Long‐distance mobile mRNA CAX3 modulates iron uptake and zinc compartmentalization

Abstract

Iron deficiency in plants can lead to excessive absorption of zinc; however, important details of this mechanism have yet to be elucidated. Here, we report that MdCAX3 mRNA is transported from the leaf to the root, and that MdCAX3 is then activated by MdCXIP1. Suppression of MdCAX3 expression leads to an increase in the root apoplastic pH, which is associated with the iron deficiency response. Notably, overexpression of MdCAX3 does not affect the apoplastic pH in a MdCXIP1 loss‐of‐function Malus baccata (Mb) mutant that has a deletion in the MdCXIP1 promoter. This deletion in Mb weakens MdCXIP1 expression. Co‐expression of MdCAX3 and MdCXIP1 in Mb causes a decrease in the root apoplastic pH. Furthermore, suppressing MdCAX3 in Malus significantly reduces zinc vacuole compartmentalization. We also show that MdCAX3 activated by MdCXIP1 is not only involved in iron uptake, but also in regulating zinc detoxification by compartmentalizing zinc in vacuoles to avoid iron starvation‐induced zinc toxicity. Thus, mobile MdCAX3 mRNA is involved in the regulation of iron and zinc homeostasis in response to iron starvation.

The long‐distance mobile mRNA MdCAX3 is upregulated in response to iron deficiency. It is transported from leaves to roots, where translated MdCAX3 is activated by the chaperone MdCXIP1 regulating iron and zinc homeostasis in response to iron starvation.

Introduction

Plants require a range of essential micronutrients for normal growth and development (Hall & Williams, 2003). A notable example is iron (Fe), which is an essential cofactor for many cellular redox reactions, such as photosynthesis and respiration (Conte & Walker, 2011). Although Fe is abundant in the earth, an oxygen‐rich atmosphere has resulted in the loss of the bioavailable form of Fe, Fe (II), and the accumulation of insoluble Fe, Fe (III), which is not readily assimilated by plants. Consequently, Fe deficiency is a common nutritional limitation for plants (Kim & Guerinot, 2007; Walker & Connolly, 2008; Morrissey & Guerinot, 2009; Ivanov et al, 2012; Thomine & Vert, 2013; Grillet & Schmidt, 2019). Zinc (Zn) and Fe share the same transporter, Zn has a stronger transporter binding affinity than Fe and so it can easily be substituted (Mellor & Maley, 1948; Martin & Marschner, 2012). Enhanced Zn uptake activates physiological and morphological Fe deficiency responses (Colangelo & Guerinot, 2006; Leskova et al, 2017; Chen et al, 2018), and toxicity caused by excessive Zn can be attenuated by Fe supplements (Fukao et al, 2011; Shanmugam et al, 2011). Cases of serious Zn pollution caused by human activities and excessive use of chemical fertilizer are becoming more increasingly frequent (Alkorta et al, 2004) and a significant side‐effect is the further deterioration of plant Fe homeostasis. This is an important driving factor in better understanding Fe and Zn interactions.

Fe deficiency leads to excessive absorption of Zn and the precise regulation of Fe and Zn homeostasis by plants is an important factor in maintaining growth. In Arabidopsis thaliana, Fe deficiency or excessive Zn treatment induces the transcription of genes encoding ZIF1, HMA3, and MTP3, which are proteins located in the tonoplast that play roles in Zn partitioning (Becher et al, 2004; Arrivault et al, 2006; Colangelo & Guerinot, 2006; van de Mortel et al, 2006; Haydon & Cobbett, 2007; Gustin et al, 2009; Yang et al, 2010; Haydon et al, 2012). Another factor in the Fe/Zn balance is Nicotianamine (NA), a non‐protein amino acid that acts as a chelator and makes an important contribution to long‐distance Fe and Zn transport (Wiren et al, 1996; Takahashi et al, 2003; Cassin et al, 2009; Klatte et al, 2009; Schuler & Bauer, 2011; Haydon et al, 2012; Schuler et al, 2012; Clemens et al, 2013). Heterologous overexpression of TcNAS1 in A. thaliana was reported to result in high NA levels that promoted root to shoot Fe transport, which, in turn, enhanced the accumulation of Zn in the roots (Pianelli et al, 2005; Cassin et al, 2009). ZIF1 functions as a NA transporter during vascular NA regionalization and affects the dynamic balance involved in Fe and Zn homeostasis (Haydon & Cobbett, 2007; Yang et al, 2010; Haydon et al, 2012; Shanmugam et al, 2012).

IRT1 is a key Fe transporter in non‐gramineous plants and is highly expressed after Fe deficiency (Eide et al, 1996; Connolly et al, 2002; Vert et al, 2002; Barberon et al, 2011; Zhang et al, 2017). However, due to the broad substrate specificity of IRT1, several heavy metals, such as Zn, manganese (Mn), cobalt (Co), nickel (Ni), and cadmium (Cd), are over‐absorbed during Fe deficiency (Henriques et al, 2002; Fukao et al, 2011; Shanmugam et al, 2011; Wu et al, 2012a; Leskova et al, 2017). Under Fe deficiency or excessive Zn conditions, IRT1 protein levels in roots increase, leading to a further imbalance in Fe and Zn homeostasis and the toxic effects of excessive Zn (Fukao et al, 2011; Shanmugam et al, 2011). IRT1 protein levels are modulated by IRT1 DEGRADATION FACTOR1 (IDF1), which regulates the degradation of IRT1 in a ubiquitination‐dependent manner (Shin et al, 2013). Further studies have shown that IRT1 directly senses elevated non‐Fe metal concentrations and integrates multiple substrate‐dependent regulation processes through a strict regulatory mechanism to avoid the toxicity caused by excessive accumulation of Zn and Mn (Dubeaux et al, 2018).

Another component of the Fe and Zn homeostasis system is the cation/H⁺ exchanger (CAX) family of divalent cations transporters. These proteins contribute to the detoxification of various metals (e.g., Mn²⁺, Cd²⁺, and Zn²⁺) when present in excess to regulate metal homeostasis via compartmentalization in the vacuole (Hirschi et al, 2000; Cheng et al, 2003, 2005; Martinoia et al, 2007; Manohar et al, 2011; Punshon et al, 2012; Wu et al, 2013; Yamada et al, 2014; Baliardini et al, 2015). They rely on a transmembrane proton gradient, which is usually present across the tonoplast (Hirschi et al, 2000; Cheng et al, 2005; Shigaki et al, 2006; Pittman & Hirschi, 2016; Zhang et al, 2016). In addition, there is evidence that in A. thaliana, AtCAX3 indirectly promotes the activity of a plasma membrane (PM) H⁺‐ATPase (Cheng et al, 2005; Pittman et al, 2005; Zhao et al, 2008; Cho et al, 2012). Interestingly, it is well known that higher PM H⁺‐ATPase activity contributes to rhizosphere acidification and promotes Fe uptake by plants by increasing the levels of soluble Fe in the rhizosphere (Santi & Schmidt, 2009; Zhao et al, 2016; Zhou et al, 2019; Sun et al, 2020).

