Banana fruits (Musa spp., AAA group, cv. Brazil) at physiological mature stage with green were harvested from an orchard in Guangzhou, Guangdong Province, P.R. China (23°30′N, 113°30′E). Harvested fruits were packed with polypropylene plastic bags (film thickness: 32 μm) and immediately transported to the laboratory within 2 h. Fruits were selected for uniformity of shape, color, and size, and rinsed the surface gently. Then fruits were placed into two groups randomly and packed in polyethylene bags (200 mm × 50 mm, 0.03 mm thickness, and three fruit per bag) and stored at 4 and 25°C, respectively. Six bags in total of 18 fingers from each treatment were randomly selected at 0, 2, 4, and 6 days during storage, and the chilling index and hue angle were evaluated. To isolate cuticular membranes, chemical reagents of analytical grade were prepared as previously described (Huang and Jiang, 2019). The peel tissues were sliced, frozen in liquid N₂, and stored at − 80°C for further analysis.

The symptom of chilling injury was estimated by the extent of browning surface with different scales: “1” smooth surface without browning point; “2” slight browning with 1/4 of the fruit surface; “3” 1/4–1/2 browning area; “4” 1/2–3/4 browning area; and “5” > 3/4 browning area. The chilling injury index was calculated as: CI (chilling injury scale × corresponding fruit fingers within each class)/(number of total fruit × the highest scale) × 100.

The peel color degree was determined using a Minolta Chroma Meter CR-400 (Minolta Camera Co., Ltd., Osaka, Japan). Nine fruit fingers from each treatment at each stored stage and three points around the middle of each fruit were measured. Color was recorded as parameters of CIE L*, a*, and b* scales. In which, L indicates the lightness or darkness, a* indicates green to red, and b* donates blue to yellow. The overall color index was indicated by L* or general value of hue angle [h° = tan^–1(b*/a*)].

The surface characteristics of each organ of the two kinds of vegetables were observed using scanning electron microscopy (SEM). Briefly, small pieces (3 mm × 3 mm) of the peel tissues were picked and fixed in 2.5% glutaraldehyde in 0.2 M sodium phosphate buffer, pH 7.2–7.4 at 4°C overnight. The fixed samples were vacuum-infiltrated in the material on ice for 1 h. The samples were washed using sodium phosphate buffer and then dehydrated in a graded alcohol series with 30, 50, 75, 90, and 100% three times. After the samples were dried completely, the small pieces were mounted on aluminum stubs using a conductive double-sided adhesive tape prior to observation. The dried samples were coated with gold:palladium (60:40) at 30 mA using a LEICA EM ACE600 (Leica Microsystems, Wetziar, Germany) sputter coater, depositing an alloy of approximately 10 nm thickness. The characteristics of the sample surfaces were examined under a field emission SEM (JEOL JSM-6360LV, Tokyo, Japan) at 15 kV accelerating voltage.

The cuticles of banana fruit were isolated enzymatically following the protocol described by Huang and Jiang (2019). Fruit samples from each group at each stage were randomly selected. Peel discs were obtained by a puncher 1.2 cm in diameter near the middle position of the fruit. The punched discs were soaked in 10 mM citric acid buffer containing 1% (w/v) pectinase and 1% (w/v) cellulase (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China). The cuticular membranes isolated from peel tissues were then washed by 10 mM sodium tetraborate decahydrate and distilled water in series. After that, cuticles were air-dried for further cuticular chemical analysis.

The cuticular waxes and cutin monomers of banana fruit were extracted using the isolated cuticular membranes following the reported procedure previously (Huang, 2017). The dry and fine intact cuticular membranes isolated from banana fruit were immersed in chloroform completely. To better release the soluble waxes, the extraction was set with a moderate temperature at 40°C. The extraction for each sample was repeated three times, and each time for 2 min. After combining the extracted solution, n-tetracosane (Sigma-Aldrich, Shanghai, China) as an internal standard was added to help detect the accumulation of cuticular waxes. Gentle stream of nitrogen gas was used to dry the extracts for further analysis.

The above matrix membrane after removing waxes was subsequently depolymerized in boron trifluoride with methanol (BF₃-methanol, 10%, ∼1.3 M) overnight at 70°C. Then, n-dotriacontane (Sigma-Aldrich, Shanghai, China) was added as an internal standard. Saturated aqueous sodium chloride solution and chloroform were added subsequently to extract the cutin monomers. The organic phase containing cutin monomers was collected and evaporated to dryness under a gentle stream of nitrogen gas for further analysis.

