The Role of Iron in the P-Acquisition Mechanisms of the Unicellular N2-Fixing Cyanobacteria Halothece sp., Found in Association With the Mediterranean Seagrass Posidonia oceanica

Posidonia oceanica, an endemic seagrass of the Mediterranean Sea harbors a high diversity of N2-fixing prokaryotes. One of these is Halothece sp., a unicellular N2-fixing cyanobacteria detected through nifH analysis from the epiphytes of P. oceanica. The most related strain in culture is Halothece sp. PCC 7418 and this was used as the test organism in this study. In the Mediterranean Sea, phosphorus (P) and iron (Fe) can be the major limiting nutrients for N2 fixation. However, information about the mechanisms of P-acquisition and the role of metals (i.e., Fe) in these processes for N2-fixing bacteria is scarce. From our genomic analyses of the test organism and other phylogenetically related N2-fixing strains, Halothece sp. PCC 7418 is one of the strains with the greatest number of gene copies (eight copies) of alkaline phosphatases (APases). Our structural analysis of PhoD (alkaline phosphatase type D) and PhoU (phosphate acquisition regulator) of Halothece sp. PCC 7418 showed the connection among metals (Ca2+ and Fe3+), and the P-acquisition mechanisms. Here, we measured the rates of alkaline phosphatase activity (APA) through MUF-P hydrolysis under different combinations of concentrations of inorganic P (PO43−) and Fe in experiments under N2-fixing (low NO3− availability) and non-N2 fixing (high NO3− availability) conditions. Our results showed that APA rates were enhanced by the increase in Fe availability under low levels of PO43−, especially under N2-fixing conditions. Moreover, the increased PO43−-uptake was reflected in the increased of the P-cellular content of the cells under N2 fixation conditions. We also found a positive significant relationship between cellular P and cellular Fe content of the cells (r2 = 0.71, p < 0.05). Our results also indicated that Fe-uptake in Halothece sp. PCC 7418 was P and Fe-dependent. This study gives first insights of P-acquisition mechanisms in the N2-fixing cyanobacteria (Halothece sp.) found in P. oceanica and highlights the role of Fe in these processes.


