1 Introduction

Potassium (K+) is a crucial monovalent cation in plant nutrition, constituting a macronutrient that can account for up to 10% of a plant's dry weight (as reviewed in [1]). This cation plays a vital role in numerous aspects of growth and survival. For example, it helps compensate for negative charges generated in processes like glycolysis; maintains turgor pressure and cell volume; enhances the activity of various enzymes; aids in protein synthesis; and regulates membrane potential and intracellular pH. Beyond these fundamental cellular functions, K+ also plays a pivotal role in whole plant physiology, influencing processes such as stomatal movement, which controls water loss through transpiration and desiccation [2, 3]. Stomatal regulation is critical since changes in stomatal characteristics can impact water use efficiency (WUE), which is a key determinant of crop yield and efficiency [4]. Stomatal movements are significantly affected by stresses like water deficiency or salinity conditions. Additionally, over the long term, plants can adapt the number and size of stomata in response to environmental conditions to optimize their WUE (as reviewed in [5]).

In most plant cells, K+ accumulates against its concentration gradient in relatively high amounts, while other cations, such as sodium (Na+), must be excluded to prevent toxicity. The uptake of K+ by plants is influenced by its availability in the soil and the presence of Na+, which can be transported by certain channels and transporters. Salt stress comprises three fundamental components: osmotic stress when concentrations are very high, Na+ toxicity due to K+ displacement, and resulting K+ deficiency that disrupts the Na+/K+ ratio in root and aerial parts of plants [6]. Precise K+ nutrition management is essential for plants to cope with salinity or drought stress. Understanding plant responses to various components of this stress is crucial. We propose that a molecular strategy identifying genes involved in the response to K+ deficiency in plants could yield knowledge beneficial for developing more salt- and drought-tolerant plant varieties.

Studies of plant genome sequences have identified three families of genes encoding plasma membrane K+ transporters: the HKT family [7], the HAK/KUP/KAT transporter family [8] and the Voltage-gated (VG) family, formerly known as Shaker channels, based on structural and phylogenetic studies [9, 10]. The first two families are found in plants, fungi, and bacteria, while the VG/Shaker family exists in these organisms, as well as in animals. The VG/K+-selective channels dominate membrane conductance in most cell types, including guard and root cells. Structural data obtained from bacterial, animal and plant channels, including KAT1, suggest a tetrameric, pore-forming structure, composed of homo- or hetero-tetramers [11,12,13].

A crucial aspect of K+ nutrition involves its uptake from the soil by the epidermal and cortical root cells. However, while soil K+ concentrations can vary widely from 0.01 to 20 mM, plant cells maintain a relatively constant concentration of 80–100 mM in the cytoplasm. Thus, coordinating various transport systems is crucial to ensure the uptake of this macronutrient for proper plant growth and development [14]. K+ uptake by roots exhibits biphasic kinetics due to the existence of two transport systems. One operates with high affinity at low external K+ concentrations (< 1 mM), while the other functions with low affinity at high external K+ concentrations (> 1 mM). At high soil K+ concentrations, channels primarily facilitate K+ movement across the membrane, allowing ions to move along the electrochemical gradient. At low K+ concentrations, active transport systems are required to bring K+ into the cell against its gradient [15]. In some cases, channels may also function in the high-affinity uptake range if the membrane is adequately hyperpolarized [16].

The main players contributing to K+ nutrition include HAK5, AKT1, and non-selective cyclic nucleotide-activated cation channels (CNGCs), each operating in different substrate availability ranges, from high to low, respectively [1]. Among these channels, only AKT1 (and AtKC1) are highly expressed in root tissues [17]. In many plant species studied to date, HAK family members, particularly HAK5, show low expression levels in roots under control conditions but are strongly induced in response to K+ deprivation, serving to maintain K+ supply and K+/Na+ homeostasis [8]. Notably, an Arabidopsis mutant lacking HAK5 and AKT1 can still take up K+ and it exhibits residual growth with external K+ concentrations above 1 mM, suggesting additional trade-offs in the transport system [16, 18, 19].

Additionally, GORK (guard cell outward-rectifying K+), expressed in the outer cell layers of the root, appears to be the primary K+ efflux pathway, precisely regulating the electrical potential of the plasma membrane in these cells. Moreover, GORK facilitates repolarization when exposed to high salinity [18]. In the root stele, once K+ has been taken up by cortical cells, the SKOR channel loads the cation into xylem vessels for transport to aerial zones [19]. The low-affinity NO3− transporter NRT1.5, serves a similar function, transporting K+ to the xylem by coupling with other transporters or directly acting as a K+/H+ antiporter [1]. Finally, a noteworthy case is the KUP9 transporter, specifically expressed in the quiescent center. This symporter appears to be key in maintaining meristematic activity under low K+ conditions by regulating both K+ and auxin homeostasis [20].

Although K+ is an essential macronutrient for proper plant development, limited studies have characterized the transcriptomic response between phylogenetically related species with different responses to K+ deficiency. Initial studies conducted in the model plant Arabidopsis thaliana more than 15 years ago using microarrays revealed a significant number of deregulated genes related to the synthesis of jasmonic acid (JA) were identified in seedlings after 2 weeks of K+ starvation [21]. JA is one of the primary hormones involved in leaf senescence, likely a consequence of inadequate K+ supply to aerial organs. Additional categories of genes with altered expression included those related to the synthesis of glucosinolates, polyamines, and other secondary metabolites. Notably, important signaling components, such as stress-related transcription factors like WRKY, AP2, or bHLH were induced. Moreover, a substantial number of genes involved in cell wall formation, such as peroxidases, extensins, and arabinogalactans, were repressed [21].

