Results

pldα1 is hypersensitive to high levels of Mg2+

We grew pldα1-1 (Bargmann et al. , 2009) seedlings in varying concentrations of diverse nutrients including Mg2+. pldα1-1 was hypersensitive to magnesium, with reduced primary root length, fresh weight (Fig. 1a,b,c), and number and length of lateral roots (Fig. S1.). A significant decrease inpldα1-1 primary root length was observed after application of 1 mM MgCl2; at this conditions, pldα1-1 primary roots were found to be 6% shorter than wt (Fig. 1a,b). Concentrations of 5 mM MgCl2 and higher had a severe effect on the growth of wt plants; however, in all studied concentrations,pldα1-1 was more sensitive. The greatest difference in primary root length between wt and pldα1-1 was observed in plants treated with 15 mM MgCl2, where pldα1-1 roots were 40% shorter (Fig. 1b). The greatest difference in fresh weight between wt and pldα1-1 was observed with 10 mM MgCl2, wherepldα1-1 was half the weight of wt (Fig. 1c).
To determine whether the MgCl2 hypersensitivity observed in seedlings persists in mature plants, wt and pldα1-1 were grown hydroponically. MgSO4 (at 10 mM) was added to the hydroponic solution and the plants were maintained for 10 days. Reduced growth in pldα1-1 compared to wt plants was markedly visible (Fig. 1d). However, because magnesium sulfate was used instead of magnesium chloride, it was necessary to rule out the possible effects of other ions. Plants were treated with 10 mM MgCl2, MgSO4, or Mg(NO3)2, and growth was assessed. Although there were visible variations in the effect of individual anions, the significant difference between wt and pldα1-1 was clearly detectable in all cases (Fig. S2). Therefore, it is possible to rule out that the anion is responsible for the observed pldα1-1 phenotype.
To ensure that the observed Mg2+ hypersensitivity was due to an insertion in PLDα1 and no other genes, we used three additional T-DNA insertion lines for PLDα1 , includingpldα1-2 (described by Bargmann et al. (2009)), pldα1-3(SALK), and pldα1-4 (GABI-KAT). We also made complementation lines by transforming pldα1-1 plants with PLDα1 driven by its native promoter. The levels of PLDα1 (PLDα1-Com) protein in seedling extracts was verified using anti-PLDα1/2 antibody. PLDα1 was not detected in any of the pldα1 lines (Fig. 2a). Levels of PLDα1 in the complementation lines were lower than in wt; therefore, the two PLDα1-Com lines with the highest PLDα1 protein levels were used for subsequent analyses. pldα1-2, pldα1-3 , pldα1-4, and complementation lines were phenotyped for Mg2+sensitivity. Primary root length was 23, 25, and 26% lower inpldα1-2 , pldα1-3, and pldα1-4 , respectively. Fresh weight was 52, 53, and 54% lower in pldα1-2 , pldα1-3, andpldα1-4 , respectively, compared to wt when treated with high Mg2+ (Fig. 2 c, d, e). Overall, all PLDα1 mutants were similarly Mg2+-sentitive to pldα1-1 . In contrast, primary root length and fresh weight in linespldα1-1 -Com1 and pldα1-1 -Com2 were similar to wt when treated with high Mg2+ (Fig. S3).
Arabidopsis pldα1 plants are more sensitive to high-Mg2+ conditions than wt; thus, PLDα1 appears to be involved in response to high-Mg conditions. These results uncovered a novel physiological role of PLDα1 in the context of Mg2+-homeostasis.

