Potassium uptake is one of the most basic processes of plant physiology. However, a comprehensive description is lacking. At a cellular level fungi have provided a helpful but imperfect plant model, which we aim to improve using Physcomitrella patens. Blast searches in expressed sequence tag databases demonstrated that Physcomitrella expresses the same families of K+ and Na+ transport systems as flowering plants. We cloned two inward rectifier channels, PpAKT1-2, and four HAK-type transporters (PpHAK1-4). In both types of transport system, phylogenetic analyses revealed that despite their high sequence conservation they could not be included in Arabidopsis or rice (Oryza sativa) clusters. Both inward rectifier channels and one HAK transporter (PpHAK1) were expressed in yeast. PpAKT1 and activated mutants of PpAKT2 and PpHAK1 showed clear functions that were similar to those of homologous systems of flowering plants. A pphak1 null mutant line of Physcomitrella failed to deplete K+ below 10 μm. Moreover, in a non-K+-limiting medium in which wild-type plants grew only as protonema, pphak1-1 plants produced leafy gametophores and contained 60% more K+. We found that Physcomitrella takes up K+ through several systems. PpHAK1 is the dominant system in plants that underwent K+ starvation for long periods but an as-yet unidentified system, which is non-selective for K+, Rb+, and Cs+, dominates in many other conditions. Finally, we discuss that, similar to PpHAK1, one of the functions of AtHAK5 may be to control cellular K+ content and that a non-selective as-yet unidentified system also exists in Arabidopsis.
Potassium is the most abundant cation in the internal milieu of all living cells. The selection of this cation for cellular electrical-charge and osmotic adjustments occurred early in the evolution of life and has been conserved with very few variations even in the most extreme halophilic species. The constant source of K+ and the high Na+ concentration in the marine environment in which early life evolved have shaped the mechanisms of ionic homeostasis of extant marine organisms. This also applies to terrestrial animal species, because they evolved so that their cells continue to be bathed in a sea-like liquid, which they produce from food minerals and ingested fresh water. These conditions have determined the animal-cell paradigm of K+ and Na+ transport. Meanwhile, in order to adapt to terrestrial environment where soluble minerals are scant, plants acquired the capacity to fulfill their cellular K+ requirements from environments in which K+ concentrations could be more than a thousand times lower than in seawater (Rodríguez-Navarro, 2000; Rodríguez-Navarro and Rubio, 2006).
We selected the moss Physcomitrella patens to be used as a plant model that may complement studies of K+ transport in Arabidopsis. Physcomitrella has well-established advantages as a plant model, especially in relation to its high rate of homologous recombination (Cove et al., 2006; Frank et al., 2005; Reski, 2005; Schaefer and Zryd, 2001). With respect to using Physcomitrella as a model for cellular K+ transport in flowering plants, it should be noted that vascular plants are not derived from bryophytes but rather the two lineages diverged in similar terrestrial environments >400 million years ago (Ma) (Goremykin and Hellwing, 2005; Palmer et al., 2004; Qiu et al., 2006; Yoon et al., 2004). Thus the components of K+ transport models that are coincident in Physcomitrella and Arabidopsis are probably shared by all flowering plants.
An additional benefit of research using Physcomitrella is that mutant lines defective in K+ transport, which can be generated by homologous recombination, can be used for functional expression of transporters in flowering plants. Currently, yeast and bacterial mutants, as well as Xenopus oocytes, are being used to express plant transport proteins (Dreyer et al., 1999). However, with the advance of knowledge, the use of non-plant models for the expression of plant transport systems becomes more problematic as more details of their function and regulation are sought. For example, there may be transport activations and signaling processes that do not exist, and therefore cannot be investigated, in distant heterologous systems. Furthermore, artifacts may be produced with the ectopic expressions of plant transport systems in distant species that may modify the proteins via inappropriate pathways. The need to co-express AtSOS2 and AtSOS3 for the functional expression of AtSOS1 in yeast (Quintero et al., 2002), AtCBL1 and AtCIPK23 for the expression of AtAKT1 in Xenopus oocytes (Li et al., 2006; Xu et al., 2006), and the modification of the voltage-dependent gating of AKT2 in animal cells (Latz et al., 2007) or the confusing results that are obtained when TaHKT1 and HvHKT1 are expressed in yeast cells (Haro et al., 2005) are just a few examples of the above-mentioned problems.
As a first step in our efforts to establish a comprehensive model of K+ and Na+ uptake in Physcomitrella, we identified several Physcomitrella genes that could encode K+ and Na+ transporters. From these genes, we cloned two that encode AKT1-type channels and four that encode HAK-type transporters. Expression in yeast of the cloned channels and of the PpHAK1 transporter, together with pphak1-1 mutant experiments, revealed that PpHAK1 participates in K+ uptake at many K+ concentrations and is involved in the control of cellular K+ content. However, PpHAK1 does not dominate K+ uptake, except after long periods of K+ deprivation. In other conditions the dominant transport system, which is non-selective for K+, Rb+, or Cs+, has not yet been identified. Interestingly, most of our findings in Physcomitrella apply to Arabidopsis as well.
K+ and Na+ transport systems in Physcomitrella
We first identified several Physcomitrella expressed sequence tags (ESTs), e.g. BY949858, whose translated sequences exhibited up to 88% identity with H+-pump ATPases of flowering plants. In the Physcomitrella genome, the sequence of which has been recently released, there are two genes (estExt_fgenesh2_pg.C_1370088 and e_gw1.404.27.1 in http://genome.jgi-psf.org//Phypa1_1/Phypa1_1.home.html) that could encode H+-pump ATPases very similar to those of flowering plants, and three genes that could encode ATPases that are similarly distant from fungal and flowering plant H+-pump ATPases. These results indicate that the fungal and plant model of energization of plasma membranes applies to Physcomitrella.
