Transcriptome Analysis of Banana ( Musa acuminate L . ) in Response to Low-Potassium Stress

Hainan Key Laboratory for Sustainable Utilization of Tropical Bioresources, Institute of Tropical Agriculture and Forestry, Hainan University, No.58 Renmin Avenue, Haikou 570228, China; xmplant@163.com (M.X.); heromano@126.com (C.-B.Z.); hrnjsfdx326@163.com (R.H.); yanzhen_yz@126.com (Z.Y.); island517@126.com (Z.Q.); xruibear@163.com (R.X.); 18389595152@163.com (Y.C.); weishuangshuang@hainu.edu.cn (S.-S.W.) * Correspondence: tanghua@hainu.edu.cn; Tel.: +86-131-3891-5025


Introduction
Potassium (K + ) is one of the most abundant and important macronutrients in high-biomass plants, especially in banana plants [1]. It accounts for 2-10% of the total dry weight in plants, and plays key roles in plant growth and development processes, such as osmoregulation, photosynthetic efficiency via balancing gas exchange, pH adjustment, the cotransport of sugars, activating enzymes, and protein biosynthesis [2][3][4]. It has been reported that the concentration of K + in soil is usually below 1 mM [5]. However, K + accumulates in plant cells and can reach concentrations of 100 mM [6]. Therefore, many plants experience low −K + stress during their lives [7]. The low −K + stress signals can be transducted across the plasma membrane and into the cytosol, and the K + homeostasis is modulated by K + channels and transporters, which facilitates plant adaptation to K + -deficient conditions [8]. When faced with the K + content, dry samples were placed into chemical decomposition tubes with a small amount of water, and 8 mL of concentrated sulfuric acid was then added for overnight. The following day, the samples were incinerated in a chemical decomposition furnace at 250 • C for 30 min; then, the temperature was raised to 400 • C until a brownish-black color appeared in the solution, after about 3 h. The solution was cooled, and 10 drops of 30% H 2 O 2 were added and mixed gently before boiling for 5 min. This step was repeated three to five times with the H 2 O 2 gradually being removed until the solution became colorless. The solution was then poured into a 100-mL volumetric flask and diluted to 100 mL with double-distilled water. The K + concentration was measured with a flame photometer, with three biological and technological replicates used for K + measurement, and SPSS software (R24.0, IBM, Chicago, IL, USA) and a significant t-test (* p < 0.05, ** p < 0.01) were used to determine the statistical significance.

