The Arabidopsis leaky allele stop1³³ and the null allele stop1¹²⁷ are both in a Col^er105 genetic background and come from an ethyl methanesulfonate mutagenesis screen (Balzergue et al., 2017). The pALMT1:GUS reporter (GUS, uidA gene encoding β-glucuronidase) (Godon et al., 2019) was introgressed by crossing in the stop1³³ and stop1¹²⁷ mutant backgrounds (Mercier et al., 2021). The pSTOP1:GFP-STOP1_#B10 construct used was described in Balzergue et al. (2017).

Plants were grown in agar in long (16 h photoperiod) or short days (8 h photoperiod), at 22°C (day)/21°C (night).

The seeds were surface-sterilized for 2 min in a solution containing 70% ethanol and 0.05% sodium dodecyl sulfate, and washed twice with 96% ethanol.

The nutrient solution contained 0.47 mM MgSO₄, 2.1 mM NH₄NO₃, 1.89 mM KNO₃, 0.67 mM CaCl₂, 0.5 μM KI, 0.79 mM H₃BO₃, 10 μM MnSO₄, 5 μM ZnSO₄, 1 μM Na₂MoO₄, 0.1 μM CuSO₄, 0.1 μM CoCl₂, 5 g L^–1 sucrose. The agar (8 g.L^–1) for plates was from Sigma-Aldrich (A7921 Lot BCBZ7284, see Mercier et al., 2021 for elemental composition). The agar media were supplemented with 10 μM and 500 μM KH₂PO₄ for low-phosphate and rich-phosphate conditions, respectively. The media were buffered with 3.4 mM 2-(N-morpholino) ethanesulfonic acid (MES) for pH range from 5 to 6, and with piperazine-N,N′-bis 2-ethanesulfonic acid (PIPES) 3.4 mM from pH 6 to 7. The pre-culture was performed in rich-phosphate conditions (P₅₀₀) and 10 μM FeCl₂ to avoid additional chlorotic stress. After transfer on exposure media, the assessment of Fe and Al effects was performed in low-phosphate condition (P₁₀). Fe (FeCl₂) and Al (AlCl₃) were added independently to the growth media.

Hydroponic culture was performed in MS/10 nutritive solution, 0.5% sucrose with 3.4 mM buffer, Homo-Pipes (pH 4.4) or MES (pH5–6.2). Plants were grown in long days (16 h photoperiod) for 3 days and transferred in a fresh medium for 12 h for stress assessment.

The speciation of Al was simulated with the JChess 2.0 software (van der Lee and De Windt, 1999), with the database released in 2004 in the framework of the Common Thermodynamic Database Project (van der Lee and Lomenech, 2004) and updated afterward (Denison and Garnier-LaPlace, 2005; Lofts et al., 2015; Vercouter et al., 2015; Février et al., 2021). In this database, thermodynamic complexation constants for Al were imported from the thermo.com database developed by the Lawrence Livermore National Laboratory. Equilibrium reactions with their associated thermodynamic constants were presented in Supplementary Table 1. They include the formation of mononuclear hydroxyl-Al and polynuclear hydroxyl-Al species, sulfate and phosphate Al-complexes, as well as solid phases such as Gibbsite, a poorly soluble Al hydroxide [Al(OH)₃], and Alunite, a hydroxylated Al potassium sulfate mineral [KAl₃(SO₄)₂(OH)₆].

Simulations of Al speciation were performed for each experiment. The entire chemical composition of the culture medium was considered as input in the modeling, with the exception of Homo-Pipes, MES, and agar, which are supposed to be inert toward Al speciation (Parent and Campbell, 1994).

Plants were grown for 5 days on P₅₀₀Fe₁₀ without AlCl₃ added to the medium then transferred for 4 h on the described media in low phosphate condition (P₁₀) before observation through confocal microscopy. The GFP-fluorescence was performed as in Godon et al. (2019). Images were collected on a Zeiss LSM780 confocal microscope (Carl Zeiss)¹ using an ×40 water objective. The GFP was excited with an Ar ion laser (488 nm). Emitted light was collected from 493 to 538 nm for GFP, using the MBS 488 filter. All nuclei were imaged using the same conditions of gain, offset, and resolution. The quantification of GFP fluorescence was carried out as follows: a Z-stack of six images (separated by a distance of 2.0 μm) imaged representative nuclei on the surface of the root. Images were acquired in 12 bits using Zen black software (SP2 v.11.0, 2012, Carl Zeiss), then converted to a maximum projection image. The average nuclear fluorescence intensities were quantified using Image J by applying a mask for nuclei detection. These nuclei formed regions of interest (ROI) in which the GFP mean intensity was quantified. The GFP fluorescence intensity of minimum 15 nuclei per plant was averaged. Six plants per condition were assessed. The average of each plant was used for statistical analysis using the Kruskal-Wallis test.