We then carried out a computer-based search among translated ESTs, using the amino acid sequences of different types of K+ and Na+ transport proteins as queries. The results revealed that Physcomitrella expresses transcripts that encode K+ and Na+ transport proteins that belong to most of the families identified in Arabidopsis and rice (Oryza sativa; Table S1). Based on these search results, we can predict that Physcomitrella expresses several shaker-type K+ channels that may belong to Groups I, II, and III (AKT1/KAT1/VFK1) and Group V (SKOR/GORK), but perhaps not to Group IV (AtKC1) (Pilot et al., 2003). We also identified CNGC-type and vacuolar TPK/KCO-type channels, as well as HAK transporters and antiporters of the SOS1, NHX, NHAD, CHX, and KEA types. No translated EST showing similarity to HKT transporters was found, but a Physcomitrella HKT1 gene exists (fgenesh1_pg.scafold_63000099 at http://genome.jgi-psf.org//Phypa1_1/Phypa1_1.home.html). Expressed sequence tags or genes that encode putative homologs of the fungal ACU K+-ATPases (Benito et al., 2004) were not found.
Cloning of PpAKT1 and PpAKT2 genes
To clone the Physcomitrella functional homolog of AtAKT1 we selected two ESTs (BJ600825 and BJ579461). Using primers that were designed from the EST sequences and a cDNA preparation of Physcomitrella plants grown in 1.8 mm K+, we cloned the corresponding full-length cDNAs, PpAKT1 and PpAKT2, by RT-PCR. The ends of the 5′ and 3′ untranslated sequences of the cDNAs were then used to design primers and clone the corresponding genes by PCR on a sample of Physcomitrella genomic DNA.
A comparison of the gene and cDNA sequences revealed the presence of 13 introns in both PpAKT1 and PpAKT2. These introns did not overlap entirely. Moreover, the intron pattern in these genes did not coincide exactly with any of those described for shaker-type K+ channels in Arabidopsis (compare Figure S1 with Table 4 in Pilot et al., 2003).
The PpAKT1 and PpAKT2 cDNAs encoded two proteins that were 944 and 967 amino acids long, respectively, slightly longer than AtAKT1. The amino acid sequences exhibited all the characteristics of shaker plant K+ channels (Gambale and Uozumi, 2006; Pilot et al., 2003) but neither of them fit any of the five phylogenetic groups into which flowering plant channels have been divided (Pilot et al., 2003). The PpAKT1 and PpAKT2 channels were equally distant from the Groups I to IV, and more distant from Group V, which includes the outward rectifiers SKOR and GORK and is the most divergent group in Arabidopsis (Figure 1a). The phylogenetic divergence of PpAKT1 from PpAKT2 was similar to that existing between the monocot and dicot channels TaAKT1 and AtAKT1. A conspicuous characteristic that divides plant shaker-type channels is the presence of ankyrin repeats after the nucleotide-binding domain. These repeats are present in all channels identified in flowering plants, with the exception of some members of Group II, e.g. KAT1, and Group IV, e.g. AtKC1 (Gambale and Uozumi, 2006; Pilot et al., 2003). Similar to wheat (Triticum aestivum) AKT1, five ankyrin repeats were detected in PpAKT1 and PpAKT2 (http://www.embl-heidelberg.de/~andrade/papers/rep/search.html); this differs from Arabidopsis AKT1 for which the same analysis predicted only four repeats.
Functional expression of PpAKT1 and PpAKT2 in yeast cells
PpAKT1 and PpAKT2 cDNAs were cloned into the yeast expression vector pYPGE15 (Brunelli and Pall, 1993) under the control of the PGK1 gene promoter and transformed into a trk1 trk2 yeast mutant defective for high-affinity K+ uptake. To optimize the yeast expression we eliminated most of the polylinker sequences of the vector (Haro et al., 2005) and modified the sequence context around the AUG codon (Hamilton et al., 1987; see Experimental procedures). In these optimized constructs, PpAKT1 clearly suppressed the defect of the trk1 trk2 yeast mutant at low K+, while the effect of PpAKT2 was barely detectable (Figure 1b).
A major obstacle in conducting a complete kinetic study of the capacity of PpAKT1 to transport alkali cations in yeast cells is the channel discrimination between K+ and Rb+ (Choe et al., 2000), which prevented the use of Rb+ as a K+ analog (Rodríguez-Navarro, 2000). To overcome this obstacle, we took advantage of the low toxicity of Rb+ in yeast (Camacho et al., 1981) and tested K+ uptake in K+-starved, Rb+-loaded yeast cells (these cells contained 20 nmol K+ and 250 nmol Rb+ mg−1 versus 250 nmol K+ mg−1 in regular K+-starved cells). In these Rb+ yeast cells, PpAKT1 showed a medium affinity for K+ (Km = 0.3 mm) while in regular K+-starved yeast cells (without Rb+) it mediated Rb+ and Cs+ influxes that exhibited low affinity for Rb+ (Km = 2 mm), and a much lower affinity for Cs+ (Km ≥ 50 mm) (Table S2). At much lower concentrations than its Km, Cs+ blocked the transport of Rb+ mediated by PpAKT1. For example, at equimolar concentrations (0.5, 5, and 10 mm) of both Rb+ and Cs+, Cs+ strongly inhibited Rb+ influx. The mean inhibition of 0.5 mm Cs+ was 85 ± 3% (±SD, n =3), and at 5 and 10 mm, the mean inhibition was 69 ± 2% (±SD, n =2 at each concentration).
In contrast with the fairly efficient transport of K+ and Rb+, or even Cs+, PpAKT1 did not significantly increase the intrinsic Na+ uptake of the yeast mutant (in uptake experiments from 1 to 100 mm Na+). The theoretical prediction of a massive Na+ influx through KIR channels at 100 mm Na+ (White, 1999) was not confirmed in PpAKT1.
PpAKT2 barely suppressed the defect of the yeast mutant (Figure 1b) and precise analyses of Rb+ uptake demonstrated that PpAKT2 exhibited a very low Vmax of Rb+ influx. This poor functional expression might be the result of the incapacity of yeast cells to activate the channel (Li et al., 2006; Xu et al., 2006), but activation might be mimicked in mutant channels. Therefore, we carried out a selection of spontaneous PpAKT2 mutants that supported the growth of the trk1 trk2 yeast cells at 100 μm K+ and found three mutants, PpAKT2-1, PpAKT2-2, and PpAKT2-3. Figure 1(b) shows growth enhancement and Figure S1 shows the position of the mutations. The PpAKT2-1 channel carried a D514V mutation that was located after the nucleotide-binding motif of the channel and just before the ankyrin repeats, in a region that is highly conserved in groups I–III. Most shaker-type channels conserve either a D or E residue at this position, and GORK/SKOR channels have an S residue, which has a low probability of phosphorylation (http://www.cbs.dtu.dk/services/NetPhos/). In the PpAKT2-2 mutant a stop signal is substituted for the triplet encoding G823; and in PpAKT2-3 there is a frame shift mutation that adds 12 new amino acids after the K771 residue of the channel, NPQCRVSAVSSA, before termination of translation. Interestingly, according to Daram et al. (1997), the PpAKT2-1 mutation was a substitution in region 2B, and the PpAKT2-2 and PpAKT2-3 mutations produced deletions in region 4; both regions are involved in the tetramerization of the AtAKT1 channel.