RNA-Seq and Bioinformatics Analysis
The RNA-Seq and bioinformatics analysis were entrusted to Shenzhen BGI Tech Company (Shenzhen, China). Six plants of each group were used for RNA extraction with an improved cetyltrimethylammonium bromide (CTAB) method [24]. The quantity and quality of the RNA were done with a spectrophotometer Ultrospec 2100 pro (Amersham Biosciences, Cambridge, UK) and 1% agarose gel electrophoresis. The extracted RNA was first treated with DNase I to degrade any possible DNA contamination. The messenger ribonucleic acid (mRNA) was enriched using oligo (dT) magnetic beads (Invitrogen) according to the manufacturer's protocol. After mixing with fragmentation buffer, the mRNA was fragmented into short fragments. The first-strand complementary DNA (cDNA) was synthesized using a random hexamer primer and the second-strand cDNA was synthesized by adding buffer, dNTPs, RNase H, and DNA polymerase I. The double-stranded cDNA was purified using magnetic beads. Then, the sequencing adaptors were ligated to the fragments. Finally, the fragments were enriched by polymerase chain reaction (PCR) amplification. The PCR products were measured by quality control (QC) step, and a 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA) and a StepOnePlus Real-Time PCR System (Applied Biosystems Inc., Foster City, CA, USA) were used to qualify and quantify sequences for the sample library. The library products were prepared for sequencing using a HiSeq 2000 system (Illumina, San Diego, CA, USA).
The original image data was transferred into sequence data via base calling, which is defined as raw data or raw reads and saved as a fastq file [25]. Those FASTQ files include the detailed read sequences and the read quality information. In addition, the filtering of raw data is needed to decrease the data noise. After filtering, the remaining reads are called "clean reads" and used for downstream bioinformatics analysis. Clean data were obtained and stored as FASTQ format after filtering by detecting the content of each base to measure the stability of the library and sequencing eligibility. Under the normal circumstances, the nucleotide distribution at each position identifies the content of stable, non-AT or GC separation. Base quality reflects the accuracy, while the sequencing, reagents, sample quality, etc., affect the quality of the base. If the percentage of the bases with low quality (<20) is low, then the sequencing quality of this lane is good. After data quality statistics, clean reads were mapped to reference sequences using Burrows-Wheeler Alignment (BWA) tool [26] and to gene reference using Bowtie [27]. In general, a higher ratio of alignment indicated a closer genetic relationship between the samples and the reference species. The genotype of the reference genome was Musa acuminata (DH-Pahang), and the latest version was improved by a combination of methods and datasets, leading to the release of version 2 of the assembly and gene annotation (https://banana-genome-hub.southgreen.fr) [23]. The process of RNA-Seq and bioinformatics analysis are presented in Figure 1. The raw data of the high-throughput sequence (RNA-Seq) results and processed files have been submitted to the Gene Expression Omnibus (GEO) database of the national center for biotechnology information (NCBI) web site; its accession number is GSE102968.

Differentially Expressed Genes (DEGs) Screening
In this article, we used RNA-Seq by the FPKM (Fragments Per Kilobase of transcript per Million mapped reads) method to calculate the expression level of the detected genes with the RNASeq by Expectation Maximization (RSEM) software package [28]. The FPKM method is able to eliminate the influence of different gene lengths and sequencing discrepancy on the calculation of gene expression. The formula of the FPKM method is shown as follows: FPKM = (10 6 C)/(NL/10 3 ), which is given to be the expression of gene A; here, C is the number of fragments that are uniquely aligned to gene A, N is the total number of fragments that are uniquely aligned to all genes, and L is the number of bases on gene A. In order to identify the low potassium-responsive genes, the DEGs were analyzed between the normal and treatment groups using P-value <0.05 and |log2 Ratio| ≥1 as thresholds. The significance of the DEGs was determined by using the false discovery rate (FDR) control method to justify the P-value [29].

Quantitative Real-Time PCR (qRT-PCR) Analysis
To validate the reliability and repeatability of the RNA-Seq results, 15 genes were selected for identification by qRT-PCR assays (three technological replicates of each gene). Detailed

Differentially Expressed Genes (DEGs) Screening
In this article, we used RNA-Seq by the FPKM (Fragments Per Kilobase of transcript per Million mapped reads) method to calculate the expression level of the detected genes with the RNASeq by Expectation Maximization (RSEM) software package [28]. The FPKM method is able to eliminate the influence of different gene lengths and sequencing discrepancy on the calculation of gene expression. The formula of the FPKM method is shown as follows: FPKM = (10 6 C)/(NL/10 3 ), which is given to be the expression of gene A; here, C is the number of fragments that are uniquely aligned to gene A, N is the total number of fragments that are uniquely aligned to all genes, and L is the number of bases on gene A. In order to identify the low potassium-responsive genes, the DEGs were analyzed between the normal and treatment groups using p-value < 0.05 and |log2 Ratio| ≥ 1 as thresholds. The significance of the DEGs was determined by using the false discovery rate (FDR) control method to justify the p-value [29].

