Identification and characterization of miRNA169 family members in banana (Musa acuminata L.) that respond to fusarium oxysporum f. sp. cubense infection in banana cultivars

Computational identification of miR169 family members

In our previous study, we identified miR169 family members in banana using small RNA sequencing (Song et al., 2016). The mature miR169 sequences of other plant species were obtained from miRbase database (Release 21.0, http://www.mirbase.org/). The BLAST parameters were the default routine settings of the database. The Latin name of miR169 members of plant species were got from the miRbase though the function of online database “Search by miRNA name or keyword.”

Prediction of miR169 target genes and functional analysis

The putative target genes for ma-miR169 were predicted using the plant miRNA target prediction online software psRNATarget (http://plantgrn.noble.org/psRNATarget/) with the default parameter settings. We selected the V2 Scoring Schema (2017 release), and applied Musa genomics (Musa acuminate V2, Banana Genome Hub, http://banana-genome-hub.southgreen.fr/home1). psRNATarget is a new web server designed to integrate and analyze reverse complementary matching between target transcripts and small RNAs. Another important function is the evaluation of target-site accessibility through calculation of the unpaired energy utilized to unfold the secondary structure around the miRNA target site in the mRNA. The following parameters were used: penalty for opening gap = 2, penalty for extending gap = 0.5, expectation = 10, penalty for GU pair = 1, penalty for other mismatches = 1, HSP size = 19 and seed region = 2–7 nucleotides. KEGG pathway analysis (Huang et al., 2018) was performed to analyze the metabolic pathways and functions of unigenes (http://www.genome.jp/kegg/).

Phylogenetic analysis of miR169 family members

Molecular evolutionary analyses were conducted using the MEGA 7.0 program. To evaluate the reliability of the different phylogenetic groups among miR169 family members, a phylogenetic tree was constructed using the maximum likelihood algorithm in the MEGA 7.0 program. Editing was performed in FigTree v1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/). The sequences were aligned using ClustalW. Based on the neighbor-joining method, the molecular evolutionary history was inferred using 1,000 bootstrap replicates to assess the robustness of the tree branches.

Plant material and growth conditions

Young banana seedlings of four cultivars, named Baxi banana (BX), HongYan banana (HY), FenJiao banana (FJ), and BaoDaojiao banana (BDJ), were obtained. BX is a triploid (AAA) cultivar with high yield, high quality, and the capacity for long-term storage. FJ is also a triploid cultivar, but with a different genotype (AAB), and it is characterized by good flavor, rapid ripening, and tolerance to abiotic stresses. BX and FJ re Fusarium wilt-susceptible cultivars that withers and die after infection. Other cultivars exhibit some resistance to Fusarium wilt, and the infection rate can be controlled at 5–8%. Both are widely cultivated in banana-planting areas. The seedlings used for RT-PCR were tissue culture seedlings that were not exposed to the environment, and the seedling height was approximately six to eight cm. qRT-PCR seedlings needed to be transplanted, and the seedling height was approximately 25–30 cm.

Pathogen infection and microscopy analysis

All banana species were cultured and propagated separately. Foc4 was confirmed to efficiently infect banana plants and induce disease symptoms. At the seedling stage, the hydroponic suspension was replaced with a Foc4 spore suspension (1.5 × 10⁶ conidia/mL). The pseudostem and roots of the banana were sampled and stored at −70 °C, and part of the postinfection banana was transverse and longitudinally cut for microscopic examination. Macroelements and microelements were supplied throughout the growing stage to maintain growth. The seedlings were grown under a long photoperiod (16:8 h/light:dark).

Primer design, cDNA synthesis and qRT-PCR analysis

In our previous study, ma-miR169a and ma-miR169b displayed high expression according to the sequencing results. We used the reads per kilobases million values (Additional File 8) to build a heatmap (Becker, Chambers & Wilks, 1988). First, we identified the expression of ma-miR169a, ma-miR169b and their targets. For this purpose, RT-PCR and qRT-PCR primers were designed according to the predicted miRNA and target sequences (Additional File 6). The small RNA samples were converted to miR-cDNA using an RT primer pool with reverse transcriptase. A specific primer pair was designed for each miRNA, after which PCR amplification using a SYBR Premix Ex Taq™ Kit (Takara, Dalian, China) was carried out with a Rotor-Gene 6000 machine (Corbett Robotics, Brisbane, QLD, Australia). Quantification of the relative expression of miRNAs was performed using the ^ΔΔCT method, and the U6 gene was used as a control (Song et al., 2018). Quantification of the target was also carried out through qRT-PCR using the actin gene as a control. The primers of the U6 and actin were listed in Additional File 6. The forward and reverse primer sequences flanked the binding region in which the miRNA interacted with its target mRNA.

Statistical analysis

The quantitative expression data were analyzed statistically using one-way ANOVA followed by post hoc Tukey’s HSD (honest significant difference) tests. The p-value < 0.01 was considered statistically significant.

