Genome-wide systematic characterization of the HAK/KUP/KT gene family and its expression profile during plant growth and in response to low-K+ stress in Saccharum

Background Plant genomes contain a large number of HAK/KUP/KT transporters, which play important roles in potassium uptake and translocation, osmotic potential regulation, salt tolerance, root morphogenesis and plant development. Potassium deficiency in the soil of a sugarcane planting area is serious. However, the HAK/KUP/KT gene family remains to be characterized in sugarcane (Saccharum). Results In this study, 30 HAK/KUP/KT genes were identified in Saccharum spontaneum. Phylogenetics, duplication events, gene structures and expression patterns were analyzed. Phylogenetic analysis of the HAK/KUP/KT genes from 15 representative plants showed that this gene family is divided into four groups (clades I-IV). Both ancient whole-genome duplication (WGD) and recent gene duplication contributed to the expansion of the HAK/KUP/KT gene family. Nonsynonymous to synonymous substitution ratio (Ka/Ks) analysis showed that purifying selection was the main force driving the evolution of HAK/KUP/KT genes. The divergence time of the HAK/KUP/KT gene family was estimated to range from 134.8 to 233.7 Mya based on Ks analysis, suggesting that it is an ancient gene family in plants. Gene structure analysis showed that the HAK/KUP/KT genes were accompanied by intron gain/loss in the process of evolution. RNA-seq data analysis demonstrated that the HAK/KUP/KT genes from clades II and III were mainly constitutively expressed in various tissues, while most genes from clades I and IV had no or very low expression in the tested tissues at different developmental stages. The expression of SsHAK1 and SsHAK21 was upregulated in response to low-K+ stress. Yeast functional complementation analysis revealed that SsHAK1 and SsHAK21 could rescue K+ uptake in a yeast mutant. Conclusions This study provided insights into the evolutionary history of HAK/KUP/KT genes. HAK7/9/18 were mainly expressed in the upper photosynthetic zone and mature zone of the stem. HAK7/9/18/25 were regulated by sunlight. SsHAK1 and SsHAK21 played important roles in mediating potassium acquisition under limited K+ supply. Our results provide valuable information and key candidate genes for further studies on the function of HAK/KUP/KT genes in Saccharum.


Background
Potassium is an essential mineral nutrient for plant growth and development and is also the most abundant monovalent cation in plants, accounting for approximately 2~10% of plant dry weight [1]. Potassium is involved in many important physiological and biochemical processes, such as cell turgor regulation, cell charge balance regulation, enzyme activity regulation and protein synthesis [1]. Symptoms of plant potassium deficiency usually manifest as weak stems, easy lodging, decreased tolerance to drought and cold and yellow leaves, due to the degradation of proteins and chlorophyll, which leads to tissue necrosis [2]. Thus, potassium is of great importance for improving crop yield and quality. Sugarcane is an important sugar and energy crop with a long growth period, large biomass and large amount of potassium fertilizer absorption. On the one hand, it is estimated that sugarcane needs to absorb approximately 2~2.5 kg of potassium to produce one ton of sugar [3,4]. On the other hand, sugarcane is mainly cultivated in subtropical and tropical regions, where soil acidification and potassium leaching are common. The contents of total potassium and available potassium in the cultivated layer of these sugarcane areas are low.
Plant cells maintain a relatively high and stable K + concentration (approximately 100~150 mM) in the cytosol, while the K + concentration is highly variable in the range of 0.01~1 mM [5]. It is generally believed that there are two mechanisms for potassium uptake by plants, namely, a high-affinity transport system (HATS) via potassium transporters at low external potassium concentrations (< 0.2 mM) and a low-affinity transport system (LATS) via potassium channels at high potassium concentrations (> 0.5 mM) [6,7]. According to their structure and function, potassium transporters in plants can be divided into five families: (1) Shaker channels; (2) TPK (tandem-pore K + ) channels; (3) HAK (high-affinity K + transporter)/KUP (K + uptake permease)/KT (K + transporter); (4) HKT transporters; and (5) CPAs (cation-proton antiporters) [2,8]. Among them, the HAK/ KUP/KT family is the largest and is widely distributed in bacteria, fungi and plants but has not been identified in animal cells [9].
