MicroRNA candidate miRcand137 in apple is induced by Botryosphaeria dothidea for impairing host defense

Abstract

MicroRNA (miRNA)-mediated gene silencing is a master gene regulatory pathway in plant–pathogen interactions. The differential accumulation of miRNAs among plant varieties alters the expression of target genes, affecting plant defense responses and causing resistance differences among varieties. Botryosphaeria dothidea is an important phytopathogenic fungus of apple (Malus domestica). Malus hupehensis (Pamp.) Rehder, a wild apple species, is highly resistant, whereas the apple cultivar “Fuji” is highly susceptible. Here, we identified a 22-nt miRNA candidate named miRcand137 that compromises host resistance to B. dothidea infection and whose processing was affected by precursor sequence variation between M. hupehensis and “Fuji.” miRcand137 guides the direct cleavage of and produced target-derived secondary siRNA against Ethylene response factor 14 (ERF14), a transcriptional activator of pathogenesis-related homologs that confers disease resistance to apple. We showed that miRcand137 acts as an inhibitor of apple immunity by compromising ERF14-mediated anti-fungal defense and revealed a negative association between miRcand137 expression and B. dothidea sensitivity in both resistant and susceptible apples. Furthermore, MIRCAND137 was transcriptionally activated by the invading fungi but not by the fungal elicitor, implying B. dothidea induced host miRcand137 as an infection strategy. We propose that the inefficient miRcand137 processing in M. hupehensis decreased pathogen-initiated miRcand137 accumulation, leading to higher resistance against B. dothidea.

A pathogen-induced miRNA compromises apple anti-fungal immunity and differentially accumulates in a wild species and cultivar, leading to the a difference in resistance.

Introduction

microRNAs (miRNAs) are 21- to 24-nucleotide (nt)-long small RNA (sRNA) molecules, mediating the posttranscriptional gene regulation by direct cleavage or translation inhibition of target mRNA (Jonas and Izaurralde, 2015), playing as master governor in plant life processes. The biogenesis of plant miRNA is multi-step, beginning with the long primary transcript (pri-miRNA) transcribed from the MIR loci, an independent transcription unit with its own promoter (Megraw et al., 2006). Being featured for pri-miRNAs, the imperfect stem–loop structure embedding the miRNA sequence is processed into a hairpin-like precursor and subsequently into the miRNA/miRNA* duplex by Dicer-like1 (DCL1; Kurihara and Watanabe, 2004) cooperating with Hyponastic leaves1 (HYL1) (Kurihara et al., 2006) and Serrate (SE) (Yang et al., 2006). The duplex is subsequently methylated by Hua enhancer1 methyltransferase, translocated into the cytoplasm via the HASTY pathway (Park et al., 2005), and unwound by sRNA degrading nuclease (Ramachandran and Chen, 2008). The released miRNA is loaded onto Argonaute1 (AGO1) to incorporate into the RNA-induced silencing complex (Voinnet, 2009).miRNA biosynthesis in response to environmental stimuli is under strict modulation at various levels, where biotic and abiotic stress-related cis-elements on MIR promoters (Megraw et al., 2006) are indispensable. For instance, the key seedling photomorphogenesis-involved transcription factor (TF) LONG HYPOCOTYL 5 binds to G/C boxes in the promoter of At-miR163, regulating Arabidopsis (Arabidopsis thaliana) primary root elongation of light-responsivity (Li et al., 2021). Sequence and functional analysis for the promoter of Md-miR285N revealed a heterogeneous network of gene regulatory elements, indicating that the expression of miR285N is regulated during various phases of the apple (Malus domestica) life cycle and in response to different stress conditions (Pompili et al., 2020). Another essential aspect of miRNA biogenesis is the precise manipulation of MIR transcripts to excise the short chain from the foldback (Moro et al., 2018). Changes in specific nucleotides, especially those located at unpaired positions, can enhance or abolish miRNA production (Rojas et al., 2020). Thus, miRNA expression is the cumulative outcome of MIR gene transcriptional regulation and the processing of miRNA precursor.

The interaction of plants and pathogens represents a dynamic competition between the robust plant immune system and efficient infection strategies of the phytopathogen. miRNA as the principal governor of gene expression is highly engaged in this competition, with multifaceted roles. Upon perceiving the invading pathogen, the host reprograms miRNA expression to induce multi-level defense responses by modulating specific target genes expression (Padmanabhan et al., 2009). Pathogens suppress plant defenses by delivering fungal miRNAs into the host cells, which hijack the host RNA interference pathway and silence resistance-related genes (Weiberg et al., 2013). Plants have also developed a miRNA delivery system to defend themselves by targeting key fungal genes (Cai et al., 2018). Previous studies revealed that some successful pathogens manipulate the expression of plant miRNAs to inactivate certain host genes to dismiss host immunity (Zhang et al., 2018). This strategy is associated with alterations in MIR gene transcription, wherein pathogen elicitors may play a role.

Ethylene response factors (ERFs) are important members of the plant-specific APETALA2 (AP2)/ ERF TF superfamily, broadly involving in various defense-related signal pathways. ERF proteins contain a conserved AP2 DNA-binding domain and specifically bind to cis-elements of GCC-box, DRE/CRT, or both (Cheng et al., 2013), which are frequently found in the promoter of defense-related genes especially PRs (Mcgrath et al., 2005). Infection-induced ERFs activate the transcription of PR genes, functioning as positive regulators of plant immunity (Tezuka, et al., 2019; Hawku et al., 2021). For apple, several ERF genes have been reported to be associated with defense responses against different fungal pathogens (Akagi et al., 2011; Gusberti et al., 2012). Particularly, a recent study demonstrated that MdERF11 activated the phytoimmunity and played a substantial role in B. dothidea resistance to apple (Wang et al., 2020). Nonetheless, the expression regulatory mechanisms underlying ERF genes during the apple–pathogens interaction have not been elucidated. A number of miRNAs were reported to target genes coding for TFs, facilitating a strategy of “regulating the regulators” to orchestrate the multiple gene expression in synchrony (Song et al. 2019). AP2/ERF family factors in several plant species have predicted characteristics for interacting with miRNA (Rakhmetullina et al., 2021), but the experimental verification was lacking.

Botryosphaeria dothidea is one of the most common and destructive fungal pathogens of woody plants, causing serious damage to fruit trees, especially to apple (Marsberg et al., 2017). The genome of B. dothidea strains originated from apple is enriched in genes encoding putative pathogenic proteins (Wang et al., 2018). The resistance of apple germplasms against B. dothidea infection varied greatly among different interspecies. However, the interactions between B. dothidea and the Malus spp. are not completely understood. Therefore, in-depth understanding of the defense mechanism is of great importance to prevent and control the damage caused by B. dothidea.

Malus hupehensis (Pamp.) Rehder is a Chinese wild apple species with high B. dothidea resistance. Several defense-related structural and regulatory genes have been demonstrated to be involved in its immunity against B. dothidea (Zhang et al., 2011, 2012). Few studies have examined the difference in the expression modulation of genes in M. hupehensis and susceptible apple cultivars such as M. domestica “Fuji” (hereafter, “Fuji”) during the infection.

In this study, we identify a B. dothidea-induced miRNA candidate (miRcand137) that was differentially expressed in the resistant M. hupehensis and the susceptible “Fuji.” By directing the silencing of ERF14 that codes for a transcription activator of several PR genes, miRcand137 interrupts apple anti-fungal defense, acting as a negative regulator of B. dothidea resistance. The MIRCAND137 gene is transcriptionally activated by the invading pathogen, representing a potential offensive strategy of the fungi. Furthermore, we elucidated that the differential accumulation of mature miRcand137 in M. hupehensis and “Fuji” was due to base variations on the precursor that affected the miRNA processing. Our results reveal a potential resistance mechanism of wild apple species interacting with B. dothidea.

Results

Identification of miRcand137 in apple

We previously conducted a sRNA deep sequencing for samples derived from mock-inoculated and B. dothidea-infected plants of the resistant M. hupehensis and the susceptible “Fuji” to investigate the involvement of miRNAs in apple immunity (Yu et al., 2014). Apart from members belonging to conserved plant miRNA families, we identified sRNAs responding to infection and being mapped to segments of the apple genome that may form hairpins, considering them as candidate miRNAs. Among miRNA candidates induced by B. dothidea, a 22-nt sequence with the highest abundance in infected plants was more abundant in the susceptible “Fuji” than in the resistant M. hupehensis (Supplemental Table S1). It was one of the sRNAs with the largest increase in expression in B. dothidea-inoculated “Fuji” compared to the control, but the expression change in M. hupehensis was much less.

