Transcriptome-wide identification and expression profiling of the ERF gene family suggest roles as transcriptional activators and repressors of fruit ripening in durian

The involvement of the phytohormone ethylene as the main trigger of climacteric fruit ripening is well documented. However, our knowledge regarding the role of ethylene response factor (ERF) transcription factor in the transcriptional regulation of ethylene biosynthesis during fruit ripening remains limited. Here, comprehensive transcriptome analysis and expression profiling revealed 63 ERFs in durian pulps, termed DzERF1–DzERF63, of which 34 exhibited ripening-associated expression patterns at three stages (unripe, midripe, and ripe) during fruit ripening. Hierarchical clustering analysis classified 34 ripening-associated DzERFs into three distinct clades, among which, clade I consisted of downregulated DzERFs and clade III included those upregulated during ripening. Phylogenetic analysis predicted the functions of some DzERFs based on orthologs of previously characterized ERFs. Among downregulated DzERFs, DzERF6 functional prediction revealed its role as a negative regulator of ripening via ethylene biosynthetic gene repression, whereas among upregulated genes, DzERF9 was predicted to positively regulate ethylene biosynthesis. Correlation network analysis of 34 ripening-associated DzERFs with potential target genes revealed a strong negative correlation between DzERF6 and ethylene biosynthetic genes and a strong positive correlation between DzERF9 and ethylene biosynthesis. DzERF6 and DzERF9 showed differential expression patterns in association with different ripening treatments (natural, ethylene-induced, and 1-methylcyclopropene-delayed ripening). DzERF6 was downregulated, whereas DzERF9 was upregulated, during ripening and after ethylene treatment. The auxin-repressed and auxin-induced expression of DzERF6 and DzERF9, respectively, confirmed its dose-dependent responsiveness to exogenous auxin. We suggest ethylene- and auxin-mediated roles of DzERF6 and DzERF9 during fruit ripening, possibly through transcriptional regulation of ethylene biosynthetic genes.

6 128 [30]. A texture analyzer was used to measure the softness of the first pulp to ensure that samples of the two 129 cultivars were compared at the same ripening stage [32]. Thereafter, the second fruit pulp was collected 130 without a seed, immediately frozen in liquid nitrogen, and stored at −80 °C until RNA extraction.
139 To analyze the expression levels of candidate ripening-associated DzERFs under auxin treatment, we used 140 the cDNA generated in our previous study (Khaksar and    148 Mapping the reads to the D. zibethinus reference genome and expression analysis 149 We used the OmicsBox program (Biobam, Spain) for transcriptome data analysis. Raw reads were 150 filtered to obtain high-quality clean reads by removing adapters, reads shorter than 60 bp, and low-quality 151 reads with a Q-value ≤ 30 using FastQC and Trimmomatic. Then, a gene-level analysis was performed by 152 aligning the reads against the reference genome of durian cv. Musang King [31] using STAR (Spliced 153 Transcripts Alignment to a Reference). Counting of reads and expression analysis were performed with . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made 179 (AFH56415.1/AID51421.1/AGA15800.1) [37]).
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

181
We used the MEME program (http://meme-suite.org) to identify the conserved motifs of ripening-182 associated DzERFs [38] with the following parameter settings: motif length = 6-100; motif sites = 2-120; 183 maximum number of motifs = 10; the distribution of a single motif was "any number of repetitions".   was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made    (Table 1). Expression profiling 248 of 63 DzERFs at three different stages during post-harvest ripening (unripe, midripe, and ripe) of durian 249 pulp cv. Monthong revealed 34 DzERFs with differential expression patterns. A heat map was constructed 250 to cluster these DzERFs according to their expression levels (RPKM) based on the Euclidian distance 251 method (Fig 1). Accordingly, the heat map classified 34 ERFs into three separate clades. Clade 1 consisted 252 of 15 ERF genes (DzERF5, 6, 11, 13,14,18,19,21,23,26,27,28,30,31 was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   Fig S1). We further investigated the conserved motifs of DzERFs and identified at least 10 271 (Fig 2 and Supplementary Fig S2). Among them, motifs 1 and 2 corresponded to the AP2/ERF domain and 272 were widely distributed in all DzERFs, except for DzERF19, which lacked motif 2 (Fig 2). Notably, the 273 DzERFs that were clustered together harbored a similar motif organization.
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made    296 for DzERF24, which was not expressed in leaf and stem tissues (Fig 4). This expression profile suggests 297 the role of ERFs in a wide range of physiological processes in various tissues.

Fig 4. Tissue-specific expression profile of ripening-associated durian ERFs (DzERFs) in the Musang
299 King cultivar at the ripe stage. We used the publicly available Illumina RNA-seq data to analyze the . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   Fig 5A) and a correlation network ( Fig 5B). As revealed by hierarchical clustering of Pearson's 310 correlations, all DzERFs for which the expression decreased during ripening were clustered together and 311 were negatively correlated with the ripening-associated genes. However, the DzERFs that increased during 312 ripening were clustered together with the ripening-associated genes, suggesting a positive correlation 313 between those DzERFs and ripening-related genes (Fig 5A). Notably, as shown in Fig 5B, all DzERFs for 314 which the expression increased during ripening exhibited positive correlations with ripening-associated 315 genes. Among these, the highest positive correlation was observed between DzERF9 and ethylene 316 biosynthetic genes (SAM synthase, ACS, and ACO), followed by DzERF9 and auxin biosynthetic genes 317 (TAA1 and YUCCA4). However, for those DzERFs that decreased during ripening, a negative correlation 318 was observed with ripening-associated genes. Among the DzERFs, the highest negative correlation was 319 found between DzERF6 and ethylene biosynthetic genes (SAM synthase, ACS, and ACO; Fig 5B). We also 320 included a member of the auxin response factor TF family (DzARF2A) in our correlation network analysis. 321 This previously characterized TF was shown to transactivate ethylene biosynthetic genes (Khaksar and 322 Sirikantaramas, 2020). A positive correlation was observed between DzERF9 and DzARF2A, whereas 323 DzERF6 was negatively correlated with DzERF2A (Fig 5B). Taking into account both the strong correlation 324 with ethylene biosynthetic genes and the pattern of expression during fruit ripening, DzERF6 and DzERF9 325 were selected as candidates for repressing and activating durian fruit ripening, respectively.   (Fig 7A), whereas for DzERF9, we observed significantly 364 higher transcript accumulation with increasing concentrations of auxin (Fig 7B). Exogenous auxin 365 treatment at 40 µΜ elicited the highest expression level of DzERF9 (Fig 7B). These results revealed the

395
In this study, based on the transcriptome data of durian fruit cv. Monthong at three different stages 396 of post-harvest ripening (unripe, midripe, and ripe), we identified 34 ripening-associated DzERFs, 397 designated DzERF1 to DzERF34. Heat map representation according to the expression levels classified 398 DzERFs into three separate clades (Fig 1). Clade I consisted of 15 members, with a decreasing expression 399 level during ripening. However, clade III comprised 16 members that were upregulated over the course of 400 ripening (Fig 1).

401
Our phylogenetic analysis clustered the 34 ripening-associated DzERFs into 15 subclades, among 402 which some DzERFs were paired with previously characterized ERFs from other fruit crops (Fig 3).
403 Increasing evidence suggests that the identification of characterized orthologues is a powerful tool to predict . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made            . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 17, 2021. ; https://doi.org/10.1101/2021.05.17.444443 doi: bioRxiv preprint