Iron (Fe) and phosphorus (P) are two essential elements for plant growth (Marschner, 1995). Both elements are abundant in most soils but are present in chemical forms generally not available to plants. As a result, plants enhance acquisition of Fe and P through morphological and physiological responses in their roots. Deficiency of either nutrient initiates similar root responses, such as development of subapical root hairs or cluster roots and the acidification of the rhizosphere or the production of organic acids (Lucena et al., 2015; Song and Liu, 2015; Neumann, 2016; Stetter et al., 2017; Bhosale et al., 2018; Lucena et al., 2018). A defining feature of Strategy I plants is the need to reduce Fe(III), the most abundant form of Fe in soils, to Fe(II), prior to acquisition. In Arabidopsis, this Fe(III) reduction is catalyzed by a ferric reductase (EC 1.16.1.17) encoded by the FRO2 gene, whereas Fe(II) uptake is mediated by a transporter encoded by the IRT1 gene (Walker and Connolly, 2008; Ivanov et al., 2012). Both genes are up-regulated under Fe deficiency and are activated by specific bHLH transcription factors, which are also up regulated by Fe deficiency. These bHLH transcription factors include FIT, bHLH38, bHLH39, and others in Arabidopsis (Yuan et al., 2008; Ivanov et al., 2012; Brumbarova et al., 2015). Some of the physiological responses to phosphate (Pi) starvation include the up-regulation of Pi transporters (e.g. AtPht1;4, also named AtPT2) and of phosphatase activities (e.g. AtPAP17, also named AtACP5) (Zhang et al., 2014). del Pozo et al. (1999) purified and sequenced the N-terminal region of a Pi starvation induced acid phosphatase (encoded by the AtPAP17 gene) from Arabidopsis thaliana. PAP17 shares the characteristics of type 5 acid phosphatases, a class of purple acid phosphatases (PAP) stimulated under low P availability. Recent studies have demonstrated that secretion of PAPs can facilitate utilization of organic P in the rhizosphere (Wang et al., 2009; Robinson et al., 2012; Mehra et al., 2017).

Studies in recent years have demonstrated an important role of MYB type transcription factors on the expression of genes that initiate responses to Fe or P deficiency. AtMYB72 increases its transcription under Fe deficiency and is involved in some Fe deficiency responses (García et al., 2010). Most transcription factors that induce P deficiency responses belong to the MYB family (Müller et al., 2007). Of particular note, the Arabidopsis MYB transcription factors AtPHR1 and AtPHR2 (“Phosphate Starvation Response” 1 and 2 respectively), are crucial for P deficiency signaling (Rubio et al., 2001; Franco-Zorrilla et al., 2004; Todd et al., 2004; Bari et al., 2006; Mehra et al., 2017).

In previous works, it has been found that P deficiency induces the expression of Fe-related genes; that Fe deficiency induces the expression of P-related genes and that both deficiencies induce the expression of ethylene-related genes (reviewed in García et al., 2015; Lucena et al., 2015; Lucena et al., 2018). Taken together, these results suggest the existence of cross talk between Fe and P deficiency. The upregulation of Fe acquisition genes under P deficiency is generally associated to the increased accumulation of Fe in leaves (Wang Y. H. et al., 2002; Misson et al., 2005; Hirsch et al., 2006; Ward et al., 2008; Zheng et al., 2009; Abel, 2011; Perea-García et al., 2013; Wang et al., 2014, Lucena et al., 2018). On the other hand, P fertilization excess causes Fe chlorosis in calcareous soils (Romera et al., 1992; Sánchez-Rodriguez et al., 2013; Henry et al., 2017) and Fe deficiency affects P content in leaves (de Kock, 1955).