Here, upon observing that broccoli (Brassica oleracea, L. var. Italica Plenck) demonstrates notable tolerance to limiting K+ conditions, compared to Arabidopsis, we were interested in investigating the molecular mechanisms involved. For this reason, we carried out comparative transcriptomic studies upon K+ starvation over several days. Arabidopsis and broccoli belong to the Brassicaceae family, and the evolutionary split between these two species occurred approximately 43.2 million years ago, making them relatively closely related. Comparative genomic studies have demonstrated a high level of synteny between the two species, with most orthologous genes distributed similarly [22]. However, the causes that explain the marked differences in the tolerance to limiting K+ conditions are still unknown. A very recent report indicated that, contrary to what was commonly assumed, K+ may be a limiting factor in many terrestrial ecosystems [23]. By conducting these comparative genomic studies, we aim to gain a comprehensive understanding of the molecular effects of K+ deficiency and to identify possible responses present in broccoli that may explain its relative tolerance. This knowledge can provide valuable tools for the development of crop plant varieties that are more productive with less added fertilizers, and to better understand the transcriptomic effects of the deficiency of a major inorganic nutrient.

2 Materials and methods

2.1 Plant material and growth conditions

Arabidopsis Col-0 and broccoli plants (Brassica oleracea, L. var Italica Plenck) cultivar Naxos, kindly provided by Sakata Seed Ibérica SLU, were used for these studies. The hydroponic media for control conditions contained 2 mM K2SO4, 1.5 mM CaCl2, 10.3 mM NH4NO3, 0.75 mM MgSO4, 0.363 mM Ca(H2PO4)2, 0.025 mg/L CuSO4·5H2O, 36.7 mg/L FeNa-EDTA, 10 mg/L H3BO3, 10 mg/L ZnSO4·7H2O, 18.94 mg/L MnSO4·H2O, 0.25 mg/L Na2MoO4·2H20 (4 mM K+). For long term experiments with limiting K+, the media was made with 0.05 mM K2SO4 (0.1 mM K+) and for short term starvation experiments, no K2SO4 was added. The plants were grown at 24 °C under long (16 h light/8 h dark photoperiod) or short (8 h light/16 h dark) day conditions, as indicated, in a growth chamber and the medium was aerated.

2.2 Growth comparison between broccoli and Arabidopsis and broccoli root length, stomatal size and number measurements

To compare the growth ratio in limiting potassium conditions, broccoli plants were grown on MS plates for 7 days, transferred to hydroponic medium and grown for 1 more week under short day conditions (8 h light/16 h darkness). After these 2 weeks, half of the plants were transferred to the hydroponic media with 0.05 mM K2SO4, (limiting K+) and the other half was maintained in 4 mM K+, under short day conditions. The same conditions and growth media were used for Arabidopsis, with the exception that Arabidopsis was grown for 12 days on MS plates. After 14 days of growth in limiting potassium, the aerial part of both species was photographed and collected. This material was dried in an oven and the dry weight was measured. For the measurement of stomatal number and size, broccoli plants were grown as described, but growth in limiting potassium was continued until day 30. At that time, root length was measured, and leaf peeling was performed on the plants grown under both conditions (4 mM or 0.1 mM K+) during the middle of the day (4 h after the lights had turned on). For Arabidopsis, the same conditions were used, except that the measurements were conducted 10 days after transfer to limiting K+. Images were immediately captured using a light microscope and processed with the Fuji software using the ImageJ program to determine the size and number of stomata and the number of epidermic cells. The stomatal index is calculated as the ratio of the number of stomata to the total number of stomata and epidermal cells.

2.3 Photosynthesis measurements

Gas exchange and chlorophyll fluorescence measurements were taken for the determination of photosynthetic related parameters with a LI-6400 (LI-COR, Lincoln, NE, USA). Net photosynthesis (CO2 fixation) rate (AN), stomatal conductance (gs), transpiration rate (E) and substomatal CO2 concentration (Ci) were measured at steady state under conditions of saturating light (1000 μmol m−2 s−1). Quantum efficiency of photosystem II (ΦPSII) was measured in the same leaves. Ten different determinations were performed in mature leaves from different broccoli plants grown in limiting potassium for 30 days, as described in the previous section.

2.4 Ion content

Arabidopsis (Col-0) and broccoli seeds were sown on MS plates for 12 and 7 days, respectively. Seedlings were transferred to hydroponic culture medium containing 4 mM K+ for 7 days and grown under long day conditions (16 h light/8 h darkness). Half of the plants were then transferred to K+-free medium, and samples were collected at days 1, 3 and 6. Extracts were made by resuspending 10 mg of dry weight in 1 mL of 0.1 M HNO3. The K+ content was determined for Arabidopsis and broccoli using plasma emission spectrophotometry (Agilent Technologies 700 series ICP-OES; Santa Clara (Ca), USA).

2.5 RNA-seq

Arabidopsis and broccoli were grown and harvested as described for the ion content measurements. RNA-seq analyses was performed as described for four biological replicates of leaves and roots for both species were collected on days 1, 3 and 6 after transfer to limiting K+ conditions as described [24]. Clean read pairs longer than 20 nt were then mapped to the TAIR10 version of the Arabidopsis thaliana genome sequence (https://www.arabidopsis.org/) and EnsemblPlants release-54 Brassica oleracea genome using STAR [25]. The number of read counts mapped to one and only one of the annotated genes of the genome (uniquely mapped) was obtained with STAR. A differential expression analysis was done with DESeq2 (Love et al., 2014). The data is deposited at GEO omnibus (NCBI) with accession number GSE255728.