According to their homology with bacterial KUP and fungal HAK transporters [10], the plant HAK/KUP/KT transporter members AtKUP1 and HvHAK1 were first cloned from Arabidopsis and barley [11,12]. Both genes could complement K + uptake-deficient strains of yeast, indicating that they had potassium transporter activity. Subsequently, several HAK/KUP/KT members were cloned and identified, such as AtKUP3 and AtHAK5 in Arabidopsis, OsHAK1 in rice and CaHAK1 in pepper, which were also shown to be highly compatible potassium transporters [13][14][15][16]. Based on comparative genomic methods, 13, 27 and 27 HAK/KUP/KT genes were identified in Arabidopsis, rice and maize, respectively [17][18][19]. These predicted HAK/KUP/KT transporters were sorted into four clusters. HAK/KUP/KT K + transporters play versatile roles in potassium ion acquisition and transport, salt stress, osmotic regulation, and the morphogenesis of root and phenotype in plants [7]. The expression of OsHAK1 was greatly induced in the roots of K + -starved rice, while OsHAK5 was less expressed in roots but abundantly expressed in shoots [20,21]. Some ions, particularly Na + and NH 4 + , can have additional effects on the expression of HAK/KUP/KT genes [22,23].
The transcriptional regulation of HAK/KUP/KT K + transporters is a universal mechanism by which different plant species respond to K + -starvation stress [8]. The HAK/KUP/KT genes in clade I, such as AtHAK5, OsHAK1, CaHAK1 and ThHAK5, display low expression levels both in roots and shoots under control conditions and are highly upregulated in roots upon K + -deficiency stress [12][13][14]16]. While the HAK/KUP/KT K + transporters in other three clades exhibit different expression patterns [24], since transcription of most K + transporters are not induced by K + starvation [25]. In Arabidopsis, several transcription factors, including bHLH121 (basic helix-loop-helix 121), DDF2 (dwarf and delayed flowering 2), JLO (jagged lateral organs) and TFII_A (transcription initiation factor II_A gamma chain), have been identified to bind the promoters of HAK5 and activate its expression under low K + stress [26]. Activation of HAK/KUP/KT K + transporters is also regulated at posttranscriptional and/or posttranslational level. AtHAK5 and its homologs from pepper and tomato can be activated by the CIPK23 (CBL-interacting protein kinase 23)/CBL (calcineurin B-like protein) complex [27].
In summary, numerous studies have been performed in the functional research of plant HAK/KUP/KT potassium transporters, and important progress has been made. However, the known functional HAK/KUP/KT genes have mainly been identified in a few plants, such as Arabidopsis, rice and maize, but their physiological functions and regulatory mechanisms in sugarcane remain unknown. In this study, based on the newly released S. spontaneum genome [28], we identified the HAK/KUP/KT gene family in S. spontaneum. Phylogenetic relationships among different species, exon/intron organization and gene expression were analyzed. Altogether, these results provide valuable information and robust candidate genes for future functional analyses for the genetic improvement of potassium-utilization efficiency in sugarcane.

Identification of HAK genes in sugarcane
Based on comparative genomics, 29 SbHAK genes were identified from sorghum (Sorghum bicolor, sugarcane's nearest relative). Using the protein sequences of sorghum HAK genes as a reference, 30 distinct S. spontaneum HAK genes (Table 1), excluding alleles, were identified from the genome of tetraploid S. spontaneum AP85-441 [28]. Each of these genes contained one to four alleles, with an average of 3 (Additional file 1). The 30 SsHAK genes were distributed on seven S. spontaneum chromosomes: chromosome 1 contained six genes; chromosome 2 contained seven genes; chromosome 3 contained four genes; chromosome 4 contained two genes; chromosome 5 contained five genes; and chromosome 6 and 8 each contained three genes. No SsHAK genes were identified on chromosome 7 (Additional file 1).