Either work in the nineties revealed links between ethylene and responses to deficiencies of Fe (Romera and Alcántara, 1994) and P (Borch et al., 1999). Although the regulation of the genes related to these responses is not totally understood, ethylene has since been implicated in the activation of both Fe-related genes (AtFIT, AtFRO2, and AtIRT1; Lucena et al., 2006; Lucena et al., 2015; Lucena et al., 2018; García et al., 2011; García et al., 2015) and P-related genes (AtPht1;4 and AtPAP17; Lei et al., 2011; Roldán et al., 2013). In all cases, the activating effect of ethylene was dependent on the Fe or P status of the plants (García et al., 2011; Lei et al., 2011; Neumann, 2016; Liu et al., 2017), which suggests the involvement of signals other than ethylene in gene activation (García et al., 2013). García et al. (2013) showed that the foliar application of Fe blocked the expression of Fe-acquisition genes in the Arabidopsis wild-type cultivar Col-0 but not in the Arabidopsisopt3-2 mutant, which suggests the existence of a Fe-related repressive signal moving in the phloem. This mutant harbors a T-DNA insertion in the AtOPT3 promoter resulting in reduced AtOPT3 expression (Stacey et al., 2008). Recently, AtOPT3 has been implicated in the transport of Fe ions into the phloem (Zhai et al., 2014). In relation to ethylene, the opt3 mutant produces more ACC in Fe-sufficient roots than the wild-type cultivar Col-0 (García et al., 2018). In agreement with this, it also presents constitutive activation of several ethylene-related genes in Fe-sufficient roots (García et al., 2018). These results suggest that a Fe-related repressive signal entering the phloem through OPT3 blocks ethylene synthesis in roots (Figure 9; García et al., 2018).

The Arabidopsis pho2 mutant is impaired in the sensing of P-related phloem signals coming from leaves to roots and overaccumulates Pi in leaves under Pi-replete conditions (Bari et al., 2006; Chiou and Lin, 2011). Micrografting experiments revealed that a pho2 root genotype is sufficient to yield leaf Pi over-accumulation (Bari et al., 2006). PHO2 encodes an ubiquitin-conjugating E2 enzyme, that targets proteins involved in P acquisition, like Pht1;4, and P loading into the xylem, like PHO1, for degradation (Figure 9; Chiou and Lin, 2011; Liu et al., 2012; Ham et al., 2018; Wang et al., 2018). Probably, PHO2 is also implicated in the regulation of PPAs protein stability (Wang and Liu, 2018). Under P deficiency, increased miRNA399, that participates in the cleavage of PHO2 mRNA, enhances plant P acquisition and translocation by reducing the level of PHO2 protein (Figure 9; Chiou and Lin, 2011; Liu et al., 2012; Ham et al., 2018; Wang et al., 2018). miRNA399 is upregulated earlier in shoots than in roots and it has been proposed that its movement through the phloem acts as a systemic signal. In this sense, PHO2 is recognized as a systemic root sensor of shoot-derived P-related signals (Chiou and Lin, 2011).

The role of signals besides ethylene in the regulation of Fe and P acquisition genes is also supported by the fact that the Arabidopsis constitutive ethylene mutant ctr1 does not present constitutive expression of either Fe-related genes (García et al., 2014) or P-related genes (Lei et al., 2011) when grown in complete nutrient solution. Among the non-ethylene signals involved in the regulation of Fe- or P-related genes have been proposed a Fe-peptide (García et al., 2013) and the miRNA399 described above (Pant et al., 2009; Kumar et al., 2017), both moving through the phloem. These signals could interact with ethylene and confer specificity to the responses to Fe or P deficiency, avoiding the induction of the specific responses when ethylene increases due to other nutrient deficiencies or stresses. Besides the specificity conferred by these non-ethylene signals, ethylene itself could confer specificity to the responses to Fe or P deficiency by acting through different signaling pathways in each case. For example, P deficiency did not induce the expression of P-related genes in the Arabidopsis ethylene insensitive mutants ein2 and etr1 (Lei et al., 2011), whereas Fe deficiency induced the expression of Fe-related genes in both mutants (Lucena et al., 2006; García et al., 2013). Taken together, the above results indicate that ethylene could regulate different nutrient responses by acting in conjunction with other signals and also by acting through different signaling pathways. More knowledge of the specific aspects of the regulation by ethylene of the responses to these deficiencies is necessary to understand the interactions between Fe and P. In this work we have further studied the existence of cross talks between the responses to Fe and P deficiency, taking into account the time course of both deficiencies and the existence of cross talks between Fe and P phloem signals. For this, we have studied Fe and P deficiency responses in A. thaliana Col-0, opt3-2, and pho2 mutants, impaired in the transport of Fe through the phloem (opt3-2) or P homeostasis (pho2). The results obtained show that Fe deficiency induces the expression of P acquisition genes and phosphatase activity while P deficiency transitorily induces the expression of Fe acquisition genes and ferric reductase activity. Moreover, the foliar application of either Fe or P diminished the phosphatase activity and the expression of P-related genes in P deficient plants and the ferric reductase activity in Fe deficient plants. Taken together, the results show additional interactions between P and Fe signals originating in the shoots, and suggest that Fe- or P-related phloem signals can interact with ethylene for the regulation of the responses to their deficiencies.