2.6 Data analysis

For heatmap representations, fold changes (FC) >|2| and padj < 0.05 between conditions and the control were represented using Rstudio (The Comprehensive R Archive Network (rstudio.com)). For cluster construction, the cuttree function was used. Heatmap representations were performed using the pheatmap function. The changes in gene expression, intensities and cluster groups are represented by different colors. To perform the principal component (PCA) and volcano plot analyses, RNA-seq data were normalized and processed using MetaboAnalyst 5.0 software (Wishart Research Group, University of Alberta, Alberta, Canada) [26]. Data normalization was performed by summation and scaling using the Pareto method. PCAs were performed using a multivariate analysis method. Volcano plots were constructed based on a FC ≥ 2 and a Student’s t-test P ≤ 0.05. Gene Ontology analysis of the DEGs was performed using the Panther classification system (Panther18.0) (https://pantherdb.org/) [27] and the functional categories are presented using REVIGO (http://revigo.irb.hr/) [28].

2.7 Phylogenetic studies

The Arabidopsis sequence was retrieved from TAIR and a protein–protein BLAST was done (NCBI) to obtain the orthologues of broccoli. These sequences were retrieved from NCBI and were used to do a BLAST analysis (Ensembl Plants: plants.ensembl.org) to identify the specific code in the broccoli genomes. Alignments and phylogenetic trees were constructed using the Molecular Evolutionary Genetics Analysis (MEGA) software (maximum likelihood) [29].

2.8 Metabolomic analyses

Broccoli plants were grown as described for the ion content measurements and the leaf tissue from day 6 was analyzed. Extraction of plant material, the instrumental conditions and processing of data for metabolomics analyses were essentially the same as those described previously [30, 31]. When no mass spectra library was available, but a potential identity was found, experimental mass spectrum was manually justified using Chemdraw software (CambridgeSoft, UK) and the in silico mass fragmentation function (level 3 annotation). A summary of the annotations is provided in Supplementary Table S2.

3 Results

3.1 Response of Arabidopsis and broccoli to growth in limiting K+

We were interested in comparing and contrasting the response of the important crop plant, broccoli, to conditions of limiting K+ with the closely related model plant, Arabidopsis. Arabidopsis and broccoli were grown for 14 days in limiting potassium (0.1 mM, Materials and Methods) and a clear difference in behavior was observed (Fig. 1A–E). While Arabidopsis displayed a growth reduction (measured as dry weight) of around 50%, broccoli showed less than a 20% reduction, indicating relative tolerance of this crop plant to limiting K+ (Fig. 1E). In addition, at this time interval, except for this minor reduction in growth, no symptoms of the lack of this nutrient were observed in broccoli plants, unlike Arabidopsis, which presented a clear phenotype (Fig. 1B). In the case of broccoli, it was necessary to wait until 30 days under limiting K+ conditions to observe a significant reduction in leaf size and the appearance of chlorotic spots (Fig. 1F). The Arabidopsis plants did not survive this amount of time in limiting K+.

Fig. 1
figure 1

Response of Arabidopsis and broccoli to growth in limiting K+. Arabidopsis and broccoli seedlings were grown on MS plates for 12 and 7 days, respectively. Seedlings were transferred to hydroponic medium with 4 mM K+ for 7 days. Plants were then transferred to K+-free hydroponic medium. At day 14, images of the leaves were taken of Arabidopsis (A and B) and broccoli (C and D) grown in the presence of 4 mM K+ (A and C) or 0.1 mM K+ (B and D). E The aerial parts of the plants were collected, dried and weighed. The box-whisker plots contain all data points (n = 6), and the boxes extend from the first quartile to the third quartile. The whiskers indicate the minimum and maximum values. The central lines show the average and x represents the median values. Data points outside the boxes are outliers. Statistical pairwise analyses were performed using the Student’s t test. ** indicates P ≤ 0.01 and *** indicates P ≤ 0.001 when comparing between treatments and ### indicates P ≤ 0.001 when comparing between species. F Broccoli plants were grown on MS plates for 7 days and transferred to hydroponic medium with 4 mM K+ for 7 days. The plants were then transferred to hydroponic medium containing either 0.1 mM (limiting K+) or 4 mM K+. At day 30 after transfer, plants grown in the indicated conditions of K.+ supplementation were photographed. (Inset = close-up of the chlorotic spots of broccoli leaves)

Full size image

To further characterize the phenotypes of broccoli after this prolonged stress (30 days), we analyzed root length, stomatal density, size, and index, compared to controls grown in parallel in media with 4 mM K+ (Fig. 2). We observed that the root length increased significantly in these broccoli plants. We also observed increases in stomatal density and size, and a slight decrease in the stomatal index. At the 30-day time point, the roots from each individual plant were difficult to separate. Therefore, to record the differences in root architecture, we repeated the experiment and photographed the plants at day 15 after transfer (Fig. 2E). In Arabidopsis, we did not observe significant phenotypes for root length or stomata 10 days after transfer (Figure S1).

