BLAST and HMMER were used to identify PPO genes with conserved structures in banana. Musa acuminate amino acid sequences (MaPPOs) were extracted from the M. acuminate assembly (https://banana-genome-hub.southgreen.fr/). To examine the presence of the conserved domain, a batch search of the sequences for all obtained MaPPO genes was performed through the online databases of SMART (http://smart.embl.de/smart/set_mode.cgi?GENOMIC=1) and NCBI CDD (https://www.ncbi.nlm.nih.gov/cdd/). The MWs, PIs, and hydrophilia parameters were evaluated using an online tool on the ExPasy server (https://web.expasy.org/protparam/).

Protein sequences of PPOs from Malus domestica (MdPPO), Oryza sativa (OsPPO), Solanum lycopersicum (SlPPO), Solanum tuberosum (StPPO), Triticum aestivum (TaPPO), Hordeum vulgare (HvPPO), Fragaria ananassa (FaPPO), and Solanum melongena (SmPPO) were obtained from the Phytozome database (Goodstein et al., 2012). Multiple sequence alignment of MaPPO proteins was performed using the MUSCLE program in MEGA11 software, and the phylogenetic tree was constructed using the NJ method with the Poisson model, pairwise deletion, and 1000 repeats. The graph of the phylogenetic tree was modified using the online tool Evolview (https://www.evolgenius.info/evolview/).

The gene structure, including introns, coding sequences (CDSs), and untranslated regions (UTRs), of banana MaPPOs was derived from the banana genome (https://banana-genome-hub.southgreen.fr/node/50/7720969) and displayed using the Gene Structure Display Server (http://gsds.cbi.pku.edu.cn/). MEME (https://meme-suite.org/meme/tools/meme) and the NCBI Conserved Structural Domain Database CDD (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) were used to identify the conserved motifs and conserved structural domains of MaPPO proteins, respectively. The conserved motifs and structures of all PPO proteins in banana were displayed using TBtools (https://github.com/CJ-Chen/TBtools).

Mature green banana fruit (Musa acuminata AAA group, Cavendish subgroup) was obtained from a local market in Guangdong, China. Ethephon-induced ripening, 1-MCP-delayed ripening, and natural ripening of the fruit were evaluated as described previously (Li et al., 2020). Each treatment contained three biological replicates and was stored at 22°C with about 90% relative humidity until fully ripe. The fruit samples from various development stages were collected from local banana plantations in Guangzhou, China. The sampling time point was determined by the number of days after flowering. The fruit samples at each time point were collected and quickly frozen in liquid nitrogen and then stored at −80°C until utilization.

The fine powder of each sample ground in liquid nitrogen was used to isolate total RNA, as described in a previous report (Asif et al., 2000). The elimination of any potentially contaminated gDNA and reverse transcription of cDNA were performed using a reverse transcription kit (TaKaRa, Dalian, China) according to the manufacturer’s protocol. qRT-PCR and data analysis were conducted as per our previously reported method (Li et al., 2020), and MaCAC (clathrin adaptor complex medium subunit, HQ853240) was used as a reference gene (Chen et al., 2011). The primers with high amplification efficiency (90–110%) were designed using Beacon Designer 7, and these are listed in Supplementary File 2 . The qRT-PCR assay was conducted using the Applied Biosystems Q5 Real-Time PCR System (ThermoFisher, USA) using TB Green^® Premix kit (Tli RNaseH Plus, TAKARA, Dalian, China).

The expression data of MaPPO genes during fruit ripening were retrieved from our previous transcriptomic data on fruit ripening (Li et al., 2020), and proteomic analysis of the same samples was performed at the same time. The protein expression level of MaPPOs was obtained using proteome data. For analysis of expression patterns during fruit development, banana fruit was collected 15, 30, 45, 60, and 75 DAF (days after flower), and total RNA was extracted for sequencing analysis by Nuoji Biotechnology Company (China). The relative expression of MaPPO genes or proteins was displayed as a heat map drawn with TBtools (Chen et al., 2020). The raw reads of the transcriptome were deposited in the National Center for Biotechnology Information Sequence Reads Archive (SRA) under accession number PRJNA598018.

