Genome-wide analyses of SWEET family proteins reveal involvement in fruit development and abiotic/biotic stress responses in banana

Sugars Will Eventually be Exported Transporters (SWEET) are a novel type of sugar transporter that plays crucial roles in multiple biological processes. From banana, for the first time, 25 SWEET genes which could be classified into four subfamilies were identified. Majority of MaSWEETs in each subfamily shared similar gene structures and conserved motifs. Comprehensive transcriptomic analysis of two banana genotypes revealed differential expression patterns of MaSWEETs in different tissues, at various stages of fruit development and ripening, and in response to abiotic and biotic stresses. More than 80% MaSWEETs were highly expressed in BaXi Jiao (BX, Musa acuminata AAA group, cv. Cavendish), in sharp contrast to Fen Jiao (FJ, M. acuminata AAB group) when pseudostem was first emerged. However, MaSWEETs in FJ showed elevated expression under cold, drought, salt, and fungal disease stresses, but not in BX. Interaction networks and co-expression assays further revealed that MaSWEET-mediated networks participate in fruit development signaling and abiotic/biotic stresses, which was strongly activated during early stage of fruit development in BX. This study provides new insights into the complex transcriptional regulation of SWEETs, as well as numerous candidate genes that promote early sugar transport to improve fruit quality and enhance stress resistance in banana.

In plants, sugars not only provide energy, essential signal molecule components (carbon), and metabolic intermediates for plant growth and development, but also play crucial roles in both abiotic and biotic stress responses [1][2][3][4][5] . Sugar transport from source-to sink cells is controlled by multiple transporters, i.e. sucrose transporters (SUT) and monosaccharide transporters 3,[6][7][8] . Several studies revealed a constantly low level of SUT expression and saturable sucrose transport kinetics, indicating that SUTs may not be the key rate-limiting factor for sucrose transport but additional transporters are responsible for sucrose allocation across the membrane [9][10][11][12][13][14] . Recently, a new class of sugar transporter, the Sugar Will Eventually be Exported Transporters (SWEET), was identified in Arabidopsis thaliana 15 . Unlike SUTs which contain 12 transmembrane domains (TMD) connected by hydrophilic loops and catalyze sucrose/H + -coupled transport 3,13,14,16 , SWEET proteins contain 7 TMD connected by a PQ-loop-repeat 15,17,18 and function as key transporters for sucrose, hexose, and fructose along a concentration gradient 19 .
To date, genome-wide analysis has identified variable number of SWEET genes in plants, including 17 in A. thaliana 15 , 21 in rice (Oryza sativa) 20 , 23 in sorghum (Sorghum bicolor) 21 , 52 in soybean (Glycine max) 22 , 35 in potato (Solanum tuberosum) 23 , 29 in tomato (Solanum lycopersicum) 18 , 33 in apple (Malus domestica) 24 , and 17 in grape (Vitis vinifera) 4 . Phylogenetic analysis revealed that SWEETs could be classified into four subfamilies (Clades I-IV) in A. thaliana, O. sativa, V. vinifera, S. lycopersicum, and M. domestica 14,18,22,24 . SWEETs in Clades I and II preferentially transport hexoses 15,25 , while those in Clade III are sucrose transporters 25,26 . Clade IV contains relatively less number of SWEET genes compared to other Clades and has been shown to largely consist of vacuolar transporters controlling the flux of fructose across the tonoplast 27,28 . Biochemical and functional analyses have shown SWEETs' involvement in multiple essential plant biological processes, including growth, senescence 26,28 , seed and pollen development 15,29,30 , environmental adaptation 20 , and host-pathogen interactions 4,15 . For example, in A. thaliana, AtSWEET17 knockout mutants exhibited a reduced root growth phenotype 28 , and AtSWEET15 was upregulated ~22-fold in senescent leaves possibly involved in the regulation of carbohydrate mobilization 31 . AtSWEET8 contributes to pollen viability, suppression of which was shown to reduce starch content in pollen grains and cause male sterility 15 . AtSWEET16-overexpressing plants were shown to display significant alterations in sugar levels and developmental processes, like germination and growth 32 . Overexpression of AtSWEET4 increased plant size 33 . In addition, expression of some SWEET genes can be induced by various stress conditions, including cold, salt, drought, and plant-pathogens 4,15,33,34 . Previously, AtSWEET4 expression was reported to enhance tolerance to freezing and drought stresses 4,33 . SISWEET15 in S. lycopersicum was identified as an important regulator of salt tolerance 34 . Pseudomonas syringae infection was shown to significantly induce the expression of AtSWEET-4, -5, -7, -8, -10, -12, and -15 in A. thaliana leaves 15 . In V. vinifera, VvSWEET4 expression was reported to participate in the interaction with Botrytis cinerea 4 . Taken together, these studies have revealed many important roles of the SWEET gene family in the regulation of plant growth, development, and responses to abiotic and biotic stresses.