Prior to the analysis of the chemical composition of cuticular waxes and cutin monomers, the dry extracts prepared in the aforementioned manner were derivatized with pyridine (Shanghai Aladdin Bio-Chem Technology Co., Ltd., Shanghai, China) and N,O-bis (trimethylsilyl)trifluoroacetamide (BSTFA; Sigma-Aldrich, Shanghai, China) for 30 min at 70°C. To quantify wax and cutin monomer components, the extracts were analyzed using a capillary gas chromatography instrument equipped with a flame ionization detector (7820A, GC System; Agilent Technologies, Santa Clara, CA, United States), and on-column injection with a capillary column (30 m × 0.32 mm, DB-1 ms, 0.1 μm film; J&W Scientific, Agilent Technologies, Santa Clara, CA, United States). To separate cuticular wax compounds, 10 μL samples were injected at 50°C; after 2 min at 50°C, the temperature was raised to 200°C at 40°C min^–1, held for 2 min at 200°C, and then raised to 320°C at 3°C min^–1 and held for 30 min at 320°C. For separation of the cutin monomers, 10 μl samples were injected at 50°C; following 1 min at 50°C, the temperature was raised to 150°C at 10°C min^–1, held for 2 min at 150°C, and then raised to 320°C at 3°C min^–1 and held for 30 min at 320°C. The area under the peaks was compared with that of the internal standards to obtain the quantity of cuticular wax and cutin monomer components.

The chemical components were analyzed using a temperature-controlled capillary gas chromatography instrument equipped with a mass spectrometric detector (m/z 50–750, MSD 5975; Agilent Technologies, Santa Clara, CA, United States). Single compounds were identified based on their electron ionization mass spectra using authentic standards, the Wiley 10^th/NIST 2014 mass spectral library (W10N14; John Wiley & Sons), or by interpretation of the spectra according to the retention times and/or by comparison with data from the literature (Holloway, 1982; Jetter et al., 2008) or online database (LipidWeb)¹. The coverage was quantified against the amount of the internal standard, and the average chain length of the VLC acyclic compounds was calculated following the method reported by Huang (2017).

Three biological replicates were performed for each sample. Total RNA was extracted and removed the contaminations with DNase I. Then, the purity and concentration of the RNA were checked by multifunction microwell plate detector (Shanghai Dobio Biology Technology Inc., Spark^®, Tecan, Shanghai, China) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States). Then the cDNA was synthesized by mixing with DNA polymerase I and RNase H. The samples were amplified and sequenced on Illumina HiSeq™ 4000 by Gene Denovo Co. (Guangzhou, China). The fine data were mapped to the reference genome using published banana data. The gene expression abundance was represented by an RPKM value. Differentially expressed genes (DEGs) were identified by the DESeq2², which indicated the change of false discovery rate genes over twofold with statistical significance (false discovery rate, FDR < 0.05). DEGs in the pointed pathways were analyzed in detail.

The total RNA of banana peel tissues using a HiPure Plant RNA Mini Kit (Magen, Guangzhou, China) and cleaned with DNase I (TaKaRa, Otsu, Japan). The DNA-free RNA was used as the template for reverse-transcription PCR (RT-PCR). The first-strand cDNA template of each gene was synthesized using PrimeScript^® RT Master Mix (Perfect Real Time) template kit (DRR036A; TaKaRa, Otsu, Japan), and their concentrations were determined using a multifunction microwell plate detector (Shanghai Dobio Biology Technology Inc., Spark^®, Tecan, Shanghai, China). Relative expression levels of the sequenced DEGs related to fatty acid elongation (MaLACS, MaKCS, MaKCR, and MaECR) and wax biosynthesis (MaCER1 and MaFAR) were carried out by qRT-PCR. The reaction system was SYBR Premix Ex Taq Kit (DRR420A; TaKaRa, Otsu, Japan) where PCR conditions for the primers were 95°C for 30 s, 95°C for 5 s, and 60°C for 34 s, 40 cycles. The Primer Premier 6.0 software (Premier, Canada) was used to design the primers (Supplementary Table 1). The qRT-PCR was performed with SYBR^® Premix Ex Taq™ II (Takara, Otsu, Japan) in an ABI7500 Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, United States). Three independent biological replicates were used in the analysis.

Statistical analyses were performed using the IBM SPSS (version 23, IBM Corp., Armonk, NY, United States) and SigmaPlot 12.5 (Systat Software, Inc., San Jose, CA, United States). Comparison analyses were performed by one-way ANOVA, and significant differences were analyzed at a level of 0.05. SigmaPlot 12.5 was used to elaborate the graphs shown in the figures.