When aligned with Malus expressed sequence tags (ESTs) and the Apple genome for the possible genomic locus, the sRNA was mapped to the first exon region of two highly conserved tobacco mosaic virus (TMV) resistance protein N-like pseudogenes in tandem on chr16 (LOC103403303 and LOC103431855). The segment embedding the 22-nt sequence being considered as the precursor had a typically hairpin-like predicted secondary structure without additional secondary branches or macrocyclic structures (Figure 1A). The sRNA originated from its 5′-stem region, and the only duplex contained five mismatched base pairs, two of which were nucleotides in asymmetric bulges. We named this miRNA candidate miRcand137, and the region in the two pseudogenes as the potential MIRCAND137 locus.

Figure 1.

Validation of miRcand137 and its MIR locus. A, Prediction of the transcription locus of miRcand137 in Apple Genome. R1 and R2 represent the first exon of LOC103403303 and LOC 103431855, respectively; P1 and P2 represent the 2,000-bp fragment upstream of R1 and R2. Predicted TSSs are indicated. B, The miRcand137 precursor transcription and the mature miRcand137 production in N. benthamiana transformed with predicted MIR fragments driven by the 35S promoter. 35SCaMV::GFP was taken as a control. C, The miRcand137 precursor transcription and the mature miRcand137 production in N. benthamiana transformed with predicted MIR fragments under the upstream regulatory sequence. NbActin was used as the loading control for transcripts and U6 for that of sRNA.

Driven by the 35SCaMV promoter, the segment of both LOC103403303 and LOC103431855 could be transcribed in Nicotiana benthamiana and produce mature miRcand137, the same as an artificial MIRCAND137 constructed using AtMIR319a backbone (aMIRCAND137 for amiRcand137) (Figure 1B). When taking their endogenous regulatory sequence up to 1,800-bp upstream of transcription start site (TSS) as the promoter, we detected a small amount of miRcand137 in N. benthamiana transformed with the segment of LOC103403303, but neither the mature form nor the precursor was detectable in that with LOC103431855 (Figure 1C). The region in LOC103403303 was confirmed to transcribe miRNA precursor under its own promoter and further generate mature miRNAcand137, being considered as the true MIRNACAND137 locus.

miRcand137 negatively contributes to apple resistance to B. dothidea

Upon B. dothidea inoculation, typical lesions and calculated colonization coefficients (ratio of pathogen and host biomass) on infected leaves demonstrated the appreciable fungi proliferation 24-h postinoculation (hpi) (Supplemental Figure S1). The delayed host symptom development and pathogen growth indicated M. hupehensis was indeed less susceptible than “Fuji.” sRNA gel blotting analysis substantiated that miRcand137 upregulated expressed in both M. hupehensis and “Fuji” 24 hpi, consistent with results acquired from the deep sequencing (Figure 2A). We noticed that mature miRcand137 accumulated at a higher level in “Fuji” than in M. hupehensis with a greater fold change between infected leaves and the mock-inoculated control, which was verified by the quantitative reverse transcription PCR (RT-qPCR) assay (Figure 2B).

Figure 2.

miRcand137 regulates apple sensitivity to B. dothidea infection. A, sRNA gel blotting showing the accumulation of mature miRcand137 in M. domestica “Fuji” and M. hupehensis under B. dothidea infection or mock inoculation (distilled water). B, Stem–loop RT-qPCR determining the abundance and changes of mature miRcand137 in “Fuji” and M. hupehensis inoculated with B. dothidea or the mock. C, Secondary structure of precursors for AtMIR319a skeleton and the modified aMIRCAND137. D, Schematic diagram of the TRV vector for miRcand137 overexpression (TRV: aMIRCAND137) and inhibition (TRV: STTMcand137). E, The abundance of miRcand137 in “Gala” of WT, TRV: 00, TRV: aMIRCAND137, and TRV: STTMcand137 determined by sRNA gel blotting. F, Typical leaves of different types of TRV-infiltrated “Gala” infected with B. dothidea. Photos were taken at 24 hpi. Scale bars, 1 cm. G, The percentages of leaves with different grades of disease progression from apple plants infiltrated with different TRV vectors upon B. dothidea infection. H, The calculated pathogen colonization coefficient (biomass ratio of fungus and host as determined by RT-qPCR) of leaves from different TRV-infiltrated “Gala.” Random 20 leaves from three biological replicates were sampled for each type of TRV vector. For the boxplot, centerline, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range. Leaves of apple plants were all collected at 24 hpi. U6 was used as an internal control for determining sRNA expression. Data shown in (B) and (G) are means ± sd (n = 3). Asterisks indicate significant differences from the WT or as indicated in the figure. **P ≤ 0.01; *P ≤ 0.05 (Student’s t test).

To clarify the role of miRcand137 in apple resistance, we introduced the AtMIR319a-skeletoned aMIRCAND137 into M. domestica “Gala” by tobacco rattle virus (TRV) system to overexpress (TRV: aMIRCAND137) (Figure 2, C and D) and introduced STTMcand137 for inhibiting endogenous miRcand137 (TRV: STTMcand137) (Figure 2D). Green fluorescent protein (GFP) fluorescence in freshly emerged leaves suggested the successful spread of the virus, validated by detected accumulation of TRV RNA (Supplemental Figure S2, A and B). TRV: aMIRCAND137 exhibited a higher miRcand137 abundance than the wild-type (WT) and plants infiltrated with empty TRV (TRV: 00), whereas the introduced STTMcand137 caused a decrease (Figure 2E).

We measured the disease progression of infected leaves from different TRV-infiltrated “Gala,” reflected by the disease severity grade based on the extension of necrotic lesions relative to the total leaf area. When challenged with B. dothidea, a larger percentage of leaves from TRV: aMIRCAND137 were of grade III or IV with severe symptoms, indicating a stronger sensitivity compared with that of WT or TRV: 00 (Figure 2, F and G). In contrast, most leaves from TRV: STTMcand137 were in grade I or II, showing a milder symptom development (Figure 2, F and G). Measured by the pathogen colonization coefficient, miRcand137 overexpression caused a relatively rapid B. dothidea proliferation, whereas the repression of endogenous miRcand137 restricted the fungal growth (Figure 2H). These results demonstrated that mature miRcand137 negatively regulated apple immunity, promoting host disease progression and facilitating the development of B. dothidea.

Base variations on the precursor affect the accumulation of mature miRcand137

We managed to amplify the miRcand137 foldback from cDNA of M. hupehensis and “Fuji,” further confirming the existence of miRcand137 transcripts in apple. Under B. dothidea infection, the miRcand137 transcript showed a significant upregulation as expected in both M. hupehensis and “Fuji” (Figure 3, A and B). Although mature miRcand137 was more induced in “Fuji” than in M. hupehensis during the infection, we observed no significant difference in the level or, the fold change for the accumulation of miRcand137 transcript between those two (Figure 3B).

Figure 3.

Differential bases in the miRcand137 foldback affect the processing of mature miRNA. A, The abundance of miRcand137 transcript was determined for “Fuji” and M. hupehensis inoculated with B. dothidea or distilled water (Mock) by northern blotting. EF-1α served as a loading control. B, RT-qPCR determining the transcription level and change of miRcand137 precursor in “Fuji” and M. hupehensis with or without B. dothidea infection. The expression levels were normalized against that of EF-1α. C, Predicted secondary structure of miRcand137 precursors of “Fuji” and M. hupehensis. Base variations at position +4 (T to C), position +46 (A to T), and position +89 (A to G) are indicated by arrows. Shannon’s entropy values calculated by RNAfold reflect the probability of the base-pair states and are color-coded. Positional entropy values (Bits) range from 0 (red) to 1.9 (purple). D, The schematic diagram of miRcand137 precusor. The base mutations in the foldback of MhMIRCAND137 are shown. E, Northern blot assay for mature miRcand137 and its transcript generated in N. benthamiana by MdMIRCAND137 and series of mutated MhMIRCAND137, along with GFP as a negative control. Actin and U6 were shown as a loading control for the transcript and sRNA, respectively. F, The abundance of miRcand137 in N. benthamiana expressing various versions of MIRCAND137 determined by RT-qPCR normalized with U6. Data are shown as means ± sd (n = 3). Asterisks indicate significant differences. **P ≤ 0.01; *P ≤ 0.05 (Student’s t test).