Banana (Musa acuminata L.) is not only one of the most favored starchy fruits worldwide, but also an important staple food in some African and Latin American countries 35,36 . Sugar transporters are extremely important for banana plant since sugars play an important role in starch accumulation and sugar content is also a major component of fruit quality and economic value. Being a large annual monocotyledonous herbaceous plant, banana is frequently affected or even destroyed by various abiotic and biotic stresses during growth and development. Investigations of key candidate genes involved in fruit development and response to different stress conditions based on complete genome sequences have been conducted in several fruit crops, but only few in banana [37][38][39] . Despite of its importance in fruit quality, plant growth and stress regulation, SWEET genes have not been studied in banana.
For the first time, we have conducted a genome-wide analysis of SWEET genes in banana, termed as MaSWEET, and analyzed their phylogenetic relationship, gene structure, protein motifs, spatial and temporal expression patterns, and expression in response to both abiotic and biotic stresses in two different M. acuminata varieties. In addition, we characterized interaction networks and co-expression patterns of MaSWEETs in early stages of fruit development and under abiotic and biotic stresses. This comprehensive study serve to facilitate our understanding of MaSWEET associated fruit development processes and stress responses, and provide a foundation for future studies of crop improvement involving SWEETs.

Identification and phylogenetic analysis of banana MaSWEET family members. To identify all
MaSWEET members in BaXi Jiao (BX) [M. acuminata L. AAA group, cv. Cavendish] and Fen Jiao (FJ) [M. acuminata AAB group] bananas, Hidden Markov Model searches using typical SWEET domains (PFAM: PF03083) as queries, as well as BLAST searches using AtSWEET and O. sativa OsSWEET sequences as queries, were performed in the banana genome database. After validating the SWEET domain using the CDD and PFAM databases, a total of 25 MaSWEETs were identified. MaSWEETs were named according to their respective orthologous genes in A. thaliana. The 25 predicted MaSWEET proteins varied from 171 (MaSWEET7b) to 333 (MaSWEET14j) amino acid residues in size, with relative molecular masses between 19.097 (MaSWEET7b) and 37.418 (MaSWEET14j) kDa, and isoelectric points ranging from 5.29 to 9.83 (Supplementary Table S1). The highly variable structure of MaSWEETs may suggest their potentially different functional roles in different biological processes or under different growth conditions.
To study the evolutionary relationships between SWEET family proteins, a neighbor-joining tree was generated by aligning 25, 17, and 21 SWEET proteins from banana, A. thaliana, and O. sativa, respectively, using ClustalX and MEGA5.0 software (Fig. 1). Based on the phylogenetic tree, all MaSWEET proteins were grouped into four Clades (I-IV); Clades II and III were large, with more than 8 MaSWEET members, whereas Clades I and IV contained less than 4 MaSWEET proteins, consistent to A. thaliana and O. sativa 15,20 . Distinct from other plant species, there were 10 MaSWEET14 members in Clade III, while only 1 each in A. thaliana and O. sativa (Fig. 1). Besides, the Ka/Ks ratio of each duplicated gene pairs between AAA and AAB, AAA and Arabidopsis, and AAA and rice was calculated to estimate the molecular evolutionary rates (Supplementary Table S2). In AAA and AAB, the Ka/Ks ratios from 22 duplicated gene pairs were less than 1, while in other 3 duplicate gene pairs are more than 1. In AAA and rice, Ka/Ks ratios of 23 duplicated gene pairs were less than 1, only 1 duplicate gene pair is more than 1. Whereas in AAA and Arabidopsis, the Ka/Ks ratios from 11 duplicated gene pairs were less than 1, the other 14 duplicate gene pairs are more than 1.