In addition to transporters, transcription factors also have a potential role in regulating the dynamic balance between Fe and Zn. Recent studies in A. thaliana have shown that basic helix–loop–helix (bHLH) transcription factors, which regulate Fe absorption and transport, also participate in the regulation of the Fe/Zn balance by activating the expression of MTP3, HMA3 and NAS1, NAS2 (Colangelo & Guerinot, 2004; Long et al, 2010; Yang et al, 2010; Wu et al, 2012a; Chen et al, 2018). A systemic signaling pathway that operates at the whole plant level has been identified that maintains nutrient homeostasis, and it has been shown that leaves also play a role in sensing nutrient state and generate long‐distance transduction signals that travel to the roots (Enomoto et al, 2007; Lin et al, 2008; Liu et al, 2009, 2011; Wu et al, 2012b; Chen et al, 2016; Grillet et al, 2018; Khan et al, 2018). Although progress has been made in understanding the interaction between Fe and Zn homeostasis, the mechanism of shoot‐to‐root signaling regulation has yet to be determined. Grafting technology has been used to show that endogenous long‐distance transported mRNAs are transported through the phloem between rootstocks and scions, and various roles for these mRNAs in the target tissue have been characterized (Kim et al, 2001; Haywood et al, 2005; Kudo & Harada, 2007; Notaguchi et al, 2012; Thieme et al, 2015; Duan et al, 2016; Xia & Zhang, 2020). However, to date, there is no conclusive evidence that long‐distance mobile mRNAs are involved in the regulation of Fe and Zn homeostasis.

In this current study, the use of highly heterozygous Malus genotypes allowed us to identify the long‐distance mobile mRNA, MdCAX3 (Md12G11658000), which is upregulated in response to Fe deficiency. Furthermore, we demonstrate that endogenous MdCAX3 mRNA can be transported from the leaves to the roots, and that translated MdCAX3 is then activated by the molecular chaperone MdCXIP1. We present the hypothesis that this represents a key component in Fe uptake regulation and involves rhizosphere acidification and Zn detoxification by vacuole compartmentalization.

Results

MdCAX3 mRNA undergoes long‐distance movement via the phloem and participates in the Fe deficiency response

We previously established a heterologous Malus plant grafting system using a scion of “Golden Delicious” (Malus domestica, Md) and two rootstocks: Malus xiaojinensis (Mx), which is a Fe uptake efficient species (FUE), and Malus baccata (Mb), which is a Fe uptake inefficient species (FUI) (Zhang et al, 2017) to identify long‐distance mRNA signal transduction in response to Fe stress (Lv et al, 2021). From the mobile transcript profile (Lv et al, 2021), the MdCAX3 gene was found to be significantly upregulated in the roots of Md/Mx grafts following the Fe deficiency treatment, but not significantly in Md/Mb roots (Fig 1A). We measured the expression of MdCAX3 in grafted (Mx/Mx, Mb/Mb) and ungrafted (Mx, Mb) roots and leaves to investigate whether the physical injury caused by grafting affects MdCAX3 transcript levels and confirmed that it did not in both roots and leaves (Appendix Fig S1A and B). To confirm MdCAX3 mRNA mobility in Malus grafts, we analyzed the MdCAX3 gene sequences from Md, Mx, and Mb. This revealed a single nucleotide polymorphism (SNP) (G/A) at position 651 from the ATG start site, which we referred to as present in the MdCAX3^G genotype in Md and the MdCAX3^A genotype in Mb and Mx (Fig 1C and Appendix Fig S2A and B, Appendix Table S2). Notably, in the heterografts (Md/Mx, Md/Mb), the MdCAX3 ^G/A was identified in roots (Fig 1D). We further used TaqMan real‐time PCR probes to discriminate between the endogenous MdCAX3^G and MdCAX3^A mRNAs. The MdCAX3^G was found to be significantly upregulated in the roots of Md/Mx grafts following the Fe deficiency treatment and observed no changes in MdCAX3^A (Fig 1B). This showed that, under Fe‐deficient growth condition, the upregulated MdCAX3 in the root originated from the scion. To investigate the mechanism by which MdCAX3 moves from leaf‐to‐root in the Malus grafts, we used phloem girdling, which disrupts phloem transport. We found that MdCAX3^G mRNA was not detected in the roots of the girdled heterografts (Fig 1D and Appendix Fig S3). Furthermore, girdling prevented MdCAX3^G upregulation under Fe deficiency conditions (Fig 1E). The MdCAX3 gene was found to be significantly downregulated in the leaves of Md/Mx grafts following the Fe deficiency treatment, girdling prevented MdCAX3 downregulation in leaves under Fe deficiency conditions (Appendix Fig S1C). These results indicated that the additional MdCAX3 transcripts in the roots originated from the scion and moved long distance from leaf to root. To examine whether MdCAX3 mRNA is translated after transport, transient expressions of MdCAX3‐GFP in in apple leaves were conducted. We found that GFP signal could be detected in the roots of the transient transgenic apple seedlings but not in the stem (Fig 1F–H and Appendix Fig S4).

Figure 1. MdCAX3 mRNA moves from shoots to roots via the phloem and participates in the Fe deficiency response.

1. Quantitative real‐time (qRT)‐PCR analysis of MdCAX3 in Md/Mx and Md/Mb under Fe deficient and Fe replete conditions.
2. TaqMAN real‐time PCR analysis of MdCAX3^A and MdCAX3^G expression in Md/Mx and Md/Mb under Fe deficient and Fe replete conditions.
3. The MdCAX3 gene sequence contains a single nucleotide polymorphism (SNP) (G/A) 651 bp downstream from the ATG, and the stable locus is G in Md and A in Mx and Mb.
4. The mobility of MdCAX3 was verified by cloning the SNP sites described in C in Md/Mx and Md/Mb.
5. TaqMAN real‐time PCR analysis of MdCAX3^G expression in girdled Md/Mx and Md/Mb under Fe deficient and Fe replete conditions for 3 days.
6. Vectors construction. Primers were used to verify the mobility of mRNA.
7. Verify the mobility of GFP (control) mRNA and GFP‐MdCAX3 mRNA by RT‐PCR in Mx leaf, stem, and root. The recombinant vectors (F) were transformed into A. tumefaciens GV3101 and introduced into Mx leaves by vacuum infiltration.
8. GFP fluorescence image of the Mx roots under different treatment. Nuclei were labeled using DAPI. Scale bar: 100 μm.

Data information: In (A, B, E), values are means ± SD, n = 3 (biological replicates). Asterisks indicate statistically significant differences (***P < 0.001; ANOVA, Tukey correction), ns = no significant difference.