All 30 predicted SsHAK proteins had a typical "K_trans" domain (PF02705), which is specific to HAK/KUP/KT potassium transporter family members. For consistency, these SsHAK genes were named based on the previously reported O. sativa HAK nomenclature and phylogenetic relationships [17]. If two SsHAK genes were equally close to a single OsHAK gene, then the same name was used, followed by the letters "a" and "b" (Table 1). Two paralogous SsHAK genes (SsHAK19a and SsHAK19b) were identified that corresponded to the same sorghum gene, Sobic.006G062100, which may be caused by gene loss in sorghum or gene duplication in sugarcane. The number of amino acids in the 30 identified SsHAKs ranged from 487 to 967, with an average of 758. The predicted isoelectric points (pI) of the SsHAKs varied from 5.88 to 9.26, and the average pI was 8.15. The molecular weight ranged from 55.84 kDa to 106.49 kDa, with an average of 84.47 kDa ( Table 1). The prediction of transmembrane domains in the SsHAK proteins indicated that most contained 11 or 12 transmembrane helices, which was similar to the findings in sorghum. The subcellular locations of the SsHAK proteins predicted by WoLF PSORT were mainly the plasma membrane, which is most suitable for their roles as transporters to maintain K + homeostasis in sugarcane. In addition, the SsHAK proteins were also located on some organelles, including the endoplasmic reticulum, vacuole, cytoplasm, Golgi body and chloroplast. Protein sequence alignment of SsHAKs with their orthologs in sorghum showed that S. spontaneum and Sorghum bicolor shared identities ranging from 81 to 98%, with an average of 92.5% (Table 1). Four hundred thirty-five pairwise protein sequence comparisons among these SsHAKs showed that SsHAK19a and SsHAK19b shared the highest identity (96%), while other gene pairs had protein sequence similarities ranging from 28 to 82% with an average of 46%, indicating that the SsHAKs are an ancient gene family with high sequence divergence (Additional file 2). The nonsynonymous to synonymous substitution ratios (Ka/Ks) between SsHAKs and their orthologous genes in sorghum were calculated to study the evolutionary functional constraints in sugarcane. The results showed that the Ka/Ks ratios were less than 0.5, except for SsHAK13, suggesting that purifying selection was the main force driving the evolution of HAK genes (Fig. 1).

Phylogenic analysis of HAK genes in S. spontaneum and representative angiosperms
To analyze the evolution of the HAK gene family in S. spontaneum and different plants, a total of 278 HAK genes from 14 representative angiosperms and a HAK member from Chlamydomonas reinhardtii as the outgroup were used to construct a phylogenetic tree using the neighbor-joining method ( Fig. 2  These HAK genes could be divided into four clades (I, II, III, IV) based on previously reported OsHAKs [17]. In A. trichopoda, the earliest diverging angiosperm, there were only 6 HAK genes, while in dicots and monocots, the number of HAKs ranged from 8 to 30 (Figs. 2 and 3), indicating that the ancient whole-genome duplication (WGD) contributed to the expansion of the HAK gene family in both dicots and monocots. Clade II and clade III included HAK genes from all 14 angiosperm genomes, indicating that the progenitors of these genes may have already existed prior to the split from angiosperms (Figs. 2 and 3). Clade I and clade IV mainly contained HAK genes from monocotyledons. Eightythree HAK genes were identified in clade I, in which only one HAK gene was from A. comosus (Aco006685, homologous with SsHAK5) and Arabidopsis (AtHAK5), and the other 81 HAK genes were from all eight examined Poaceae species (Figs. 2 and 3). Twenty-nine HAKs were grouped into clade IV, and only 2 of them were from dicotyledons. These results indicated that the HAKs were unevenly distributed.
According to the Ks value in sorghum and sugarcane (Additional file 5), the divergence time of four clusters of HAKs was estimated. The median value of Ks was between 1.644 and 2.851, and its corresponding divergence time was between 134.8 and 233.7 Mya, indicating that the HAK was an ancient and divergent family. Furthermore, two pairs of duplicated SsHAKs (SsHAK5a/5b and SsHAK16a/16b) diverged at 18.94 and 58.14 Mya (Additional file 6). These results suggested that the SsHAK family is an ancient gene family with recent gene duplication events.