INTRODUCTION
Posidonia oceanica is an endemic seagrass in the Mediterranean Sea, forming extensive meadows with valuable established key ecological services: high primary productivity, as a carbon sink, as a habitat and nursery for a variety of micro-and macroorganisms, as sediment stabilizers, as buffers for ocean acidification, and as an important site for biogeochemical processes (e.g., nitrogen cycles) (Gutiérrez et al., 2012;Campagne et al., 2015;Agawin et al., 2016). Atmospheric nitrogen (N 2 ) fixation associated with P. oceanica meadows are similar in rates or even higher than tropical seagrasses and may play a key role in maintaining the high productivity of the P. oceanica in oligotrophic waters (Agawin et al., 2016(Agawin et al., , 2017. N 2 fixation in P. oceanica is carried out by microorganisms called diazotrophs that can be found on the surface of the leaves, roots, and rhizomes (epiphytic population) or even on the inside of the roots (endophytic population) (Sohm et al., 2011;Agawin et al., 2019). Among the diazotrophic prokaryotes, a huge variety of diazotrophic cyanobacteria have been detected based on the sequence analysis of nifH gene (gene coding for the nitrogenase enzyme responsible for the N 2 fixation) on the leaves of P. oceanica (Agawin et al., 2016(Agawin et al., , 2017. In general, cyanobacteria are key components in the marine food web, contributing significantly to primary production in oligotrophic oceans (Agawin et al., 2000;Herrero and Flores, 2008). Compared with other phytoplankton taxa, cyanobacteria have elevated ratio of nitrogen (N):phosphorus (P) (a molar ratio above 25 compared with the general Redfield ratio of 16 in marine phytoplankton) and can be a consequence of having two light-harvesting complexes (Redfield, 1934;Geider and La Roche, 2002;Quigg et al., 2011). Changes affecting the N:P ratios in their environment by limiting concentration of N or P, could change their N:P tissue composition and may have consequences in their adaptation and survival and possibly the N 2 fixation activities of diazotrophic cyanobacteria (Sañudo-Wilhelmy et al., 2001;Sohm et al., 2011). Nonetheless, these versatile microorganisms may have several adaptive mechanisms to changes in their dynamic marine environment (e.g., nutrient availability) (Tandeau de Marsac and Houmard, 1993;Schwarz and Forchhammer, 2005;Herrero and Flores, 2008).
Phosphorus, (i.e., inorganic phosphorus, PO 4 3-), together with iron (Fe) are hypothesized to be the major limiting nutrients for N 2 fixation (Mills et al., 2004;Moore et al., 2009Moore et al., , 2013. Phosphorus is vital for the storage and retrieval system of genetic information (DNA/RNA), for the energy metabolism through ATP dependence (Kornberg, 1995;Santos-Beneit, 2015;Tiwari et al., 2015), and in most bacteria, it is important for the structure of the cell membrane. During P-starvation, microorganisms produce enzymes that are hydrolyze P-esters contained in dissolved organic phosphorus (DOP) releasing dissolved inorganic phosphorus (DIP), that the cells can utilize. These enzymes are called alkaline phosphatases (APases) and in marine bacteria they are included in three main families: PhoA, PhoX, and PhoD. APases are metalloenzymes that require metal co-factors. PhoA forms a coordinate with two zinc (Zn 2+ ) and one magnesium (Mg 2+ ) ions; PhoX forms a coordinate with three calcium (Ca 2+ ) and one/two Fe 3+ ions (Yong et al., 2014); and PhoD coordinates with an unknown number of Ca 2+ ions. In Bacillus subtilis model, PhoD has an active site formed with one Fe 3+ and two Ca 2+ ions . This information suggests the possible interaction between metals (e.g., Fe 3+ , Ca 2+ , Mg 2+ , and Zn 2+ ) in the mechanisms of P-acquisition involving APases. In Halothece sp. PCC 7418, two types of APases have been reported: PhoA (two copies) and PhoD (one copy). Calcium dependence was proven in PhoD in Halothece sp. PCC 7418 (Kageyama et al., 2011). However, Fe dependence of PhoD and the relative importance between these two types of APases (PhoA and PhoD) have not been demonstrated in Halothece sp. PCC 7418.
APases are included in what is known as the Pho regulon. It is a huge regulatory group of genes that control P-acquisition. Pho regulon is composed of elements related with (1) highaffinity phosphate transport (PstS, PstC, PstA, and PstB) and low-affinity phosphate transport, (2) extracellular enzymes capable of obtaining PO 4 3from organic phosphates (APases), and (3) polyphosphate metabolism (PpK, PpX, and PpA) as P reservoir or elements with unknown functions (PhoU) (Blanco et al., 2002;Yuan et al., 2006;Santos-Beneit, 2015). PhoU coordinates with metal cluster (Zn 2+ or Fe 3+ ), and may have a role in the control of autokinase activity of the PhoR and Pst systems (Gardner et al., 2014). The Pho regulon is mainly controlled by PhoR-PhoB, a two-component regulatory system (Santos-Beneit, 2015). PhoR is an inner-membrane histidine kinase, while PhoB is a transcriptional factor that recognizes and binds to consensus sequence named PHO box. In cyanobacteria, PHO box is formed by three tandem repeats of 8 bp separated by 5 bp, unlike PHO Box from Escherichia coli, formed by two direct repeats of 7 bp separated by 5 bp (Yuan et al., 2006;Su et al., 2007;Tiwari et al., 2015).
The P-acquisition mechanisms in bacteria are well studied in the Atlantic ocean, where Fe is shown to enhance the P-acquisition mechanisms in N 2 -fixing cyanobacterial species, Trichodesmium spp. and Crocosphaera watsonii (Fu et al., 2005;Dyhrman and Haley, 2006;Browning et al., 2017). However, there is scarcely any information about the relation between metals (e.g., Fe) and P-acquisition mechanisms in N 2 -fixing cyanobacteria found in association with the Mediterranean seagrass, P. oceanica, taking into account the multiple ecological benefits of this seagrass in the region. The Mediterranean Sea is oligotrophic, characterized by low water column PO 4 3concentrations and a decreasing gradient of PO 4 3concentrations from west to east basins (Tanhua et al., 2013). Knowledge on the P-acquisition mechanisms of N 2 -fixing organisms in an environment with limiting levels of PO 4 3is particularly important. Moreover, the Mediterranean Sea is subject to Saharan atmospheric dust deposition containing Fe (Statham and Hart, 2005), which can play a role in the P-acquisition mechanisms of the organisms.
To study, for the first time, the P-acquisition mechanisms in N 2 -fixing cyanobacteria associated with the dominant coastal ecosystem in the region (P. oceanica seagrass beds), we selected a diazotrophic unicellular cyanobacteria, Halothece sp. found on the leaves of P. oceanica (Agawin et al., 2017)  The most related culturable strain is Halothece sp. PCC 7418, and this was used as the test organism in this study. The halotolerant Halothece sp. PCC 7418 (originally called Synechococcus PCC 7418), also known as Aphanothece halophytica, was originally isolated from Solar Lake on the eastern shore of the Sinai Peninsula in 1972 (UniProt source). First, we made a genomic analyses of the Pho regulon to check the regulatory group of genes that control the P-acquisition mechanisms and then a structural analysis of PhoD (alkaline phosphatase type D) and PhoU (phosphate acquisition regulator) of Halothece sp. PCC 7418 to investigate the connection among metals (e.g., Ca 2+ and Fe 3+ ) and the P-acquisition mechanisms of this species. Second, we experimentally investigated how the availability of Fe affects the alkaline phosphatase activity (APA), their PO 4 3--uptake rates, and the magnitude of the effect under different levels of PO 4 3and NO 3 availability, and how the availability of PO 4 3and Fe affect Fe-uptake rates of the cells.