Plants were grown for 10 days before transfer for 24 h. The roots were collected to perform RT-PCR. Total RNA was extracted from whole roots using the RNeasy Plant Mini Kit (Qiagen, France) and treated with the RNase-free DNase Set (Qiagen, France) according to the manufacturer’s instructions. Reverse transcription was performed on 400 ng total RNA using the SuperScript VILO DNA Synthesis Kit (Invitrogen). Real time quantitative reverse transcription PCR (qRT–PCR) was performed on a 480 LightCycler thermocycler (Roche) using the manufacturer’s instructions with Light cycler 480 sybr green I master (Roche) and with primers listed in Supplementary Table 2. We used tubulin (AT5G62690) as a reference gene for normalization.

The GUS staining of Arabidopsis seedlings was conducted as previously described (Balzergue et al., 2017) for different time specified in each result.

As described in the introduction, it is assumed that at pH > 5.5, Al has low toxic effects on plants and on the STOP1-ALMT1 pathway. In order to test whether the expression of ALMT1 is induced by Al at pH > 5.5, 4 days-old WT seedlings carrying the pALMT1:GUS reporter were transferred for 3 days on growth media supplemented with 20 μM AlCl₃ and buffered at pH from 5.7 to 7. As a negative control, some seedlings were transferred on the same media but not supplemented with AlCl₃. As shown on Figure 1, on control conditions without AlCl₃, no or low GUS staining is detected on the root tip. On media supplemented with AlCl₃, the GUS staining is visible in the root tip of seedlings transferred up to a pH 6.5. No staining is observed at pH 7.0.

In the literature, it is widely described that ALMT1-induced expression results directly from the activity of the STOP1 transcription factor. Here, we assessed the expression of pALMT1:GUS marker in WT seedlings grown at pH 5.7, supplemented with concentrations of AlCl₃ ranging from 0 to 20 μM (Figure 2A). The leaky stop1³³ allele displays a decreased ALMT1 expression resulting from a partial STOP1 activity while the stop1¹²⁷ null allele does not show GUS staining as a consequence of the absence of STOP1 (Balzergue et al., 2017; Mercier et al., 2021). In WT, we observed that the intensity of the staining increases with the concentration of Al. In these conditions, ALMT1 is induced at Al concentration as low as 2 μM AlCl₃, showing that this reporter is highly sensitive. Furthermore, this activation of ALMT1 depends on STOP1 because in a stop1¹²⁷ null mutant no GUS staining is detected independently from the Al concentration added (Supplementary Figure 1A). To check the absence of effect from the agar-solidified medium, a similar experiment was performed in a hydroponic solution. Supplementary Figure 1B shows the induction of ALMT1 when 20 μM AlCl₃ was added in the medium at pH 5.7. The pALMT1:GUS is a sensitive visual reporter, however, the ALMT1 expression was quantified using RT-PCR in the whole root. The relative expression of ALMT1 was measured in plants after 24 h transfer from phosphate-rich media to low phosphate media containing 0, 5, and 20 μM AlCl₃. Figure 2B shows that ALMT1 is induced when AlCl₃ is added in the media. Furthermore, the rate of induction correlates with Al concentration, as with the pALMT1:GUS reporter.

It was previously shown that Al³⁺ increases the stability of the STOP1 protein in root nuclei (Godon et al., 2019). To determine whether the increased ALMT1 induction is correlated with an increased STOP1 abundance, we used the pSTOP1:GFP-STOP1 reporter. Five days old seedlings that grew in phosphate-rich medium were transferred for 4 h on the described media and then observed by confocal microscopy. Figure 2C shows that STOP1 accumulated in root nuclei 4 h after transfer of the seedlings on a medium supplemented with AlCl₃. The fluorescence in the nuclei was quantified (Figure 2D). While no statistical differences were observed between 0 and 5 μM AlCl₃ nor between 5 μM and 20 μM AlCl₃, the STOP1 abundance was significantly different comparing the negative control and 20 μM AlCl₃ (p < 0.01, Kruskal-Walis and Dunn’s multiple comparison test). These results show a correlation between Al-induced STOP1 accumulation in root nuclei and the ALMT1 expression assessed by the GUS stainings (Figure 2A) and the RT-PCR (Figure 2B). Together, these experiments show that, at pH 5.7, low concentrations of AlCl₃ are able to activate the STOP1-ALMT1 pathway in root seedlings.