Gene structure and conserved motif analysis of banana MaSWEETs. Evolutionary analysis further supported the classification of 25 MaSWEETs into four groups, in accordance with cluster analysis data (Fig. 2). The exon-intron structural evolution using Gene Structure Display Server software showed that the MaSWEETs contained 6 exons in Clades I and IV, 4-5 exons in Clade II, and 4-6 exons in Clade III. This suggests that MaSWEETs in the same group share similar exon-intron organizations.
The conserved motif analysis and functional annotation prediction of banana MaSWEETs were then performed based on phylogenetic relationships ( Fig. 3; Supplementary Table S3). A total of 9 conserved motifs in all 25 MaSWEETs were identified using MEME software and annotated with the InterPro database. Motifs 1-6 and 8 were annotated as 7 TMD helices, and all 25 identified MaSWEETs contained typical SWEET-type TMD helices. In particular, 7 TMD helices were found in all MaSWEETs in Clades I and IV and most in Clade II including  MaSWEET-4a-e, and -6a and Clade III including MaSWEET-14a, -14c-g, -14i, and -14j. In Clade II, motif 5, 7, and 8 were lacking in MaSWEET7a, while motifs 2 and 4 were lacking in MaSWEET7b. In Clade III, motif 8 was found in neither MaSWEET-14b nor MaSWEET-14h. These results suggest that MaSWEETs clustered in the same group share conserved motifs which further supports the phylogenetic results.

Expression patterns of MaSWEETs in different banana tissues.
To investigate the MaSWEET spatial expression patterns in banana, roots, leaves, and fruits of BX and FJ were sampled and subjected to transcriptional analysis. 19 (excluding MaSWEET-2b, -4e, -6a, -7b, -14a, and -16a) of the 25 MaSWEETs were expressed in at least one of the tested tissue in both varieties ( Fig. 4 and Supplementary Table S4).
Comparison of spatial expression profiles between the two banana varieties revealed 19 MaSWEET genes were expressed in BX compared to 17 in FJ. Moreover, similar spatial expression patterns of 5 MaSWEET genes (MaSWEET-4b, -14b, -14c, -14d, and -14h) were found in BX and FJ root and leaf. Significantly more MaSWEET genes (MaSWEET-4a, -4c, -14c, -14d, -14g, and -14h) were highly expressed in fruit than leaf in both genotypes. In contrast, some spatial expression patterns of MaSWEET were different between the two genotypes. In roots, for example, MaSWEET4c showed relatively lower expression (FPKM < 5.5) in BX compared to FJ (FPKM > 22.4). In fruit, 5 genes (MaSWEET-1a, -4a, -4c, -7a, and -14g) were highly expressed in BX (527 > FPKM > 13) in contrast to rather low expression in FJ (0 < FPKM < 5), suggesting MaSWEET members may play different roles in different organs in each banana variety. Furthermore, in all tested organs including roots, stems, and fruits, MaSWEET4b was highly expressed (FPKM > 10) in both BX and FJ, indicating a key role for this particular gene in all tissues, but further research is necessary. Together, these genotype-based tissue expression patterns provide insight for further study of the MaSWEET gene family in banana in relevance to plant and fruit development.  Table S5). Twenty of the 25 MaSWEETs were expressed during fruit development and ripening of BX and FJ bananas, taken together.
These results clearly demonstrated that more MaSWEET genes were significantly up-regulated in leaf (FPKM > 10) in FJ than BX by cold stress. In addition, the number of genes that were induced to up-regulated express in response to salt and osmotic stresses was also higher in FJ than in BX.