PLDα1 activity increases after treatment with Mg2+

Next, we investigated whether high levels of Mg2+could trigger changes in PLDα1 activity. Arabidopsis has 12 genes encoding PLDs, which differ biochemically and require different in vitro conditions for activation (Hong et al. , 2016). PLDs cleave ordinary phospholipids such as phosphatidylcholine, releasing PA and free head group, e.g. choline. PA is also the product of diacylglycerol kinase activity, as well as the substrate for PA phosphatase, among other enzymes (Ruelland et al. , 2015). Hence, PA levels do not necessarily correlate with PLD activity. A unique feature of PLDs is their so-called transphosphatidylation activity, where, in the presence of primary alcohols such as n -butanol, PLD transfers the phosphatidyl group from its substrate to n -butanol, releasing phosphatidylbutanol (PBut). PBut-formation therefore directly corresponds to PLD activity (deVrije & Munnik, 1997).
PLDα1 is known to be both membrane-associated and cytosolic (Fan, Zheng, Cui & Wang, 1999). Predominant cytosolic localization was reported by Novák et al. (2018) in Arabidopsis expressing PLDα1-YFP, therefore we determined PLDα activity in the soluble fraction. Plants were treated with 10 or 40 mM MgSO4, after which root samples were taken at 10, 30, and 180 min. The soluble fraction was prepared, and the activity of PLDα was determined using fluorescently labeled phosphatidylcholine as a substrate under conditions optimal for PLDα (Hong, Zheng & Wang, 2008). Lipids, including PBut, were extracted and separated using high-performance thin-layer chromatography (HP-TLC), and the amount of fluorescently labeled PBut was quantified (Fig. 3). PLDα activity was also measured in samples prepared from pldα1-1plants, where PLDα activity was found to be negligible (Fig. 3a). Hence, we concluded that the quantity of released PBut corresponds to PLDα1 activity.
PLDα1 activity in the soluble fraction increased after MgSO4 treatment (Fig. 3b). The increase was concentration-dependent, as higher concentrations of MgSO4 consistently led to an increase in PLDα1 activity (Fig. 3b,c). Interestingly, the increase in PLDa1 activity was transient, reaching 2.5-fold after 30-minutes of treatment with 10 mM MgSO4 (Fig. 3d).
An increase in PLDα1 activity could be due to higher rates of transcription of PLDα1 , activation of PLDα1, or a combination of the two. Thus, we measured transcriptional levels of PLDα1 in control and high-Mg2+ treated (10 mM MgSO4, for 24h) plants using quantitative RT-PCR. We found no increase in PLDα1 transcript levels following Mg2+ treatment (Fig. 3e).
These results demonstrate that PLDα1 is activated by Mg2+ shortly after treatment, though not at the transcriptional level.

PLDα1 activity contributes to high- Mg2+tolerance

To confirm that the activity of PLDα1 is essential for high-magnesium tolerance in wt plants, we introduced an inactive mutant forPLDα1 into pldα1-1 plants (p PLDα1::PLDα1-Mut/pldα1 ). Members of the PLD superfamily retain the highly conserved HKD motif, which is encoded twice in higher-plant PLDs (Wang et al. , 2014). Point mutations in HKD motifs result in the complete loss of PLD activity in Brassica oleracea (Lerchner, Mansfeld, Kuppe & Ulbrich-Hofmann, 2006), as well as in humans and mice (Sung et al. , 1997).
Transgenic pldα1-1 plants expressingpPLDα1 ::PLDα1 K334R;K663R (lines Mut1 and Mut2) at levels consistent with wt (Fig. 2a) showed similar sensitivity to MgCl2 as pldα1-1 , for both primary root length (Fig. 4a) and fresh weight (Fig. 4b).
These results demonstrate that Arabidopsis PLDα1 activity is critical in mediation of the response to high-magnesium conditions.

pldα1 accumulates less Mg2+ and K+ under high-Mg2+conditions

To elucidate the possible mechanism responsible for the higher susceptibility of pldα1 , we measured Mg2+-content in wt and mutant plants under control and high-Mg2+ conditions. After high-Mg2+ (10 mM) treatment, seedling Mg2+ levels were elevated by about five times in wt and pldα1-1. Nevertheless, pldα1 showed significantly lower Mg2+-content than wt (Fig. 5a).
There is known to be an antagonistic relationship between the uptake of Mg2+ and Ca2+ (Guo, Babourina, Christopher, Borsic & Rengel, 2010, Yamanaka et al. , 2010). Moreover, increased levels of Mg2+ application in rice results in lower uptake of calcium and potassium (Fageria, 2001), and transcription of the potassium transporter HAK5 increases following treatment with Mg2+ in Arabidopsis (Tang & Luan, 2017, Visscher et al. , 2010). Therefore, we measured Ca2+ and K+ content in the wt andpldα1-1 plants to investigate these relationships.
In agreement with Mg-Ca antagonism, seedling Ca2+content was lower when 10 mM MgCl2 was added to the agar medium. However, we did not observe any difference between wt andpldα1-1 (Fig. 5b). Interestingly, K+ levels in wt and pldα1-1 were lower in Mg2+-treated plants, with pldα1-1 plants retaining even less K+ than wt (Fig. 5c).
These results demonstrate that PLDα1 is involved in the regulation of Mg2+- and K+- content in Arabidopsis seedlings grown in high-Mg2+ conditions.