The precedeing experiments show that Arabidopsis seedlings detect Al in conditions where it is not toxic. To assess which concentrations of Al are toxic in our growth conditions without iron (P₁₀Fe₀), we measured the length of the primary root (Figure 2E). The stop1¹²⁷ null mutant and the stop1³³ leaky mutant were included as Al-sensitive controls (Balzergue et al., 2017; Mercier et al., 2021). The dose-response curve shows that at 5, 20, and 50 μM AlCl₃, the three lines were not different from their untreated controls. By contrast, at a higher concentration, 200 μM AlCl₃, the root growth of stop1¹²⁷ was blocked; the stop1³³ was slightly less inhibited than stop1¹²⁷ while the WT was about 80% the size of the control at 0 μM AlCl₃. This experiment shows that at pH 5.7, Al is not toxic at concentrations ≤ 50 μM AlCl₃, even in the hypersensitive stop1 mutants. Combined with the previous results, these experiments demonstrate that a growth medium with agar, at pH 5.7 with as low as 2 μM AlCl₃, is a condition uncoupling Al-sensing from Al-toxicity.

In order to assess which Al species are present in the nutrient solution of our growth media, a simulation was performed with the JChess 2.0 software (see section “Materials and Methods”). To note, these simulations did not take account of the agar and the buffers. According to preliminary simulations, the culture medium was oversaturated with gibbsite and with alunite also at low pH (between 4.4 and 4.5) (not shown). However, although gibbsite is a poorly soluble mineral, its formation in the culture medium has been shown to be kinetically limited (Parent, 1991) and might be also impacted by the presence of agar. Therefore, precipitation of Al was discarded in the speciation simulations. In the first simulation with 20 μM AlCl₃ at different pH, the predominant form of Al was Al³⁺ at low pH (4.4 < pH < 5.1). As soon as the pH increased above 5, the polynuclear specie Al₁₃O₄(OH)₂₄⁷⁺ (referred to as Al₁₃) increased up to 70% of the total Al. At pH 6.0 and above, the mononuclear hydroxy species—neutral Al(OH)_3(aq) and aluminate (Al(OH)₄^–) became dominant (Supplementary Figure 2A).

In the second simulation, we used different concentrations of Al at pH 5.7 (Supplementary Figure 2B). In contrast to the first simulation, the speciation of Al was affected by the concentration of Al. At low Al concentration, mononuclear Al hydroxyl species were the main Al species in the culture medium, with about 48% of neutral Al(OH)_3(aq) and 19% of Al(OH)₂⁺. However, at a concentration above 10 μM of AlCl₃, the polynuclear species Al₁₃ prevailed over other Al species. At 40 μM AlCl₃, Al₁₃ represented more than 80% of Al species. To note, except Al₁₃ that exceeds 10 μM the values of each other species remained below 5 μM independently from the AlCl₃ concentrations assessed (up to 200 μM) (Supplementary Figure 2B inset). Thus, no species prevailed the others. According to these simulations, our medium supplemented with 20 μM AlCl₃, at pH 5.7, should be toxic, at least for the Al-sensitive stop1 mutants due to the extreme toxicity of polynuclear Al₁₃ species (Kinraide, 1991). Since we do not observe toxic symptoms (Figure 2E), we infer that the agar might prevent toxicity of Al₁₃, and possibly of other species.