Validation of the differentially expressed banana
MaSWEET by qRT-PCR. Based on RNA-seq data, MaSWEET-4b and -14c were highly expressed (FPKM > 4) in all tested banana tissues including root, leaf, and fruit, while differential expression between the two genotypes was detected in MaSWEET-1a and -14g. MaSWEET-4b and -14d were highly expressed (FRKM > 2) in all tested fruit development and ripening stages in both BX and FJ genotypes, while MaSWEET-14h and -14g showed genotypic variation of expression at these stages. MaSWEET-4b and -14c were upregulated by cold, salt, and osmotic stresses in BX and FJ, while MaSWEET-4c and -14d showed genotypic variations in response to each abiotic stress. Furthermore, MaSWEET-4d and -14h were upregulated by Foc 4 TR4 infection in BX and FJ, whereas MaSWEET-4c and -14b exhibited genotype-specific response to Foc 4 TR4 infection.
For verification, the above mentioned differentially expressed MaSWEETs were selected for qRT-PCR analysis to validate RNA-seq data. After normalization, all the examined MaSWEETs with the exception of MaSWEET-4b and -14c in BX leaves, MaSWEET4b in BX 0 DAF, MaSWEET14d in BX 14 DPH and FJ 6 DPH, and MaSWEET-4d and -14h in FJ 0 and 2 DPI showed trends consistent with the RNA-seq analysis (Fig. 8). The correlation coefficient between RNA-seq and qRT-PCR data ranged from 0.7493 to 0.9997 (Supplementary Table S8), except MaSWEET-4d, -14h, and -4c in 0 and 2DPI of BX and FJ (Supplementary Table S8M,N, and S8O). These results indicate that RNA-seq analysis supplied suitable expression results for both banana varieties.

SWEET family interaction networks and their co-expression in early fruit development stages and under abiotic/biotic stresses. Studying gene family interaction networks is very useful for investigat-
ing potential gene functions 40 . SWEET genes have been documented as playing active roles in plant development and abiotic/biotic stresses in many species 4,15,[28][29][30]33 . However, due to the paucity of such reports in banana, we investigated the potential MaSWEET protein interaction and co-expression networks to aid future studies investigating their biological function based on experimentally validated interactions. An AtSWEET network was constructed, and 27 interactive proteins (with high confidence; score > 0.9), including 17 SWEETs and 10 other interactive proteins [alcohol dehydrogenase-like protein (AT1G66800), photoassimilate-responsive protein 1 (AT5G52390), WD40 domain-containing protein (AT1G47610), remorin-like protein (AT2G41870), tyrosine aminotransferase family protein (AT4G23590), self-incompatibility protein S1-like protein (AT1G26795), armadillo/beta-catenin-like repeat family protein (AT5G62580), BAHD acyltransferase (AT5G47980), annexin 7 (AT5G10230), and aspartyl protease family protein (AT4G22050), were identified with STRING 41 . Then, orthologs of these interaction proteins in banana were identified with reciprocal BLASTP analyses, and the expression profiles of these genes in BX and FJ at early development stages (0 and 20 DAF) were extracted from RNA-seq data sets ( Fig. 9 and Supplementary Tables S9 and S10). At 0 DAF in BX, 38 gene pairs showed uniform up-regulation, but down-regulated gene pairs were not detected (

Discussion
In spite of the economic and social importance of bananas, research on banana plants has generally be slow relative to other crops, especially with respect to fruit development and responsiveness to abiotic and biotic stresses 37 . The SWEET family represents a novel type of sugar transporter that has participated in the regulation of plant growth, development, and stress responses in many plant species 4,15,18,20,22 . Using the genome wide identification approach, we have identified 25 MaSWEET genes in M. acuminata, and made a comparative analysis in two genotypes including BX and FJ. These genes were classified into four subfamilies (Clades I-IV) according to their phylogenetic evolutionary relationship (Fig. 1), which was consistent with the classification of SWEETs in A. thaliana and O. sativa 14,22 . However, there are also discrepancies in the size of subfamilies, for instance in Clade III, a staggering 10 members of MaSWEET14 was found, which is in sharp contrast to only one member in AtSWEET and OsSWEET (Fig. 1). In addition, the number of duplicated gene pairs (Ka/Ks ratio <1 or >1) was different between AAA and AAB, AAA and Arabidopsis, and AAA and rice (Supplementary Table S2). This may suggest a broad number event of gene duplication and divergence occurred in MaSWEET14 members during banana evolutionary history compared to A. thaliana and O. sativa.