Addition of Ca2+ and K+ alleviates Mg2+-hypersensitivity in pldα1 plants

Aware that there is an antagonistic relationship between some of the essential nutrients, we investigated whether excess Ca2+ or K+ could affect pldα1hypersensitivity to Mg2+. Application of both Ca2+ and K+ ameliorate growth inpldα1-1 in high-Mg2+ (Fig. 6a). Addition of Ca2+ completely restored the growth of pldα1-1to wt levels, in both root length and fresh weight. Roots from bothpldα1-1 and wt were smaller (Fig. 6b), while fresh weight for both was similar, compared to control conditions (Fig. 6c). Application of K+ lowered the root-length difference between wt and pldα1 . With Mg2+, root length ofpldα1-1 was 82.5% that of wt. However, with K+, root length of pldα1-1 increased to 94.9% that of wt. Fresh weight under high-Mg2+ inpldα1-1 was brought up to wt levels with K+(Fig. 6c), though root length and fresh weight in both pldα1-1and wt were lower compared to the control conditions (Fig. 6).
Addition of Ca2+ increased plant growth in both wt andpldα1-1 plants (Fig. 6). However, no difference in Ca2+ content between wt and pldα1-1 was detected (Fig 5b). These results, along with what is known about the antagonistic relationship between Ca and Mg, suggest that Ca2+ deficiency is not the underlying factor behind growth defects in pldα1 grown under high-Mg2+, but that high Mg2+ or low K+ content is responsible. Under high-Mg2+ conditions,pldα1‑1 retained less Mg2+ than wt, though the lower Mg2+-content was more toxic to pldα1-1than the higher-Mg2+ content in wt. This phenomenon may be explained by impairment of Mg2+ sequestration in pldα1-1 , which would result in higher cytosolic concentrations of Mg2+. Additionally, lower K+content may contribute to impaired growth in pldα1-1 , or a combination of the two mechanisms.

K+-related genes CIPK9 and HAK5 are not transcriptionally upregulated in pldα1

Ten members of the MGT family were identified in the Arabidopsis genome, (Li, Tutone, Drummond, Gardner & Luan, 2001). Therefore, we investigated transcriptional response of MGT family genes, of which Arabidopsis has 10 (Li et al. , 2001), to high-magnesium stress. Transcript levels were determined using quantitative RT-PCR in roots and leaves of wt and pldα1-1 plants, separately (Fig. 7a). Transcript levels of MGT1 in roots andMGT7 in leaves was slightly elevated in Mg2+-treated plants, though there was no difference between wt and pldα1-1 (Fig. 7a). Transcript levels for two genes are not shown, as MGT5 was under the detection limit andMGT8 was found to be a pseudogene (Zhang et al. , 2019).
In Arabidopsis, CAX1 is known to be involved in high-Mg2+ resistance (Bradshaw, 2005). Moreover, transcription of CAX1 is downregulated under high-Mg2+ conditions (Visscher et al. , 2010). Hence, we determined transcriptional levels of CAX1 in control and Mg2+-treated wt and pldα1-1 plants.CAX1 transcription was decreased in the roots and leaves of Mg2+-treated plants; however, as in case of MGT genes, there was no difference between wt and pldα1-1 (Fig. 7b).
Next, we looked at transcription of CIPK9 and HAK5 , both of which are known to be involved in potassium homeostasis under low-potassium conditions (Coskun, Britto & Kronzucker, 2014), and are reportedly upregulated in high-magnesium conditions (Tang et al. , 2015, Visscher et al. , 2010). Furthermore, Arabidopsis CIPK9 has also been shown to participate in high-Mg2+ response (Tang et al. , 2015). In agreement with Visscher et al. (2010), we observed high upregulation of CIPK9 and HAK5 in wt roots after Mg2+ treatment. However, the increase inCIPK9 and HAK5 transcript levels was almost completely abolished in pldα1-1 roots (Fig. 7b). In Mg2+-treated pldα1-1 leaves, CIPK9transcript levels were slightly increased (~doubled), and HAK5 was not detected; however, there was no difference between wt and pldα1-1 (Fig. 7b).
These results indicate that PLDα1 is essential in a signaling mechanism which leads to an increased expression of HAK5 and CIPK9in roots upon high-Mg2+ treatment.

The hak5, akt1 double mutant is hypersensitive to high-magnesium

Based on our previous observations, we speculated that proper regulation of potassium homeostasis is essential in high Mg2+-conditions. We examined whether hak5plants are hypersensitive to high-magnesium, and found no difference between hak5 and wt (Fig. 8). Next, we tested the sensitivity of a hak5 , akt1 double mutant to high-Mg2+and found that it was significantly more sensitive than wt. Root length in hak5 , akt1 plants was 9.5% (Fig. 8b), and fresh weight 14.5%, less than in wt (Fig. 8c). Under control conditions,hak5, akt1 growth did not differ from wt (Fig. 8).
However, there was still a significant difference between pldα1and hak5, akt1 sensitivity to high-Mg2+.pldα1-1 roots were 24%, and fresh weight 47%, less than wt; thus, hak5, akt1 is less sensitive to high-Mg2+ compared to plda1-1(Fig. 8b,c).
These results revealed that plants impaired in K+uptake are also compromised in their tolerance to high levels of Mg2+. Therefore, appropriate regulation of potassium homeostasis is key to that of magnesium.
We conclude that in Arabidopsis, K+-homeostasis is involved in response to high-Mg2+, and that this mechanism is at least partially mediated by PLDα1.