Exon/intron organization of the HAK family in S. spontaneum and other angiosperms To investigate the structural characteristics and evolution of the HAK gene family, the exon/intron organization in The exon number of SsHAKs varied from 2 to 12, and half of the SsHAKs possessed 8 or 9 exons. The pattern of SsHAK gene structure was similar to that of HAK gene structure from sorghum and maize in the same clade, suggesting that the HAK gene structure in the Panicoideae was relatively conserved. In clade I, the exon number in HAK genes varied from 2 to 12, which was the most variation among these 4 clades. Notably, the HAK genes in the subfamily with SsHAK22 had only 2 to 4 exons; however, the protein size remained consistent, which was likely due to the loss of introns. Clade II had the largest number of HAK genes, with 60 out of 98 HAKs possessing 9 exons and 5 out of 9 SsHAKs harboring 8 exons. SsHAK3/8/10 had one less exon than their orthologous genes in sorghum; the first exon in SsHAK13 and seventh exon in SsHAK24 were smaller than the corresponding exons in sorghum, and both resulted in shorter amino acid sequences in S. spontaneum (Table 1, Fig. 2). In clade III, the exon number was relatively conserved, with 61 out of 68 HAK genes possessing 8 to 10 exons, while the gene size varied greatly, mainly due to the different sizes of introns. The exon number in clade IV ranged from 2 to 8, with an average of 7, which was smaller than that in other clades. Notably, the HAK genes in the subfamily with SsHAK4 had only 2 to 5 exons, which was likely caused by intron loss during the process of evolution. The results indicated that HAKs underwent gene structure reconstruction under different evolutionary dynamics in S. spontaneum and other angiosperms in this study.

Expression analysis of HAK genes in Saccharum species
To study the expression profiles and potential functions of HAKs in Saccharum, we compared the gene expression patterns according to 4 sets of RNA-seq data: 1) different developmental stages and tissues; 2) a leaf gradient; 3) the circadian rhythm; and 4) treatment under low-potassium stress. The FPKM values of HAK1, HAK7 and HAK20b in YT55 at 0 h, 6 h, 12 h, 24 h, 48 h and 72 h under K + -starvation conditions were verified by RT-qPCR. The relative expression level was positively correlated with the FPKM value (R 2 = 0.8419, Additional file 9), suggesting the reliability of gene expression based on the RNA-seq analysis.

Expression pattern of HAKs in different tissues at different stages
To study gene functional divergence among the Saccharum species, transcriptome profiles of HAKs between two Saccharum species, S. officinarum and S. spontaneum, were analyzed based on RNA-seq at three developmental stages (seedling, premature and mature stages) in five different tissues, 2 leaf (leaf and leaf roll) and 3 stalk (immature, maturing and mature) tissues (Fig. 4). Among the 30 HAK genes analyzed, 18 genes (HAK3/4/ 5a/5b/12/13/14/15/16a/16b/17/19a/19b/20a/20b/21/22/ 26) showed very low or undetectable expression levels in all examined tissues of the two Saccharum species. HAK1 and HAK2 had different expression patterns in the two Saccharum species. HAK1 had higher expression levels in S. spontaneum than in S. officinarum, and the expression level in leaves was higher than that in stems   at three different stages. HAK2 had higher expression levels in S. officinarum than in S. spontaneum, and the expression level in stems was higher than that in leaves.
HAK8 was mainly expressed in the upper stems, while the expression levels in the middle and lower stems were very low. HAK9 and HAK10 had higher expression levels in stems than in leaves. HAK18 was expressed in all examined tissues, with higher expression levels, especially in leaves at the seedling stage and in mature stems. Notably, HAK27 was highly expressed in leaves at all examined three stages, but the expression level in stems was very low or undetectable.

Expression pattern of HAKs across a leaf segment gradient
To further explore the functional divergence of HAK genes for photosynthesis in the source tissues, we studied the expression pattern of HAKs in continuously developing leaf segment gradients from S. officinarum and S. spontaneum (Fig. 5). Saccharum leaves were divided into four zones: the basal zone (sink tissue), transitional zone (undergoing sink-source transition), maturing zone and mature zone (fully differentiated zone with active photosynthesis), following the method described in maize [30]. Consistent with the expression pattern at different developmental stages, 18 HAK genes (HAK3/4/ 5a/5b/12/13/14/15/16a/16b/17/19a/19b/20a/20b/21/22/ 26) showed very low or undetected expression levels in all examined leaf segments, suggesting their limited roles in sugar transport (Fig. 5). HAK1 and HAK2 showed higher expression levels in the basal zone than in the other 3 zones. The expression level of HAK7 increased gradually from the base to the tip of the S. spontaneum leaf, while in S. officinarum, HAK7 displayed higher expression levels in the maturing zone than in the other 3 zones. The expression level of HAK8 decreased gradually from the base to the tip of the leaf in both S. officinarum and S. spontaneum. HAK9 showed different expression patterns in S. spontaneum and S. officinarum. In S. spontaneum, the expression level of HAK9 increased gradually from the basal zone to the maturing zone and then decreased in the mature zone. In S. spontaneum, the expression level of HAK9 decreased from the transition zone to the maturing zone and then increased in the mature zone, and the expression level was much higher in S. officinarum, suggesting gene functional divergence after the split of these two Saccharum species. HAK10 showed higher expression levels in the transition zone in S. spontaneum and higher expression levels in the mature zone in S. officinarum. HAK18 displayed higher expression levels in the maturing zone in both S. spontaneum and S. officinarum, while HAK23 showed higher expression levels in the basal zone in the two Saccharum species. HAK25 displayed higher expression levels in the maturing zone in S. officinarum but higher expression levels in the basal zone in S. spontaneum.