Fig. 2
figure 2

Analysis of broccoli root length, stomatal density, size, and index in long-term K+ limitation K+. Broccoli plants were grown on MS plates for 7 days and transferred to hydroponic medium with 4 mM K+ for 7 days. The plants were then transferred to hydroponic medium containing either 0.1 mM (limiting K+) or 4 mM K+. At day 30 after transfer, root length (n = 14) (A) stomatal density (B) and area (C) were measured and used to calculate the stomatal index, as described in Materials and Methods (D). 90 stomata from three different leaves per condition were analyzed. Data are mean values and error bars represent the standard deviation. ** indicates P ≤ 0.01 and *** indicates P ≤ 0.001. S: Stomata; E: Epidermic cells. E Representative photographs of the roots of broccoli plants grown for 15 days in limiting K+ (0.1 mM) or control conditions (4 mM K+). F, G Representative photographs of the samples used for the analyses presented in (B–D)

Full size image

We next measured several physiological parameters in broccoli plants after 30 days in limiting K+, observing a reduction in net photosynthesis, quantum efficiency of PSII, stomatal conductance, and transpiration rate (Table S1). These data suggest reduced stomatal function and the corresponding response of the plant to increase stomatal density to counteract this defect in photosynthesis. Therefore, broccoli presents a physiological response to limiting K+, but the symptoms are less severe and take twice as long to appear, as compared to Arabidopsis, indicating a relative tolerance of broccoli plants to this abiotic stress.

3.2 Conditions for short-term limiting K+ treatment to study transcriptional responses in Arabidopsis and broccoli

Having confirmed the relative tolerance of broccoli to limiting K+ conditions as compared to Arabidopsis, we were interested in studying the transcriptional response during the early phases of K+ deficit. We decided to take a comparative transcriptomic approach to discover the genetic regulation that could explain our observations. As a first step, we performed preliminary experiments to set up the proper conditions for the transcriptional analysis, focusing on short time intervals. We reasoned that the early transcriptional response of broccoli may mediate its better performance in limiting K+. We analyzed the ion content during a time-course of K+ limitation in both species. Figure 3 shows the K+ content in both leaves and roots at 1, 3 and 6 days after the transfer of the plants from 4 mM K+ to media containing no added K+. As observed, the K+ content of Arabidopsis roots decreases at a considerably faster rate than broccoli (Fig. 3A). However, after 6 days, an approximately 50% reduction in the K+ content for both tissues and both species was observed (Fig. 3A, B). These disparate kinetics suggest that there are differences in the transport systems governing these responses in the roots of each species.

Fig. 3
figure 3

Determination of the plant K+ content during short-term K+ limitation in Arabidopsis and broccoli. Arabidopsis and broccoli seedlings were grown on MS plates for 12 and 7 days, respectively. Seedlings were transferred to hydroponic medium with 4 mM K+ for 7 days. Plants were then transferred to K+-free hydroponic medium, and samples were collected on days 0, 1, 3 and 6. K+ content was determined for roots (A) and leaves (B) using plasma emission spectrophotometry. Data are mean values of 4 biological replicates and error bars represent the standard deviation. AR Arabidosis roots, BR broccoli roots, AL Arabidopsis leaves, BL broccoli leaves, dw dry weight

Full size image

3.3 Transcriptomic analysis of initial response to limiting K+ in Arabidopsis and broccoli

We used samples from this same experiment to perform RNA-seq analyses for both leaves and roots of Arabidopsis and broccoli. Four biological replicates for each sample were analyzed in parallel with controls grown at 4 mM K+ and harvested on the same days. This strategy was employed to obtain more robust data for statistical analyses and to reduce the possible differences in the developmental stages of the plants. We observed a robust transcriptional response (DESeq2, P < 0.05) for all conditions tested, which generally increased with time, except for day 3 in Arabidopsis leaves and days 1 and 3 in broccoli leaves, where we observed a much more modest response (Table 1).

Table 1 Number of differentially expressed genes in Arabidopsis and broccoli under limiting K+
Full size table

Although we do not have an explanation for the transient decrease in the response of Arabidopsis leaves at day 3, the delay in the broccoli leaf response in is likely due to the slower reduction of the internal K+ content (Fig. 3). This delayed response suggests that significant transcriptional changes in broccoli leaves do not occur until day 6. PCAs of the DEGs with FC ≥ 2 in expression showed tight grouping for the 4 biological replicates, reflecting the different expression patterns for root vs. leaves (Fig. 4). We also observed a marked separation of the day 6 samples in Arabidopsis leaves, indicating that this species' most significant transcriptional response occurs under these conditions.

Fig. 4
figure 4

Principal component analysis of RNA-seq data from Arabidopsis and broccoli grown in limiting K+. The MetaboAnalyst 5.0 software was used to generate PCA plots of the DEGs (P > 0.05 and FC ≥ 2) for all of the conditions and tissues assayed in Arabidopsis (A) and broccoli (B). In the legend, the first position indicates the species (Arabidopsis or broccoli), the second position indicates the tissue (Leaf or Root), the third position is the number of days in limiting K+ (1, 3 or 6) and the last position is the [K+] (0 or 4 mM)

Full size image

To visualize the magnitude and statistical significance of the changes in the expression patterns at day 6, we generated volcano plots (Fig. 5). These plots revealed notable differences, such as the relatively low level of gene repression in Arabidopsis leaves (Fig. 5A), where most genes showed only modest repression. Other interesting observations that can be made based on these analyses are the induction of the Bo4g027570 gene encoding a PYL4-like abscisic acid receptor in broccoli leaves or the two homologues of the PHT1;3 phosphate transporter gene in broccoli roots. A list of the top differentially regulated genes is provided in Figure S2.