A kinetic study of the three PpAKT2 mutant channels revealed that their cation selectivities were identical, which might be due to the fact that none of the mutations affected the channel’s selectivity filter. In comparison with PpAKT1, PpAKT2 mutant channels exhibited higher Kms for K+ and Rb+, and lower K+/Rb+ selectivity, as well as null capacity for transporting Cs+ (Table S2). However, Cs+ blocked the activity of the channel and inhibited Rb+ influx by 70% in equimolar Rb+/Cs+ experiments. The mean inhibition of uptake experiments with 5.0 mm and 10.0 mm of both Rb+ and Cs+ in the three mutants was 68.1 ± 3.8% (±SD, n =2 for each mutant at each concentration).
Cloning of PpHAK1, PpHAK2, PpHAK3, and PpHAK4 genes
Our computer-based search identified 32 ESTs that could correspond to the transcripts of at least four HAK genes. Using RT-PCR we cloned four HAK cDNAs that included all ESTs. Then the ends of the 5′ and 3′ untranslated sequences of the four cDNAs were used to design primers and to clone the corresponding four genes, PpHAK1, PpHAK2, PpHAK3, and PpHAK4, by PCR on a genomic DNA sample of Physcomitrella. In the recently released genome sequence of Physcomitrella, there are 18 genes encoding putative HAK transporters (Figure S2). Before the release of the genome sequence, we carried out an extensive cloning of HAK cDNA fragments using degenerated primers (Rubio et al., 2000; Santa-María et al., 1997) in normal and K+-starved plants, but never identified a HAK cDNA fragment different from those that we cloned. These results and the identity of the HAK ESTs suggest that the expression of most of the HAK genes, except PpHAK1, -2, -3, and -4, is low.
Alignment of the cloned HAK cDNAs and genes revealed the existence of seven introns in PpHAK1 and eight introns in identical positions in PpHAK2, PpHAK3, and PpHAK4 (Figure S3). The intron distribution in PpHAK2, PpHAK3, and PpHAK4 was similar to that of AtKT1/KUP1 (Ahn et al., 2004) and to that of OsHAK1 (Bañuelos et al., 2002) with an additional intron in the 5′ untranslated region (5′ UTR). The intron distribution of the PpHAK1 gene was similar to that of AtKT12/KUP12, which encodes a chloroplast transporter (Kleffmann et al., 2004; Peltier et al., 2004).
The cloned cDNAs of PpHAK1, -2, -3, and -4 encoded proteins of 822, 825, 820, and 819 amino acids, respectively. The analysis of the transporter sequences revealed that PpHAK1 was quite divergent from PpHAK2, PpHAK3, and PpHAK4. The latter three transporters showed approximately 80% identity among them and 45% with PpHAK1. Phylogenetic analysis of these and related transporters (Figure 2) revealed that PpHAK1 did not belong to any of the previously described groups (Rubio et al., 2000), whereas the other three transporters might belong to group III. This uncertainty regarding the phylogenetic position of the cloned HAK transporters did not apply to 10 transporters of this family, which clearly belong to group II (Figure S2).
In-frame GFP fusions to the PpHAK1 and PpHAK2 cDNAs and transient expression in Physcomitrella protoplasts did not enable us to accurately predict the localization of the PpHAK1 transporters. However, the location of PpHAK1 at the plasma membrane could be deduced from its function (see below). In contrast, comparison of the position of the GFP signal with reference to the autofluorescence of the chloroplasts strongly suggested that PpHAK2 was not located at the plasma membrane (Figure S4) and could not mediate K+ uptake from the external medium.
Expression of PpHAK1 in yeast cells
PpHAK1 cDNA was cloned into vector pYPGE15 and transformed into the yeast trk1 trk2 mutant. Even in optimized constructs, in which no sequence remains from polylinkers were present and with an optimum context around the first in-frame AUG codon, PpHAK1 only slightly suppressed the defect of the yeast mutant. However, precise tests of Rb+ uptake revealed that PpHAK1 mediated a Rb+ influx that exhibited a Km of 0.2 mm Rb+ and a Vmax of 0.9 nmol Rb+ mg−1 min−1. Similar experiments with Cs+ demonstrated that PpHAK1 also transported Cs+ at rates that were similar to those found for Rb+. Assuming that K+ and Rb+ were taken up at the same rates, as normally occurs in HAK transporters (Rodríguez-Navarro, 2000), the slow K+ uptake mediated by PpHAK1 could be expected not to improve the growth of the trk1 trk2 yeast mutant at low K+ concentrations.
Using the same reasoning employed with PpAKT2, that yeast cells express a low-activity form of PpHAK1, we searched for activated mutants, which appeared spontaneously with high frequency. Among the identified mutations, we did not find one equivalent to that which activates AtHAK5, L776H at the C-terminus of the protein (Rubio et al., 2000). Therefore, we constructed the equivalent mutant in PpHAK1 but it did not improve the growth of the trk1 trk2 yeast cells. To test whether the region has a regulatory function, we constructed a mutant, PpHAK1-5, with four amino acid changes that reproduced the C-terminus of the very active OsHAK1 transporter. This mutant suppressed defective growth of trk1 trk2 yeast cells at 100 μm K+.