Banana fruit exhibited an obvious browning appearance as the chilling injury symptom with an index over 80% of surface-browned at 6 days under 4°C after harvest (Figures 1A,B). The hue angle of fruit peel decreased rapidly from 120° to less than 100° following the chilling injury process, inducing the discoloration of green to brown at 4°C during storage (Figure 1C). In contrast, banana fruit stored at 25°C as control, the green peel appearance showed no obvious changes with trace of browning pot with most of them for mechanical damages during treatments after harvest (Figures 1A,B). The hue angle values of fruit exhibited relatively stable around 120° at 6 days stored at 25°C (Figure 1C). The green mature banana fruit changed slowly without ripening changes in the early storage period under room temperature, whereas suffered chilling injury rapidly as browning peel appearance under low-temperature storage (Huang et al., 2016).

The SEM observation showed the morphological changes of banana peel surface stored at 6 days under different temperatures. The epidermis of banana fruit consists of small square-, rectangular-, and gourd-shaped epidermal cells. These epidermal cells were outward forming numerous plateau ridges (Figure 2A). Similar to the previous anatomical analysis, the epidermal cells covered a thin layer of cell wall and cuticle on the outer surface of banana fruit (Amnuaysin et al., 2012). The outermost surface accumulated sparse small wax crystals (Figure 2C). After 6 days, the green mature banana fruit stored at 25°C showed clear cell shapes with crystals, indicating the well status for the morphological appearance in fruit (Figures 2A,C). However, the appearance of morphology was observed to be modified by the chilling injury under low temperature. Followed by the development of chilling-browning symptoms, the protruding epidermal cells collapsed and the surface of the fruit became flat (Figures 2B,D).

The main pattern of cuticular components of cuticular waxes and cutin monomers was detected and quantitated thoroughly. The overall accumulation of cutin monomers was detected as 23.1 μg cm^–2 in green mature fruit stored at 25°C, which was shifted to increase significantly (p < 0.05) as 31.2 μg cm^–2 by chilling injury after 6 days (Table 1). Similarly, the total wax coverage was stimulated to increase from 30.0 to 36.6 μg cm^–2 at 6 days under 25 and 4°C, respectively (Table 1). As a result, the ratio of total wax over cutin monomers showed no significant changes, which were about 1.2–1.3 (Table 1).

The cutin monomers in the series detected were prominently mid-chain epoxy ω-hydroxy fatty acids followed by fatty acids, primary alcohols, ω-mid-dihydroxy fatty acids, and 2-hydroxy fatty acids in banana fruit cuticles (Figure 3). The most abundant chain length of monomers was C₁₆ and C₁₈ fatty acids without or with branched hydroxyl or epoxy group. The predominant C₁₈ monomer was 9,10-epoxy-18-hydroxyoctadecanoic acid, which showed no significant changes around 10 μg cm^–2 in banana fruit cuticles under different temperatures (Figure 3 and Supplementary Table 2). Fatty acids without added groups accumulated in a wide range of chain-length distribution from C₁₆ to C₃₀, with most abundant of C₁₆, C₁₈, C₂₄, and C₃₀. The chilling stress stimulated the increase of C₁₆- and C₁₈-chain fatty acids remarkably, whereas C₂₄ and C₃₀ were relatively stable, resulting in the total fatty acids raised from 5.5 to 10.0 μg cm^–2 (Figure 3 and Supplementary Table 1). The 9(10), 16-dihydroxyhexadecanoic acid was the main dihydroxy fatty acid increased slightly from 0.98 to 1.3 μg cm^–2 (Figure 3 and Supplementary Table 2). Consequently, the prominent carbon chain of C₁₆ over C₁₈ in cutin monomers was shifted from 0.21 to 0.35 by chilling stimulation (Table 1). Noteworthy, another monomer of 2-hydroxy fatty acids accumulated with a chain length of C₂₂–C₂₆ with most abundant of C₂₄, raised from 0.6 to 2.1 μg cm^–2 after 6 days of low-temperature storage. Primary alcohols ranged from C₂₀ to C₃₀ with even-numbered carbon chains showed no obvious changes. In addition, cyclic monomers, predominantly coumaric acid and their derivatives, were also detected and increased slightly under low-temperature storage (Figure 3 and Supplementary Table 2).

