To assay targeting, we constructed a vector containing the miR397a gene and a series of reporter vectors containing firefly luciferase (LUC) gene fused to a miR397a target sequences or a mutated target sequences amplified from LAC genes (Figure 3b). When young Nicotiana benthamiana leaves were infiltrated with a reporter vector only, LUC activities were similar among leaves infiltrated with the control reporter or with a test reporter (Figure S3). This result suggests that there was little endogenous miR397a in N. benthamiana leaves. When leaves were infiltrated with a reporter vector together with the miR397a effector vector, LUC activities were significantly lower in leaves infiltrated with LUC reporter fused to a miR397a target sequence than those in leaves infiltrated with LUC reporter fused to a mutated miR397a target sequence (Figure 3c). This result indicated that miR397a might act on the target sequence but not the mutated target sequence to regulate the expression of LUC. The results of both 5′-RACE and reporter gene analysis verified the computational prediction and supported a regulatory role of PbrmiR397a in suppressing these LACs.

The 24 PbrLACs targeted by PbrmiR397a belonged to clades 1, 2, 3, 4 and 6 (Figure 2a), but none of them were classified into clade 5. The relative expression of PbrmiR397a was abundant at 21 DAF but decreased rapidly thereafter. PbrmiR397a expression contrasted with that of PbrLACs at 21 and 35 DAF, but the expression of both decreased rapidly thereafter (Figure 3d). This suggests that PbrmiR397a plays an important role by regulating the expression of PbrLACs in the early stages of fruit development.

Overexpression of PbrmiR397a or the simultaneous silencing of PbrLAC1, 2, and 18 reduces lignin contents in pear fruits

To further elucidate the roles of PbrmiR397a and its targets PbrLAC1, 2 and 18 in lignin biosynthesis, miR397a overexpression or LAC antisense constructs were agroinfiltrated into ‘Dangshansuli’ pear fruit at 35 DAF. Ten days after infiltration, a strong reduction in lignin staining was observed at infiltration sites both for miR397a overexpression and for LAC1, 2, and 18 silencing together, but a reduction in lignin staining was not observed at the injection sites for the control vector or for single LAC antisense constructs (Figures 4a and S4).

The lignin content in the fleshy tissue around the infiltration sites showed 30.5% and 25.9% reduction for 35S-PbrmiR397a and 35S-antiPbrLAC1, 2, 18 together, respectively, when compared with that of the corresponding noninfiltrated sites (Figure 4b). The lignin content at other sites infiltrated with 35S-antiPbrLAC1, 35S-antiPbrLAC2 or 35S-antiPbrLAC18 decreased moderately. However, no significant differences were observed compared with that of their corresponding noninfiltrated sites. In addition, the lignin content in the fleshy tissue around the infiltration sites was unchanged when the control empty vector (EV) was used.

The expression of six genes (PbrLAC1, PbrLAC2, PbrLAC18, PbrPOD1, PbrPOD2 and PbrUGT72E) at the injection sites was analysed by q-PCR. PbrPOD1, PbrPOD2 and PbrUGT72E are putative orthologous genes involved in the transfer of monolignols in Arabidopsis (Lin et al., 2016). Injection of the PbrmiR397a overexpression construct decreased the expression of PbrLAC1, PbrLAC2 and PbrLAC18 but increased the expression of PbrPOD1, PbrPOD2 and PbrUGT72E. This gene expression pattern was repeated after simultaneously injecting the PbrLAC1, 2 and 18 silencing constructs (Figure 4c). However, injection of each individual PbrLAC gene-silencing construct suppressed only its own expression and not the expression of the other five genes in the lignin pathway.