In addition to this C-terminal constructed mutant, we studied four additional mutants that allowed the growth of the trk1 trk2 strain at 100 μm K+ and exhibited high-affinity Rb+ influx. Figure 3 shows the position of these mutations and the yeast growth-enhancing effect at low K+. Table 1 summarizes the kinetic characteristics of the Rb+ influx mediated by the PpHAK1 mutants in trk1 trk2 yeast cells. One of them, R443S (PpHAK1-1), was equivalent to R397S described previously in the barley (Hordeum vulgare) transporter HvHAK1 (Senn et al., 2001), but the other three have not been described. PpHAK1-–3 yeast cells depleted external K+, Rb+, or Cs+ very efficiently (Figure S5), as HAK1-type transporters typically do (Bañuelos et al., 2002; (Rodríguez-Navarro and Rubio, 2006). In PpHAK1-1 and PpHAK1-2 the mutation produced a pronounced effect on the Km and very little effect on the Vmax, which strongly suggests that these phenotypes did not result from an increase in the amount of protein targeted to the plasma membrane. Expression of PpHAK2 in yeast cells did not enhance K+ uptake, which is consistent with its probable location in the tonoplast of yeast cells (Figure S4).
Table 1. Kinetic parameters of Rb+ influx mediated by PpHAK1 and its mutantsa
Vmax (nmol mg−1 min−1)
aThe SD of three or four independent determinations was typically around or below 10% of the mean for Km values of 50–100 μm and around 30–40% of the mean for Km values around 10 μm. ≈10 means a value of between 8 and 15; <10 means a value of around 5.
Effects of the pphak1-1 mutation on high-affinity K+ uptake
To test whether PpHAK1 mediated high-affinity K+ uptake in Physcomitrella, we isolated the pphak1-1 mutant line, which carries the insertion of a neomycin resistance cassette into the PpHAK1 gene. As described in the Experimental procedures, PCR amplifications and sequencing of the PpHAK1 locus in the pphak1-1 line confirmed that the neomycin-resistant cassette was inserted into the PpHAK1 gene as expected (Figure 4a,b). To rule out the possibility that another copy of the cassette was inserted elsewhere in the genome, we performed Southern blot analyses with genomic DNA from pphak1-1 plants using several restriction enzymes and a probe that hybridized to the resistance cassette. As shown in Figure 4(c), the estimated sizes of the hybridized bands corresponded exclusively to the DNA fragments expected from the insertion of the cassette into the PpHAK1 gene. Lines with multiple insertions would have produced multiple bands of exactly the same intensity in every digestion because the probability that the digestion fragments of two insertions have the same lengths is very low. After these tests the pphak1-1 line was used for physiological analyses.
The first difference that we found between wild-type and pphak1-1 plants was the lower capacity of mutant plants to deplete external K+. Plants of both lines that had been grown under high-K+ conditions (1.8 mm) in PPNH4 medium lost a small amount of K+ when transferred to the K+-free medium (KFM) but recovered within 2–3 days. After another 2–3 days wild-type plants were able to deplete external K+ down to concentrations that occasionally were undetectable by atomic emission spectrophotometry (<0.1 μm). In contrast, in the majority of experiments pphak1-1 plants were unable to deplete the external K+ below 10 μm. In only 20% of the experiments pphak1-1 plants depleted K+ to 7–8 μm K+.
Kinetic analyses of K+ and Rb+ influxes in wild-type plants that were grown in PPNH4 medium and then K+ starved 4 weeks in KFM revealed that a high-affinity transport system (Kms ≈ 2 μm K+ or Rb+) mediated most of these influxes. Similar experiments with pphak1-1 plants demonstrated that the high-affinity transport system was completely absent in these mutant plants (Figure 5a), which was consistent with their impaired capacity to deplete external K+ below 7–10 μm.
To investigate the involvement of non-PpHAK1 K+ transport systems in K+ uptake at low K+ concentrations (∼100 μm) we carried out further comparisons between pphak1-1 and wild-type plants. For these experiments the reduction of the K+ starvation period to only 1 week and the use of plants that had been grown previously in different media, either PPNH4 medium or in the two Knop-KFMK (K+-free medium plus KCl) media (described in Experimental procedures), produced the best results (Figure 5b). In 1-week K+-starved wild-type plants the high-affinity uptake process was dominant in PPNH4 plants (Km ≈ 8 μm) but was almost absent in Knop-KFMK plants (Km > 200–300 μm). This was concluded because in Knop-KFMK plants the semi-logarithmic plot of Rb+ or K+ depletion (logarithm of the external cation concentration versus time) was almost linear, which corresponds to first-order kinetics. This substrate concentration dependence takes place in Michaelis–Menten processes when the Km is much higher than the substrate tested-concentrations (< 50 μm in Figure 5b), then k ≈ Vmax/Km.
Continuing the study with 1-week K+-starved PPNH4-grown plants we found that the pphak1-1 mutation eliminated the high-affinity uptake of K+ (Figure 5c) and Rb+ (Figure 5d). Therefore, again, the depletion time courses followed first-order kinetics (linear semi-logarithmic plots). As shown in Figure 5(c) and (d), these experiments were started at 120–150 μm to calculate the kinetic constants with more precision (see the values in the legend). To determine the complete concentration dependence of Rb+ influx in pphak1-1 plants, we measured Rb+ uptake at 0.5 and 5 mm Rb+ in pphak1-1 plants that had been prepared as in Figure 5(d). The obtained data, coupled with those presented in Figure 5(d), showed that pphak1-1 plants took up Rb+ following a Michaelis–Menten process with a Km of 700 μm Rb+.
In summary, in our tests, the high-affinity K+ or Rb+ uptake exhibited by K+-starved wild-type plants, Kms of 2–8 μm, was lost in pphak1-1 plants, in which the Km of Rb+ influx (700 μm Rb+) cannot be considered to correspond to a high-affinity process. The Km for K+ could not be determined for technical reasons, but assuming the same Vmax for both cations, the K+Km would be 420 μm.
A further characterization of the cation uptake systems operating in 1-week K+-starved pphak1-1 plants was possible using Cs+. In these plants, Cs+ was taken up very rapidly for 5 min and was then dramatically auto-inhibited when the Cs+ content reached 5–10 μmol Cs+ g−1 dry weight (Figure 5e). The Cs+ uptake rate before the delayed auto-inhibition was intermediate between the rates of K+ and Rb+ uptake. In contrast, in wild-type plants prepared in the same way, the delayed auto-inhibited Cs+ influx represented only a portion of the total Cs+ influx (Figure 5f). These results demonstrated that PpHAK1 mediated a Cs+ uptake that was not auto-inhibited, as described for other plants (Bañuelos et al., 2002; Rodríguez-Navarro, 2000). They also supported the notion that PpHAK1 and the system that exhibited a delayed Cs+ auto-inhibition worked in parallel in wild-type plants.