Three different bases were recognized in sequences of the foldback in MIRCAND137 (pre-miRcand137) obtained from M. hupehensis and “Fuji,” resulting in slight alteration in the folded hairpin structure outside the core region of mature miRNA (+4 T to C, +46 A to T, and +89 G to A) (Figure 3C). Intrigued that precursor structure influenced processing efficiency and accuracy of plant miRNAs (Cuperus et al., 2010a), we conjectured that the differential accumulation of mature miRcand137 in apple varieties was resulted from sequence variations in the MIRCAND137 foldback. To address this, MIRCAND137 derivatives with substitutions at either positions 4, 46, and/or 89 were constructed (Figure 3D) and transiently transformed into N. benthamiana. Compared to the WT of MIRCAND137 in M. hupehensis, the mutated version containing the substitution at position +46 yielded a significantly higher miRcand137 abundance (Figure 3, E and F). The corresponding result was obtained by mutating this base in the foldback of MdMIRCAND137 (Supplemental Figure S3). For position +89, base substitutions caused a moderately changed miRcand137 accumulation, and a single substitution at position +4 brought not any alteration ((Figure 3, E and F; Supplemental Figure S3). These results suggested that the A to T at +46 position in the precursor was responsible for the differential accumulation of mature miRcand137 between M. hupehensis and “Fuji.”

miRcand137 directs the regulation of ERF14

Two homologous ERF genes with a sequence identity of 87.81% registered in Genome Database for Rosaceae (GDR), MD08G1166100, and MD15G1352400, were predicted to be targeted by miRcand137 (Supplemental Figure S4A). As there was much lower expression level of MD15G1352400 in both apple varieties of either infection or control (Supplemental Figure S4B), we focused on MD08G1166100 (recognized as ERF14) in this study. Despite several mismatches found in the coding region, ERF14 of “Fuji” and M. hupehensis shared a same miRcand137 binding sequence (Supplemental Figure S5). Roughly contrary fluctuation patterns were observed when monitoring the expression of miRcand137 and ERF14 in the disease progression, where the early ERF14 induction was partially suppressed by the subsequently upregulated miRcand137 (Figure 4A). Corresponding to the lower miRcand137 level, ERF14 transcripts was more abundant in M. hupehensis than in “Fuji” throughout the infection. Two 3′ cleavage products of ERF14 transcripts were detected by 5′-RNA ligase-mediated-rapid amplification of cDNA ends (RLM-RACE) (Figure 4B). The major one (band a) indicated the cleavage directly mediated by miRcand137, and band b suggested an extra one downstream (Figure 4C).

Figure 4.

miRcand137 mediates the direct cleavage of ERF14 mRNA. A, The abundance of miRcand137 and ERF14 in “Fuji” and M. hupehensis during B. dothidea infection determined by RT-qPCR normalized by EF-1α and U6, respectively. Lines represent the relative expression level of mature miRcand137 and the bars indicate the transcription level of ERF14. B, 5′-RACE products obtained from the purified mRNA of “Fuji” and M. hupehensis in agarose gel. The band of “a” and “b” indicate two different products derived from different cleavage sites. M, DNA Marker DL2000. C, The structure diagram of ERF14 mRNA. Nucleotide sequences of miRcand137 TS in WT and mutated (m) ERF14 were aligned against miRcand137. D, Schematic diagram of constructs used in the co-expression assay. WT or mutated (m) ERF14 fragments from M. hupehensis or “Fuji” were fused with GUS to generate GUS-(wt or m) ERF14 reporters. Three tandem TSs completely base complementary pairing with miRcand137 were used as the positive control (3×TS), and three mutated TS (mTS) that did not bind with miRcand137 in tandem were used as the negative control. E, Co-expression analysis for miRcand137 and WT (or m) ERF14 fused with GUS reporter. The abundance of the recombinant mRNA and miRcand137 in transiently transformed N. benthamiana were detected by northern blot assay with probe derived from ERF14. NbActin and U6 were used as the loading control. F, Histochemical staining of GUS shows the expression of recombinant GUS in N. benthamiana co-expressing miRcand137. The schematic diagram of constructs are shown in (D). Images of stained leaves were digitally extracted for comparison. Scale bars, 1 cm. G, Quantification of GUS activity in co-transformed N. benthamiana. Different letters indicate significant difference P < 0.05 (one-way ANOVA followed by post hoc Tukey test). All data are shown as means ± sd (n = 3).

To verify the miRNA–target interaction we conducted a co-expressing assay in N. benthamiana using GUS fused with WT or mutant (m) ERF14 as reporters (Figure 4D). The mutation in mERF14 altered the miRcand137 binding site without changing the amino acid sequence (Figure 4C). The aMIRCAND137 was used to express miRcand137 for refraining from the influence of precursor variation. Using a probe spanning the junction of ERF14 and GUS, northern blot analysis revealed that reporters of wtERF14 decreased by co-expressed miRcand137; in contrast, those of mERF14 were not affected (Figure 4E). Similar results were obtained by determining the accumulation of recombinant GUS protein (Figure 4, F and G). Besides, data from GUS activity measurement demonstrated that miRcand137 silenced ERF14 of M. hupehensis and “Fuji” by a similar efficiency (Figure 4G). Thus, miRcand137 directed the cleavage of ERF14 via complementary base-pairing, and sequence variations outside the miRNA-binding site did not affect the cleavage ability.

Various factors, including 5′-nt identity, miRNA length, and the miRNA/miRNA* duplex structure, determine the sorting of miRNAs to AGO family members (Rogers and Chen, 2013). As miRcand137 is 22-nt long with U as the 5′-end, we investigated whether it depended on AGO1 to function, similar to miRNAs of this type in Arabidopsis (Mi et al., 2008; Mallory and Vaucheret, 2010). AGO1 homologs in N. benthamiana were silenced using TRV-mediated virus-induced gene silencing (VIGS, TRV: ΔNbAGO1), and those of AGO4 were silenced as the control (TRV: ΔNbAGO4; Supplemental Figure S6, A and B). ERF14 accumulation in AGO4-silenced seedlings was highly inhibited by co-expression, similar to that observed in TRV: 00 infiltrated with the empty TRV. To the opposite, exogenous ERF14 in TRV: NbAGO1 was no longer affected by miRcand137 co-expression (Supplemental Figure S6C). These results clearly indicated that miRcand137 silenced ERF14 depending on the participation of AGO1, but not AGO4.

For band b in the 5′-RLM-RACE assay, we performed an sRNA gel blotting using a probe derived from flanking of the cleavage site and detected signals in low molecular weight (LMW) RNA samples (Supplemental Figure S7A). Considering miRcand137 is 22 nt in length, we conjectured that sRNA possibly associated with this cleavage was phasiRNA derived from the cleavage products of ERF14 (Chen et al., 2010). From sRNA libraries of B. dothidea-infected apple samples, we identified specific 21-nt sequences that exactly matched ERF14 in sense or antisense, aligned from the miRcand137 target site (TS) to the 3′-end. The abundance of putative ERF14-derived phasiRNAs was higher in “Fuji” than that in M. hupehensis, corresponding to the accumulation level of miRcand137, which implied a relevance between phasiRNA production and miRcand137 (Supplemental Figure S7B; Supplemental Table S2). For further verification, a GFPcand137 sensor was constructed and co-expressed with miRcand137 in N. benthamiana (Figure 5A). As the essential role of RDR6 in the secondary siRNA biosynthesis (Cuperus et al., 2010b), we silenced NbSDE1, which encode a protein in N. benthamiana that is homologous to Arabidopsis RDR6 (AEE78550) with the highest sequence identity of 65.17%, by TRV-VIGS (TRV: ΔNbSDE1) (Supplemental Figure S7, C–E). The amount of full-length GFPcand137 transcripts in SDE1-silenced seedlings was greater than that in TRV: 00 upon the miRcand137 co-expression, indicating a more efficient attenuation of GFPcand137 in the presence of RDR6 homolog (Figure 5B). Correspondingly, GFP fluorescence generated by GFPcand137 in TRV: ΔNbSDE1 was only partially quenched upon miRcand137 co-expression (Figure 5C). Two fragments of different sizes were identified in the sRNA sample of TRV: 00 using a 3′-probe, whereas only the larger one was detected in that of SDE1-silenced (Figure 5B). Therefore, miRcand137 triggered the phasiRNA derived from the 3′-cleavage product in an RDR6-dependent manner, which was closely related to the completed target inhibition. Co-expression assay using a modified aMIRCAND137₂₁ to produce miRcand137 of 21 nt, which retained the ERF14-binding ability, suggested that the 22 nt form of miRcand137 was crucial for phasiRNA triggering (Supplemental Figure S7, F and H). The much lower target accumulation of 22-nt miRcand137 co-expression than that of the 21 nt one supported the critical role of triggered phasiRNA in efficient ERF14 silencing.