These evolutionary and phylogenetic analyses were supported by gene structure and conserved motif analyses. Gene structure analysis indicated that MaSWEETs within each Clade shared a similar exon-intron organization (Fig. 2). For example, each member of Clades I, III (except MaSWEET-14h and -14b) and IV harbors 5 introns, whereas Clade II members contain fewer (4 introns). According to a previous report in O. sativa, the rate of intron loss is faster than that of intron gain after segmental duplication 42 . We therefore propose that Clades I, III, and IV might contain the original ancestor MaSWEET genes, and Clade II might be derived by gene duplication and divergence with subsequent intron loss. This structural feature of MaSWEETs has also been observed in other plant species, such as S. lycopersicum 18 and G. max 22 . Moreover, conserved motif analysis indicated that all the MaSWEETs harbored the seven conserved TMDs, while gene members within each subfamily shared other more specific motifs (Fig. 3). Such an observation is consistent with other plants, such as A. thaliana 15 , S. lycopersicum 18 , and O. sativa 20 .
The fruit development and ripening processes are crucial for banana yield and quality. The SWEET family has been reported to participate in the fruit development process in many plant species, such as M. domestica 24 , S. lycopersicum 18 , and Vitis vinifera 4 . In tomato, expression levels of SWEET12C and SWEET14 were relatively high in ripening fruit 43 . SWEETs putatively control sucrose accumulation in sweet sorghum stems 19 . However, whether MaSWEETs in banana have also participated in fruit development and postharvest ripening, so far remained elusive. In the current study, we found that more than 80% MaSWEETs were expressed during the five developmental and ripening stages tested in BX and FJ varieties, and more than 27% of them were highly expressed (FPKM > 10) at each stage (Fig. 5), implying that MaSWEETs are extensively involved in banana fruit development and ripening. Interestingly, MaSWEET expression patterns at a certain stage was substantially different between genotypes, for instance, 20 MaSWEETs was highly expressed (FPKM > 10) in BX compared to only 7 in FJ, in the initial stage of pseudostem emergence (0 DAF). The greater number of highly expressed genes during early development in BX versus FJ may be explained by the genome variations between these two genotypes. The majority of edible cultivated bananas are triploid, originating from intra-or interspecific hybridization between the diploid M. acuminata (A genome) and M. balbisiana (B genome), producing AAA, AAB, and ABB genotypes among others 44,45 . The BX genotypes, with AAA genome, is known to produce high yield and quality fruit with long fingers and extended shelf life and storage period compared to the FJ genotype with AAB genome 37,46 . It is tempting to assume that the increase of MaSWEET expression during early stage of fruit development augments sugar transport ability, thereby enhancing quality and yield of banana fruit. Furthermore, interaction network and co-expression analyses showed more gene pairs were uniformly up-regulated in BX than in FJ in the early stages of fruit development (Fig. 9). Such findings in this study is consistent with previous studies suggesting that the A-genome harbors more genes that are important for banana yield and quality and could be used as a target in breeding programs 37,46 .
Bananas are large, monocotyledonous, herbaceous plants that are extremely sensitive to abiotic stress, such as those imposed by drought, salt, or cold conditions 46 . However, the response of banana to abiotic stresses is largely inadequate 47 . The expression of SWEETs in various plant species reportedly changes in response to multiple abiotic stresses 18,32,48 . For example, in S. lycopersicum, Class II SlSWEETs were found to be upregulated several fold in leaves by sugar, salt, and temperature stress treatments 18 . Under cold stress conditions, AtSWEET16 overexpression give rise to plants with increased freezing tolerance compared to wild type, probably due to a greater accumulation of vacuolar sucrose and glucose 32 . AtSWEET15, also known as SAG29, was strongly induced during senescence and abiotic stresses, such as cold, high salt, and drought treatments 48 . Under high salt conditions, AtSWEET15-overexpressing lines exhibited reduced root growth and cell viability compared to control plants 48 . Under these stress conditions, the expression of SWEETs tended to promote sugar accumulation in plant vacuoles and change plant organ developmental states. In the present study, 76% MaSWEETs showed transcriptional changes after abiotic stress treatment, including osmotic, salt, and cold stresses, in both BX and FJ genotypes (Fig. 6). Previous gene expression analysis have shown increased SWEET transcription when subjected to cold, drought, and salt abiotic stresses in other plants, such as S. lycopersicum 18 , A. thaliana 32 and O. sativa 26 . These results indicate that SWEET genes exhibit an extensive response to abiotic stresses. Comparative analysis clearly demonstrated that more MaSWEET genes were significantly upregulated in FJ than in BX tissues subjected to cold treatments. In salt and osmotic stresses, the number of genes that were induced to up-regulated express was also higher in FJ than in BX (Fig. 6). B-genomecontaining banana varieties have previously been reported to confer strong tolerance to abiotic stress 45,46 . In addition, numerous studies have confirmed that SWEETs are involved in maintaining sugar homeostasis in plant organs, facitating adaptation to low temperatures 27,48 . Taken together, it is reasonable to propose that significant upregulation of MaSWEETs in FJ banana plants subjected to cold stress could be attributable to thermal tolerance.