The K+-starved plants exhibited a high-affinity Na+ uptake that was not affected by the pphak1-1 mutation. Occasionally, this uptake was faster in pphak1-1 plants than in wild-type plants. This suggests that the aforementioned HKT1 gene that exists in the genome of Physcomitrella encodes a high-affinity Na+ uptake system (Rodríguez-Navarro and Rubio, 2006).
Effects of the pphak1-1 mutation in plants grown at high K+
At non-limiting K+ concentrations a notable morphological difference between wild-type and pphak1-1 plants occurred in Knop medium (5.2 mm K+): wild-type plants grew exclusively as protonema whereas the pphak1-1 line produced many leafy gametophores (Figure 6a,b). Analyses of the K+ content of these plants revealed that the K+ content of pphak1-1 mutant plants was roughly 60% higher than that of wild-type plants (1220 ± 120 versus 780 ± 50 nmol K+ mg−1 dry weight). After transferring Knop plants to KFMK (1.8 mm K+) for 2–3 weeks the morphology of the plants changed but wild-type and mutant plants were still different. Wild-type plants started to produce leafy gametophores, resembling the pphak1-1 plants in Knop medium, and the pphak1-1 plants were transformed into very long gametophores with some rhizoids (Figure 6c,d). In contrast to these differences, wild-type and pphak1-1 plants were not appreciably different when grown in PPNH4 solid medium.
Consistent with the aforementioned morphological differences, we found that PpHAK1 was functional in non-K+-starved wild-type plants, although high-affinity K+ uptake was not detectable in these plants. For example, the Rb+ uptake rate in the pphak1-1 mutant line at 0.5 mm Rb+ was half that in the wild type (0.25 versus 0.51 nmol Rb+ min−1 mg−1 in plants that were growing in KFMK with 1.8 mm K+). However, the defective function was more clearly observed with Cs+ uptake tests because in pphak1-1 plants Cs+ uptake was completely abolished while wild-type plants took up Cs+ at a rate that was only 25% slower than that of Rb+ (0.37 versus 0.51 nmol min−1 mg−1) (Figure 7). It is worth observing that the types of experiments shown in Figure 7 cannot detect a Cs+ uptake that is auto-inhibited at a content of 5–10 μmol Cs+ g−1 dry weight.
Searches in EST databases and genome sequences indicate that all typical plant K+ and Na+ transport system families exist in Physcomitrella. The absence of a certain functional form of a transporter in a family, e.g. an AtKC1-like channel, cannot be established with certainty by sequence analysis. This conclusion was suggested by the phylogenetic analyses of the AKT channels and HAK transporters that we have cloned, which do not clearly belong to any of the previously established phylogenetic groups (Figures 1a and 2). Despite this, PpHAK1 may fulfill the same function as AtHAK5 (see below). Therefore, it seems now that not only functional equivalences but also orthological relationships between Physcomitrella and Arabidopsis K+ transporters cannot be determined by using the phylogenetic analyses that have been performed with flowering plant sequences. This includes the analysis of the intron patterns of the genes and can be explained by the aforementioned 400 million years of independent evolution of mosses and flowering plants. Therefore, the remarkable findings of our study are the identities of the transport system families. The genetic coincidences in this case are considerably greater than more universal genomic comparisons, in which the coincidence may amount to <70% (Nishiyama et al., 2003). This observation emphasizes the importance of the K+ and Na+ transport systems for all land plants and suggests that they were functional before the divergence of bryophytes and vascular plants.
As already mentioned, PpAKT1 may be functionally homologous to AtAKT1. This possibility needs to be tested with ppakt1 plants, which can be obtained by homologous recombination. The mutational analysis that we carried out in yeast cells in combination with expression of the mutant channels in ppakt1 plants and electrophysiology (Ermolayeva et al., 1996, 1997; Johannes et al., 1997; Klein et al., 2003) in the same plants open exciting new possibilities for studying the functions of inward rectifier K+ channels.
Similarly, Physcomitrella offers a great opportunity to discover the functions of the many members of the large family of KT/HAK/KUP transporters that do not mediate high-affinity K+ uptake. This applies to the cloned PpHAK2, -3, and -4, which may not be located on the plasma membrane. Some transporters of the KT/HAK/KUP family (At4g23640, At4g33530, and At5g09400 in Arabidopsis and OsHAK10 in rice) localize to the tonoplast (Bañuelos et al., 2002; Jaquinod et al., 2007), while another (At1g60160) is found in the chloroplast (Kleffmann et al., 2004; Peltier et al., 2004). The large number of HAK transporters in Physcomitrella, 18 versus 13 in Arabidopsis (Mäser et al., 2001) and 26 in rice (Amrutha et al., 2007), indicates that these transporters may fulfill important cellular functions that can be investigated in Physcomitrella.
The functions of PpHAK1
The kinetics of Rb+ or K+ influxes in 1-week K+-starved wild-type plants of Physcomitrella can be explained by the parallel function of at least two processes, one of them being the high-affinity transport mediated by PpHAK1, which disappeared in pphak1-1 plants. In addition, PpHAK1 may mediate approximately 50% of the Rb+ uptake at 0.5 mm Rb+ and 100% of the uptake of Cs+ when compared with wild-type plants (Figure 7). Unexpectedly, the disruption of the PpHAK1 gene dramatically increased the K+ content of the plants.
The absence of a K+ transporter does not necessarily imply a lower K+ content, even when it results in lower K+ influx, because influx may greatly exceed the net uptake that is required to support growth. In fact, low-affinity K+ transport is characterized by futile cycling of K+ at the plasma membrane of barley (Szczerba et al., 2006). Our results suggest that PpHAK1 participates in the control of K+ content in plants growing under high-K+ conditions, which may occur by controlling the K+ content set point regulating either K+ efflux or K+ influx. Remarkably, findings obtained with recombinant inbred lines (RILs) derived from the parental accessions Cape Verde Islands and Landsberg erecta (Alonso-Blanco et al., 1998) indicate that AtHAK5 probably fulfills a similar function in Arabidopsis. In these RILs the K+ content varies from 640 to 1200 nmol K+ mg−1 (dry weight) in plants grown at 2 mm K+ (in medium with very low, 0.12 μm, NH4+) and AtHAK5 coincides with a quantitative trait locus for K+ content in these lines (Harada and Leigh, 2006).