Sequence and expression analyses of PbrmiR397a and PbrLACs in pear varieties reveal the basis of stone cell content

The stone cell content of 304 pear varieties has been assessed (Zhang et al., 2016). Of these varieties, 30 high- and 30 low-stone cell-content varieties were selected to evaluate the relationship between stone cell content and genetic variation in PbrmiR397a and PbrLACs. The average stone cell content ranged from 10.53% to 20.11% for the high group (HG) and from 3.71% to 6.78% for the low group (LG) (Figure 5a). After sequencing the genome of these 60 pear varieties, we identified SNPs in the genomic region of PbrmiR397a (3107 bp), PbrLAC1 (4985 bp), PbrLAC2 (4943 bp), and PbrLAC18 (4822 bp). There was no correlation between the SNPs of PbrLACs and stone cell content (Table S6). Twenty-five SNPs (named #1 to #25) were identified in the 3-kb promoter region of the PbrmiR397a gene among the sixty pear varieties, but none were identified in the PbrmiR397a precursor. Of these 25 SNPs, four (#1, #7, #12 and #15) were highly correlated with stone cell content in the HG and LG (Figure 5a). A phylogenetic tree based on the upstream sequences of PbrmiR397a classified all 30 HG varieties into one clade and all 30 LG varieties into another clade (Figure 5b), which indicated that these four SNPs are useful for marker-assisted breeding. To verify the identified SNPs, the 3-kb promoter region of PbrmiR397a was cloned from three randomly selected varieties in both the HG and LG. The complete sequencing results of these 3-kb fragments confirmed all previously identified SNPs and revealed six new SNPs, but the results did not reveal any indels. These six new SNPs did not correlate with stone cell content (Figure S5).

Using the PlantCARE database for analyses of conserved promoter elements (Rombauts et al., 1999), we identified several known cis-elements in the promoter region of PbrmiR397a in the HG and LG. SNP #7 (G>A) creates a TCA element, and SNP #15 (G>A) disrupts a TGACG motif in the LG varieties. All other identified cis-elements were the same between the HG and LG groups. The TCA element and TGACG motif are hormone-responsive elements. To determine whether the SNPs affect promoter activity, dual-luciferase reporter assays were performed using Arabidopsis leaf protoplasts. The assay results showed that the levels of luciferase driven by pmiR397a-HG and pmiR397a-LG were low and did not respond to auxin treatments (Figure S6). However, application of 20, 30 or 40 μm salicylic acid (SA) greatly enhanced the levels of luciferase driven by pmiR397a-LG but not by pmiR397a-HG (Figure 5c). To determine whether the alteration in cis-elements caused by SNP #7 is essential for the SA-inducible luciferase activity, deletion variants of the pmiR397a-LG promoter were tested using the assay. The promoter fragments pmiR397a-LG and pmiR397a-LG1, both containing a TCA element, were activated by treatment with 30 μm SA. Other promoters that lack the TCA element or that contain the mutant TCA element displayed low levels of luciferase activity (Figure 5d). In pear fruit, the expression level of PbrmiR397a was significantly lower in the 15 HG varieties than in 15 LG varieties. In contrast, the expression level of PbrLACs was higher in the 15 HG varieties than in the 15 LG varieties (Figure 5e). Based on these results, we propose that the alterations to cis-elements caused by SNP #7 may be responsible for the different expression levels of PbrmiR397a between the HG and LG varieties.

Overexpression of PbrmiR397a reduces NbLAC transcript levels and lignin content in transgenic tobacco plants

Nine hygromycin-resistant primary transgenic plants were confirmed to contain the transgene PbrmiR397a (Figure 6a). These plants were grown to T2 generation and used for q-PCR analysis to determine the expression of PbrmiR397a. Four transgenic lines (397a-3, 397a-4, 397a-8 and 397a-11) that exhibited relatively high levels of PbrmiR397a expression (Figure 6b) were selected for further analyses in T2 generation. The biomass measurements, including plant height, leaf area, and fresh weight of leaves and whole plant of T2 homozygous transgenic tobacco showed no significant differences from that of wild-type (WT) control plants (Table 1). However, the T2 plants showed 18%–29% reduction in lignin content compared to WT control plants (Table 2). Even with this reduction, there was no change in the ratio of S-/G-lignin monomers compared to WT plants (Table 2).

These tobacco lines also showed a different arrangement of vascular tissue; the vascular tissue had highly reduced numbers of vessel elements (Figure 6c–f) and thinner xylem regions (Figure 6k). Ultraviolet (UV) autofluorescence further confirmed that lignin autofluorescence was stronger in the xylem tissues of WT than in those of the 397a-8 line (Figure 6g,h). The secondary wall thickness of transgenic PbrmiR397a lines was also significantly lower than that of WT plants (Figure 6i,j and l).