Currently, banana production has been devastated by fungal infestations caused by Foc TR4 49 . However, it remains unresolved whether SWEET genes play a role in banana disease resistance. In cassava, the bacterial blight Xanthomonas axonopodis was found to directly induce MeSWEET10a transcription 50 . Similarly, expression of OsSWEET14 confers resistance against bacterial blight in wild O. sativa 51 . Recently, it was reported that in addition to Xanthomonas, other bacterial and fungal pathogens, such as Botrytis cinerea and some mycorrhizal fungi, also modulate expression of plant SWEETs 4, 52 . In the current study, we found transcriptional up-regulation of 3 MaSWEET genes (MaSWEET-4c, -4d, and -14h) in response to Foc TR4 infection in both BX and FJ bananas, suggesting a possible crucial role for these MaSWEETs to play in response to fungal infections in banana. Additionally, higher number of MaSWEET genes were highly expressed in FJ than in BX, suggesting that the B-genome which is not present in BX, may also be involve in protection against the biotic stress in banana.
In conclusion, we identified 25 MaSWEET genes from the M. acuminata genome, which could be classified into four clades following phylogenetic analysis, gene structure, and conserved protein motif analyses. Spatial and temporal expression profiles of MaSWEETs in two triploid banana varieties revealed that MaSWEETs may have distinct functions during the fruit development and ripening. The expression patterns of MaSWEETs in response to both abiotic and biotic stresses indicate that they are also involved in stress signaling pathway regulation, especially in B-genomecontaining banana varieties. Furthermore, interaction networks and co-expression assays indicated the strong transcriptional response of banana SWEET-mediated networks in early fruit development and under abiotic/biotic stresses of A-genomecontaining banana varieties, suggesting MaSWEETs enhance sugar transport during early fruit development and under abiotic/biotic stresses. These results serve to advance the current understanding of SWEET functional characteristics in banana developmental processes and stress responses and lay a solid foundation for further genetic improvement of banana quality, yield, and resistance to both biotic and abiotic stresses.

Methods
Plant materials and treatments. BX [M. acuminate L. AAA group, cv. Cavendish] is a triploid variety of banana widely cultivated in tropical and subtropical regions in China, which produces a high yield of high quality fruits fit for long-term storage. FJ [M. acuminata AAB group] is also a triploid variety cultivated in the Guangdong and Hainan provinces of China, which had good fruit flavor, ripens quickly, and exhibits high tolerance to abiotic stresses. Samples of both genotypes were obtained from the banana plantation at the Chinese Academy of Tropical Agricultural Sciences (Danzhou, Hainan, China). Five-leaf stage young BX and FJ banana seedlings with uniform growth were grown in soil at 28 °C with 70% relative humidity and 16 h light/8 h dark cycle with 200 μmol·m −2 ·s −1 light intensity.