Quantitative Real-Time PCR (qRT-PCR) Analysis
To validate the reliability and repeatability of the RNA-Seq results, 15 genes were selected for identification by qRT-PCR assays (three technological replicates of each gene). Detailed information about these genes is provided in Supplemental Material 2. Tissues of six banana seedlings of each group were mixed and used for RNA extraction with the improved cetyltrimethylammonium bromide (CTAB) method [24]. Total RNA was treated with DNase I and RNase-free (Thermo Fisher Scientific, Waltham, MA, USA) to eliminate genomic DNA contamination. For first-strand cDNA synthesis, 1.0 µg of total RNA was used with random and oligo(dT) 18 primers (Revert Aid First Strand cDNA Synthesis Kit, #K1621, Thermo Fisher Scientific, Waltham, MA, USA). SYBR ® Premix Ex Taq II (Takara, Dalian, China) was used in the PCR reaction system on a 7500 RT-PCR System (Applied Biosystems, Waltham, MA, USA) according to the manufacturer's protocols. The qRT-PCR was performed using a 20-µL reaction system containing 10 µL of SYBR ® Premix Ex Taq II (Takara, Tli RNase Plus, Dalian, China) (2×), 2 µL of cDNA template, 0.8 µL of PCR forward primer (10 µM), 0.8 µL of PCR reverse primer (10 Agronomy 2019, 9, 169 5 of 20 µM), 0.4 µL of ROX reference DyeII (50×), and 6 µL of nuclease-free water. The PCR amplification program was as follows: 50 • C for 2 min, 95 • C for 10 min, followed by 40 cycles of 95 • C for 15 s, and 60 • C for 1 min. The 2 −∆∆CT method was used for qPCR data analysis, in which ∆C T(test) = C T(target, test) − C T(ref, test) , ∆C T(calibrator) = C T(target, calibrator) − C T(ref, calibrator) , and ∆∆C T = ∆C T(test) − ∆C T(calibrator) , and the expression ratio of the experimental and control groups was 2 −∆∆CT . The , primers of the selected genes were designed using Primer Premier 6 (www.premierbiosoft.com). The amplification of Actin1 (GenBank Accession No. AF246288, submitted by National Taiwan University, Taipei, China) was used as an internal reference gene control for data normalization [30].

K + Deficiency Inhibits the Growth of Banana Seedlings
In the nutriculture experiments with low −K + stress, at the early stage (20 days), the edges of the leaves of the banana plants in the low −K + group (LK) were slightly yellow, as though they had been burned, compared with those of the normal −K + group (NK). However, there were no obvious differences in the phenotypic characteristics of plant height or number of leaves (Table1). After 60 days, the lower leaves of the seedlings in the LK group became yellow, with brown stains, as though boiling water had been poured over them (the red arrow in Figure 2A). Then, the lower leaves became wilted, and the total number of leaves was less than that in the NK group ( Figure 2B). The LK plants were thin and short, with a small leaf area and fewer roots ( Figure 2C). These symptoms indicated that stress due to K + deficiency dramatically inhibited the growth and development of banana seedlings.
To investigate the changes between the NK and LK treatments, each plant was divided into an aerial part (leaves and pseudostems) and an underground part (roots). The phenotypic traits of each group were measured, which contained 20 banana seedlings. The phenotypic traits including leaf numbers, plant height, and fresh and dry weights. Spearman analysis was used for correlation analysis of the statistical variables with SPSS software. The results are shown in Table 1.
There were significant differences between the NK and LK treatments in terms of total leaf number and the number of leaves showing symptoms. The fresh and dry weights of the aerial and underground parts in the LK group were significantly less than those in the NK group. This indicated that low −K + stress significantly inhibited the growth and development of the banana leaf, pseudostem, and root, and led to a decrease in the biomass of banana plants. We also measured the K + content of the aerial and underground parts of the plants in the NK (NKL and NKR) and LK (LKL and LKR) treatments, respectively. SPSS software was applied to the results (Figure 3), which confirmed that the K + content in the LK group was significantly lower than that in the NK group, indicating that a restricted K + supply affected K + uptake and K + content in banana tissue. In conclusion, the K + content and biomass index of banana seedlings decreased under the stress of low −K + conditions.   . K + content of plants under different treatments: normal -K + leaves (NKL) and low -K + leaves (LKL) and normal -K + roots (NKR) and low -K + roots (LKR). Error bars with different letters indicate significant differences (three technical replicates, p < 0.05).