Figure 5.

The RDR6-dependent siRNA biosynthesis pathway contributed to the miRcand137-mediated silencing of ERF14. A, The schematic diagram of GFP-miRcand137 sensor (GFPcand137). The stop codon of GFP is underlined, and the adaptor is indicated in red. The TS of miRcand137 was aligned with the miRNA. B, The abundance of full-length GFPcand137 mRNA (F-GFPcand137), miRcand137, GFPcand137-derived sRNA in SDE1-silenced N. benthamiana and the TRV: 00 control upon GFPcand137 and miRcand137 co-expression. TRV-infiltrated plants transformed with none (N) or empty vector served as control. The full-length GFPcand137 mRNA was determined by a probe spanning the miRcand137TS. The location of probes for detecting sRNA generated by the miRcand137 cleavage products in LMW RNA are shown. NbActin and U6 was used as the loading control. C, Expression of GFPcand137 in the presence of miRcand137 in TRV-mediated SDE1-silenced N. benthamiana and TRV: 00 control as determined by GFP fluorescence under UV illumination. D, The location of PCR primers for amplifying apple RDR6 fragment used in TRV-VIGS of “Gala” is shown. E, sRNA gel blotting showing the abundance of two apple tasiRNAs derived from ARF and MYB, respectively, along with miRcand137, in TRV: ΔRDR6 (RDR6-silenced) and TRV: 00. U6 served as a loading control. F, 5′-RACE products obtained from the purified mRNA of TRV: ΔRDR6 and TRV: 00 were in agarose gel. The band of “a” and “b” indicate two products of different cleavage sites. M, DNA Marker DL500. G, The transcript levels of RDR6 and ERF14 in TRV: ΔRDR6 and TRV: 00 are determined by RT-qPCR with the normalization with EF-1α. Data are shown as means ± sd (n = 3). Asterisks indicate significant differences. **P ≤ 0.01; *P ≤ 0.05 (Student’s t test).

Next, we silenced RDR6 in “Gala” (TRV: ΔRDR6) to halt the secondary siRNA biosynthesis, which was confirmed by decreased expression of several tasiRNAs (Figure 5, D and E; Xia et al., 2012). The minor ERF14 cleavage product was not detectable in TRV: ΔRDR6, indicating its connection with the secondary siRNA pathway (Figure 5F). ERF14 expressed at a significantly higher level in TRV: ΔRDR6 than in TRV: 00 control, providing further evidence for the importance of phasiRNA triggering in miRcand137-directed ERF14 regulation (Figure 5G).

ERF14 induces apple anti-fungal defense and promoted B. dothidea resistance

ERF14 of M. hupehensis and “Fuji” encoded the same putative residue of 239 amino acid (aa) (Supplemental Figure S5), containing one and only one AP2 domain of 58 aa highly conserved among ERF14 protein of plant species (Figure 6A; Supplemental Figure S8A). The basic amino acid region (TSACKKKKFK) in the N-terminus as a potential nuclear localization signal indicated that ERF14 localized in the nucleus, as verified by subcellular localization analysis (Supplemental Figure S8B). The acidic C-terminus (pI 3.92) of even merely 18 aa could serve as the transcriptional activation region, demonstrated by the yeast self-activation assay using series of ERF14 deletion to determine the minimal activation domain (Supplemental Figure S8C). Results of the electrophoretic mobility shift assay (EMSA) demonstrated the dual-specific binding capability of ERF14 to GCC-box and DRE/CRT (Supplemental Figure S8D), which was further ascertained in eukaryotic cells using yeast one-hybrid (Y1H) assay (Supplemental Figure S8E). In the cross-competition, cold GCC-box was more competitive than cold DRE/CRT against both labeled probes, suggesting a stronger affinity of ERF14 to GCC-box than DRE/CRT (Figure 6B).

Figure 6.

ERF14 directly regulates the transcription of PR homologs and confers B. dothidea immunity to apple. A, The structure diagram of ERF14 protein. The conserved motif in the APETAL2 (AP2) DNA binding domain was analyzed for ERF14 from apple and various plant species using MEME (http://meme-suite.org/tools/meme). B, EMSA showing the DNA binding affinity of ERF14 to GCC-box and DRE/CRT. ERF14 was tagged with GST and prokaryotic expressed by E. coli strain BL21. Lane 1, free probe; Lane 2, free probe with GST protein; Lanes 3 to 10, recombinant GST-ERF14 protein. For competition assays, the GST-ERF14 protein was incubated with the indicated amount of either the cold DRE/CRT or GCC before the labeled probe being added. Sequence of the oligonucleotides used in the DNA-binding assay is shown. C, Schematic diagram of GCC-box and DRE/CRT cis-elements in the upstream regulatory region of apple PRs. D, Analysis for the activation ability of ERF14 to apple PR promoters by Y1H assay. The combination of p53His2 and pGADT7-Rec2-53 works as a positive control. E, GUS reporter system to test the transcriptional activation ability of ERF14wt and ERF14mut to PR promoters isolated from “Fuji.” The schematic diagrams of the WT and mutated DNA binding site in ERF14wt and ERF14mut proteins are shown. The activity of GUS was quantified by a fluorometric 4-methyl-lumbelliferyl-β-d-glucuronide method. GFP acts as a negative control. Data are shown as means ± sd (n = 3). Different letters labeled significant difference P < 0.05 (one-way ANOVA followed by post hoc Tukey test). F, RT-qPCR determining the transcript abundance of PR genes in WT, TRV: 00, TRV:ΔERF14, TRV: ERF14 and TRV: mERF14 “Gala.” The transcript level was normalized against that of EF-1α. Data are shown as means ± sd (n = 3). Asterisks indicate significant differences. **P ≤ 0.01; (Student’s t test). G, Typical leaves of WT, TRV: 00, TRV:ΔERF14, TRV: ERF14 and TRV: mERF14 “Gala” infected by B. dothidea. Photos were taken at 24 hpi. Images of infected leaves have been digitally extracted for comparison. Scale bars, 1 cm. H, The percentages of leaves with different grades of disease progression from apple plants infiltrated with different TRV vectors upon B. dothidea infection. Data are shown as means ± sd (n = 3). I, Calculated pathogen colonization coefficient of leaves from indicated TRV-infiltrated “Gala.” Random 20 leaves from three biological replicates were sampled for each TRV type. For the boxplot, centerline, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range. Asterisks indicate significant differences. **P ≤ 0.01; *P ≤ 0.05 (Student’s t test).

GCC-box existed in the promoter of PR homologs belonging to different families in many plant species (Mcgrath et al., 2005). We searched upstream regulatory regions of apple anti-fungal PR homologs and identified GCC-box or DRE/CRT in that of PR1 (MD13G1265800), PR2 (Glu, MD12G1089500), PR3 (Chitinase, MD01G1212900), PR4 (MD04G1225300), and PR10 (MD13G1161400) (Figure 6C; Supplemental Table S3). Confirmed by the Y1H, ERF14 could activate these PR promoters but not those being mutated at the ERF-binding motif or the PR5 (Thaumatin, MD09G1256000) promoter with no such motifs (Figure 6D; Supplemental Figure S8F). Moreover, upon the EREBP/AP2 motif mutation, ERF14 no longer activated these PR promoters, which was further supported by the N. benthamiana co-transformed assay (Figure 6E; Supplemental Figure S8, F and G).

Next, we overexpressed ERF14 in “Gala” by TRV-mediated virus-assisted gene expression (VAGE, TRV: ERF14), including mERF14 insensitive to miRcand137 to avoid the effect of endogenous miRcand137 (TRV: mERF14), and silenced ERF14 by TRV-VIGS (TRV: ΔERF14). (Supplemental Figure 2, C–G). TRV-mediated ERF14 overexpression/inhibition did not alter the miRcand137 accumulation (Supplemental Figure S2E). Interestingly, no differential ERF14 abundance was observed between TRV: ERF14 and TRV: mERF14 (Supplemental Figure S2D), which maybe because the strong virus-assisted target expression compensated the silencing effect of miRcand137 in the absence of infection. We assessed the expression of PR homologs in TRV-infiltrated “Gala,” finding them, corresponding to the ERF14 abundance, notably upregulated in TRV: ERF14 and TRV: mERF14 and downregulated in TRV: ΔERF14, especially that of PR1 (Figure 6F). These results together suggested that ERF14 induced the transcription of apple PRs by directly binding to their promoters.