Genome Analysis
With the goal of comparing Halothece sp. PCC 7418 Pho regulon with its closest genomes (Luo et al., 2009), the distribution of the number of copies of Pho regulon components in selected strains was analyzed. The genome from Halothece sp. PCC 7418 (GenBank: NC_019779.1) and genomes from other closely related microorganisms were compared using the dedicated bacterial information system Pathosystems Resource Integration Center (PATRIC). This database, and the analysis tools included, offers an easy interface in which annotated genes that are included in different subsystems can be searched (Wattam et al., 2017).

Three-Dimensional Predicted Structures
Sequences of PhoD and PhoU in FASTA format were sent to the I-Tasser server for protein 3D-structure prediction (Zhang, 2008), with their domains previously checked in Pfam 32.0 (Finn et al., 2016

Strain and Culture Conditions
Halothece sp. PCC 7418, was obtained from the Pasteur Culture Collection of Cyanobacteria (PCC) and maintained in 250 ml acid-cleaned Quartz Erlenmeyer flasks containing 150 ml of ASNIII + Tu4X medium (initial pH 7.5) (Stanier et al., 1979).
The medium was supplemented with 0. and optimal NO 3 conditions. Cultures were maintained at the same conditions as described above for over 12 days. During the last day, PO 4 3-, Fe, and/or NO 3 were added to the different treatments to achieve optimal concentrations of PO 4 3-(45 μM), Fe (7.5 μM), and NO 3 -(4.4 mM) to evaluate the changes in the APA rates, and the new conditions were maintained for over 4 days. The different conditions of the experiments are shown in Table 1.
The importance of PhoD in Halothece sp. PCC 7418 was investigated by changing the availability of the metal co-factors for PhoA (Zn 2+ and Mg 2+ ). The method used was as described above in the initial main experiments except that the medium was depleted with Mg and Zn and the condition of PO 4 3and Fe was: [Medium PO 4 3--High Fe] under optimal NO 3 -.
All cultures were performed in duplicate, and the study parameters (APA, N 2 fixation, uptake rates of PO 4 3and Fe, TDP and/or P/Fe/Mn cellular content) were evaluated during the different phases of the culture (O.D 750 nm ≅ 0.01-0.2). A subsample of the cells (1.5 ml) was taken from the culture flasks during the experiment and were counted through flow cytometric analysis (as described below) to normalize the results per cell. All samples were manipulated in a class-100 clean hood, to avoid Fe contamination.

Alkaline Phosphatase Activity
Alkaline phosphatase activity (APA) was evaluated through a fluorometric assay, in which the hydrolysis of the fluorogenic substrate (S) 4-methylumbelliferyl phosphate (MUF-P, Sigma-Aldrich) to 4-methylumbelliferyl (MUF) was measured. Generally, an end point enzymatic assay was conducted with a concentration of 2 μM MUF-P during the exponential phase of the culture (O.D 750 nm ≅ 0.1). After 1 h incubation in darkness at room temperature, APA was measured in a microtiter plate that contained borate buffer at pH 10 (3:1 of sample:buffer). The MUF production (fmole MUF cell −1 h −1 ) was measured with a Cary Eclipse spectrofluorometer (FL0902M009, Agilent Technologies) at 359 nm (excitation) and 449 nm (emission) and using a calibration standard curve with commercial MUF (Sigma-Aldrich). conditions and the APA rate (fmole MUF cell −1 h −1 ) was calculated as the slope of the fitted line.