Expression pattern of HAKs during the circadian rhythm
Acting as an enzyme activator, potassium ions participate in a series of photosynthetic processes [31]. To analyze the expression pattern of HAKs during diurnal cycles, we investigated the transcriptome profiles of the mature leaves in the two Saccharum species at 2 h intervals over a 24 h period and at 4 h intervals over an additional 24 h. Consistent with the transcriptome profiles at different developmental stages and in the leaf segment gradient, 18 genes (HAK3/4/5a/5b/12/13/14/15/16a/16b/17/19a/19b/20a/20b /21/22/26) displayed very low or undetectable expression levels in the two Saccharum species, further supporting their limited roles in growth and development (Fig. 6). In addition, HAK8 and HAK24 also showed low expression levels over the two 24 h periods. HAK1, HAK2, HAK7, HAK18 and HAK27 showed higher expression levels in S. officinarum than in S. spontaneum, while HAK9 and HAK10 displayed higher expression levels in S. spontaneum than in S. officinarum. HAK1 and HAK2 had no diurnal expression pattern in the two Saccharum species. HAK7 displayed a higher expression level at night than in the daytime and showed the lowest expression level at noon in S. officinarum but showed no diurnal expression pattern in S. spontaneum. HAK10 displayed a higher expression level at night than in the daytime in S. spontaneum but showed no diurnal expression pattern in S. officinarum. HAK9 displayed a higher expression level at night than in the daytime in both Saccharum species. HAK18 and HAK27 displayed higher expression in the morning in the two Saccharum species. These findings suggested the functional divergence of the HAK genes in diurnal rhythms.

Expression pattern of HAKs under K + -deficiency stress
To investigate the functional divergence of HAK genes in response to low-potassium stress in sugarcane, we studied the expression profiles of HAKs in roots from the Saccharum hybrid variety YT55 at 0 h, 6 h, 12 h, 24 h, 48 h and 72 h under low-K + stress (Fig. 7)   SsHAK1 and SsHAK21 were selected for complementary validation in yeast because they were both induced in response to low-K + stress. The transformed yeast strain carrying only the empty vector pYES2.0 was used as a control. There were no obvious growth differences between yeast transformed with pYES2.0 and yeast transformed with pYES2.0-SsHAK1 or pYES2.0-SsHAK21 in SC/−ura medium containing 100 mM KCl (Fig. 8). However, when the KCl concentration decreased to 10 mM, the growth of yeast transformed with SsHAK1 and SsHAK21 was better than that of yeast transformed with the empty vector. When the KCl concentration decreased to 1 mM, the growth of yeast transformed with the empty vector was significantly inhibited, while the growth of yeast transformed with SsHAK1 or SsHAK21 was almost unaffected (Fig. 8). These results suggested that both SsHAK1 and SsHAK21 could recover the K + absorption function in the yeast mutant strain R5421, indicating that they had potassium transporter activity.

Discussion
The HAK/KUP/KT family of potassium transporters have been widely reported to be associated with K + transport across membranes in plants. They also play vital roles in response to salt and drought stress, as well as morphogenesis of root and shoot [7]. However, genome-wide analysis of the HAK/KUP/KT gene family has not been conducted in Saccharum due to its complex genetic background. The recently released S. spontaneum genome allowed us to identify 30 HAK genes from S. spontaneum. In addition, 248 HAK genes from 13 other representative plant species and an outgroup were used to construct a phylogenetic tree and study the evolution of HAK genes in Saccharum. Furthermore, expression analysis based on RNA-seq revealed spatiotemporal expression and functional divergence in the HAK family, which provides valuable information and robust candidate genes for future functional analysis.