We next investigated the tissue localization of MdCAX3 expression by in situ hybridization using stem cross‐sections and observed a MdCAX3 mRNA signal in phloem sieve element companion cells. In the roots, the MdCAX3 mRNA signal was diffuse and present in all vascular bundle cells and the epidermis (Fig 2A). In addition, we examined the subcellular localization of MdCAX3 by expressing it as a fusion protein with the green fluorescent protein reporter in tobacco (Nicotiana benthamiana) leaves using transient Agrobacterium tumefaciens‐mediated transformation. To clarify MdCAX3 subcellular localization, the leaves were stained with DAPI for 5 min, and then treated with hyperosmotic (2 M NaCl) for 15 min before confocal imaging. The plasmolysis experiment indicated that the GFP‐MdCAX3 fusion protein localized in the tonoplast (Fig 2B).

Figure 2. In situ hybridization analysis of MdCAX3 mRNA and Subcellular localization of MdCAX3.

1. In situ hybridization analysis of MdCAX3 in Malus domestica root and stem cross sections. Se, sieve tube; cc, companion cell. Scale bars: 100 μm.
2. Subcellular localization of MdCAX3 in N. benthamiana leaves. The eGFP‐MdCAX3 vector was transformed into A. tumefaciens GV3101 and injected into N. benthamiana leaves. To clarify MdCAX3 subcellular localization, the leaves were stained with DAPI for 5 min, and then treated with hyperosmotic (2 M NaCl) for 15 min before confocal imaging. N, Nucleus; V, Vacuole. Scale bars: 25 μm.

CAX3 upregulates apoplastic proton extrusion during Fe starvation in FUE species

It has been reported that in A. thaliana, AtCAX3 can indirectly promote the activity of PM H⁺‐ATPases (Cheng et al, 2005; Zhao et al, 2008; Cho et al, 2012). To investigate whether MdCAX3 contributes to the downregulation of the apoplast pH as part of the Fe deficiency response, we carried out virus‐induced gene silencing (VIGS) to block the expression of MdCAX3 in the FUE species Mx, which has a high rhizosphere acidification ability (Wu et al, 2012b). Compared with the untransformed control, the apoplastic pH was notably higher in Mx when MdCAX3 expression was transiently suppressed and plants were subjected to an Fe deficiency treatment (Fig 3A–D). To determine whether the mobility of MdCAX3 is actually necessary to increase iron uptake in roots, transient suppression of MdCAX3 in apple leaves were conducted. The MdCAX3 ^G was barely found in the roots of apple plants which MdCAX3 was suppressed in leaves (Fig 3E and F). Suppression MdCAX3 in leaves prevented MdCAX3^G upregulation following the Fe deficiency treatment. Interestingly, suppression of MdCAX3 in apple leaves mimicked physiological phenomenon of the suppression of MdCAX3 in apple seedlings, indicating that the MdCAX3 mobility is involved in the Fe deficiency induced downregulation of apoplast pH in root (Fig 3G–I). We also overexpressed MdCAX3 in the FUI species Mb, which has a known low rhizosphere acidification capacity, and observed no changes in apoplastic pH (Appendix Fig S5A–D). Taken together, these results suggested that MdCAX3 is involved in the regulation of the apoplastic pH in FUE species, but not in FUI species.

Figure 3. MdCAX3 is involved in the regulation of apoplastic pH in Mx and Md/Mx .

1. qRT‐PCR analysis of MdCAX3 in the roots of MdCAX3‐suppressed Mx.
2. Fluorescence ratio (488/440 nm excitations) of BCECF fluorescence in the roots of MdCAX3‐suppressed Mx.
3. Ratio (488/440) image of BCECF fluorescence in the roots of MdCAX3‐suppressed Mx. Scale bars: 100 μm.
4. Rhizosphere acidification in MdCAX3‐suppressed Mx Bromocresol purple was used as a pH indicator for visualization. Scale bars: 1 cm.
5. qRT‐PCR analysis of MdCAX3 in the leaves of MdCAX3‐suppressed in leaves Md/Mx.
6. TaqMAN real‐time PCR analysis of MdCAX3^G expression in the roots of MdCAX3‐suppressed in leaves Md/Mx under Fe deficient and Fe replete conditions for 3 days.
7. Fluorescence ratio (488/440 nm excitations) of BCECF fluorescence in the roots of MdCAX3‐suppressed in leaves Md/Mx.
8. Ratio (488/440) image of BCECF fluorescence in the roots of MdCAX3‐suppressed in leaves Md/Mx. Scale bars: 100 μm.
9. Rhizosphere acidification of MdCAX3‐suppressed in leaves Md/Mx. Bromocresol purple was used as a pH indicator for visualization. Scale bars: 1 cm.

Data information: In (A, E, F), values are means ± SD, n = 3 (biological replicates). Asterisks indicate statistically significant differences (***P < 0.001; ANOVA, Tukey correction), ns = no significant difference. In (B, G), the central band represents the median and the box ranges showing the interquartile range and covers the central 50% of the data. The whiskers showing the minimum and maximum of the data. Three biological replicates; three roots were quantified for each replicate. Asterisks indicate statistically significant differences (***P < 0.001; **P < 0.01; ANOVA, Tukey correction), ns = no significant difference.

The mechanistic basis of the different roles of MdCAX3 in FUE and FUI species

Plant CAX proteins have an N‐terminal autoinhibitory domain (NRR), that is activated when the NRR is removed or bound by a CAX interacting protein (CXIP1, CXIP4) or CAX1 (Shigaki et al, 2001; Pittman et al, 2002; Cheng & Hirschi, 2003; Cheng et al, 2004; Shigaki & Hirschi, 2006; Zhao et al, 2009a, 2009b; Hocking et al, 2017). To investigate why MdCAX3 overexpression in FUI species (Mb) did not affect the apoplastic pH, we measured MdCXIPs expression levels and saw a significant difference in MdCXIP1 (MD04G1241900) transcript abundance between the roots of Md/Mb and Md/Mx plants, with MdCXIP1 barely being detected in the Md/Mb roots (Fig 4A and B, Appendix Fig S6, Appendix Table S2). To elucidate the reason for the difference in MdCXIP1 expression between Md/Mb and Md/Mx, we analyzed its sequence in Mb and Mx. Compared with Mx, we identified a long deletion in the MdCXIP1 promoter region in Mb, −486 to −454 bp upstream from the ATG initiation site (Fig 4C and Appendix Fig S7). Next, we separately fused the MdCXIP1 promoter sequences from Mx and Mb with the GUS coding sequence and transiently co‐expressed each with GFP, driven by CaMV 35S promoter, in tobacco (Nicotiana benthamiana) leaves. The expression of GUS and GFP in the tobacco leaves was detected by quantitative real‐time (qRT) PCR, and we observed that the relative expression of GUS driven by the Mb MdCXIP1 promoter to GFP expression was significantly lower than relative expression of GUS driven by Mx MdCXIP1 promoter to GFP (Fig 4D–F). Taken together, these results suggest that the deletion in the MdCXIP1 promoter weakens its basal activity, and that the natural variation in MdCXIP1 promoter activity may explain why MdCAX3 overexpression in Mb did not affect apoplastic pH.