3-Uptake Rates, Nutrient Concentrations in the Culture Medium and in the Cells
Samples for the determination of PO 4 3and total dissolved P (TDP) were centrifuged for 15 min at 16,000 ×g under 4°C. The supernatant was collected from the centrifuged tubes and used for PO 4 3determinations following standard spectrophotometric methods (Hansen and Koroleff, 2007). TDP concentrations were also analyzed using the latter method after persulfate digestion. Samples for Fe analyses of culture media were filtered through sterile 0.2 μm filters (MFV5-025, FilterLab) at different times (initial and final) during the experiments. The metal (Fe) concentrations of culture medium were measured by inductively coupled plasma mass spectrometry (ICP-MS; iCap, Thermo Scientific), following the trace-metal clean techniques described in Tovar-Sanchez et al. (2006)  conditions. Specific PO 4 3uptake rates (pmole PO 4 3cell −1 day −1 ) were calculated as described in (Ghaffar et al., 2017). Briefly, specific PO 4 3uptake rates were calculated as the mass balance of PO 4 3over the multiple days by taking the differences of PO 4 3concentrations at two different times (T 0 -T 1 , T 0 -T 4 , and T 0 -T 10 ) and normalized by the number of cells counted at different time points (0, 1, 4, and 10) through the following equation: conditions). Initial and final Fe concentrations of the culture media were measured, and the difference between time = 0 and time = 10 (T 0 -T 10 ) was used to determine the Fe-uptake during the 10 days of the experiment. Specific Fe-uptake (fmole Fe cell −1 day −1 ) was calculated the same way as the specific PO 4 3--uptake rates described above.

Acetylene Reduction Assay
N 2 -fixing activities were measured with the acetylene reduction assay (ARA) method under known N 2 -fixing conditions for unicellular cyanobacteria (i.e., low NO 3 concentrations, anaerobic environment, dark phase of the photoperiod, Reddy et al., 1993), and under low-medium levels of Fe and in low-medium-high levels of PO 4 3-. A volume of 50 ml from treatments with [Low NO 3 -] condition at day 8 of the experiment was transferred to anaerobic tubes for cultivation for 2 days, and after which, ARA measurements were done following the method described in Agawin et al. (2014). Duplicate 10 ml samples of culture from each experimental tube, were filtered through 0.45 μm GF/F filters (MFV5-025, FilterLab). The filters were deposited in hermetic vials containing 1 ml of the corresponding culture medium. Acetylene (C 2 H 2 ) was added at 20% (v/v) final concentration in each vial using gas-tight Hamilton syringes. The filters were incubated in the vials for 3 h at room temperature in the dark. After 3 h incubation time, 10 ml of headspace gas were removed with a gas-tight Hamilton syringe from the incubation vials or tubes, transferred and stored in Hungate tubes and sealed with hot melt adhesive glue (SALKI, ref. 0430308) to minimize gas losses (Agawin et al., 2014). Ethylene and acetylene were determined using a GC (model HP-5890, Agilent Technologies) equipped with a flame ionization detector (FID). The column was a Varian wide-bore column (ref. CP7584) packed with CP-PoraPLOT U (27.5 m length, 0.53 mm inside diameter, 0.70 mm outside diameter, and 20 μm film thickness). Helium was used as carrier gas at a flow rate of 30 ml min −1 . Hydrogen and airflow rates were set at 30 and 365 ml min −1 , respectively. The split flow was used so that the carrier gas flow through the column was 4 ml min −1 at a pressure of 5 psi. Oven, injection, and detector temperatures were set at 52, 120, and 170°C, respectively. Ethylene produced was calculated using the equations in Stal (1988). The acetylene reduction rates were converted to N 2 fixation rates (pmole N 2 ml −1 h −1 ) using a factor of 4:1 (Jensen and Cox, 1983).

Statistical Analyses
Univariate Analysis of variance (ANOVA) factor analyses and post-hoc (Bonferroni) was used to study the effect of the nutrient treatment conditions to APA rates, P-cellular content and specific PO 4 3-, and Fe uptake rates. In other cases, where we want to highlight a specific point, we use individual t tests. Regression analyses were used to determine the relationships between P-cellular content vs. N 2 rates fixation, P-cellular content vs. Fe-cellular content and P/Fe-cellular content vs. other metals (i.e., Mn). The statistical analyses were performed using the SPSS program version 21 (IBM Corp year 2012).

Pho Regulon of Halothece sp. PCC 7418
The distribution of the number of copies of Pho regulon components of Halothece sp. PCC 7418 and its closest genomes (Luo et al., 2009) are shown in Figure 1 and Supplementary Table S1. The Gloeocapsa sp. PCC 7428 genome had the highest number of copies detected (up to 45), suggesting that this species is one of the better adapted species to P-limitation. On the other hand, Nostoc punctiforme PCC 73102 and Chroococcidiopsis thermalis PCC 7203 genomes had the lowest number of copies of the Pho regulon components. Our test microorganism Halothece sp. PCC 7418 genome was the fourth cyanobacterium containing more copies of the Pho regulon (26): 1 for phoU, 4 for pstS, 3 for pstC, 2 for pstA, 3 for pstB, 1 for phoR-phoB, 8 for APases, 1 for ppK, 1 for ppX, and 1 for ppA. With eight copies of APases, it was the second cyanobacterium containing more APases (8) (Figure 2B). All these amino acids are described in 2YEQ as the active site and coordinate with two Ca 2+ and one Fe 3+ ions . Only one substitution was detected in Asp 249, where in 2YEQ is Asn 216. The in-silico results described above of PhoD and how it coordinates with Ca 2+ and Fe 3+ ions in its active site in Halothece sp. PCC 7418 corroborates with the results of the experiment testing the relative importance of PhoD and PhoA in Halothece sp. PCC 7418, showing that the APA rates, with the depletion of Mg 2+ and Zn 2+ which are the metal co-factors of PhoA, did not differ considerably with sufficient availability of Mg 2+ and Zn 2+ (Figure 3). This suggests that PhoD (and not PhoA) is the more active APase in Halothece sp. PCC 7418.