The CDSs of each MaPPO without the stop codon were amplified by PCR from cDNA and introduced into the pCambia1300-GFP vector (modified from pCambia1300). The expression of GFP and GFP-infused MaPPOs was driven by the cauliflower mosaic virus (CaMV)35S promoter. The fusion constructs and the control GFP vector were transiently expressed in tobacco leaf cells using the Agrobacterium-mediated method. The assay was performed as described previously (Sparkes et al., 2006). The ER tracker (Mravec et al., 2009) was co-expressed to indicate the localization of ER. GFP signals were examined with a fluorescence microscope (Zeiss LSM 710).

The CDSs of MaPPO genes were introduced into the E. coli expression vector pCZN1. Protein expression was performed under the following conditions: an initial culture (OD600 = 0.5), followed by addition of 0.2 mM IPTG, and incubation at 15°C overnight. Cell lysates were examined by immunity blots using an anti-His tag antibody (Tiangen, Beijing, China). The recombinant MaPPO proteins were purified on Ni-NTA His-Bind resin (Novagen, USA), and the protein concentration was assayed using an Easy Protein Quantitative Kit (Trans, Beijing, China). PPO activity was conducted as described in the previous report using a polyphenol oxidase activity assay kit (Beijing Solarbio Science & Technology Co., Ltd., Beijing)(https://doi.org/10.1007/s11947-018-2232-0). PPO activity was calculated based on the amount of recombinant protein or the fresh weight of the tissue sample in the reaction system.

The CDS of MaPPO gene were subcloned into pCMABIA1300 vector. The A. tumefaciens strain EHA105 including the constructed plasmids or control vector was injected into the mature-green banana fruits through the distal end using a syringe. The middle three hands from each banana bunch were used the experiment. At least eight fruit fingers were used for each treatment, and four ml A. tumefaciens was delivered into each fruit finger. One day after bacterium infiltration, injected fruits were dipped into ethephon solution (1000 times dilution from 40% ethephon) for 1 min, and stored at 22°C with 90% relative humidity for 5 d. Samples were harvested on Day 3 for examination of gene expression and PPO activity.

Through a genome-wide search in the genome database of M. acuminate (https://banana-genome-hub.southgreen.fr/tripal_megasearch), a total of 10 full-length MaPPOs were identified. To identify the candidate coding sequences (CDSs) of the 10 MaPPOs, the coding regions were amplified with PCR using a cDNA template. The final MaPPOs were referred to as MaPPO1-MaPPO10 according to their order on the chromosomes. The concrete sequences of each PPO gene in cultivated banana (Musa acuminata AAA group, Cavendish subgroup) were determined by PCR amplification from cDNA and sequencing, and the detailed sequences are listed in Supplementary File 1 . The characterization of these MaPPOs is presented in Table 1 . The deduced MaPPO proteins had amino acid numbers from 238 to 595, molecular weights (MWs) from 25.76 to 67.57 kDa, hydrophilia parameters from -0.572 to -0.157, and isoelectric points (PIs) from 5.87 to 9.09.

To investigate the evolutionary relationship of the PPO gene family, an unrooted neighbor-joining (NJ) tree was constructed with 65 PPO proteins from wheat, barley, rice, banana, apple, strawberry, tomato, eggplant, and potato. The homologs of MaPPO4 were not included in the analysis because the conserved domain of MaPPOs was incomplete. As shown in Figure 1 , the phylogenetic tree was characterized by five subgroups. The largest number of PPOs was observed in Group 2, with 26 PPOs, followed by Groups 5 (20) and 4 (9). Remarkably, MaPPOs were only distributed in Groups 1 and 5, and Group 1 had only 5 PPO proteins from banana and none from other examined plant species, suggesting that these PPOs might have been obtained in banana after their divergence or loss in other plant species. The remaining four PPO proteins in banana were concentrated in Group 5 and clustered in the same branch as other PPOs from monocots (wheat, barley, rice), indicating that they might have evolved from common ancestors. Interestingly, each group included PPO proteins from either monocots or dicots, and the PPO proteins from dicots were clustered into three individual groups (groups 2, 3, and 4), indicating that PPOs from monocots and dicots evolved independently from different lineages.