The expression of fifteen of the 20 LAC genes that are targets of PbrmiR397a in Nicotiana tabacum in stem-differentiating xylem (SDX) of the second internode of 2-month-old tobacco plants was measured by q-RT-PCR analysis (Figure S7). Nine of the 15 SDX-expressed LACs were down-regulated by 28%–90% in the PbrmiR397a transgenic line compared with the WT. Nontarget genes of miR397a, which included POD, PRX and UGT72E, were up-regulated more than twofold. There was no significant change in expression of other genes tested (Table S7), which indicated that the reduced lignin content is specifically caused by the down-regulation of the nine LACs.

Manipulation of miR397 alters lignin content

In rice, overexpression of miR397 promotes increased branch numbers and increased grain size via the suppression of OsLAC (Zhang et al., 2013). In poplar, stem lignin content is reduced in transgenic poplar plants overexpressing PtrmiR397a (Lu et al., 2013). In Arabidopsis, overexpression of miR397b resulted in normal plant development but reduced lignin content, increased number of seeds and enlarged seed size (Wang et al., 2014). In the present study, overexpression of PbrmiR397a in tobacco plant led to reduced lignin content; however, no developmental effects were observed. In contrast, disruption of AtLAC4 and AtLAC17, two target genes of miR397, leads to a semidwarf phenotype that has reduced lignin content under conditions of continuous light (Berthet et al., 2011); triple mutants (AtLAC4, AtLAC11 and AtLAC17) are severely dwarfed and exhibit both very low lignin levels and severely collapsed xylem cells (Zhao et al., 2013). Overexpression of OsLAC in WT plants results in arrested plant growth, reduced survival rates, and reduced grain yield and biomass (Zhang et al., 2013). All these results suggest that manipulation of miR397 might be a more effective way for reducing lignin biosynthesis than manipulation of individual LAC genes, as miR397 can regulate the transcript levels of several LAC genes simultaneously without causing apparent adverse effects on plant development.

Plant hormones may regulate the expression level of miR397

In pear, PbrmiR397a expression levels were significantly higher in the varieties with low levels of stone cells (LG), compared with high stone cell (HG) varieties, whereas the expression levels of PbrLACs were opposite to those of PbrmiR397a. A strong association between high levels of PbrmiR397a and low levels of lignin and stone cell content in different pear varieties was revealed. A SNP in the promoter of the PbrmiR397a gene was identified by genomewide sequencing to be associated with stone cell content in sixty pear varieties. In addition, we found that the SNP created a TCA element that may respond to SA signals and induce the expression of PbrmiR397a. In fact, most plant hormones affect the lignin biosynthesis pathway. Auxin negatively regulates vessel lignification during peach endocarp lignification (Zhang et al., 2015). Also, decreased levels of cytokinin in response to the reduced expression of ABCG14, a newly identified cytokinin transporter, lead to delayed lignin biosynthesis and reduced numbers of lignified cells (Zhang et al., 2014). Our results showed that alterations to cis-elements caused by SNP #7 might result in the markedly different expression of PbrmiR397a observed between the HG and LG. This finding will help to develop DNA markers to detect pear varieties or progenies that have low lignin and stone cell contents.

Relationships of PbrLACs, PbrPOD and PbrUGT72E with lignin synthesis in pear fruit

LACs have similar structural conservation between bacteria, fungi and plants (Dwivedi et al., 2011). However, the number of LAC genes has increased; throughout plant evolution, the number of LACs ranges from three in Chlamydomonas reinhardtii to 12 in Physcomitrella patens, 10 in Selaginella moellendorffii and 17–39 in angiosperms (Weng and Chapple, 2010). These numbers indicate that LAC members have expanded during the evolution from lower plant species to higher plant species. In Arabidopsis, the lac4 and lac17 double-knockout mutant shows semidwarf phenotype and lower levels of lignin content under continuous light conditions (Berthet et al., 2011). Triple mutants of AtLAC4, AtLAC11 and AtLAC17 have significant lignin reduction and blocked development (Zhao et al., 2013), which indicates that all three LAC genes contribute to lignin synthesis and are functionally redundant. Our results reveal that the involvement of LACs during stone cell lignification is supported by specific LAC transcription profiles: expression levels of PbrLAC genes were positively correlated with lignin contents during fruit development and were higher in the HG varieties than in the LG varieties. Furthermore, the simultaneous silencing of PbrLAC1, PbrLAC2 and PbrLAC18 significantly reduced lignification, but silencing one LAC gene led to only a slight reduction in lignin content, which indicates functional redundancy among PbrLAC genes.