In parallel with the described changes, the pphak1-1 disruption affected the morphology of plants grown under high-K+ conditions (Figure 6). It is not clear whether K+ content and morphological changes in pphak1-1 plants are pleiotropic independent effects that are produced by the pphak1-1 mutation or whether there is a causal relationship. However, it would not be surprising if a change in K+ content of a plant produced a physical change because K+ content is a relevant parameter for plant growth. K+ content does not correlate with either fresh or dry weight but correlates strongly with the fresh to dry weight ratio or with the water content across many species (Broadley et al., 2004). Indeed, two Arabidopsis mutants with morphologic defects, root hair development (Rigas et al., 2001) and hypocotyl growth (Elumalai et al., 2002), have mutations in the AtKT3/KUP4 and AtKT2/KUP2 genes, which encode transporters of the same family as AtHAK5.
The diversity of mutations that affect the kinetic parameters of PpHAK1 in yeast cells (Figure 3 and Table 1) suggests that the activity of PpHAK1 is highly regulated and that the transporter may operate with a diversity of Kms. One of the mutations that activated PpHAK1 (R443S) was also obtained in a mutagenic analysis of the barley HvHAK1 transporter (Senn et al., 2001), and the regulatory functions of the C-terminus of PpHAK1 that can be deduced from our mutational analysis also apply to AtHAK5 (Rubio et al., 2000).
In summary, PpHAK1 has several important functions in Physcomitrella; high-affinity K+ uptake, control of K+ content, and a link between nutrition and morphogenesis. This suggests that the roles of PpHAK1 go beyond merely allowing K+ to cross the plasma membrane at low K+ concentrations.
Other K+ uptake systems
One-week K+-starved pphak1-1 plants took up K+ or Rb+ at fairly rapid rates at cation concentrations around 100 μm (Figure 5c,d), which is a normal K+ concentration in the soil solution (Schneider, 2003). This uptake could be produced by two or several systems, which is particularly important regarding PpAKT1 or another inward rectifier channel with a function similar to that of AtAKT1 in Arabidopsis (Dennison et al., 2001; Hirsch et al., 1998; Spalding et al., 1999). Considering their kinetic parameters (Table S1), neither PpAKT1 nor PpAKT2 could mediate the K+ and Rb+ uptake shown in Figure 5(c) and (d), which was too similar for channels that discriminate between K+ and Rb+. Therefore, the results obtained with pphak1-1 plants strongly suggest the existence of another unknown system. If this system operates alone, it might exhibit the aforementioned Kms of 0.7 mm Rb+ and 0.4 mm K+ and similar Vmaxs. If PpAKT1 (or PpAKT2) operates in parallel with the unknown system, a simple kinetic analysis using the kinetic data found for PpAKT1 in yeast (Kms = 0.3 mm K+, 2 mm Rb+) indicates that PpAKT1 could account for 40% of the K+ uptake (Figure 5c) and a negligible amount of the Rb+ uptake (Figure 5d). In this case, the second unknown system would exhibit similar Kms for K+ and Rb+ (0.7 mm) and would account for 60% of the K+ uptake and almost 100% of the Rb+ uptake. The remarkable conclusion in the two cases is the existence of an unknown system, which exhibits a low discrimination between K+ and Rb+ and performs a relevant function.
Our study indicates the existence of a Cs+ uptake system in Physcomitrella that is independent of PpHAK1 and that shows delayed auto-inhibition (Figure 5e). Interestingly a homologous system may have been described in Arabidopsis (Broadley et al., 2001), although it has received very little attention. In plants grown at 10 mm K+, a rapid Cs+ uptake (Km = 2.6 mm Cs+, Vmax = 17 μmol Cs+ g−1 fresh weight root h−1) was found that was entirely conserved in atakt1 plants. Although not at a high rate, such an active transporter is still effective at 100 μm Cs+. Similar to our findings in Physcomitrella, Cs+ uptake in Arabidopsis plants stopped when the Cs+ content in roots was 1 μmol g−1 fresh weight (Figure 2 in Broadley et al., 2001), an amount similar to the 5–10 μmol Cs+ g−1 dry weight that Physcomitrella plants had taken up when Cs+ uptake stopped (Figure 5e,f).
It may be questioned why Cs+ uptake stops in the aforementioned experiments with Arabidopsis plants, in which AtHAK5 should transport Cs+ without experiencing auto-inhibition (Rubio et al., 2000). One possibility is that AtHAK5 functioned poorly in the tested plants due to the growth conditions. In Physcomitrella PpHAK1 was scarcely functional in plants grown in Knop-KFMK media (Figure 5b) and the presence of NH4+in the medium affected the expression of the HAK1 transporter in pepper (Capsicum annuum) (Martínez-Cordero et al., 2004, 2005) and barley (Santa-María et al., 2000) plants. In addition, the AtHAK5 transporter may have a low activity in the Wassilevskija ecotype. This can be deduced from the ratios between the Vmaxs of the high- and low-affinity systems in K+-starved Arabidopsis plants, which are 1 and 0.26 in Columbia and in Wassilevskija ecotypes, respectively (Gierth et al., 2005). Therefore, the combined effect of the growth medium and the use of non-K+-starved plants from the Wassilevskija ecotype could explain the functional absence of AtHAK5 in the aforementioned study of Cs+ uptake (Broadley et al., 2001).