As described in the literature, STOP1 was firstly identified as required for tolerance to acidity (Iuchi et al., 2007), then toxic Al (Sawaki et al., 2009; Kobayashi et al., 2013). However, as previously mentioned in the introduction, the speciation of Al makes it mostly studied at low pH due to its increased toxicity. As consequence, it is difficult to differentiate responses to Al from responses to low pH in these conditions. Here, our objective was to identify conditions to dissociate both mechanisms. WT plants carrying the pALMT1:GUS reporter were grown for 4 days on a phosphate-rich medium at the indicated pH before GUS staining (Supplementary Figure 3). This result shows that ALMT1 expression was induced when the pH is below 5.5 in absence of Al. To note, similar results when plants were grown on a low-phosphate medium (Data not shown). Three independent experiments were performed on agar-grown plants. Although this agar was selected for its low amount of Al, the decreasing pH could increase the Al³⁺ fraction of the remaining Al in the agar, and as the pALMT1:GUS was shown to be highly sensitive to Al, a similar experiment was performed in hydroponics. In Figure 3, WT plants carrying the pALMT1:GUS reporter were grown 3 days in hydroponic medium (see section “Materials and Methods”) in phosphate-rich conditions not supplemented with Fe or Al, at the indicated pH, and then transferred in a fresh phosphate-rich medium for one more day before staining for GUS activity. The results show that, in absence of Al, ALMT1 expression is induced at low pH. Therefore, these growth conditions uncouple the low pH-dependent ALMT1 expression form the Al- dependent.

We previously demonstrated that both Fe and Al trigger STOP1 signaling toward ALMT1 expression (Godon et al., 2019) and that, under low-phosphate conditions, Fe inhibits root growth (Svistoonoff et al., 2007; Balzergue et al., 2017). Here, we tested the hypothesis that Fe and Al interfere with the expression of ALMT1 and the root growth under low-phosphate conditions. Dose-response curve of Fe was performed on WT, stop1^33, and stop1¹²⁷ seedlings (see section “Materials and Methods”). Figure 4A shows that the WT and the two mutants display similar responses, except at 10 μM FeCl₂, where the WT is inhibited. The stop1 mutants are insensitive to 10 μM FeCl₂ whereas root growth of WT plants was significantly reduced. Surprisingly, the WT growth is not inhibited by 20–50 μM of FeCl₂ and displays the same primary root length as the mutants. At concentrations above 50 μM, the root growth is inhibited similarly for the three lines. This experiment shows that the growth inhibition of the WT is restricted to a narrow range of Fe concentration (i.e., around 10 μM). Figure 4B shows the corresponding ALMT1 expression staining to the Fe dose-response curve. In the WT we observe a slight induction of the ALMT1 expression at 10 μM FeCl₂. The ALMT1 expression is increased along with the Fe concentration from 50 μM. In the stop1³³ leaky mutant, the ALMT1 expression is detected only at 200 μM Fe.

We then assessed the effect of Al on the Fe-dependent inhibition of WT root growth. The seedlings were grown 4 days on a P₅₀₀ medium and transferred 3 days on a P₁₀ medium supplemented with 0–200 μM FeCl₂, and with 5 μM AlCl₃ (Figure 4C). We observed that Al enhances the Fe-dependent inhibition of WT root growth under low-phosphate and at 10 μM the WT root is inhibited. Intriguingly, at intermediate concentrations of FeCl₂ (20–100 μM) the root growth is partially restored (compared to that at 10 μM). The overall pattern of response is similar to the one observed in absence of Al (Figure 4A), but Al strengthens the inhibiting effect of the Fe on the primary root growth. At 200 μM FeCl₂, the root growth is blocked. In these growth conditions, the expression of ALMT1 in WT root tips is induced at each Fe concentration tested (Figure 4D). In the stop1³³ mutant, ALMT1 is slightly induced in the root cap and strongly expressed in the whole root at 200 μM Fe (Figure 4D).

This experiment shows that Al enhances both the Fe-dependent root growth inhibition and ALMT1 expression. In addition, comparing the dose-response curves obtained with or without Al unveiled a complex effect of intermediate concentrations of Fe on root growth. Altogether the GUS staining in Figures 4B,D show that Fe and Al independently activate ALMT1 expression in WT.