RNA Sequencing, Alignment, and Bioinformatics Analysis in Banana Leaves and Roots
To investigate the differences in gene expression between the normal -K + (NK) and low -K + (LK) treatments in the leaves and roots of banana plants, NKL, LKL, NKR, and LKR samples were selected for RNA-Seq. From the four cDNA libraries, a total of 48, 767, and 309 raw reads were generated (Table 2). After the removal of low quality sequences and short  Figure 3. K + content of plants under different treatments: normal −K + leaves (NKL) and low −K + leaves (LKL) and normal −K + roots (NKR) and low −K + roots (LKR). Error bars with different letters indicate significant differences (three technical replicates, p < 0.05).  The genes that were only expressed in NKL, LKL, NKR, or LKR were defined as specifically expressed genes (SEGs). The number of specifically expressed genes in the LK samples was greater than that in the NK samples ( Figure 5A); the low −K + treatment resulted in an increase in the number  Figure 5B). There were 266 differentially up-reglated genes and 33 differentially down-regulated genes that were common to both leaves and roots ( Figure 6). The statistical analysis suggested that these genes may be expressed conservatively in banana and play important roles under low −K + conditions. All of the DEGs were mapped to the terms of the gene ontology (GO) database (http://www.geneontology.org), which calculated the gene numbers of each term. Then, the hypergeometric test was used to find the enriched GO terms of DEGs compared to the whole genome. P-value was calculated by Gene Ontology TermFinder, and corrected by Bonferroni. GO terms whose corrected p-value ≤ 0.05 were defined as prominent enriched GO terms. In leaves, the GO analysis identified 34, 122, and 197 terms, which contained 1472, 1760, and 2495 genes relevant to cellular components, molecular functions, and biological processes, respectively. In roots, there were 100, 224, and 404 terms, which contained 8483, 9038, and 12,541 genes relevant to cellular components, molecular functions, and biological processes, respectively. Genes that had a cluster frequency ≥ 20% were selected and compared among all of the GO categories in leaves and roots (Figure 7).
In all of the samples of leaves and roots, the GO terms for cells and cell parts were highly represented in the cellular components, catalytic activity was highly represented in the molecular functions, and metabolic processes were highly represented in the biological processes. Interestingly, some GO terms were represented in roots, but not in leaves. These data indicated that the DEGs for cells, catalytic activity, and metabolic processes were most relevant to low -K + stress. All of the DEGs were mapped to the terms of the gene ontology (GO) database (http://www. geneontology.org), which calculated the gene numbers of each term. Then, the hypergeometric test was used to find the enriched GO terms of DEGs compared to the whole genome. p-Value was calculated by Gene Ontology TermFinder, and corrected by Bonferroni. GO terms whose corrected p-value ≤ 0.05 were defined as prominent enriched GO terms. In leaves, the GO analysis identified 34, 122, and 197 terms, which contained 1472, 1760, and 2495 genes relevant to cellular components, molecular functions, and biological processes, respectively. In roots, there were 100, 224, and 404 terms, which contained 8483, 9038, and 12,541 genes relevant to cellular components, molecular functions, and biological processes, respectively. Genes that had a cluster frequency ≥20% were selected and compared among all of the GO categories in leaves and roots (Figure 7).