An important question concerns the identity of the Cs+ transporter that operates in parallel with PpHAK1. It is unlikely to be specific for Cs+ and probably also transports K+ and Rb+ (White and Broadley, 2000; Zhu and Smolders, 2000). This suggests that the Cs+ transporter may be the aforementioned, thus far unidentified, K+ and Rb+ transporter (Figure 5c,d). This possibility urges its identification and cloning because its function seems relevant both for Physcomitrella and Arabidopsis. A low discrimination among K+, Rb+, and Cs+ is a characteristic of HAK transporters, and members of this large family are candidates. However, because auto-inhibition of Cs+ uptake has not been described in HAK transporters, other transport systems must also be considered. Among them members of the CNGC family are obvious candidates because they have been related to Cs+ accumulation and K+ content in Arabidopsis (Harada and Leigh, 2006; Payne et al., 2004; Vreugdenhil et al., 2004). Furthermore: (i) antisense AtCNGC10 plants have 50% less K+ than wild-type plants (Li et al., 2005) and show several morphological modifications (Borsics et al., 2007); (ii) atcngc3-null plants are more resistant to toxic concentrations of NaCl and KCl (Gobert et al., 2006); and (iii) AtCNGC2 is permeable to Cs+ (Leng et al., 2002).
In conclusion, our results suggest that bryophytes and vascular plants use very similar K+ and Na+ transporters. It is very likely that the bulk of K+ uptake is mediated by the combination of three systems: (i) a high-affinity HAK transporter, which may exhibit a wide range of K+Kms and be involved in controlling the set point of K+ content; (ii) an inward-rectifier K+ channel; and (iii) a non-selective as-yet unidentified system that may dominate uptake in many conditions. Physcomitrella is an excellent and simple tool for the study of these transporters at the cellular level.
Plants and growth conditions
The moss P. patens was maintained axenically in the PPNH4 medium described elsewhere (Ashton et al., 1979), supplemented with 7 g l−1 agar when required. Plants were grown either in biofermenters or jars in a phytochamber with a discontinuous white light (16 h light at 25°C/8 h dark at 18°C) at quantum irradiance of 200 μmol m−2 sec−1. Moss cultures were propagated by fragmenting old cultures in sterile water and transferring aliquots of this suspension to fresh medium. For physiological tests we used three additional media: Knop medium with reduced calcium concentration (Frank et al., 2005), in which the wild type grew exclusively as protonema; KFMK; and KFM. The K+ contents of these media (in mm) were: PPNH4, 1.8; Knop, 5.2; KFMK, 1.8. Table S3 summarizes the different compositions of these media. K+-starved plants were obtained by growing plants in KFM for different periods from 4 days to 4 weeks. Two types of plants were K+ starved, either plants grown for 3 weeks in PPNH4 or plants grown for 3 weeks in Knop medium and then for two or three additional weeks in KFMK (Knop-KFMK plants). In both cases the material tested was almost exclusively protonema cultures in the wild-type line. The morphology of the pphak1-1 mutant is described in text. Plants that had been K+-starved for 4–7 days had an almost normal K+ content, at most 10% lower than plants before starvation.
Bacterial and yeast strains, growth conditions, and plasmids
Escherichia coli strain DH5α was routinely used for plasmid DNA propagation. Functional tests in Saccharomyces cerevisiae were carried out in the strain WΔ6 (Mat a ade2 ura3 trp1 trk1Δ::LEU2 trk2Δ::HIS3; Haro and Rodríguez-Navarro, 2003), deficient in the TRK1 and TRK2 K+ uptake systems. This strain was maintained in YPD (2% peptone, 1% yeast extract, 2% glucose) supplemented with 50 mm K+. Yeast growth tests were carried out in arginine phosphate (AP) medium whose basic formulation does not contain K+, Na+, or NH4 + (Rodríguez-Navarro and Ramos, 1984). For functional expression tests in yeast, the cDNAs were cloned into the plasmid pYPGE15 (Brunelli and Pall, 1993), which has the PGK constitutive expression promoter. For the isolation of PpAKT2 and PpHAK1 mutants that suppressed the defective growth of the yeast trk1 trk2 mutant at low K+, 105–106 cells of the WΔ6 strain transformed with plasmids containing either PpAKT2 or PpHAK1 were inoculated in plates of solid AP medium containing 100 μm K+. The clones growing in this medium were isolated and studied.
Recombinant DNA techniques
Manipulation of nucleic acids was performed using standard protocols or, when appropriate, according to the manufacturers’ instructions. Polymerase chain reaction was performed in a Perkin-Elmer thermocycler using the Expand-High-Fidelity PCR System (Roche Diagnostics, http://www.roche.com/). The DNA sequencing was performed in an automated ABI PRISM 3100 DNA sequencer (Applied Biosystems, http://www.appliedbiosystems.com/). The PCR amplifications of HAK and AKT cDNA fragments were carried out on double-stranded cDNA synthesized from total RNA by using the cDNA Synthesis System Kit (GE Healthcare, http://www.gehealthcare.com/) and primers (Table S4) designed from the EST sequences identified in databases. Total P. patens RNA and DNA were prepared using the RNeasy Plant Kit and DNeasy Plant Kit (Qiagen, http://www.qiagen.com/). The 5′ ends of the cDNAs were obtained by using the 5′/3′ RACE Kit (Roche Diagnostics). Full-length cDNAs were amplified from Physcomitrella mRNA by standard RT-PCR methods using the specific forward and reverse primers that amplified fragments that included the ATG and STOP triplets (Table S4). The resulting PCR fragments were first cloned into the PCR2.1-Topo vector using the TOPO TA Cloning Kit (Invitrogen, http://www.invitrogen.com/). For expression in yeast, the PCR2.1-Topo constructs were digested with XbaI and KpnI and the fragments containing the cDNAs were then ligated into the yeast expression vector pYPGE15 (Brunelli and Pall, 1993) that had been previously digested with the same enzymes. In all cases we suppressed most of the polylinker sequences that preceded the translation initiation codon and created a sequence environment around this codon as similar as possible to A(U)AA(C)A AUGUCU(C) (Hamilton et al., 1987).