To better characterize the effect of Al on the Fe-dependent responses, we performed an Al dose-response experiment in a medium supplemented, or not, with 10 μM FeCl₂ (Figures 4E,F). In Figure 4E (top row) in absence of Al, the WT displays a light ALMT1 expression while no ALMT1 induction is observed in the mutants. When 20 μM or higher concentrations of AlCl₃ are added to the medium without supplemented Fe, ALMT1 expression is largely induced in the WT. However, in the stop1³³ leaky mutant, we observe low staining that does not increase with AlCl₃ concentration. By contrast, when 10 μM FeCl₂ are added in the Al-containing media, a strong ALMT1 induction is observed in the stop1³³ leaky allele (Figure 4E, bottom row). Fe and Al independently activate ALMT1 expression in WT but not, or low ALMT1 induction in stop1³³. However, Fe and Al trigger higher ALMT1 expression in the leaky allele. The observed staining of WT and stop1³³ on Al concentration range in presence of 10μM FeCl₂ is in correlation with the respective primary root responses (Figure 4F). The WT shows a strong primary root inhibition while the stop1³³ mutant displays an intermediate root arrest compared to the null mutant stop1¹²⁷ (Figure 4F). In these conditions, the root growth of the ALMT1 expressing lines (WT and stop1³³) is more sensitive than the stop1¹²⁷ null allele. Comparison with Figure 2E indicates that WT inhibition at ≤ 50 μM Al depends on the added Fe. Of note, as in Figure 4C, a release of the WT primary root inhibition is also observed at 100 μM AlCl₃. These results suggest that Fe and Al on ALMT1 induction, but also root growth, are distinct signal response mechanisms as they independently induce STOP1. However, Fe and Al response mechanisms interact when both ions are combined. Based on the dose-response curve, Al strengthens the inhibitory effect of Fe on growth of primary roots. The cumulated stress up-regulate ALMT1 in the WT but also stop1³³ while both stresses independently applied cannot.

To confirm the effect of Al and Fe on the STOP1 pathway, RT-PCR was performed for two direct targets genes of STOP1: ALMT1 and CIPK23 (Figure 5A). WT seedlings were grown for 10 days on a rich-phosphate medium and transferred 24 h on low-phosphate media supplemented with Fe or Al conditions described in the figure. The roots were harvested for RNA extraction. Concerning CIPK23 expression, whatever the conditions tested, the CIPK23 expression remained stable. Instead, the ALMT1 relative expression tends to show an upregulation in presence of Al alone and with Fe. Surprisingly, a low tendency was observed in the presence of Fe alone. No statistical differences could be observed using the multiple comparison Kruskal-Wallis test applied with Dunn correction, however, these tendencies were observed within 4 independent experiments (data not shown). In Supplementary Figure 4, we observed the ALMT1 pattern induced by the Al, Fe independently and combined stresses by transferring 4 days old WT plants on the indicated media for 24 h. The Al conditions show an ALMT1 induction in the whole root, that is coherent with the RT-PCR results in Figure 5A. The combined Fe, Al condition shows a localized upregulation in the root tip that is also visible through the RT-PCR. To note, this stress induces a strong root growth arrest characterized by a bulge of the root tip in accordance with Figure 4C. However, without Al, the Fe₁₀ condition induces ALMT1 in the root tip only display a low but not significant induction by RT-PCR (Figure 5A). One explanation could be that the expression of ALMT1 is more localized at the root tip than in the rest of the root, and taking the whole root for RT-PCR dilute its signal.

Thus, to assess the effect of these stresses on the STOP1 abundance in the root tip specifically, we used plants carrying the construct pSTOP1:GFP-STOP1 to measure the accumulation of STOP1 at the nuclei through GFP fluorescence. In Figure 5B, 5 days old plants were transferred 4 h on the indicated media before confocal observation. As positive control, the addition of 5 μM AlCl₃ shows an accumulation of GFP at the nuclei. Increasing doses of FeCl₂ from 10 to 80 μM were added to the media tends to show a correlated GFP accumulation at the nuclei. In Fe₅₀Al₅ and Fe₈₀Al₅ conditions, the GFP intensity is higher than in their respective control, Fe₅₀, Fe₈₀, and Al₅ and tend to show a synergistic effect between Fe and Al.

Altogether these results show that Fe alone triggers a STOP1 dependent primary root growth arrest at a specific concentration Fe₁₀ in the WT. In this condition, ALMT1 is induced in the root tip, however, no STOP1 accumulation was detected in this condition. Instead, the STOP accumulation was observed in correlation with the increasing dose of Fe in the media. Furthermore, we observed that the addition of low Al concentration (Al₅) in the media strengthens the Fe-dependent root growth arrest and the ALMT1 induction in the root tip. Furthermore, the combined stress of Fe and Al is the only condition able to trigger an upregulation of ALMT1 in stop1³³. Finally, Al interacts with Fe signal pathway on STOP1 by displaying a synergistic effect on STOP1 accumulation when high doses of Fe are added to the media.