Figure 4. The MbCXIP1 promoter is weaker than the MxCXIP1 promoter.

1. Transcriptional profiling of MdCXIP1 in Md/Mx and Md/Mb under Fe deficient and Fe replete conditions.
2. Quantitative real‐time (qRT)‐PCR analysis of MdCXIP1 expression in Mx and Mb.
3. Sequence differences of CXIP1 promoters between Mx and Mb.
4. Schematic diagram of vectors used for promoter activity analysis.
5. qRT‐PCR analysis of β‐glucuronidase (GUS)/green fluorescent protein (GFP) expression in Nicotiana benthamiana leaves after transient co‐expression of pro35S::GFP with proMxCXIP1:GUS or proMbCXIP1:GUS, and co‐expression with pBI101:GUS as control.
6. GUS staining in N. benthamiana leaves after transient co‐expression of pro35S::GFP with proMxCXIP1:GUS or proMbCXIP1:GUS, and co‐expression with pBI101:GUS as a control.
7. Interaction between MdCXIP1 and MdCAX3 tested by yeast two‐hybrid (Y2H) assays. A dilution series of NMY51 yeast cells co‐expressing MdCXIP1 and MdCAX3 cultured on SD/II and SDIV screening media. pPBT3‐N (empty vector) co‐expressed with MdCXIP1 was used as a negative control. X‐gal was used in SD/‐Trp/‐Leu/‐His/‐Ade screening medium to further test for possible interactions.
8. Interaction between MdCXIP1 and MdCAX3 tested by bimolecular fluorescence complementation (BiFC). Transient co‐expression of YNE and YCE‐MdCXIP1 was used as a negative control. Scale bar: 20 μm.
9. Interaction between MdCXIP1 and the MdCAX3 N‐terminal autoinhibitory domain in an in vitro pull‐down assay. The His‐MdCXIP1, GST, and GST‐N‐MdCAX3 fusion proteins were expressed in the Escherichia coli BL21 strain, and the GST protein alone was used as a negative control. “+” indicates presence and “−” indicates absence. IB, immunoblot.

Data information: In (B, E) values are means ± SD, n = 3 (biological replicates). Asterisks indicate statistically significant differences (***P < 0.001, ANOVA, Tukey correction).

Source data are available online for this figure.

Based on the above results, we hypothesized that MdCAX3 may function during Fe starvation via its interaction with MdCXIP1. We tested this by yeast two‐hybrid (Y2H) analysis and found that an interaction between MdCAX3 and MdCXIP1. Compared with the negative control, yeast cells co‐expressing pPBT3‐MdCAX3 and pPR3N‐MdCXIP1 grew grow normally on SD/‐Trp/‐Leu/‐His/‐Ade screening medium and were stained with X‐gal, indicating that there was indeed an interaction between MdCAX3 and MdCXIP1 (Fig 4G). To verify the interaction in vivo, a bimolecular fluorescence complementation (BiFC) assay was performed. Co‐expression of MdCXIP1‐cYFP (MdCXIP1 fused with a yellow fluorescent protein (YFP) C‐terminal fragment), with nYFP‐MdCAX3 (MdCAX3 fused with a YFP N‐terminal fragment) resulted in strong YFP signals in the tonoplast (Fig 4H), while the negative controls showed no fluorescence. We also expressed and purified a recombinant His‐tagged MdCXIP1 protein (His‐MdCXIP1), GST‐tagged N‐MdCAX3 (GST fused with the N‐terminal of MdCAX3), and then performed a pull‐down assay using His‐MdCXIP1 and GST‐N‐MdCAX3. After immobilization of GST‐N‐MdCAX3, His‐MdCXIP1 was detected in the immunoaffinity‐associated product, suggesting an interaction between the two proteins which occurred in the N‐terminal autoinhibitory domain of CAX (Fig 4I).

To further confirm the mechanistic basis for the loss function of MdCAX3 in Mb, we transiently co‐overexpressed MdCAX3 and MdCXIP1 in this Malus species. We observed a decrease in the apoplastic pH in the co‐overexpressing lines compared with the control (Fig 5A–E). We further transiently overexpressed sMdCAX3 (MdCAX3 with the N‐terminal autoinhibitory domain removed) in Mb which we observed a significant decrease in the apoplastic pH in the root tips (Appendix Fig S8A–D). We also suppressed MdCXIP1 in the FUE species Mx, and Fe deficiency induced decrease in apoplastic pH was blocked (Appendix Fig S8E–H). Based on these results, we concluded that MdCAX3 can enhance rhizosphere acidification and participate in the Fe deficiency response, and that MdCXIP1 activates MdCAX3.

Figure 5. Co‐overexpression of MdCAX3 and MdCXIP1 in Mb enhances rhizosphere acidification as part of the Fe deficiency response.

• A, B
  Quantitative real‐time (qRT) PCR analysis of MdCAX3 and MdCXIP1 expression in the roots of MdCAX3 and MdCXIP1 co‐overexpressing Mb. Values are means ± SD, n = 3 (biological replicates). Asterisks indicate statistically significant differences (***P < 0.001, ANOVA, Tukey correction).
• C
  Fluorescence ratio (488/440 nm excitations) of BCECF fluorescence in the roots of MdCAX3 and MdCXIP1 co‐overexpressing Mb. The central band represents the median and the box ranges showing the interquartile range and covers the central 50% of the data. The whiskers showing the minimum and maximum of the data. Three biological replicates; 3 roots were quantified for each replicate. Asterisks indicate statistically significant differences (***P < 0.001; ANOVA, Tukey correction).
• D
  Ratio (488/440) image of BCECF fluorescence in MdCAX3 and MdCXIP1 co‐overexpressing Mb. Scale bar: 100 μm.
• E
  Rhizosphere acidification in MdCAX3 and MdCXIP1 co‐overexpressing Mb. Bromocresol purple was used as a pH indicator for visualization. Scale bar: 1 cm.