To determine the role of ERF14 in apple immunity, TRV-infiltrated plants were then challenged with B. dothidea. Compared with the controls, plants overexpressing ERF14 exhibited milder symptoms with lower colonization coefficients, and changes for TRV: mERF14 were more pronounced than that of TRV: ERF14 (Figure 6, G and H). Rapidly developed lesions and higher pathogen colonization coefficient in TRV: ΔERF14 indicated an increased rate of disease progression caused by ERF14 silencing (Figure 6I). These results suggested a positive role of ERF14 for apple in resisting B. dothidea infection, mediating the anti-fungi defense by promoting PRs expression.

miRcand137 compromises apple immunity via modulating ERF14-mediated defense

To determine the impact of miRcand137 on ERF14 in inducing apple immunity, we tested whether miRcand137 suppresses the activation of ERF14 to the PR promoter. Co-expressed miRcand137, but not GFP or AtMIR319a, reduced the transcriptional activation of the PR1 promoter by ERF14 (Figure 7A). Checking the expression of PR homologs in TRV: aMIRCAND137 and TRV: STTMcand137, we found they were repressed by miRcand137 overexpression and induced by the inhibition (Figure 7B). These results imply an attenuation of miRcand137 on apple anti-fungal immune response induced by ERF14.

Figure 7.

miRcand137 modulates apple immunity against B. dothidea through repressing ERF14-mediated immune response. A, GUS reporter system to test the transcriptional activation ability of ERF14 to the PR1 promoter influenced by co-transformed aMIRCAND137. AtMIR319a and GFP act as the negative control. The activity of GUS was quantified by a fluorometric 4-methyl-lumbelliferyl-β-d-glucuronide method. B, RT-qPCR determining the transcript abundance of PR genes in plants of WT, TRV: 00, TRV: aMIRCAND137, and TRV: STTMcand137. C, Typical leaves of TRV: 00, TRV: ΔERF14, TRV: ERF14, and TRV: mERF14 “Gala” transiently expressing aMIRCAND137 or STTMcand137 infected by B. dothidea. Photos were taken at 24 hpi. Images of infected leaves have been digitally extracted for comparison. Scale bars, 1 cm. D, The percentages of leaves with different disease progression grades from “Gala” infiltrated with different TRV vectors transiently transformed with different constructs under B. dothidea infection. E, The calculated pathogen colonization coefficient of leaves from different TRV-infiltrated “Gala” expressing the indicated constructs. Random 20 leaves from three biological replicates for each TRV type each transient expression vector. F, The percentages of B. dothidea-infected leaves with different disease progression grades from “Fuji” transformed with STTMcand137 and M. hupehensis with aMIRCAND137. G, The relative expression level (in a Log₂ scale) of miRcand137 and ERF14 in leaves with different grades from “Fuji” expressing STTMcand137 or M. hupehensis expressing aMIRCAND137. Random 10 leaves for each grade, each transiently transformed apple species were detected. Images of typical leaves for different grades have been digitally extracted and shown below. Scale bars, 1 cm. GFP driven by 35SCaMV was served as a control for transient transformation, and the empty TRV carrying GFP was as that of TRV-infiltration (TRV: 00). U6 and EF-1α were used to normalize the expression level of sRNA and transcripts, respectively. Leaves of apple plants were all collected at 24 hpi. Data shown in the bar and column charts are means ± sd (n = 3). For boxplots, centerline, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range. Different letters labeled significantly difference P < 0.05 (one-way ANOVA followed by the post hoc Tukey test). Asterisks indicate significant differences. **P ≤ 0.01; *P ≤ 0.05 (Student’s t test).

Next, the miRcand137–ERF14 interaction during B. dothidea infection was investigated by expressing aMIRCAND137 or STTMcand137 in TRV: ERF14, TRV: mERF14, and TRV: ΔERF14. The overexpressed miRcand137 partially eliminated the enhanced resistance caused by ERF14 overexpression in TRV: ERF14, but did not affect that in TRV: mERF14 (Figure 7, C–E). When endogenous miRcand137 was suppressed, TRV: ERF14 and TRV: mERF14 exhibited similar resistance with no observable symptoms (Figure 7C). For TRV: ΔERF14, no difference in the lesion extension or B. dothidea colonization was observed upon the overexpression or the inhibition of miRcand137, meaning the exogenous miRcand137 did not cause a much heavier susceptibility, and endogenous miRcand137 suppression was inadequate in restoring the compromised resistance, implying that ERF14 silencing invalidated the effect of miRcand137 in B. dothidea sensitivity (Figure 7, C–E). These results suggest that miRcand137 interferes with apple immunity against B. dothidea through targeting ERF14.

To further elucidate the influence of phasiRNA triggering on miRcand137-interupped apple defense, we performed B. dothidea challenge for TRV: ΔRDR6 with overexpressed miRcand137 or inhibited ERF14. Unanticipatedly, RDR6 silencing boosted the disease progression (Supplemental Figure S9, A–C), which may be caused by the blocking of secondary siRNA biosynthesis that represses important siRNA-dependent defense pathways. Nevertheless, in TRV: RDR6, the sensitivity increase caused by miRcand137 overexpression was not as notable as that caused by ERF14 silencing, whereas TRV: 00 plants expressing aMIRCAND137 showed similar resistance to those in which endogenous ERF14 silenced (Supplemental Figure S9, D–F), suggesting that miRcand137-mediated compromise of apple immunity was amplified by RDR6-associated phasiRNA biosynthesis. To relate the differential accumulation of miRcand137 between M. hupehensis and “Fuji” to their B. dothidea resistance variation, we overexpressed miRcand137 in M. hupehensis and diluted it in “Fuji.” A higher proportion of leaves from miRcand137-overexpressing M. hupehensis exhibited severe symptoms (>grade III), whereas most of the leaves from miRcand137-suppressing “Fuji” exhibited milder ones (Figure 7F). On analyzing leaves of different disease progression, those of the higher grades showed greater miRcand137 abundance and those of the lower showed less (Figure 7G). These results support the oppositional correspondence of B. dothidea sensitivity and the miRcand137 expression level, which was the case in both the resistant and the susceptible apple.

MIRCAND137 was induced by live B. dothidea

Given the increased miRcand137 transcripts level, the upregulation of mature miRcand137 during the infection was at least partly because the transcriptional activation of the MIR gene. To investigate how the infection signal acts on MIRCAND137, we scanned its promoter for pathogen-associated cis-acting elements and recognized two BIHD1OS, three W-box, and one GT-1 (Figure 8A; Supplemental Figure S10; Supplemental Table S3). Malus hupehensis had one more W-box and “Fuji” had one more BIHD1OS. The promoter of MIRCAND137 was then tested for the reaction to B. dothidea. Under the control of MIRCAND137pro, the exogenous GUS in “Gala” was actively transcribed after the mycelial suspension treatment (Figure 8, B and C), concomitant with the increased endogenous miRcand137 transcripts expression (Figure 8D). Despite the extra W-box or the absent BIHD1OS, reporters driven by the promoter of M. hupehensis and “Fuji” shared a similar expression level and pattern (Figure 8, B and C), corresponding with the observation that miRcand137 transcription levels of “Fuji” and M. hupehensis showed no notable difference in the early infection.

Figure 8.

Botryosphaeria dothidea induced MIRCAND137 transcription through cis-elements in the promoter. A, Schematic diagram of pathogen-responsive cis-elements in the upstream regulatory region of MIRCAND137. B, GUS activity quantification for “Gala transiently expressed GUS reporter under the control of the MIRCAND137 promoter treated with B. dothidea mycelium suspension or the mock 12-h posttreatment. 35SCaMV-driven GUS served as a control. C, Time course analysis of transcript levels of GUS under the control of 35SCaMV, MdMIRCAND137pro, or MhMIRCAND137pro in transiently transformed “Gala” with mycelium suspension treatment by RT-qPCR. D, Time course analysis of transcript levels of mature miRcand137, miRcand137 transcript, and miR168 transcript driven by their self-promoter in “Gala” with mycelium suspension treatment by RT-qPCR. E, The schematic diagram of cis-elements deletions for the MIRCAND137 promoter is shown. F, Dual-LUC reporter system to test the transcriptional activation of LUC driven by indicated cis-elements-deleted MIRCAND137 promoter in “Gala” under mycelium suspension or mock treatment. The ratio of measured activities of firefly LUC and REN LUC was calculated as the final transcriptional activity. The 35SCaMV-driven LUC served as a control. G, The Schematic diagram showing the W-box element mutations for the MIRCAND137 promoter. H, The transcriptional activation activity of B. dothidea against LUC under the control of indicated W-box-deleted MIRCAND137 promoter for “Fuji.” The activities of firefly LUC and REN LUC were measured sequentially, and the LUC/REN ratio was calculated as the final transcriptional activity. U6 and EF-1α were used to normalize the expression level of sRNA and transcripts, respectively. All data shown are means ± sd (n = 3). Different letters labeled significantly difference P < 0.05 (one-way ANOVA followed by the post hoc Tukey test). Asterisks indicate significant differences. **P ≤ 0.01; *P ≤ 0.05 (Student’s t test).