Three-Dimensional Structure of PhoU
Annotated PhoU had 224 amino acids (aa) and presented two PhoU domains. The predicted structure of PhoU (C-score = 0.55, estimated TM-score = 0.79 ± 0.09, estimated RMSD = 4.5 ± 2.9 Å) had seven α-helix without β-chains. The protein with more structure homology was PhoU of P. aeruginosa (4Q25) of 250 aa with an identity of 32.5% and coverage of 93.3%. Sequence alignment with 4Q25 showed 27.45% of identity and we used this alignment to describe its metal clusters (Figures 2C,D). Results showed that Halothece sp. PCC 7418 using 4Q25 as a template displayed at least one metal cluster, and possibly a second one, forming a trinuclear metal site with three Fe and tetranuclear metal site with three Fe and one nickel (Ni). The first cluster was complete and had the same aa as P. aeruginosa

Alkaline Phosphatase Activity in Halothece sp. PCC 7418
Generally, APA rates were significantly higher (p < 0.05) in [Low NO 3 -] conditions compared with optimal NO 3 conditions ( Figure 4A). Under [Low NO 3 -] APA rates were ≈ 7 times higher in [Low-Medium PO 4 3-] and ≈ 77 times higher in [High PO 4 3-] compared with their rates under optimal NO 3 conditions. Moreover, under optimal NO 3 conditions, APA rates did not have significant differences among the treatments ( Figure 4A). Under [Low NO 3 -], treatment combinations of PO 4 3and Fe levels had a significant effect on APA rates (ANOVA, p < 0.05), where the rates were significant higher (p < 0.05) at the highest Fe levels and at low to medium PO 4 3levels, compared with other treatment combinations ( Figure 4A). Figure 4B shows the differences in the kinetics of APA for treatments under [Low NO 3 -] at low and medium PO 4 3levels and low and high Fe levels. At high Fe levels with low to medium PO 4 3levels, the V vs. S curve did not reach saturation levels with the maximum S added (5 μM MUF-P). The Vmax and Km, calculated using the available data for these treatments, were: Vmax   (Figure 6A). On the other hand, specific PO 4 3--uptake rates under N 2 -fixing conditions [Low NO 3 -] and optimal NO 3 conditions generally did not vary significantly (ANOVA, p > 0.05) among treatment combinations ( Figure 6B). However, specific t-tests conducted under [Low NO 3 -] conditions, showed PO 4 3--uptake rates to be on average 200 times significantly higher (p < 0.05) than the rates under optimal conditions of NO 3 in T 0 -T 4 and T 0 -T 10 in low to medium Fe levels ( Figure 6B). Different concentrations of Fe in [High PO 4 3-] did not show significant differences in PO 4 3--uptake rates (p > 0.05) (Figure 6B). The time course of depletion of total dissolved phosphate (TDP) in the culture media showed that under optimal NO 3 conditions, the media were depleted with TDP while under NO 3 starvation conditions, the cells were not capable in POis represented as Pi. Values are the mean, and the error bar is the spanning range between the duplicate measurements. Different letters indicate pairwise significant differences (p < 0.05) among treatments using a post-hoc test (Bonferroni) after ANOVA over the whole dataset was done, and asterisks (*) indicate significant differences (p < 0.05) between [Low depleting TDP from the media (Figure 7A). Fe did not have a significant effect in TDP depletion (p > 0.05). The time course of depletion of TDP in the re-inoculum conditions at [Low PO 4 3--Low Fe] (under NO 3 starvation and NO 3 optimal conditions), showed the same tendency, in which under NO 3 starvation conditions, TDP was not depleted (Figure 7B). Figure 8 shows the specific Fe-uptake rates at different levels of PO 4 3and Fe under N 2 -fixing conditions. Results showed that generally, specific Fe-uptake rates varied significantly at different treatment combinations of PO 4 3and Fe (ANOVA, p < 0.05). Fe-uptake rates were significantly higher (p < 0.05) at [High PO 4 3-] conditions compared to [Low PO 4 3-] and [Medium PO 4 3-] conditions. There were also significant differences (p < 0.05) of increased Fe-uptake with increasing availability of Fe.