Localization of PpHAK1-GFP and PpHAK2-GFP in Physcomitrella protoplasts and yeast cells
For Physcomitrella protoplasts, the PpHAK1-GFP and PpHAK2-GFP constructs were in-frame fusions of the 3′ end of the PpHAK1 and PpHAK2 open reading frames to the GFP gene of the plasmid 35S-AdhI::GFP (Rubio-Somoza et al., 2006). To generate these constructs, the PpHAK1 full-length cDNA was amplified using the BglII-HAK1 primers, which include the BglII restriction site, and the PpHAK2 full-length cDNA using the BamHI-HAK2 primers, which include the BamHI restriction site (restriction sites are underlined in Table S4). The PpHAK1 or PpHAK2 PCR fragments were cloned into the BamHI site of the plasmid 35S-AdhI::GFP, which is at the 5′ end of GFP gene. The resulting constructs were then used for transient expression in Physcomitrella protoplasts, which were transformed following the polyethylene glycol method (Hohe et al., 2004). For large-scale protoplast isolation, we used pH-controlled bioreactor cultivation in modified Knop medium with reduced calcium concentration according to the protocol described by Hohe et al. (2004). After transformation, the protoplasts were kept in the dark for 24 h in PPNH4 medium supplemented with 3% mannitol and 5% glucose, followed by cultivation in the same medium for 4–6 days under normal growth conditions.
For yeast cells, the PpHAK1-GFP and PpHAK2-GFP fusions were cloned into the plasmid pYPGE15 (Brunelli and Pall, 1993). These constructs were transformed into the aforementioned trk1 trk2 yeast mutant.
The GFP fluorescence signals in Physcomitrella protoplasts and yeast cells were visualized by using a confocal ultraspectral Leica microscope TCS-Sp2-AOBS-UV (Leica Microsystems, http://www.leica.com/).
Generation of the pphak1-1 knockout line
The PpHAK1 knockout fragment that was used to disrupt the PpHAK1 gene was constructed by inserting the neomycin resistance cassette into a unique SalI site of a fragment of the PpHAK1 gene cloned in the PCRTopo 2.1 vector. In this construct two fragments of 1499 and 2738 bp of the PpHAK1 gene flanked the selection cassette. The cassette included the nptII gene under the control of the nos 5′ promoter and nos 3′ terminator and was obtained by PCR amplification from the pBIN19 vector using the primers nos F and R (Table S4). Knockout mutants were generated by transforming Physcomitrella protoplasts with 25 μg of a linear DNA fragment obtained by digesting the knockout construct vector with the AvaI restriction enzyme, which cuts twice into the PpHAK1 gene leaving two PpHAK1 fragments at the 5′ and 3′ ends (Figure 4). Stable antibiotic-resistant clones were selected after two rounds of incubation in PPNH4 medium supplemented with 25 μg ml−1 G418 (Sigma-Aldrich Chemie, http://www.sigmaaldrich.com/).
The first screening of putative disrupted clones was carried out by three independent PCR reactions on genomic DNA purified from transformant plants, one spanning the chromosomal targeted region using the Knockout 1 primers, and two amplifying part of the marker cassette and part of the chromosomal regions 5′ and 3′ outside the knockout construct, Knockout 2 and 3 primers (Table S4). In clones in which these amplifications produced the expected fragments, the fragments were sequenced to check that integration occurred as expected (Figure 4).
Southern blot analyses of targeted knockout Physcomitrella lines were carried out according to standard procedures. Genomic DNA (20 μg) was digested with NcoI, PstI, or SphI enzymes and hybridized with a probe that includes the neomycin resistance cassette, as shown in Figure 4. The DNA probe was synthesized and digoxigenin-labeled by PCR. Hybridization and detection was carried out according to the supplier’s manual (Roche Applied Science, http://www.roche.com/).
Cation uptake experiments in Physcomitrella
Uptake rates at micromolar concentrations of the tested cation, always with K+-starved cells, were calculated from cation depletion experiments. For these experiments, Physcomitrella samples were transferred to KFM containing the required concentration of the tested cation, K+, Na+, Rb+, or Cs+, and then measuring the cation concentration in the bathing medium at intervals by atomic emission spectrophotometry. Sampling did not significantly modify the volume of the testing medium. Fitting of the data points to a concentration-dependent influx that follows a Michaelis–Menten equation was performed as described previously (Bañuelos et al., 2002). This fitting could be applied to K+ and Rb+ depletion curves, but only its application to Rb+ uptake is formally correct. Determinations of uptake rate at millimolar concentrations were carried out by transferring Physcomitrella plants to KFM containing the required concentration of the tested cation, Rb+ or Cs+. At intervals, Physcomitrella samples were transferred to 0.8-μm pore Millipore (http://www.millipore.com/) membrane filters, washed with 10 mm KCl, 1 mm MgCl2 for 1 min, and dried overnight. After weighing the plant samples, the internal cations were acid extracted overnight in a 0.1 m ClH solution and the cation concentrations in the supernatant were determined by atomic emission spectrophotometry. For concentration-dependent kinetic analyses the initial rate of uptake was taken as the influx of the cation at the tested concentration.
Cation uptake experiments in yeast
The K+-starved yeast cells were prepared from cultures grown overnight in AP medium supplemented with 50 mm K+ that were then transferred for 4 h to K+-free AP medium. High-affinity K+ or Rb+ uptake experiments were carried out in 10 mm Ca2+-2-(N-morpholine)-ethanesulfonic acid (MES), pH 6.0 buffer supplemented with 2% glucose following the depletion of the cation in the assay medium measured by atomic emission spectrophotometry as described previously (Bañuelos et al., 2002). When influx was slow in the micromolar range of concentrations, Rb+ or Cs+ influxes were determined as the initial rates of uptake in cells that were collected at intervals in 0.8-μm pore Millipore membrane filters and rapidly washed with 20 mm MgCl2 solution. Cells were extracted overnight with 0.1 m HCl, and the cation concentrations in the supernatant were measured by atomic emission spectrophotometry. Similar experiments with K+ were performed in K+-starved Rb+ yeast cells that were obtained by growing the yeast cells overnight in AP medium supplemented with 25 mm Rb+ and 5 mm K+ (Rb+-yeast cells) and then growing them for an additional period of 4 h in K+- and Rb+-free AP medium. These cells were used as described above for Rb+ and Cs+ uptake experiments.
We would like to thank Marcel Velduizen and Elena Pérez-Carbajo for their skilful technical assistance. Financial support for this work was provided by Ministerio de Educación y Ciencia and FEDER program of the EU (no. AGL2004-05153), which also funded a Ramón y Cajal contract to BB. Additional financial support was provided by DGUI-UPM Research Group Program (no. 05/10719).