Up-regulation of POD and UGT72E expression was observed in pear fruit cells when PbrmiR397a was overexpressed or when PbrLACs were simultaneously silenced. To prevent the accumulation of monolignols in plasma membranes or inner cell walls when oxidative capacity is diminished, some form of feedback regulatory mechanism may exist (Vanholme et al., 2013). In Arabidopsis, the glycosyltransferases UGT72E2 and UGT72E3 synthesize glucosylate monolignols into coniferin or syringin (Lanot et al., 2006). Monolignol glycosylation could increase the solubility and decrease the reactivity of monolignols, and the up-regulation of PbrUGT72E could be a detoxification mechanism.

Reverse genetic approaches have shown an active role of PODs during lignin formation in multiple species. The down-regulation of PRX60 in tobacco plants (Blee et al., 2003) or of PRX3 in aspen (Li et al., 2003) results in a reduction of lignin accumulation and altered lignin composition. Moreover, Arabidopsis AtPRX2, AtPRX71 and AtPRX25 single- or double-knockout mutants exhibit reduced lignin accumulation but normal stem height (Shigeto et al., 2013). However, we observed that lignin content was not recovered by the twofold up-regulation of POD expression after LAC genes were repressed in pear fruit. This observation indicated that LAC and POD activities are nonredundant and that both are necessary for monolignol polymerization during lignin accumulation. This finding subsequently led to a hypothetical model in which LACs and PODs act in sequential order during cell lignification, starting with LACs and followed by PODs, to achieve full cell wall lignification. Once the initial lignin polymers are catalysed by LACs, the late intervention of PODs is further supported, as PODs can form rigid crosslinks between lignin, extensins and hemicelluloses in the secondary cell wall (Lagrimini et al., 1987; Passardi et al., 2004).

The results of this research will aid future studies to reveal the regulatory network of lignin biosynthesis. Revealing of the regulatory network not only will help to improve pear fruit quality but also will promote the production of wood and biofuels.

Plant materials

‘Dangshansuli’ (P. bretschneideri, white pear group) fruit samples were collected from the same trees at 21, 35, 49, 63, 77, 91, 105 and 150 DAF in an orchard (Gaoyou County, Jiangsu Province, China). The fruit of 60 varieties was collected at 35 DAF from the Chinese National Pear Germplasm Repository in Wuhan.

Histology

Hand-cut sections of pear fruit were treated with 30% HCl (v/v) for 1 min, stained in 12% phloroglucinol and 95% ethyl alcohol (w/v) for 7 min, and then washed with water. Cell walls containing lignin can be stained by phloroglucinol-HCl (Wiesner reagent); however, nonlignified cells are not clearly stained (Wan et al., 2014; Zhong et al., 2000). Pear fruit sections were imaged using a hand-held camera.

Analysing stone cell and lignin contents in fruit flesh

Stone cell content was carried out using frozen-HCl treatment as previously described (Tao et al., 2009), and data were shown as dry weight (g) per pear fruit. The acetyl bromide-based method was used to detect lignin in fruit. The lignin content was shown as the percentage (calculated lignin content/stone cells dry weight × 100). The methods were the same as those previously described (Tao et al., 2009). Three independent experiments were performed (at least ten fruits were used in each experiment).

LAC gene identification and phylogenetic analysis

PbrLACs, MdLACs, PpLACs, FvLACs and NbLACs were identified by the comparison of 17 Arabidopsis LAC protein sequences against the predicted protein sequences of pear, apple, peach, strawberry and tobacco using the tBLASTn and BLASTp algorithms (Gertz et al., 2006) and an e-value cut-off of 1E⁻¹⁰. The BLAST results detected copper-oxidase domains in National Centre for Biotechnology Information (NCBI) conserved domain database. A phylogenetic tree was constructed using MEGA 5.1 (Tamura et al., 2011). The reliability of branching was assessed by bootstrap resampling using 1000 replicates.