Fig. 5
figure 5

Volcano plot representation of RNA-seq data from Arabidopsis and broccoli grown in limiting K+. The MetaboAnalyst 5.0 software was used to generate volcano plots of the DEGs (P > 0.05 and FC ≥ 2) of K+ Arabidopsis leaves (A) and roots (B), and broccoli leaves (C) and roots (D) grown for 6 days with 0 mM K+, as compared to controls grown in 4 mM K+. Genes mentioned in the text are indicated in the corresponding plots. Red and blue dots represent up and down-regulated genes, respectively. Gray dots represent non-significantly differentially expressed genes. The X-axis represents the log2 of FC in expression while the Y-axis shows log10 of the Q-value

Full size image

Heat maps of DEG expression patterns over the experimental time course (Figure S3) were used to generate clusters for further analyses of the data (Supplementary files S1–4). These clusters were then analyzed for the enrichment of the GO terms identified for the DEGs in leaves (Figure S4) and roots (Figure S5) in both Arabidopsis and broccoli. Here, we made several interesting observations. For example, in the leaves of both species, a common response involving the induction of genes related to response to oxidative stress, hypoxia, bacterial and fungal defense responses, JA, abscisic and salicylic acid (SA) signaling, salt stress, water deprivation, and wounding was observed. Additionally, both Arabidopsis and broccoli roots displayed a common response to hypoxia. Unique responses for each species were also identified. For instance, the most highly induced class of genes in Arabidopsis roots were related to heat shock and this response was not observed in broccoli (Supplementary File S2, Cluster 11). Heat shock activation probably is related to tissue damage and indicative of an acute stress response, which was absent in broccoli.

Another interesting finding is the expression pattern of genes related to Fe2+ starvation and homeostasis in Arabidopsis roots and leaves. These genes were initially repressed, compared to control samples at day 1 and then became up-regulated over the course of the K+ starvation (Supplementary Files S1 and S2). The activation of the Fe2+ homeostasis may be related to the hypoxia response, as hemoglobin is the main protein that Arabidopsis uses to counteract hypoxia [32]. The interesting question is: Why do K+-depleted plants behave as hypoxic plants? One possible explanation may be related to considerations of basic biochemistry in which low K+ alters the pH of the cytoplasm and this alters the affinity for oxygen of the hemoglobin, leading to a hypoxia response. An alternative explanation may involve the requirement of some antioxidant enzymes for Fe2+. When it becomes limiting and there is a concomitant decrease Fe2+-containing heme groups, the hypoxia response may be activated.

3.4 Identifying candidates that may mediate broccoli tolerance to limiting K+

We next examined in more detail the transcriptional response of several classes of genes looking for possible candidates to explain the relative tolerance of broccoli to limiting K+. We began by comparing the expression profiles of transporter genes related to K+ uptake in Arabidopsis and broccoli. It has been previously reported that, in Arabidopsis, the gene encoding the high affinity K+ transporter HAK5 is highly induced in roots upon K+ starvation [33]. As shown in Fig. 6, we observed a similar response on day 6. Interestingly, BoHAK5, the broccoli orthologue, showed much higher expression under control conditions and this elevated expression persisted in the K+-starved plants. Notably, the lowest levels of BoHAK5 expression under limiting K+ conditions were higher than the maximum expression observed in Arabidopsis on day 6. This level of expression, which reaches up to 13,000 transcripts per million (TPM), is very high compared to the other transporters, especially under normal growth conditions. Regarding the data corresponding to several other genes previously implicated in K+ transport, we observed a higher relative expression of AKT1 in Arabidopsis, compared with the orthologue in broccoli. NRT1.5, which was previously shown to transport K+ to the xylem by acting as a K+/H+ antiporter or by coupling with other transporters [1], was highly expressed in both species, compared to the other transporters. Regarding K+-responsive genes, in broccoli the expression of the BoHAK5 gene responded more robustly to changes in the K+ concentration., In contrast, several Arabidopsis genes, including AtHAK5, AtSKOR and AtNRT1.5, were differentially expressed in response to the K+ treaments.

Fig. 6
figure 6

Comparison of the expression profiles of genes involved in K+ uptake and transport in Arabidopsis and broccoli grown in limiting K+. The TPMs in the RNA-seq data for the indicated genes are represented for Arabidopsis (A) and broccoli (B). Data are mean values for 4 biological replicates and error bars represent the standard deviation. In the legend, the first position indicates the species (Arabidopsis or broccoli), the second position indicates the tissue (Leaf or Root), the third position is the number of days in limiting K+ (1, 3 or 6) and the last position is the [K+] (0 or 4 mM)

Full size image

As stated, one of the most striking differences between broccoli and Arabidopsis is the high expression level of HAK5 in broccoli, which may explain the slower reduction in the K+ content observed in broccoli. Since it is possible that other genes from the KUP family in broccoli may be highly induced in response to K+ limitation like AtHAK5, we performed a phylogenetic analysis of this gene family and examined their expression patterns. However, none of the identified broccoli isoforms showed a robust induction under these conditions (Figure S6). Therefore, it appears that BoHAK5 is a major player in K+ uptake in broccoli roots. It will be interesting to study the transport characteristics of BoHAK5 and its promotor, as it may represent a potential tool to improve the growth of plants under low K+ conditions.