MdCXIP1‐activated MdCAX3 regulates Zn detoxification by promoting sequestration of Zn in vacuoles

The CAX family plays an important role as cation transporters in the detoxification of excess heavy metals to the vacuolar compartments of plants (Cheng et al, 2003, 2005; Bradshaw, 2005; Kamiya et al, 2012; Punshon et al, 2012). Normally, under Fe deficiency conditions, upregulation of IRT1 will lead to the accumulation of heavy metals, especially Zn ions. After Fe deficiency treatment of the plants in the Malus homograft system, we observed a significant negative correlation between the Fe and Zn content in the roots of both Md/Mx (Correlation coefficient = −0.781**) and Md/Mb (Correlation coefficient = −0.846**), which indicated that a large amount of Zn had accumulated in the roots (Fig 6A). The Zn content in leaves of Md/Mb was much higher compared with Md/Mx (Fig 6B). To date, the transport of specific metal ions and the identification of CAX domains has mainly been verified by heterologous expression in S. cerevisiae strain K667 (Cheng et al, 2005; Shigaki et al, 2005; Shigaki & Hirschi, 2006; Edmond et al, 2009). To determine whether MdCAX3 is involved in the regulation of Zn detoxification, we heterologously expressed sMdCAX3 (MdCAX3 with the N‐terminal autoinhibitory domain removed) and MdCAX3 in S. cerevisiae strain K667. The results showed that heterologous expression of the sMdCAX3 sequence strongly increased the tolerance of S. cerevisiae to excess Zn, while expression of the MdCAX3 sequence only slightly restored the growth of the yeast. We also co‐expressed MdCAX3 and MdCXIP1 in S. cerevisiae and the resulting strong recovery of yeast cell growth indicated that MdCXIP1 activates MdCAX3 transport activity and improves the tolerance of yeast to excess Zn (Fig 6C and Appendix Fig S9A–C). To further investigate the effect of MdCAX3 on Zn compartmentalization in vacuoles, we measured the Zn content of vacuoles from plants with silenced MdCAX3 expression. Compared with the control, in Mx lines in which MdCAX3 was transiently suppressed (pTRV‐CAX3) the Zn content in the vacuole was significantly lower (Fig 6D and E). Conversely, leaf chlorosis was stronger in Md/Mb comparing with Md/Mx, which further indicated that MdCAX3 regulates Zn detoxification by compartmentalizing Zn in vacuoles (Fig 6F). To further investigate whether MdCAX3 and MdCXIP1 are involved in Zn detoxification, we transiently suppressed MdCAX3 or MdCXIP1 in apple plantlets. We examined the effect of an excessive Zn treatment on MdCAX3 and MdCXIP1 expression in MdCAX3‐suppressed or MdCXIP1‐suppressed apple plants. MdCAX3 expression was upregulated with excessive Zn treatment in MdCXIP1‐suppressed apple plants (Fig 7A and B). Furthermore, the MdCAX3‐suppressed or MdCXIP1‐suppressed apple plants displayed a Zn‐sensitive phenotype and significantly reduced Fe uptake capacity compared with control plants under excessive Zn treatment (Fig 7C and D). To sum up, MdCAX3 can positively involve in Zn vacuole compartmentalization, but inhibition of the N‐terminal autoinhibitory domain requires activation by MdCXIP1.

Figure 6. MdCAX3 activated by MdCXIP1 is involved in regulating Zn detoxification by sequestering Zn in vacuoles.

1. Positive correlation of Fe (up) and Zn (down) content in Md/Mx and Md/Mb under Fe deficient and Fe replete conditions. Values are means ± SD, n = 3 (biological replicates).
2. Zn content in leaves of Md/Mx and Md/Mb under Fe deficient and Fe replete conditions. Values are means ± SD, n = 3 (biological replicates). Asterisks indicate statistically significant differences (***P < 0.001, Student’s t‐test), ns = no significant difference.
3. Suppression of Zn²⁺ sensitivity in S. cerevisiae strain K667 (pmc1 vcx1 cnb1) determined by expressing sMdCAX3 (N‐terminally truncated) and full‐length MdCAX3 with or without MdCXIP1. Growth of a dilution series of yeast cells expressing the various plasmids on selection medium lacking histidine (−Ura) and spotted onto YPD medium containing 10 mM ZnSO₄. The W303‐1A expression of pYES2 (empty vector) was used as a positive control.
4. Cellular localization of Zn²⁺ (blue) in Mx and MdCAX3‐suppressed Mx roots under Fe deficient conditions. Arrowhead indicates the cytoplasm. Scale bars: 20 μm.
5. Fluorescence intensity detection at the position indicated by the long and narrow arrow in the Fig 6D.
6. Shoot phenotype of Md/Mx and Md/Mb under Fe deficient and Fe replete conditions. Scale bar: 2 cm.

Figure 7. Silencing MdCAX3 or MdCXIP1 greatly reduced the tolerance of Mx to excess Zn and reduced Fe absorption.

• A, B
  Quantitative real‐time (qRT) PCR analysis of MdCAX3 or MdCXIP1 in the silencing MdCAX3 or MdCXIP1 Mx roots under control (Hoagland nutrient solution) and excess Zn (300 μM Zn) conditions for 3 weeks. Values are means ± SD, n = 3 (biological replicates). Asterisks indicate statistically significant differences (***P < 0.001, Student’s t‐test), ns = no significant difference.
• C
  Shoot phenotypes of Mx under control (Hoagland nutrient solution) and excess Zn (300 μM Zn) conditions for 3 weeks. Scale bar: 1 cm.
• D
  Perls stain and Zinquin ethyl ester assays of Mx under control (Hoagland nutrient solution) and excess Zn (300 μM Zn) conditions for 3 weeks. Leaf Scale bar: 0.5 cm, Root Scale bar: 0.5 mm.

To investigate whether CAX3 mediated Fe/Zn homeostasis may be a common phenomenon in Arabidopsis thaliana, we evaluated the phenotypes of the cax3 mutant lines with Fe‐deficient or Zn‐excessive treatment (Appendix Fig S10A–D). The results showed that the expressions of the IRT1 in roots of the cax3 mutant compared with the control was downregulated with Fe‐deficient or Zn‐excessive treatment (Appendix Fig S10E). Conversely, compared with the control, in cax3 mutant, the Fe content in the roots was significantly lower (Appendix Fig S10G). Furthermore, the cax3 mutant displayed a Zn sensitive phenotype and significantly reduced Fe uptake capacity compared with control plants under excessive Zn treatment (Appendix Fig S10D and F–H).

Discussion

Based on the above results, we propose the following model: during Fe deficiency, MdCAX3 mRNA in leaves is continuously transported through the phloem to the roots. There it is translated to a protein whose function is to pump Zn ions into vacuoles in a partition detoxification process that stops the degradation of IRT1 induced by excessive Zn (Dubeaux et al, 2018) and promotes the plasma membrane HA to pump protons to the apoplast, thereby causing rhizosphere acidification. This, in turn, promotes Fe absorption and regulates Fe and Zn homeostasis. However, in Mb, this process cannot be activated because of the lack of MdCXIP1 (Fig 8).

Figure 8. Proposed model for long‐distance transport of mobile MdCAX3 mRNA modulates both Fe uptake and Zn compartmentalization during Fe starvation.

Under Fe deficient conditions, MdCAX3 mRNA is continuously transported from leaves to the roots through the phloem, and is translated in the roots to proteins that pump zinc ions into vacuoles for partition detoxification. This relieves the degradation of MdIRT1 induced by excessive Zn levels and promotes the PM H⁺‐ATPase to pump protons to the apoplast to promote rhizosphere acidification. In Mb, the process is hampered by the lack of MdCXIP1.