For salt and osmotic stress treatments, 5-leaf stage banana seedlings were irrigated with 300 mM NaCl and 200 mM mannitol for 7 d, respectively. For cold treatment, banana seedlings were maintained at 4 °C for 22 h. For examination of banana fruit development, fruits were collected 0 (budding), 20 (cutting flower), and 80 (ready for harvest) DAF from both BX and FJ plants. To assess banana fruit postharvest ripening, BX fruits stored for 8 and 14 DPH and FJ fruits stored for 3 and 6 DPH, which represent fruits ripening stages of "more green than yellow" and "full yellow" according to Pua et al. 53 in each variety, were sampled. For Foc TR4 treatments, roots of 5-leaf stage banana seedlings were dipped in a Foc TR4 spore suspension of 1.5 × 10 6 condia/mL. The entire root system was harvested at 0 and 2 DPI, immediately frozen in liquid nitrogen, and stored at −80 °C until expression analysis.  Table S1. Phylogenetic analysis of SWEET amino acids sequences from banana, A. thaliana and O. sativa were carried out using Clustal X2.0 and MEGA5.0 software with bootstrap values for 1000 replicates. The Ka/Ks ratio was used to show the selection pressure for the duplicate genes 54 .

Identification and phylogenetic analyses of
Protein properties and gene structure analysis. Molecular mass and isoelectric points of MaSWEET proteins were predicted using the ExPASy database (http://expasy.org/). Motifs of MaSWEET proteins were analyzed using MEME software (http://meme.nbcr.net/meme/cgi-bin/meme.cgi) and annotated by InterProScan database search (http://www.ebi.ac.uk/Tools/pfa/iprscan/). Structural features of MaSWEET genes were identified using Gene Structure Display Server software (http://gsds.cbi.pku.edu.cn/). The interactions between SWEETs and their transcriptional factors were determined using STRING (http://string-db.org/) with a confidence score >0.9 and Cytoscape software.

Transcriptomic analysis.
To investigate the transcriptional accumulation of MaSWEET genes in different organs of the BX and FJ varieties, roots, leaves, and fruits at 80 DAF were collected. Banana fruits at different development stages, i.e. 0, 20, and 80 DAF, were collected to assess the expression patterns of banana MaSWEETs during fruit development. To examine the expression of MaSWEETs during postharvest ripening processes, fruits at 8 (more green than yellow) and 14 (full yellow) DPH in BX and at 3 (more green than yellow) and 6 (full yellow) DPH in FJ were collected. The 5-leaf stage banana seedlings were treated with 200 mM mannitol for 7 days, 300 mM NaCl for 7 days, low temperature (4 °C) for 22 h, or Foc TR4 spore suspension (1.5 × 10 6 condia/mL) dipping roots for 2 days, respectively. All samples were collected and frozen quickly in liquid nitrogen and stored at −80 °C for RNA extraction and transcriptome analysis. Each sample contains three biological replicates.
Total RNA from each sample was extracted using a plant RNA extraction kit (Tiangen Biotech, Beijing, China), and then cDNA libraries constructed according to the protocols supplied by Illumina. RNA-seq was performed with an Illumina GAII following the manufacturer's instructions, with two replicates per sample. The sequencing depth was 5.34X on average. Adapter sequences in the raw sequence reads and low quality sequences were removed using FASTX-toolkit and FastQC, respectively. The obtained clean reads were mapped to the DH-Pahang genome (M. acuminata, diploid A genome) 35 . Transcriptome assemblies were performed using Cufflinks 55 . Gene expression levels were calculated as FPKM. Differentially expressed genes were then sequenced from RNA-seq data.
qRT-PCR analysis. The expression levels of MaSWEETs in different tissues, different stages of fruit development and ripening, and in response to cold, salt, osmotic, and Foc TR4 stresses were further assessed by qRT-PCR using a Stratagene Mx3000P detection system (Stratagene, San Diego, CA) with SYBR ® Premix Ex Taq TM kit (TaKaRa Bio Inc, Otsu Shiga, Japan). Primers that had high specificity and efficiency according to melting curve analysis and agarose gel electrophoresis were used to conduct quantification analysis (Supplementary Table S13). Amplification efficiencies of primer pairs ranged from 0.9 to 1.1, ACTIN (GenBank accession no.EF672732) and UBQ (GenBank accession no. XP009390884.1) were verified as being constitutively expressed and suitable as internal controls were used as reference genes to normalize transcriptional levels of MaSWEET genes. Relative expression levels of MaSWEETs were analyzed in triplicate and calculated using the 2 −ΔΔCT method 56 .