In all of the samples of leaves and roots, the GO terms for cells and cell parts were highly represented in the cellular components, catalytic activity was highly represented in the molecular functions, and metabolic processes were highly represented in the biological processes. Interestingly, some GO terms were represented in roots, but not in leaves. These data indicated that the DEGs for cells, catalytic activity, and metabolic processes were most relevant to low −K + stress.   The Ca 2+ ion is a ubiquitous second messenger in plant signal transduction in response to various biotic and abiotic stresses and developmental stimuli [39]. Plants recognize the Ca 2+ signal through Ca 2+ sensors, such as calmodulins, calmodulin-like proteins (CMLs), CBLs, and Ca 2+ -dependent protein kinases (CDPKs) [40]. Calcium ion sensors (e.g., CBLs) and their targets (e.g., CIPKs) interact with each other as a CBL-CIPK Ca 2+ complex, which activates gene expression downstream of the signal transduction pathway in response to stress conditions [5,41]. It has been reported that the CBL-CIPK complex is related to K + uptake by the modulation of AKT channels in rice and Arabidopsis thaliana [16,18]. Two CBL/CIPK pairs were identified in grape vine, which could activate the inward K + channel VvK1.2 [42]. Based on a KEGG pathway analysis and GO function analysis, three up-regulated (Ma07_g17570, Ma07_g05310, and Ma09_g04710) CIPK DEGs and one down-regulated (Ma11_g04530) CIPK DEG were detected in leaves, and nine up-regulated (Ma06_g34240, Ma06_g28510, Ma07_g23630, Ma01_g17400, Ma04_g29510, Ma07_g29120, Ma03_g14540, Ma04_g15200, and Ma08_g10700) CIPK DEGs and one down-regulated (Ma06_g26140) CIPK DEG were detected in roots in this study. We conducted a phylogenetic analysis of these CBL family DEGs in banana leaves and roots by MEGA7.0 ( Figure 9). Under low -K + conditions, these DEGs participate in K + uptake and transport in banana to a substantial extent. The Ca 2+ ion is a ubiquitous second messenger in plant signal transduction in response to various biotic and abiotic stresses and developmental stimuli [39]. Plants recognize the Ca 2+ signal through Ca 2+ sensors, such as calmodulins, calmodulin-like proteins (CMLs), CBLs, and Ca 2+ -dependent protein kinases (CDPKs) [40]. Calcium ion sensors (e.g., CBLs) and their targets (e.g., CIPKs) interact with each other as a CBL-CIPK Ca 2+ complex, which activates gene expression downstream of the signal transduction pathway in response to stress conditions [5,41]. It has been reported that the CBL-CIPK complex is related to K + uptake by the modulation of AKT channels in rice and Arabidopsis thaliana [16,18]. Two CBL/CIPK pairs were identified in grape vine, which could activate the inward K + channel VvK1.2 [42]. Based on a KEGG pathway analysis and GO function analysis, three up-regulated (Ma07_g17570, Ma07_g05310, and Ma09_g04710) CIPK DEGs and one down-regulated (Ma11_g04530) CIPK DEG were detected in leaves, and nine up-regulated (Ma06_g34240, Ma06_g28510, Ma07_g23630, Ma01_g17400, Ma04_g29510, Ma07_g29120, Ma03_g14540, Ma04_g15200, and Ma08_g10700) CIPK DEGs and one down-regulated (Ma06_g26140) CIPK DEG were detected in roots in this study. We conducted a phylogenetic analysis of these CBL family DEGs in banana leaves and roots by MEGA7.0 ( Figure 9). Under low −K + conditions, these DEGs participate in K + uptake and transport in banana to a substantial extent.