Plants tightly regulate Fe acquisition and distribution (Palmer & Guerinot, 2009). However, besides a shortage of Fe for cellular redox reactions, when there is an insufficient Fe supply the resulting Fe deficiency responses cause over‐accumulation of other transition metals, such as Zn (Haydon et al, 2012). Therefore, mechanisms have evolved to coordinate the metal balance (Arrivault et al, 2006; Schaaf et al, 2006; Morrissey & Guerinot, 2009). Previous physiological, transcriptome, metabolic, and signal transmission studies have provided insights into the mechanisms involved in nutrient homeostasis and evidence exists for shoot‐to‐root mRNA signaling in response to nutrient status (Enomoto et al, 2007; Lin et al, 2008; Liu et al, 2009; Wu et al, 2012b; Chen et al, 2016; Grillet et al, 2018). However, a role for mobile RNAs in Zn and Fe homeostasis has yet to be demonstrated. Zn toxicity or Fe deficiency led to chlorosis in leaves and so there is a possibility that leaves could play a role in Zn and Fe sensing. In this current study, we present several lines of evidence that MdCAX3 can undergo leaf‐to‐root movement in Malus grafts, and that the MdCAX3 mRNA was transported to coordinate the Fe deficiency response and Zn detoxification. MdCAX3 mRNA can be transported to the root to promote PM H⁺‐ATPase activity there, which, in turn, increases Fe uptake by increasing the soluble Fe in the rhizosphere. Thus, MdCAX3 makes a significant contribution to the detoxification of excess Zn by vacuole compartmentalization. Our studies provide the first mechanistic insights into the mRNA signaling mechanism underlying metal balancing during Fe deficiency stress.

Various methods, such as grafting, annulus stripping, phloem sap testing, and differences of the phloem RNA content, have shown that there are different kinds and quantities of RNA molecules in the phloem of different plant species (Kim et al, 2001; Yoo et al, 2004; Haywood et al, 2005; Kudo & Harada, 2007; Pant et al, 2008; Pant et al, 2008; Ham et al, 2009; Notaguchi et al, 2012, 2015; Thieme et al, 2015; Chen et al, 2016; Duan et al, 2016; Xia et al, 2018). Previous studies have indicated that graft‐transmissible mRNA orthologs are shared among different plant species, which, in turn, indicates a conserved role in modulating physiological processes. By searching the data from other heterograft systems, shoot‐to‐root mobile MdCAX3 was found to be conserved in species such as A. thaliana, and N. benthamiana/tomato (Thieme et al, 2015; Xia et al, 2018). The conserved mobility of CAX3 among plant species indicates that the mobile CAX3 signaling pathway may be conserved and so may be broadly involved in modulating the Fe deficiency responses. Our study showed cax3 mutant displayed a Zn‐sensitive phenotype and significantly reduced Fe uptake capacity compared with control plants under excessive Zn treatment (Appendix Fig S10). AtCXIP1/GRX14 is also referred to as Arabidopsis class II GRXs (GRXcp) (Cheng et al, 2006; Valassakis et al, 2019). Plant glutaredoxins (GRXs) are able to bind iron–sulfur clusters in vitro, which are conserved across species, have been shown to be crucial in regulating iron homeostasis (Cheng et al, 2006). Those containing stable Fe–S clusters could serve as a redox or iron sensor (Cheng et al, 2006; Valassakis et al, 2019). GRXS14 deficiency results in drastically reduced chlorophyll content in dark conditions and that its overexpression is associated with both modified chlorophyll content (Rey et al, 2017). Nagels Durand et al (2016) showed heterologous expression of GRXS17 suppresses the over‐accumulation of iron in the Saccharomyces cerevisiae Grx3/Grx4 mutant and disruption of GRXS17 causes plant sensitivity to iron deficiency stress. Furthermore, it was proved that GRXS17 is required for tolerance to iron deficiency (Yu et al, 2017; Cheng et al, 2020). We suppose that the CAX3/CXIP1‐mediated Fe/Zn homeostasis may be a common phenomenon in plant species and related mechanism remains to be studied.

We inferred from our results that MdCAX3 promoted proton extrusion during Fe starvation in the roots of Mx, but not Mb (Fig 3 and Appendix Fig S5). Self‐incompatible Malus species have particularly high levels of genetic variation, which in our study was sufficient to allow us to demonstrate that the MdCAX interacting protein (MdCXIP1) transcript levels were different in Mx and Mb. We identified a deletion in the MbCXIP1 promoter that is responsible for weakening the basal activity of the MdCXIP1 promoter, and consequently the expression of the corresponding gene (Fig 4). These results explained why the over‐expression of MdCAX3 in Mb did not affect the apoplastic pH. However, the biological significance of the MdCXIP1 promoter variation in nutrient homeostasis remains to be established. Previously, we suggested that the MdIRT1 mutant allele with a TATA‐box insertion caused increased MdIRT1 expression, and consequently higher Fe uptake in Mx, whereas this TATA‐box insertion is absent in the equivalent sequence of Mb resulting in inefficient Fe uptake (Zhang et al, 2017). Due to the broad substrate specificity of IRT1, heavy metals such as Zn, Mn, Co, Ni, and Cd are taken up to toxic levels under Fe deficient conditions (Henriques et al, 2002; Fukao et al, 2011; Shanmugam et al, 2011; Wu et al, 2012a; Leskova et al, 2017). IRT1 acts as a transceptor that directly senses excess non‐Fe metal substrates in the cytoplasm, leading to its own degradation (Dubeaux et al, 2018). As IRT1 is strongly expressed in Mx, to avoid over‐absorbing Zn, MdCXIP1 which can activate MdCAX3 existed enhanced basal activity. In contrast, MdIRT1 is weakly expressed in Mb and so does not take up heavy metals efficiently, and so Mb does not over‐absorb heavy metals during Fe deficiency stress. This might explain why MdCXIP1 expression is much lower in Mb than in Mx. These results suggest that natural variation between plant species has evolved to maintain a dynamic balance of Fe and Zn.