Previous works have shown the existence of cross talk between the responses to Fe and P deficiency in such a way that P deficiency induces the expression of Fe responses and Fe deficiency induces the expression of P responses. Those works have preferentially analyzed the expression of Fe- and P-related genes and Fe and P content in leaves (de Kock, 1955; Wang et al., 2002; Misson et al., 2005; Hirsch et al., 2006; Ward et al., 2008; Zheng et al., 2009; Abel, 2011; Perea-García et al., 2013; Wang et al., 2014). However, no studies have investigated the effect of each deficiency on some activities related to the other deficiency, such as ferric reductase activity (encoded by the AtFRO2 gene in Arabidopsis) or phosphatase activity (encoded by the AtPAP17 gene and other related genes in Arabidopsis). We think that, besides gene expression, it is important to determine these activities to better understand the cross talk between Fe and P deficiency because of the possible effects of post-transcriptional regulation. Additionally, we consider it important to study the time course of each deficiency on the induction of genes related to the other deficiency and the possible cross talk between shoot derived signals associated with each deficiency.

In this work, similar to previous works (see above), the results obtained show that P deficiency can induce the expression of some Fe acquisition genes, like AtFRO2 and AtIRT1 (Figures 1C, D; Wang et al., 2002; Misson et al., 2005; Wang et al., 2014), while Fe deficiency can induce the expression of some P acquisition genes, like AtPht1;4 and AtPAP17 (Figure 2). These results match with others obtained by microarray analysis showing an early transcriptional response of both AtPht1;4 and AtPAP17 to Fe deficiency (1–3 days growing without Fe) in roots of Arabidopsis plants (Thimm et al., 2001). The induction of the Fe acquisition genes by P deficiency occurred early, after 3 h (AtFRO2) and 6 h (AtIRT1) of the deficiency treatment, and then the induction tended to disappear (Figures 1C, D), which suggests the presence of specific Fe inhibitory signals. Moreover, the ferric reductase activity was induced only transitorily (Figures 1A, B) and decreased faster than the expression of the AtFRO2 gene (Figures 1C, D), suggesting the possible existence of a post-transcriptional regulation by Fe (Connolly et al., 2003). In relation to Fe deficiency, it induced the expression of P acquisition genes and phosphatase activity after 2 days of deficiency (Figures 2 and 5), even before the induction caused by P deficiency itself (Figure 5). The induction of the phosphatase activity by Fe deficiency (Figure 5A) was not as transitory as the induction of ferric reductase by P deficiency (Figure 1A). This result is somewhat surprising since the induction of phosphatase activity is a response traditionally associated with P deficiency (Bari et al., 2006; Lei et al., 2011). The possible role of phosphatase activity on Fe nutrition is not known and warrants further investigation.

The existence of cross talk between the responses to Fe and P deficiency could be associated with common signals in the activation of both nutrient deficiency responses, such as ethylene (Nagarajan and Smith, 2012; García et al., 2015; Lucena et al., 2015; Song and Liu, 2015; Neumann, 2016; Lucena et al., 2018), nitric oxide (NO; Graziano and Lamattina, 2007; Meng et al., 2012) and auxin (Chen et al., 2010; Miura et al., 2011; Bhosale et al., 2018). In relation to ethylene, the results presented in Figure 3 show that opt3-2 and pho2, which exhibit constitutive Fe or P responses, respectively, have greater ACC content in roots than Col-0. Since both mutants are affected in the transport of Fe (opt3-2) or in the sensing of P-related signals (pho2) in the phloem (Bari et al., 2006; García et al., 2013), their higher ACC accumulation suggests that phloem-Fe signals and phloem-P signals can inhibit directly, or indirectly, ACC synthesis and, presumably, ethylene synthesis. These phloem signals should come from the shoots, as previously suggested (Buhtz et al., 2010; García et al., 2013).