To gain more insights from the evolutionary relationship within the banana MaPPO gene family, a phylogenetic tree of 10 PPO proteins from banana was constructed using MEGA 11 using the NJ method with 1000 bootstrap replicates ( Figure 2A ). The motif architectures, conserved protein structures, and gene structures of 10 MaPPOs were examined within the phylogenetic context and visualized using TBtools. Seven out of 10 MaPPOs had one intron; one MaPPO had two introns, and two MaPPOs had no introns ( Figure 2B ). Interestingly, two MaPPOs (MaPPO3/7) had no untranslated regions (UTR) structure ( Figure 2B ). The phylogenetic tree of the MaPPO genes contained two subgroups. Generally, a gene within the same group had a similar and conserved structure in terms of exon number and intron length; for instance, all members in Group 2 had two exons and one intron. Based on searching the CDD and SMART databases, we identified three conserved domains (PPO1_KFDV, tyrosinase, and PPO1_DWL) in the MaPPO proteins and found them in all MaPPOs, except in MaPPO4, indicating their conserved biological functions ( Figure 2D ). Moreover, we identified 10 conserved motifs in the PPO protein sequence using MEME ( Figure 2C ), and five motifs (1, 2, 3, 5, and 8), one motif (6), and two motifs (4 and 9) were related to the N-terminal tyrosinase, intermediate PPO1_DWL, and C-terminal PPO1_KFDV domains, respectively. In addition, among most MaPPOs, two novel motifs (10 and 7) were identified at the N-terminal of the protein sequences.

To investigate the roles of MaPPOs, we examined the expression levels of nine MaPPO genes in roots, pseudostems, corms, young leaves, mature leaves, bracts, and mature green fruit of banana using qRT-PCR ( Figure 3 ). Three MaPPO genes (MaPPO3/8/9) were undetectable in all examined tissues due to their low transcript abundance (data not shown). MaPPO1 was preferentially expressed in mature green fruit, and MaPPO2 and MaPPO10 were predominantly expressed in bract tissue. MaPPO6 and MaPPO7 had similar expression patterns and were highly expressed in young leaves, whereas MaPPO5 was highly expressed in mature leaves ( Figure 3 ). Differential expression of various MaPPOs could imply their different functions in various plant tissues. To further understand the role of MaPPO genes in fruit tissue, the transcript levels of all MaPPOs were compared. The data revealed that MaPPO1 had the highest expression in fruit, followed by MaPPO6, MaPPO10, MaPPO7, and MaPPO5 ( Figure 4A ). Moreover, MaPPO1 and MaPPO6 had the highest protein expression levels in the proteome data of banana fruit ( Figure 4B ), suggesting that they might play an important role in fruit tissues.