Broccoli and Arabidopsis diverged about 43 Mya [34]. Between 22.5 Mya [35] or 7.9 Mya [34] the Brassicaceae tribe underwent a whole genome triplication that did not happen in the Arabidopsis and Thellungiella tribes. Interestingly, when comparing the Arabidopsis and broccoli genomes, most of the genes encoding K+ transporters are not triplicated. For example, KUP1, 2, and 6 are duplicated, while the other KUPs (with the exception of KUP9) and HAK5 are present as single copies. This limited diversity points to a very specific role for each K+ transporter and that redundancy has been negatively selected, probably because maintaining K+ transporters with overlapping functions in the genome could have adverse effects (i.e. lack of regulation, entry of toxic cations, K+ depletion). AtKUP5 seems to have arisen in the Arabidopsis genome recently, after the divergence, while BoKUP9a and BoKUP9b seem to have originated in a later duplication after the whole genome triplication.

A unique response observed in broccoli roots, revealed upon examination of the clusters of genes that were differentially expressed, was the induction of genes related to the cellular response and transport of NO3− (Supplementary File S4, Cluster 9). We examined in more detail the phylogeny and expression patterns of this family of genes in broccoli. As shown in Fig. 7, we observed the induction for 8 different NRT2 family members in broccoli, and 6 responded to limiting K+. Interestingly, we also observed high levels of expression for the Bo7g115930, which was also strongly and rapidly induced in limiting K+ (FCday 1 = 5.15; FCday 3 = 5.10; FCday 6 = 2.97). This gene encodes an orthologue of AtCEPH, a type 2C protein phosphatase gene, which in Arabidopsis has been described to dephosphorylate and activate the NRT2.1 nitrate transporter [36]. Thus, this appears to be a very robust response in broccoli and suggests that the lack of K+ leads to increased expression and activation of high affinity NO3− transporters. We analyzed the genes that are co-regulated with the NRT2 Arabidopsis orthologues of this subset of NO3− transport genes using the ATTED-II software (Version11.1) [37]. We identified homologs of the transcription factors of the WRKY family, WRKY29 and WRKY72 as being co-regulated with these NO3− transport genes, suggesting that this family of transcription factors may be involved in mediating this response in broccoli. Accordingly, we studied the phylogeny of these two gene families and their expression patterns in our broccoli RNA-seq experiments. Our analysis suggests that BoWRKY29a, BoWRKY72b and BoWRKY72c would appear to be the best candidates for future studies to confirm the role of these transcription factors in this regulatory circuit in broccoli (Figure S7).

Fig. 7
figure 7

Identification, phylogenic analysis, and comparison of the expression profiles of the NRT2-family genes in Arabidopsis and broccoli grown in limiting K+. A Arabidopsis and broccoli orthologues of the NRT2 family. B Phylogenetic analysis of the NRT2 genes from Arabidopsis and broccoli. C The TPMs in the RNA-seq data for the indicated genes are represented for the NRT2 family genes and the regulatory phosphatase (CEPH) from broccoli. Data are mean values for 4 biological replicates and error bars represent the standard deviation. In the legend, the first position indicates the species (broccoli), the second position indicates the tissue (Leaf or Root), the third position is the number of days in limiting K+ (1, 3 or 6) and the last position is the [K+] (0 or 4 mM)

Full size image

Another of the more striking transcriptional responses to K+ starvation in broccoli is the alteration of gene expression of glucosinolate biosynthesis pathway genes. This response was also observed in Arabidopsis, but to a lesser extent. More specifically, we observed a total of 54 broccoli genes that are predicted orthologues of 41 different Arabidopsis genes related to glucosinolate biosynthesis whose expression was altered during K+ starvation in broccoli (Figure S8). Based on these results, we performed a metabolic analysis of leaves at day 6, comparing plants grown in 4 mM K+ to those transferred to media without K+ supplementation. We were able to detect several glucosinolate-related metabolites. Many of these molecules accumulated in response to K+ starvation stress (Fig. 8), whereas others did not (Figure S9). Interestingly, citric acid levels markedly decreased (Figure S9), which may reflect a decrease in the rate of primary metabolism under conditions of abiotic stress, possibly favoring the accumulation of secondary metabolites. The accumulation of glucosinolates in response to various stress conditions has been reported previously and may contribute to a general defense mechanism in Brassicas [38].

Fig. 8
figure 8

Determination of glucosinolate-related metabolite content in broccoli leaves grown in limiting K+. Leaves from plants grown for 6 days at 0 or 4 mM K.+ were collected and processed for as described in Materials and Methods. Metabolites are indicated in each graph. The box-whisker plots contain all data points (n = 8) and the boxes extend from the first quartile to the third quartile. The whiskers indicate the minimum and maximum values. The central lines show the average and x represents the median values. Data points outside the boxes are outliers. Statistical pairwise analyses were performed using the Student’s t test. * indicates P ≤ 0.05; ** indicates P ≤ 0.01; and *** indicates P ≤ 0.001