Evolution of the HAK gene family in Saccharum and representative angiosperms WGD or polyploidization, gene loss and diploidization are considered important evolutionary forces in plants [32,33]. Angiosperms, pancore eudicots and monocots originated from ε, γ and σ WGD events, which have been revealed by a rigorous phylogenomic approach [33]. A recent study showed that pineapple had one fewer ancient ρ WGD events than other gramineous plants [34]. A. trichopoda is the earliest known angiosperm to have evolved separately from other angiosperms and has attracted much attention from botanists. In this study, 279 HAKs from 15 plant and after the ε WGD event (approximately 220 Mya) [33]. The number of HAKs in the four clades varied greatly (from 29 to 98, Fig. 3), which is consistent with a previous study in which HAKs were unevenly distributed in different clades among angiosperms [35]. In clade I, only one HAK gene member was from A. comosus and Arabidopsis, while in Poaceae species, the HAK number ranged from 6 to 13. This result indicated that WGD or recent gene duplication contributed greatly to the expansion of HAKs. SsHAK5a/5b, SsHAK16a/16b, and SsHAK19a/19b were from tandem duplications, while SsHAK20a/20b may have originated from a transposed duplication. The LCAs of SsHAK5 and SsHAK18 (in clade III) may have occurred before the split of monocotyledonous and dicotyledonous plants. HAK5 was speculated to be lost in other dicotyledons except for Arabidopsis, which may be due to the gene functional redundancy of the HAK family. HAK18 was retained in all monocotyledonous and dicotyledonous plants, showing its functional constraint for the HAK family, and the expression profile analysis of HAK18 also confirmed this.
In clade II and clade III, SsHAK2 and SsHAK7 were retained from the ε WGD event, and in dicotyledons, these two orthologous genes were lost. SsHAK3 and SsHAK13 originated after A. trichopoda had evolved separately from other angiosperms. SsHAK8, SsHAK9 and SsHAK10 were assumed to be retained from the ε WGD event; SsHAK11, SsHAK12, SsHAK15, SsHAK24 and SsHAK25 were retained from the σ WGD event, as only monocotyledons contained these genes. SsHAK14 and SsHAK23 were assumed to be retained from the ε WGD event, but HAK14 was probably lost in dicotyledons. Clade IV contained the lowest number of HAKs. SsHAK4 and SsHAK17 originated before the split of monocotyledons and dicotyledons and after the split of A. trichopoda from angiosperms. The LCA of SsHAK26 originated after the split of the Gramineae and pineapple.
The HAK gene family in plants exhibited a less conserved exon/intron structure. The exon number in Saccharum ranged from 2 to 12 (Fig. 1, Additional file 7), and the variation range in Saccharum was larger than that in rice [17], maize [19] and wheat [36]. Three types of mechanisms, exon/intron gain/loss, exonization/pseudoexonization and insertion/deletion, mainly led to exon-intron structure differences in paralogous or orthologous genes [37]. Although the gene structure of SsHAKs changed greatly, the protein size was relatively conserved, suggesting that exon-intron structure differences in SsHAKs were mainly caused by intron gain/loss. Clade I and clade IV belong to the older HAK family in Saccharum, so the HAKs in these two clades were speculated to have more intron gain/loss events based on the 'introns-early' theory during the lengthy evolutionary process [38,39]. The results in this study also support this view because the variation in exon number in clade I and clade IV was much greater than that in clade II and clade III.

Gene expression and functional divergence of HAKs in Saccharum
The transcriptional regulation of K + transporters is a common mechanism by which plant species respond to low-K + stress [8], and expression pattern analysis can provide insight into the potential functions of the HAK gene family. In this study, we found that most HAK genes in clade I and clade IV showed low or undetectable expression levels across all examined samples. Most HAK genes in clade II and clade III were strongly expressed in all tested tissues. These results were consistent with the results of previous studies on HAK genes in rice [17], Arabidopsis [25] and wheat [36]. Five OsHAK genes (OsHAK2/10/15/23/25) from clades II and III were expressed in all examined tissues of three different genotypes [17]. In Arabidopsis, 12 out of 13 HAK/KUP/KT genes were from clades II and III, most of which were expressed in the roots, leaves, siliques and flowers [25]. Similarly, most TaHAKs in wheat belonging to clades II and III were constitutively expressed in all tissues [36].