Subcellular localization of PbrLACs

The full-length cDNAs of PbrLAC1, 2 or 18 were fused in frame to the N-terminus of GFP cDNA to form fusion vectors 35S-PbrLAC1-GFP, 35S-PbrLAC2-GFP or 35S-PbrLAC18-GFP (Figure S8a). The fusion constructs and the control vector (35S-GFP) were transferred into Agrobacterium tumefaciens strain GV3101 by the freeze–thaw method. The constructs were introduced into onion epidermal cells by Agrobacterium-mediated transformation in accordance with previously described methods (Sun et al., 2007). One day after transformation, the onion epidermal cells were plasmolysed with a 0.3 g/mL sucrose solution and examined for green fluorescence signals using a Leica TCs SP2 spectral confocal microscope (Leica Microsystems, Wetzlar, Germany). To obtain comparable images, laser intensity, pinhole and photomultiplier gain settings were kept constant between samples.

RNA and DNA isolation

Total RNA and DNA were extracted using the cetyl-trimethylammonium bromide (CTAB)-based method (Porebski et al., 1997). The RNA and DNA samples were quantified with a Nanodrop spectrophotometer (Thermo Fisher Scientific, Massachusetts, USA).

Expression profiles of genes

First-strand cDNA synthesis was performed using the SYBR PrimeScript miRNA RT-PCR Kit (Takara, Kusatsu, Japan). Q-PCR was performed using the LightCycler 480 (Roche, Basel, Switzerland). Each gene was repeated for three biological samples and three technical repeats. Relative expression levels of each gene were calculated using the 2^−ΔΔCp algorithm. PbrGAPDH and U6 were used as reference genes for PbrLACs and PbrmiR397a, respectively. NbActin served as a housekeeping gene for lignin biosynthesis genes.

Reverse transcription was performed using one-step gDNA removal and cDNA synthesis kit (Transgen, Beijing, China). Q-RT-PCR analysis was performed using a thermocycler. NbActin served as the housekeeping gene for NbLACs.

Computational prediction and experimental validation of PbrmiR397a targets

PbrmiR397a targets were predicted from the pear and tobacco genome sequences using the psRNATarget server (Dai and Zhao, 2011). A penalty score of 2.5 calculated as previously described (Jiu et al., 2015; Leng et al., 2016) was applied to potential target sites in the miRNA:mRNA duplexes (Figure S2b). Experimental validation of predicted targets was carried out using a modified RNA ligase-mediated 5′-RACE and a GeneRacer kit (Thermo Fisher Scientific, Massachusetts, USA). PCR was performed on cDNA from ‘Dangshansuli’ pear fruit at 21 DAF and stem tissue of 2-month-old tobacco plants. Primers used for amplifying are shown in Table S8. The PCR products were gel purified, subcloned into pEasy vectors (Transgen) and sequenced using the plasmid DNA of at least eight independent clones.

The pGreenII Dual-Luciferase miRNA Target Expression Vector, derived from pGreenII 0800-LUC (Hellens et al., 2005), was used to quantitatively evaluate miRNA activity. For this evaluation, an effector and three control vectors were constructed to contain CaMV35S::miR397a, CaMV35S::LUC, 35S::LUC::5× ts-miR397 and 35S::LUC::5× its-miR397, respectively (Figure 3b). In the positive control vector 35S::LUC::5× ts-miR397, LUC was fused to five copies of target sequence of PbrmiR397a. In the negative control vector 35S::LUC::5× its-miR397, LUC was fused to five copies of inverted target sequence of PbrmiR397a. The candidate miR397a target and its flanking sequence (200- to 300-bp upstream and downstream of target sequence) were PCR amplified from cDNA of ‘Dangshansuli’ pear fruit at 21 DAF and stem tissue of 2-month-old tobacco plants using primers designed to bind to LAC genes of pear and tobacco (Table S8). Candidate target vectors were constructed by fusing these PCR fragments to the 3′ of the LUC gene in the basal plasmid (35S::LUC) using ClonExpress Entry One Step Cloning Kit (Vazyme, Nanjing, China). The mutant candidate target and its flanking sequence were amplified with alterations of two bases (GC to AA) at cleavage site of the target sequence using the method described previously (Nicolas, 2011; Yan and Cullen, 2003) and primers listed in Table S8. Renilla luciferase (REN) was in the same vector as LUC and was used as an internal control for normalization of LUC expression.