The presence of a wide range of proteins in the phloem sap supports the idea that they are involved in long‐distance transport at the whole plant level (Nakamura et al, 1993; Balachandran et al, 1997; Ishiwatari et al, 1998; Kehr et al, 1999; Schobert et al, 2000; Aoki et al, 2002; Yoo et al, 2002; Walz et al, 2004). Recently, increasing evidence has been provided that phloem proteins are involved in long‐distance shoot‐to‐root coordination. For example, HY5 is a shoot‐to‐root mobile protein signal that mediates light regulation of root growth and maintain the balance of carbon(C) and nitrogen (N) metabolism (Chen et al, 2016), and IRON MAN (IMA) phloem‐mobile peptides have been shown to positively regulate Fe uptake in roots (Grillet et al, 2018). Proteins transported to destination tissues can function directly without first having to be translated and so genetic and physiological studies have been combined to establish a paradigm for mobile proteins in plants, targeting effects in distinct and remote organs. The question of how mobile RNAs function in destination tissues, whether by translation, processing to small RNAs, or an alternative mechanism, is interesting and will be the target of future studies. Proteomic studies of grafted plants indicate the presence of proteins translated from mobile RNAs, suggesting that they move through complex routes and that their function may be spatially uncoupled from their cells of origin (Thieme et al, 2015). Recent studies showed that long‐distance mobile TCTP1 is translated after transport in distinct root cells and affects root growth (Branco & Masle, 2019). Our results showed that translated MdCAX3 is activated by the molecular chaperone, MdCXIP1. Furthermore, our study suggests that mobile RNAs are critical in orchestrating the nutrient demand of the plant by long‐distance shoot‐to‐root communication, and that key proteins may not originate in situ but rather result from the transport and translation of their cognate mRNAs.

Materials and Methods

Plant materials and growth conditions

Malus domestica ‘Golden Delicious’ (Md), M. baccata (Mb) and M. xiaojinensis (Mx) tissue culture seedlings were cultured in Murashige & Skoog (MS, PhytoTechnology, M519) medium containing 0.5 mg/l 6‐benzylaminopurine (6‐BA) and 0.5 mg/l indole‐3‐butytric acid (IBA) for subculturing. Seedlings longer than 3 cm were selected and cultured on ½ MS medium containing 0.5 mg/l IBA for rooting. After rooting, the seedlings were transferred to Hoagland nutrient solution (Han et al, 1994) and cultured for 1 month before treatment. Malus plants used for the Fe deficiency experiment were cultured in the presence of normal Fe concentrations (40 μM) or with no Fe. The excess Zn treatment experiment plants were cultured in the presence of 300 μM Zn for 3 weeks. Malus plant materials were grown under a 16‐h day (25 ± 2°C)/8‐h night (21 ± 2°C) cycle with a 250 μmol m⁻² s⁻¹ light intensity.

The grafting experiments were performed as described previously (Sun et al, 2020). Mx and Mb were used as rootstocks to graft with Md, respectively (Md/Mx and Md/Mb) or itself (Mx/Mx and Mb/Mb) and were cultured on ½ MS medium containing 0.5 mg/L IBA for rooting. They were then transferred to Hoagland nutrient solution and cultured for 1 month before girdling and Fe deficient treatment experiments.

The Phloem girdling experiments were performed as described previously (lv et al, 2021). Grafted seedlings were selected for girdling experiments which involved using a scalpel to remove a 1‐mm‐wide section of bark at the stem base to the depth of the xylem. The treatment of iron deficiency was carried out immediately after girdling.

Arabidopsis thaliana lines included WT (Col‐0), and a T‐DNA insertion line cax3 (SALK_094565C) which confirmed by PCR. The seeds were surface‐sterilized and rinsed three times with sterile water, seeded on solid half MS medium (pH 5.8) under sterile conditions. After 48 h of incubation at 4°C to promote germination, seedlings were cultured vertically for 7 days (16‐/8‐h light/dark cycle, temperature 22°C). Seedlings were transferred to half MS medium and half MS medium with Fe deficient (0 μM Fe + 300 μM Ferrozine) or Zn excessive (300 μM Zn) for treatment. Arabidopsis thaliana was sampled for RNA extraction with grown for 3 days, and for phenotyping, elemental analysis, grown for 9 days.

qRT‐PCR

Total RNA was extracted from Malus and N. benthamiana using an RNAprep Pure Plant Plus Kit (TIANGEN, DP441). First‐strand cDNA was synthesized using a first‐strand cDNA synthesis kit (Vazyme, R232‐01), according to the manufacturer’s instructions. QRT‐PCR was performed with an ABI QuantStudio TM6 Flex system (ThermoFisher Scientific) using the SensiFAST™ SYBR Lo‐ROX Kit (Bioline, BIO‐94005) in a 20 μl reaction volume (SYBR Lo‐ROX mix 10 μl, cDNA 2 μl, forward and reverse primer 0.5 μl each, ddH₂O 7 μl), with a thermal cycling as follows: 95°C for 2 min, 40 cycles for 10 s at 95°C and 30 s at 60°C. TaqMAN real‐time PCR was performed with an ABI QuantStudio TM6 Flex system (ThermoFisher Scientific) using the Premix Ex Taq™ Kit (TaKaRa, RR390L) in a 20 μl reaction volume (Premix Ex Taq™ mix 10 μl, cDNA 2 μl, forward and reverse primer 0.4 μl each, Probe 0.8 μl, ROX Reference Dye (50X) 0.04 μl, ddH₂O 6.36 μl), with a thermal cycling as follows: 95°C for 3 min, 40 cycles for 5 s at 95°C and 34 s at 60°C. The probe was modified by 5’‐FAM and 3’‐MGB. β‐ACTIN (MD07G1012400) was used as internal control and transcript abundance was calculated using the 2^{−ΔΔ C T} method (Livak & Schmittgen, 2001). The primers used are listed in Appendix Table S1.

SNP typing

To study the mRNA mobility of MdCAX3, we detected the stable SNP of MdCAX3 in the roots and leaves of Md/Mb and Md/Mx grafted seedlings under Fe deficient and Fe replete conditions. Monoclonal sequencing technique was used to detect the stable SNP A/G of different samples, 3 biological repeats, and 6 random single spots were sequenced for each repeat.

In situ hybridization of MdCAX3

Apple roots and stems were sampled for fluorescence in situ hybridization (FISH) analysis as previously described (Yang et al, 2020) with an MdCAX3 gene‐specific probe. The probes used are listed in Appendix Table S1. Nuclei were labeled using 4’,6‐diamidino‐2‐phenylindole (DAPI, D9542; Sigma). Confocal imaging was performed using an FV3000 microscope (Olympus).

Subcellular localization of MdCAX3

The full‐length MdCAX3 coding sequence (CDS) was inserted into the PRI101‐eGFP vector at the BsrG1 site to generate a GFP‐MdCAX3 fusion protein. Subsequently, the recombinant vectors were transformed into A. tumefaciens GV3101 and injected into N. benthamiana leaves. To clarify MdCAX3 subcellular localization, the leaves were stained with DAPI for 5 min, and then treated with hyperosmotic (2 M NaCl) for 15 min before confocal imaging. Confocal imaging was performed with an FV3000 microscope (Olympus) as previously described (Sun et al, 2020). Primers used are listed in Appendix Table S1.

Vector construction and transient expression

DNA fragments corresponding to the MdCAX3 and MdCXIP1 full‐length CDS were separately inserted into the PRI101 vector at the BamH1 site to generate overexpression vectors. VIGS (Ramegowda et al, 2014) was used to suppress MdCAX3 and MdCXIP1 expression by inserting the CDS fragments of MdCAX3 (308 bp) and MdCXIP1 (350 bp) into the pTRV2 vector at the XhoI site.