Phosphorus-Cellular Content and Its Relationship With N 2 Fixation and Iron-Cellular Content
Phosphorus cellular content of Halothece sp. PCC 7418 showed significant positive linear correlation with N 2 fixation rates (p < 0.05, r 2 = 0.86, n = 12) ( Figure 9A). Moreover, the P-cellular content of the cells showed significant positive linear correlation with their Fe contents (p < 0.05, r 2 = 0.71, n = 18) ( Figure 9B). The P and Fe-cellular contents of the cells did not show significant correlations with other metals (i.e., Mn).

DISCUSSION
Pho Regulon and the Three-Dimensional Structure of PhoD and PhoU of Halothece sp. PCC 7418: Elucidating the Role of Iron as Co-factor The Pho regulon of Halothece sp. PCC 7418 is composed of genes whose protein products are involved in different functions: autokinase activity of PhoR and phosphate transport (PhoU); high-affinity phosphate transport (PstS, PstC, PstA, and PstB), in a two-component regulatory system (PhoR-PhoB); extracellular enzymes capable of obtaining PO 4 3from organic phosphates (Alkaline Phosphatases, APases); and polyphosphate metabolism (PpK, PpX, and PpA) (Santos-Beneit, 2015). However, no low-affinity transporters were annotated while some studies demonstrated that this strain exhibited low-affinities transporters (Tripathi et al., 2013). Halothece sp. PCC 7418 contains a Pho regulon with 11 distinct genes in single or multiple copies altogether accounting 26 distinct loci in the whole genome, suggesting that Halothece sp. PCC 7418 is well adapted to survive to P-limiting conditions. In model strains whose P-acquisition mechanisms are well studied such as Trichodesmium spp. and Crocosphaera watsonii they only have 15 copies and 19 copies respectively in their Pho regulon (Fu et al., 2005;Dyhrman and Haley, 2006). Genome analysis indicated that Halothece sp. PCC 7418 and Gloeocapsa sp. PCC 7428 were the strains with more copies of Alkaline Phosphatases (APase), 8 and 19, respectively (Figure 1). These two cyanobacteria are halotolerant species, and there are studies that suggest that salt stress enhance APA in halophytic strains (Kageyama et al., 2011). In a previous study (Kageyama et al., 2011), Halothece sp. PCC 7418 only showed A B FIGURE 7 | TDP (μM) under three APases: two PhoA and one PhoD. Of the eight APases found in our study for the same species, one of them is also annotated as PhoD and the rest are not annotated to a specific type of APase. PhoD, together with PhoX, is one of the most abundant APases in marine habitats and its activity may be controlled by availability of its metal co-factor(s) (e.g., Fe 3+ , Ca 2+ , Mg 2+ , and Zn 2+ ) (Luo et al., 2009;Zeng et al., 2011).
Three-dimensional analyses with PhoD of Halothece sp. PCC 7418 revealed its active site as a homologue to the crystal structure of PhoD of B. subtilis with two Ca 2+ and one Fe 3+ ions as co-factors ( Figure 2B; Rodriguez et al., 2014). Previous studies on APase activity in Halothece sp. PCC 7418 indicated Ca 2+ dependence of PhoD (Kageyama et al., 2011) but the Fe 3+ dependence was not investigated. The experiment conducted here wherein the omission of Mg 2+ and Zn 2+ (but not Fe 3+ in the culture medium) did not result in any significant changes in APase activity (Figure 3), suggesting that the APases of Halothece sp. PCC 7418 (i.e., PhoD) do not require these metals (Mg 2+ and Zn 2+ ) as co-factors as in the case of PhoA (Kageyama et al., 2011), and the most active APase could be PhoD.
Iron is not only important as a co-factor for the activities of APase but can be essential in other components of Pho regulon like PhoU in which the results of the 3D-dimensional analyses in this study showed PhoU of Halothece sp. PCC 7418 forming at least one Fe-containing metal cluster, and possible a second cluster (Figures 2C,D), using as a model, the PhoU of P. aeruginosa (4Q25). PhoU can participate in the PO 4 3transport across the cell membranes of bacteria in PO -is represented as Pi. Values are the mean, and the error bar is the spanning range between the duplicate measurements, and letters indicate significant differences (p < 0.05) between treatments using a post-hoc test (Bonferroni) after ANOVA over the whole dataset was done.