Low-K + stress tends to induce the upregulated expression of K + transporter genes [40]. Previous studies demonstrated that the expression of OsHAK1 in rice [20], TaHAK1 in wheat [36] and PbrHAK1 in pear [41] was induced by K + starvation. In this study, the expression level of SsHAK1 increased rapidly under low-K + stress, and this result is in good agreement with previous studies. Notably, SsHAK21 was substantially upregulated after a short period of K + -starvation treatment and then rapidly downregulated (transient activation), indicating that SsHAK21 was involved in the low-K + stress response in sugarcane. Similar results were found in rice, as OsHAK21 functions in the maintenance of ion homeostasis and tolerance to salt stress [42]. SsHAK1, SsHAK17 and SsHAK21 displayed upregulated expression, suggesting that they may play important roles in maintaining normal growth and mediating potassium acquisition under K + deficiency. In addition, nearly half of the SsHAK genes were not expressed or had very low levels of expression in all tested tissues at all stages or even under low-K + stress, which may be caused by the gene functional redundancy due to WGD events in sugarcane.
The root system acquires K + from the soil solution, and then K + is transported among compartments within cells and from the roots to the shoots. A schematic model was proposed based on the expression profiles of the 30 SsHAK genes to illustrate the spatial and temporal gene expression in plant tissues and root hair cells of sugarcane (Fig. 9). HAK7/9/18 were mainly expressed in the tissues of maturing and mature stems and leaves, indicating their important roles in K + transport in these tissues. HAK7/9/18/25 also showed a circadian rhythm expression pattern, suggesting that these genes were regulated by sunlight. Low-K + stress induced the upregulation of the transcriptional expression of HAK genes. In Arabidopsis, transcription factors, such as DDF2, JLO, ARF2, RAP2.11, TFII_A, and bHLH121, directly bind the promoter of AtHAK5 to induce its expression and increase tolerance to low-K + and salt stress [26]. In this study, the expression of HAK1 and HAK21 was greatly upregulated, which may also be positively regulated by transcription factors (TFs), such as DDF2 and JLO, and further experiments, such as yeast one-hybrid assays, can be used to screen the TFs. AtHAK5 and its homologs from pepper and tomato can be activated by the CIPK23 (CBL-interacting protein kinase 23)/CBL1 (calcineurin B-like protein) complex [27]. In rice, OsHAK1/ 19/20 can be phosphorylated by a receptor-like protein kinase, RUPO (ruptured pollen tube) [43]. In this study, the CBL-CIPK complex and the receptor-like kinase RUPO may also act as regulators of high-affinity potassium transporters, such as HAK1, via phosphorylationdependent interactions.

Conclusions
In this study, 30 HAK (high-affinity K + transporter) genes were identified through comparative genomics analyses of sugarcane. Evolutionary analysis revealed that both ancient whole-genome duplication (WGD) and recent gene duplication contributed to the expansion of the gene family, and purifying selection was the main force driving evolution. HAK/KUP/KT genes were accompanied by intron gain/loss in the process of evolution. Expression analysis based on RNA-seq under low-K + stress and at different developmental stages revealed spatiotemporal expression and functional divergence in the HAK/KUP/KT gene family. Yeast functional complementation analysis showed that SsHAK1 and SsHAK21 mediated K + transport under low-K + stress. Altogether, these results provide valuable information and robust candidate genes for future functional analyses for the genetic improvement of potassiumutilization efficiency in sugarcane.

Plant materials
Two Saccharum species, LA-Purple (S. officinarum, 2n = 8x = 80, originated in the USA and was introduced into China; the plants were provided by Zhang's laboratory at Fujian Agriculture and Forestry University) and SES-208 (S. spontaneum, 2n = 8x = 64, originated in the USA and was introduced into China; the plants were provided by Zhang's laboratory in Fujian Agriculture and Forestry University), were cultivated at Fujian Agricultural and Forestry University (Fuzhou, 119°16′48″E, 26°4′48″N, Fujian, China) and sampled for gene expression pattern analysis.