All the vectors were transferred into A. tumefaciens GV3101. Agrobacterium containing the effector vector or report vector were cultured separately and then mixed in a ratio of 10:1 in an infiltration buffer (10 mm MgCl₂; 200 μm acetosyringone; 10 mm MES, pH 5.5) to a final concentration of OD₆₀₀ between 0.7 and 1.0. Young N. benthamiana leaves were infiltrated with the mixture of Agrobacterium cultures using needleless syringes. The N. benthamiana plants were grown in a glasshouse with daylight extension to 16 h. For each effector–reporter combination, two independent experiments were performed using three infiltrated leaves from three different plants in each experiment. Three days after infiltration, LUC and REN were assayed using dual-luciferase assay reagents (Promega, Madison, USA). The relative ratio of LUC/REN calculated and compared to the corresponding control value set at 100.

Functional study of PbrmiR397a and PbrLAC using transiently transformed pear fruit flesh

We prepared a pCambia 1301 construct to overexpress PbrmiR397a (35S-PbrmiR397a; Figure S8b). Three unique parts of PbrLAC1, 2 and 18 were inserted into the vector in a reverse orientation to form the antisense constructs 35S-antiPbrLAC1, 35S-antiPbrLAC2, and 35S-antiPbrLAC18 (Figure S8b). These constructs and EVs were individually transformed into A. tumefaciens GV3101. For infiltration, GV3101 cells were inoculated in Luria-Bertani (LB) medium and cultured at 28 °C while shaking at 200 r.p.m. for 1 day. Afterwards, these cells were centrifuged, resuspended in an infiltration buffer (10 mm MgCl₂; 200 μm acetosyringone; 10 mm MES, pH 5.5) and then maintained at 21 °C for 6 h. The cells were infiltrated into the flesh of ‘Dangshansuli’ fruit at 35 DAF using needleless syringes. Six fruits were injected with each construct in an experiment that was repeated three times. The transformed fruit was placed in the dark at 21 °C overnight and then transferred to a growth chamber (21 °C, 16-h light/8-h dark photoperiod) under low light conditions for 10 days before being examined and imaged (Zhou et al., 2015).

Tobacco plant transformation and lignin analysis

The overexpression vector PbrmiR397a used in fruit infiltration was also used for tobacco (N. tabacum) transformation via the Agrobacterium-mediated leaf disc transformation method (Thomashow, 1999).

Nine randomly selected hygromycin-resistant independent T0 transgenic tobacco plants were analysed by PCR using PbrmiR397a-specific primers (Table S8) to confirm the presence of PbrmiR397a transgene. Seeds of a T0 plant were sown on hygromycin selection medium to generate T1 plants. Ten T1 plants from a T0 were planted in a greenhouse for seed collection from each plant separately. These seeds were sown on hygromycin containing medium to generate T2 plants. If 100% of T2 plants were hygromycin resistance, the T1 plant was considered as a homozygous transgenic plant and its seeds were resown directly into soil without hygromycin selection to generate T2 plants that were used for the subsequent experiments. Transcript levels of PbrmiR397a in SDX at the second internode from the shoot tips of 8-week-old T2-generation plants were analysed by q-PCR using primers described in Table S8. The experiment was repeated for three biological samples from three different plants. Four of these lines (397a-3, 397a-4, 397a-8 and 397a-11) that exhibited relatively high levels of PbrmiR397a expression were used for the subsequent experiments. Plants of each line were grown in a glasshouse with daylight extension to 16 h. Plant height, fresh weight of leaves per plant and whole plant, and leaf area were determined for eight plants of each transgenic line and WT control when the plants were at 16 weeks old.

Whole stems from WT and T2 transgenic plants were frozen in liquid nitrogen. After freezing, the xylem ring from five plants of each line and WT was manually collected by removing the pith and cortex. The xylem tissues were pooled from five plants, milled to a fine powder and sequentially extracted with acetone, a mix of toluene and ethanol (2/1, v/v), ethanol, and water. The resulting cell wall residue was dried for lignin analysis. Klason lignin content was estimated according to the standard procedures (Lin and Dence, 2012). The lignin composition was studied by thioacidolysis, as previously described (Lapierre et al., 1995). The lignin-derived thioacidolysis monomers were identified by GC-MS as their trimethylsilylated derivatives. All the results were the mean value obtained from four independent experiments performed on the same amount of starting materials derived from pooled xylem tissues of five plants.