Transcription Factors (TFs) Analysis in Banana Leaves and Roots
Transporters and channels have important roles in the accumulation of nutrient elements from the soil by plants, and in the maintenance of homeostasis. How the transporters and channels are regulated by upstream transcription factors is of critical importance. Transcription factors are transcriptional regulators in plants and form an integral part of the signaling networks that modulate many plant processes. Transcription factors are proteins that bind to specific DNA sequences or interact with other transcription factor proteins, thereby controlling the transcription rate of genetic information from DNA to mRNA. Previous studies have shown that MYB-type, bHLH, AP2/ERF, NAC, and WRKY transcription factors play important roles in regulating K + , Na + , and phosphate (PO4 3− ) transport and salt tolerance in rice and Arabidopsis thaliana [43][44][45][46][47][48][49]. Our results revealed that genes encoding MYB, MYB-related, bHLH, AP2-EREBP, WRKY, and NAC transcription factors comprised the top six classes of transcription factor genes differentially expressed in banana leaves and roots ( Figure 10). These transcription factors may be among the most critical factors affecting the growth and development of banana seedlings under K + -deficient conditions.

Transcription Factors (TFs) Analysis in Banana Leaves and Roots
Transporters and channels have important roles in the accumulation of nutrient elements from the soil by plants, and in the maintenance of homeostasis. How the transporters and channels are regulated by upstream transcription factors is of critical importance. Transcription factors are transcriptional regulators in plants and form an integral part of the signaling networks that modulate many plant processes. Transcription factors are proteins that bind to specific DNA sequences or interact with other transcription factor proteins, thereby controlling the transcription rate of genetic information from DNA to mRNA. Previous studies have shown that MYB-type, bHLH, AP2/ERF, NAC, and WRKY transcription factors play important roles in regulating K + , Na + , and phosphate (PO 4 3− ) transport and salt tolerance in rice and Arabidopsis thaliana [43][44][45][46][47][48][49]. Our results revealed that genes encoding MYB, MYB-related, bHLH, AP2-EREBP, WRKY, and NAC transcription factors comprised the top six classes of transcription factor genes differentially expressed in banana leaves and roots ( Figure 10). These transcription factors may be among the most critical factors affecting the growth and development of banana seedlings under K + -deficient conditions.

Validation of Gene Expression by qRT-PCR
To validate the reliability and repeatability of gene expression in RNA-Seq, a total of 15 genes were selected randomly for identification by qRT-PCR assays (three technical replicates of each gene). The results of the qRT-PCR analysis for these genes mostly matched the expression patterns found by RNA-Seq analysis (Figure 11), which indicated that the RNA-Seq results were reliable.

Validation of Gene Expression by qRT-PCR
To validate the reliability and repeatability of gene expression in RNA-Seq, a total of 15 genes were selected randomly for identification by qRT-PCR assays (three technical replicates of each gene). The results of the qRT-PCR analysis for these genes mostly matched the expression patterns found by RNA-Seq analysis (Figure 11), which indicated that the RNA-Seq results were reliable.