A dual-luciferase (LUC) reporter system was applied to determine whether identified pathogen-associated cis-elements mediate the inducing of MIRCAND137. Deleting all W-box abolished the pathogen-induced activation of the MIRCAND137 promoter, in that the increase in relative firefly LUC activity was barely observed after mycelium suspension treatment (Figure 8, E and F). Deleting all two BIHD1OS or the only GT-1 has no such effect. Next, we tested which W-box took major responsibility. Mutating on the extra W-box in the promoter of M. hupehensis did not affect the pathogen-induced LUC expression, neither did the reverse mutation for the promoter of “Fuji” (Supplemental Figure S11, A and B). Regardless in the W-box shared by “Fuji” and M. hupehensis, single deletion for the first one (w-1) hardly affected the activation of MIRCAND137pro by B. dothidea, while the absent of the second (w-2) or third (w-3) one caused a significant reduction. w-23 retaining only the first W-box was no longer activated by the fungal stimulation; on the contrary, those merely retaining the second (w-13) or the third W-box (w-12) showed a moderate activation (Figure 8, G and H; Supplemental Figure S11C). Therefore, the two W-box near the TSS are required for B. dothidea to initiate the transcription of MIRCAND137 and may be partially functionally redundant.

W-box in promoters was implicated to confer defense-related genes of responsiveness to pathogen elicitors. We collected the cell wall extraction of B. dothidea to determine the response of the MIRCAND137 promoter to the elicitor (Supplemental Figure S11, D and E). Defense enhancer miR168, which was induced in apple infected by B. dothidea (Yu et al., 2017) and in Arabidopsis treated with Fusarium oxysporum elicitor (Baldrich et al., 2014), showed an upregulation in “Gala” as early as 1 h after the B. dothidea elicitor treatment (Supplemental Figure S11E). However, neither the MIRCAND137pro::GUS nor the transcript or the mature form of miRcand137 exhibited detectable expression fluctuation upon the elicitor treatment (Supplemental Figure S11, D and E), indicating that live pathogen but not the fungal elicitor triggered the MIRCAND137 transcription during the B. dothidea infection.

In summary, we found that apple miRcand137 is induced specifically by B. dothidea via W-box in the promoter region of MIRCAND137. By directly mediating mRNA cleavage and triggering phasiRNA, miRcand137 directs the regulation of ERF14, blocking its transcriptional activation of PR genes which promotes apple anti-fungi immunity, thereby negatively contributed to the resistance against B. dothidea. Bases variation in the miRcand137 precursor altering the folded hairpin structure affected the mature miRcand137 processing, resulting in differential miRcand137 accumulation in M. hupehensis and “Fuji” during the infection, further leading to the difference in B. dothidea sensitivity. Therefore, the miRcand137/ERF14 module may be a node for the infection strategy of B. dothidea and the immunity mechanism of apple.

Discussion

MiRNA-mediated gene silencing intricately influences plant–microbe interactions (Padmanabhan et al., 2009). As a part of host immunity or elicited by pathogens, miRNAs actively modulate plant defense responses, thereby promoting host resistance or facilitating pathogen invasion (Li et al., 2017). The differences in the expression of specific miRNA result in different disease susceptibility between varieties. In this study, we characterized a pathogen-induced miRNA candidate, namely miRcand137, targeting apple ERF14 to impede the anti-fungal defense, being a factor determining the interaction of apple and B. dothidea.

With the development of plant sRNA study, numbers of recently recognized miRNAs have been reported and declared species specific. The definition of confident annotations for plant miRNAs have been updated to minimize the false positive (Axtell and Meyers, 2018). miRcand137 characterized in this study derived from a typical hairpin-like precursor without secondary stems or large loops in the duplex (Axtell and Meyers, 2018), and was detectable in apple varieties upon B. dothidea infection and mock-treatment, meeting the criteria with evident expression and biogenesis (Taylor et al., 2017). The precursor transcription and mature miRNA generation in N. benthamiana transformed with DNA fragments containing the foldback further verified the existence of the MIRCAND137 locus.

The dominant target of miRcand137 include ERF14, encoding a transcription activator of the ERF family that participate in defense-related pathways. ERFs with transcriptional activation activity generally act positively on plant immunity while those with that of inhibitory act negatively (Liu et al., 2012). A single ERF gene could confer resistance to specific or broad-spectrum pathogens in plants (Cheng et al., 2013). By activating the promoter via ERF-binding motifs, ERF14 induced the expression of several anti-fungi PR homologs, facilitating apple resistance against B. dothidea. Here, we provide evidence that the negative contribution of miRcand137 to apple immunity relied on its silence on ERF14. miRcand137 partially eliminated the resistance strengthened by wtERF14, but not that by mERF14. Upon ERF14 silencing, the susceptibility of plants to B. dothidea was no longer influenced by altered miRcand137 expression, neither aggravated by exogenous miRcand137 expression nor reduced by the endogenous miRcand137 inhibition. The accumulation of ERF14-activited PRs changed in TRV-infiltrated plants with altered miRcand137/ERF14 expression, further relating the miRcand137-ERF14 module with PRs-related defenses.

Function of ERFs in plant immunity is complicated and includes the signal integration of reactive oxygen species (ROS) (Wang et al., 2013), mitogen-activated protein kinase (MAPK) cascade (Liu et al., 2017), and programmed cell death (Ogata et al., 2015). ERFs function in plant hormones crosstalk, serving as signaling hub of biotic and abiotic stress responses (Liu et al., 2018). The response pattern of host ERFs concerns the tropism of invading fungi (Berrocal‐Lobo et al., 2002). Regarding B. dothidea, MdERF11 was reported to interfere with the SA synthesis pathway, elevating immune responses of apple (Wang et al., 2020). Whether ERF14 also intervenes in other plant defense pathways directly or by interacting with other proteins remains unknown. Further study with more focus on ERF14 is therefore required.

We confirmed that miRcand137 was induced at the transcriptional level, supported by the increased transcripts level and the B. dothidea-activated promoter. W-box in the upstream regulatory region was determined to be in charge of the transcriptional activation of MIRCAND137. Being key to plant immunity, W-box was indicated to respond to fungal elicitors, inducing the expression of resistance-related genes such as Cytosolic Thioredoxin h5 (Laloi et al., 2004) and Chitins (Yamamoto et al., 2004). Intriguingly, Magnaporthe oryzae elicitor activated the promoter of OsMIR319 through W-box, induced the accumulation of resistance inhibitor miR319b in rice (Oryza sativa; Zhang et al., 2018). W-box exerts a dual role in plant–pathogen interactions. Despite the elicitor-responsiveness of W-box, the cell wall extraction of B. dothidea triggered the expression of miRcand137 at neither the transcript nor the mature miRNA level, consistent with the uninduced MIRCAND137pro-drived reporter. These observations indicate that live B. dothidea triggers the transcription of apple MIRCAND137 in an unknown indirect mechanism, which needs to be further elucidated. Botryosphaeria dothidea was revealed to also induce the expression of miR397, an resistance suppressor targeting LAC7 (Yu et al., 2020). Therefore, eliciting host miRNA to silence resistance-related genes may be an inherent strategy of this pathogen to escape plant defense.