Microscopy

The second highest internode of 8-week-old tobacco stems from five plants of WT and T2-generation PbrmiR397a transgenic lines were fixed in formalin-acetic acid-alcohol fixative at 4 °C for 1 week. Samples were dehydrated in an ethanol series and embedded in paraffin. Cross-sections (5 μm) were cut with a Leica RM 2015 ultramicrotome (Leica Microsystems, Wetzlar, Germany). The sections were placed onto glass slides and stained with 1% (w/v) Toluidine Blue O [with 1% (w/v) sodium borate] for 5 min, after which the sections were observed using a Leica TCs SP2 spectral confocal microscope (Leica Microsystems). The width of xylem in WT and transgenic plants was measured with the images of Toluidine Blue O-stained sections by Image-Pro Plus software. The width of xylem from each line was the mean value of eight sections at the diagonal position in the image of the stem cross-section from five plants of each lines.

For visualization of total lignin autofluorescence, 5-μm sections were observed using a Leica TCS SP2 AOBS confocal laser scanning microscope (Leica Microsystems) illuminated with a 405-nm blue diode laser, and emission was detected at 460 nm (Vanholme et al., 2013). To obtain comparable images, laser intensity, pinhole, and photomultiplier gain settings were kept constant between samples.

Regarding transmission electron microscopy (TEM), the same materials of paraffin section were fixed in fixative for 12 h. The method was used as described previously (Whitehill et al., 2016). The secondary wall thickness of vessel cells was measured with the images of TEM. Five TEM images were selected randomly from each line, and eight values of cell wall thickness at the diagonal position were obtained from each image, after which the final mean values of five images were calculated for each line.

SNP calling, cis-element predicting and phylogenetic analysis

Sixty varieties were collected from the Chinese National Pear Germplasm Repository in Wuhan. The sequences of 150 bp of paired-end reads were then filtered with Trimmomatic software (Bolger et al., 2014). The total clean data for each sample were at least 5.3 Gb (10× genome coverage). Sequence read mapping (default parameters) was carried out using Burrows-Wheeler alignment (Li and Durbin, 2009). The module ‘mpileup’ in SAMtools (Li et al., 2009) was used for SNP calling. An in-house Perl script was then used to analyse the effectiveness of the SNPs. The lowest number of reads to support the SNPs was four, and the individual base percent was greater than 25%. Cis-elements in promoter region were predicted with PlantCARE (Rombauts et al., 1999). Phylogenetic analysis was performed based on sequence data of the 3.0-kb promoter region of the PbrmiR397a gene.

Dual-luciferase assays of transiently transformed leaf protoplasts

The 3.0-kb promoter region of the PbrmiR397a gene was amplified from ‘Chenjiadamali’ (HG pear variety) and ‘Cuilv’ (LG pear variety) and inserted into a pGreenII 0800-LUC vector (Hellens et al., 2005). Four partial promoter sequences of the PbrmiR397a gene from the LG (pmiR397a-LG1, pmiR397a-LG1m, pmiR397a-LG-2 and pmiR397a-LGD) were amplified and inserted into the same vector (see Figure 5d). Primers used for amplifying are shown in Table S8. The transformed colonies were selected and cultured for plasmid DNA that was purified with a Plasmid Maxprep Kit (Vigorous, Beijing, China). The plasmid DNA of reporter constructs (promoter-luciferase) was transferred into the protoplasts from leaves of 4-week-old short-day Arabidopsis plants as previously described (Yoo et al., 2007). After transfection, protoplasts were cultured for 18 h. The method used for hormone treatment as previously described with modifications (Wang et al., 2016). A series of concentrations of auxin and SA were applied to each sample at 2 h after transfection. Luciferase activities were determined using a dual-luciferase reporter assay system (Promega). Relative luciferase activities of treatment samples were calculated by normalizing against the activity of control samples.

Statistical analysis

The data were statistically processed using the SAS software package (SAS Institute, North Carolina, USA); analysis of variance was used to compare the statistical difference based on t-test at the significance levels of P < 0.05 (*), P < 0.01 (**).