The cuticular waxes in banana fruit cuticles were accumulated with a variety of typical VLC aliphatic and cyclic compounds. The VLC aliphatic pattern contained most abundant of fatty acids followed by n-alkanes, primary alcohols, and aldehydes. The cyclic pattern in waxes was dominated by pentacyclic triterpenoids and sterols (Figure 4 and Supplementary Table 3). Generally, the aliphatics accumulated higher than cyclics in banana fruit cuticles. The accumulation of aliphatics increased significantly to 26.8 μg cm^–2 in banana fruit stored under 4°C compared with that of 19.9 μg cm^–2 in fruit under room temperature (Table 1). The cyclics exhibited no obvious changes (about 8.0 μg cm^–2) under both temperature storage. As a result, the ratio of aliphatics over cyclics was shifted from 2.6 to 3.3 (Table 1).

The various cuticular wax components accumulated into a homologous series. The VLC aliphatics (> C₁₈) were detected within a fairly broad range of carbon chain lengths. Fatty acids as the most abundant aliphatic pattern ranged from C₂₀ to C₃₀ with most abundant of C₂₂ and C₂₄, which raised significantly under low-temperature storage at 6 days compared to that stored under 25°C (Figure 5A). Carbon chain ranged between C₂₀ and C₃₂ was detected for n-alkanes, almost no changes for all the n-alkane series of components, with an increase for C₂₅ at 6 days under 4°C (Figure 5C). Aldehydes (C₂₆ to C₃₀) and primary alcohols (C₂₀ to C₃₂) accumulated with prominent of C₂₈ and C₃₀, which increased for C₃₀ under low temperature (Figures 5B,D).

The pentacyclic triterpenoids dominated the cyclic pattern containing a variety of members with the most abundant being uvaol, followed by α-amyrin, β-amyrin, epi-lupeol, and epi-lupeol acetate. Sterols were comprised of β-sitosterol and stigmasterol (Figure 6 and Supplementary Table 3). After 6 days, besides uvaol, which showed a slight increase, the content of all the other triterpenoids was lower in fruit cuticles under 4°C than that under 25°C. In contrast, both β-sitosterol and stigmasterol were stimulated to increase under chilling temperature (Figure 6 and Supplementary Table 3).

The gene response to the chilling injury on mRNA level was thoroughly analyzed using RNA-sequencing together with RT-qPCR. The results showed that in total of 16,707 DEGs were annotated with 9,420 DEGs (56%) downregulation and 7,287 DEGs (44%) upregulation in this study. The Gene Ontology (GO) annotation indicated that 7,576 DEGs (47.7% of total GO-annotated DEGs) were identified as “biological process” with 18 functional groups; 4,834 DEGs (30.5%) were annotated as “cellular component” with 14 functional groups; and 3,464 DEGs (21.8%) were involved in “molecular function” with 10 functional groups (Supplementary Figure 1). The Kyoto Encyclopedia of Genes and Genomes (KEGG) mapping analysis revealed that the DEGs were mainly involved in seven classes of various metabolism pathways (61.6% of total annotated DEGs), when compared to low temperature with room temperature storage (Supplementary Figure 2).

As cuticular constituents were mostly constituted by lipids, a total of 392 DEGs in 15 different pathways involved in lipid metabolism were detected. Fatty acid elongation, ether lipid metabolism, biosynthesis of unsaturated fatty acids, etc., were enriched markedly (rich factor > 0.5, Figure 7A). Further analysis revealed that 111 DEGs involved in biosynthesis and formation of cuticle were identified. These DEGs were mainly structural genes in five pathways for the biosynthesis of cuticle, i.e., fatty acid elongation, VLC alcohol and n-alkane pathways, cyclic waxes, cutin biosynthesis, and lipid-transfer proteins (Figure 7B).

Among these cuticle-related genes, 23 DEGs were involved in fatty acid elongation including MaLCAS1/2/9, MaKCS1/2/3/4/11, MaKCR1, MaECR, and MaCER26/26L (Figure 8A and Supplementary Table 4). Besides numerous DEGs in sterol biosynthesis, 22 DEGs were identified as structure genes in wax biosynthesis respond to chilling injury, which were MaCER1/3/7, MaFAR1/4, MaWSD1, MaDGAT1/2, MaSQS1/2, and Masqe1/3 (Figure 8B and Supplementary Table 4). About 26 DEGs in cutin monomer biosynthesis, MaCYP74A, MaCYP74A2, MaCYP86A1/2, MaCYP86B1, MaHHT1, MaHHT, MaPXG4, MaDCR1/2/3, MaGPAT1/3/4/5/6/7, and MaAGPAT6 were differentially regulated (Figure 8C and Supplementary Table 5). And 12 DEGs of lipid transporters MaLTP1/2, MaLTPG1/2, and MaABCG11, which were potentially related to the transportation of cutin monomers and waxes were detected (Figure 8D and Supplementary Table 5).