Verified by the level of promoter activation and transcripts accumulation in early infection, the degree of B. dothidea-induced MIRCAND137 transcription was nearly the same among apple varieties. However, the upregulation of mature miRcand137 in M. hupehensis was not as notable as that in “Fuji.” Base variations in the precursor were validated to influence the efficiency of mature miRcand137 processing to varying degrees, resulting in different miRcand137 expression levels. Precursors of plant miRNAs are relatively highly diverse in sequences and sizes; as such, its impact on miRNA processing was initially accepted to be un remarkable (Axtell, 2008). Nevertheless, studies on Arabidopsis revealed that the region from the 15th nt at the end of miRNA/miRNA* duplex to the stem–loop junction is vital for accurate miRNA processing (Mateos et al., 2010, Song et al., 2010). Influencing the tail-loop structure, bases at the stem–tail junction interacted with the DCL1–HYL1–SE complex, responsible for positioning the first cut at the proximal duplex-end during miRNA processing (Cuperus et al., 2010a). For miRcand137, a single-nucleotide variation at position +46 could almost fully dictate the expression of mature miRNA. This position of the miRcand137 precursor is exactly where the paired and unpaired regions join, in where the base difference between M. hupehensis and “Fuji” caused alterations in the tail-loop.

Upon miRcand137 overexpressing in M. hupehensis or silencing in “Fuji,” remarkable changes in B. dothidea sensitivity were observed, indicating a similar mechanism of the miRcand137-ERF14 module in regulating the immunity of resistant and susceptible apples. The symptom severity of infected leaves was negatively associated with the abundance of miRcand137 and positively to that of ERF14. Given this, the pronounced resistance of M. hupehensis could be attributed to the inefficient accumulation of pathogen-induced miRcand137. These findings establish a miRcand137-ERF14 module interfering in apple immunity, deepen our understanding of the regulatory mechanism for miRNA expression in apple–fungus interactions, and provide insights into the molecular basis underlying the resistance of wild apple species against B. dothidea.

Materials and methods

Plant growth conditions

Malus hupehensis, as well as apple (M. domestica) cultivars “Fuji” and “Gala,” were cultured in vitro using Murashige and Skoog (MS) medium supplemented with 0.3 mg L⁻¹ indole-3-acetic acid (IAA), 0.2 mg L⁻¹ 6-benzyl amino purine (6-BA), and 0.1 mg L⁻¹ gibberellic acid at 25°C under 16-h light/8-h dark cycles. Half MS medium supplemented with 1.0 mg L⁻¹ IAA and 0.4 mg L⁻¹ indole-3-butyric acid (IBA) was used for rooting. Four-week-old, rooted plants were transplanted into a mixture of loam, vermiculite, and perlite (1:1:1) and cultured in a growth chamber under a 16-h light period at 22°C. The surviving plants were used for experiments 4 weeks after transplantation.

Nicotiana benthamiana seedlings were grown in a greenhouse at 24°C under conditions of 16-h light/8-h dark cycles and used four weeks after germination.

Pathogen culture and inoculation

Botryosphaeria dothidea strain LW48 was cultured on potato dextrose agar in the dark at 25°C, and induced to sporulate using UV irradiation. The conidial was suspended in distilled water and adjusted to the final concentration of 6 × 10⁻⁶ conidia mL^–1.

The third and fourth fully expanded apple leaves were inoculated with 10 mL of conidial suspension on either side of the main vein. Distilled water was used for mock treatment. Inoculated leaves were maintained at 100% RH and 25°C before sampled at different time points.

For fungal elicitor, the collected B. dothidea mycelia were sonicated at 100 W for 20 min and autoclaved at 121°C (15 psi) for 20 min. The cell wall extract was lyophilized and suspended in distilled water at a final concentration of 500 μg mL⁻¹.

Evaluation of disease resistance

The symptom severity of infected leaves was measured according to the scale of necrotic lesions, which was categorized into six levels by the relative area to the total leaf: I, <1%; II, 1%–10%; III, 11%–25%; IV, 26%–40%; V 5, 41%–65%; and VI, >66%. This grade was used to reflect the disease progression of each individual leaf (Yu et al., 2017). All experiments were performed using three replicates, with at least 20 leaves for each type of plants for each treatment in each replicate.

The colonization coefficient was calculated as the ratio of pathogen and host biomass (Yu et al., 2017), where 20 leaves were randomly sampled from three replicates and pooled.

Gene expression analysis

Total RNA was isolated using the RNAprep pure plant kit (Tiangen Biotech, Beijing, China). LMW RNA was extracted using a polyethylene glycol method (Wang et al., 2010). PrimeScript RT reagent kit with gDNA Eraser (Takara, Kyoto, Japan) was used for first-strand cDNA synthesis. For sRNAs, the provided oligonucleotide was replaced with stem–loop reverse transcription primers.

Total or LMW RNA transferred onto a Hybond-NX nylon membrane (GE Healthcare Life Sciences, Uppsala, Sweden) was hybridized with digoxigenin (DIG)-labeled nucleic acid probes. Signals were detected using the DIG Luminescent Detection Kit (Roche, Basel, Switzerland). For miRcand137 transcripts, a probe was designed at the region of sequence identity in the foldback of MIRCAND137 of “Fuji” and M. hupehensis. Probes for RNA gel blot assay are listed in Supplemental Table S4. EF-1α and Actin were used as loading controls for total RNA of apple and N. benthamiana, respectively, whereas U6 was for LMW RNA.

An ABI 7500 Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) with SYBR Premix Ex Taq reaction system (Takara, Kyoto, Japan) was used for RT-qPCR. Primers for stem–loop reverse transcription were used to determine sRNAs and U6 (internal control), pairing with gene-specific ones. The transcript level of MIRCAND137 and coding genes were normalized by that of EF-1α.

Gene cloning and DNA constructs

Apple genomic DNA (gDNA) was extracted using the Plant Genomic DNA Kit (Tiangen Biotech). DNA fragment harboring the miRcand137 foldback and its flanking regions was isolated as the MIRCAND137. ERF14 was amplified from apple cDNA based on sequence downloaded from the Genome Database for Rosaceae (https://www.rosaceae.org/).

The MIRCAND137 fragments isolated from apple gDNA were cloned under the 35SCaMV promoter to express miRcand137, and miR319/miR319* duplex in AtMIR319 was replaced with miRcand137/miRcand137* to generate aMIRCAND137 (Liang et al., 2012). A fragment consisting of two copies of partially complementary sequences of miRcand137 with a 48-bp short spacer was synthesized as STTMcand137 for miRcand137 inhibition. The ERF14 RNAi fragment harbored both sense and antisense sequences of a 25-bp ERF14 segment separated by a 20-bp GUS intron.

GeneArt Site-Directed Mutagenesis System kit (Invitrogen, Carlsbad, CA, USA) was used to introduce point mutations. Easygeno rapid recombinant clone kit (Tiangen Biotech) was employed for homologous recombination.

5′-RLM-RACE assay

5′-RLM-RACE was performed to determine the cleavage site of target using the FirstChoice RLM-RACE Kit (Ambion, Austin, TX, USA). The purified mRNA was ligated with a 5′-adaptor and amplified using random decamers to obtain 5′-RACE cDNA. Nested PCR was performed using adaptor-derived and gene-specific primers to obtain the 3′-cleavage products.

Nicotiana benthamiana co-expression assay

WT or mutated ERF14 fragments from M. hupehensis or “Fuji” were fused with GUS to generate GUS-(m)ERF14 reporters. Three in tandem miRcand137 target sequences or mutated ones were cloned into GUS before the stop codon to construct GUS-3×TS (positive control) and GUS-3×mTS (negative control), respectively. Nicotiana benthamiana transiently co-transformed with recombinant reporters and aMIRCAND137 were harvested to detect ERF14 level and GUS activity.

For analyzing miRcand137-triggered phasiRNA biosynthesis, the GFP-miRcand137 sensor (GFPcand137) was constructed by introducing miRcand137 TS downstream of the stop codon of GFP with an ATT adaptor as part of the 3′-UTR. A thymine (T) was inserted into miRcand137* in aMIRCAND137 to eliminate the structural asymmetry of miRcand137/miRcand137* duplex to result in aMIRCAND137₂₁.

Agrobacterium-mediated transient transformation

Agrobacterium tumefaciens EHA105 containing recombinant plasmids were pelleted, resuspended in a buffer containing 10-mM 2-(N-morpholino)ethanesulfonic acid (MES, pH 5.6) and 10-mM MgCl₂ to an OD₆₀₀ of 0.5. Acetosyringone (100 mM) was added before use. For vacuum infiltration, N. benthamiana seedlings or apple plants were inverted and immersed into the suspension and vacuumed using a vacuum dryer at 40 kPa. For injection–infiltration, the bacterial suspension was injected into leaves using a needleless syringe (200 μL once).