For expression pattern analysis at different developmental stages, tissue samples including leaf roll, leaf (fully expanded leaf), top immature internode, premature internode and mature internode were collected from premature plants (9-month-old plants) and mature plants (12-month-old plants). The sugarcane internodes were numbered from top to bottom. Leaf and stem Fig. 9 Schematic models for the roles of HAKs based on gene expression profiles in sugarcane. In the maturing and mature zones of the leaves and stems, HAK7/9/18 were the main expressed genes. Moreover, these genes also presented a diurnal expression pattern. HAK25 was mainly expressed in the maturing and mature zone of leaf tissues, while HAK2 was mainly expressed in the stem. Low-K + stress induced the upregulation of the expression of HAK1 and HAK21. Transcription factors such as DDF2 and JLO may directly bind to the promoters of HAK1/21 to induce gene expression and subsequently promote HAK transporters, such as HAK1 and HAK 21, to acquire K + in roots. HAK1 may be phosphorylated and activated by the CBL1-CIPK23 complex or receptor-like kinase, RUPO (ruptured pollen tube). The K + concentration in the vacuole is highly varied to maintain cellular K + homeostasis. Some HAK transporters, such as HAK10, located in the tonoplast of vacuoles may play a role in regulating the K + concentration in vacuoles tissues in the seedling stage were collected from 35-dayold plants as previously described [44,45].
For expression pattern analysis of the leaf gradient, the two Saccharum species were grown in a greenhouse with light intensities of 350 μmol/m 2 /sec, 14:10 L/D, 30°C L/ 22°C D and 60% relative humidity. The second leaf of 15-day-old LA-Purple and 11-day-old SES208 after planting at 3 h into the light period and samples preparation method was as described by Li et al. [30].
For expression pattern analysis of the diurnal cycle, leaves of the mature plants of LA-Purple and SES208 were sampled consecutively 12 times with 2 h intervals from 6:00 a.m. on March 2, 2017, then sampled consecutively 7 times with 4 h intervals from 6:00 a.m. on March 3, 2017. The sunrise and sunset times on March 2, 2017 in Fuzhou were 6:25 a.m. and 6:05 p.m. respectively. Tissue collection was performed following a previously described method [34].
For expression pattern analysis under low-potassium stress, Saccharum hybrid variety YT55 (this variety was bred by Guangzhou Sugarcane Industry Research Institute and was planted in breeding bases for sugarcane in Wengyuan, Guangdong Province) was cultured at a normal potassium level (3.0 mmol /L) for 20 days in a greenhouse and then transferred to the K + -deficient nutrient solution (0.1 mmol /L) for starvation treatment. Mixed samples of roots from 6 plants in a pot (a biological replicate and three biological replicates in total were collected) were collected at 0 h, 6 h, 12 h, 24 h, 48 h and 72 h after starvation and stored in liquid nitrogen for total RNA isolation.
The evolutionary history of 14 representative angiosperms was inferred by the neighbor-jointing (NJ) method [47]. Based on the protein sequence alignment, the phylogenetic tree of the HAK/KUP/KT gene family was constructed using NJ methods. The construction of the NJ tree was performed using MEGA7 [48] with the "pair deletion" and "Poisson correction" models. The reliability of the internal branches of the tree was evaluated by a bootstrap test (1000 replicates), and the percentages are shown next to the branches.
Analysis of the expression profiling of HAKs in Saccharum based on RNA-seq RNA preparation, cDNA libraries construction and RNA-seq libraries sequencing were performed as previously described [51,52]. Raw data were aligned to available S. spontaneum AP85-441 reference gene models using Trinity (https:// github.com/trinityrnaseq/trinityrnaseq/wiki). Fragments per kilobase per million mapped fragments (FPKM) values were calculated to represent gene expression levels as previously described [51,52].

Validation of HAK gene expression levels by RT-qPCR
The expression level of three HAK genes (HAK1, HAK7 and HAK20b) in the roots of Saccharum hybrid variety YT55 at 6 time points (0 h, 6 h, 12 h, 24 h, 48 h and 72 h) under K + -starvation conditions was validated by RT-qPCR, to normalize the expression levels, 2 constitutively expressed genes, the eukaryotic elongation factor 1a (eEF-1a) and actin were used as reference genes, each sample had 3 biological replicates and 3 technical replicates. (Additional file 10). The reaction program of reverse transcription, real-time PCR and the relative expression levels calculation were carried out as Wang et al. described [52].