Full size image

A key metabolite that has been shown to accumulate in Arabidopsis in response to K+-limiting conditions is SA, probably through the isochorismate pathway [39]. Indeed, according to our transcriptomic data, after 3 days of growth in limiting K+, AtICS2 expression had a slight but significant increase (1.7 times) in roots, and AtICS1 increased expression by fourfold on day 6, which is in good agreement with the previous metabolomic studies carried out in Arabidopsis by Cui and colleagues [39]. However, this response was faster and more robust in broccoli. In this species, up to 3 isoforms of ICS1 were induced in root tissue during the course of the experiment. Bo6g086860 and Bo2g082320 already appeared induced in roots on the first day of treatment, while these two, along with Bo6g118920, underwent quite significant inductions on day 6 (Fig. 9A). In addition, the orthologue of AtICS2 in broccoli (Bo8g068170) was induced on days 3 and 6. These data indicate a rapid and robust response in broccoli to synthesize SA under conditions of K+ limitation. We also observed induction of the only isoform of the EDS5 gene and fast and robust induction of 3 of the 4 isoforms of encoding the PBS3 enzyme. On the other hand, only a slight induction on day 1 was observed for the PAL orthologue and no differential expression was detected for genes encoding the AIM enzyme. These results suggest that SA biosynthesis through the isochorismate pathway is quickly and robustly induced in broccoli in response to K+ limitation. The full list of expression data for all isoforms, tissues and treatments is provided in Table S3. This induction of SA biosynthesis genes was also reflected in the induction of SA response genes (GO statistically overrepresented at day 6) in both leaves (30 genes induced) and roots (12 genes induced) in broccoli (Supplementary file S4). Importantly, in our metabolomic analysis, we observed an increase in SA accumulation of close to sevenfold in broccoli plants grown in limiting K+, as compared to control plants (Fig. 9B), corroborating the transcriptional response data.

Fig. 9
figure 9

Induction of SA biosynthesis genes and phytohormone accumulation in broccoli leaves grown in limiting K+. A Schematic representation of the SA biosynthetic pathways showing the FC log2 of the corresponding broccoli genes 1, 3 and 6 days after transfer to limiting K+. Adapted from [61]. The plastid is indicated in green. For the gene expression data, green denotes induction and red denotes repression, as compared to the samples grown in 4 mM K.+, and the intensity of the color indicates higher induction. NS = Not significant. B Determination of SA accumulation in broccoli leaves. The experimental procedure and data presentation are the same as described in Fig. 8. Statistical pairwise analyses were performed using the Student’s t test. ** indicates P ≤ 0.01

Full size image

4 Discussion

In this study, we aimed to identify candidate genes and pathways that may explain the relative tolerance of broccoli to limiting K+, compared to the phylogenetically related species Arabidopsis. We observed that Arabidopsis plants displayed a marked response to this stress after only 14 days, whereas broccoli plants were only mildly affected at this time point. We did not observe significant effects until 30 days of K+ limitation in broccoli, when Arabidopsis plants had already died. Studying the different aspects of the long- and short-term response to limiting K+ is important because several studies indicate widespread K+ limitation in terrestrial ecosystems leads to decreased biomass production in many different crop plants [40]. Therefore, understanding how plants respond on the molecular level to limiting K+ may provide valuable information to increase K+ uptake efficiency and reduce the negative effects associated with the lack of this important cation.

In order to gain insight into the possible mechanisms of broccoli tolerance to limiting K+, we studied the transcriptional response of both Arabidopsis and broccoli to the early stages of exposure to limiting K+ concentrations. Our experimental design included 4 biological replicates for each RNA-seq analysis and compared age-matched samples grown in normal (4 mM) or limiting (0 mM) K+. We collected samples of both Arabidopsis and broccoli roots and leaves at days 1, 3 and 6 after transfer to limiting K+. The idea of using both Arabidopsis and broccoli was to take advantage of the extensive information available for the most commonly studied model plant, which is more sensitive to this treatment and a relatively close relative that is an increasingly important crop plant and shows comparative tolerance to this abiotic stress. By comparing and contrasting the changes evoked in the gene expression patterns, we aimed to generate knowledge useful for the improvement of plant responses to this abiotic stress condition. We were able to identify a large number of DEGs in both species and in both roots and leaves, likely due to the statistical advantage of including 4 biological replicates. Our data thus provides a comprehensive, time resolved and tissue-specific analysis of the transcriptional response to K+ deficiency in two phylogenetically related species with different levels of relative tolerance.

In terms of common responses, both species showed robust responses to oxidative stress and hypoxia. This is in agreement with several previous studies examining the responses of different plant species [41, 42]. Several categories related to the response to different biotic and abiotic stresses were also enriched among the genes induced in leaves in both species, similar to what was observed in previous studies in Arabidopsis [43]. Accordingly, GO terms corresponding to several hormone response pathways were also enriched in both species. In particular, genes related to the response to JA and abscisic acid were identified as DEGs in both broccoli and Arabidopsis, again supporting previous studies based on transcriptomic data from Arabidopsis [21]. Interestingly, the gene corresponding to the Bo4g027570 locus encoding a PYL4 orthologue showed the most marked up-regulation in broccoli leaves after 6 days of treatment. This abscisic acid receptor represents an interesting candidate for future studies in broccoli to identify key players in different stress responses. Regarding other common responses observed in previous studies, we observed some evidence for alterations in genes related to Ca+ signaling, such as CBL9, which was highly induced in broccoli leaves (FCday 6 = 4.1) (Supplementary File S4). We did not observe significant enrichment in GO terms related to gibberellins, auxin, or ethylene biosynthesis, although a more detailed analysis of the response of key regulators of these pathways may reveal candidates for further studies.