Discussion
Potassium is the most important and abundant cation in plant cells, playing critical roles in plant physiological function [50]. Reports showed that potassium mainly participates in the electrical neutralization of inorganic and organic anions and macromolecules, pH homeostasis, the control of membrane electrical potential, the regulation of cell osmotic pressure, optimal protein synthesis, and photosynthesis [51]. In addition, as cofactors for many enzymes, K ions participate in many metabolic processes, such as nitrogen (N), sulfur (S), and phosphorus (P) assimilation, pyruvate synthesis, and sugar metabolism [52]. Potassium is taken up by the plant root through the epidermal and cortical cells, and then transported to the shoot and distributed to the leaves [53].
In general, there are two primary signaling pathways that play important roles in regulating K uptake, which include the CBL-CIPK network and ethylene-mediated pathways [54,55]. In the CBL-CIPK pathway, the Ca 2+ sensor proteins (CBLs) interact with their target kinases (CIPKs). The balance of the cations was regulated by a number of transporting proteins that are involved in the K + and Na + uptake and translocation [31]. The report showed that both ethylene production and the transcription of genes of the ethylene biosynthesis increased when met with low potassium stress [56]. Reactive oxygen species (ROS) have been suggested to be upstream regulators of calcium signaling and participate in low potassium-signaling pathways [57,58]. In addition, ethylene acts upstream of the ROS in response to potassium deprivation by increasing the H 2 O 2 concentrations [55]. Therefore, it is necessary to clarify the molecular mechanisms of K ions' uptake and transport in the CBL-CIPK network and ethylene-mediated pathways, especially in response to low −K + conditions. However, the regulation of potassium in plants is a complicated process that is coordinated with Ca, Na, N, and S ions [6,52]. As is well known, K + absorption and translocation are mainly mediated by plant K + transporters and channels [59]. At present, there are mainly three families of ion channels (Shaker, TPK/KCO, and TPC) and three families of transporters (HAK, HKT, and CPA) have been identified as contributing to K and Na ions transport across the plasmalemma and internal membranes [58,60]. Potassium transporters may function in both low-affinity and high-affinity transport as members of the KT/KUP/HAK family [31]. It has been reported that the protein kinase CIPK9 interacts with the calcium sensor CBL3 and plays an important role in K + homeostasis under low-potassium stress in Arabidopsis [59]. In vivo and vitro experiments showed that the CBL1, CBL8, CBL9, and CBL10, together with a CBL-interacting serine/threonine protein kinase23 (CIPK23) activated a low K + inducible potassium transporter HAK5 to regulate K + transport in Arabidopsis [61]. In this data, we also identified 14 differentially expressed CBL-interacting protein kinases genes (Figures 8 and 9) in a low-potassium group of banana plants. Among these genes, two genes (Ma06_g28510 and Ma09_g04710) were annotated to CBL-interacting serine/threonine protein kinases, which may play crucial roles in protein phosphorylation during low-potassium starvation.
Transcription factors (TFs) determine the temporal and spatial features of gene expression by binding to specific promoter sequences that comprise cis-regulatory elements [62]. Treated with low-potassium stress, significantly differentially expressed banana TFs about the MYB and MYB-related family genes were the largest subgroup of all the TFs in banana. It was reported that the AP2/ERF transcription factor RAP2.11 was a component in response to low potassium through regulation of the low-potassium signal transduction pathway via the high-affinity K + uptake transporter AtHAK5, which was bound to a GCC-box of the AtHAK5 promoter [44]. At present, a total number of 112 DEGs about the AP2-EREBP transcription factors were identified, of which 25 of the DEGs were in banana leaves and 87 of the DEGs were in banana roots. Transcriptome analysis is a common but excellent technological means to find genes that have participated in a variety of biotic or abiotic stresses in plants. Using RNA-Seq, this study provides important transcriptome information for banana (Musa acuminata L. AAA group, cv. Cavendish) in response to K + starvation. The transcriptional profiles of banana genes have revealed that many genes are differentially regulated in leaves and roots after low potassium stress. Our data showed important genes that were related to K + transport and uptake, including genes of the HAK/KUP/KT system, AKT family, CBLs, and so on. These genes were differentially expressed between LK and NK treatments, especially in roots.
Here, we compared the mRNA expression levels of different banana tissues in low-potassium conditions. The expression patterns of mRNA among different tissues revealed that many important DEGs might regulate potassium expression at the transcriptional level. At present, transgenic research is the best method to clarify the functions of a gene. The successful acquisition of transgenic Cavendish bananas with resistance to fusarium wilt tropical race 4 (Foc4) brings a bright future for the study of genes response for K + stress [63]. Many DEGs encoding protein kinases and ion transporters were simultaneously down/up-regulated in our data, suggesting that they may participate in the responses of banana to K + deficiency. The data obtained in this study may enable a better understanding of the mechanisms controlling banana K + -deficiency tolerance and will be useful in informing studies of K + deficiency in other plant species. However, further research is required to validate the functions of the DEGs related to K + stress.

Conclusions
The results of this study suggested that lots of DEGs related to K + homeostasis participated in the mechanisms for low-potassium tolerance in banana. Although the genome of banana has been sequenced and assembled, genetic expression on the transcription level under low-potassium stress has not been reported in banana. This study main focused on analyzing the DEGs via the bioinformatics method and qRT-PCR assay. Our report provided a basic information for genes regarding transcription levels, and could be used to better understand the function of these genes under low-potassium tolerance.