A B
FIGURE 9 | Linear regression analyses (A) between P-cellular content and N 2 fixation and (B) between P and Fe-cellular content. Cellular content was measured in [Low Frontiers in Microbiology | www.frontiersin.org the regulation of the phosphate-specific transport systems (Santos-Beneit, 2015) and in controlling cellular phosphate metabolism (Lubin et al., 2015). The specific role of PhoU in Halothece sp. PCC 7418, however, remains to be investigated.  (Figures 4A,B and Figure 5A), confirming the regulatory role of Fe in the APase (i.e., PhoD) in this species as we predicted in our 3D-structural analyses of its PhoD ( Figure 2B). However, the effect of Fe availability on the rates of APA in Halothece sp. PCC 7418 depends on the availability of inorganic sources of nitrogen (i.e., NO 3 -) wherein at low NO 3 concentrations, increasing Fe availability enhanced the APA rates ( Figures 4A,B). We showed that under [Low NO 3 -] and at high Fe levels, APA was not saturated ( Figure 4B). We hypothesized that under these conditions, the Vmax of APases from Halothece sp. PCC 7418 is so high that increasing MUF-P concentrations, up to 10 μM (in the other assays that were additionally conducted),was not high enough to saturate the enzyme because of the enhancement of APA by high levels of the Fe co-factor. At high or optimal NO 3 concentrations, APA rates in general are lower than in NO 3 starvation conditions ( Figure 5A) and even lower than in [Low NO 3 -] treatments ( Figure 4A).