TRV-mediated VIGS and VAGE

For VIGS in N. benthamiana, partial fragments of NbPDS, NbAGO1-1, NbAGO1-2, NbAGO4-1, and NbAGO4-2 (Liu et al., 2014), and NbSDE1 were cloned into the pTRV2 vector. For apple, a pTRV2-GFP was obtain by constructed GFP-tagged coat protein fragments into pTRV2. aMIRCAND137 and STTMcand137 were then cloned into for miRcand137 overexpression or inhibition, respectively; and full-length or partial ERF14 were used to overexpress or silence the target (Spitzer-Rimon et al., 2013).

Suspension of A. tumefaciens containing pTRV1 or pTRV2 derivatives was mixed in a ratio of 1:1 (v/v). The third, fourth, and fifth leaves of N. benthamiana were injected with the mixture and cultivated for 14 d. Rooted apple plants were vacuum infiltrated before transplantation and illuminated with a long-wave ultraviolet lamp (Luyor-3260RB, Shanghai, China) after 4 weeks to check the spread of TRV. The freshly expanded leaves of the TRV-infiltrated plants were used to determine the efficiency of TRV-manipulated gene expression and for subsequent experiments.

GUS assay

GUS expression was determined using histochemical staining and GUS activity quantification. Leaves to be tested were incubated in a solution of 0.2-M NaH₂PO₄·2H₂O, 0.2-M Na₂HPO₄·12H₂O, 100-mM X-Gluc, and 10% (v/v) methyl alcohol and removed chlorophyll in a mixture of ethanol and glycerol (v/v; 3:1). A fluorometric 4-methylumbelliferyl-β-d-glucuronide method was used to measure GUS activity. Define one unit as 1 nM of 4-methylumbelliferon generated per minute per milligram of soluble protein. For each treatment, 10 leaves with each construct were pooled for detection.

Yeast assay

Series of deleted ERF14 were fused downstream of the GAL4 DNA-binding domain in the pGBKT7. The resulting constructs were transformed into Saccharomyces cerevisiae strain AH109. After selection on -Trp dropout medium, the transformants were transferred to -Trp, -His (DDO) medium containing 50-mM 3-AT, along with assays of colony lift filter and liquid culture using o-Nitrophenyl β-d-galactopyranoside as a substrate.

For the Y1H assay, five tandem GCC-box or DRE/CRT, or upstream regions of apple PRs were cloned upstream of His3 promoter in pHis2.1 to obtain reporters. ERF14 was cloned in-frame with the GAL4 activation domain of pGADT7-Rec2 to obtain the effecter. AH109 co-transformants were grown on -Trp, -Leu, -His (TDO) medium for analysis. The combination of p53His2 with pGADT7-Rec2-53 or pGADT7-Rec2-ERF14 served as positive and negative controls, respectively.

EMSA

ERF14 was cloned downstream of GST in pGEX-6P-1. The GST-tagged recombinant protein was induced to express in Escherichia coli BL21(DE3) and purified using affinity chromatography and a Glutathione Sepharose 4B Microspin column (Amersham, Arlington Heights, IL, USA). DIG–ddUTP-labeled double-stranded oligonucleotide of WT or mutated GCC-box and DRE/CRT were synthesized as probes.

Subcellular localization

Nicotiana benthamiana leaves infiltrated with 35SCaMV::ERF14-GFP were collected after incubation in the dark for 2 d to observe GFP fluorescence using a confocal microscope (Leica Microsystems, Wetzlar, Germany). Empty 35SCaMV::GFP vector served as the control. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). The lasers used for GFP and DAPI were 488 nm and 405 nm, respectively, at 18% intensity, with a collection bandwidth of 6 nm.

Functional analysis of the MIRCAND137 promoter

Region of 1,800-bp upstream of predicted TSS was isolated from apple gDNA as the MIRCAND137 promoter and sequenced. Putative cis-elements were screened by New PLACE (https://www.dna.affrc.go.jp/PLACE/?action=newplace).

“Gala” infiltrated with MIRCAND137pro::GUS were treated with the mycelial suspension or fungal elicitor of B. dothidea. The treatment of distilled water served as the control. At least eight leaves per treatment were sampled at each time point for determining the expression of GUS and miRNAs (precursor or mature). Samples of 12-h posttreatment were analyzed for GUS activity.

Series of mutated MIRCAND137pro were cloned into the pGreenII-0800-LUC vector to drive the LUC reporter. Upon treated with mycelial suspension for 12 h, 10 infiltrated “Gala” leaves were collected for each construct. The expression of LUC was determined using dual-LUC assay reagents (Promega, Madison, WI, USA) and normalized using 35SCaMV::REN (Hellens et al., 2005).

Bioinformatic analyses

The precursor of miRcand137 was folded by RNAfold (Yoshikawa et al., 2005) using the Vienna package (Allen et al., 2005) with the parameters -p -T 22 -d2. Shannon’s entropy at each position was calculated by summing entropy values for all probabilities using RNApdist.pl and color-coded using relplot.pl (Peragine et al., 2004). Targets of miRcand137 were predicted by psRNATarget (http://plantgrn.noble.org/psRNATarget/) with M. domestica CDS downloaded from GDR. The MEME suite program (http://meme-suite.org/tools/meme) was employed to identify conserved YRG and RAYD elements in the AP2 motif among ERF14 proteins.

Statistical analyses

All data were presented with means ± sd with the (n) of biological replicates as indicated in the legends. Data were compared using a two-tailed Student’s t test (∗P ≤ 0.05 and ∗∗P ≤ 0.01) or Tukey’s HSD (honestly significant difference) post hoc test, followed by one-way ANOVA where different letters denoted P < 0.05.

Accession numbers

Sequence data for genes used in this article can be found under accession numbers in GDR or NCBI: MD08G1166100 (MdERF14), MD15G1352400, MD13G1265800 (MdPR1), MD12G1089500 (MdPR2), MD01G1212900 (MdPR3), MD04G1225300 (MdPR4), MD09G1256000 (MdPR5), MD13G1161400 (MdPR10), MD15G1124800 (MdRDR6), MD04G1011000 (MdEF-1α), CN928184 (tasiARF), CN490819 (tasiMYB), EU549286 (AtMIR319a), DQ469932 (NbPDS), AY722008 (NbSDE1), DQ321488 (NbAGO1-1), DQ321489 (NbAGO1-2), DQ321490 (NbAGO4-1), DQ321491 (NbAGO4-2), and AY594294 (NbActin).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Malus hupehensis is resistant to B. dothidea infection and “Fuji” is susceptible (related to Figure 2).

Supplemental Figure S2. TRV-mediated VIGS and VAGE in “Gala” (related to Figures 2 and 6).

Supplemental Figure S3. Variations in the miRcand137 precursor of M. hupehensis debilitate the mature miRNA processing (related to Figure 3).

Supplemental Figure S4. Two genes coding for ERF protein are predicted miRcand137 targets (related to Figure 4).

Supplemental Figure S5. Alignment for the ERF14 coding region of “Fuji” and M. hupehensis.

Supplemental Figure S6. miRcand137 mediating the regulation of ERF14 required AGO1 (related to Figure 4).

Supplemental Figure S7. The complete suppression of ERF14 requires the 22-nt miRcand137 triggered phasiRNA (related to Figure 5).

Supplemental Figure S8. ERF14 regulates the expression of apple PRs via activating their promoters (related to Figure 6).

Supplemental Figure S9. miRcand137 needs RDR6-dependent secondary siRNA biosynthesis to impede apple defense against B. dothidea (related to Figure 7).

Supplemental Figure S10. Alignment for the MIRCAND137 promoter of “Fuji” and M. hupehensis.

Supplemental Figure S11. The promoter of MIRCAND137 is activated by live B. dothidea but not the fungal elicitor (related to Figure 8).

Supplemental Table S1. miRNA candidates responding to B. dothidea.

Supplemental Table S2. Mapped phased reads matching ERF14 in B. dothidea-infected apple samples.

Supplemental Table S3. Analysis for cis-elements in apple promoters.

Supplemental Table S4. Oligonucleotides used in this study.

 

X.Y. designed the experiments. X.Y., Y. H., and L.C. carried out most of the experiments. T.Z. and K.H. conducted the infection experiment and evaluated the plant disease resistance. J.C. contributed to plant materials cultivation and assisted with sample preparation. X.Y. performed data analysis and interpretation of results, and wrote the manuscript in consultation with S.W. S.Q. conceived the project and coordinated the research. All authors read and approved the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is Shenchun Qu (qscnj@njau.edu.cn).