Transient expression experiments were performed as previously described (Sun et al, 2020) with the empty vectors used as negative controls. Plants were selected and divided into two batches for separate transformation, and each batch of six cultures was denoted as #1 and #2 for treatments. Primers used are listed in Appendix Table S1.

Rhizosphere apoplast pH fluorescence analysis

Apple roots from the different treatments were sampled for rhizosphere apoplast pH fluorescence analysis using the pH indicator 2’,7’‐Bis‐(2‐carboxyethyl)‐5‐(and‐6‐) carboxyfluorescein‐acetoxymethyl ester (BCECF‐AM, Beyotime, C1006) as previously described (Tang et al, 2012). The fluorescence emission window for BCECF was set to 488 and 440 nm. Confocal imaging and fluorescence ratio (488/440 nm excitations) were performed with an LSM 710 microscope (Zeiss).

Rhizosphere acidification assay

Rhizosphere acidification assays were performed as previously described (Zhao et al, 2016). Apples from different treatments were transferred to medium containing 1% w/v agarose and 0.06% bromocresol purple (pH adjusted to 6.5 with NaOH) for 1 h.

Promoter activity analysis

To investigate the CXIP1 promoter activity in Mx and Mb, the promoters were inserted into the pBI101‐GUS vector at the SalI site. To eliminate any difference due to transformation efficiency, the recombinant vectors were co‐transformed into N. benthamiana leaves with the PRI101‐eGFP empty vector driven by the CaMV 35S promoter. Transformation was performed as described above. The expression of GUS and GFP in leaves was determined by qRT‐PCR as above. GUS staining was performed as previously described (Jefferson, 1987). Primers used are listed in Appendix Table S1.

Y2H assay

The full‐length MdCAX3 CDS sequence was inserted into the pPBT3‐N (bait vector) at the NcoI site, and the full‐length MdCXIP1 CDS was inserted into the pPR3‐N (prey vector) at the EcoRI site (Snider et al, 2010). The recombinant vectors were verified by sequencing and then co‐transformed into the NMY51 strain of Saccharomyces cerevisiae. pPBT3‐N (empty vector) co‐expressed with MdCXIP1 was used as a negative control. An NMY51 yeast clone dilution series was cultured on SD/‐Trp/‐Leu (SD/II) and SD/‐Trp/‐Leu/‐His/‐Ade (SD/IV) screening media. Y2H assays were performed as previously described (Zhao et al, 2016). Primers used are listed in Appendix Table S1.

BiFC assay

The MdCAX3 and MdCXIP1 full‐length CDS sequences without stop codons were separately inserted into the pSPYNE‐35S and pSPYCE‐35S vectors at the BamHI site (Hu et al, 2002). Subsequently, the recombinant vectors were transformed into A. tumefaciens GV3101 and co‐injected into N. benthamiana leaves. Confocal imaging was performed with an FV3000 microscope (Olympus). Primers used are listed in Appendix Table S1.

Pull‐down assays

The N‐terminal MdCAX3 CDS (encoding amino acids 1‐62) was inserted into the pGEX‐4T‐1 vector (GE) at the SalI site to generate the GST‐N‐MdCAX3 fusion protein. The full‐length MdCXIP1 CDS was inserted into the pET‐32a vector (Novagen) at the SalI site to generate the His‐MdCXIP1 fusion protein. Pull‐down assays were performed as previously described (Zhao et al, 2016). Primers used are listed in Appendix Table S1.

Metal ion contents

Malus plant and Arabidopsis thaliana from different treatment were sampled to measure metal ion content. Samples were dried at 65°C for 1 days and incubated with 1.0 M HCl with gentle shaking for 5 h. The metal ion contents of extracts (filtered with 0.45 μm filter membrane) were measured by inductively coupled plasma mass spectrometer (ICP‐MS, OPTIMA3300DV; Perkin Elmer, Shelton, CT).

Yeast transformation and metal tolerance assays

The WT S. cerevisiae strain W303‐1A and mutant strain K667 (vcx1::hisG cnb1::LEU2 pmc1::TRP1 ade2‐1 can1‐100 his3‐11,15 leu2‐3,112 trp1‐1 ura3‐1) were used in the experiment (Cunningham & Fink, 1996). The MdCAX3, MdCXIP1, and sMdCAX3 (1‐96 truncated) sequences were separately inserted into the pYES2 vector (Invitrogen) at the BamHI site. Yeast cell dilution series expressing the various plasmids were grown on selection medium lacking histidine (−Ura) and spotted onto YPD medium containing 100 mM CaCl₂ or 10 mM ZnSO₄, and 2 mM MnCl₂. The pYES2 (empty vector) expression of W303‐1A was used as a positive control. Yeast transformation and growth assays were performed as previously described (Cheng & Hirschi, 2003). Primers used are listed in Appendix Table S1.

Perls staining

For visualization of Fe localization, apple roots and leaves from different treatments were sampled for Perls staining. The samples were transferred into 75% alcohol, decolorized for 3 h in a constant temperature shaker at 37°C, and then stained with Perls staining solution for 1 h. Perls staining solution is made of equal volumes of 4% [v/v] HCl and 4% [w/v] K‐ferrocyanide. Imaging was performed with a SEX16 (Olympus) after washed with deionized water.

Intracellular Zn assays

Apple roots from different treatments were sampled for intracellular Zn assays using the fluorescent Zn indicator Zinquin ethyl ester (Mkbio, CAS NO.181530‐09‐6). The experiments were carried out according to the manufacturer’s instructions. The fluorescence emission window for zinquin ethyl ester was set to 407–457 nm. Confocal imaging and fluorescence intensity detection were performed with a Ti‐E microscope (Nikon).

Author contributions

Pengbo Hao: Data curation; Formal analysis; Methodology; Writing—original draft. Xinmin Lv: Data curation. Mengmeng Fu: Methodology. Zhen Xu: Methodology. Ji Tian: Writing—review and editing. Yi Wang: Methodology. Xinzhong Zhang: Resources. Xuefeng Xu: Resources. Ting Wu: Conceptualization; Supervision; Funding acquisition; Writing—original draft; Project administration; Writing—review and editing. Zhenhai Han: Conceptualization; Supervision; Funding acquisition.

In addition to the CRediT author contributions listed above, the contributions in detail are:

TW and ZH conceived and designed the research. PH, XL, MF, and ZX conducted the experiments. YW, XX, and XZ contributed reagents and analytical tools. TW and PH wrote the manuscript. JT gave advice and edited the manuscript. All authors read and approved the manuscript.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Source Data for Figure 4

EMBO reports (2022) 23: e53698.

Data availability

The authors declare that no data have been deposited in external repositories.