Our data revealed several interesting observations regarding the response of broccoli to K+ limitation. First, we observed very high levels of BoHAK5 expression under all conditions tested, compared to Arabidopsis. In fact, BoHAK5 was more highly expressed than BoAKT1, in contrast to what is observed in Arabidopsis. This points to differences in the relative contributions of these K+ transporters in these two species and may suggest that, at least in this respect, broccoli may be more similar to rice, where OsHAK1 plays an important role in K+ uptake in roots [42, 44]. Therefore, BoHAK5 is an excellent candidate for future studies. Secondly, we observed a robust induction of many genes belonging to the NRT2 family of high affinity NO3− transporters in broccoli roots, including an orthologue of the PP2C protein phosphatase isoform that was recently shown to dephosphorylate and activate AtNRT2.1 [36]. K+ and NO3− have been proposed to be coordinately regulated [45] and the expression of Nitrate transporter 1/peptide transporter (NRT1/PTR) genes have been reported to be regulated by external K+ in Arabidopsis [21], rice [46], wheat [47], banana [48], pear [49] and cotton [50]. This is especially interesting in terms of the requirements of fertilizers, since most formulations contain both K+ and nitrogen sources. This relationship in broccoli warrants future investigation, as does the relationship between K+ and PO43−, as two orthologues of PHT1;3 (Bo2g119310 and Bo7g065360) were among the most significantly induced genes in broccoli roots (see Fig. 5). These data suggest that there is a coordinated response in broccoli upon K+ limitation to increase high affinity uptake of NO3−and PO43− in roots. Our data identify interesting candidates to study the interrelationship between the regulation and response to these key nutrients in broccoli.

Glucosinolate biosynthesis is clearly affected by K+ limitation in both broccoli and Arabidopsis. Armengaud and colleagues observed similar results in their transcriptomic analysis of the response of Arabidopsis to external K+ [51]. We confirmed these observations in Arabidopsis and observed an even more robust response in broccoli. The expression of a total of 54 genes related to glucosinolate biosynthesis was observed to be differentially regulated in broccoli in our RNA-seq data (Figure S8). This is especially interesting when considering that glucosinolates have been proposed as important contributors to the classification of broccoli as a health-promoting food [52]. Isothiocyanates, which are bioactive glucosinolate metabolites produced by gut microbes have been shown to have beneficial effects in the prevention of cancer [53,54,55]. In particular, glucoraphanin has been identified as being especially important as a precursor for sulforaphane, which is being intensively studied as a natural molecule effective in the treatment of cancer and other diseases [56]. In fact, a variety of broccoli, called Beneforté, which contains a threefold increase in the amount of glucophanin in florets, was developed and commercialized [57]. In our study, we observed an increase in the concentrations of several glucosinolate metabolites, including glucoraphanin (1.9-fold increase) in response to limiting K+ concentrations. This treatment may be explored as a method to increase the internal concentrations of these added-value metabolites.

One of the most striking results found here relates to the accumulation of SA in broccoli. We observed rapid and robust induction of genes of the isochorismate pathway and marked accumulation of SA in broccoli leaves. We also observed induction, although to a lesser extent of genes involved in this branch of the SA biosynthesis pathway in Arabidopsis leaves. Our data are in good agreement with the metabolomic study carried out by Cui and colleagues, where they observed accumulation of SA metabolites in response to K+ limitation in Arabidopsis [39]. Although the time course of treatments is quite different, this study showed that the accumulation of these SA metabolites was not observed until 10 to 20 days after transfer to low K+ media. In contrast, here, we observed a marked increase in SA after only 6 days. Our transcriptomic data also supports a more rapid response of broccoli to induce SA accumulation, as compared to Arabidopsis. Further studies will be required to determine whether this induction of SA mediates tolerance to limiting K+, but previous studies suggest that SA can enhance the activity of the H+-ATPase, thereby altering membrane potential and limiting K+ leakage through GORK channels under conditions of saline stress [58]. It will be interesting to investigate whether this mechanism can contribute, along with the high levels of BoHAK5 expression, to the slower decrease in K+ content in broccoli roots observed here. Notably, we analyzed the BoHAK5 promoter using the PlantCARE database [59] and observed that it contains an SA-responsive TCA-element 440 bp upstream of the ATG. However, given the multiple roles of SA in both biotic and abiotic stress responses [60], other mechanisms must also be considered. For example, many studies have documented the crosstalk between nitric oxide (NO) and SA signaling and their relation to oxidative stress[38]. Our transcriptional data indicate that K+ limitation leads to iron deficiency, which may induce the hypoxia response, as discussed above. In turn, NO signaling and subsequent increased NO3− requirements, may lead to the activation of SA biosynthesis and signaling observed here. In any case, it is clear that future studies in this area are warranted.

5 Conclusions

In this study, we examined the short and long-term response of broccoli to limiting K+ conditions and we performed a comprehensive analysis of the transcriptional response of broccoli and Arabidopsis to the initial stages of K+ limitation in a time-resolved and tissue specific manner. Our results provide a wealth of information regarding K+-responsive genes in an increasingly important crop species. We have identified key ion transporters, like BoHAK5, the NRT2 family and PHT1;3 orthologues, which are now viable candidates for future studies aimed at characterizing the molecular details of ion transport and the coordination of nutrient uptake. Furthermore, we provide data regarding the expression of genes involved in the glucosinolate pathway in broccoli, which will be helpful for future studies aimed at fortifying the accumulation of these health-promoting metabolites. Lastly, our results reveal a robust and fast accumulation of SA mediated by the induction biosynthesis genes of the isochorismate pathway. Taken together, our data reveal several excellent candidates that can be explored as biotechnological targets to improve the performance of crop plants to limiting K+ conditions.