To determine the role of MaPPOs in fruit development and ripening, the transcriptomic data of the MaPPOs were analyzed. The expression of four genes (MaPPO1/6/7/10) had a relatively high transcript abundance and varied with stage ( Figure 5 ), whereas MaPPO2/3/5/8/9 were almost not detected throughout fruit development (data not shown). qRT-PCR analysis indicated that the expression patterns of these MaPPOs were similar to those observed in the transcriptomic analysis ( Figure 5 ). At the early developmental stage of fruit, MaPPO6/7/10 exhibited a high expression level, which decreased gradually to a very low level at 75 days after flowering (DAF) ( Figure 5 ). The MaPPO1 transcript was present at a relatively high level at 15 DAF, which decreased at 30 DAF and then increased to the highest expression level at later stages of development ( Figure 5 ). During the postharvest ripening of banana fruit, transcriptomic data showed that MaPPO1/4/5/6/7/10 were detectable after ethephon treatment, and MaPPO1 presented the highest expression level among them, whereas the other four MaPPOs were almost undetectable ( Figure 6A ). To confirm the reliability of PPO gene expression in the transcriptome, the expression of the MaPPO genes was examined using qRT-PCR. MaPPO4 was not included in the following experiments due to its missing PPO-KFDV and PPO_DWL domains. A similar expression pattern of the MaPPO gene to that in the transcriptome data was observed during the ripening process ( Figure 6B ). In response to ethephon treatment, the expression of MaPPO1 increased to the highest level at day 3 and then decreased gradually; the other four detectable MaPPO genes showed relatively low abundance. MaPPO6 and MaPPO7 exhibited gradually decreased expression patterns. MaPPO5 presented a relatively higher expression in the fruit senescence stage, whereas MaPPO10 presented an expression trend of decreasing first and then increasing throughout the ripening process. Under natural ripening and 1-Methylcyclopropene (1-MCP) treatment, a trend similar to that of ethephon treatment was observed, and the expression of each MaPPO was delayed with the postponement of the fruit ripening process ( Figure 6 ). Interestingly, MaPPO1 and MaPPO6 exhibited higher transcript levels both in the early fruit development stage ( Figure 5A , 15 d) and the mature green stage ( Figure 4A ; Figure 6A , 0 d) and had higher protein expression levels in the mature green fruit stage ( Figure 4B , control), compared to other MaPPO genes.

To further investigate the role of MaPPOs in fruit development and ripening, four relatively high-expression MaPPO genes were selected to examine subcellular localization. The CDS of each PPO gene was infused to the C-terminus of GFP ( Figure 7 ). These GFP chimeras were transiently expressed in tomato leaf cells, and the localization of the expressed proteins was examined using confocal laser microscopy 48 h after Agrobacterium infection. The fluorescence signal of 35S::GFP was observed in the nucleus and cytoplasm ( Figure 7 ), whereas the signal of MaPPO1::GFP and MaPPO7::GFP overlapped with the auto-fluorescence of chloroplasts ( Figure 7A ), suggesting that MaPPO1 and MaPPO7 are chloroplast-localized proteins. The fluorescence signal of MaPPO6::GFP was observed both in chloroplasts and ER, and the signal of MaPPO10::GFP was co-localized with that of the endoplasmic reticulum (ER) marker, indicating that MaPPO6 was a chloroplast- and ER-localized protein and MaPPO10 was an ER-localized protein ( Figure 7 ).

To determine whether MaPPO genes encoded an active polyphenol oxidase enzyme, MaPPO1/6/7/10 were chosen for prokaryotic expression in Escherichia coli. Recombinant MaPPO proteins with a His-tag were isolated and purified. The molecular weight of these proteins was verified with Western blot analysis using an anti-His antibody ( Figure 8 ), and the molecular mass of recombinant PPOs was consistent with the predicted molecular mass, indicating successful expression in the prokaryotic expression system. The enzyme activity assay indicated that MaPPO1 had the highest activity among the examined MaPPO proteins, followed by MaPPO6, MaPPO7, and MaPPO10, and MaPPO1 yielded a PPO activity 7-fold higher than that of MaPPO10. The results indicate that MaPPO1 and MaPPO6 might contribute to the main PPO activity in banana fruit. To further examine the in vivo activity of MaPPO proteins, MaPPO1 and MaPPO6 were transiently overexpressed in banana fruit ( Figure 9 ). In the MaPPO-overexpressing fruit, the expression level of MaPPO1 and MaPPO6 was about 20 times and 35 times that in the control fruit, suggesting successful expression of target genes ( Figure 9B ). PPO activity analysis indicated that MaPPO1- and MaPPO6-overexpressing fruit had 6- and 5-fold higher PPO activity than the control fruit, implying that they are active polyphenol oxidase enzymes in banana ( Figure 9 ).