To study the possible cross talk of Fe and P shoot-derived signals with the responses to both deficiencies, we sprayed leaves of Col-0, opt3-2 and pho2 plants with Fe or P, and analyzed their effects in roots. The foliar application of Fe inhibited the ferric reductase activity in Col-0 (Figure 4A) and in the pho2 mutant (Figure 4C) while it had less effect in the opt3-2 mutant (Figure 4B). This latter effect agrees with previous results (García et al., 2013). Curiously, the foliar application of P also inhibited the ferric reductase activity in Col-0 plants, less than the Fe application (Figure 4A), but had no effect on pho2 (Figure 4C) and hardly in opt3-2 (Figure 4B). These results suggest that there are Fe and P shoot-derived signals moving through the phloem that are inhibitory and that both Fe and P signals are interrelated. This interrelation is further supported when analysing phosphatase activity and the expression of one of its encoding genes, AtPAP17. As shown in Figure 5, the foliar application of either Fe or P drastically inhibited the phosphatase activity and AtPAP17 expression in Col-0. However, neither Fe nor P application appreciably inhibited these P responses in either mutant (Figures 6 and 7).

Despite the existence of common signals involved in the activation of responses to both nutrient deficiencies, like ethylene (Figure 3), and the cross talk between shoot-derived signals (see previous paragraph), some results presented in this work suggest the existence of specific signals that block the activation of the responses to one deficiency when the deficiency is caused by the other element. For example, the expression of Fe acquisition genes is almost constitutively activated in the opt3-2 mutant (Figures 8A, B) while the expression of P acquisition genes is not (Figures 8C, D). This suggests that the absence of the phloem-Fe signal related to OPT3 depresses the expression of Fe acquisition genes (García et al., 2013) but not that of P acquisition genes, probably because there are specific P signals that block it.

According to the results obtained in this study we propose a Working Model to explain the role of ethylene and P-related and Fe-related phloem signals on the regulation of Fe and P acquisition genes (Figure 9). Once inside roots, Fe (black arrows) is translocated to leaves through the xylem, bound to citrate (provided by the FRD3 transporter). In shoots, some Fe can enter the phloem through the OPT3 transporter, and moves back to roots probably bound to a chelating agent (Fe)?. In roots, this Fe? can be sensed by the Brutus protein (BTS) that blocks the expression of the Fe acquisition genes FRO2 and IRT1, probably because it inhibits the synthesis of ethylene (ET), which has been involved in their upregulation. P (blue arrows) is absorbed through P transporters, like Pht1;4, and then loaded into the xylem through transporters like PHO1. Under P deficiency, miR399 in shoots increases and moves through the phloem to roots where it suppresses PHO2. PHO2 participates in the degradation of Pht1;4 and PHO1, and probably inhibits ethylene synthesis. Consequently, under P deficiency, the suppression of PHO2 by miR399 can permit the stabilization of Pht1;4 and PHO1, and the synthesis of ethylene, which has been involved in the upregulation of Pht1;4 and PAP17, encoding a phosphatase (PAP).

In conclusion, the results obtained in this work further support the cross talk between Fe and P deficiency responses since Fe deficiency induces the expression of P acquisition genes and the phosphatase activity, and P deficiency induces the expression of Fe acquisition genes and the ferric reductase activity. In most cases, like the induction of Fe acquisition genes and ferric reductase activity by P deficiency, this occurs very transitorily, probably due to the existence of Fe-related inhibitory signals and because of the post-transcriptional regulation of FRO2 by Fe. The cross talk between both deficiencies could be related to the existence of common signals, like ethylene, implicated in the activation of their responses. Besides ethylene, the results obtained with the foliar application of Fe or P show additional interactions between P and Fe inhibitory signals coming from the shoots, and suggest that Fe- or P-related phloem signals could interact with ethylene in the regulation of the responses to their deficiencies.

CL and FR designed the experiments after discussions with JG-M and MG; RP and MG conducted the laboratory work; ÁZ and EB determined ACC; AS, EA, RP-V, and CL wrote the manuscript.

This work was supported by the European Regional Development Fund from the European Union, the “Ministerio de Economía y Competitividad” (Projects AGL2013- 40822-R and AGL2015-65104-P); the “Junta de Andalucía” (Research Groups AGR115 and BIO159) and the economic contribution of the Company TimacAgro Spain (Roullier Group).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.