Identification and genomic distribution of ma-miR169 family members in banana

Based on our previous miRNA transcriptomic data (Song et al., 2016), a total of 18 ma-miR169 members were identified from the banana genome. The sequences of the ma-miR169 family were 18–22 nucleotides in length, and the sequence similarity reached 73.67% (Fig. 1A). Evolutionary analysis clustered the ma-miR169 family members into two groups (Fig. 1B). To identify other various plant ma-miR169 members, a genome-wide search was carried out using the miRNAs database, miRbase (http://www.mirbase.org/cgi-bin/browse.pl). A total of 209 mature miR169 sequences were predicted in 28 plant species (Additional File 1). The numbers of miR169 family members differed greatly among the various species. Compared with the number of miR169 members in other plant species, including 11 in Arabidopsis thaliana, three in Oryza sativa, eight in Zea mays, and 17 in Glycine max, the number in banana, 18 was the highest (Fig. 2). The sequences of the miR169 family members were 18–23 nucleotides in length, and the sequence similarity reached 68.85% (Additional File 2), suggesting that the mature sequences of these miR169 members were highly conserved.

Prediction of ma-miR169 targets

Based on the 18 mature ma-miR169 sequences in banana, we used all the banana gene coding sequences to predict the ma-miR169 targets with the psRNATarget online software (Li et al., 2017). A total of 25 targets were obtained, and the number of targets differed greatly for the different miR169 members. Compared with other miR169 members, ma-miR169h, ma-miR169f, ma-miR169c, ma-miR169d, ma-miR169i, and ma-miR169k had more targets, with 3, 3, 2, 2, and 2 targets, respectively (Additional File 3). In addition, we found a complicated relationship between miRNAs and their targets. ma-miR169b, ma-miR169c, ma-miR169e, ma-miR169g, ma-miR169i, ma-miR169j, ma-miR169k, ma-miR169l, ma-miR169m, ma-miR169n, ma-miR169p, and ma-miR169q had the same target (GSMUA_Achr8T16690_001); ma-miR169c and ma-miR169k had the same target (GSMUA_Achr11T16520_001); and ma-miR169h and ma-miR169f shared three targets (GSMUA_Achr10T02910_001, GSMUA_Achr4T09610_001, and GSMUA_Achr8T20870_001). The targets of ma-miR169a (GSMUA_Achr4T12800_001) and ma-miR169o (GSMUA_Achr2T19800_001) were unique and were not common to other ma-miR169 members; a similar situation was observed for ma-miR169d (GSMUA_Achr8T24960_001 and Achr9T12370_001) and ma-miR169i (GSMUA_Achr1T03730_001 and Achr8T16690_001), which each had two targets.

Functional annotation of target genes

Within the molecular function category, a large number of target genes were assigned to transcription factors, cellular processes, binding, metabolic processes, and organelles. To understand the biological function of these target genes, we got the metabolic pathway of those targets using the KEGG database (KEGG PATHWAY Database, 2017; http://www.kegg.jp/kegg/). Five target genes (GSMUA_Achr1T03730_001, GSMUA_Achr2T19800_001, GSMUA_Achr4T12800_001, GSMUA_Achr8T20870_001, and GSMUA_Achr11T16520_001) were marked in five metabolic pathways. These pathways included glycine, serine and threonine metabolism; pentose and glucuronate interconversion; glycerolipid metabolism; amino sugar and nucleotide sugar metabolism; and starch and sucrose metabolism (Additional Files 4 and 5).

Expression profiles of ma-miR169 members in two banana varieties

To investigate the expression patterns of the ma-miR169 family members, roots of HG and BX were collected for transcriptome sequencing (Fig. 3). Generally, when the value of FPKM (Reads per Kilo bases per Million reads) is more than 20, it is considered to be high expression level (Song et al., 2016; Chen et al., 2015), the FPKM of ma-miR169a and ma-miR169b were 68.41 and 36.17, respectively in HG, showing a high expression levels, In contrast, the other 16 ma-miR169 members FPKM values were between 0 and 12.32, exhibited low expression levels. Thus, ma-miR169a and ma-miR169b may be important candidates for the next research.

The involvement of ma-miR169a and ma-miR169b in the banana response to Foc4

The most typical symptoms of the banana infected by Foc4 were brown spots or stains on the roots and protocorms. At 30 days after Foc4 inoculation, no or slight brown staining in the roots and protocorms of HY (Figs. 4A and 4J) and BDJ (Figs. 4B, 4C, 4K and 4L) were observed, whereas there were obvious brown stains in the roots and protocorms of FJ and BX (Figs. 4D and 4G). These results indicate that HY and BDJ are more resistant to Foc4 than FJ and BX.

Because of the obvious phenotypic differences, this set of samples was also used for the expression analysis of ma-miR169a and ma-miR169b. We synthesized reverse transcription primers and qRT-PCR primers (Additional File 7), and cloned those miRNAs into HG and BX, and confirmed the success of the ma-miR169a and ma-miR169b cloning through sequencing (Fig. 5). The expression levels of ma-miR169a and ma-miR169b were significantly increased in the roots of HY and BDJ by Foc4 treatment comparing with controls and their increased fold were approximately 3–43, in contrast, ma-miR169a and ma-miR169b showed low expression or repression in the roots of FJ and BX after Foc4 treatment. Further experiments indicate that there were negative relationships between ma-miR169a, ma-miR169b and their targets basing on their expression levels to Foc4 infection in the roots of HY and BDJ, but they were low expressed in FJ and BX (Fig. 6; Additional File 8). Together, it suggested that the expression characteristics of ma-miR169a and ma-miR169b were consistent with the resistance to Foc4 in banana cultivars.