Alkaline Phosphatase Activity in
These results can be due to peculiar characteristics of the N 2 fixation process. High concentrations of readily assimilable forms of dissolved inorganic nitrogen (DIN,i.e.,NH 4 , NO 3 -) are known to inhibit N 2 fixation as evidenced by DIN inhibition studies (Knapp, 2012). The N 2 fixation process (N 2 + 8e − + 16ATP + 8H + → 2NH 3 + H 2 + 16ADP + 16 PO 4 3-) is an energetically costly processes requiring 16 ATPs and 25% more energy is needed to reduce N 2 than to reduce NO 3 to NH 4 . A N 2 -fixing cell such as Halothece sp. PCC 7418 would rather reduce first the available NO 3 than to fix N 2 . Conversely, the N 2 -fixing process is stimulated with low NO 3 availability (Manhart and Wong, 1980;Nelson et al., 1982). Since the energy (ATP) to fuel N 2 fixation is dependent on PO 4 3-, the demand for PO 4 3is theoretically enhanced when the cells are doing N 2 fixation (in conditions under low NO 3 availability).
Thus, APase activities are expected to be stimulated under low NO 3 conditions, and consequently depend on the availability of Fe because APases such as PhoD may have Fe as co-factor. Moreover, Fe is an important structural component of the nitrogenase enzyme catalyzing the N 2 fixation process. Nitrogenase contains 38 Fe atoms per holoenzyme since nitrogenase is characterized by slow reaction rates the N 2 -fixers need a large cellular pool of this enzyme, and thus more Fe is needed (Hoffman et al., 2014). The enhanced rates of APase under N 2 -fixing conditions (low NO 3 availability) and high Fe availability with low PO 4 3levels is expected as APases activities are induced with low PO 4 3levels in the medium (Romano et al., 2017). The control of NO 3 and PO 4 3-availabilities in APase activities for N 2 -fixing cells such as Halothece sp. PCC 7418 is further supported here with the results of decreased APA rates when NO 3 -, PO 4 3and Fe were added to cells growing previously with low PO 4 3-, low Fe and/or low NO 3 levels ( Figure 5B).
Phosphorus and Iron-Uptake and Cellular Contents in Halothece sp. PCC 7418 The PO 4 3--uptake measurements in Halothece sp. PCC 7418 were done in the experimental units with high PO 4 3levels because (1) the method used for PO 4 3analyses was not sensitive enough to measure very low levels of PO 4 3-(≤ 0.1 μM), and (2) APase activities are not induced at high PO 4 3levels allowing us to evaluate if Fe is also important in PO 4 3transport mechanisms and not only in APase activities. PO 4 3--uptake rates of Halothece sp. PCC 7418 was significantly higher under N 2 -fixing conditions ([Low NO 3 -]) than in non-N 2 fixing conditions due to the high demand of P for the energy costly N 2 fixation ( Figure 6B). The dependence of N 2 fixation on P in Halothece sp. PCC 7418 is evidenced here with the significant linear correlation between cellular P content of the cells and their rates of N 2 fixation ( Figure 9A), consistent with studies carried out in Trichodesmium spp. in the Atlantic (Sañudo-Wilhelmy et al., 2001). In addition, not only N 2 -fixing conditions can enhance the P-requirements of cyanobacteria. It is also reported that under nitrogen limitation, phytoplankton can accumulate carbohydrates and phospholipids, increasing their P-cellular content (Liefer et al., 2019). Different concentrations of Fe, however, did not show significant differences in PO 4 3--uptake at high levels of PO 4 3availability. This suggests that PO 4 3--uptake mechanisms in this case are not dependent on Fe levels or the Fe present in all treatments (from low to high Fe concentrations) are sufficient for the cells (Figure 6B). The latter case may be most likely since we found significant correlations between the P-cellular and Fe-cellular content of the cells (Figure 9B). These results are also consistent with our data that the highest P-cellular content was found at high Fe levels (Figure 6A), suggesting the narrow connection between P and Fe. The relation between P and Fe cellular contents is also supported by evidences that high concentrations of elemental P are found associated (or co-localized spatially) with Fe within the cells of phytoplankton [Chlorella sp. and Chlamydomonas sp. (Diaz et al., 2009)]. The Fe-uptake measurements in Halothece sp. PCC 7418 in N 2 -fixing conditions revealed that Fe-uptake was correlated with P with high Fe-uptake rates at higher PO 4 3levels (Figure 8). This may be due to the P-dependence (ATP) of Fe transporters (Noinaj et al., 2010;Kranzler et al., 2013). Results also show the tendency of higher Fe-uptake rates in higher concentrations of Fe in the media, suggesting a passive transport of this metal in Halothece sp. PCC 7418. However, this needs to be further investigated.
The time course of depletion of total dissolved phosphate (TDP) in the media (Figure 7A), showed that under NO 3 starvation conditions, cells did not deplete TDP, and even increased at the final stage of the experiment suggesting a liberation of cellular TDP of dying cells. Extreme NO 3 starvation conditions are suggested here to be detrimental to the growth of Halothece sp. PCC 7418 and may have consequences on their P-uptake mechanisms, explaining why APA rates were lower than in [Low NO 3 -] conditions. Even when the nutrients ( PO 4 3-, Fe and/or NO 3 -) were re-inoculated in the cultures that were previously starved with NO 3 -, the cells did not acclimate and were not capable of depleting TDP from the media ( Figure 7B). Whereas, much of the previous research has focused on the inhibition or sensitivity of N 2 fixation to increased availability of dissolved inorganic nitrogen (e.g., NO 3 -, NH 4 + ) (Knapp, 2012), investigations on the physiological conditions for growth of N 2 -fixers are few. Spiller and Shanmugam (1987), gave some evidences that a unicellular species of marine N 2 -fixer Synechococcus sp. strain SF1 (isolated from macroalgae, Sargassum fluitans) is dependent on the presence and type of carbon (C) source to support its growth with N 2 as the sole nitrogen source. Their results showed, for example, that without the addition of C source (e.g., HCO 3 − ), there was no growth of the species tested with N 2 as the sole source. Moreover, some studies have reported less cell yield of unicellular N 2 fixers when grown with N 2 as sole N source compared with addition of NO 3 since N 2 fixation is an energetically costly process (Spiller and Shanmugam, 1987;Agawin et al., 2007). Our result that extreme NO 3 starvation condition (at nM levels close to N 2 as sole source) is suggested to be detrimental to the growth of Halothece sp. PCC 7418 may be due to the type of C source (glucose and citrate) in our treatments which may not be the optimum for growth of this species with N 2 as sole N source. This hypothesis however needs more investigations.
In summary, this is the first study investigating the interaction between PO 4 3-, Fe, and NO 3 availabilities in the P-acquisition mechanisms of a unicellular N 2 -fixing bacteria found in association with the Mediterranean seagrass P. oceanica. The results suggest that APase activities under inorganic P-limited conditions are enhanced by increased Fe availabilities. The PO 4 3and Fe dependence of Halothece sp. PCC 7418 depends whether they are grown in N 2 -fixing conditions (i.e., low NO 3 levels) or not. Genomic and structural analyses have also shown the tight association between P-acquisition mechanisms and Fe in Halothece sp. PCC 7418. Studies combining genomic and protein structural analyses and experimental approaches are important to investigate in detail the control of environmental factors (e.g., availability of metals and nutrients) to the functioning of N 2 -fixing organisms found in important species of seagrasses.

DATA AVAILABILITY
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

AUTHOR CONTRIBUTIONS
VF-J and NA designed the experiments. VF-J conducted all experiments and led the writing of the paper. All authors contributed to the writing and review of the manuscript, and NA is the supervisor of the laboratory.

FUNDING
This work was supported by funding to NA through the Agencia Estatal de Investigación and the European Regional Development Funds project (CTM2016-75457-P).